Parent Directory
|
Revision Log
|
Patch
revision 630 by ph10, Fri Jul 22 10:00:10 2011 UTC | revision 933 by ph10, Sat Feb 25 12:18:23 2012 UTC | |
---|---|---|
# | Line 6 | Line 6 |
6 | and semantics are as close as possible to those of the Perl 5 language. | and semantics are as close as possible to those of the Perl 5 language. |
7 | ||
8 | Written by Philip Hazel | Written by Philip Hazel |
9 | Copyright (c) 1997-2011 University of Cambridge | Copyright (c) 1997-2012 University of Cambridge |
10 | ||
11 | ----------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
12 | Redistribution and use in source and binary forms, with or without | Redistribution and use in source and binary forms, with or without |
# | Line 82 negative to avoid the external error cod | Line 82 negative to avoid the external error cod |
82 | #define MATCH_SKIP_ARG (-993) | #define MATCH_SKIP_ARG (-993) |
83 | #define MATCH_THEN (-992) | #define MATCH_THEN (-992) |
84 | ||
/* This is a convenience macro for code that occurs many times. */ | ||
#define MRRETURN(ra) \ | ||
{ \ | ||
md->mark = markptr; \ | ||
RRETURN(ra); \ | ||
} | ||
85 | /* Maximum number of ints of offset to save on the stack for recursive calls. | /* Maximum number of ints of offset to save on the stack for recursive calls. |
86 | If the offset vector is bigger, malloc is used. This should be a multiple of 3, | If the offset vector is bigger, malloc is used. This should be a multiple of 3, |
87 | because the offset vector is always a multiple of 3 long. */ | because the offset vector is always a multiple of 3 long. */ |
# | Line 121 Returns: nothing | Line 113 Returns: nothing |
113 | */ | */ |
114 | ||
115 | static void | static void |
116 | pchars(const uschar *p, int length, BOOL is_subject, match_data *md) | pchars(const pcre_uchar *p, int length, BOOL is_subject, match_data *md) |
117 | { | { |
118 | unsigned int c; | unsigned int c; |
119 | if (is_subject && length > md->end_subject - p) length = md->end_subject - p; | if (is_subject && length > md->end_subject - p) length = md->end_subject - p; |
# | Line 148 Arguments: | Line 140 Arguments: |
140 | md points to match data block | md points to match data block |
141 | caseless TRUE if caseless | caseless TRUE if caseless |
142 | ||
143 | Returns: < 0 if not matched, otherwise the number of subject bytes matched | Returns: >= 0 the number of subject bytes matched |
144 | -1 no match | |
145 | -2 partial match; always given if at end subject | |
146 | */ | */ |
147 | ||
148 | static int | static int |
149 | match_ref(int offset, register USPTR eptr, int length, match_data *md, | match_ref(int offset, register PCRE_PUCHAR eptr, int length, match_data *md, |
150 | BOOL caseless) | BOOL caseless) |
151 | { | { |
152 | USPTR eptr_start = eptr; | PCRE_PUCHAR eptr_start = eptr; |
153 | register USPTR p = md->start_subject + md->offset_vector[offset]; | register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset]; |
154 | ||
155 | #ifdef PCRE_DEBUG | #ifdef PCRE_DEBUG |
156 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
# | Line 171 pchars(p, length, FALSE, md); | Line 165 pchars(p, length, FALSE, md); |
165 | printf("\n"); | printf("\n"); |
166 | #endif | #endif |
167 | ||
168 | /* Always fail if reference not set (and not JavaScript compatible). */ | /* Always fail if reference not set (and not JavaScript compatible - in that |
169 | case the length is passed as zero). */ | |
170 | ||
171 | if (length < 0) return -1; | if (length < 0) return -1; |
172 | ||
# | Line 181 ASCII characters. */ | Line 176 ASCII characters. */ |
176 | ||
177 | if (caseless) | if (caseless) |
178 | { | { |
179 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
180 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
181 | if (md->utf8) | if (md->utf) |
182 | { | { |
