1 | /**********************************************************************
|
---|
2 | regexec.c - Onigmo (Oniguruma-mod) (regular expression library)
|
---|
3 | **********************************************************************/
|
---|
4 | /*-
|
---|
5 | * Copyright (c) 2002-2008 K.Kosako <sndgk393 AT ybb DOT ne DOT jp>
|
---|
6 | * Copyright (c) 2011-2016 K.Takata <kentkt AT csc DOT jp>
|
---|
7 | * All rights reserved.
|
---|
8 | *
|
---|
9 | * Redistribution and use in source and binary forms, with or without
|
---|
10 | * modification, are permitted provided that the following conditions
|
---|
11 | * are met:
|
---|
12 | * 1. Redistributions of source code must retain the above copyright
|
---|
13 | * notice, this list of conditions and the following disclaimer.
|
---|
14 | * 2. Redistributions in binary form must reproduce the above copyright
|
---|
15 | * notice, this list of conditions and the following disclaimer in the
|
---|
16 | * documentation and/or other materials provided with the distribution.
|
---|
17 | *
|
---|
18 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
|
---|
19 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
---|
20 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
---|
21 | * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
|
---|
22 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
---|
23 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
---|
24 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
---|
25 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
---|
26 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
---|
27 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
---|
28 | * SUCH DAMAGE.
|
---|
29 | */
|
---|
30 |
|
---|
31 | #include "regint.h"
|
---|
32 |
|
---|
33 | #ifdef RUBY
|
---|
34 | # undef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
35 | #else
|
---|
36 | # define USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
37 | #endif
|
---|
38 |
|
---|
39 | #ifndef USE_TOKEN_THREADED_VM
|
---|
40 | # ifdef __GNUC__
|
---|
41 | # define USE_TOKEN_THREADED_VM 1
|
---|
42 | # else
|
---|
43 | # define USE_TOKEN_THREADED_VM 0
|
---|
44 | # endif
|
---|
45 | #endif
|
---|
46 |
|
---|
47 | #ifdef RUBY
|
---|
48 | # define ENC_DUMMY_FLAG (1<<24)
|
---|
49 | static inline int
|
---|
50 | rb_enc_asciicompat(OnigEncoding enc)
|
---|
51 | {
|
---|
52 | return ONIGENC_MBC_MINLEN(enc)==1 && !((enc)->ruby_encoding_index & ENC_DUMMY_FLAG);
|
---|
53 | }
|
---|
54 | # undef ONIGENC_IS_MBC_ASCII_WORD
|
---|
55 | # define ONIGENC_IS_MBC_ASCII_WORD(enc,s,end) \
|
---|
56 | (rb_enc_asciicompat(enc) ? (ISALNUM(*s) || *s=='_') : \
|
---|
57 | onigenc_ascii_is_code_ctype( \
|
---|
58 | ONIGENC_MBC_TO_CODE(enc,s,end),ONIGENC_CTYPE_WORD,enc))
|
---|
59 | #endif /* RUBY */
|
---|
60 |
|
---|
61 | #ifdef USE_CRNL_AS_LINE_TERMINATOR
|
---|
62 | # define ONIGENC_IS_MBC_CRNL(enc,p,end) \
|
---|
63 | (ONIGENC_MBC_TO_CODE(enc,p,end) == 13 && \
|
---|
64 | ONIGENC_MBC_TO_CODE(enc,(p+enclen(enc,p,end)),end) == 10)
|
---|
65 | # define ONIGENC_IS_MBC_NEWLINE_EX(enc,p,start,end,option,check_prev) \
|
---|
66 | is_mbc_newline_ex((enc),(p),(start),(end),(option),(check_prev))
|
---|
67 | static int
|
---|
68 | is_mbc_newline_ex(OnigEncoding enc, const UChar *p, const UChar *start,
|
---|
69 | const UChar *end, OnigOptionType option, int check_prev)
|
---|
70 | {
|
---|
71 | if (IS_NEWLINE_CRLF(option)) {
|
---|
72 | if (ONIGENC_MBC_TO_CODE(enc, p, end) == 0x0a) {
|
---|
73 | if (check_prev) {
|
---|
74 | const UChar *prev = onigenc_get_prev_char_head(enc, start, p, end);
|
---|
75 | if ((prev != NULL) && ONIGENC_MBC_TO_CODE(enc, prev, end) == 0x0d)
|
---|
76 | return 0;
|
---|
77 | else
|
---|
78 | return 1;
|
---|
79 | }
|
---|
80 | else
|
---|
81 | return 1;
|
---|
82 | }
|
---|
83 | else {
|
---|
84 | const UChar *pnext = p + enclen(enc, p, end);
|
---|
85 | if (pnext < end &&
|
---|
86 | ONIGENC_MBC_TO_CODE(enc, p, end) == 0x0d &&
|
---|
87 | ONIGENC_MBC_TO_CODE(enc, pnext, end) == 0x0a)
|
---|
88 | return 1;
|
---|
89 | if (ONIGENC_IS_MBC_NEWLINE(enc, p, end))
|
---|
90 | return 1;
|
---|
91 | return 0;
|
---|
92 | }
|
---|
93 | }
|
---|
94 | else {
|
---|
95 | return ONIGENC_IS_MBC_NEWLINE(enc, p, end);
|
---|
96 | }
|
---|
97 | }
|
---|
98 | #else /* USE_CRNL_AS_LINE_TERMINATOR */
|
---|
99 | # define ONIGENC_IS_MBC_NEWLINE_EX(enc,p,start,end,option,check_prev) \
|
---|
100 | ONIGENC_IS_MBC_NEWLINE((enc), (p), (end))
|
---|
101 | #endif /* USE_CRNL_AS_LINE_TERMINATOR */
|
---|
102 |
|
---|
103 | #ifdef USE_CAPTURE_HISTORY
|
---|
104 | static void history_tree_free(OnigCaptureTreeNode* node);
|
---|
105 |
|
---|
106 | static void
|
---|
107 | history_tree_clear(OnigCaptureTreeNode* node)
|
---|
108 | {
|
---|
109 | int i;
|
---|
110 |
|
---|
111 | if (IS_NOT_NULL(node)) {
|
---|
112 | for (i = 0; i < node->num_childs; i++) {
|
---|
113 | if (IS_NOT_NULL(node->childs[i])) {
|
---|
114 | history_tree_free(node->childs[i]);
|
---|
115 | }
|
---|
116 | }
|
---|
117 | for (i = 0; i < node->allocated; i++) {
|
---|
118 | node->childs[i] = (OnigCaptureTreeNode* )0;
|
---|
119 | }
|
---|
120 | node->num_childs = 0;
|
---|
121 | node->beg = ONIG_REGION_NOTPOS;
|
---|
122 | node->end = ONIG_REGION_NOTPOS;
|
---|
123 | node->group = -1;
|
---|
124 | xfree(node->childs);
|
---|
125 | node->childs = (OnigCaptureTreeNode** )0;
|
---|
126 | }
|
---|
127 | }
|
---|
128 |
|
---|
129 | static void
|
---|
130 | history_tree_free(OnigCaptureTreeNode* node)
|
---|
131 | {
|
---|
132 | history_tree_clear(node);
|
---|
133 | xfree(node);
|
---|
134 | }
|
---|
135 |
|
---|
136 | static void
|
---|
137 | history_root_free(OnigRegion* r)
|
---|
138 | {
|
---|
139 | if (IS_NOT_NULL(r->history_root)) {
|
---|
140 | history_tree_free(r->history_root);
|
---|
141 | r->history_root = (OnigCaptureTreeNode* )0;
|
---|
142 | }
|
---|
143 | }
|
---|
144 |
|
---|
145 | static OnigCaptureTreeNode*
|
---|
146 | history_node_new(void)
|
---|
147 | {
|
---|
148 | OnigCaptureTreeNode* node;
|
---|
149 |
|
---|
150 | node = (OnigCaptureTreeNode* )xmalloc(sizeof(OnigCaptureTreeNode));
|
---|
151 | CHECK_NULL_RETURN(node);
|
---|
152 | node->childs = (OnigCaptureTreeNode** )0;
|
---|
153 | node->allocated = 0;
|
---|
154 | node->num_childs = 0;
|
---|
155 | node->group = -1;
|
---|
156 | node->beg = ONIG_REGION_NOTPOS;
|
---|
157 | node->end = ONIG_REGION_NOTPOS;
|
---|
158 |
|
---|
159 | return node;
|
---|
160 | }
|
---|
161 |
|
---|
162 | static int
|
---|
163 | history_tree_add_child(OnigCaptureTreeNode* parent, OnigCaptureTreeNode* child)
|
---|
164 | {
|
---|
165 | # define HISTORY_TREE_INIT_ALLOC_SIZE 8
|
---|
166 |
|
---|
167 | if (parent->num_childs >= parent->allocated) {
|
---|
168 | int n, i;
|
---|
169 |
|
---|
170 | if (IS_NULL(parent->childs)) {
|
---|
171 | n = HISTORY_TREE_INIT_ALLOC_SIZE;
|
---|
172 | parent->childs =
|
---|
173 | (OnigCaptureTreeNode** )xmalloc(sizeof(OnigCaptureTreeNode*) * n);
|
---|
174 | CHECK_NULL_RETURN_MEMERR(parent->childs);
|
---|
175 | }
|
---|
176 | else {
|
---|
177 | OnigCaptureTreeNode** tmp;
|
---|
178 | n = parent->allocated * 2;
|
---|
179 | tmp =
|
---|
180 | (OnigCaptureTreeNode** )xrealloc(parent->childs,
|
---|
181 | sizeof(OnigCaptureTreeNode*) * n);
|
---|
182 | if (tmp == 0) {
|
---|
183 | history_tree_clear(parent);
|
---|
184 | return ONIGERR_MEMORY;
|
---|
185 | }
|
---|
186 | parent->childs = tmp;
|
---|
187 | }
|
---|
188 | for (i = parent->allocated; i < n; i++) {
|
---|
189 | parent->childs[i] = (OnigCaptureTreeNode* )0;
|
---|
190 | }
|
---|
191 | parent->allocated = n;
|
---|
192 | }
|
---|
193 |
|
---|
194 | parent->childs[parent->num_childs] = child;
|
---|
195 | parent->num_childs++;
|
---|
196 | return 0;
|
---|
197 | }
|
---|
198 |
|
---|
199 | static OnigCaptureTreeNode*
|
---|
200 | history_tree_clone(OnigCaptureTreeNode* node)
|
---|
201 | {
|
---|
202 | int i, r;
|
---|
203 | OnigCaptureTreeNode *clone, *child;
|
---|
204 |
|
---|
205 | clone = history_node_new();
|
---|
206 | CHECK_NULL_RETURN(clone);
|
---|
207 |
|
---|
208 | clone->beg = node->beg;
|
---|
209 | clone->end = node->end;
|
---|
210 | for (i = 0; i < node->num_childs; i++) {
|
---|
211 | child = history_tree_clone(node->childs[i]);
|
---|
212 | if (IS_NULL(child)) {
|
---|
213 | history_tree_free(clone);
|
---|
214 | return (OnigCaptureTreeNode* )0;
|
---|
215 | }
|
---|
216 | r = history_tree_add_child(clone, child);
|
---|
217 | if (r != 0) {
|
---|
218 | history_tree_free(child);
|
---|
219 | history_tree_free(clone);
|
---|
220 | return (OnigCaptureTreeNode* )0;
|
---|
221 | }
|
---|
222 | }
|
---|
223 |
|
---|
224 | return clone;
|
---|
225 | }
|
---|
226 |
|
---|
227 | extern OnigCaptureTreeNode*
|
---|
228 | onig_get_capture_tree(OnigRegion* region)
|
---|
229 | {
|
---|
230 | return region->history_root;
|
---|
231 | }
|
---|
232 | #endif /* USE_CAPTURE_HISTORY */
|
---|
233 |
|
---|
234 | extern void
|
---|
235 | onig_region_clear(OnigRegion* region)
|
---|
236 | {
|
---|
237 | int i;
|
---|
238 |
|
---|
239 | for (i = 0; i < region->num_regs; i++) {
|
---|
240 | region->beg[i] = region->end[i] = ONIG_REGION_NOTPOS;
|
---|
241 | }
|
---|
242 | #ifdef USE_CAPTURE_HISTORY
|
---|
243 | history_root_free(region);
|
---|
244 | #endif
|
---|
245 | }
|
---|
246 |
|
---|
247 | extern int
|
---|
248 | onig_region_resize(OnigRegion* region, int n)
|
---|
249 | {
|
---|
250 | region->num_regs = n;
|
---|
251 |
|
---|
252 | if (n < ONIG_NREGION)
|
---|
253 | n = ONIG_NREGION;
|
---|
254 |
|
---|
255 | if (region->allocated == 0) {
|
---|
256 | region->beg = (OnigPosition* )xmalloc(n * sizeof(OnigPosition));
|
---|
257 | if (region->beg == 0)
|
---|
258 | return ONIGERR_MEMORY;
|
---|
259 |
|
---|
260 | region->end = (OnigPosition* )xmalloc(n * sizeof(OnigPosition));
|
---|
261 | if (region->end == 0) {
|
---|
262 | xfree(region->beg);
|
---|
263 | return ONIGERR_MEMORY;
|
---|
264 | }
|
---|
265 |
|
---|
266 | region->allocated = n;
|
---|
267 | }
|
---|
268 | else if (region->allocated < n) {
|
---|
269 | OnigPosition *tmp;
|
---|
270 |
|
---|
271 | region->allocated = 0;
|
---|
272 | tmp = (OnigPosition* )xrealloc(region->beg, n * sizeof(OnigPosition));
|
---|
273 | if (tmp == 0) {
|
---|
274 | xfree(region->beg);
|
---|
275 | xfree(region->end);
|
---|
276 | return ONIGERR_MEMORY;
|
---|
277 | }
|
---|
278 | region->beg = tmp;
|
---|
279 | tmp = (OnigPosition* )xrealloc(region->end, n * sizeof(OnigPosition));
|
---|
280 | if (tmp == 0) {
|
---|
281 | xfree(region->beg);
|
---|
282 | xfree(region->end);
|
---|
283 | return ONIGERR_MEMORY;
|
---|
284 | }
|
---|
285 | region->end = tmp;
|
---|
286 |
|
---|
287 | region->allocated = n;
|
---|
288 | }
|
---|
289 |
|
---|
290 | return 0;
|
---|
291 | }
|
---|
292 |
|
---|
293 | static int
|
---|
294 | onig_region_resize_clear(OnigRegion* region, int n)
|
---|
295 | {
|
---|
296 | int r;
|
---|
297 |
|
---|
298 | r = onig_region_resize(region, n);
|
---|
299 | if (r != 0) return r;
|
---|
300 | onig_region_clear(region);
|
---|
301 | return 0;
|
---|
302 | }
|
---|
303 |
|
---|
304 | extern int
|
---|
305 | onig_region_set(OnigRegion* region, int at, int beg, int end)
|
---|
306 | {
|
---|
307 | if (at < 0) return ONIGERR_INVALID_ARGUMENT;
|
---|
308 |
|
---|
309 | if (at >= region->allocated) {
|
---|
310 | int r = onig_region_resize(region, at + 1);
|
---|
311 | if (r < 0) return r;
|
---|
312 | }
|
---|
313 |
|
---|
314 | region->beg[at] = beg;
|
---|
315 | region->end[at] = end;
|
---|
316 | return 0;
|
---|
317 | }
|
---|
318 |
|
---|
319 | extern void
|
---|
320 | onig_region_init(OnigRegion* region)
|
---|
321 | {
|
---|
322 | region->num_regs = 0;
|
---|
323 | region->allocated = 0;
|
---|
324 | region->beg = (OnigPosition* )0;
|
---|
325 | region->end = (OnigPosition* )0;
|
---|
326 | region->history_root = (OnigCaptureTreeNode* )0;
|
---|
327 | }
|
---|
328 |
|
---|
329 | extern OnigRegion*
|
---|
330 | onig_region_new(void)
|
---|
331 | {
|
---|
332 | OnigRegion* r;
|
---|
333 |
|
---|
334 | r = (OnigRegion* )xmalloc(sizeof(OnigRegion));
|
---|
335 | if (r)
|
---|
336 | onig_region_init(r);
|
---|
337 | return r;
|
---|
338 | }
|
---|
339 |
|
---|
340 | extern void
|
---|
341 | onig_region_free(OnigRegion* r, int free_self)
|
---|
342 | {
|
---|
343 | if (r) {
|
---|
344 | if (r->allocated > 0) {
|
---|
345 | if (r->beg) xfree(r->beg);
|
---|
346 | if (r->end) xfree(r->end);
|
---|
347 | r->allocated = 0;
|
---|
348 | }
|
---|
349 | #ifdef USE_CAPTURE_HISTORY
|
---|
350 | history_root_free(r);
|
---|
351 | #endif
|
---|
352 | if (free_self) xfree(r);
|
---|
353 | }
|
---|
354 | }
|
---|
355 |
|
---|
356 | extern void
|
---|
357 | onig_region_copy(OnigRegion* to, const OnigRegion* from)
|
---|
358 | {
|
---|
359 | #define RREGC_SIZE (sizeof(int) * from->num_regs)
|
---|
360 | int i, r;
|
---|
361 |
|
---|
362 | if (to == from) return;
|
---|
363 |
|
---|
364 | r = onig_region_resize(to, from->num_regs);
|
---|
365 | if (r) return;
|
---|
366 |
|
---|
367 | for (i = 0; i < from->num_regs; i++) {
|
---|
368 | to->beg[i] = from->beg[i];
|
---|
369 | to->end[i] = from->end[i];
|
---|
370 | }
|
---|
371 | to->num_regs = from->num_regs;
|
---|
372 |
|
---|
373 | #ifdef USE_CAPTURE_HISTORY
|
---|
374 | history_root_free(to);
|
---|
375 |
|
---|
376 | if (IS_NOT_NULL(from->history_root)) {
|
---|
377 | to->history_root = history_tree_clone(from->history_root);
|
---|
378 | }
|
---|
379 | #endif
|
---|
380 | }
|
---|
381 |
|
---|
382 |
|
---|
383 | /** stack **/
|
---|
384 | #define INVALID_STACK_INDEX -1
|
---|
385 |
|
---|
386 | /* stack type */
|
---|
387 | /* used by normal-POP */
|
---|
388 | #define STK_ALT 0x0001
|
---|
389 | #define STK_LOOK_BEHIND_NOT 0x0002
|
---|
390 | #define STK_POS_NOT 0x0003
|
---|
391 | /* handled by normal-POP */
|
---|
392 | #define STK_MEM_START 0x0100
|
---|
393 | #define STK_MEM_END 0x8200
|
---|
394 | #define STK_REPEAT_INC 0x0300
|
---|
395 | #define STK_STATE_CHECK_MARK 0x1000
|
---|
396 | /* avoided by normal-POP */
|
---|
397 | #define STK_NULL_CHECK_START 0x3000
|
---|
398 | #define STK_NULL_CHECK_END 0x5000 /* for recursive call */
|
---|
399 | #define STK_MEM_END_MARK 0x8400
|
---|
400 | #define STK_POS 0x0500 /* used when POP-POS */
|
---|
401 | #define STK_STOP_BT 0x0600 /* mark for "(?>...)" */
|
---|
402 | #define STK_REPEAT 0x0700
|
---|
403 | #define STK_CALL_FRAME 0x0800
|
---|
404 | #define STK_RETURN 0x0900
|
---|
405 | #define STK_VOID 0x0a00 /* for fill a blank */
|
---|
406 | #define STK_ABSENT_POS 0x0b00 /* for absent */
|
---|
407 | #define STK_ABSENT 0x0c00 /* absent inner loop marker */
|
---|
408 |
|
---|
409 | /* stack type check mask */
|
---|
410 | #define STK_MASK_POP_USED 0x00ff
|
---|
411 | #define STK_MASK_TO_VOID_TARGET 0x10ff
|
---|
412 | #define STK_MASK_MEM_END_OR_MARK 0x8000 /* MEM_END or MEM_END_MARK */
|
---|
413 |
|
---|
414 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
415 | # define MATCH_ARG_INIT(msa, arg_option, arg_region, arg_start, arg_gpos) do {\
|
---|
416 | (msa).stack_p = (void* )0;\
|
---|
417 | (msa).options = (arg_option);\
|
---|
418 | (msa).region = (arg_region);\
|
---|
419 | (msa).start = (arg_start);\
|
---|
420 | (msa).gpos = (arg_gpos);\
|
---|
421 | (msa).best_len = ONIG_MISMATCH;\
|
---|
422 | } while(0)
|
---|
423 | #else
|
---|
424 | # define MATCH_ARG_INIT(msa, arg_option, arg_region, arg_start, arg_gpos) do {\
|
---|
425 | (msa).stack_p = (void* )0;\
|
---|
426 | (msa).options = (arg_option);\
|
---|
427 | (msa).region = (arg_region);\
|
---|
428 | (msa).start = (arg_start);\
|
---|
429 | (msa).gpos = (arg_gpos);\
|
---|
430 | } while(0)
|
---|
431 | #endif
|
---|
432 |
|
---|
433 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
434 |
|
---|
435 | # define STATE_CHECK_BUFF_MALLOC_THRESHOLD_SIZE 16
|
---|
436 |
|
---|
437 | # define STATE_CHECK_BUFF_INIT(msa, str_len, offset, state_num) do { \
|
---|
438 | if ((state_num) > 0 && str_len >= STATE_CHECK_STRING_THRESHOLD_LEN) {\
|
---|
439 | unsigned int size = (unsigned int )(((str_len) + 1) * (state_num) + 7) >> 3;\
|
---|
440 | offset = ((offset) * (state_num)) >> 3;\
|
---|
441 | if (size > 0 && offset < size && size < STATE_CHECK_BUFF_MAX_SIZE) {\
|
---|
442 | if (size >= STATE_CHECK_BUFF_MALLOC_THRESHOLD_SIZE) {\
|
---|
443 | (msa).state_check_buff = (void* )xmalloc(size);\
|
---|
444 | CHECK_NULL_RETURN_MEMERR((msa).state_check_buff);\
|
---|
445 | }\
|
---|
446 | else \
|
---|
447 | (msa).state_check_buff = (void* )xalloca(size);\
|
---|
448 | xmemset(((char* )((msa).state_check_buff)+(offset)), 0, \
|
---|
449 | (size_t )(size - (offset))); \
|
---|
450 | (msa).state_check_buff_size = size;\
|
---|
451 | }\
|
---|
452 | else {\
|
---|
453 | (msa).state_check_buff = (void* )0;\
|
---|
454 | (msa).state_check_buff_size = 0;\
|
---|
455 | }\
|
---|
456 | }\
|
---|
457 | else {\
|
---|
458 | (msa).state_check_buff = (void* )0;\
|
---|
459 | (msa).state_check_buff_size = 0;\
|
---|
460 | }\
|
---|
461 | } while(0)
|
---|
462 |
|
---|
463 | # define MATCH_ARG_FREE(msa) do {\
|
---|
464 | if ((msa).stack_p) xfree((msa).stack_p);\
|
---|
465 | if ((msa).state_check_buff_size >= STATE_CHECK_BUFF_MALLOC_THRESHOLD_SIZE) { \
|
---|
466 | if ((msa).state_check_buff) xfree((msa).state_check_buff);\
|
---|
467 | }\
|
---|
468 | } while(0)
|
---|
469 | #else /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
470 | # define MATCH_ARG_FREE(msa) if ((msa).stack_p) xfree((msa).stack_p)
|
---|
471 | #endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
472 |
|
---|
473 |
|
---|
474 |
|
---|
475 | #define MAX_PTR_NUM 100
|
---|
476 |
|
---|
477 | #define STACK_INIT(alloc_addr, heap_addr, ptr_num, stack_num) do {\
|
---|
478 | if (ptr_num > MAX_PTR_NUM) {\
|
---|
479 | alloc_addr = (char* )xmalloc(sizeof(OnigStackIndex) * (ptr_num));\
|
---|
480 | heap_addr = alloc_addr;\
|
---|
481 | if (msa->stack_p) {\
|
---|
482 | stk_alloc = (OnigStackType* )(msa->stack_p);\
|
---|
483 | stk_base = stk_alloc;\
|
---|
484 | stk = stk_base;\
|
---|
485 | stk_end = stk_base + msa->stack_n;\
|
---|
486 | } else {\
|
---|
487 | stk_alloc = (OnigStackType* )xalloca(sizeof(OnigStackType) * (stack_num));\
|
---|
488 | stk_base = stk_alloc;\
|
---|
489 | stk = stk_base;\
|
---|
490 | stk_end = stk_base + (stack_num);\
|
---|
491 | }\
|
---|
492 | } else if (msa->stack_p) {\
|
---|
493 | alloc_addr = (char* )xalloca(sizeof(OnigStackIndex) * (ptr_num));\
|
---|
494 | heap_addr = NULL;\
|
---|
495 | stk_alloc = (OnigStackType* )(msa->stack_p);\
|
---|
496 | stk_base = stk_alloc;\
|
---|
497 | stk = stk_base;\
|
---|
498 | stk_end = stk_base + msa->stack_n;\
|
---|
499 | }\
|
---|
500 | else {\
|
---|
501 | alloc_addr = (char* )xalloca(sizeof(OnigStackIndex) * (ptr_num)\
|
---|
502 | + sizeof(OnigStackType) * (stack_num));\
|
---|
503 | heap_addr = NULL;\
|
---|
504 | stk_alloc = (OnigStackType* )(alloc_addr + sizeof(OnigStackIndex) * (ptr_num));\
|
---|
505 | stk_base = stk_alloc;\
|
---|
506 | stk = stk_base;\
|
---|
507 | stk_end = stk_base + (stack_num);\
|
---|
508 | }\
|
---|
509 | } while(0)
|
---|
510 |
|
---|
511 | #define STACK_SAVE do{\
|
---|
512 | if (stk_base != stk_alloc) {\
|
---|
513 | msa->stack_p = stk_base;\
|
---|
514 | msa->stack_n = stk_end - stk_base; /* TODO: check overflow */\
|
---|
515 | };\
|
---|
516 | } while(0)
|
---|
517 |
|
---|
518 | static unsigned int MatchStackLimitSize = DEFAULT_MATCH_STACK_LIMIT_SIZE;
|
---|
519 |
|
---|
520 | extern unsigned int
|
---|
521 | onig_get_match_stack_limit_size(void)
|
---|
522 | {
|
---|
523 | return MatchStackLimitSize;
|
---|
524 | }
|
---|
525 |
|
---|
526 | extern int
|
---|
527 | onig_set_match_stack_limit_size(unsigned int size)
|
---|
528 | {
|
---|
529 | MatchStackLimitSize = size;
|
---|
530 | return 0;
|
---|
531 | }
|
---|
532 |
|
---|
533 | static int
|
---|
534 | stack_double(OnigStackType** arg_stk_base, OnigStackType** arg_stk_end,
|
---|
535 | OnigStackType** arg_stk, OnigStackType* stk_alloc, OnigMatchArg* msa)
|
---|
536 | {
|
---|
537 | size_t n;
|
---|
538 | OnigStackType *x, *stk_base, *stk_end, *stk;
|
---|
539 |
|
---|
540 | stk_base = *arg_stk_base;
|
---|
541 | stk_end = *arg_stk_end;
|
---|
542 | stk = *arg_stk;
|
---|
543 |
|
---|
544 | n = stk_end - stk_base;
|
---|
545 | if (stk_base == stk_alloc && IS_NULL(msa->stack_p)) {
|
---|
546 | x = (OnigStackType* )xmalloc(sizeof(OnigStackType) * n * 2);
|
---|
547 | if (IS_NULL(x)) {
|
---|
548 | STACK_SAVE;
|
---|
549 | return ONIGERR_MEMORY;
|
---|
550 | }
|
---|
551 | xmemcpy(x, stk_base, n * sizeof(OnigStackType));
|
---|
552 | n *= 2;
|
---|
553 | }
|
---|
554 | else {
|
---|