183 | /* Match characters up to the end of the reference. NOTE: the number of | /* Match characters up to the end of the reference. NOTE: the number of |
184 | bytes matched may differ, because there are some characters whose upper and | bytes matched may differ, because there are some characters whose upper and |
# | Line 193 if (caseless) | Line 188 if (caseless) |
188 | the latter. It is important, therefore, to check the length along the | the latter. It is important, therefore, to check the length along the |
189 | reference, not along the subject (earlier code did this wrong). */ | reference, not along the subject (earlier code did this wrong). */ |
190 | ||
191 | USPTR endptr = p + length; | PCRE_PUCHAR endptr = p + length; |
192 | while (p < endptr) | while (p < endptr) |
193 | { | { |
194 | int c, d; | int c, d; |
195 | if (eptr >= md->end_subject) return -1; | if (eptr >= md->end_subject) return -2; /* Partial match */ |
196 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
197 | GETCHARINC(d, p); | GETCHARINC(d, p); |
198 | if (c != d && c != UCD_OTHERCASE(d)) return -1; | if (c != d && c != UCD_OTHERCASE(d)) return -1; |
# | Line 210 if (caseless) | Line 205 if (caseless) |
205 | /* The same code works when not in UTF-8 mode and in UTF-8 mode when there | /* The same code works when not in UTF-8 mode and in UTF-8 mode when there |
206 | is no UCP support. */ | is no UCP support. */ |
207 | { | { |
if (eptr + length > md->end_subject) return -1; | ||
208 | while (length-- > 0) | while (length-- > 0) |
209 | { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; } | { |
210 | if (eptr >= md->end_subject) return -2; /* Partial match */ | |
211 | if (TABLE_GET(*p, md->lcc, *p) != TABLE_GET(*eptr, md->lcc, *eptr)) return -1; | |
212 | p++; | |
213 | eptr++; | |
214 | } | |
215 | } | } |
216 | } | } |
217 | ||
# | Line 221 are in UTF-8 mode. */ | Line 220 are in UTF-8 mode. */ |
220 | ||
221 | else | else |
222 | { | { |
223 | if (eptr + length > md->end_subject) return -1; | while (length-- > 0) |
224 | while (length-- > 0) if (*p++ != *eptr++) return -1; | { |
225 | if (eptr >= md->end_subject) return -2; /* Partial match */ | |
226 | if (*p++ != *eptr++) return -1; | |
227 | } | |
228 | } | } |
229 | ||
230 | return eptr - eptr_start; | return (int)(eptr - eptr_start); |
231 | } | } |
232 | ||
233 | ||
# | Line 277 enum { RM1=1, RM2, RM3, RM4, RM5, RM | Line 279 enum { RM1=1, RM2, RM3, RM4, RM5, RM |
279 | RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, | RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, |
280 | RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, | RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, |
281 | RM51, RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60, | RM51, RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60, |
282 | RM61, RM62, RM63 }; | RM61, RM62, RM63, RM64, RM65, RM66 }; |
283 | ||
284 | /* These versions of the macros use the stack, as normal. There are debugging | /* These versions of the macros use the stack, as normal. There are debugging |
285 | versions and production versions. Note that the "rw" argument of RMATCH isn't | versions and production versions. Note that the "rw" argument of RMATCH isn't |
# | Line 290 actually used in this definition. */ | Line 292 actually used in this definition. */ |
292 | #define RMATCH(ra,rb,rc,rd,re,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rw) \ |
293 | { \ | { \ |
294 | printf("match() called in line %d\n", __LINE__); \ | printf("match() called in line %d\n", __LINE__); \ |
295 | rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \ | rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \ |
296 | printf("to line %d\n", __LINE__); \ | printf("to line %d\n", __LINE__); \ |
297 | } | } |
298 | #define RRETURN(ra) \ | #define RRETURN(ra) \ |
# | Line 300 actually used in this definition. */ | Line 302 actually used in this definition. */ |
302 | } | } |
303 | #else | #else |
304 | #define RMATCH(ra,rb,rc,rd,re,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rw) \ |
305 | rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1) | rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1) |
306 | #define RRETURN(ra) return ra | #define RRETURN(ra) return ra |
307 | #endif | #endif |
308 | ||
# | Line 315 argument of match(), which never changes | Line 317 argument of match(), which never changes |
317 | ||
318 | #define RMATCH(ra,rb,rc,rd,re,rw)\ | #define RMATCH(ra,rb,rc,rd,re,rw)\ |
319 | {\ | {\ |
320 | heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\ | heapframe *newframe = frame->Xnextframe;\ |
321 | if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\ | if (newframe == NULL)\ |
322 | frame->Xwhere = rw; \ | {\ |
323 | newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\ | |
324 | if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\ | |
325 | newframe->Xnextframe = NULL;\ | |
326 | frame->Xnextframe = newframe;\ | |
327 | }\ | |
328 | frame->Xwhere = rw;\ | |
329 | newframe->Xeptr = ra;\ | newframe->Xeptr = ra;\ |
330 | newframe->Xecode = rb;\ | newframe->Xecode = rb;\ |
331 | newframe->Xmstart = mstart;\ | newframe->Xmstart = mstart;\ |
newframe->Xmarkptr = markptr;\ | ||
332 | newframe->Xoffset_top = rc;\ | newframe->Xoffset_top = rc;\ |
333 | newframe->Xeptrb = re;\ | newframe->Xeptrb = re;\ |
334 | newframe->Xrdepth = frame->Xrdepth + 1;\ | newframe->Xrdepth = frame->Xrdepth + 1;\ |
# | Line 337 argument of match(), which never changes | Line 344 argument of match(), which never changes |
344 | {\ | {\ |
345 | heapframe *oldframe = frame;\ | heapframe *oldframe = frame;\ |
346 | frame = oldframe->Xprevframe;\ | frame = oldframe->Xprevframe;\ |
(pcre_stack_free)(oldframe);\ | ||
347 | if (frame != NULL)\ | if (frame != NULL)\ |
348 | {\ | {\ |
349 | rrc = ra;\ | rrc = ra;\ |
# | Line 351 argument of match(), which never changes | Line 357 argument of match(), which never changes |
357 | ||
358 | typedef struct heapframe { | typedef struct heapframe { |
359 | struct heapframe *Xprevframe; | struct heapframe *Xprevframe; |
360 | struct heapframe *Xnextframe; | |
361 | ||
362 | /* Function arguments that may change */ | /* Function arguments that may change */ |
363 | ||
364 | USPTR Xeptr; | PCRE_PUCHAR Xeptr; |
365 | const uschar *Xecode; | const pcre_uchar *Xecode; |
366 | USPTR Xmstart; | PCRE_PUCHAR Xmstart; |
USPTR Xmarkptr; | ||
367 | int Xoffset_top; | int Xoffset_top; |
368 | eptrblock *Xeptrb; | eptrblock *Xeptrb; |
369 | unsigned int Xrdepth; | unsigned int Xrdepth; |
370 | ||
371 | /* Function local variables */ | /* Function local variables */ |
372 | ||
373 | USPTR Xcallpat; | PCRE_PUCHAR Xcallpat; |
374 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
375 | USPTR Xcharptr; | PCRE_PUCHAR Xcharptr; |
376 | #endif | #endif |
377 | USPTR Xdata; | PCRE_PUCHAR Xdata; |
378 | USPTR Xnext; | PCRE_PUCHAR Xnext; |
379 | USPTR Xpp; | PCRE_PUCHAR Xpp; |
380 | USPTR Xprev; | PCRE_PUCHAR Xprev; |
381 | USPTR Xsaved_eptr; | PCRE_PUCHAR Xsaved_eptr; |
382 | ||
383 | recursion_info Xnew_recursive; | recursion_info Xnew_recursive; |
384 | ||
# | Line 385 typedef struct heapframe { | Line 391 typedef struct heapframe { |
391 | int Xprop_value; | int Xprop_value; |
392 | int Xprop_fail_result; | int Xprop_fail_result; |
393 | int Xoclength; | int Xoclength; |
394 | uschar Xocchars[8]; | pcre_uchar Xocchars[6]; |
395 | #endif | #endif |
396 | ||
397 | int Xcodelink; | int Xcodelink; |
# | Line 427 returns a negative (error) response, the | Line 433 returns a negative (error) response, the |
433 | same response. */ | same response. */ |
434 | ||
435 | /* These macros pack up tests that are used for partial matching, and which | /* These macros pack up tests that are used for partial matching, and which |