555 | unsigned int limit_size = MatchStackLimitSize;
|
---|
556 | n *= 2;
|
---|
557 | if (limit_size != 0 && n > limit_size) {
|
---|
558 | if ((unsigned int )(stk_end - stk_base) == limit_size)
|
---|
559 | return ONIGERR_MATCH_STACK_LIMIT_OVER;
|
---|
560 | else
|
---|
561 | n = limit_size;
|
---|
562 | }
|
---|
563 | x = (OnigStackType* )xrealloc(stk_base, sizeof(OnigStackType) * n);
|
---|
564 | if (IS_NULL(x)) {
|
---|
565 | STACK_SAVE;
|
---|
566 | return ONIGERR_MEMORY;
|
---|
567 | }
|
---|
568 | }
|
---|
569 | *arg_stk = x + (stk - stk_base);
|
---|
570 | *arg_stk_base = x;
|
---|
571 | *arg_stk_end = x + n;
|
---|
572 | return 0;
|
---|
573 | }
|
---|
574 |
|
---|
575 | #define STACK_ENSURE(n) do {\
|
---|
576 | if (stk_end - stk < (n)) {\
|
---|
577 | int r = stack_double(&stk_base, &stk_end, &stk, stk_alloc, msa);\
|
---|
578 | if (r != 0) {\
|
---|
579 | STACK_SAVE;\
|
---|
580 | if (xmalloc_base) xfree(xmalloc_base);\
|
---|
581 | return r;\
|
---|
582 | }\
|
---|
583 | }\
|
---|
584 | } while(0)
|
---|
585 |
|
---|
586 | #define STACK_AT(index) (stk_base + (index))
|
---|
587 | #define GET_STACK_INDEX(stk) ((stk) - stk_base)
|
---|
588 |
|
---|
589 | #define STACK_PUSH_TYPE(stack_type) do {\
|
---|
590 | STACK_ENSURE(1);\
|
---|
591 | stk->type = (stack_type);\
|
---|
592 | STACK_INC;\
|
---|
593 | } while(0)
|
---|
594 |
|
---|
595 | #define IS_TO_VOID_TARGET(stk) (((stk)->type & STK_MASK_TO_VOID_TARGET) != 0)
|
---|
596 |
|
---|
597 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
598 | # define STATE_CHECK_POS(s,snum) \
|
---|
599 | (((s) - str) * num_comb_exp_check + ((snum) - 1))
|
---|
600 | # define STATE_CHECK_VAL(v,snum) do {\
|
---|
601 | if (state_check_buff != NULL) {\
|
---|
602 | int x = STATE_CHECK_POS(s,snum);\
|
---|
603 | (v) = state_check_buff[x/8] & (1<<(x%8));\
|
---|
604 | }\
|
---|
605 | else (v) = 0;\
|
---|
606 | } while(0)
|
---|
607 |
|
---|
608 |
|
---|
609 | # define ELSE_IF_STATE_CHECK_MARK(stk) \
|
---|
610 | else if ((stk)->type == STK_STATE_CHECK_MARK) { \
|
---|
611 | int x = STATE_CHECK_POS(stk->u.state.pstr, stk->u.state.state_check);\
|
---|
612 | state_check_buff[x/8] |= (1<<(x%8)); \
|
---|
613 | }
|
---|
614 |
|
---|
615 | # define STACK_PUSH(stack_type,pat,s,sprev,keep) do {\
|
---|
616 | STACK_ENSURE(1);\
|
---|
617 | stk->type = (stack_type);\
|
---|
618 | stk->u.state.pcode = (pat);\
|
---|
619 | stk->u.state.pstr = (s);\
|
---|
620 | stk->u.state.pstr_prev = (sprev);\
|
---|
621 | stk->u.state.state_check = 0;\
|
---|
622 | stk->u.state.pkeep = (keep);\
|
---|
623 | STACK_INC;\
|
---|
624 | } while(0)
|
---|
625 |
|
---|
626 | # define STACK_PUSH_ENSURED(stack_type,pat) do {\
|
---|
627 | stk->type = (stack_type);\
|
---|
628 | stk->u.state.pcode = (pat);\
|
---|
629 | stk->u.state.state_check = 0;\
|
---|
630 | STACK_INC;\
|
---|
631 | } while(0)
|
---|
632 |
|
---|
633 | # define STACK_PUSH_ALT_WITH_STATE_CHECK(pat,s,sprev,snum,keep) do {\
|
---|
634 | STACK_ENSURE(1);\
|
---|
635 | stk->type = STK_ALT;\
|
---|
636 | stk->u.state.pcode = (pat);\
|
---|
637 | stk->u.state.pstr = (s);\
|
---|
638 | stk->u.state.pstr_prev = (sprev);\
|
---|
639 | stk->u.state.state_check = ((state_check_buff != NULL) ? (snum) : 0);\
|
---|
640 | stk->u.state.pkeep = (keep);\
|
---|
641 | STACK_INC;\
|
---|
642 | } while(0)
|
---|
643 |
|
---|
644 | # define STACK_PUSH_STATE_CHECK(s,snum) do {\
|
---|
645 | if (state_check_buff != NULL) {\
|
---|
646 | STACK_ENSURE(1);\
|
---|
647 | stk->type = STK_STATE_CHECK_MARK;\
|
---|
648 | stk->u.state.pstr = (s);\
|
---|
649 | stk->u.state.state_check = (snum);\
|
---|
650 | STACK_INC;\
|
---|
651 | }\
|
---|
652 | } while(0)
|
---|
653 |
|
---|
654 | #else /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
655 |
|
---|
656 | # define ELSE_IF_STATE_CHECK_MARK(stk)
|
---|
657 |
|
---|
658 | # define STACK_PUSH(stack_type,pat,s,sprev,keep) do {\
|
---|
659 | STACK_ENSURE(1);\
|
---|
660 | stk->type = (stack_type);\
|
---|
661 | stk->u.state.pcode = (pat);\
|
---|
662 | stk->u.state.pstr = (s);\
|
---|
663 | stk->u.state.pstr_prev = (sprev);\
|
---|
664 | stk->u.state.pkeep = (keep);\
|
---|
665 | STACK_INC;\
|
---|
666 | } while(0)
|
---|
667 |
|
---|
668 | # define STACK_PUSH_ENSURED(stack_type,pat) do {\
|
---|
669 | stk->type = (stack_type);\
|
---|
670 | stk->u.state.pcode = (pat);\
|
---|
671 | STACK_INC;\
|
---|
672 | } while(0)
|
---|
673 | #endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
674 |
|
---|
675 | #define STACK_PUSH_ALT(pat,s,sprev,keep) STACK_PUSH(STK_ALT,pat,s,sprev,keep)
|
---|
676 | #define STACK_PUSH_POS(s,sprev,keep) STACK_PUSH(STK_POS,NULL_UCHARP,s,sprev,keep)
|
---|
677 | #define STACK_PUSH_POS_NOT(pat,s,sprev,keep) STACK_PUSH(STK_POS_NOT,pat,s,sprev,keep)
|
---|
678 | #define STACK_PUSH_ABSENT STACK_PUSH_TYPE(STK_ABSENT)
|
---|
679 | #define STACK_PUSH_STOP_BT STACK_PUSH_TYPE(STK_STOP_BT)
|
---|
680 | #define STACK_PUSH_LOOK_BEHIND_NOT(pat,s,sprev,keep) \
|
---|
681 | STACK_PUSH(STK_LOOK_BEHIND_NOT,pat,s,sprev,keep)
|
---|
682 |
|
---|
683 | #define STACK_PUSH_REPEAT(id, pat) do {\
|
---|
684 | STACK_ENSURE(1);\
|
---|
685 | stk->type = STK_REPEAT;\
|
---|
686 | stk->u.repeat.num = (id);\
|
---|
687 | stk->u.repeat.pcode = (pat);\
|
---|
688 | stk->u.repeat.count = 0;\
|
---|
689 | STACK_INC;\
|
---|
690 | } while(0)
|
---|
691 |
|
---|
692 | #define STACK_PUSH_REPEAT_INC(sindex) do {\
|
---|
693 | STACK_ENSURE(1);\
|
---|
694 | stk->type = STK_REPEAT_INC;\
|
---|
695 | stk->u.repeat_inc.si = (sindex);\
|
---|
696 | STACK_INC;\
|
---|
697 | } while(0)
|
---|
698 |
|
---|
699 | #define STACK_PUSH_MEM_START(mnum, s) do {\
|
---|
700 | STACK_ENSURE(1);\
|
---|
701 | stk->type = STK_MEM_START;\
|
---|
702 | stk->u.mem.num = (mnum);\
|
---|
703 | stk->u.mem.pstr = (s);\
|
---|
704 | stk->u.mem.start = mem_start_stk[mnum];\
|
---|
705 | stk->u.mem.end = mem_end_stk[mnum];\
|
---|
706 | mem_start_stk[mnum] = GET_STACK_INDEX(stk);\
|
---|
707 | mem_end_stk[mnum] = INVALID_STACK_INDEX;\
|
---|
708 | STACK_INC;\
|
---|
709 | } while(0)
|
---|
710 |
|
---|
711 | #define STACK_PUSH_MEM_END(mnum, s) do {\
|
---|
712 | STACK_ENSURE(1);\
|
---|
713 | stk->type = STK_MEM_END;\
|
---|
714 | stk->u.mem.num = (mnum);\
|
---|
715 | stk->u.mem.pstr = (s);\
|
---|
716 | stk->u.mem.start = mem_start_stk[mnum];\
|
---|
717 | stk->u.mem.end = mem_end_stk[mnum];\
|
---|
718 | mem_end_stk[mnum] = GET_STACK_INDEX(stk);\
|
---|
719 | STACK_INC;\
|
---|
720 | } while(0)
|
---|
721 |
|
---|
722 | #define STACK_PUSH_MEM_END_MARK(mnum) do {\
|
---|
723 | STACK_ENSURE(1);\
|
---|
724 | stk->type = STK_MEM_END_MARK;\
|
---|
725 | stk->u.mem.num = (mnum);\
|
---|
726 | STACK_INC;\
|
---|
727 | } while(0)
|
---|
728 |
|
---|
729 | #define STACK_GET_MEM_START(mnum, k) do {\
|
---|
730 | int level = 0;\
|
---|
731 | k = stk;\
|
---|
732 | while (k > stk_base) {\
|
---|
733 | k--;\
|
---|
734 | if ((k->type & STK_MASK_MEM_END_OR_MARK) != 0 \
|
---|
735 | && k->u.mem.num == (mnum)) {\
|
---|
736 | level++;\
|
---|
737 | }\
|
---|
738 | else if (k->type == STK_MEM_START && k->u.mem.num == (mnum)) {\
|
---|
739 | if (level == 0) break;\
|
---|
740 | level--;\
|
---|
741 | }\
|
---|
742 | }\
|
---|
743 | } while(0)
|
---|
744 |
|
---|
745 | #define STACK_GET_MEM_RANGE(k, mnum, start, end) do {\
|
---|
746 | int level = 0;\
|
---|
747 | while (k < stk) {\
|
---|
748 | if (k->type == STK_MEM_START && k->u.mem.num == (mnum)) {\
|
---|
749 | if (level == 0) (start) = k->u.mem.pstr;\
|
---|
750 | level++;\
|
---|
751 | }\
|
---|
752 | else if (k->type == STK_MEM_END && k->u.mem.num == (mnum)) {\
|
---|
753 | level--;\
|
---|
754 | if (level == 0) {\
|
---|
755 | (end) = k->u.mem.pstr;\
|
---|
756 | break;\
|
---|
757 | }\
|
---|
758 | }\
|
---|
759 | k++;\
|
---|
760 | }\
|
---|
761 | } while(0)
|
---|
762 |
|
---|
763 | #define STACK_PUSH_NULL_CHECK_START(cnum, s) do {\
|
---|
764 | STACK_ENSURE(1);\
|
---|
765 | stk->type = STK_NULL_CHECK_START;\
|
---|
766 | stk->u.null_check.num = (cnum);\
|
---|
767 | stk->u.null_check.pstr = (s);\
|
---|
768 | STACK_INC;\
|
---|
769 | } while(0)
|
---|
770 |
|
---|
771 | #define STACK_PUSH_NULL_CHECK_END(cnum) do {\
|
---|
772 | STACK_ENSURE(1);\
|
---|
773 | stk->type = STK_NULL_CHECK_END;\
|
---|
774 | stk->u.null_check.num = (cnum);\
|
---|
775 | STACK_INC;\
|
---|
776 | } while(0)
|
---|
777 |
|
---|
778 | #define STACK_PUSH_CALL_FRAME(pat) do {\
|
---|
779 | STACK_ENSURE(1);\
|
---|
780 | stk->type = STK_CALL_FRAME;\
|
---|
781 | stk->u.call_frame.ret_addr = (pat);\
|
---|
782 | STACK_INC;\
|
---|
783 | } while(0)
|
---|
784 |
|
---|
785 | #define STACK_PUSH_RETURN do {\
|
---|
786 | STACK_ENSURE(1);\
|
---|
787 | stk->type = STK_RETURN;\
|
---|
788 | STACK_INC;\
|
---|
789 | } while(0)
|
---|
790 |
|
---|
791 | #define STACK_PUSH_ABSENT_POS(start, end) do {\
|
---|
792 | STACK_ENSURE(1);\
|
---|
793 | stk->type = STK_ABSENT_POS;\
|
---|
794 | stk->u.absent_pos.abs_pstr = (start);\
|
---|
795 | stk->u.absent_pos.end_pstr = (end);\
|
---|
796 | STACK_INC;\
|
---|
797 | } while(0)
|
---|
798 |
|
---|
799 |
|
---|
800 | #ifdef ONIG_DEBUG
|
---|
801 | # define STACK_BASE_CHECK(p, at) \
|
---|
802 | if ((p) < stk_base) {\
|
---|
803 | fprintf(stderr, "at %s\n", at);\
|
---|
804 | goto stack_error;\
|
---|
805 | }
|
---|
806 | #else
|
---|
807 | # define STACK_BASE_CHECK(p, at)
|
---|
808 | #endif
|
---|
809 |
|
---|
810 | #define STACK_POP_ONE do {\
|
---|
811 | stk--;\
|
---|
812 | STACK_BASE_CHECK(stk, "STACK_POP_ONE"); \
|
---|
813 | } while(0)
|
---|
814 |
|
---|
815 | #define STACK_POP do {\
|
---|
816 | switch (pop_level) {\
|
---|
817 | case STACK_POP_LEVEL_FREE:\
|
---|
818 | while (1) {\
|
---|
819 | stk--;\
|
---|
820 | STACK_BASE_CHECK(stk, "STACK_POP"); \
|
---|
821 | if ((stk->type & STK_MASK_POP_USED) != 0) break;\
|
---|
822 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
823 | }\
|
---|
824 | break;\
|
---|
825 | case STACK_POP_LEVEL_MEM_START:\
|
---|
826 | while (1) {\
|
---|
827 | stk--;\
|
---|
828 | STACK_BASE_CHECK(stk, "STACK_POP 2"); \
|
---|
829 | if ((stk->type & STK_MASK_POP_USED) != 0) break;\
|
---|
830 | else if (stk->type == STK_MEM_START) {\
|
---|
831 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
832 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
833 | }\
|
---|
834 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
835 | }\
|
---|
836 | break;\
|
---|
837 | default:\
|
---|
838 | while (1) {\
|
---|
839 | stk--;\
|
---|
840 | STACK_BASE_CHECK(stk, "STACK_POP 3"); \
|
---|
841 | if ((stk->type & STK_MASK_POP_USED) != 0) break;\
|
---|
842 | else if (stk->type == STK_MEM_START) {\
|
---|
843 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
844 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
845 | }\
|
---|
846 | else if (stk->type == STK_REPEAT_INC) {\
|
---|
847 | STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
|
---|
848 | }\
|
---|
849 | else if (stk->type == STK_MEM_END) {\
|
---|
850 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
851 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
852 | }\
|
---|
853 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
854 | }\
|
---|
855 | break;\
|
---|
856 | }\
|
---|
857 | } while(0)
|
---|
858 |
|
---|
859 | #define STACK_POP_TIL_POS_NOT do {\
|
---|
860 | while (1) {\
|
---|
861 | stk--;\
|
---|
862 | STACK_BASE_CHECK(stk, "STACK_POP_TIL_POS_NOT"); \
|
---|
863 | if (stk->type == STK_POS_NOT) break;\
|
---|
864 | else if (stk->type == STK_MEM_START) {\
|
---|
865 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
866 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
867 | }\
|
---|
868 | else if (stk->type == STK_REPEAT_INC) {\
|
---|
869 | STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
|
---|
870 | }\
|
---|
871 | else if (stk->type == STK_MEM_END) {\
|
---|
872 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
873 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
874 | }\
|
---|
875 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
876 | }\
|
---|
877 | } while(0)
|
---|
878 |
|
---|
879 | #define STACK_POP_TIL_LOOK_BEHIND_NOT do {\
|
---|
880 | while (1) {\
|
---|
881 | stk--;\
|
---|
882 | STACK_BASE_CHECK(stk, "STACK_POP_TIL_LOOK_BEHIND_NOT"); \
|
---|
883 | if (stk->type == STK_LOOK_BEHIND_NOT) break;\
|
---|
884 | else if (stk->type == STK_MEM_START) {\
|
---|
885 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
886 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
887 | }\
|
---|
888 | else if (stk->type == STK_REPEAT_INC) {\
|
---|
889 | STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
|
---|
890 | }\
|
---|
891 | else if (stk->type == STK_MEM_END) {\
|
---|
892 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
893 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
894 | }\
|
---|
895 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
896 | }\
|
---|
897 | } while(0)
|
---|
898 |
|
---|
899 | #define STACK_POP_TIL_ABSENT do {\
|
---|
900 | while (1) {\
|
---|
901 | stk--;\
|
---|
902 | STACK_BASE_CHECK(stk, "STACK_POP_TIL_ABSENT"); \
|
---|
903 | if (stk->type == STK_ABSENT) break;\
|
---|
904 | else if (stk->type == STK_MEM_START) {\
|
---|
905 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
906 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
907 | }\
|
---|
908 | else if (stk->type == STK_REPEAT_INC) {\
|
---|
909 | STACK_AT(stk->u.repeat_inc.si)->u.repeat.count--;\
|
---|
910 | }\
|
---|
911 | else if (stk->type == STK_MEM_END) {\
|
---|
912 | mem_start_stk[stk->u.mem.num] = stk->u.mem.start;\
|
---|
913 | mem_end_stk[stk->u.mem.num] = stk->u.mem.end;\
|
---|
914 | }\
|
---|
915 | ELSE_IF_STATE_CHECK_MARK(stk);\
|
---|
916 | }\
|
---|
917 | } while(0)
|
---|
918 |
|
---|
919 | #define STACK_POP_ABSENT_POS(start, end) do {\
|
---|
920 | stk--;\
|
---|
921 | STACK_BASE_CHECK(stk, "STACK_POP_ABSENT_POS"); \
|
---|
922 | (start) = stk->u.absent_pos.abs_pstr;\
|
---|
923 | (end) = stk->u.absent_pos.end_pstr;\
|
---|
924 | } while(0)
|
---|
925 |
|
---|
926 | #define STACK_POS_END(k) do {\
|
---|
927 | k = stk;\
|
---|
928 | while (1) {\
|
---|
929 | k--;\
|
---|
930 | STACK_BASE_CHECK(k, "STACK_POS_END"); \
|
---|
931 | if (IS_TO_VOID_TARGET(k)) {\
|
---|
932 | k->type = STK_VOID;\
|
---|
933 | }\
|
---|
934 | else if (k->type == STK_POS) {\
|
---|
935 | k->type = STK_VOID;\
|
---|
936 | break;\
|
---|
937 | }\
|
---|
938 | }\
|
---|
939 | } while(0)
|
---|
940 |
|
---|
941 | #define STACK_STOP_BT_END do {\
|
---|
942 | OnigStackType *k = stk;\
|
---|
943 | while (1) {\
|
---|
944 | k--;\
|
---|
945 | STACK_BASE_CHECK(k, "STACK_STOP_BT_END"); \
|
---|
946 | if (IS_TO_VOID_TARGET(k)) {\
|
---|
947 | k->type = STK_VOID;\
|
---|
948 | }\
|
---|
949 | else if (k->type == STK_STOP_BT) {\
|
---|
950 | k->type = STK_VOID;\
|
---|
951 | break;\
|
---|
952 | }\
|
---|
953 | }\
|
---|
954 | } while(0)
|
---|
955 |
|
---|
956 | #define STACK_NULL_CHECK(isnull,id,s) do {\
|
---|
957 | OnigStackType* k = stk;\
|
---|
958 | while (1) {\
|
---|
959 | k--;\
|
---|
960 | STACK_BASE_CHECK(k, "STACK_NULL_CHECK"); \
|
---|
961 | if (k->type == STK_NULL_CHECK_START) {\
|
---|
962 | if (k->u.null_check.num == (id)) {\
|
---|
963 | (isnull) = (k->u.null_check.pstr == (s));\
|
---|
964 | break;\
|
---|
965 | }\
|
---|
966 | }\
|
---|
967 | }\
|
---|
968 | } while(0)
|
---|
969 |
|
---|
970 | #define STACK_NULL_CHECK_REC(isnull,id,s) do {\
|
---|
971 | int level = 0;\
|
---|
972 | OnigStackType* k = stk;\
|
---|
973 | while (1) {\
|
---|
974 | k--;\
|
---|
975 | STACK_BASE_CHECK(k, "STACK_NULL_CHECK_REC"); \
|
---|
976 | if (k->type == STK_NULL_CHECK_START) {\
|
---|
977 | if (k->u.null_check.num == (id)) {\
|
---|
978 | if (level == 0) {\
|
---|
979 | (isnull) = (k->u.null_check.pstr == (s));\
|
---|
980 | break;\
|
---|
981 | }\
|
---|
982 | else level--;\
|
---|
983 | }\
|
---|
984 | }\
|
---|
985 | else if (k->type == STK_NULL_CHECK_END) {\
|
---|
986 | level++;\
|
---|
987 | }\
|
---|
988 | }\
|
---|
989 | } while(0)
|
---|
990 |
|
---|
991 | #define STACK_NULL_CHECK_MEMST(isnull,id,s,reg) do {\
|
---|
992 | OnigStackType* k = stk;\
|
---|
993 | while (1) {\
|
---|
994 | k--;\
|
---|
995 | STACK_BASE_CHECK(k, "STACK_NULL_CHECK_MEMST"); \
|
---|
996 | if (k->type == STK_NULL_CHECK_START) {\
|
---|
997 | if (k->u.null_check.num == (id)) {\
|
---|
998 | if (k->u.null_check.pstr != (s)) {\
|
---|
999 | (isnull) = 0;\
|
---|
1000 | break;\
|
---|
1001 | }\
|
---|
1002 | else {\
|
---|
1003 | UChar* endp;\
|
---|
1004 | (isnull) = 1;\
|
---|
1005 | while (k < stk) {\
|
---|
1006 | if (k->type == STK_MEM_START) {\
|
---|
1007 | if (k->u.mem.end == INVALID_STACK_INDEX) {\
|
---|
1008 | (isnull) = 0; break;\
|
---|
1009 | }\
|
---|
1010 | if (BIT_STATUS_AT(reg->bt_mem_end, k->u.mem.num))\
|
---|
1011 | endp = STACK_AT(k->u.mem.end)->u.mem.pstr;\
|
---|
1012 | else\
|
---|
1013 | endp = (UChar* )k->u.mem.end;\
|
---|
1014 | if (STACK_AT(k->u.mem.start)->u.mem.pstr != endp) {\
|
---|
1015 | (isnull) = 0; break;\
|
---|
1016 | }\
|
---|
1017 | else if (endp != s) {\
|
---|
1018 | (isnull) = -1; /* empty, but position changed */ \
|
---|
1019 | }\
|
---|
1020 | }\
|
---|
1021 | k++;\
|
---|
1022 | }\
|
---|
1023 | break;\
|
---|
1024 | }\
|
---|
1025 | }\
|
---|
1026 | }\
|
---|
1027 | }\
|
---|
1028 | } while(0)
|
---|
1029 |
|
---|
1030 | #define STACK_NULL_CHECK_MEMST_REC(isnull,id,s,reg) do {\
|
---|
1031 | int level = 0;\
|
---|
1032 | OnigStackType* k = stk;\
|
---|
1033 | while (1) {\
|
---|
1034 | k--;\
|
---|
1035 | STACK_BASE_CHECK(k, "STACK_NULL_CHECK_MEMST_REC"); \
|
---|
1036 | if (k->type == STK_NULL_CHECK_START) {\
|
---|
1037 | if (k->u.null_check.num == (id)) {\
|
---|
1038 | if (level == 0) {\
|
---|
1039 | if (k->u.null_check.pstr != (s)) {\
|
---|
1040 | (isnull) = 0;\
|
---|
1041 | break;\
|
---|
1042 | }\
|
---|
1043 | else {\
|
---|
1044 | UChar* endp;\
|
---|
1045 | (isnull) = 1;\
|
---|
1046 | while (k < stk) {\
|
---|
1047 | if (k->type == STK_MEM_START) {\
|
---|
1048 | if (k->u.mem.end == INVALID_STACK_INDEX) {\
|
---|
1049 | (isnull) = 0; break;\
|
---|
1050 | }\
|
---|
1051 | if (BIT_STATUS_AT(reg->bt_mem_end, k->u.mem.num))\
|
---|
1052 | endp = STACK_AT(k->u.mem.end)->u.mem.pstr;\
|
---|
1053 | else\
|
---|
1054 | endp = (UChar* )k->u.mem.end;\
|
---|
1055 | if (STACK_AT(k->u.mem.start)->u.mem.pstr != endp) {\
|
---|
1056 | (isnull) = 0; break;\
|
---|
1057 | }\
|
---|
1058 | else if (endp != s) {\
|
---|
1059 | (isnull) = -1; /* empty, but position changed */ \
|
---|
1060 | }\
|
---|
1061 | }\
|
---|
1062 | k++;\
|
---|
1063 | }\
|
---|
1064 | break;\
|
---|
1065 | }\
|
---|
1066 | }\
|
---|
1067 | else {\
|
---|
1068 | level--;\
|
---|
1069 | }\
|
---|
1070 | }\
|
---|
1071 | }\
|
---|
1072 | else if (k->type == STK_NULL_CHECK_END) {\
|
---|
1073 | if (k->u.null_check.num == (id)) level++;\
|
---|
1074 | }\
|
---|
1075 | }\
|
---|
1076 | } while(0)
|
---|
1077 |
|
---|
1078 | #define STACK_GET_REPEAT(id, k) do {\
|
---|
1079 | int level = 0;\
|
---|
1080 | k = stk;\
|
---|
1081 | while (1) {\
|
---|
1082 | k--;\
|
---|
1083 | STACK_BASE_CHECK(k, "STACK_GET_REPEAT"); \
|
---|
1084 | if (k->type == STK_REPEAT) {\
|
---|
1085 | if (level == 0) {\
|
---|
1086 | if (k->u.repeat.num == (id)) {\
|
---|
1087 | break;\
|
---|
1088 | }\
|
---|
1089 | }\
|
---|
1090 | }\
|
---|
1091 | else if (k->type == STK_CALL_FRAME) level--;\
|
---|
1092 | else if (k->type == STK_RETURN) level++;\
|
---|
1093 | }\
|
---|
1094 | } while(0)
|
---|
1095 |
|
---|
1096 | #define STACK_RETURN(addr) do {\
|
---|
1097 | int level = 0;\
|
---|
1098 | OnigStackType* k = stk;\
|
---|
1099 | while (1) {\
|
---|
1100 | k--;\
|
---|
1101 | STACK_BASE_CHECK(k, "STACK_RETURN"); \
|
---|
1102 | if (k->type == STK_CALL_FRAME) {\
|
---|
1103 | if (level == 0) {\
|
---|
1104 | (addr) = k->u.call_frame.ret_addr;\
|
---|
1105 | break;\
|
---|
1106 | }\
|
---|
1107 | else level--;\
|
---|
1108 | }\
|
---|
1109 | else if (k->type == STK_RETURN)\
|
---|
1110 | level++;\
|
---|
1111 | }\
|
---|
1112 | } while(0)
|
---|
1113 |
|
---|
1114 |
|
---|
1115 | #define STRING_CMP(s1,s2,len) do {\
|
---|
1116 | while (len-- > 0) {\
|
---|
1117 | if (*s1++ != *s2++) goto fail;\
|
---|
1118 | }\
|
---|
1119 | } while(0)
|
---|
1120 |
|
---|
1121 | #define STRING_CMP_IC(case_fold_flag,s1,ps2,len,text_end) do {\
|
---|
1122 | if (string_cmp_ic(encode, case_fold_flag, s1, ps2, len, text_end) == 0) \
|
---|
1123 | goto fail; \
|
---|
1124 | } while(0)
|
---|
1125 |
|
---|
1126 | static int string_cmp_ic(OnigEncoding enc, int case_fold_flag,
|
---|
1127 | UChar* s1, UChar** ps2, OnigDistance mblen, const UChar* text_end)