436 | appears several times in the code. We set the "hit end" flag if the pointer is | appear several times in the code. We set the "hit end" flag if the pointer is |
437 | at the end of the subject and also past the start of the subject (i.e. | at the end of the subject and also past the start of the subject (i.e. |
438 | something has been matched). For hard partial matching, we then return | something has been matched). For hard partial matching, we then return |
439 | immediately. The second one is used when we already know we are past the end of | immediately. The second one is used when we already know we are past the end of |
# | Line 438 the subject. */ | Line 444 the subject. */ |
444 | eptr > md->start_used_ptr) \ | eptr > md->start_used_ptr) \ |
445 | { \ | { \ |
446 | md->hitend = TRUE; \ | md->hitend = TRUE; \ |
447 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \ | if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \ |
448 | } | } |
449 | ||
450 | #define SCHECK_PARTIAL()\ | #define SCHECK_PARTIAL()\ |
451 | if (md->partial != 0 && eptr > md->start_used_ptr) \ | if (md->partial != 0 && eptr > md->start_used_ptr) \ |
452 | { \ | { \ |
453 | md->hitend = TRUE; \ | md->hitend = TRUE; \ |
454 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \ | if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \ |
455 | } | } |
456 | ||
457 | ||
458 | /* Performance note: It might be tempting to extract commonly used fields from | /* Performance note: It might be tempting to extract commonly used fields from |
459 | the md structure (e.g. utf8, end_subject) into individual variables to improve | the md structure (e.g. utf, end_subject) into individual variables to improve |
460 | performance. Tests using gcc on a SPARC disproved this; in the first case, it | performance. Tests using gcc on a SPARC disproved this; in the first case, it |
461 | made performance worse. | made performance worse. |
462 | ||
# | Line 459 Arguments: | Line 465 Arguments: |
465 | ecode pointer to current position in compiled code | ecode pointer to current position in compiled code |
466 | mstart pointer to the current match start position (can be modified | mstart pointer to the current match start position (can be modified |
467 | by encountering \K) | by encountering \K) |
markptr pointer to the most recent MARK name, or NULL | ||
468 | offset_top current top pointer | offset_top current top pointer |
469 | md pointer to "static" info for the match | md pointer to "static" info for the match |
470 | eptrb pointer to chain of blocks containing eptr at start of | eptrb pointer to chain of blocks containing eptr at start of |
# | Line 474 Returns: MATCH_MATCH if matched | Line 479 Returns: MATCH_MATCH if matched |
479 | */ | */ |
480 | ||
481 | static int | static int |
482 | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart, | match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode, |
483 | const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb, | PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb, |
484 | unsigned int rdepth) | unsigned int rdepth) |
485 | { | { |
486 | /* These variables do not need to be preserved over recursion in this function, | /* These variables do not need to be preserved over recursion in this function, |
# | Line 485 so they can be ordinary variables in all | Line 490 so they can be ordinary variables in all |
490 | register int rrc; /* Returns from recursive calls */ | register int rrc; /* Returns from recursive calls */ |
491 | register int i; /* Used for loops not involving calls to RMATCH() */ | register int i; /* Used for loops not involving calls to RMATCH() */ |
492 | register unsigned int c; /* Character values not kept over RMATCH() calls */ | register unsigned int c; /* Character values not kept over RMATCH() calls */ |
493 | register BOOL utf8; /* Local copy of UTF-8 flag for speed */ | register BOOL utf; /* Local copy of UTF flag for speed */ |
494 | ||
495 | BOOL minimize, possessive; /* Quantifier options */ | BOOL minimize, possessive; /* Quantifier options */ |
496 | BOOL caseless; | BOOL caseless; |
497 | int condcode; | int condcode; |
498 | ||
499 | /* When recursion is not being used, all "local" variables that have to be | /* When recursion is not being used, all "local" variables that have to be |
500 | preserved over calls to RMATCH() are part of a "frame" which is obtained from | preserved over calls to RMATCH() are part of a "frame". We set up the top-level |
501 | heap storage. Set up the top-level frame here; others are obtained from the | frame on the stack here; subsequent instantiations are obtained from the heap |
502 | heap whenever RMATCH() does a "recursion". See the macro definitions above. */ | whenever RMATCH() does a "recursion". See the macro definitions above. Putting |
503 | the top-level on the stack rather than malloc-ing them all gives a performance | |
504 | boost in many cases where there is not much "recursion". */ | |
505 | ||
506 | #ifdef NO_RECURSE | #ifdef NO_RECURSE |
507 | heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe)); | heapframe *frame = (heapframe *)md->match_frames_base; |
if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY); | ||
frame->Xprevframe = NULL; /* Marks the top level */ | ||
508 | ||
509 | /* Copy in the original argument variables */ | /* Copy in the original argument variables */ |
510 | ||
511 | frame->Xeptr = eptr; | frame->Xeptr = eptr; |
512 | frame->Xecode = ecode; | frame->Xecode = ecode; |
513 | frame->Xmstart = mstart; | frame->Xmstart = mstart; |
frame->Xmarkptr = markptr; | ||
514 | frame->Xoffset_top = offset_top; | frame->Xoffset_top = offset_top; |
515 | frame->Xeptrb = eptrb; | frame->Xeptrb = eptrb; |
516 | frame->Xrdepth = rdepth; | frame->Xrdepth = rdepth; |
# | Line 520 HEAP_RECURSE: | Line 524 HEAP_RECURSE: |
524 | #define eptr frame->Xeptr | #define eptr frame->Xeptr |
525 | #define ecode frame->Xecode | #define ecode frame->Xecode |
526 | #define mstart frame->Xmstart | #define mstart frame->Xmstart |
#define markptr frame->Xmarkptr | ||
527 | #define offset_top frame->Xoffset_top | #define offset_top frame->Xoffset_top |
528 | #define eptrb frame->Xeptrb | #define eptrb frame->Xeptrb |
529 | #define rdepth frame->Xrdepth | #define rdepth frame->Xrdepth |
530 | ||
531 | /* Ditto for the local variables */ | /* Ditto for the local variables */ |
532 | ||
533 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
534 | #define charptr frame->Xcharptr | #define charptr frame->Xcharptr |
535 | #endif | #endif |
536 | #define callpat frame->Xcallpat | #define callpat frame->Xcallpat |
# | Line 585 declarations can be cut out in a block. | Line 588 declarations can be cut out in a block. |
588 | below are for variables that do not have to be preserved over a recursive call | below are for variables that do not have to be preserved over a recursive call |
589 | to RMATCH(). */ | to RMATCH(). */ |
590 | ||
591 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
592 | const uschar *charptr; | const pcre_uchar *charptr; |
593 | #endif | #endif |
594 | const uschar *callpat; | const pcre_uchar *callpat; |
595 | const uschar *data; | const pcre_uchar *data; |
596 | const uschar *next; | const pcre_uchar *next; |
597 | USPTR pp; | PCRE_PUCHAR pp; |
598 | const uschar *prev; | const pcre_uchar *prev; |
599 | USPTR saved_eptr; | PCRE_PUCHAR saved_eptr; |
600 | ||
601 | recursion_info new_recursive; | recursion_info new_recursive; |
602 | ||
# | Line 606 int prop_type; | Line 609 int prop_type; |
609 | int prop_value; | int prop_value; |
610 | int prop_fail_result; | int prop_fail_result; |