|
---|
1128 | {
|
---|
1129 | UChar buf1[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
---|
1130 | UChar buf2[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
---|
1131 | UChar *p1, *p2, *end1, *s2;
|
---|
1132 | int len1, len2;
|
---|
1133 |
|
---|
1134 | s2 = *ps2;
|
---|
1135 | end1 = s1 + mblen;
|
---|
1136 | while (s1 < end1) {
|
---|
1137 | len1 = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &s1, text_end, buf1);
|
---|
1138 | len2 = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &s2, text_end, buf2);
|
---|
1139 | if (len1 != len2) return 0;
|
---|
1140 | p1 = buf1;
|
---|
1141 | p2 = buf2;
|
---|
1142 | while (len1-- > 0) {
|
---|
1143 | if (*p1 != *p2) return 0;
|
---|
1144 | p1++;
|
---|
1145 | p2++;
|
---|
1146 | }
|
---|
1147 | }
|
---|
1148 |
|
---|
1149 | *ps2 = s2;
|
---|
1150 | return 1;
|
---|
1151 | }
|
---|
1152 |
|
---|
1153 | #define STRING_CMP_VALUE(s1,s2,len,is_fail) do {\
|
---|
1154 | is_fail = 0;\
|
---|
1155 | while (len-- > 0) {\
|
---|
1156 | if (*s1++ != *s2++) {\
|
---|
1157 | is_fail = 1; break;\
|
---|
1158 | }\
|
---|
1159 | }\
|
---|
1160 | } while(0)
|
---|
1161 |
|
---|
1162 | #define STRING_CMP_VALUE_IC(case_fold_flag,s1,ps2,len,text_end,is_fail) do {\
|
---|
1163 | if (string_cmp_ic(encode, case_fold_flag, s1, ps2, len, text_end) == 0) \
|
---|
1164 | is_fail = 1; \
|
---|
1165 | else \
|
---|
1166 | is_fail = 0; \
|
---|
1167 | } while(0)
|
---|
1168 |
|
---|
1169 |
|
---|
1170 | #define IS_EMPTY_STR (str == end)
|
---|
1171 | #define ON_STR_BEGIN(s) ((s) == str)
|
---|
1172 | #define ON_STR_END(s) ((s) == end)
|
---|
1173 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
1174 | # define DATA_ENSURE_CHECK1 (s < right_range)
|
---|
1175 | # define DATA_ENSURE_CHECK(n) (s + (n) <= right_range)
|
---|
1176 | # define DATA_ENSURE(n) if (s + (n) > right_range) goto fail
|
---|
1177 | # define ABSENT_END_POS right_range
|
---|
1178 | #else
|
---|
1179 | # define DATA_ENSURE_CHECK1 (s < end)
|
---|
1180 | # define DATA_ENSURE_CHECK(n) (s + (n) <= end)
|
---|
1181 | # define DATA_ENSURE(n) if (s + (n) > end) goto fail
|
---|
1182 | # define ABSENT_END_POS end
|
---|
1183 | #endif /* USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE */
|
---|
1184 |
|
---|
1185 |
|
---|
1186 | #ifdef USE_CAPTURE_HISTORY
|
---|
1187 | static int
|
---|
1188 | make_capture_history_tree(OnigCaptureTreeNode* node, OnigStackType** kp,
|
---|
1189 | OnigStackType* stk_top, UChar* str, regex_t* reg)
|
---|
1190 | {
|
---|
1191 | int n, r;
|
---|
1192 | OnigCaptureTreeNode* child;
|
---|
1193 | OnigStackType* k = *kp;
|
---|
1194 |
|
---|
1195 | while (k < stk_top) {
|
---|
1196 | if (k->type == STK_MEM_START) {
|
---|
1197 | n = k->u.mem.num;
|
---|
1198 | if (n <= ONIG_MAX_CAPTURE_HISTORY_GROUP &&
|
---|
1199 | BIT_STATUS_AT(reg->capture_history, n) != 0) {
|
---|
1200 | child = history_node_new();
|
---|
1201 | CHECK_NULL_RETURN_MEMERR(child);
|
---|
1202 | child->group = n;
|
---|
1203 | child->beg = k->u.mem.pstr - str;
|
---|
1204 | r = history_tree_add_child(node, child);
|
---|
1205 | if (r != 0) {
|
---|
1206 | history_tree_free(child);
|
---|
1207 | return r;
|
---|
1208 | }
|
---|
1209 | *kp = (k + 1);
|
---|
1210 | r = make_capture_history_tree(child, kp, stk_top, str, reg);
|
---|
1211 | if (r != 0) return r;
|
---|
1212 |
|
---|
1213 | k = *kp;
|
---|
1214 | child->end = k->u.mem.pstr - str;
|
---|
1215 | }
|
---|
1216 | }
|
---|
1217 | else if (k->type == STK_MEM_END) {
|
---|
1218 | if (k->u.mem.num == node->group) {
|
---|
1219 | node->end = k->u.mem.pstr - str;
|
---|
1220 | *kp = k;
|
---|
1221 | return 0;
|
---|
1222 | }
|
---|
1223 | }
|
---|
1224 | k++;
|
---|
1225 | }
|
---|
1226 |
|
---|
1227 | return 1; /* 1: root node ending. */
|
---|
1228 | }
|
---|
1229 | #endif /* USE_CAPTURE_HISTORY */
|
---|
1230 |
|
---|
1231 | #ifdef USE_BACKREF_WITH_LEVEL
|
---|
1232 | static int mem_is_in_memp(int mem, int num, UChar* memp)
|
---|
1233 | {
|
---|
1234 | int i;
|
---|
1235 | MemNumType m;
|
---|
1236 |
|
---|
1237 | for (i = 0; i < num; i++) {
|
---|
1238 | GET_MEMNUM_INC(m, memp);
|
---|
1239 | if (mem == (int )m) return 1;
|
---|
1240 | }
|
---|
1241 | return 0;
|
---|
1242 | }
|
---|
1243 |
|
---|
1244 | static int backref_match_at_nested_level(regex_t* reg,
|
---|
1245 | OnigStackType* top, OnigStackType* stk_base,
|
---|
1246 | int ignore_case, int case_fold_flag,
|
---|
1247 | int nest, int mem_num, UChar* memp, UChar** s, const UChar* send)
|
---|
1248 | {
|
---|
1249 | UChar *ss, *p, *pstart, *pend = NULL_UCHARP;
|
---|
1250 | int level;
|
---|
1251 | OnigStackType* k;
|
---|
1252 |
|
---|
1253 | level = 0;
|
---|
1254 | k = top;
|
---|
1255 | k--;
|
---|
1256 | while (k >= stk_base) {
|
---|
1257 | if (k->type == STK_CALL_FRAME) {
|
---|
1258 | level--;
|
---|
1259 | }
|
---|
1260 | else if (k->type == STK_RETURN) {
|
---|
1261 | level++;
|
---|
1262 | }
|
---|
1263 | else if (level == nest) {
|
---|
1264 | if (k->type == STK_MEM_START) {
|
---|
1265 | if (mem_is_in_memp(k->u.mem.num, mem_num, memp)) {
|
---|
1266 | pstart = k->u.mem.pstr;
|
---|
1267 | if (pend != NULL_UCHARP) {
|
---|
1268 | if (pend - pstart > send - *s) return 0; /* or goto next_mem; */
|
---|
1269 | p = pstart;
|
---|
1270 | ss = *s;
|
---|
1271 |
|
---|
1272 | if (ignore_case != 0) {
|
---|
1273 | if (string_cmp_ic(reg->enc, case_fold_flag,
|
---|
1274 | pstart, &ss, pend - pstart, send) == 0)
|
---|
1275 | return 0; /* or goto next_mem; */
|
---|
1276 | }
|
---|
1277 | else {
|
---|
1278 | while (p < pend) {
|
---|
1279 | if (*p++ != *ss++) return 0; /* or goto next_mem; */
|
---|
1280 | }
|
---|
1281 | }
|
---|
1282 |
|
---|
1283 | *s = ss;
|
---|
1284 | return 1;
|
---|
1285 | }
|
---|
1286 | }
|
---|
1287 | }
|
---|
1288 | else if (k->type == STK_MEM_END) {
|
---|
1289 | if (mem_is_in_memp(k->u.mem.num, mem_num, memp)) {
|
---|
1290 | pend = k->u.mem.pstr;
|
---|
1291 | }
|
---|
1292 | }
|
---|
1293 | }
|
---|
1294 | k--;
|
---|
1295 | }
|
---|
1296 |
|
---|
1297 | return 0;
|
---|
1298 | }
|
---|
1299 | #endif /* USE_BACKREF_WITH_LEVEL */
|
---|
1300 |
|
---|
1301 |
|
---|
1302 | #ifdef ONIG_DEBUG_STATISTICS
|
---|
1303 |
|
---|
1304 | # ifdef _WIN32
|
---|
1305 | # include <windows.h>
|
---|
1306 | static LARGE_INTEGER ts, te, freq;
|
---|
1307 | # define GETTIME(t) QueryPerformanceCounter(&(t))
|
---|
1308 | # define TIMEDIFF(te,ts) (unsigned long )(((te).QuadPart - (ts).QuadPart) \
|
---|
1309 | * 1000000 / freq.QuadPart)
|
---|
1310 | # else /* _WIN32 */
|
---|
1311 |
|
---|
1312 | # define USE_TIMEOFDAY
|
---|
1313 |
|
---|
1314 | # ifdef USE_TIMEOFDAY
|
---|
1315 | # ifdef HAVE_SYS_TIME_H
|
---|
1316 | # include <sys/time.h>
|
---|
1317 | # endif
|
---|
1318 | # ifdef HAVE_UNISTD_H
|
---|
1319 | # include <unistd.h>
|
---|
1320 | # endif
|
---|
1321 | static struct timeval ts, te;
|
---|
1322 | # define GETTIME(t) gettimeofday(&(t), (struct timezone* )0)
|
---|
1323 | # define TIMEDIFF(te,ts) (((te).tv_usec - (ts).tv_usec) + \
|
---|
1324 | (((te).tv_sec - (ts).tv_sec)*1000000))
|
---|
1325 | # else /* USE_TIMEOFDAY */
|
---|
1326 | # ifdef HAVE_SYS_TIMES_H
|
---|
1327 | # include <sys/times.h>
|
---|
1328 | # endif
|
---|
1329 | static struct tms ts, te;
|
---|
1330 | # define GETTIME(t) times(&(t))
|
---|
1331 | # define TIMEDIFF(te,ts) ((te).tms_utime - (ts).tms_utime)
|
---|
1332 | # endif /* USE_TIMEOFDAY */
|
---|
1333 |
|
---|
1334 | # endif /* _WIN32 */
|
---|
1335 |
|
---|
1336 | static int OpCounter[256];
|
---|
1337 | static int OpPrevCounter[256];
|
---|
1338 | static unsigned long OpTime[256];
|
---|
1339 | static int OpCurr = OP_FINISH;
|
---|
1340 | static int OpPrevTarget = OP_FAIL;
|
---|
1341 | static int MaxStackDepth = 0;
|
---|
1342 |
|
---|
1343 | # define MOP_IN(opcode) do {\
|
---|
1344 | if (opcode == OpPrevTarget) OpPrevCounter[OpCurr]++;\
|
---|
1345 | OpCurr = opcode;\
|
---|
1346 | OpCounter[opcode]++;\
|
---|
1347 | GETTIME(ts);\
|
---|
1348 | } while(0)
|
---|
1349 |
|
---|
1350 | # define MOP_OUT do {\
|
---|
1351 | GETTIME(te);\
|
---|
1352 | OpTime[OpCurr] += TIMEDIFF(te, ts);\
|
---|
1353 | } while(0)
|
---|
1354 |
|
---|
1355 | extern void
|
---|
1356 | onig_statistics_init(void)
|
---|
1357 | {
|
---|
1358 | int i;
|
---|
1359 | for (i = 0; i < 256; i++) {
|
---|
1360 | OpCounter[i] = OpPrevCounter[i] = 0; OpTime[i] = 0;
|
---|
1361 | }
|
---|
1362 | MaxStackDepth = 0;
|
---|
1363 | # ifdef _WIN32
|
---|
1364 | QueryPerformanceFrequency(&freq);
|
---|
1365 | # endif
|
---|
1366 | }
|
---|
1367 |
|
---|
1368 | extern void
|
---|
1369 | onig_print_statistics(FILE* f)
|
---|
1370 | {
|
---|
1371 | int i;
|
---|
1372 | fprintf(f, " count prev time\n");
|
---|
1373 | for (i = 0; OnigOpInfo[i].opcode >= 0; i++) {
|
---|
1374 | fprintf(f, "%8d: %8d: %10lu: %s\n",
|
---|
1375 | OpCounter[i], OpPrevCounter[i], OpTime[i], OnigOpInfo[i].name);
|
---|
1376 | }
|
---|
1377 | fprintf(f, "\nmax stack depth: %d\n", MaxStackDepth);
|
---|
1378 | }
|
---|
1379 |
|
---|
1380 | # define STACK_INC do {\
|
---|
1381 | stk++;\
|
---|
1382 | if (stk - stk_base > MaxStackDepth) \
|
---|
1383 | MaxStackDepth = stk - stk_base;\
|
---|
1384 | } while(0)
|
---|
1385 |
|
---|
1386 | #else /* ONIG_DEBUG_STATISTICS */
|
---|
1387 | # define STACK_INC stk++
|
---|
1388 |
|
---|
1389 | # define MOP_IN(opcode)
|
---|
1390 | # define MOP_OUT
|
---|
1391 | #endif /* ONIG_DEBUG_STATISTICS */
|
---|
1392 |
|
---|
1393 |
|
---|
1394 | #ifdef ONIG_DEBUG_MATCH
|
---|
1395 | static char *
|
---|
1396 | stack_type_str(int stack_type)
|
---|
1397 | {
|
---|
1398 | switch (stack_type) {
|
---|
1399 | case STK_ALT: return "Alt ";
|
---|
1400 | case STK_LOOK_BEHIND_NOT: return "LBNot ";
|
---|
1401 | case STK_POS_NOT: return "PosNot";
|
---|
1402 | case STK_MEM_START: return "MemS ";
|
---|
1403 | case STK_MEM_END: return "MemE ";
|
---|
1404 | case STK_REPEAT_INC: return "RepInc";
|
---|
1405 | case STK_STATE_CHECK_MARK: return "StChMk";
|
---|
1406 | case STK_NULL_CHECK_START: return "NulChS";
|
---|
1407 | case STK_NULL_CHECK_END: return "NulChE";
|
---|
1408 | case STK_MEM_END_MARK: return "MemEMk";
|
---|
1409 | case STK_POS: return "Pos ";
|
---|
1410 | case STK_STOP_BT: return "StopBt";
|
---|
1411 | case STK_REPEAT: return "Rep ";
|
---|
1412 | case STK_CALL_FRAME: return "Call ";
|
---|
1413 | case STK_RETURN: return "Ret ";
|
---|
1414 | case STK_VOID: return "Void ";
|
---|
1415 | case STK_ABSENT_POS: return "AbsPos";
|
---|
1416 | case STK_ABSENT: return "Absent";
|
---|
1417 | default: return " ";
|
---|
1418 | }
|
---|
1419 | }
|
---|
1420 | #endif
|
---|
1421 |
|
---|
1422 | /* match data(str - end) from position (sstart). */
|
---|
1423 | /* if sstart == str then set sprev to NULL. */
|
---|
1424 | static OnigPosition
|
---|
1425 | match_at(regex_t* reg, const UChar* str, const UChar* end,
|
---|
1426 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
1427 | const UChar* right_range,
|
---|
1428 | #endif
|
---|
1429 | const UChar* sstart, UChar* sprev, OnigMatchArg* msa)
|
---|
1430 | {
|
---|
1431 | static const UChar FinishCode[] = { OP_FINISH };
|
---|
1432 |
|
---|
1433 | int i, num_mem, pop_level;
|
---|
1434 | ptrdiff_t n, best_len;
|
---|
1435 | LengthType tlen, tlen2;
|
---|
1436 | MemNumType mem;
|
---|
1437 | RelAddrType addr;
|
---|
1438 | OnigOptionType option = reg->options;
|
---|
1439 | OnigEncoding encode = reg->enc;
|
---|
1440 | OnigCaseFoldType case_fold_flag = reg->case_fold_flag;
|
---|
1441 | UChar *s, *q, *sbegin;
|
---|
1442 | UChar *p = reg->p;
|
---|
1443 | UChar *pkeep;
|
---|
1444 | char *alloca_base;
|
---|
1445 | char *xmalloc_base = NULL;
|
---|
1446 | OnigStackType *stk_alloc, *stk_base, *stk, *stk_end;
|
---|
1447 | OnigStackType *stkp; /* used as any purpose. */
|
---|
1448 | OnigStackIndex si;
|
---|
1449 | OnigStackIndex *repeat_stk;
|
---|
1450 | OnigStackIndex *mem_start_stk, *mem_end_stk;
|
---|
1451 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
1452 | int scv;
|
---|
1453 | unsigned char* state_check_buff = msa->state_check_buff;
|
---|
1454 | int num_comb_exp_check = reg->num_comb_exp_check;
|
---|
1455 | #endif
|
---|
1456 |
|
---|
1457 | #if USE_TOKEN_THREADED_VM
|
---|
1458 | # define OP_OFFSET 1
|
---|
1459 | # define VM_LOOP JUMP;
|
---|
1460 | # define VM_LOOP_END
|
---|
1461 | # define CASE(x) L_##x: sbegin = s; OPCODE_EXEC_HOOK;
|
---|
1462 | # define DEFAULT L_DEFAULT:
|
---|
1463 | # define NEXT sprev = sbegin; JUMP
|
---|
1464 | # define JUMP goto *oplabels[*p++]
|
---|
1465 |
|
---|
1466 | static const void *oplabels[] = {
|
---|
1467 | &&L_OP_FINISH, /* matching process terminator (no more alternative) */
|
---|
1468 | &&L_OP_END, /* pattern code terminator (success end) */
|
---|
1469 |
|
---|
1470 | &&L_OP_EXACT1, /* single byte, N = 1 */
|
---|
1471 | &&L_OP_EXACT2, /* single byte, N = 2 */
|
---|
1472 | &&L_OP_EXACT3, /* single byte, N = 3 */
|
---|
1473 | &&L_OP_EXACT4, /* single byte, N = 4 */
|
---|
1474 | &&L_OP_EXACT5, /* single byte, N = 5 */
|
---|
1475 | &&L_OP_EXACTN, /* single byte */
|
---|
1476 | &&L_OP_EXACTMB2N1, /* mb-length = 2 N = 1 */
|
---|
1477 | &&L_OP_EXACTMB2N2, /* mb-length = 2 N = 2 */
|
---|
1478 | &&L_OP_EXACTMB2N3, /* mb-length = 2 N = 3 */
|
---|
1479 | &&L_OP_EXACTMB2N, /* mb-length = 2 */
|
---|
1480 | &&L_OP_EXACTMB3N, /* mb-length = 3 */
|
---|
1481 | &&L_OP_EXACTMBN, /* other length */
|
---|
1482 |
|
---|
1483 | &&L_OP_EXACT1_IC, /* single byte, N = 1, ignore case */
|
---|
1484 | &&L_OP_EXACTN_IC, /* single byte, ignore case */
|
---|
1485 |
|
---|
1486 | &&L_OP_CCLASS,
|
---|
1487 | &&L_OP_CCLASS_MB,
|
---|
1488 | &&L_OP_CCLASS_MIX,
|
---|
1489 | &&L_OP_CCLASS_NOT,
|
---|
1490 | &&L_OP_CCLASS_MB_NOT,
|
---|
1491 | &&L_OP_CCLASS_MIX_NOT,
|
---|
1492 |
|
---|
1493 | &&L_OP_ANYCHAR, /* "." */
|
---|
1494 | &&L_OP_ANYCHAR_ML, /* "." multi-line */
|
---|
1495 | &&L_OP_ANYCHAR_STAR, /* ".*" */
|
---|
1496 | &&L_OP_ANYCHAR_ML_STAR, /* ".*" multi-line */
|
---|
1497 | &&L_OP_ANYCHAR_STAR_PEEK_NEXT,
|
---|
1498 | &&L_OP_ANYCHAR_ML_STAR_PEEK_NEXT,
|
---|
1499 |
|
---|
1500 | &&L_OP_WORD,
|
---|
1501 | &&L_OP_NOT_WORD,
|
---|
1502 | &&L_OP_WORD_BOUND,
|
---|
1503 | &&L_OP_NOT_WORD_BOUND,
|
---|
1504 | # ifdef USE_WORD_BEGIN_END
|
---|
1505 | &&L_OP_WORD_BEGIN,
|
---|
1506 | &&L_OP_WORD_END,
|
---|
1507 | # else
|
---|
1508 | &&L_DEFAULT,
|
---|
1509 | &&L_DEFAULT,
|
---|
1510 | # endif
|
---|
1511 | &&L_OP_ASCII_WORD,
|
---|
1512 | &&L_OP_NOT_ASCII_WORD,
|
---|
1513 | &&L_OP_ASCII_WORD_BOUND,
|
---|
1514 | &&L_OP_NOT_ASCII_WORD_BOUND,
|
---|
1515 | # ifdef USE_WORD_BEGIN_END
|
---|
1516 | &&L_OP_ASCII_WORD_BEGIN,
|
---|
1517 | &&L_OP_ASCII_WORD_END,
|
---|
1518 | # else
|
---|
1519 | &&L_DEFAULT,
|
---|
1520 | &&L_DEFAULT,
|
---|
1521 | # endif
|
---|
1522 |
|
---|
1523 | &&L_OP_BEGIN_BUF,
|
---|
1524 | &&L_OP_END_BUF,
|
---|
1525 | &&L_OP_BEGIN_LINE,
|
---|
1526 | &&L_OP_END_LINE,
|
---|
1527 | &&L_OP_SEMI_END_BUF,
|
---|
1528 | &&L_OP_BEGIN_POSITION,
|
---|
1529 |
|
---|
1530 | &&L_OP_BACKREF1,
|
---|
1531 | &&L_OP_BACKREF2,
|
---|
1532 | &&L_OP_BACKREFN,
|
---|
1533 | &&L_OP_BACKREFN_IC,
|
---|
1534 | &&L_OP_BACKREF_MULTI,
|
---|
1535 | &&L_OP_BACKREF_MULTI_IC,
|
---|
1536 | # ifdef USE_BACKREF_WITH_LEVEL
|
---|
1537 | &&L_OP_BACKREF_WITH_LEVEL, /* \k<xxx+n>, \k<xxx-n> */
|
---|
1538 | # else
|
---|
1539 | &&L_DEFAULT,
|
---|
1540 | # endif
|
---|
1541 | &&L_OP_MEMORY_START,
|
---|
1542 | &&L_OP_MEMORY_START_PUSH, /* push back-tracker to stack */
|
---|
1543 | &&L_OP_MEMORY_END_PUSH, /* push back-tracker to stack */
|
---|
1544 | # ifdef USE_SUBEXP_CALL
|
---|
1545 | &&L_OP_MEMORY_END_PUSH_REC, /* push back-tracker to stack */
|
---|
1546 | # else
|
---|
1547 | &&L_DEFAULT,
|
---|
1548 | # endif
|
---|
1549 | &&L_OP_MEMORY_END,
|
---|
1550 | # ifdef USE_SUBEXP_CALL
|
---|
1551 | &&L_OP_MEMORY_END_REC, /* push marker to stack */
|
---|
1552 | # else
|
---|
1553 | &&L_DEFAULT,
|
---|
1554 | # endif
|
---|
1555 |
|
---|
1556 | &&L_OP_KEEP,
|
---|
1557 |
|
---|
1558 | &&L_OP_FAIL, /* pop stack and move */
|
---|
1559 | &&L_OP_JUMP,
|
---|
1560 | &&L_OP_PUSH,
|
---|
1561 | &&L_OP_POP,
|
---|
1562 | # ifdef USE_OP_PUSH_OR_JUMP_EXACT
|
---|
1563 | &&L_OP_PUSH_OR_JUMP_EXACT1, /* if match exact then push, else jump. */
|
---|
1564 | # else
|
---|
1565 | &&L_DEFAULT,
|
---|
1566 | # endif
|
---|
1567 | &&L_OP_PUSH_IF_PEEK_NEXT, /* if match exact then push, else none. */
|
---|
1568 | &&L_OP_REPEAT, /* {n,m} */
|
---|
1569 | &&L_OP_REPEAT_NG, /* {n,m}? (non greedy) */
|
---|
1570 | &&L_OP_REPEAT_INC,
|
---|
1571 | &&L_OP_REPEAT_INC_NG, /* non greedy */
|
---|
1572 | &&L_OP_REPEAT_INC_SG, /* search and get in stack */
|
---|
1573 | &&L_OP_REPEAT_INC_NG_SG, /* search and get in stack (non greedy) */
|
---|
1574 | &&L_OP_NULL_CHECK_START, /* null loop checker start */
|
---|
1575 | &&L_OP_NULL_CHECK_END, /* null loop checker end */
|
---|
1576 | # ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT
|
---|
1577 | &&L_OP_NULL_CHECK_END_MEMST, /* null loop checker end (with capture status) */
|
---|
1578 | # else
|
---|
1579 | &&L_DEFAULT,
|
---|
1580 | # endif
|
---|
1581 | # ifdef USE_SUBEXP_CALL
|
---|
1582 | &&L_OP_NULL_CHECK_END_MEMST_PUSH, /* with capture status and push check-end */
|
---|
1583 | # else
|
---|
1584 | &&L_DEFAULT,
|
---|
1585 | # endif
|
---|
1586 |
|
---|
1587 | &&L_OP_PUSH_POS, /* (?=...) start */
|
---|
1588 | &&L_OP_POP_POS, /* (?=...) end */
|
---|
1589 | &&L_OP_PUSH_POS_NOT, /* (?!...) start */
|
---|
1590 | &&L_OP_FAIL_POS, /* (?!...) end */
|
---|
1591 | &&L_OP_PUSH_STOP_BT, /* (?>...) start */
|
---|
1592 | &&L_OP_POP_STOP_BT, /* (?>...) end */
|
---|
1593 | &&L_OP_LOOK_BEHIND, /* (?<=...) start (no needs end opcode) */
|
---|
1594 | &&L_OP_PUSH_LOOK_BEHIND_NOT, /* (?<!...) start */
|
---|
1595 | &&L_OP_FAIL_LOOK_BEHIND_NOT, /* (?<!...) end */
|
---|
1596 | &&L_OP_PUSH_ABSENT_POS, /* (?~...) start */
|
---|
1597 | &&L_OP_ABSENT, /* (?~...) start of inner loop */
|
---|
1598 | &&L_OP_ABSENT_END, /* (?~...) end */
|
---|
1599 |
|
---|
1600 | # ifdef USE_SUBEXP_CALL
|
---|
1601 | &&L_OP_CALL, /* \g<name> */
|
---|
1602 | &&L_OP_RETURN,
|
---|
1603 | # else
|
---|
1604 | &&L_DEFAULT,
|
---|
1605 | &&L_DEFAULT,
|
---|
1606 | # endif
|
---|
1607 | &&L_OP_CONDITION,
|
---|
1608 |
|
---|
1609 | # ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
1610 | &&L_OP_STATE_CHECK_PUSH, /* combination explosion check and push */
|
---|
1611 | &&L_OP_STATE_CHECK_PUSH_OR_JUMP, /* check ok -> push, else jump */
|
---|
1612 | &&L_OP_STATE_CHECK, /* check only */
|
---|
1613 | # else
|
---|
1614 | &&L_DEFAULT,
|
---|
1615 | &&L_DEFAULT,
|
---|
1616 | &&L_DEFAULT,
|
---|
1617 | # endif
|
---|
1618 | # ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
1619 | &&L_OP_STATE_CHECK_ANYCHAR_STAR,
|
---|
1620 | &&L_OP_STATE_CHECK_ANYCHAR_ML_STAR,
|
---|
1621 | # else
|
---|
1622 | &&L_DEFAULT,
|
---|
1623 | &&L_DEFAULT,
|
---|
1624 | # endif
|
---|
1625 | /* no need: IS_DYNAMIC_OPTION() == 0 */
|
---|
1626 | # if 0 /* no need: IS_DYNAMIC_OPTION() == 0 */
|
---|
1627 | &&L_OP_SET_OPTION_PUSH, /* set option and push recover option */
|
---|
1628 | &&L_OP_SET_OPTION /* set option */
|
---|
1629 | # else
|
---|
1630 | &&L_DEFAULT,
|
---|
1631 | &&L_DEFAULT
|
---|
1632 | # endif
|
---|
1633 | };
|
---|
1634 | #else /* USE_TOKEN_THREADED_VM */
|
---|
1635 |
|
---|
1636 | # define OP_OFFSET 0
|
---|
1637 | # define VM_LOOP \
|
---|
1638 | while (1) { \
|
---|
1639 | OPCODE_EXEC_HOOK; \
|
---|
1640 | sbegin = s; \
|
---|
1641 | switch (*p++) {
|
---|
1642 | # define VM_LOOP_END } sprev = sbegin; }
|
---|
1643 | # define CASE(x) case x:
|
---|
1644 | # define DEFAULT default:
|
---|
1645 | # define NEXT break
|
---|
1646 | # define JUMP continue; break
|
---|
1647 | #endif /* USE_TOKEN_THREADED_VM */
|
---|
1648 |
|
---|
1649 |
|
---|
1650 | #ifdef USE_SUBEXP_CALL
|
---|
1651 | /* Stack #0 is used to store the pattern itself and used for (?R), \g<0>,
|
---|
1652 | etc. Additional space is required. */
|
---|
1653 | # define ADD_NUMMEM 1
|
---|
1654 | #else
|
---|
1655 | /* Stack #0 not is used. */
|
---|
1656 | # define ADD_NUMMEM 0
|
---|
1657 | #endif
|
---|
1658 |
|
---|
1659 | n = reg->num_repeat + (reg->num_mem + ADD_NUMMEM) * 2;
|
---|
1660 |
|
---|
1661 | STACK_INIT(alloca_base, xmalloc_base, n, INIT_MATCH_STACK_SIZE);
|
---|
1662 | pop_level = reg->stack_pop_level;
|
---|
1663 | num_mem = reg->num_mem;
|
---|
1664 | repeat_stk = (OnigStackIndex* )alloca_base;
|
---|
1665 |
|
---|
1666 | mem_start_stk = (OnigStackIndex* )(repeat_stk + reg->num_repeat);
|
---|
1667 | mem_end_stk = mem_start_stk + (num_mem + ADD_NUMMEM);