611 | int oclength; | int oclength; |
612 | uschar occhars[8]; | pcre_uchar occhars[6]; |
613 | #endif | #endif |
614 | ||
615 | int codelink; | int codelink; |
# | Line 622 int save_offset1, save_offset2, save_off | Line 625 int save_offset1, save_offset2, save_off |
625 | int stacksave[REC_STACK_SAVE_MAX]; | int stacksave[REC_STACK_SAVE_MAX]; |
626 | ||
627 | eptrblock newptrb; | eptrblock newptrb; |
628 | ||
629 | /* There is a special fudge for calling match() in a way that causes it to | |
630 | measure the size of its basic stack frame when the stack is being used for | |
631 | recursion. The second argument (ecode) being NULL triggers this behaviour. It | |
632 | cannot normally ever be NULL. The return is the negated value of the frame | |
633 | size. */ | |
634 | ||
635 | if (ecode == NULL) | |
636 | { | |
637 | if (rdepth == 0) | |
638 | return match((PCRE_PUCHAR)&rdepth, NULL, NULL, 0, NULL, NULL, 1); | |
639 | else | |
640 | { | |
641 | int len = (char *)&rdepth - (char *)eptr; | |
642 | return (len > 0)? -len : len; | |
643 | } | |
644 | } | |
645 | #endif /* NO_RECURSE */ | #endif /* NO_RECURSE */ |
646 | ||
647 | /* To save space on the stack and in the heap frame, I have doubled up on some | /* To save space on the stack and in the heap frame, I have doubled up on some |
# | Line 634 the alternative names that are used. */ | Line 654 the alternative names that are used. */ |
654 | #define code_offset codelink | #define code_offset codelink |
655 | #define condassert condition | #define condassert condition |
656 | #define matched_once prev_is_word | #define matched_once prev_is_word |
657 | #define foc number | |
658 | #define save_mark data | |
659 | ||
660 | /* These statements are here to stop the compiler complaining about unitialized | /* These statements are here to stop the compiler complaining about unitialized |
661 | variables. */ | variables. */ |
# | Line 659 defined). However, RMATCH isn't like a f | Line 681 defined). However, RMATCH isn't like a f |
681 | complicated macro. It has to be used in one particular way. This shouldn't, | complicated macro. It has to be used in one particular way. This shouldn't, |
682 | however, impact performance when true recursion is being used. */ | however, impact performance when true recursion is being used. */ |
683 | ||
684 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
685 | utf8 = md->utf8; /* Local copy of the flag */ | utf = md->utf; /* Local copy of the flag */ |
686 | #else | #else |
687 | utf8 = FALSE; | utf = FALSE; |
688 | #endif | #endif |
689 | ||
690 | /* First check that we haven't called match() too many times, or that we | /* First check that we haven't called match() too many times, or that we |
# | Line 701 for (;;) | Line 723 for (;;) |
723 | switch(op) | switch(op) |
724 | { | { |
725 | case OP_MARK: | case OP_MARK: |
726 | markptr = ecode + 2; | md->nomatch_mark = ecode + 2; |
727 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | md->mark = NULL; /* In case previously set by assertion */ |
728 | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, | |
729 | eptrb, RM55); | eptrb, RM55); |
730 | if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && | |
731 | md->mark == NULL) md->mark = ecode + 2; | |
732 | ||
733 | /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an | /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an |
734 | argument, and we must check whether that argument matches this MARK's | argument, and we must check whether that argument matches this MARK's |
# | Line 712 for (;;) | Line 737 for (;;) |
737 | position and return MATCH_SKIP. Otherwise, pass back the return code | position and return MATCH_SKIP. Otherwise, pass back the return code |
738 | unaltered. */ | unaltered. */ |
739 | ||
740 | if (rrc == MATCH_SKIP_ARG && | else if (rrc == MATCH_SKIP_ARG && |
741 | strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0) | STRCMP_UC_UC(ecode + 2, md->start_match_ptr) == 0) |
742 | { | { |
743 | md->start_match_ptr = eptr; | md->start_match_ptr = eptr; |
744 | RRETURN(MATCH_SKIP); | RRETURN(MATCH_SKIP); |
745 | } | } |
if (md->mark == NULL) md->mark = markptr; | ||
746 | RRETURN(rrc); | RRETURN(rrc); |
747 | ||
748 | case OP_FAIL: | case OP_FAIL: |
749 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
750 | ||
751 | /* COMMIT overrides PRUNE, SKIP, and THEN */ | /* COMMIT overrides PRUNE, SKIP, and THEN */ |
752 | ||
753 | case OP_COMMIT: | case OP_COMMIT: |
754 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
755 | eptrb, RM52); | eptrb, RM52); |
756 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && |
757 | rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG && | rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG && |
758 | rrc != MATCH_THEN) | rrc != MATCH_THEN) |
759 | RRETURN(rrc); | RRETURN(rrc); |
760 | MRRETURN(MATCH_COMMIT); | RRETURN(MATCH_COMMIT); |
761 | ||
762 | /* PRUNE overrides THEN */ | /* PRUNE overrides THEN */ |
763 | ||
764 | case OP_PRUNE: | case OP_PRUNE: |
765 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
766 | eptrb, RM51); | eptrb, RM51); |
767 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
768 | MRRETURN(MATCH_PRUNE); | RRETURN(MATCH_PRUNE); |
769 | ||
770 | case OP_PRUNE_ARG: | case OP_PRUNE_ARG: |
771 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | md->nomatch_mark = ecode + 2; |
772 | md->mark = NULL; /* In case previously set by assertion */ | |
773 | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, | |
774 | eptrb, RM56); | eptrb, RM56); |
775 | if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && | |
776 | md->mark == NULL) md->mark = ecode + 2; | |
777 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
md->mark = ecode + 2; | ||
778 | RRETURN(MATCH_PRUNE); | RRETURN(MATCH_PRUNE); |
779 | ||
780 | /* SKIP overrides PRUNE and THEN */ | /* SKIP overrides PRUNE and THEN */ |
781 | ||
782 | case OP_SKIP: | case OP_SKIP: |
783 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
784 | eptrb, RM53); | eptrb, RM53); |
785 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) |
786 | RRETURN(rrc); | RRETURN(rrc); |
787 | md->start_match_ptr = eptr; /* Pass back current position */ | md->start_match_ptr = eptr; /* Pass back current position */ |
788 | MRRETURN(MATCH_SKIP); | RRETURN(MATCH_SKIP); |
789 | ||
790 | /* Note that, for Perl compatibility, SKIP with an argument does NOT set | |
791 | nomatch_mark. There is a flag that disables this opcode when re-matching a | |
792 | pattern that ended with a SKIP for which there was not a matching MARK. */ | |
793 | ||
794 | case OP_SKIP_ARG: | case OP_SKIP_ARG: |
795 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | if (md->ignore_skip_arg) |
796 | { | |
797 | ecode += PRIV(OP_lengths)[*ecode] + ecode[1]; | |
798 | break; | |
799 | } | |
800 | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, | |
801 | eptrb, RM57); | eptrb, RM57); |
802 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) |
803 | RRETURN(rrc); | RRETURN(rrc); |
804 | ||
805 | /* Pass back the current skip name by overloading md->start_match_ptr and | /* Pass back the current skip name by overloading md->start_match_ptr and |
806 | returning the special MATCH_SKIP_ARG return code. This will either be | returning the special MATCH_SKIP_ARG return code. This will either be |
807 | caught by a matching MARK, or get to the top, where it is treated the same | caught by a matching MARK, or get to the top, where it causes a rematch |
808 | as PRUNE. */ | with the md->ignore_skip_arg flag set. */ |
809 | ||
810 | md->start_match_ptr = ecode + 2; | md->start_match_ptr = ecode + 2; |
811 | RRETURN(MATCH_SKIP_ARG); | RRETURN(MATCH_SKIP_ARG); |
812 | ||