|
---|
1668 | {
|
---|
1669 | OnigStackIndex *pp = mem_start_stk;
|
---|
1670 | for (; pp < repeat_stk + n; pp += 2) {
|
---|
1671 | pp[0] = INVALID_STACK_INDEX;
|
---|
1672 | pp[1] = INVALID_STACK_INDEX;
|
---|
1673 | }
|
---|
1674 | }
|
---|
1675 | #ifndef USE_SUBEXP_CALL
|
---|
1676 | mem_start_stk--; /* for index start from 1,
|
---|
1677 | mem_start_stk[1]..mem_start_stk[num_mem] */
|
---|
1678 | mem_end_stk--; /* for index start from 1,
|
---|
1679 | mem_end_stk[1]..mem_end_stk[num_mem] */
|
---|
1680 | #endif
|
---|
1681 |
|
---|
1682 | #ifdef ONIG_DEBUG_MATCH
|
---|
1683 | fprintf(stderr, "match_at: str: %"PRIuPTR" (%p), end: %"PRIuPTR" (%p), start: %"PRIuPTR" (%p), sprev: %"PRIuPTR" (%p)\n",
|
---|
1684 | (uintptr_t )str, str, (uintptr_t )end, end, (uintptr_t )sstart, sstart, (uintptr_t )sprev, sprev);
|
---|
1685 | fprintf(stderr, "size: %d, start offset: %d\n",
|
---|
1686 | (int )(end - str), (int )(sstart - str));
|
---|
1687 | fprintf(stderr, "\n ofs> str stk:type addr:opcode\n");
|
---|
1688 | #endif
|
---|
1689 |
|
---|
1690 | STACK_PUSH_ENSURED(STK_ALT, (UChar* )FinishCode); /* bottom stack */
|
---|
1691 | best_len = ONIG_MISMATCH;
|
---|
1692 | s = (UChar* )sstart;
|
---|
1693 | pkeep = (UChar* )sstart;
|
---|
1694 |
|
---|
1695 |
|
---|
1696 | #ifdef ONIG_DEBUG_MATCH
|
---|
1697 | # define OPCODE_EXEC_HOOK \
|
---|
1698 | if (s) { \
|
---|
1699 | UChar *op, *q, *bp, buf[50]; \
|
---|
1700 | int len; \
|
---|
1701 | op = p - OP_OFFSET; \
|
---|
1702 | fprintf(stderr, "%4"PRIdPTR"> \"", (*op == OP_FINISH) ? (ptrdiff_t )-1 : s - str); \
|
---|
1703 | bp = buf; \
|
---|
1704 | q = s; \
|
---|
1705 | if (*op != OP_FINISH) { /* s may not be a valid pointer if OP_FINISH. */ \
|
---|
1706 | for (i = 0; i < 7 && q < end; i++) { \
|
---|
1707 | len = enclen(encode, q, end); \
|
---|
1708 | while (len-- > 0) *bp++ = *q++; \
|
---|
1709 | } \
|
---|
1710 | if (q < end) { xmemcpy(bp, "...", 3); bp += 3; } \
|
---|
1711 | } \
|
---|
1712 | xmemcpy(bp, "\"", 1); bp += 1; \
|
---|
1713 | *bp = 0; \
|
---|
1714 | fputs((char* )buf, stderr); \
|
---|
1715 | for (i = 0; i < 20 - (bp - buf); i++) fputc(' ', stderr); \
|
---|
1716 | fprintf(stderr, "%4"PRIdPTR":%s %4"PRIdPTR":", \
|
---|
1717 | stk - stk_base - 1, \
|
---|
1718 | (stk > stk_base) ? stack_type_str(stk[-1].type) : " ", \
|
---|
1719 | (op == FinishCode) ? (ptrdiff_t )-1 : op - reg->p); \
|
---|
1720 | onig_print_compiled_byte_code(stderr, op, reg->p+reg->used, NULL, encode); \
|
---|
1721 | fprintf(stderr, "\n"); \
|
---|
1722 | }
|
---|
1723 | #else
|
---|
1724 | # define OPCODE_EXEC_HOOK ((void) 0)
|
---|
1725 | #endif
|
---|
1726 |
|
---|
1727 |
|
---|
1728 | VM_LOOP {
|
---|
1729 | CASE(OP_END) MOP_IN(OP_END);
|
---|
1730 | n = s - sstart;
|
---|
1731 | if (n > best_len) {
|
---|
1732 | OnigRegion* region;
|
---|
1733 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
1734 | if (IS_FIND_LONGEST(option)) {
|
---|
1735 | if (n > msa->best_len) {
|
---|
1736 | msa->best_len = n;
|
---|
1737 | msa->best_s = (UChar* )sstart;
|
---|
1738 | }
|
---|
1739 | else
|
---|
1740 | goto end_best_len;
|
---|
1741 | }
|
---|
1742 | #endif
|
---|
1743 | best_len = n;
|
---|
1744 | region = msa->region;
|
---|
1745 | if (region) {
|
---|
1746 | region->beg[0] = ((pkeep > s) ? s : pkeep) - str;
|
---|
1747 | region->end[0] = s - str;
|
---|
1748 | for (i = 1; i <= num_mem; i++) {
|
---|
1749 | if (mem_end_stk[i] != INVALID_STACK_INDEX) {
|
---|
1750 | if (BIT_STATUS_AT(reg->bt_mem_start, i))
|
---|
1751 | region->beg[i] = STACK_AT(mem_start_stk[i])->u.mem.pstr - str;
|
---|
1752 | else
|
---|
1753 | region->beg[i] = (UChar* )((void* )mem_start_stk[i]) - str;
|
---|
1754 |
|
---|
1755 | region->end[i] = (BIT_STATUS_AT(reg->bt_mem_end, i)
|
---|
1756 | ? STACK_AT(mem_end_stk[i])->u.mem.pstr
|
---|
1757 | : (UChar* )((void* )mem_end_stk[i])) - str;
|
---|
1758 | }
|
---|
1759 | else {
|
---|
1760 | region->beg[i] = region->end[i] = ONIG_REGION_NOTPOS;
|
---|
1761 | }
|
---|
1762 | }
|
---|
1763 |
|
---|
1764 | #ifdef USE_CAPTURE_HISTORY
|
---|
1765 | if (reg->capture_history != 0) {
|
---|
1766 | int r;
|
---|
1767 | OnigCaptureTreeNode* node;
|
---|
1768 |
|
---|
1769 | if (IS_NULL(region->history_root)) {
|
---|
1770 | region->history_root = node = history_node_new();
|
---|
1771 | CHECK_NULL_RETURN_MEMERR(node);
|
---|
1772 | }
|
---|
1773 | else {
|
---|
1774 | node = region->history_root;
|
---|
1775 | history_tree_clear(node);
|
---|
1776 | }
|
---|
1777 |
|
---|
1778 | node->group = 0;
|
---|
1779 | node->beg = ((pkeep > s) ? s : pkeep) - str;
|
---|
1780 | node->end = s - str;
|
---|
1781 |
|
---|
1782 | stkp = stk_base;
|
---|
1783 | r = make_capture_history_tree(region->history_root, &stkp,
|
---|
1784 | stk, (UChar* )str, reg);
|
---|
1785 | if (r < 0) {
|
---|
1786 | best_len = r; /* error code */
|
---|
1787 | goto finish;
|
---|
1788 | }
|
---|
1789 | }
|
---|
1790 | #endif /* USE_CAPTURE_HISTORY */
|
---|
1791 | } /* if (region) */
|
---|
1792 | } /* n > best_len */
|
---|
1793 |
|
---|
1794 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
1795 | end_best_len:
|
---|
1796 | #endif
|
---|
1797 | MOP_OUT;
|
---|
1798 |
|
---|
1799 | if (IS_FIND_CONDITION(option)) {
|
---|
1800 | if (IS_FIND_NOT_EMPTY(option) && s == sstart) {
|
---|
1801 | best_len = ONIG_MISMATCH;
|
---|
1802 | goto fail; /* for retry */
|
---|
1803 | }
|
---|
1804 | if (IS_FIND_LONGEST(option) && DATA_ENSURE_CHECK1) {
|
---|
1805 | goto fail; /* for retry */
|
---|
1806 | }
|
---|
1807 | }
|
---|
1808 |
|
---|
1809 | /* default behavior: return first-matching result. */
|
---|
1810 | goto finish;
|
---|
1811 | NEXT;
|
---|
1812 |
|
---|
1813 | CASE(OP_EXACT1) MOP_IN(OP_EXACT1);
|
---|
1814 | DATA_ENSURE(1);
|
---|
1815 | if (*p != *s) goto fail;
|
---|
1816 | p++; s++;
|
---|
1817 | MOP_OUT;
|
---|
1818 | NEXT;
|
---|
1819 |
|
---|
1820 | CASE(OP_EXACT1_IC) MOP_IN(OP_EXACT1_IC);
|
---|
1821 | {
|
---|
1822 | int len;
|
---|
1823 | UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
---|
1824 |
|
---|
1825 | DATA_ENSURE(1);
|
---|
1826 | len = ONIGENC_MBC_CASE_FOLD(encode,
|
---|
1827 | /* DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag), */
|
---|
1828 | case_fold_flag,
|
---|
1829 | &s, end, lowbuf);
|
---|
1830 | DATA_ENSURE(0);
|
---|
1831 | q = lowbuf;
|
---|
1832 | while (len-- > 0) {
|
---|
1833 | if (*p != *q) {
|
---|
1834 | goto fail;
|
---|
1835 | }
|
---|
1836 | p++; q++;
|
---|
1837 | }
|
---|
1838 | }
|
---|
1839 | MOP_OUT;
|
---|
1840 | NEXT;
|
---|
1841 |
|
---|
1842 | CASE(OP_EXACT2) MOP_IN(OP_EXACT2);
|
---|
1843 | DATA_ENSURE(2);
|
---|
1844 | if (*p != *s) goto fail;
|
---|
1845 | p++; s++;
|
---|
1846 | if (*p != *s) goto fail;
|
---|
1847 | sprev = s;
|
---|
1848 | p++; s++;
|
---|
1849 | MOP_OUT;
|
---|
1850 | JUMP;
|
---|
1851 |
|
---|
1852 | CASE(OP_EXACT3) MOP_IN(OP_EXACT3);
|
---|
1853 | DATA_ENSURE(3);
|
---|
1854 | if (*p != *s) goto fail;
|
---|
1855 | p++; s++;
|
---|
1856 | if (*p != *s) goto fail;
|
---|
1857 | p++; s++;
|
---|
1858 | if (*p != *s) goto fail;
|
---|
1859 | sprev = s;
|
---|
1860 | p++; s++;
|
---|
1861 | MOP_OUT;
|
---|
1862 | JUMP;
|
---|
1863 |
|
---|
1864 | CASE(OP_EXACT4) MOP_IN(OP_EXACT4);
|
---|
1865 | DATA_ENSURE(4);
|
---|
1866 | if (*p != *s) goto fail;
|
---|
1867 | p++; s++;
|
---|
1868 | if (*p != *s) goto fail;
|
---|
1869 | p++; s++;
|
---|
1870 | if (*p != *s) goto fail;
|
---|
1871 | p++; s++;
|
---|
1872 | if (*p != *s) goto fail;
|
---|
1873 | sprev = s;
|
---|
1874 | p++; s++;
|
---|
1875 | MOP_OUT;
|
---|
1876 | JUMP;
|
---|
1877 |
|
---|
1878 | CASE(OP_EXACT5) MOP_IN(OP_EXACT5);
|
---|
1879 | DATA_ENSURE(5);
|
---|
1880 | if (*p != *s) goto fail;
|
---|
1881 | p++; s++;
|
---|
1882 | if (*p != *s) goto fail;
|
---|
1883 | p++; s++;
|
---|
1884 | if (*p != *s) goto fail;
|
---|
1885 | p++; s++;
|
---|
1886 | if (*p != *s) goto fail;
|
---|
1887 | p++; s++;
|
---|
1888 | if (*p != *s) goto fail;
|
---|
1889 | sprev = s;
|
---|
1890 | p++; s++;
|
---|
1891 | MOP_OUT;
|
---|
1892 | JUMP;
|
---|
1893 |
|
---|
1894 | CASE(OP_EXACTN) MOP_IN(OP_EXACTN);
|
---|
1895 | GET_LENGTH_INC(tlen, p);
|
---|
1896 | DATA_ENSURE(tlen);
|
---|
1897 | while (tlen-- > 0) {
|
---|
1898 | if (*p++ != *s++) goto fail;
|
---|
1899 | }
|
---|
1900 | sprev = s - 1;
|
---|
1901 | MOP_OUT;
|
---|
1902 | JUMP;
|
---|
1903 |
|
---|
1904 | CASE(OP_EXACTN_IC) MOP_IN(OP_EXACTN_IC);
|
---|
1905 | {
|
---|
1906 | int len;
|
---|
1907 | UChar *q, *endp, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
---|
1908 |
|
---|
1909 | GET_LENGTH_INC(tlen, p);
|
---|
1910 | endp = p + tlen;
|
---|
1911 |
|
---|
1912 | while (p < endp) {
|
---|
1913 | sprev = s;
|
---|
1914 | DATA_ENSURE(1);
|
---|
1915 | len = ONIGENC_MBC_CASE_FOLD(encode,
|
---|
1916 | /* DISABLE_CASE_FOLD_MULTI_CHAR(case_fold_flag), */
|
---|
1917 | case_fold_flag,
|
---|
1918 | &s, end, lowbuf);
|
---|
1919 | DATA_ENSURE(0);
|
---|
1920 | q = lowbuf;
|
---|
1921 | while (len-- > 0) {
|
---|
1922 | if (*p != *q) goto fail;
|
---|
1923 | p++; q++;
|
---|
1924 | }
|
---|
1925 | }
|
---|
1926 | }
|
---|
1927 |
|
---|
1928 | MOP_OUT;
|
---|
1929 | JUMP;
|
---|
1930 |
|
---|
1931 | CASE(OP_EXACTMB2N1) MOP_IN(OP_EXACTMB2N1);
|
---|
1932 | DATA_ENSURE(2);
|
---|
1933 | if (*p != *s) goto fail;
|
---|
1934 | p++; s++;
|
---|
1935 | if (*p != *s) goto fail;
|
---|
1936 | p++; s++;
|
---|
1937 | MOP_OUT;
|
---|
1938 | NEXT;
|
---|
1939 |
|
---|
1940 | CASE(OP_EXACTMB2N2) MOP_IN(OP_EXACTMB2N2);
|
---|
1941 | DATA_ENSURE(4);
|
---|
1942 | if (*p != *s) goto fail;
|
---|
1943 | p++; s++;
|
---|
1944 | if (*p != *s) goto fail;
|
---|
1945 | p++; s++;
|
---|
1946 | sprev = s;
|
---|
1947 | if (*p != *s) goto fail;
|
---|
1948 | p++; s++;
|
---|
1949 | if (*p != *s) goto fail;
|
---|
1950 | p++; s++;
|
---|
1951 | MOP_OUT;
|
---|
1952 | JUMP;
|
---|
1953 |
|
---|
1954 | CASE(OP_EXACTMB2N3) MOP_IN(OP_EXACTMB2N3);
|
---|
1955 | DATA_ENSURE(6);
|
---|
1956 | if (*p != *s) goto fail;
|
---|
1957 | p++; s++;
|
---|
1958 | if (*p != *s) goto fail;
|
---|
1959 | p++; s++;
|
---|
1960 | if (*p != *s) goto fail;
|
---|
1961 | p++; s++;
|
---|
1962 | if (*p != *s) goto fail;
|
---|
1963 | p++; s++;
|
---|
1964 | sprev = s;
|
---|
1965 | if (*p != *s) goto fail;
|
---|
1966 | p++; s++;
|
---|
1967 | if (*p != *s) goto fail;
|
---|
1968 | p++; s++;
|
---|
1969 | MOP_OUT;
|
---|
1970 | JUMP;
|
---|
1971 |
|
---|
1972 | CASE(OP_EXACTMB2N) MOP_IN(OP_EXACTMB2N);
|
---|
1973 | GET_LENGTH_INC(tlen, p);
|
---|
1974 | DATA_ENSURE(tlen * 2);
|
---|
1975 | while (tlen-- > 0) {
|
---|
1976 | if (*p != *s) goto fail;
|
---|
1977 | p++; s++;
|
---|
1978 | if (*p != *s) goto fail;
|
---|
1979 | p++; s++;
|
---|
1980 | }
|
---|
1981 | sprev = s - 2;
|
---|
1982 | MOP_OUT;
|
---|
1983 | JUMP;
|
---|
1984 |
|
---|
1985 | CASE(OP_EXACTMB3N) MOP_IN(OP_EXACTMB3N);
|
---|
1986 | GET_LENGTH_INC(tlen, p);
|
---|
1987 | DATA_ENSURE(tlen * 3);
|
---|
1988 | while (tlen-- > 0) {
|
---|
1989 | if (*p != *s) goto fail;
|
---|
1990 | p++; s++;
|
---|
1991 | if (*p != *s) goto fail;
|
---|
1992 | p++; s++;
|
---|
1993 | if (*p != *s) goto fail;
|
---|
1994 | p++; s++;
|
---|
1995 | }
|
---|
1996 | sprev = s - 3;
|
---|
1997 | MOP_OUT;
|
---|
1998 | JUMP;
|
---|
1999 |
|
---|
2000 | CASE(OP_EXACTMBN) MOP_IN(OP_EXACTMBN);
|
---|
2001 | GET_LENGTH_INC(tlen, p); /* mb-len */
|
---|
2002 | GET_LENGTH_INC(tlen2, p); /* string len */
|
---|
2003 | tlen2 *= tlen;
|
---|
2004 | DATA_ENSURE(tlen2);
|
---|
2005 | while (tlen2-- > 0) {
|
---|
2006 | if (*p != *s) goto fail;
|
---|
2007 | p++; s++;
|
---|
2008 | }
|
---|
2009 | sprev = s - tlen;
|
---|
2010 | MOP_OUT;
|
---|
2011 | JUMP;
|
---|
2012 |
|
---|
2013 | CASE(OP_CCLASS) MOP_IN(OP_CCLASS);
|
---|
2014 | DATA_ENSURE(1);
|
---|
2015 | if (BITSET_AT(((BitSetRef )p), *s) == 0) goto fail;
|
---|
2016 | p += SIZE_BITSET;
|
---|
2017 | s += enclen(encode, s, end); /* OP_CCLASS can match mb-code. \D, \S */
|
---|
2018 | MOP_OUT;
|
---|
2019 | NEXT;
|
---|
2020 |
|
---|
2021 | CASE(OP_CCLASS_MB) MOP_IN(OP_CCLASS_MB);
|
---|
2022 | if (! ONIGENC_IS_MBC_HEAD(encode, s, end)) goto fail;
|
---|
2023 |
|
---|
2024 | cclass_mb:
|
---|
2025 | GET_LENGTH_INC(tlen, p);
|
---|
2026 | {
|
---|
2027 | OnigCodePoint code;
|
---|
2028 | UChar *ss;
|
---|
2029 | int mb_len;
|
---|
2030 |
|
---|
2031 | DATA_ENSURE(1);
|
---|
2032 | mb_len = enclen(encode, s, end);
|
---|
2033 | DATA_ENSURE(mb_len);
|
---|
2034 | ss = s;
|
---|
2035 | s += mb_len;
|
---|
2036 | code = ONIGENC_MBC_TO_CODE(encode, ss, s);
|
---|
2037 |
|
---|
2038 | #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
|
---|
2039 | if (! onig_is_in_code_range(p, code)) goto fail;
|
---|
2040 | #else
|
---|
2041 | q = p;
|
---|
2042 | ALIGNMENT_RIGHT(q);
|
---|
2043 | if (! onig_is_in_code_range(q, code)) goto fail;
|
---|
2044 | #endif
|
---|
2045 | }
|
---|
2046 | p += tlen;
|
---|
2047 | MOP_OUT;
|
---|
2048 | NEXT;
|
---|
2049 |
|
---|
2050 | CASE(OP_CCLASS_MIX) MOP_IN(OP_CCLASS_MIX);
|
---|
2051 | DATA_ENSURE(1);
|
---|
2052 | if (ONIGENC_IS_MBC_HEAD(encode, s, end)) {
|
---|
2053 | p += SIZE_BITSET;
|
---|
2054 | goto cclass_mb;
|
---|
2055 | }
|
---|
2056 | else {
|
---|
2057 | if (BITSET_AT(((BitSetRef )p), *s) == 0)
|
---|
2058 | goto fail;
|
---|
2059 |
|
---|
2060 | p += SIZE_BITSET;
|
---|
2061 | GET_LENGTH_INC(tlen, p);
|
---|
2062 | p += tlen;
|
---|
2063 | s++;
|
---|
2064 | }
|
---|
2065 | MOP_OUT;
|
---|
2066 | NEXT;
|
---|
2067 |
|
---|
2068 | CASE(OP_CCLASS_NOT) MOP_IN(OP_CCLASS_NOT);
|
---|
2069 | DATA_ENSURE(1);
|
---|
2070 | if (BITSET_AT(((BitSetRef )p), *s) != 0) goto fail;
|
---|
2071 | p += SIZE_BITSET;
|
---|
2072 | s += enclen(encode, s, end);
|
---|
2073 | MOP_OUT;
|
---|
2074 | NEXT;
|
---|
2075 |
|
---|
2076 | CASE(OP_CCLASS_MB_NOT) MOP_IN(OP_CCLASS_MB_NOT);
|
---|
2077 | DATA_ENSURE(1);
|
---|
2078 | if (! ONIGENC_IS_MBC_HEAD(encode, s, end)) {
|
---|
2079 | s++;
|
---|
2080 | GET_LENGTH_INC(tlen, p);
|
---|
2081 | p += tlen;
|
---|
2082 | goto cc_mb_not_success;
|
---|
2083 | }
|
---|
2084 |
|
---|
2085 | cclass_mb_not:
|
---|
2086 | GET_LENGTH_INC(tlen, p);
|
---|
2087 | {
|
---|
2088 | OnigCodePoint code;
|
---|
2089 | UChar *ss;
|
---|
2090 | int mb_len = enclen(encode, s, end);
|
---|
2091 |
|
---|
2092 | if (! DATA_ENSURE_CHECK(mb_len)) {
|
---|
2093 | DATA_ENSURE(1);
|
---|
2094 | s = (UChar* )end;
|
---|
2095 | p += tlen;
|
---|
2096 | goto cc_mb_not_success;
|
---|
2097 | }
|
---|
2098 |
|
---|
2099 | ss = s;
|
---|
2100 | s += mb_len;
|
---|
2101 | code = ONIGENC_MBC_TO_CODE(encode, ss, s);
|
---|
2102 |
|
---|
2103 | #ifdef PLATFORM_UNALIGNED_WORD_ACCESS
|
---|
2104 | if (onig_is_in_code_range(p, code)) goto fail;
|
---|
2105 | #else
|
---|
2106 | q = p;
|
---|
2107 | ALIGNMENT_RIGHT(q);
|
---|
2108 | if (onig_is_in_code_range(q, code)) goto fail;
|
---|
2109 | #endif
|
---|
2110 | }
|
---|
2111 | p += tlen;
|
---|
2112 |
|
---|
2113 | cc_mb_not_success:
|
---|
2114 | MOP_OUT;
|
---|
2115 | NEXT;
|
---|
2116 |
|
---|
2117 | CASE(OP_CCLASS_MIX_NOT) MOP_IN(OP_CCLASS_MIX_NOT);
|
---|
2118 | DATA_ENSURE(1);
|
---|
2119 | if (ONIGENC_IS_MBC_HEAD(encode, s, end)) {
|
---|
2120 | p += SIZE_BITSET;
|
---|
2121 | goto cclass_mb_not;
|
---|
2122 | }
|
---|
2123 | else {
|
---|
2124 | if (BITSET_AT(((BitSetRef )p), *s) != 0)
|
---|
2125 | goto fail;
|
---|
2126 |
|
---|
2127 | p += SIZE_BITSET;
|
---|
2128 | GET_LENGTH_INC(tlen, p);
|
---|
2129 | p += tlen;
|
---|
2130 | s++;
|
---|
2131 | }
|
---|
2132 | MOP_OUT;
|
---|
2133 | NEXT;
|
---|
2134 |
|
---|
2135 | CASE(OP_ANYCHAR) MOP_IN(OP_ANYCHAR);
|
---|
2136 | DATA_ENSURE(1);
|
---|
2137 | n = enclen(encode, s, end);
|
---|
2138 | DATA_ENSURE(n);
|
---|
2139 | if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
|
---|
2140 | s += n;
|
---|
2141 | MOP_OUT;
|
---|
2142 | NEXT;
|
---|
2143 |
|
---|
2144 | CASE(OP_ANYCHAR_ML) MOP_IN(OP_ANYCHAR_ML);
|
---|
2145 | DATA_ENSURE(1);
|
---|
2146 | n = enclen(encode, s, end);
|
---|
2147 | DATA_ENSURE(n);
|
---|
2148 | s += n;
|
---|
2149 | MOP_OUT;
|
---|
2150 | NEXT;
|
---|
2151 |
|
---|
2152 | CASE(OP_ANYCHAR_STAR) MOP_IN(OP_ANYCHAR_STAR);
|
---|
2153 | while (DATA_ENSURE_CHECK1) {
|
---|
2154 | STACK_PUSH_ALT(p, s, sprev, pkeep);
|
---|
2155 | n = enclen(encode, s, end);
|
---|
2156 | DATA_ENSURE(n);
|
---|
2157 | if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
|
---|
2158 | sprev = s;
|
---|
2159 | s += n;
|
---|
2160 | }
|
---|
2161 | MOP_OUT;
|
---|
2162 | JUMP;
|
---|
2163 |
|
---|
2164 | CASE(OP_ANYCHAR_ML_STAR) MOP_IN(OP_ANYCHAR_ML_STAR);
|
---|
2165 | while (DATA_ENSURE_CHECK1) {
|
---|
2166 | STACK_PUSH_ALT(p, s, sprev, pkeep);
|
---|
2167 | n = enclen(encode, s, end);
|
---|
2168 | if (n > 1) {
|
---|
2169 | DATA_ENSURE(n);
|
---|
2170 | sprev = s;
|
---|
2171 | s += n;
|
---|
2172 | }
|
---|
2173 | else {
|
---|
2174 | sprev = s;
|
---|
2175 | s++;
|
---|
2176 | }
|
---|
2177 | }
|
---|
2178 | MOP_OUT;
|
---|
2179 | JUMP;
|
---|
2180 |
|
---|
2181 | CASE(OP_ANYCHAR_STAR_PEEK_NEXT) MOP_IN(OP_ANYCHAR_STAR_PEEK_NEXT);
|
---|
2182 | while (DATA_ENSURE_CHECK1) {
|
---|
2183 | if (*p == *s) {
|
---|
2184 | STACK_PUSH_ALT(p + 1, s, sprev, pkeep);
|
---|
2185 | }
|
---|
2186 | n = enclen(encode, s, end);
|
---|
2187 | DATA_ENSURE(n);
|
---|
2188 | if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
|
---|
2189 | sprev = s;
|
---|
2190 | s += n;
|
---|
2191 | }
|
---|
2192 | p++;
|
---|
2193 | MOP_OUT;
|
---|
2194 | NEXT;
|
---|
2195 |
|
---|
2196 | CASE(OP_ANYCHAR_ML_STAR_PEEK_NEXT)MOP_IN(OP_ANYCHAR_ML_STAR_PEEK_NEXT);
|
---|
2197 | while (DATA_ENSURE_CHECK1) {
|
---|
2198 | if (*p == *s) {
|
---|
2199 | STACK_PUSH_ALT(p + 1, s, sprev, pkeep);
|
---|
2200 | }
|
---|
2201 | n = enclen(encode, s, end);
|
---|
2202 | if (n > 1) {
|
---|
2203 | DATA_ENSURE(n);
|
---|
2204 | sprev = s;
|
---|
2205 | s += n;
|
---|
2206 | }
|
---|
2207 | else {
|
---|
2208 | sprev = s;
|
---|
2209 | s++;
|
---|
2210 | }
|
---|
2211 | }
|
---|
2212 | p++;
|
---|
2213 | MOP_OUT;
|
---|
2214 | NEXT;
|
---|
2215 |
|
---|
2216 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
2217 | CASE(OP_STATE_CHECK_ANYCHAR_STAR) MOP_IN(OP_STATE_CHECK_ANYCHAR_STAR);
|
---|
2218 | GET_STATE_CHECK_NUM_INC(mem, p);
|
---|
2219 | while (DATA_ENSURE_CHECK1) {
|
---|
2220 | STATE_CHECK_VAL(scv, mem);
|
---|
2221 | if (scv) goto fail;
|
---|
2222 |
|
---|
2223 | STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem, pkeep);
|
---|
2224 | n = enclen(encode, s, end);
|
---|
2225 | DATA_ENSURE(n);
|
---|
2226 | if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 0)) goto fail;
|
---|
2227 | sprev = s;
|
---|
2228 | s += n;
|
---|
2229 | }
|
---|
2230 | MOP_OUT;
|
---|
2231 | NEXT;
|
---|
2232 |
|
---|
2233 | CASE(OP_STATE_CHECK_ANYCHAR_ML_STAR)
|
---|
2234 | MOP_IN(OP_STATE_CHECK_ANYCHAR_ML_STAR);
|
---|
2235 |
|
---|
2236 | GET_STATE_CHECK_NUM_INC(mem, p);
|
---|
2237 | while (DATA_ENSURE_CHECK1) {
|
---|
2238 | STATE_CHECK_VAL(scv, mem);
|
---|
2239 | if (scv) goto fail;
|
---|
2240 |
|
---|
2241 | STACK_PUSH_ALT_WITH_STATE_CHECK(p, s, sprev, mem, pkeep);
|
---|
2242 | n = enclen(encode, s, end);
|
---|
2243 | if (n > 1) {
|
---|
2244 | DATA_ENSURE(n);
|
---|
2245 | sprev = s;
|
---|
2246 | s += n;
|
---|
2247 | }
|
---|
2248 | else {
|
---|
2249 | sprev = s;
|
---|
2250 | s++;
|
---|
2251 | }
|
---|
2252 | }
|
---|
2253 | MOP_OUT;
|
---|
2254 | NEXT;
|
---|
2255 | #endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
2256 |
|
---|
2257 | CASE(OP_WORD) MOP_IN(OP_WORD);
|
---|
2258 | DATA_ENSURE(1);
|
---|
2259 | if (! ONIGENC_IS_MBC_WORD(encode, s, end))
|
---|
2260 | goto fail;
|
---|
2261 |
|
---|
2262 | s += enclen(encode, s, end);
|
---|
2263 | MOP_OUT;
|
---|
2264 | NEXT;
|
---|
2265 |
|
---|
2266 | CASE(OP_ASCII_WORD) MOP_IN(OP_ASCII_WORD);
|
---|
2267 | DATA_ENSURE(1);
|
---|
2268 | if (! ONIGENC_IS_MBC_ASCII_WORD(encode, s, end))
|
---|
2269 | goto fail;
|
---|
2270 |
|
---|
2271 | s += enclen(encode, s, end);
|
---|
2272 | MOP_OUT;
|
---|
2273 | NEXT;
|
---|
2274 |
|
---|
2275 | CASE(OP_NOT_WORD) MOP_IN(OP_NOT_WORD);
|
---|
2276 | DATA_ENSURE(1);
|
---|
2277 | if (ONIGENC_IS_MBC_WORD(encode, s, end))
|
---|
2278 | goto fail;
|
---|
2279 |
|
---|
2280 | s += enclen(encode, s, end);
|
---|
2281 | MOP_OUT;
|
---|
2282 | NEXT;
|
---|
2283 |
|
---|
2284 | CASE(OP_NOT_ASCII_WORD) MOP_IN(OP_NOT_ASCII_WORD);
|
---|
2285 | DATA_ENSURE(1);
|
---|
2286 | if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end))
|
---|
2287 | goto fail;
|
---|
2288 |
|
---|
2289 | s += enclen(encode, s, end);
|
---|
2290 | MOP_OUT;
|
---|
2291 | NEXT;
|
---|
2292 |
|
---|
2293 | CASE(OP_WORD_BOUND) MOP_IN(OP_WORD_BOUND);
|
---|
2294 | if (ON_STR_BEGIN(s)) {
|
---|
2295 | DATA_ENSURE(1);
|
---|
2296 | if (! ONIGENC_IS_MBC_WORD(encode, s, end))
|
---|
2297 | goto fail;
|
---|
2298 | }
|
---|
2299 | else if (ON_STR_END(s)) {
|
---|
2300 | if (! ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
---|
2301 | goto fail;
|
---|
2302 | }
|
---|
2303 | else {
|
---|
2304 | if (ONIGENC_IS_MBC_WORD(encode, s, end)
|
---|
2305 | == ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
---|
2306 | goto fail;
|
---|
2307 | }
|
---|
2308 | MOP_OUT;
|
---|
2309 | JUMP;
|
---|
2310 |
|
---|
2311 | CASE(OP_ASCII_WORD_BOUND) MOP_IN(OP_ASCII_WORD_BOUND);
|
---|
2312 | if (ON_STR_BEGIN(s)) {
|
---|
2313 | DATA_ENSURE(1);
|
---|
2314 | if (! ONIGENC_IS_MBC_ASCII_WORD(encode, s, end))
|
---|
2315 | goto fail;
|
---|
2316 | }
|
---|
2317 | else if (ON_STR_END(s)) {
|
---|
2318 | if (! ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end))
|
---|
2319 | goto fail;
|
---|
2320 | }
|
---|
2321 | else {
|
---|
2322 | if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)
|
---|
2323 | == ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end))
|
---|
2324 | goto fail;
|
---|
2325 | }
|
---|
2326 | MOP_OUT;
|
---|
2327 | JUMP;
|
---|
2328 |
|
---|
2329 | CASE(OP_NOT_WORD_BOUND) MOP_IN(OP_NOT_WORD_BOUND);
|
---|
2330 | if (ON_STR_BEGIN(s)) {
|
---|
2331 | if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_WORD(encode, s, end))
|
---|
2332 | goto fail;
|
---|
2333 | }
|
---|
2334 | else if (ON_STR_END(s)) {
|
---|
2335 | if (ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
---|
2336 | goto fail;
|
---|
2337 | }
|
---|
2338 | else {
|
---|
2339 | if (ONIGENC_IS_MBC_WORD(encode, s, end)
|
---|
2340 | != ONIGENC_IS_MBC_WORD(encode, sprev, end))
|
---|
2341 | goto fail;
|
---|
2342 | }
|
---|
2343 | MOP_OUT;
|
---|
2344 | JUMP;
|
---|
2345 |
|
---|
2346 | CASE(OP_NOT_ASCII_WORD_BOUND) MOP_IN(OP_NOT_ASCII_WORD_BOUND);
|
---|
2347 | if (ON_STR_BEGIN(s)) {
|
---|
2348 | if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_ASCII_WORD(encode, s, end))
|
---|
2349 | goto fail;
|
---|
2350 | }
|
---|
2351 | else if (ON_STR_END(s)) {
|
---|
2352 | if (ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end))
|
---|
2353 | goto fail;
|
---|
2354 | }
|
---|
2355 | else {
|
---|
2356 | if (ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)
|
---|
2357 | != ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end))
|
---|
2358 | goto fail;
|
---|
2359 | }
|
---|
2360 | MOP_OUT;
|
---|
2361 | JUMP;
|
---|
2362 |
|
---|
2363 | #ifdef USE_WORD_BEGIN_END
|
---|
2364 | CASE(OP_WORD_BEGIN) MOP_IN(OP_WORD_BEGIN);
|
---|
2365 | if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_WORD(encode, s, end)) {
|
---|
2366 | if (ON_STR_BEGIN(s) || !ONIGENC_IS_MBC_WORD(encode, sprev, end)) {
|
---|
2367 | MOP_OUT;
|
---|
2368 | JUMP;
|
---|
2369 | }
|
---|
2370 | }
|
---|
2371 | goto fail;
|
---|
2372 | NEXT;
|
---|
2373 |
|
---|
2374 | CASE(OP_ASCII_WORD_BEGIN) MOP_IN(OP_ASCII_WORD_BEGIN);
|
---|
2375 | if (DATA_ENSURE_CHECK1 && ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) {
|
---|
2376 | if (ON_STR_BEGIN(s) || !ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) {
|
---|
2377 | MOP_OUT;
|
---|
2378 | JUMP;
|
---|
2379 | }
|
---|
2380 | }
|
---|
2381 | goto fail;
|
---|
2382 | NEXT;
|
---|
2383 |
|
---|
2384 | CASE(OP_WORD_END) MOP_IN(OP_WORD_END);
|
---|
2385 | if (!ON_STR_BEGIN(s) && ONIGENC_IS_MBC_WORD(encode, sprev, end)) {
|
---|
2386 | if (ON_STR_END(s) || !ONIGENC_IS_MBC_WORD(encode, s, end)) {
|
---|
2387 | MOP_OUT;
|
---|
2388 | JUMP;
|
---|
2389 | }
|
---|
2390 | }
|
---|
2391 | goto fail;
|
---|
2392 | NEXT;
|
---|
2393 |
|
---|
2394 | CASE(OP_ASCII_WORD_END) MOP_IN(OP_ASCII_WORD_END);
|
---|
2395 | if (!ON_STR_BEGIN(s) && ONIGENC_IS_MBC_ASCII_WORD(encode, sprev, end)) {
|
---|
2396 | if (ON_STR_END(s) || !ONIGENC_IS_MBC_ASCII_WORD(encode, s, end)) {
|
---|
2397 | MOP_OUT;
|
---|
2398 | JUMP;
|
---|
2399 | }
|
---|
2400 | }
|
---|
2401 | goto fail;
|
---|
2402 | NEXT;
|
---|
2403 | #endif
|
---|
2404 |
|
---|
2405 | CASE(OP_BEGIN_BUF) MOP_IN(OP_BEGIN_BUF);
|
---|
2406 | if (! ON_STR_BEGIN(s)) goto fail;
|
---|
2407 | if (IS_NOTBOS(msa->options)) goto fail;
|
---|
2408 |
|
---|
2409 | MOP_OUT;
|
---|
2410 | JUMP;
|
---|
2411 |
|
---|
2412 | CASE(OP_END_BUF) MOP_IN(OP_END_BUF);
|
---|
2413 | if (! ON_STR_END(s)) goto fail;
|
---|
2414 | if (IS_NOTEOS(msa->options)) goto fail;
|
---|
2415 |
|
---|
2416 | MOP_OUT;
|
---|
2417 | JUMP;
|
---|
2418 |
|
---|
2419 | CASE(OP_BEGIN_LINE) MOP_IN(OP_BEGIN_LINE);
|
---|
2420 | if (ON_STR_BEGIN(s)) {
|
---|
2421 | if (IS_NOTBOL(msa->options)) goto fail;
|
---|
2422 | MOP_OUT;
|
---|
2423 | JUMP;
|
---|
2424 | }
|
---|
2425 | else if (ONIGENC_IS_MBC_NEWLINE(encode, sprev, end)
|
---|
2426 | #ifdef USE_CRNL_AS_LINE_TERMINATOR
|
---|
2427 | && !(IS_NEWLINE_CRLF(option)
|
---|
2428 | && ONIGENC_IS_MBC_CRNL(encode, sprev, end))
|
---|
2429 | #endif
|
---|
2430 | && !ON_STR_END(s)) {
|
---|
2431 | MOP_OUT;
|
---|
2432 | JUMP;
|
---|
2433 | }
|
---|
2434 | goto fail;
|
---|
2435 | NEXT;
|
---|
2436 |
|
---|
2437 | CASE(OP_END_LINE) MOP_IN(OP_END_LINE);
|
---|
2438 | if (ON_STR_END(s)) {
|
---|
2439 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
2440 | if (IS_EMPTY_STR || !ONIGENC_IS_MBC_NEWLINE_EX(encode, sprev, str, end, option, 1)) {
|
---|
2441 | #endif
|
---|
2442 | if (IS_NOTEOL(msa->options)) goto fail;
|
---|
2443 | MOP_OUT;
|
---|
2444 | JUMP;
|
---|
2445 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
2446 | }
|
---|
2447 | #endif
|
---|
2448 | }
|
---|
2449 | else if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 1)) {
|
---|
2450 | MOP_OUT;
|
---|
2451 | JUMP;
|
---|
2452 | }
|
---|
2453 | goto fail;
|
---|
2454 | NEXT;
|
---|
2455 |
|
---|
2456 | CASE(OP_SEMI_END_BUF) MOP_IN(OP_SEMI_END_BUF);
|
---|
2457 | if (ON_STR_END(s)) {
|
---|
2458 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
2459 | if (IS_EMPTY_STR || !ONIGENC_IS_MBC_NEWLINE_EX(encode, sprev, str, end, option, 1)) {
|
---|
2460 | #endif
|
---|
2461 | if (IS_NOTEOL(msa->options)) goto fail;
|
---|
2462 | MOP_OUT;
|
---|
2463 | JUMP;
|
---|
2464 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
2465 | }
|
---|
2466 | #endif
|
---|
2467 | }
|
---|
2468 | else if (ONIGENC_IS_MBC_NEWLINE_EX(encode, s, str, end, option, 1)) {
|
---|
2469 | UChar* ss = s + enclen(encode, s, end);
|
---|
2470 | if (ON_STR_END(ss)) {
|
---|
2471 | MOP_OUT;
|
---|
2472 | JUMP;
|
---|
2473 | }
|
---|
2474 | #ifdef USE_CRNL_AS_LINE_TERMINATOR
|
---|
2475 | else if (IS_NEWLINE_CRLF(option)
|
---|
2476 | && ONIGENC_IS_MBC_CRNL(encode, s, end)) {
|
---|
2477 | ss += enclen(encode, ss, end);
|
---|
2478 | if (ON_STR_END(ss)) {
|
---|
2479 | MOP_OUT;
|
---|
2480 | JUMP;
|
---|
2481 | }
|
---|
2482 | }
|
---|
2483 | #endif
|
---|
2484 | }
|
---|
2485 | goto fail;
|
---|
2486 | NEXT;
|
---|
2487 |
|
---|
2488 | CASE(OP_BEGIN_POSITION) MOP_IN(OP_BEGIN_POSITION);
|
---|
2489 | if (s != msa->gpos)
|
---|
2490 | goto fail;
|
---|
2491 |
|
---|
2492 | MOP_OUT;
|
---|
2493 | JUMP;
|
---|
2494 |
|
---|
2495 | CASE(OP_MEMORY_START_PUSH) MOP_IN(OP_MEMORY_START_PUSH);
|
---|
2496 | GET_MEMNUM_INC(mem, p);
|
---|
2497 | STACK_PUSH_MEM_START(mem, s);
|
---|
2498 | MOP_OUT;
|
---|
2499 | JUMP;
|
---|
2500 |
|
---|
2501 | CASE(OP_MEMORY_START) MOP_IN(OP_MEMORY_START);
|
---|
2502 | GET_MEMNUM_INC(mem, p);
|
---|
2503 | mem_start_stk[mem] = (OnigStackIndex )((void* )s);
|
---|
2504 | MOP_OUT;
|
---|
2505 | JUMP;
|
---|
2506 |
|
---|
2507 | CASE(OP_MEMORY_END_PUSH) MOP_IN(OP_MEMORY_END_PUSH);
|
---|
2508 | GET_MEMNUM_INC(mem, p);
|
---|
2509 | STACK_PUSH_MEM_END(mem, s);
|
---|
2510 | MOP_OUT;
|
---|
2511 | JUMP;
|
---|
2512 |
|
---|
2513 | CASE(OP_MEMORY_END) MOP_IN(OP_MEMORY_END);
|
---|
2514 | GET_MEMNUM_INC(mem, p);
|
---|
2515 | mem_end_stk[mem] = (OnigStackIndex )((void* )s);
|
---|
2516 | MOP_OUT;
|
---|
2517 | JUMP;
|
---|
2518 |
|
---|
2519 | CASE(OP_KEEP) MOP_IN(OP_KEEP);
|
---|
2520 | pkeep = s;
|
---|
2521 | MOP_OUT;
|
---|
2522 | JUMP;
|
---|
2523 |
|
---|
2524 | #ifdef USE_SUBEXP_CALL
|
---|
2525 | CASE(OP_MEMORY_END_PUSH_REC) MOP_IN(OP_MEMORY_END_PUSH_REC);
|
---|
2526 | GET_MEMNUM_INC(mem, p);
|
---|
2527 | STACK_GET_MEM_START(mem, stkp); /* should be before push mem-end. */
|
---|
2528 | STACK_PUSH_MEM_END(mem, s);
|
---|
2529 | mem_start_stk[mem] = GET_STACK_INDEX(stkp);
|
---|
2530 | MOP_OUT;
|
---|
2531 | JUMP;
|
---|
2532 |
|
---|
2533 | CASE(OP_MEMORY_END_REC) MOP_IN(OP_MEMORY_END_REC);
|
---|
2534 | GET_MEMNUM_INC(mem, p);
|
---|
2535 | mem_end_stk[mem] = (OnigStackIndex )((void* )s);
|
---|
2536 | STACK_GET_MEM_START(mem, stkp);
|
---|
2537 |
|
---|
2538 | if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
---|
2539 | mem_start_stk[mem] = GET_STACK_INDEX(stkp);
|
---|
2540 | else
|
---|
2541 | mem_start_stk[mem] = (OnigStackIndex )((void* )stkp->u.mem.pstr);
|
---|
2542 |
|
---|
2543 | STACK_PUSH_MEM_END_MARK(mem);
|
---|
2544 | MOP_OUT;
|
---|
2545 | JUMP;
|
---|
2546 | #endif
|
---|
2547 |
|
---|
2548 | CASE(OP_BACKREF1) MOP_IN(OP_BACKREF1);
|
---|
2549 | mem = 1;
|
---|
2550 | goto backref;
|
---|
2551 | NEXT;
|
---|
2552 |
|
---|
2553 | CASE(OP_BACKREF2) MOP_IN(OP_BACKREF2);
|
---|
2554 | mem = 2;
|
---|
2555 | goto backref;
|
---|
2556 | NEXT;
|
---|
2557 |
|
---|
2558 | CASE(OP_BACKREFN) MOP_IN(OP_BACKREFN);
|
---|
2559 | GET_MEMNUM_INC(mem, p);
|
---|
2560 | backref:
|
---|
2561 | {
|
---|
2562 | int len;
|
---|
2563 | UChar *pstart, *pend;
|
---|
2564 |
|
---|
2565 | /* if you want to remove following line,
|
---|
2566 | you should check in parse and compile time. */
|
---|
2567 | if (mem > num_mem) goto fail;
|
---|
2568 | if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
---|
2569 | if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
---|
2570 |
|
---|
2571 | if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
---|
2572 | pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
---|
2573 | else
|
---|
2574 | pstart = (UChar* )((void* )mem_start_stk[mem]);
|
---|
2575 |
|
---|
2576 | pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
---|
2577 | ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
---|
2578 | : (UChar* )((void* )mem_end_stk[mem]));
|
---|
2579 | n = pend - pstart;
|
---|
2580 | DATA_ENSURE(n);
|
---|
2581 | sprev = s;
|
---|
2582 | STRING_CMP(pstart, s, n);
|
---|
2583 | while (sprev + (len = enclen(encode, sprev, end)) < s)
|
---|
2584 | sprev += len;
|
---|
2585 |
|
---|
2586 | MOP_OUT;
|
---|
2587 | JUMP;
|
---|
2588 | }
|
---|
2589 |
|
---|
2590 | CASE(OP_BACKREFN_IC) MOP_IN(OP_BACKREFN_IC);
|
---|
2591 | GET_MEMNUM_INC(mem, p);
|
---|
2592 | {
|
---|
2593 | int len;
|
---|
2594 | UChar *pstart, *pend;
|
---|
2595 |
|
---|
2596 | /* if you want to remove following line,
|
---|
2597 | you should check in parse and compile time. */
|
---|
2598 | if (mem > num_mem) goto fail;
|
---|
2599 | if (mem_end_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
---|
2600 | if (mem_start_stk[mem] == INVALID_STACK_INDEX) goto fail;
|
---|
2601 |
|
---|
2602 | if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
---|
2603 | pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
---|
2604 | else
|
---|
2605 | pstart = (UChar* )((void* )mem_start_stk[mem]);
|
---|
2606 |
|
---|
2607 | pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
---|
2608 | ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
---|
2609 | : (UChar* )((void* )mem_end_stk[mem]));
|
---|
2610 | n = pend - pstart;
|
---|
2611 | DATA_ENSURE(n);
|
---|
2612 | sprev = s;
|
---|
2613 | STRING_CMP_IC(case_fold_flag, pstart, &s, (int)n, end);
|
---|
2614 | while (sprev + (len = enclen(encode, sprev, end)) < s)
|
---|
2615 | sprev += len;
|
---|
2616 |
|
---|
2617 | MOP_OUT;
|
---|
2618 | JUMP;
|
---|
2619 | }
|
---|
2620 | NEXT;
|
---|
2621 |
|
---|
2622 | CASE(OP_BACKREF_MULTI) MOP_IN(OP_BACKREF_MULTI);
|
---|
2623 | {
|
---|
2624 | int len, is_fail;
|
---|
2625 | UChar *pstart, *pend, *swork;
|
---|
2626 |
|
---|
2627 | GET_LENGTH_INC(tlen, p);
|
---|
2628 | for (i = 0; i < tlen; i++) {
|
---|
2629 | GET_MEMNUM_INC(mem, p);
|
---|
2630 |
|
---|
2631 | if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue;
|
---|
2632 | if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue;
|
---|
2633 |
|
---|
2634 | if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
---|
2635 | pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
---|
2636 | else
|
---|
2637 | pstart = (UChar* )((void* )mem_start_stk[mem]);
|
---|
2638 |
|
---|
2639 | pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
---|
2640 | ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
---|
2641 | : (UChar* )((void* )mem_end_stk[mem]));
|
---|
2642 | n = pend - pstart;
|
---|
2643 | DATA_ENSURE(n);
|
---|
2644 | sprev = s;
|
---|
2645 | swork = s;
|
---|
2646 | STRING_CMP_VALUE(pstart, swork, n, is_fail);
|
---|
2647 | if (is_fail) continue;
|
---|
2648 | s = swork;
|
---|
2649 | while (sprev + (len = enclen(encode, sprev, end)) < s)
|
---|
2650 | sprev += len;
|
---|
2651 |
|
---|
2652 | p += (SIZE_MEMNUM * (tlen - i - 1));
|
---|
2653 | break; /* success */
|
---|
2654 | }
|
---|
2655 | if (i == tlen) goto fail;
|
---|
2656 | MOP_OUT;
|
---|
2657 | JUMP;
|
---|
2658 | }
|
---|
2659 | NEXT;
|
---|
2660 |
|
---|
2661 | CASE(OP_BACKREF_MULTI_IC) MOP_IN(OP_BACKREF_MULTI_IC);
|
---|
2662 | {
|
---|
2663 | int len, is_fail;
|
---|
2664 | UChar *pstart, *pend, *swork;
|
---|
2665 |
|
---|
2666 | GET_LENGTH_INC(tlen, p);
|
---|
2667 | for (i = 0; i < tlen; i++) {
|
---|
2668 | GET_MEMNUM_INC(mem, p);
|
---|
2669 |
|
---|
2670 | if (mem_end_stk[mem] == INVALID_STACK_INDEX) continue;
|
---|
2671 | if (mem_start_stk[mem] == INVALID_STACK_INDEX) continue;
|
---|
2672 |
|
---|
2673 | if (BIT_STATUS_AT(reg->bt_mem_start, mem))
|
---|
2674 | pstart = STACK_AT(mem_start_stk[mem])->u.mem.pstr;
|
---|
2675 | else
|
---|
2676 | pstart = (UChar* )((void* )mem_start_stk[mem]);
|
---|
2677 |
|
---|
2678 | pend = (BIT_STATUS_AT(reg->bt_mem_end, mem)
|
---|
2679 | ? STACK_AT(mem_end_stk[mem])->u.mem.pstr
|
---|
2680 | : (UChar* )((void* )mem_end_stk[mem]));
|
---|
2681 | n = pend - pstart;
|
---|
2682 | DATA_ENSURE(n);
|
---|
2683 | sprev = s;
|
---|
2684 | swork = s;
|
---|
2685 | STRING_CMP_VALUE_IC(case_fold_flag, pstart, &swork, n, end, is_fail);
|
---|
2686 | if (is_fail) continue;
|
---|
2687 | s = swork;
|
---|
2688 | while (sprev + (len = enclen(encode, sprev, end)) < s)
|
---|
2689 | sprev += len;
|
---|
2690 |
|
---|
2691 | p += (SIZE_MEMNUM * (tlen - i - 1));
|
---|
2692 | break; /* success */
|
---|
2693 | }
|
---|
2694 | if (i == tlen) goto fail;
|
---|
2695 | MOP_OUT;
|
---|
2696 | JUMP;
|
---|
2697 | }
|
---|
2698 |
|
---|
2699 | #ifdef USE_BACKREF_WITH_LEVEL
|
---|
2700 | CASE(OP_BACKREF_WITH_LEVEL)
|
---|
2701 | {
|
---|
2702 | int len;
|
---|
2703 | OnigOptionType ic;
|
---|
2704 | LengthType level;
|
---|
2705 |
|
---|
2706 | GET_OPTION_INC(ic, p);
|
---|
2707 | GET_LENGTH_INC(level, p);
|
---|
2708 | GET_LENGTH_INC(tlen, p);
|
---|
2709 |
|
---|
2710 | sprev = s;
|
---|
2711 | if (backref_match_at_nested_level(reg, stk, stk_base, ic,
|
---|
2712 | case_fold_flag, (int )level, (int )tlen, p, &s, end)) {
|
---|
2713 | while (sprev + (len = enclen(encode, sprev, end)) < s)
|
---|
2714 | sprev += len;
|
---|
2715 |
|
---|
2716 | p += (SIZE_MEMNUM * tlen);
|
---|
2717 | }
|
---|
2718 | else
|
---|
2719 | goto fail;
|
---|
2720 |
|
---|
2721 | MOP_OUT;
|
---|
2722 | JUMP;
|
---|
2723 | }
|
---|
2724 |
|
---|
2725 | #endif
|
---|
2726 |
|
---|
2727 | #if 0 /* no need: IS_DYNAMIC_OPTION() == 0 */
|
---|
2728 | CASE(OP_SET_OPTION_PUSH) MOP_IN(OP_SET_OPTION_PUSH);
|
---|
2729 | GET_OPTION_INC(option, p);
|
---|
2730 | STACK_PUSH_ALT(p, s, sprev, pkeep);
|
---|
2731 | p += SIZE_OP_SET_OPTION + SIZE_OP_FAIL;
|
---|
2732 | MOP_OUT;
|
---|
2733 | JUMP;
|
---|
2734 |
|
---|
2735 | CASE(OP_SET_OPTION) MOP_IN(OP_SET_OPTION);
|
---|
2736 | GET_OPTION_INC(option, p);
|
---|
2737 | MOP_OUT;
|
---|
2738 | JUMP;
|
---|
2739 | #endif
|
---|
2740 |
|
---|
2741 | CASE(OP_NULL_CHECK_START) MOP_IN(OP_NULL_CHECK_START);
|
---|
2742 | GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
---|
2743 | STACK_PUSH_NULL_CHECK_START(mem, s);
|
---|
2744 | MOP_OUT;
|
---|
2745 | JUMP;
|
---|
2746 |
|
---|
2747 | CASE(OP_NULL_CHECK_END) MOP_IN(OP_NULL_CHECK_END);
|
---|
2748 | {
|
---|
2749 | int isnull;
|
---|
2750 |
|
---|
2751 | GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
---|
2752 | STACK_NULL_CHECK(isnull, mem, s);
|
---|
2753 | if (isnull) {
|
---|
2754 | #ifdef ONIG_DEBUG_MATCH
|
---|
2755 | fprintf(stderr, "NULL_CHECK_END: skip id:%d, s:%"PRIuPTR" (%p)\n",
|
---|
2756 | (int )mem, (uintptr_t )s, s);
|
---|
2757 | #endif
|
---|
2758 | null_check_found:
|
---|
2759 | /* empty loop founded, skip next instruction */
|
---|
2760 | switch (*p++) {
|
---|
2761 | case OP_JUMP:
|
---|
2762 | case OP_PUSH:
|
---|
2763 | p += SIZE_RELADDR;
|
---|
2764 | break;
|
---|
2765 | case OP_REPEAT_INC:
|
---|
2766 | case OP_REPEAT_INC_NG:
|
---|
2767 | case OP_REPEAT_INC_SG:
|
---|
2768 | case OP_REPEAT_INC_NG_SG:
|
---|
2769 | p += SIZE_MEMNUM;
|
---|
2770 | break;
|
---|
2771 | default:
|
---|
2772 | goto unexpected_bytecode_error;
|
---|
2773 | break;
|
---|
2774 | }
|
---|
2775 | }
|
---|
2776 | }
|
---|
2777 | MOP_OUT;
|
---|
2778 | JUMP;
|
---|
2779 |
|
---|
2780 | #ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT
|
---|
2781 | CASE(OP_NULL_CHECK_END_MEMST) MOP_IN(OP_NULL_CHECK_END_MEMST);
|
---|
2782 | {
|
---|
2783 | int isnull;
|
---|
2784 |
|
---|
2785 | GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
---|
2786 | STACK_NULL_CHECK_MEMST(isnull, mem, s, reg);
|
---|
2787 | if (isnull) {
|
---|
2788 | # ifdef ONIG_DEBUG_MATCH
|
---|
2789 | fprintf(stderr, "NULL_CHECK_END_MEMST: skip id:%d, s:%"PRIuPTR" (%p)\n",
|
---|
2790 | (int )mem, (uintptr_t )s, s);
|
---|
2791 | # endif
|
---|
2792 | if (isnull == -1) goto fail;
|
---|
2793 | goto null_check_found;
|
---|
2794 | }
|
---|
2795 | }
|
---|
2796 | MOP_OUT;
|
---|
2797 | JUMP;
|
---|
2798 | #endif
|
---|
2799 |
|
---|
2800 | #ifdef USE_SUBEXP_CALL
|
---|
2801 | CASE(OP_NULL_CHECK_END_MEMST_PUSH)
|
---|
2802 | MOP_IN(OP_NULL_CHECK_END_MEMST_PUSH);
|
---|
2803 | {
|
---|
2804 | int isnull;
|
---|
2805 |
|
---|
2806 | GET_MEMNUM_INC(mem, p); /* mem: null check id */
|
---|
2807 | # ifdef USE_MONOMANIAC_CHECK_CAPTURES_IN_ENDLESS_REPEAT
|
---|
2808 | STACK_NULL_CHECK_MEMST_REC(isnull, mem, s, reg);
|
---|
2809 | # else
|
---|
2810 | STACK_NULL_CHECK_REC(isnull, mem, s);
|
---|
2811 | # endif
|
---|
2812 | if (isnull) {
|
---|
2813 | # ifdef ONIG_DEBUG_MATCH
|
---|
2814 | fprintf(stderr, "NULL_CHECK_END_MEMST_PUSH: skip id:%d, s:%"PRIuPTR" (%p)\n",
|
---|
2815 | (int )mem, (uintptr_t )s, s);
|
---|
2816 | # endif
|
---|
2817 | if (isnull == -1) goto fail;
|
---|
2818 | goto null_check_found;
|
---|
2819 | }
|
---|
2820 | else {
|
---|
2821 | STACK_PUSH_NULL_CHECK_END(mem);
|
---|
2822 | }
|
---|
2823 | }
|
---|
2824 | MOP_OUT;
|
---|
2825 | JUMP;
|
---|
2826 | #endif
|
---|
2827 |
|
---|
2828 | CASE(OP_JUMP) MOP_IN(OP_JUMP);
|
---|
2829 | GET_RELADDR_INC(addr, p);
|
---|
2830 | p += addr;
|
---|
2831 | MOP_OUT;
|
---|
2832 | CHECK_INTERRUPT_IN_MATCH_AT;
|
---|
2833 | JUMP;
|
---|
2834 |
|
---|
2835 | CASE(OP_PUSH) MOP_IN(OP_PUSH);
|
---|
2836 | GET_RELADDR_INC(addr, p);
|
---|
2837 | STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
|
---|
2838 | MOP_OUT;
|
---|
2839 | JUMP;
|
---|
2840 |
|
---|
2841 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
2842 | CASE(OP_STATE_CHECK_PUSH) MOP_IN(OP_STATE_CHECK_PUSH);
|
---|
2843 | GET_STATE_CHECK_NUM_INC(mem, p);
|
---|
2844 | STATE_CHECK_VAL(scv, mem);
|
---|
2845 | if (scv) goto fail;
|
---|
2846 |
|
---|
2847 | GET_RELADDR_INC(addr, p);
|
---|
2848 | STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem, pkeep);
|
---|
2849 | MOP_OUT;
|
---|
2850 | JUMP;
|
---|
2851 |
|
---|
2852 | CASE(OP_STATE_CHECK_PUSH_OR_JUMP) MOP_IN(OP_STATE_CHECK_PUSH_OR_JUMP);
|
---|
2853 | GET_STATE_CHECK_NUM_INC(mem, p);
|
---|
2854 | GET_RELADDR_INC(addr, p);
|
---|
2855 | STATE_CHECK_VAL(scv, mem);
|
---|
2856 | if (scv) {
|
---|
2857 | p += addr;
|
---|
2858 | }
|
---|
2859 | else {
|
---|
2860 | STACK_PUSH_ALT_WITH_STATE_CHECK(p + addr, s, sprev, mem, pkeep);
|
---|
2861 | }
|
---|
2862 | MOP_OUT;
|
---|
2863 | JUMP;
|
---|
2864 |
|
---|
2865 | CASE(OP_STATE_CHECK) MOP_IN(OP_STATE_CHECK);
|
---|
2866 | GET_STATE_CHECK_NUM_INC(mem, p);
|
---|
2867 | STATE_CHECK_VAL(scv, mem);
|
---|
2868 | if (scv) goto fail;
|
---|
2869 |
|
---|
2870 | STACK_PUSH_STATE_CHECK(s, mem);
|
---|
2871 | MOP_OUT;
|
---|
2872 | JUMP;
|
---|
2873 | #endif /* USE_COMBINATION_EXPLOSION_CHECK */
|
---|
2874 |
|
---|
2875 | CASE(OP_POP) MOP_IN(OP_POP);
|
---|
2876 | STACK_POP_ONE;
|
---|
2877 | MOP_OUT;
|
---|
2878 | JUMP;
|
---|
2879 |
|
---|
2880 | #ifdef USE_OP_PUSH_OR_JUMP_EXACT
|
---|
2881 | CASE(OP_PUSH_OR_JUMP_EXACT1) MOP_IN(OP_PUSH_OR_JUMP_EXACT1);
|
---|
2882 | GET_RELADDR_INC(addr, p);
|
---|
2883 | if (*p == *s && DATA_ENSURE_CHECK1) {
|
---|
2884 | p++;
|
---|
2885 | STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
|
---|
2886 | MOP_OUT;
|
---|
2887 | JUMP;
|
---|
2888 | }
|
---|
2889 | p += (addr + 1);
|
---|
2890 | MOP_OUT;
|
---|
2891 | JUMP;
|
---|
2892 | #endif
|
---|
2893 |
|
---|
2894 | CASE(OP_PUSH_IF_PEEK_NEXT) MOP_IN(OP_PUSH_IF_PEEK_NEXT);
|
---|
2895 | GET_RELADDR_INC(addr, p);
|
---|
2896 | if (*p == *s) {
|
---|
2897 | p++;
|
---|
2898 | STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
|
---|
2899 | MOP_OUT;
|
---|
2900 | JUMP;
|
---|
2901 | }
|
---|
2902 | p++;
|
---|
2903 | MOP_OUT;
|
---|
2904 | JUMP;
|
---|
2905 |
|
---|
2906 | CASE(OP_REPEAT) MOP_IN(OP_REPEAT);
|
---|
2907 | {
|
---|
2908 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
2909 | GET_RELADDR_INC(addr, p);
|
---|
2910 |
|
---|
2911 | STACK_ENSURE(1);
|
---|
2912 | repeat_stk[mem] = GET_STACK_INDEX(stk);
|
---|
2913 | STACK_PUSH_REPEAT(mem, p);
|
---|
2914 |
|
---|
2915 | if (reg->repeat_range[mem].lower == 0) {
|
---|
2916 | STACK_PUSH_ALT(p + addr, s, sprev, pkeep);
|
---|
2917 | }
|
---|
2918 | }
|
---|
2919 | MOP_OUT;
|
---|
2920 | JUMP;
|
---|
2921 |
|
---|
2922 | CASE(OP_REPEAT_NG) MOP_IN(OP_REPEAT_NG);
|
---|
2923 | {
|
---|
2924 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
2925 | GET_RELADDR_INC(addr, p);
|
---|
2926 |
|
---|
2927 | STACK_ENSURE(1);
|
---|
2928 | repeat_stk[mem] = GET_STACK_INDEX(stk);
|
---|
2929 | STACK_PUSH_REPEAT(mem, p);
|
---|
2930 |
|
---|
2931 | if (reg->repeat_range[mem].lower == 0) {
|
---|
2932 | STACK_PUSH_ALT(p, s, sprev, pkeep);
|
---|
2933 | p += addr;
|
---|
2934 | }
|
---|
2935 | }
|
---|
2936 | MOP_OUT;
|
---|
2937 | JUMP;
|
---|
2938 |
|
---|
2939 | CASE(OP_REPEAT_INC) MOP_IN(OP_REPEAT_INC);
|
---|
2940 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
2941 | si = repeat_stk[mem];
|
---|
2942 | stkp = STACK_AT(si);
|
---|
2943 |
|
---|
2944 | repeat_inc:
|
---|
2945 | stkp->u.repeat.count++;
|
---|
2946 | if (stkp->u.repeat.count >= reg->repeat_range[mem].upper) {
|
---|
2947 | /* end of repeat. Nothing to do. */
|
---|
2948 | }
|
---|
2949 | else if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) {
|
---|
2950 | STACK_PUSH_ALT(p, s, sprev, pkeep);
|
---|
2951 | p = STACK_AT(si)->u.repeat.pcode; /* Don't use stkp after PUSH. */
|
---|
2952 | }
|
---|
2953 | else {
|
---|
2954 | p = stkp->u.repeat.pcode;
|
---|
2955 | }
|
---|
2956 | STACK_PUSH_REPEAT_INC(si);
|
---|
2957 | MOP_OUT;
|
---|
2958 | CHECK_INTERRUPT_IN_MATCH_AT;
|
---|
2959 | JUMP;
|
---|
2960 |
|
---|
2961 | CASE(OP_REPEAT_INC_SG) MOP_IN(OP_REPEAT_INC_SG);
|
---|
2962 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
2963 | STACK_GET_REPEAT(mem, stkp);
|
---|
2964 | si = GET_STACK_INDEX(stkp);
|
---|
2965 | goto repeat_inc;
|
---|
2966 | NEXT;
|
---|
2967 |
|
---|
2968 | CASE(OP_REPEAT_INC_NG) MOP_IN(OP_REPEAT_INC_NG);
|
---|
2969 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
2970 | si = repeat_stk[mem];
|
---|
2971 | stkp = STACK_AT(si);
|
---|
2972 |
|
---|
2973 | repeat_inc_ng:
|
---|
2974 | stkp->u.repeat.count++;
|
---|
2975 | if (stkp->u.repeat.count < reg->repeat_range[mem].upper) {
|
---|
2976 | if (stkp->u.repeat.count >= reg->repeat_range[mem].lower) {
|
---|
2977 | UChar* pcode = stkp->u.repeat.pcode;
|
---|
2978 |
|
---|
2979 | STACK_PUSH_REPEAT_INC(si);
|
---|
2980 | STACK_PUSH_ALT(pcode, s, sprev, pkeep);
|
---|
2981 | }
|
---|
2982 | else {
|
---|
2983 | p = stkp->u.repeat.pcode;
|
---|
2984 | STACK_PUSH_REPEAT_INC(si);
|
---|
2985 | }
|
---|
2986 | }
|
---|
2987 | else if (stkp->u.repeat.count == reg->repeat_range[mem].upper) {
|
---|
2988 | STACK_PUSH_REPEAT_INC(si);
|
---|
2989 | }
|
---|
2990 | MOP_OUT;
|
---|
2991 | CHECK_INTERRUPT_IN_MATCH_AT;
|
---|
2992 | JUMP;
|
---|
2993 |
|
---|
2994 | CASE(OP_REPEAT_INC_NG_SG) MOP_IN(OP_REPEAT_INC_NG_SG);
|
---|
2995 | GET_MEMNUM_INC(mem, p); /* mem: OP_REPEAT ID */
|
---|
2996 | STACK_GET_REPEAT(mem, stkp);
|
---|
2997 | si = GET_STACK_INDEX(stkp);
|
---|
2998 | goto repeat_inc_ng;
|
---|
2999 | NEXT;
|
---|
3000 |
|
---|
3001 | CASE(OP_PUSH_POS) MOP_IN(OP_PUSH_POS);
|
---|
3002 | STACK_PUSH_POS(s, sprev, pkeep);
|
---|
3003 | MOP_OUT;
|
---|
3004 | JUMP;
|
---|
3005 |
|
---|
3006 | CASE(OP_POP_POS) MOP_IN(OP_POP_POS);
|
---|
3007 | {
|
---|
3008 | STACK_POS_END(stkp);
|
---|
3009 | s = stkp->u.state.pstr;
|
---|
3010 | sprev = stkp->u.state.pstr_prev;
|
---|
3011 | }
|
---|
3012 | MOP_OUT;
|
---|
3013 | JUMP;
|
---|
3014 |
|
---|
3015 | CASE(OP_PUSH_POS_NOT) MOP_IN(OP_PUSH_POS_NOT);
|
---|
3016 | GET_RELADDR_INC(addr, p);
|
---|
3017 | STACK_PUSH_POS_NOT(p + addr, s, sprev, pkeep);
|
---|
3018 | MOP_OUT;
|
---|
3019 | JUMP;
|
---|
3020 |
|
---|
3021 | CASE(OP_FAIL_POS) MOP_IN(OP_FAIL_POS);
|
---|
3022 | STACK_POP_TIL_POS_NOT;
|
---|
3023 | goto fail;
|
---|
3024 | NEXT;
|
---|
3025 |
|
---|
3026 | CASE(OP_PUSH_STOP_BT) MOP_IN(OP_PUSH_STOP_BT);
|
---|
3027 | STACK_PUSH_STOP_BT;
|
---|
3028 | MOP_OUT;
|
---|
3029 | JUMP;
|
---|
3030 |
|
---|
3031 | CASE(OP_POP_STOP_BT) MOP_IN(OP_POP_STOP_BT);
|
---|
3032 | STACK_STOP_BT_END;
|
---|
3033 | MOP_OUT;
|
---|
3034 | JUMP;
|
---|
3035 |
|
---|
3036 | CASE(OP_LOOK_BEHIND) MOP_IN(OP_LOOK_BEHIND);
|
---|
3037 | GET_LENGTH_INC(tlen, p);
|
---|
3038 | s = (UChar* )ONIGENC_STEP_BACK(encode, str, s, end, (int )tlen);
|
---|
3039 | if (IS_NULL(s)) goto fail;
|
---|
3040 | sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s, end);
|
---|
3041 | MOP_OUT;
|
---|
3042 | JUMP;
|
---|
3043 |
|
---|
3044 | CASE(OP_PUSH_LOOK_BEHIND_NOT) MOP_IN(OP_PUSH_LOOK_BEHIND_NOT);
|
---|
3045 | GET_RELADDR_INC(addr, p);
|
---|
3046 | GET_LENGTH_INC(tlen, p);
|
---|
3047 | q = (UChar* )ONIGENC_STEP_BACK(encode, str, s, end, (int )tlen);
|
---|
3048 | if (IS_NULL(q)) {
|
---|
3049 | /* too short case -> success. ex. /(?<!XXX)a/.match("a")
|
---|
3050 | If you want to change to fail, replace following line. */
|
---|
3051 | p += addr;
|
---|
3052 | /* goto fail; */
|
---|
3053 | }
|
---|
3054 | else {
|
---|
3055 | STACK_PUSH_LOOK_BEHIND_NOT(p + addr, s, sprev, pkeep);
|
---|
3056 | s = q;
|
---|
3057 | sprev = (UChar* )onigenc_get_prev_char_head(encode, str, s, end);
|
---|
3058 | }
|
---|
3059 | MOP_OUT;
|
---|
3060 | JUMP;
|
---|
3061 |
|
---|
3062 | CASE(OP_FAIL_LOOK_BEHIND_NOT) MOP_IN(OP_FAIL_LOOK_BEHIND_NOT);
|
---|
3063 | STACK_POP_TIL_LOOK_BEHIND_NOT;
|
---|
3064 | goto fail;
|
---|
3065 | NEXT;
|
---|
3066 |
|
---|
3067 | CASE(OP_PUSH_ABSENT_POS) MOP_IN(OP_PUSH_ABSENT_POS);
|
---|
3068 | /* Save the absent-start-pos and the original end-pos. */
|
---|
3069 | STACK_PUSH_ABSENT_POS(s, ABSENT_END_POS);
|
---|
3070 | MOP_OUT;
|
---|
3071 | JUMP;
|
---|
3072 |
|
---|
3073 | CASE(OP_ABSENT) MOP_IN(OP_ABSENT);
|
---|
3074 | {
|
---|
3075 | const UChar* aend = ABSENT_END_POS;
|
---|
3076 | UChar* absent;
|
---|
3077 | UChar* selfp = p - 1;
|
---|
3078 |
|
---|
3079 | STACK_POP_ABSENT_POS(absent, ABSENT_END_POS); /* Restore end-pos. */
|
---|
3080 | GET_RELADDR_INC(addr, p);
|
---|
3081 | #ifdef ONIG_DEBUG_MATCH
|
---|
3082 | fprintf(stderr, "ABSENT: s:%p, end:%p, absent:%p, aend:%p\n", s, end, absent, aend);
|
---|
3083 | #endif
|
---|
3084 | if ((absent > aend) && (s > absent)) {
|
---|
3085 | /* An empty match occurred in (?~...) at the start point.
|
---|
3086 | * Never match. */
|
---|
3087 | STACK_POP;
|
---|
3088 | goto fail;
|
---|
3089 | }
|
---|
3090 | else if ((s >= aend) && (s > absent)) {
|
---|
3091 | if (s > aend) {
|
---|
3092 | /* Only one (or less) character matched in the last iteration.
|
---|
3093 | * This is not a possible point. */
|
---|
3094 | goto fail;
|
---|
3095 | }
|
---|
3096 | /* All possible points were found. Try matching after (?~...). */
|
---|
3097 | DATA_ENSURE(0);
|
---|
3098 | p += addr;
|
---|
3099 | }
|
---|
3100 | else {
|
---|
3101 | STACK_PUSH_ALT(p + addr, s, sprev, pkeep); /* Push possible point. */
|
---|
3102 | n = enclen(encode, s, end);
|
---|
3103 | STACK_PUSH_ABSENT_POS(absent, ABSENT_END_POS); /* Save the original pos. */
|
---|
3104 | STACK_PUSH_ALT(selfp, s + n, s, pkeep); /* Next iteration. */
|
---|
3105 | STACK_PUSH_ABSENT;
|
---|
3106 | ABSENT_END_POS = aend;
|
---|
3107 | }
|
---|
3108 | }
|
---|
3109 | MOP_OUT;
|
---|
3110 | JUMP;
|
---|
3111 |
|
---|
3112 | CASE(OP_ABSENT_END) MOP_IN(OP_ABSENT_END);
|
---|
3113 | /* The pattern inside (?~...) was matched.
|
---|
3114 | * Set the end-pos temporary and go to next iteration. */
|
---|
3115 | if (sprev < ABSENT_END_POS)
|
---|
3116 | ABSENT_END_POS = sprev;
|
---|
3117 | #ifdef ONIG_DEBUG_MATCH
|
---|
3118 | fprintf(stderr, "ABSENT_END: end:%p\n", ABSENT_END_POS);
|
---|
3119 | #endif
|
---|
3120 | STACK_POP_TIL_ABSENT;
|
---|
3121 | goto fail;
|
---|
3122 | NEXT;
|
---|
3123 |
|
---|
3124 | #ifdef USE_SUBEXP_CALL
|
---|
3125 | CASE(OP_CALL) MOP_IN(OP_CALL);
|
---|
3126 | GET_ABSADDR_INC(addr, p);
|
---|
3127 | STACK_PUSH_CALL_FRAME(p);
|
---|
3128 | p = reg->p + addr;
|
---|
3129 | MOP_OUT;
|
---|
3130 | JUMP;
|
---|
3131 |
|
---|
3132 | CASE(OP_RETURN) MOP_IN(OP_RETURN);
|
---|
3133 | STACK_RETURN(p);
|
---|
3134 | STACK_PUSH_RETURN;
|
---|
3135 | MOP_OUT;
|
---|
3136 | JUMP;
|
---|
3137 | #endif
|
---|
3138 |
|
---|
3139 | CASE(OP_CONDITION) MOP_IN(OP_CONDITION);
|
---|
3140 | GET_MEMNUM_INC(mem, p);
|
---|
3141 | GET_RELADDR_INC(addr, p);
|
---|
3142 | if ((mem > num_mem) ||
|
---|
3143 | (mem_end_stk[mem] == INVALID_STACK_INDEX) ||
|
---|
3144 | (mem_start_stk[mem] == INVALID_STACK_INDEX)) {
|
---|
3145 | p += addr;
|
---|
3146 | }
|
---|
3147 | MOP_OUT;
|
---|
3148 | JUMP;
|
---|
3149 |
|
---|
3150 | CASE(OP_FINISH)
|
---|
3151 | goto finish;
|
---|
3152 | NEXT;
|
---|
3153 |
|
---|
3154 | CASE(OP_FAIL)
|
---|
3155 | if (0) {
|
---|
3156 | /* fall */
|
---|
3157 | fail:
|
---|
3158 | MOP_OUT;
|
---|
3159 | }
|
---|
3160 | MOP_IN(OP_FAIL);
|
---|
3161 | STACK_POP;
|
---|
3162 | p = stk->u.state.pcode;
|
---|
3163 | s = stk->u.state.pstr;
|
---|
3164 | sprev = stk->u.state.pstr_prev;
|
---|
3165 | pkeep = stk->u.state.pkeep;
|
---|
3166 |
|
---|
3167 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
3168 | if (stk->u.state.state_check != 0) {
|
---|
3169 | stk->type = STK_STATE_CHECK_MARK;
|
---|
3170 | stk++;
|
---|
3171 | }
|
---|
3172 | #endif
|
---|
3173 |
|
---|
3174 | MOP_OUT;
|
---|
3175 | JUMP;
|
---|
3176 |
|
---|
3177 | DEFAULT
|
---|
3178 | goto bytecode_error;
|
---|
3179 | } VM_LOOP_END
|
---|
3180 |
|
---|
3181 | finish:
|
---|
3182 | STACK_SAVE;
|
---|
3183 | if (xmalloc_base) xfree(xmalloc_base);
|
---|
3184 | return best_len;
|
---|
3185 |
|
---|
3186 | #ifdef ONIG_DEBUG
|
---|
3187 | stack_error:
|
---|
3188 | STACK_SAVE;
|
---|
3189 | if (xmalloc_base) xfree(xmalloc_base);
|
---|
3190 | return ONIGERR_STACK_BUG;
|
---|
3191 | #endif
|
---|
3192 |
|
---|
3193 | bytecode_error:
|
---|
3194 | STACK_SAVE;
|
---|
3195 | if (xmalloc_base) xfree(xmalloc_base);
|
---|
3196 | return ONIGERR_UNDEFINED_BYTECODE;
|
---|
3197 |
|
---|
3198 | unexpected_bytecode_error:
|
---|
3199 | STACK_SAVE;
|
---|
3200 | if (xmalloc_base) xfree(xmalloc_base);
|
---|
3201 | return ONIGERR_UNEXPECTED_BYTECODE;
|
---|
3202 | }
|
---|
3203 |
|
---|
3204 |
|
---|
3205 | static UChar*
|
---|
3206 | slow_search(OnigEncoding enc, UChar* target, UChar* target_end,
|
---|
3207 | const UChar* text, const UChar* text_end, UChar* text_range)
|
---|
3208 | {
|
---|
3209 | UChar *t, *p, *s, *end;
|
---|
3210 |
|
---|
3211 | end = (UChar* )text_end;
|
---|
3212 | end -= target_end - target - 1;
|
---|
3213 | if (end > text_range)
|
---|
3214 | end = text_range;
|
---|
3215 |
|
---|
3216 | s = (UChar* )text;
|
---|
3217 |
|
---|
3218 | if (enc->max_enc_len == enc->min_enc_len) {
|
---|
3219 | int n = enc->max_enc_len;
|
---|
3220 |
|
---|
3221 | while (s < end) {
|
---|
3222 | if (*s == *target) {
|
---|
3223 | p = s + 1;
|
---|
3224 | t = target + 1;
|
---|
3225 | if (target_end == t || memcmp(t, p, target_end - t) == 0)
|
---|
3226 | return s;
|
---|
3227 | }
|
---|
3228 | s += n;
|
---|
3229 | }
|
---|
3230 | return (UChar* )NULL;
|
---|
3231 | }
|
---|
3232 | while (s < end) {
|
---|
3233 | if (*s == *target) {
|
---|
3234 | p = s + 1;
|
---|
3235 | t = target + 1;
|
---|
3236 | if (target_end == t || memcmp(t, p, target_end - t) == 0)
|
---|
3237 | return s;
|
---|
3238 | }
|
---|
3239 | s += enclen(enc, s, text_end);
|
---|
3240 | }
|
---|
3241 |
|
---|
3242 | return (UChar* )NULL;
|
---|
3243 | }
|
---|
3244 |
|
---|
3245 | static int
|
---|
3246 | str_lower_case_match(OnigEncoding enc, int case_fold_flag,
|
---|
3247 | const UChar* t, const UChar* tend,
|
---|
3248 | const UChar* p, const UChar* end)
|
---|
3249 | {
|
---|
3250 | int lowlen;
|
---|
3251 | UChar *q, lowbuf[ONIGENC_MBC_CASE_FOLD_MAXLEN];
|
---|
3252 |
|
---|
3253 | while (t < tend) {
|
---|
3254 | lowlen = ONIGENC_MBC_CASE_FOLD(enc, case_fold_flag, &p, end, lowbuf);
|
---|
3255 | q = lowbuf;
|
---|
3256 | while (lowlen > 0) {
|
---|
3257 | if (*t++ != *q++) return 0;
|
---|
3258 | lowlen--;
|
---|
3259 | }
|
---|
3260 | }
|
---|
3261 |
|
---|
3262 | return 1;
|
---|
3263 | }
|
---|
3264 |
|
---|
3265 | static UChar*
|
---|
3266 | slow_search_ic(OnigEncoding enc, int case_fold_flag,
|
---|
3267 | UChar* target, UChar* target_end,
|
---|
3268 | const UChar* text, const UChar* text_end, UChar* text_range)
|
---|
3269 | {
|
---|
3270 | UChar *s, *end;
|
---|
3271 |
|
---|
3272 | end = (UChar* )text_end;
|
---|
3273 | end -= target_end - target - 1;
|
---|
3274 | if (end > text_range)
|
---|
3275 | end = text_range;
|
---|
3276 |
|
---|
3277 | s = (UChar* )text;
|
---|
3278 |
|
---|
3279 | while (s < end) {
|
---|
3280 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
3281 | s, text_end))
|
---|
3282 | return s;
|
---|
3283 |
|
---|
3284 | s += enclen(enc, s, text_end);
|
---|
3285 | }
|
---|
3286 |
|
---|
3287 | return (UChar* )NULL;
|
---|
3288 | }
|
---|
3289 |
|
---|
3290 | static UChar*
|
---|
3291 | slow_search_backward(OnigEncoding enc, UChar* target, UChar* target_end,
|
---|
3292 | const UChar* text, const UChar* adjust_text,
|
---|
3293 | const UChar* text_end, const UChar* text_start)
|
---|
3294 | {
|
---|
3295 | UChar *t, *p, *s;
|
---|
3296 |
|
---|
3297 | s = (UChar* )text_end;
|
---|
3298 | s -= (target_end - target);
|
---|
3299 | if (s > text_start)
|
---|
3300 | s = (UChar* )text_start;
|
---|
3301 | else
|
---|
3302 | s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, adjust_text, s, text_end);
|
---|
3303 |
|
---|
3304 | while (s >= text) {
|
---|
3305 | if (*s == *target) {
|
---|
3306 | p = s + 1;
|
---|
3307 | t = target + 1;
|
---|
3308 | while (t < target_end) {
|
---|
3309 | if (*t != *p++)
|
---|
3310 | break;
|
---|
3311 | t++;
|
---|
3312 | }
|
---|
3313 | if (t == target_end)
|
---|
3314 | return s;
|
---|
3315 | }
|
---|
3316 | s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s, text_end);
|
---|
3317 | }
|
---|
3318 |
|
---|
3319 | return (UChar* )NULL;
|
---|
3320 | }
|
---|
3321 |
|
---|
3322 | static UChar*
|
---|
3323 | slow_search_backward_ic(OnigEncoding enc, int case_fold_flag,
|
---|
3324 | UChar* target, UChar* target_end,
|
---|
3325 | const UChar* text, const UChar* adjust_text,
|
---|
3326 | const UChar* text_end, const UChar* text_start)
|
---|
3327 | {
|
---|
3328 | UChar *s;
|
---|
3329 |
|
---|
3330 | s = (UChar* )text_end;
|
---|
3331 | s -= (target_end - target);
|
---|
3332 | if (s > text_start)
|
---|
3333 | s = (UChar* )text_start;
|
---|
3334 | else
|
---|
3335 | s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(enc, adjust_text, s, text_end);
|
---|
3336 |
|
---|
3337 | while (s >= text) {
|
---|
3338 | if (str_lower_case_match(enc, case_fold_flag,
|
---|
3339 | target, target_end, s, text_end))
|
---|
3340 | return s;
|
---|
3341 |
|
---|
3342 | s = (UChar* )onigenc_get_prev_char_head(enc, adjust_text, s, text_end);
|
---|
3343 | }
|
---|
3344 |
|
---|
3345 | return (UChar* )NULL;
|
---|
3346 | }
|
---|
3347 |
|
---|
3348 | #ifndef USE_SUNDAY_QUICK_SEARCH
|
---|
3349 | /* Boyer-Moore-Horspool search applied to a multibyte string */
|
---|
3350 | static UChar*
|
---|
3351 | bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
3352 | const UChar* text, const UChar* text_end,
|
---|
3353 | const UChar* text_range)
|
---|
3354 | {
|
---|
3355 | const UChar *s, *se, *t, *p, *end;
|
---|
3356 | const UChar *tail;
|
---|
3357 | ptrdiff_t skip, tlen1;
|
---|
3358 |
|
---|
3359 | # ifdef ONIG_DEBUG_SEARCH
|
---|
3360 | fprintf(stderr, "bm_search_notrev: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n",
|
---|
3361 | (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range);
|
---|
3362 | # endif
|
---|
3363 |
|
---|
3364 | tail = target_end - 1;
|
---|
3365 | tlen1 = tail - target;
|
---|
3366 | end = text_range;
|
---|
3367 | if (end + tlen1 > text_end)
|
---|
3368 | end = text_end - tlen1;
|
---|
3369 |
|
---|
3370 | s = text;
|
---|
3371 |
|
---|
3372 | if (IS_NULL(reg->int_map)) {
|
---|
3373 | while (s < end) {
|
---|
3374 | p = se = s + tlen1;
|
---|
3375 | t = tail;
|
---|
3376 | while (*p == *t) {
|
---|
3377 | if (t == target) return (UChar* )s;
|
---|
3378 | p--; t--;
|
---|
3379 | }
|
---|
3380 | skip = reg->map[*se];
|
---|
3381 | t = s;
|
---|
3382 | do {
|
---|
3383 | s += enclen(reg->enc, s, end);
|
---|
3384 | } while ((s - t) < skip && s < end);
|
---|
3385 | }
|
---|
3386 | }
|
---|
3387 | else {
|
---|
3388 | # if OPT_EXACT_MAXLEN >= ONIG_CHAR_TABLE_SIZE
|
---|
3389 | while (s < end) {
|
---|
3390 | p = se = s + tlen1;
|
---|
3391 | t = tail;
|
---|
3392 | while (*p == *t) {
|
---|
3393 | if (t == target) return (UChar* )s;
|
---|
3394 | p--; t--;
|
---|
3395 | }
|
---|
3396 | skip = reg->int_map[*se];
|
---|
3397 | t = s;
|
---|
3398 | do {
|
---|
3399 | s += enclen(reg->enc, s, end);
|
---|
3400 | } while ((s - t) < skip && s < end);
|
---|
3401 | }
|
---|
3402 | # endif
|
---|
3403 | }
|
---|
3404 |
|
---|
3405 | return (UChar* )NULL;
|
---|
3406 | }
|
---|
3407 |
|
---|
3408 | /* Boyer-Moore-Horspool search */
|
---|
3409 | static UChar*
|
---|
3410 | bm_search(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
3411 | const UChar* text, const UChar* text_end, const UChar* text_range)
|
---|
3412 | {
|
---|
3413 | const UChar *s, *t, *p, *end;
|
---|
3414 | const UChar *tail;
|
---|
3415 |
|
---|
3416 | # ifdef ONIG_DEBUG_SEARCH
|
---|
3417 | fprintf(stderr, "bm_search: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n",
|
---|
3418 | (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range);
|
---|
3419 | # endif
|
---|
3420 |
|
---|
3421 | end = text_range + (target_end - target) - 1;
|
---|
3422 | if (end > text_end)
|
---|
3423 | end = text_end;
|
---|
3424 |
|
---|
3425 | tail = target_end - 1;
|
---|
3426 | s = text + (target_end - target) - 1;
|
---|
3427 | if (IS_NULL(reg->int_map)) {
|
---|
3428 | while (s < end) {
|
---|
3429 | p = s;
|
---|
3430 | t = tail;
|
---|
3431 | # ifdef ONIG_DEBUG_SEARCH
|
---|
3432 | fprintf(stderr, "bm_search_loop: pos: %"PRIdPTR" %s\n",
|
---|
3433 | (intptr_t )(s - text), s);
|
---|
3434 | # endif
|
---|
3435 | while (*p == *t) {
|
---|
3436 | if (t == target) return (UChar* )p;
|
---|
3437 | p--; t--;
|
---|
3438 | }
|
---|
3439 | s += reg->map[*s];
|
---|
3440 | }
|
---|
3441 | }
|
---|
3442 | else { /* see int_map[] */
|
---|
3443 | # if OPT_EXACT_MAXLEN >= ONIG_CHAR_TABLE_SIZE
|
---|
3444 | while (s < end) {
|
---|
3445 | p = s;
|
---|
3446 | t = tail;
|
---|
3447 | while (*p == *t) {
|
---|
3448 | if (t == target) return (UChar* )p;
|
---|
3449 | p--; t--;
|
---|
3450 | }
|
---|
3451 | s += reg->int_map[*s];
|
---|
3452 | }
|
---|
3453 | # endif
|
---|
3454 | }
|
---|
3455 | return (UChar* )NULL;
|
---|
3456 | }
|
---|
3457 |
|
---|
3458 | /* Boyer-Moore-Horspool search applied to a multibyte string (ignore case) */
|
---|
3459 | static UChar*
|
---|
3460 | bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
3461 | const UChar* text, const UChar* text_end,
|
---|
3462 | const UChar* text_range)
|
---|
3463 | {
|
---|
3464 | const UChar *s, *se, *t, *end;
|
---|
3465 | const UChar *tail;
|
---|
3466 | ptrdiff_t skip, tlen1;
|
---|
3467 | OnigEncoding enc = reg->enc;
|
---|
3468 | int case_fold_flag = reg->case_fold_flag;
|
---|
3469 |
|
---|
3470 | # ifdef ONIG_DEBUG_SEARCH
|
---|
3471 | fprintf(stderr, "bm_search_notrev_ic: text: %d (%p), text_end: %d (%p), text_range: %d (%p)\n",
|
---|
3472 | (int )text, text, (int )text_end, text_end, (int )text_range, text_range);
|
---|
3473 | # endif
|
---|
3474 |
|
---|
3475 | tail = target_end - 1;
|
---|
3476 | tlen1 = tail - target;
|
---|
3477 | end = text_range;
|
---|
3478 | if (end + tlen1 > text_end)
|
---|
3479 | end = text_end - tlen1;
|
---|
3480 |
|
---|
3481 | s = text;
|
---|
3482 |
|
---|
3483 | if (IS_NULL(reg->int_map)) {
|
---|
3484 | while (s < end) {
|
---|
3485 | se = s + tlen1;
|
---|
3486 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
3487 | s, se + 1))
|
---|
3488 | return (UChar* )s;
|
---|
3489 | skip = reg->map[*se];
|
---|
3490 | t = s;
|
---|
3491 | do {
|
---|
3492 | s += enclen(reg->enc, s, end);
|
---|
3493 | } while ((s - t) < skip && s < end);
|
---|
3494 | }
|
---|
3495 | }
|
---|
3496 | else {
|
---|
3497 | # if OPT_EXACT_MAXLEN >= ONIG_CHAR_TABLE_SIZE
|
---|
3498 | while (s < end) {
|
---|
3499 | se = s + tlen1;
|
---|
3500 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
3501 | s, se + 1))
|
---|
3502 | return (UChar* )s;
|
---|
3503 | skip = reg->int_map[*se];
|
---|
3504 | t = s;
|
---|
3505 | do {
|
---|
3506 | s += enclen(reg->enc, s, end);
|
---|
3507 | } while ((s - t) < skip && s < end);
|
---|
3508 | }
|
---|
3509 | # endif
|
---|
3510 | }
|
---|
3511 |
|
---|
3512 | return (UChar* )NULL;
|
---|
3513 | }
|
---|
3514 |
|
---|
3515 | /* Boyer-Moore-Horspool search (ignore case) */
|
---|
3516 | static UChar*
|
---|
3517 | bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
3518 | const UChar* text, const UChar* text_end, const UChar* text_range)
|
---|
3519 | {
|
---|
3520 | const UChar *s, *p, *end;
|
---|
3521 | const UChar *tail;
|
---|
3522 | OnigEncoding enc = reg->enc;
|
---|
3523 | int case_fold_flag = reg->case_fold_flag;
|
---|
3524 |
|
---|
3525 | # ifdef ONIG_DEBUG_SEARCH
|
---|
3526 | fprintf(stderr, "bm_search_ic: text: %d (%p), text_end: %d (%p), text_range: %d (%p)\n",
|
---|
3527 | (int )text, text, (int )text_end, text_end, (int )text_range, text_range);
|
---|
3528 | # endif
|
---|
3529 |
|
---|
3530 | end = text_range + (target_end - target) - 1;
|
---|
3531 | if (end > text_end)
|
---|
3532 | end = text_end;
|
---|
3533 |
|
---|
3534 | tail = target_end - 1;
|
---|
3535 | s = text + (target_end - target) - 1;
|
---|
3536 | if (IS_NULL(reg->int_map)) {
|
---|
3537 | while (s < end) {
|
---|
3538 | p = s - (target_end - target) + 1;
|
---|
3539 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
3540 | p, s + 1))
|
---|
3541 | return (UChar* )p;
|
---|
3542 | s += reg->map[*s];
|
---|
3543 | }
|
---|
3544 | }
|
---|
3545 | else { /* see int_map[] */
|
---|
3546 | # if OPT_EXACT_MAXLEN >= ONIG_CHAR_TABLE_SIZE
|
---|
3547 | while (s < end) {
|
---|
3548 | p = s - (target_end - target) + 1;
|
---|
3549 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
3550 | p, s + 1))
|
---|
3551 | return (UChar* )p;
|
---|
3552 | s += reg->int_map[*s];
|
---|
3553 | }
|
---|
3554 | # endif
|
---|
3555 | }
|
---|
3556 | return (UChar* )NULL;
|
---|
3557 | }
|
---|
3558 |
|
---|
3559 | #else /* USE_SUNDAY_QUICK_SEARCH */
|
---|
3560 |
|
---|
3561 | /* Sunday's quick search applied to a multibyte string */
|
---|
3562 | static UChar*
|
---|
3563 | bm_search_notrev(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
3564 | const UChar* text, const UChar* text_end,
|
---|
3565 | const UChar* text_range)
|
---|
3566 | {
|
---|
3567 | const UChar *s, *se, *t, *p, *end;
|
---|
3568 | const UChar *tail;
|
---|
3569 | ptrdiff_t skip, tlen1;
|
---|
3570 | OnigEncoding enc = reg->enc;
|
---|
3571 |
|
---|
3572 | # ifdef ONIG_DEBUG_SEARCH
|
---|
3573 | fprintf(stderr, "bm_search_notrev: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n",
|
---|
3574 | (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range);
|
---|
3575 | # endif
|
---|
3576 |
|
---|
3577 | tail = target_end - 1;
|
---|
3578 | tlen1 = tail - target;
|
---|
3579 | end = text_range;
|
---|
3580 | if (end + tlen1 > text_end)
|
---|
3581 | end = text_end - tlen1;
|
---|
3582 |
|
---|
3583 | s = text;
|
---|
3584 |
|
---|
3585 | if (IS_NULL(reg->int_map)) {
|
---|
3586 | while (s < end) {
|
---|
3587 | p = se = s + tlen1;
|
---|
3588 | t = tail;
|
---|
3589 | while (*p == *t) {
|
---|
3590 | if (t == target) return (UChar* )s;
|
---|
3591 | p--; t--;
|
---|
3592 | }
|
---|
3593 | if (s + 1 >= end) break;
|
---|
3594 | skip = reg->map[se[1]];
|
---|
3595 | t = s;
|
---|
3596 | do {
|
---|
3597 | s += enclen(enc, s, end);
|
---|
3598 | } while ((s - t) < skip && s < end);
|
---|
3599 | }
|
---|
3600 | }
|
---|
3601 | else {
|
---|
3602 | # if OPT_EXACT_MAXLEN >= ONIG_CHAR_TABLE_SIZE
|
---|
3603 | while (s < end) {
|
---|
3604 | p = se = s + tlen1;
|
---|
3605 | t = tail;
|
---|
3606 | while (*p == *t) {
|
---|
3607 | if (t == target) return (UChar* )s;
|
---|
3608 | p--; t--;
|
---|
3609 | }
|
---|
3610 | if (s + 1 >= end) break;
|
---|
3611 | skip = reg->int_map[se[1]];
|
---|
3612 | t = s;
|
---|
3613 | do {
|
---|
3614 | s += enclen(enc, s, end);
|
---|
3615 | } while ((s - t) < skip && s < end);
|
---|
3616 | }
|
---|
3617 | # endif
|
---|
3618 | }
|
---|
3619 |
|
---|
3620 | return (UChar* )NULL;
|
---|
3621 | }
|
---|
3622 |
|
---|
3623 | /* Sunday's quick search */
|
---|
3624 | static UChar*
|
---|
3625 | bm_search(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
3626 | const UChar* text, const UChar* text_end, const UChar* text_range)
|
---|
3627 | {
|
---|
3628 | const UChar *s, *t, *p, *end;
|
---|
3629 | const UChar *tail;
|
---|
3630 | ptrdiff_t tlen1;
|
---|
3631 |
|
---|
3632 | # ifdef ONIG_DEBUG_SEARCH
|
---|
3633 | fprintf(stderr, "bm_search: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n",
|
---|
3634 | (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range);
|
---|
3635 | # endif
|
---|
3636 |
|
---|
3637 | tail = target_end - 1;
|
---|
3638 | tlen1 = tail - target;
|
---|
3639 | end = text_range + tlen1;
|
---|
3640 | if (end > text_end)
|
---|
3641 | end = text_end;
|
---|
3642 |
|
---|
3643 | s = text + tlen1;
|
---|
3644 | if (IS_NULL(reg->int_map)) {
|
---|
3645 | while (s < end) {
|
---|
3646 | p = s;
|
---|
3647 | t = tail;
|
---|
3648 | while (*p == *t) {
|
---|
3649 | if (t == target) return (UChar* )p;
|
---|
3650 | p--; t--;
|
---|
3651 | }
|
---|
3652 | if (s + 1 >= end) break;
|
---|
3653 | s += reg->map[s[1]];
|
---|
3654 | }
|
---|
3655 | }
|
---|
3656 | else { /* see int_map[] */
|
---|
3657 | # if OPT_EXACT_MAXLEN >= ONIG_CHAR_TABLE_SIZE
|
---|
3658 | while (s < end) {
|
---|
3659 | p = s;
|
---|
3660 | t = tail;
|
---|
3661 | while (*p == *t) {
|
---|
3662 | if (t == target) return (UChar* )p;
|
---|
3663 | p--; t--;
|
---|
3664 | }
|
---|
3665 | if (s + 1 >= end) break;
|
---|
3666 | s += reg->int_map[s[1]];
|
---|
3667 | }
|
---|
3668 | # endif
|
---|
3669 | }
|
---|
3670 | return (UChar* )NULL;
|
---|
3671 | }
|
---|
3672 |
|
---|
3673 | /* Sunday's quick search applied to a multibyte string (ignore case) */
|
---|
3674 | static UChar*
|
---|
3675 | bm_search_notrev_ic(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
3676 | const UChar* text, const UChar* text_end,
|
---|
3677 | const UChar* text_range)
|
---|
3678 | {
|
---|
3679 | const UChar *s, *se, *t, *end;
|
---|
3680 | const UChar *tail;
|
---|
3681 | ptrdiff_t skip, tlen1;
|
---|
3682 | OnigEncoding enc = reg->enc;
|
---|
3683 | int case_fold_flag = reg->case_fold_flag;
|
---|
3684 |
|
---|
3685 | # ifdef ONIG_DEBUG_SEARCH
|
---|
3686 | fprintf(stderr, "bm_search_notrev_ic: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n",
|
---|
3687 | (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range);
|
---|
3688 | # endif
|
---|
3689 |
|
---|
3690 | tail = target_end - 1;
|
---|
3691 | tlen1 = tail - target;
|
---|
3692 | end = text_range;
|
---|
3693 | if (end + tlen1 > text_end)
|
---|
3694 | end = text_end - tlen1;
|
---|
3695 |
|
---|
3696 | s = text;
|
---|
3697 |
|
---|
3698 | if (IS_NULL(reg->int_map)) {
|
---|
3699 | while (s < end) {
|
---|
3700 | se = s + tlen1;
|
---|
3701 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
3702 | s, se + 1))
|
---|
3703 | return (UChar* )s;
|
---|
3704 | if (s + 1 >= end) break;
|
---|
3705 | skip = reg->map[se[1]];
|
---|
3706 | t = s;
|
---|
3707 | do {
|
---|
3708 | s += enclen(enc, s, end);
|
---|
3709 | } while ((s - t) < skip && s < end);
|
---|
3710 | }
|
---|
3711 | }
|
---|
3712 | else {
|
---|
3713 | # if OPT_EXACT_MAXLEN >= ONIG_CHAR_TABLE_SIZE
|
---|
3714 | while (s < end) {
|
---|
3715 | se = s + tlen1;
|
---|
3716 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
3717 | s, se + 1))
|
---|
3718 | return (UChar* )s;
|
---|
3719 | if (s + 1 >= end) break;
|
---|
3720 | skip = reg->int_map[se[1]];
|
---|
3721 | t = s;
|
---|
3722 | do {
|
---|
3723 | s += enclen(enc, s, end);
|
---|
3724 | } while ((s - t) < skip && s < end);
|
---|
3725 | }
|
---|
3726 | # endif
|
---|
3727 | }
|
---|
3728 |
|
---|
3729 | return (UChar* )NULL;
|
---|
3730 | }
|
---|
3731 |
|
---|
3732 | /* Sunday's quick search (ignore case) */
|
---|
3733 | static UChar*
|
---|
3734 | bm_search_ic(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
3735 | const UChar* text, const UChar* text_end, const UChar* text_range)
|
---|
3736 | {
|
---|
3737 | const UChar *s, *p, *end;
|
---|
3738 | const UChar *tail;
|
---|
3739 | ptrdiff_t tlen1;
|
---|
3740 | OnigEncoding enc = reg->enc;
|
---|
3741 | int case_fold_flag = reg->case_fold_flag;
|
---|
3742 |
|
---|
3743 | # ifdef ONIG_DEBUG_SEARCH
|
---|
3744 | fprintf(stderr, "bm_search_ic: text: %"PRIuPTR" (%p), text_end: %"PRIuPTR" (%p), text_range: %"PRIuPTR" (%p)\n",
|
---|
3745 | (uintptr_t )text, text, (uintptr_t )text_end, text_end, (uintptr_t )text_range, text_range);
|
---|
3746 | # endif
|
---|
3747 |
|
---|
3748 | tail = target_end - 1;
|
---|
3749 | tlen1 = tail - target;
|
---|
3750 | end = text_range + tlen1;
|
---|
3751 | if (end > text_end)
|
---|
3752 | end = text_end;
|
---|
3753 |
|
---|
3754 | s = text + tlen1;
|
---|
3755 | if (IS_NULL(reg->int_map)) {
|
---|
3756 | while (s < end) {
|
---|
3757 | p = s - tlen1;
|
---|
3758 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
3759 | p, s + 1))
|
---|
3760 | return (UChar* )p;
|
---|
3761 | if (s + 1 >= end) break;
|
---|
3762 | s += reg->map[s[1]];
|
---|
3763 | }
|
---|
3764 | }
|
---|
3765 | else { /* see int_map[] */
|
---|
3766 | # if OPT_EXACT_MAXLEN >= ONIG_CHAR_TABLE_SIZE
|
---|
3767 | while (s < end) {
|
---|
3768 | p = s - tlen1;
|
---|
3769 | if (str_lower_case_match(enc, case_fold_flag, target, target_end,
|
---|
3770 | p, s + 1))
|
---|
3771 | return (UChar* )p;
|
---|
3772 | if (s + 1 >= end) break;
|
---|
3773 | s += reg->int_map[s[1]];
|
---|
3774 | }
|
---|
3775 | # endif
|
---|
3776 | }
|
---|
3777 | return (UChar* )NULL;
|
---|
3778 | }
|
---|
3779 | #endif /* USE_SUNDAY_QUICK_SEARCH */
|
---|
3780 |
|
---|
3781 | #ifdef USE_INT_MAP_BACKWARD
|
---|
3782 | static int
|
---|
3783 | set_bm_backward_skip(UChar* s, UChar* end, OnigEncoding enc ARG_UNUSED,
|
---|
3784 | int** skip)
|
---|
3785 | {
|
---|
3786 | int i, len;
|
---|
3787 |
|
---|
3788 | if (IS_NULL(*skip)) {
|
---|
3789 | *skip = (int* )xmalloc(sizeof(int) * ONIG_CHAR_TABLE_SIZE);
|
---|
3790 | if (IS_NULL(*skip)) return ONIGERR_MEMORY;
|
---|
3791 | }
|
---|
3792 |
|
---|
3793 | len = (int )(end - s);
|
---|
3794 | for (i = 0; i < ONIG_CHAR_TABLE_SIZE; i++)
|
---|
3795 | (*skip)[i] = len;
|
---|
3796 |
|
---|
3797 | for (i = len - 1; i > 0; i--)
|
---|
3798 | (*skip)[s[i]] = i;
|
---|
3799 |
|
---|
3800 | return 0;
|
---|
3801 | }
|
---|
3802 |
|
---|
3803 | static UChar*
|
---|
3804 | bm_search_backward(regex_t* reg, const UChar* target, const UChar* target_end,
|
---|
3805 | const UChar* text, const UChar* adjust_text,
|
---|
3806 | const UChar* text_end, const UChar* text_start)
|
---|
3807 | {
|
---|
3808 | const UChar *s, *t, *p;
|
---|
3809 |
|
---|
3810 | s = text_end - (target_end - target);
|
---|
3811 | if (text_start < s)
|
---|
3812 | s = text_start;
|
---|
3813 | else
|
---|
3814 | s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, adjust_text, s, text_end);
|
---|
3815 |
|
---|
3816 | while (s >= text) {
|
---|
3817 | p = s;
|
---|
3818 | t = target;
|
---|
3819 | while (t < target_end && *p == *t) {
|
---|
3820 | p++; t++;
|
---|
3821 | }
|
---|
3822 | if (t == target_end)
|
---|
3823 | return (UChar* )s;
|
---|
3824 |
|
---|
3825 | s -= reg->int_map_backward[*s];
|
---|
3826 | s = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, adjust_text, s, text_end);
|
---|
3827 | }
|
---|
3828 |
|
---|
3829 | return (UChar* )NULL;
|
---|
3830 | }
|
---|
3831 | #endif
|
---|
3832 |
|
---|
3833 | static UChar*
|
---|
3834 | map_search(OnigEncoding enc, UChar map[],
|
---|
3835 | const UChar* text, const UChar* text_range, const UChar* text_end)
|
---|
3836 | {
|
---|
3837 | const UChar *s = text;
|
---|
3838 |
|
---|
3839 | while (s < text_range) {
|
---|
3840 | if (map[*s]) return (UChar* )s;
|
---|
3841 |
|
---|
3842 | s += enclen(enc, s, text_end);
|
---|
3843 | }
|
---|
3844 | return (UChar* )NULL;
|
---|
3845 | }
|
---|
3846 |
|
---|
3847 | static UChar*
|
---|
3848 | map_search_backward(OnigEncoding enc, UChar map[],
|
---|
3849 | const UChar* text, const UChar* adjust_text,
|
---|
3850 | const UChar* text_start, const UChar* text_end)
|
---|
3851 | {
|
---|
3852 | const UChar *s = text_start;
|
---|
3853 |
|
---|
3854 | while (s >= text) {
|
---|
3855 | if (map[*s]) return (UChar* )s;
|
---|
3856 |
|
---|
3857 | s = onigenc_get_prev_char_head(enc, adjust_text, s, text_end);
|
---|
3858 | }
|
---|
3859 | return (UChar* )NULL;
|
---|
3860 | }
|
---|
3861 |
|
---|
3862 | extern OnigPosition
|
---|
3863 | onig_match(regex_t* reg, const UChar* str, const UChar* end, const UChar* at, OnigRegion* region,
|
---|
3864 | OnigOptionType option)
|
---|
3865 | {
|
---|
3866 | ptrdiff_t r;
|
---|
3867 | UChar *prev;
|
---|
3868 | OnigMatchArg msa;
|
---|
3869 |
|
---|
3870 | MATCH_ARG_INIT(msa, option, region, at, at);
|
---|
3871 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
3872 | {
|
---|
3873 | int offset = at - str;
|
---|
3874 | STATE_CHECK_BUFF_INIT(msa, end - str, offset, reg->num_comb_exp_check);
|
---|
3875 | }
|
---|
3876 | #endif
|
---|
3877 |
|
---|
3878 | if (region) {
|
---|
3879 | r = onig_region_resize_clear(region, reg->num_mem + 1);
|
---|
3880 | }
|
---|
3881 | else
|
---|
3882 | r = 0;
|
---|
3883 |
|
---|
3884 | if (r == 0) {
|
---|
3885 | prev = (UChar* )onigenc_get_prev_char_head(reg->enc, str, at, end);
|
---|
3886 | r = match_at(reg, str, end,
|
---|
3887 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
3888 | end,
|
---|
3889 | #endif
|
---|
3890 | at, prev, &msa);
|
---|
3891 | }
|
---|
3892 |
|
---|
3893 | MATCH_ARG_FREE(msa);
|
---|
3894 | return r;
|
---|
3895 | }
|
---|
3896 |
|
---|
3897 | static int
|
---|
3898 | forward_search_range(regex_t* reg, const UChar* str, const UChar* end, UChar* s,
|
---|
3899 | UChar* range, UChar** low, UChar** high, UChar** low_prev)
|
---|
3900 | {
|
---|
3901 | UChar *p, *pprev = (UChar* )NULL;
|
---|
3902 |
|
---|
3903 | #ifdef ONIG_DEBUG_SEARCH
|
---|
3904 | fprintf(stderr, "forward_search_range: str: %"PRIuPTR" (%p), end: %"PRIuPTR" (%p), s: %"PRIuPTR" (%p), range: %"PRIuPTR" (%p)\n",
|
---|
3905 | (uintptr_t )str, str, (uintptr_t )end, end, (uintptr_t )s, s, (uintptr_t )range, range);
|
---|
3906 | #endif
|
---|
3907 |
|
---|
3908 | p = s;
|
---|
3909 | if (reg->dmin > 0) {
|
---|
3910 | if (ONIGENC_IS_SINGLEBYTE(reg->enc)) {
|
---|
3911 | p += reg->dmin;
|
---|
3912 | }
|
---|
3913 | else {
|
---|
3914 | UChar *q = p + reg->dmin;
|
---|
3915 |
|
---|
3916 | if (q >= end) return 0; /* fail */
|
---|
3917 | while (p < q) p += enclen(reg->enc, p, end);
|
---|
3918 | }
|
---|
3919 | }
|
---|
3920 |
|
---|
3921 | retry:
|
---|
3922 | switch (reg->optimize) {
|
---|
3923 | case ONIG_OPTIMIZE_EXACT:
|
---|
3924 | p = slow_search(reg->enc, reg->exact, reg->exact_end, p, end, range);
|
---|
3925 | break;
|
---|
3926 | case ONIG_OPTIMIZE_EXACT_IC:
|
---|
3927 | p = slow_search_ic(reg->enc, reg->case_fold_flag,
|
---|
3928 | reg->exact, reg->exact_end, p, end, range);
|
---|
3929 | break;
|
---|
3930 |
|
---|
3931 | case ONIG_OPTIMIZE_EXACT_BM:
|
---|
3932 | p = bm_search(reg, reg->exact, reg->exact_end, p, end, range);
|
---|
3933 | break;
|
---|
3934 |
|
---|
3935 | case ONIG_OPTIMIZE_EXACT_BM_NOT_REV:
|
---|
3936 | p = bm_search_notrev(reg, reg->exact, reg->exact_end, p, end, range);
|
---|
3937 | break;
|
---|
3938 |
|
---|
3939 | case ONIG_OPTIMIZE_EXACT_BM_IC:
|
---|
3940 | p = bm_search_ic(reg, reg->exact, reg->exact_end, p, end, range);
|
---|
3941 | break;
|
---|
3942 |
|
---|
3943 | case ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC:
|
---|
3944 | p = bm_search_notrev_ic(reg, reg->exact, reg->exact_end, p, end, range);
|
---|
3945 | break;
|
---|
3946 |
|
---|
3947 | case ONIG_OPTIMIZE_MAP:
|
---|
3948 | p = map_search(reg->enc, reg->map, p, range, end);
|
---|
3949 | break;
|
---|
3950 | }
|
---|
3951 |
|
---|
3952 | if (p && p < range) {
|
---|
3953 | if (p - reg->dmin < s) {
|
---|
3954 | retry_gate:
|
---|
3955 | pprev = p;
|
---|
3956 | p += enclen(reg->enc, p, end);
|
---|
3957 | goto retry;
|
---|
3958 | }
|
---|
3959 |
|
---|
3960 | if (reg->sub_anchor) {
|
---|
3961 | UChar* prev;
|
---|
3962 |
|
---|
3963 | switch (reg->sub_anchor) {
|
---|
3964 | case ANCHOR_BEGIN_LINE:
|
---|
3965 | if (!ON_STR_BEGIN(p)) {
|
---|
3966 | prev = onigenc_get_prev_char_head(reg->enc,
|
---|
3967 | (pprev ? pprev : str), p, end);
|
---|
3968 | if (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0))
|
---|
3969 | goto retry_gate;
|
---|
3970 | }
|
---|
3971 | break;
|
---|
3972 |
|
---|
3973 | case ANCHOR_END_LINE:
|
---|
3974 | if (ON_STR_END(p)) {
|
---|
3975 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
3976 | prev = (UChar* )onigenc_get_prev_char_head(reg->enc,
|
---|
3977 | (pprev ? pprev : str), p);
|
---|
3978 | if (prev && ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 1))
|
---|
3979 | goto retry_gate;
|
---|
3980 | #endif
|
---|
3981 | }
|
---|
3982 | else if (! ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, p, str, end, reg->options, 1))
|
---|
3983 | goto retry_gate;
|
---|
3984 | break;
|
---|
3985 | }
|
---|
3986 | }
|
---|
3987 |
|
---|
3988 | if (reg->dmax == 0) {
|
---|
3989 | *low = p;
|
---|
3990 | if (low_prev) {
|
---|
3991 | if (*low > s)
|
---|
3992 | *low_prev = onigenc_get_prev_char_head(reg->enc, s, p, end);
|
---|
3993 | else
|
---|
3994 | *low_prev = onigenc_get_prev_char_head(reg->enc,
|
---|
3995 | (pprev ? pprev : str), p, end);
|
---|
3996 | }
|
---|
3997 | }
|
---|
3998 | else {
|
---|
3999 | if (reg->dmax != ONIG_INFINITE_DISTANCE) {
|
---|
4000 | if (p < str + reg->dmax) {
|
---|
4001 | *low = (UChar* )str;
|
---|
4002 | if (low_prev)
|
---|
4003 | *low_prev = onigenc_get_prev_char_head(reg->enc, str, *low, end);
|
---|
4004 | }
|
---|
4005 | else {
|
---|
4006 | *low = p - reg->dmax;
|
---|
4007 | if (*low > s) {
|
---|
4008 | *low = onigenc_get_right_adjust_char_head_with_prev(reg->enc, s,
|
---|
4009 | *low, end, (const UChar** )low_prev);
|
---|
4010 | if (low_prev && IS_NULL(*low_prev))
|
---|
4011 | *low_prev = onigenc_get_prev_char_head(reg->enc,
|
---|
4012 | (pprev ? pprev : s), *low, end);
|
---|
4013 | }
|
---|
4014 | else {
|
---|
4015 | if (low_prev)
|
---|
4016 | *low_prev = onigenc_get_prev_char_head(reg->enc,
|
---|
4017 | (pprev ? pprev : str), *low, end);
|
---|
4018 | }
|
---|
4019 | }
|
---|
4020 | }
|
---|
4021 | }
|
---|
4022 | /* no needs to adjust *high, *high is used as range check only */
|
---|
4023 | *high = p - reg->dmin;
|
---|
4024 |
|
---|
4025 | #ifdef ONIG_DEBUG_SEARCH
|
---|
4026 | fprintf(stderr,
|
---|
4027 | "forward_search_range success: low: %"PRIdPTR", high: %"PRIdPTR", dmin: %"PRIdPTR", dmax: %"PRIdPTR"\n",
|
---|
4028 | *low - str, *high - str, reg->dmin, reg->dmax);
|
---|
4029 | #endif
|
---|
4030 | return 1; /* success */
|
---|
4031 | }
|
---|
4032 |
|
---|
4033 | return 0; /* fail */
|
---|
4034 | }
|
---|
4035 |
|
---|
4036 | #define BM_BACKWARD_SEARCH_LENGTH_THRESHOLD 100
|
---|
4037 |
|
---|
4038 | static int
|
---|
4039 | backward_search_range(regex_t* reg, const UChar* str, const UChar* end,
|
---|
4040 | UChar* s, const UChar* range, UChar* adjrange,
|
---|
4041 | UChar** low, UChar** high)
|
---|
4042 | {
|
---|
4043 | UChar *p;
|
---|
4044 |
|
---|
4045 | range += reg->dmin;
|
---|
4046 | p = s;
|
---|
4047 |
|
---|
4048 | retry:
|
---|
4049 | switch (reg->optimize) {
|
---|
4050 | case ONIG_OPTIMIZE_EXACT:
|
---|
4051 | exact_method:
|
---|
4052 | p = slow_search_backward(reg->enc, reg->exact, reg->exact_end,
|
---|
4053 | range, adjrange, end, p);
|
---|
4054 | break;
|
---|
4055 |
|
---|
4056 | case ONIG_OPTIMIZE_EXACT_IC:
|
---|
4057 | case ONIG_OPTIMIZE_EXACT_BM_IC:
|
---|
4058 | case ONIG_OPTIMIZE_EXACT_BM_NOT_REV_IC:
|
---|
4059 | p = slow_search_backward_ic(reg->enc, reg->case_fold_flag,
|
---|
4060 | reg->exact, reg->exact_end,
|
---|
4061 | range, adjrange, end, p);
|
---|
4062 | break;
|
---|
4063 |
|
---|
4064 | case ONIG_OPTIMIZE_EXACT_BM:
|
---|
4065 | case ONIG_OPTIMIZE_EXACT_BM_NOT_REV:
|
---|
4066 | #ifdef USE_INT_MAP_BACKWARD
|
---|
4067 | if (IS_NULL(reg->int_map_backward)) {
|
---|
4068 | int r;
|
---|
4069 | if (s - range < BM_BACKWARD_SEARCH_LENGTH_THRESHOLD)
|
---|
4070 | goto exact_method;
|
---|
4071 |
|
---|
4072 | r = set_bm_backward_skip(reg->exact, reg->exact_end, reg->enc,
|
---|
4073 | &(reg->int_map_backward));
|
---|
4074 | if (r) return r;
|
---|
4075 | }
|
---|
4076 | p = bm_search_backward(reg, reg->exact, reg->exact_end, range, adjrange,
|
---|
4077 | end, p);
|
---|
4078 | #else
|
---|
4079 | goto exact_method;
|
---|
4080 | #endif
|
---|
4081 | break;
|
---|
4082 |
|
---|
4083 | case ONIG_OPTIMIZE_MAP:
|
---|
4084 | p = map_search_backward(reg->enc, reg->map, range, adjrange, p, end);
|
---|
4085 | break;
|
---|
4086 | }
|
---|
4087 |
|
---|
4088 | if (p) {
|
---|
4089 | if (reg->sub_anchor) {
|
---|
4090 | UChar* prev;
|
---|
4091 |
|
---|
4092 | switch (reg->sub_anchor) {
|
---|
4093 | case ANCHOR_BEGIN_LINE:
|
---|
4094 | if (!ON_STR_BEGIN(p)) {
|
---|
4095 | prev = onigenc_get_prev_char_head(reg->enc, str, p, end);
|
---|
4096 | if (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0)) {
|
---|
4097 | p = prev;
|
---|
4098 | goto retry;
|
---|
4099 | }
|
---|
4100 | }
|
---|
4101 | break;
|
---|
4102 |
|
---|
4103 | case ANCHOR_END_LINE:
|
---|
4104 | if (ON_STR_END(p)) {
|
---|
4105 | #ifndef USE_NEWLINE_AT_END_OF_STRING_HAS_EMPTY_LINE
|
---|
4106 | prev = onigenc_get_prev_char_head(reg->enc, adjrange, p);
|
---|
4107 | if (IS_NULL(prev)) goto fail;
|
---|
4108 | if (ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 1)) {
|
---|
4109 | p = prev;
|
---|
4110 | goto retry;
|
---|
4111 | }
|
---|
4112 | #endif
|
---|
4113 | }
|
---|
4114 | else if (! ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, p, str, end, reg->options, 1)) {
|
---|
4115 | p = onigenc_get_prev_char_head(reg->enc, adjrange, p, end);
|
---|
4116 | if (IS_NULL(p)) goto fail;
|
---|
4117 | goto retry;
|
---|
4118 | }
|
---|
4119 | break;
|
---|
4120 | }
|
---|
4121 | }
|
---|
4122 |
|
---|
4123 | /* no needs to adjust *high, *high is used as range check only */
|
---|
4124 | if (reg->dmax != ONIG_INFINITE_DISTANCE) {
|
---|
4125 | *low = p - reg->dmax;
|
---|
4126 | *high = p - reg->dmin;
|
---|
4127 | *high = onigenc_get_right_adjust_char_head(reg->enc, adjrange, *high, end);
|
---|
4128 | }
|
---|
4129 |
|
---|
4130 | #ifdef ONIG_DEBUG_SEARCH
|
---|
4131 | fprintf(stderr, "backward_search_range: low: %d, high: %d\n",
|
---|
4132 | (int )(*low - str), (int )(*high - str));
|
---|
4133 | #endif
|
---|
4134 | return 1; /* success */
|
---|
4135 | }
|
---|
4136 |
|
---|
4137 | fail:
|
---|
4138 | #ifdef ONIG_DEBUG_SEARCH
|
---|
4139 | fprintf(stderr, "backward_search_range: fail.\n");
|
---|
4140 | #endif
|
---|
4141 | return 0; /* fail */
|
---|
4142 | }
|
---|
4143 |
|
---|
4144 |
|
---|
4145 | extern OnigPosition
|
---|
4146 | onig_search(regex_t* reg, const UChar* str, const UChar* end,
|
---|
4147 | const UChar* start, const UChar* range, OnigRegion* region, OnigOptionType option)
|
---|
4148 | {
|
---|
4149 | return onig_search_gpos(reg, str, end, start, start, range, region, option);
|
---|
4150 | }
|
---|
4151 |
|
---|
4152 | extern OnigPosition
|
---|
4153 | onig_search_gpos(regex_t* reg, const UChar* str, const UChar* end,
|
---|
4154 | const UChar* global_pos,
|
---|
4155 | const UChar* start, const UChar* range, OnigRegion* region, OnigOptionType option)
|
---|
4156 | {
|
---|
4157 | ptrdiff_t r;
|
---|
4158 | UChar *s, *prev;
|
---|
4159 | OnigMatchArg msa;
|
---|
4160 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
4161 | const UChar *orig_start = start;
|
---|
4162 | const UChar *orig_range = range;
|
---|
4163 | #endif
|
---|
4164 |
|
---|
4165 | #ifdef ONIG_DEBUG_SEARCH
|
---|
4166 | fprintf(stderr,
|
---|
4167 | "onig_search (entry point): str: %"PRIuPTR" (%p), end: %"PRIuPTR", start: %"PRIuPTR", range: %"PRIuPTR"\n",
|
---|
4168 | (uintptr_t )str, str, end - str, start - str, range - str);
|
---|
4169 | #endif
|
---|
4170 |
|
---|
4171 | if (region) {
|
---|
4172 | r = onig_region_resize_clear(region, reg->num_mem + 1);
|
---|
4173 | if (r) goto finish_no_msa;
|
---|
4174 | }
|
---|
4175 |
|
---|
4176 | if (start > end || start < str) goto mismatch_no_msa;
|
---|
4177 |
|
---|
4178 |
|
---|
4179 | #ifdef USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE
|
---|
4180 | # ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
4181 | # define MATCH_AND_RETURN_CHECK(upper_range) \
|
---|
4182 | r = match_at(reg, str, end, (upper_range), s, prev, &msa); \
|
---|
4183 | if (r != ONIG_MISMATCH) {\
|
---|
4184 | if (r >= 0) {\
|
---|
4185 | if (! IS_FIND_LONGEST(reg->options)) {\
|
---|
4186 | goto match;\
|
---|
4187 | }\
|
---|
4188 | }\
|
---|
4189 | else goto finish; /* error */ \
|
---|
4190 | }
|
---|
4191 | # else
|
---|
4192 | # define MATCH_AND_RETURN_CHECK(upper_range) \
|
---|
4193 | r = match_at(reg, str, end, (upper_range), s, prev, &msa); \
|
---|
4194 | if (r != ONIG_MISMATCH) {\
|
---|
4195 | if (r >= 0) {\
|
---|
4196 | goto match;\
|
---|
4197 | }\
|
---|
4198 | else goto finish; /* error */ \
|
---|
4199 | }
|
---|
4200 | # endif /* USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE */
|
---|
4201 | #else
|
---|
4202 | # ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
4203 | # define MATCH_AND_RETURN_CHECK(none) \
|
---|
4204 | r = match_at(reg, str, end, s, prev, &msa);\
|
---|
4205 | if (r != ONIG_MISMATCH) {\
|
---|
4206 | if (r >= 0) {\
|
---|
4207 | if (! IS_FIND_LONGEST(reg->options)) {\
|
---|
4208 | goto match;\
|
---|
4209 | }\
|
---|
4210 | }\
|
---|
4211 | else goto finish; /* error */ \
|
---|
4212 | }
|
---|
4213 | # else
|
---|
4214 | # define MATCH_AND_RETURN_CHECK(none) \
|
---|
4215 | r = match_at(reg, str, end, s, prev, &msa);\
|
---|
4216 | if (r != ONIG_MISMATCH) {\
|
---|
4217 | if (r >= 0) {\
|
---|
4218 | goto match;\
|
---|
4219 | }\
|
---|
4220 | else goto finish; /* error */ \
|
---|
4221 | }
|
---|
4222 | # endif /* USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE */
|
---|
4223 | #endif /* USE_MATCH_RANGE_MUST_BE_INSIDE_OF_SPECIFIED_RANGE */
|
---|
4224 |
|
---|
4225 |
|
---|
4226 | /* anchor optimize: resume search range */
|
---|
4227 | if (reg->anchor != 0 && str < end) {
|
---|
4228 | UChar *min_semi_end, *max_semi_end;
|
---|
4229 |
|
---|
4230 | if (reg->anchor & ANCHOR_BEGIN_POSITION) {
|
---|
4231 | /* search start-position only */
|
---|
4232 | begin_position:
|
---|
4233 | if (range > start)
|
---|
4234 | {
|
---|
4235 | if (global_pos > start)
|
---|
4236 | {
|
---|
4237 | if (global_pos < range)
|
---|
4238 | range = global_pos + 1;
|
---|
4239 | }
|
---|
4240 | else
|
---|
4241 | range = start + 1;
|
---|
4242 | }
|
---|
4243 | else
|
---|
4244 | range = start;
|
---|
4245 | }
|
---|
4246 | else if (reg->anchor & ANCHOR_BEGIN_BUF) {
|
---|
4247 | /* search str-position only */
|
---|
4248 | if (range > start) {
|
---|
4249 | if (start != str) goto mismatch_no_msa;
|
---|
4250 | range = str + 1;
|
---|
4251 | }
|
---|
4252 | else {
|
---|
4253 | if (range <= str) {
|
---|
4254 | start = str;
|
---|
4255 | range = str;
|
---|
4256 | }
|
---|
4257 | else
|
---|
4258 | goto mismatch_no_msa;
|
---|
4259 | }
|
---|
4260 | }
|
---|
4261 | else if (reg->anchor & ANCHOR_END_BUF) {
|
---|
4262 | min_semi_end = max_semi_end = (UChar* )end;
|
---|
4263 |
|
---|
4264 | end_buf:
|
---|
4265 | if ((OnigDistance )(max_semi_end - str) < reg->anchor_dmin)
|
---|
4266 | goto mismatch_no_msa;
|
---|
4267 |
|
---|
4268 | if (range > start) {
|
---|
4269 | if ((OnigDistance )(min_semi_end - start) > reg->anchor_dmax) {
|
---|
4270 | start = min_semi_end - reg->anchor_dmax;
|
---|
4271 | if (start < end)
|
---|
4272 | start = onigenc_get_right_adjust_char_head(reg->enc, str, start, end);
|
---|
4273 | }
|
---|
4274 | if ((OnigDistance )(max_semi_end - (range - 1)) < reg->anchor_dmin) {
|
---|
4275 | range = max_semi_end - reg->anchor_dmin + 1;
|
---|
4276 | }
|
---|
4277 |
|
---|
4278 | if (start > range) goto mismatch_no_msa;
|
---|
4279 | /* If start == range, match with empty at end.
|
---|
4280 | Backward search is used. */
|
---|
4281 | }
|
---|
4282 | else {
|
---|
4283 | if ((OnigDistance )(min_semi_end - range) > reg->anchor_dmax) {
|
---|
4284 | range = min_semi_end - reg->anchor_dmax;
|
---|
4285 | }
|
---|
4286 | if ((OnigDistance )(max_semi_end - start) < reg->anchor_dmin) {
|
---|
4287 | start = max_semi_end - reg->anchor_dmin;
|
---|
4288 | start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, start, end);
|
---|
4289 | }
|
---|
4290 | if (range > start) goto mismatch_no_msa;
|
---|
4291 | }
|
---|
4292 | }
|
---|
4293 | else if (reg->anchor & ANCHOR_SEMI_END_BUF) {
|
---|
4294 | UChar* pre_end = ONIGENC_STEP_BACK(reg->enc, str, end, end, 1);
|
---|
4295 |
|
---|
4296 | max_semi_end = (UChar* )end;
|
---|
4297 | if (ONIGENC_IS_MBC_NEWLINE(reg->enc, pre_end, end)) {
|
---|
4298 | min_semi_end = pre_end;
|
---|
4299 |
|
---|
4300 | #ifdef USE_CRNL_AS_LINE_TERMINATOR
|
---|
4301 | pre_end = ONIGENC_STEP_BACK(reg->enc, str, pre_end, end, 1);
|
---|
4302 | if (IS_NOT_NULL(pre_end) &&
|
---|
4303 | IS_NEWLINE_CRLF(reg->options) &&
|
---|
4304 | ONIGENC_IS_MBC_CRNL(reg->enc, pre_end, end)) {
|
---|
4305 | min_semi_end = pre_end;
|
---|
4306 | }
|
---|
4307 | #endif
|
---|
4308 | if (min_semi_end > str && start <= min_semi_end) {
|
---|
4309 | goto end_buf;
|
---|
4310 | }
|
---|
4311 | }
|
---|
4312 | else {
|
---|
4313 | min_semi_end = (UChar* )end;
|
---|
4314 | goto end_buf;
|
---|
4315 | }
|
---|
4316 | }
|
---|
4317 | else if ((reg->anchor & ANCHOR_ANYCHAR_STAR_ML)) {
|
---|
4318 | goto begin_position;
|
---|
4319 | }
|
---|
4320 | }
|
---|
4321 | else if (str == end) { /* empty string */
|
---|
4322 | static const UChar address_for_empty_string[] = "";
|
---|
4323 |
|
---|
4324 | #ifdef ONIG_DEBUG_SEARCH
|
---|
4325 | fprintf(stderr, "onig_search: empty string.\n");
|
---|
4326 | #endif
|
---|
4327 |
|
---|
4328 | if (reg->threshold_len == 0) {
|
---|
4329 | start = end = str = address_for_empty_string;
|
---|
4330 | s = (UChar* )start;
|
---|
4331 | prev = (UChar* )NULL;
|
---|
4332 |
|
---|
4333 | MATCH_ARG_INIT(msa, option, region, start, start);
|
---|
4334 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
4335 | msa.state_check_buff = (void* )0;
|
---|
4336 | msa.state_check_buff_size = 0; /* NO NEED, for valgrind */
|
---|
4337 | #endif
|
---|
4338 | MATCH_AND_RETURN_CHECK(end);
|
---|
4339 | goto mismatch;
|
---|
4340 | }
|
---|
4341 | goto mismatch_no_msa;
|
---|
4342 | }
|
---|
4343 |
|
---|
4344 | #ifdef ONIG_DEBUG_SEARCH
|
---|
4345 | fprintf(stderr, "onig_search(apply anchor): end: %d, start: %d, range: %d\n",
|
---|
4346 | (int )(end - str), (int )(start - str), (int )(range - str));
|
---|
4347 | #endif
|
---|
4348 |
|
---|
4349 | MATCH_ARG_INIT(msa, option, region, start, global_pos);
|
---|
4350 | #ifdef USE_COMBINATION_EXPLOSION_CHECK
|
---|
4351 | {
|
---|
4352 | int offset = (MIN(start, range) - str);
|
---|
4353 | STATE_CHECK_BUFF_INIT(msa, end - str, offset, reg->num_comb_exp_check);
|
---|
4354 | }
|
---|
4355 | #endif
|
---|
4356 |
|
---|
4357 | s = (UChar* )start;
|
---|
4358 | if (range > start) { /* forward search */
|
---|
4359 | if (s > str)
|
---|
4360 | prev = onigenc_get_prev_char_head(reg->enc, str, s, end);
|
---|
4361 | else
|
---|
4362 | prev = (UChar* )NULL;
|
---|
4363 |
|
---|
4364 | if (reg->optimize != ONIG_OPTIMIZE_NONE) {
|
---|
4365 | UChar *sch_range, *low, *high, *low_prev;
|
---|
4366 |
|
---|
4367 | sch_range = (UChar* )range;
|
---|
4368 | if (reg->dmax != 0) {
|
---|
4369 | if (reg->dmax == ONIG_INFINITE_DISTANCE)
|
---|
4370 | sch_range = (UChar* )end;
|
---|
4371 | else {
|
---|
4372 | sch_range += reg->dmax;
|
---|
4373 | if (sch_range > end) sch_range = (UChar* )end;
|
---|
4374 | }
|
---|
4375 | }
|
---|
4376 |
|
---|
4377 | if ((end - start) < reg->threshold_len)
|
---|
4378 | goto mismatch;
|
---|
4379 |
|
---|
4380 | if (reg->dmax != ONIG_INFINITE_DISTANCE) {
|
---|
4381 | do {
|
---|
4382 | if (! forward_search_range(reg, str, end, s, sch_range,
|
---|
4383 | &low, &high, &low_prev)) goto mismatch;
|
---|
4384 | if (s < low) {
|
---|
4385 | s = low;
|
---|
4386 | prev = low_prev;
|
---|
4387 | }
|
---|
4388 | while (s <= high) {
|
---|
4389 | MATCH_AND_RETURN_CHECK(orig_range);
|
---|
4390 | prev = s;
|
---|
4391 | s += enclen(reg->enc, s, end);
|
---|
4392 | }
|
---|
4393 | } while (s < range);
|
---|
4394 | goto mismatch;
|
---|
4395 | }
|
---|
4396 | else { /* check only. */
|
---|
4397 | if (! forward_search_range(reg, str, end, s, sch_range,
|
---|
4398 | &low, &high, (UChar** )NULL)) goto mismatch;
|
---|
4399 |
|
---|
4400 | if ((reg->anchor & ANCHOR_ANYCHAR_STAR) != 0) {
|
---|
4401 | do {
|
---|
4402 | MATCH_AND_RETURN_CHECK(orig_range);
|
---|
4403 | prev = s;
|
---|
4404 | s += enclen(reg->enc, s, end);
|
---|
4405 |
|
---|
4406 | if ((reg->anchor & (ANCHOR_LOOK_BEHIND | ANCHOR_PREC_READ_NOT)) == 0) {
|
---|
4407 | while (!ONIGENC_IS_MBC_NEWLINE_EX(reg->enc, prev, str, end, reg->options, 0)
|
---|
4408 | && s < range) {
|
---|
4409 | prev = s;
|
---|
4410 | s += enclen(reg->enc, s, end);
|
---|
4411 | }
|
---|
4412 | }
|
---|
4413 | } while (s < range);
|
---|
4414 | goto mismatch;
|
---|
4415 | }
|
---|
4416 | }
|
---|
4417 | }
|
---|
4418 |
|
---|
4419 | do {
|
---|
4420 | MATCH_AND_RETURN_CHECK(orig_range);
|
---|
4421 | prev = s;
|
---|
4422 | s += enclen(reg->enc, s, end);
|
---|
4423 | } while (s < range);
|
---|
4424 |
|
---|
4425 | if (s == range) { /* because empty match with /$/. */
|
---|
4426 | MATCH_AND_RETURN_CHECK(orig_range);
|
---|
4427 | }
|
---|
4428 | }
|
---|
4429 | else { /* backward search */
|
---|
4430 | if (reg->optimize != ONIG_OPTIMIZE_NONE) {
|
---|
4431 | UChar *low, *high, *adjrange, *sch_start;
|
---|
4432 |
|
---|
4433 | if (range < end)
|
---|
4434 | adjrange = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc, str, range, end);
|
---|
4435 | else
|
---|
4436 | adjrange = (UChar* )end;
|
---|
4437 |
|
---|
4438 | if (reg->dmax != ONIG_INFINITE_DISTANCE &&
|
---|
4439 | (end - range) >= reg->threshold_len) {
|
---|
4440 | do {
|
---|
4441 | sch_start = s + reg->dmax;
|
---|
4442 | if (sch_start > end) sch_start = (UChar* )end;
|
---|
4443 | if (backward_search_range(reg, str, end, sch_start, range, adjrange,
|
---|
4444 | &low, &high) <= 0)
|
---|
4445 | goto mismatch;
|
---|
4446 |
|
---|
4447 | if (s > high)
|
---|
4448 | s = high;
|
---|
4449 |
|
---|
4450 | while (s >= low) {
|
---|
4451 | prev = onigenc_get_prev_char_head(reg->enc, str, s, end);
|
---|
4452 | MATCH_AND_RETURN_CHECK(orig_start);
|
---|
4453 | s = prev;
|
---|
4454 | }
|
---|
4455 | } while (s >= range);
|
---|
4456 | goto mismatch;
|
---|
4457 | }
|
---|
4458 | else { /* check only. */
|
---|
4459 | if ((end - range) < reg->threshold_len) goto mismatch;
|
---|
4460 |
|
---|
4461 | sch_start = s;
|
---|
4462 | if (reg->dmax != 0) {
|
---|
4463 | if (reg->dmax == ONIG_INFINITE_DISTANCE)
|
---|
4464 | sch_start = (UChar* )end;
|
---|
4465 | else {
|
---|
4466 | sch_start += reg->dmax;
|
---|
4467 | if (sch_start > end) sch_start = (UChar* )end;
|
---|
4468 | else
|
---|
4469 | sch_start = ONIGENC_LEFT_ADJUST_CHAR_HEAD(reg->enc,
|
---|
4470 | start, sch_start, end);
|
---|
4471 | }
|
---|
4472 | }
|
---|
4473 | if (backward_search_range(reg, str, end, sch_start, range, adjrange,
|
---|
4474 | &low, &high) <= 0) goto mismatch;
|
---|
4475 | }
|
---|
4476 | }
|
---|
4477 |
|
---|
4478 | do {
|
---|
4479 | prev = onigenc_get_prev_char_head(reg->enc, str, s, end);
|
---|
4480 | MATCH_AND_RETURN_CHECK(orig_start);
|
---|
4481 | s = prev;
|
---|
4482 | } while (s >= range);
|
---|
4483 | }
|
---|
4484 |
|
---|
4485 | mismatch:
|
---|
4486 | #ifdef USE_FIND_LONGEST_SEARCH_ALL_OF_RANGE
|
---|
4487 | if (IS_FIND_LONGEST(reg->options)) {
|
---|
4488 | if (msa.best_len >= 0) {
|
---|
4489 | s = msa.best_s;
|
---|
4490 | goto match;
|
---|
4491 | }
|
---|
4492 | }
|
---|
4493 | #endif
|
---|
4494 | r = ONIG_MISMATCH;
|
---|
4495 |
|
---|
4496 | finish:
|
---|
4497 | MATCH_ARG_FREE(msa);
|
---|
4498 |
|
---|
4499 | /* If result is mismatch and no FIND_NOT_EMPTY option,
|
---|
4500 | then the region is not set in match_at(). */
|
---|
4501 | if (IS_FIND_NOT_EMPTY(reg->options) && region) {
|
---|
4502 | onig_region_clear(region);
|
---|
4503 | }
|
---|
4504 |
|
---|
4505 | #ifdef ONIG_DEBUG
|
---|
4506 | if (r != ONIG_MISMATCH)
|
---|
4507 | fprintf(stderr, "onig_search: error %"PRIdPTRDIFF"\n", r);
|
---|
4508 | #endif
|
---|
4509 | return r;
|
---|
4510 |
|
---|
4511 | mismatch_no_msa:
|
---|
4512 | r = ONIG_MISMATCH;
|
---|
4513 | finish_no_msa:
|
---|
4514 | #ifdef ONIG_DEBUG
|
---|
4515 | if (r != ONIG_MISMATCH)
|
---|
4516 | fprintf(stderr, "onig_search: error %"PRIdPTRDIFF"\n", r);
|
---|
4517 | #endif
|
---|
4518 | return r;
|
---|
4519 |
|
---|
4520 | match:
|
---|
4521 | MATCH_ARG_FREE(msa);
|
---|
4522 | return s - str;
|
---|
4523 | }
|
---|
4524 |
|
---|
4525 | extern OnigPosition
|
---|
4526 | onig_scan(regex_t* reg, const UChar* str, const UChar* end,
|
---|
4527 | OnigRegion* region, OnigOptionType option,
|
---|
4528 | int (*scan_callback)(OnigPosition, OnigPosition, OnigRegion*, void*),
|
---|
4529 | void* callback_arg)
|
---|
4530 | {
|
---|
4531 | OnigPosition r;
|
---|
4532 | OnigPosition n;
|
---|
4533 | int rs;
|
---|
4534 | const UChar* start;
|
---|
4535 |
|
---|
4536 | n = 0;
|
---|
4537 | start = str;
|
---|
4538 | while (1) {
|
---|
4539 | r = onig_search(reg, str, end, start, end, region, option);
|
---|
4540 | if (r >= 0) {
|
---|
4541 | rs = scan_callback(n, r, region, callback_arg);
|
---|
4542 | n++;
|
---|
4543 | if (rs != 0)
|
---|
4544 | return rs;
|
---|
4545 |
|
---|
4546 | if (region->end[0] == start - str) {
|
---|
4547 | if (start >= end) break;
|
---|
4548 | start += enclen(reg->enc, start, end);
|
---|
4549 | }
|
---|
4550 | else
|
---|
4551 | start = str + region->end[0];
|
---|
4552 |
|
---|
4553 | if (start > end)
|
---|
4554 | break;
|
---|
4555 | }
|
---|
4556 | else if (r == ONIG_MISMATCH) {
|
---|
4557 | break;
|
---|
4558 | }
|
---|
4559 | else { /* error */
|
---|
4560 | return r;
|
---|
4561 | }
|
---|
4562 | }
|
---|
4563 |
|
---|
4564 | return n;
|
---|
4565 | }
|
---|
4566 |
|
---|
4567 | extern OnigEncoding
|
---|
4568 | onig_get_encoding(const regex_t* reg)
|
---|
4569 | {
|
---|
4570 | return reg->enc;
|
---|
4571 | }
|
---|
4572 |
|
---|
4573 | extern OnigOptionType
|
---|
4574 | onig_get_options(const regex_t* reg)
|
---|
4575 | {
|
---|
4576 | return reg->options;
|
---|
4577 | }
|
---|
4578 |
|
---|
4579 | extern OnigCaseFoldType
|
---|
4580 | onig_get_case_fold_flag(const regex_t* reg)
|
---|
4581 | {
|
---|
4582 | return reg->case_fold_flag;
|
---|
4583 | }
|
---|
4584 |
|
---|
4585 | extern const OnigSyntaxType*
|
---|
4586 | onig_get_syntax(const regex_t* reg)
|
---|
4587 | {
|
---|
4588 | return reg->syntax;
|
---|
4589 | }
|
---|
4590 |
|
---|
4591 | extern int
|
---|
4592 | onig_number_of_captures(const regex_t* reg)
|
---|
4593 | {
|
---|
4594 | return reg->num_mem;
|
---|
4595 | }
|
---|
4596 |
|
---|
4597 | extern int
|
---|
4598 | onig_number_of_capture_histories(const regex_t* reg)
|
---|
4599 | {
|
---|
4600 | #ifdef USE_CAPTURE_HISTORY
|
---|
4601 | int i, n;
|
---|
4602 |
|
---|
4603 | n = 0;
|
---|
4604 | for (i = 0; i <= ONIG_MAX_CAPTURE_HISTORY_GROUP; i++) {
|
---|
4605 | if (BIT_STATUS_AT(reg->capture_history, i) != 0)
|
---|
4606 | n++;
|
---|
4607 | }
|
---|
4608 | return n;
|
---|
4609 | #else
|
---|
4610 | return 0;
|
---|
4611 | #endif
|
---|
4612 | }
|
---|
4613 |
|
---|
4614 | extern void
|
---|
4615 | onig_copy_encoding(OnigEncodingType *to, OnigEncoding from)
|
---|
4616 | {
|
---|
4617 | *to = *from;
|
---|
4618 | }
|
---|
4619 |
|
---|