813 | /* For THEN (and THEN_ARG) we pass back the address of the bracket or | /* For THEN (and THEN_ARG) we pass back the address of the opcode, so that |
814 | the alt that is at the start of the current branch. This makes it possible | the branch in which it occurs can be determined. Overload the start of |
815 | to skip back past alternatives that precede the THEN within the current | match pointer to do this. */ |
branch. */ | ||
816 | ||
817 | case OP_THEN: | case OP_THEN: |
818 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
819 | eptrb, RM54); | eptrb, RM54); |
820 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
821 | md->start_match_ptr = ecode - GET(ecode, 1); | md->start_match_ptr = ecode; |
822 | MRRETURN(MATCH_THEN); | RRETURN(MATCH_THEN); |
823 | ||
824 | case OP_THEN_ARG: | case OP_THEN_ARG: |
825 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE], | md->nomatch_mark = ecode + 2; |
826 | offset_top, md, eptrb, RM58); | md->mark = NULL; /* In case previously set by assertion */ |
827 | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, | |
828 | md, eptrb, RM58); | |
829 | if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && | |
830 | md->mark == NULL) md->mark = ecode + 2; | |
831 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
832 | md->start_match_ptr = ecode - GET(ecode, 1); | md->start_match_ptr = ecode; |
md->mark = ecode + LINK_SIZE + 2; | ||
833 | RRETURN(MATCH_THEN); | RRETURN(MATCH_THEN); |
834 | ||
835 | /* Handle an atomic group that does not contain any capturing parentheses. | |
836 | This can be handled like an assertion. Prior to 8.13, all atomic groups | |
837 | were handled this way. In 8.13, the code was changed as below for ONCE, so | |
838 | that backups pass through the group and thereby reset captured values. | |
839 | However, this uses a lot more stack, so in 8.20, atomic groups that do not | |
840 | contain any captures generate OP_ONCE_NC, which can be handled in the old, | |
841 | less stack intensive way. | |
842 | ||
843 | Check the alternative branches in turn - the matching won't pass the KET | |
844 | for this kind of subpattern. If any one branch matches, we carry on as at | |
845 | the end of a normal bracket, leaving the subject pointer, but resetting | |
846 | the start-of-match value in case it was changed by \K. */ | |
847 | ||
848 | case OP_ONCE_NC: | |
849 | prev = ecode; | |
850 | saved_eptr = eptr; | |
851 | save_mark = md->mark; | |
852 | do | |
853 | { | |
854 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM64); | |
855 | if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ | |
856 | { | |
857 | mstart = md->start_match_ptr; | |
858 | break; | |
859 | } | |
860 | if (rrc == MATCH_THEN) | |
861 | { | |
862 | next = ecode + GET(ecode,1); | |
863 | if (md->start_match_ptr < next && | |
864 | (*ecode == OP_ALT || *next == OP_ALT)) | |
865 | rrc = MATCH_NOMATCH; | |
866 | } | |
867 | ||
868 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
869 | ecode += GET(ecode,1); | |
870 | md->mark = save_mark; | |
871 | } | |
872 | while (*ecode == OP_ALT); | |
873 | ||
874 | /* If hit the end of the group (which could be repeated), fail */ | |
875 | ||
876 | if (*ecode != OP_ONCE_NC && *ecode != OP_ALT) RRETURN(MATCH_NOMATCH); | |
877 | ||
878 | /* Continue as from after the group, updating the offsets high water | |
879 | mark, since extracts may have been taken. */ | |
880 | ||
881 | do ecode += GET(ecode, 1); while (*ecode == OP_ALT); | |
882 | ||
883 | offset_top = md->end_offset_top; | |
884 | eptr = md->end_match_ptr; | |
885 | ||
886 | /* For a non-repeating ket, just continue at this level. This also | |
887 | happens for a repeating ket if no characters were matched in the group. | |
888 | This is the forcible breaking of infinite loops as implemented in Perl | |
889 | 5.005. */ | |
890 | ||
891 | if (*ecode == OP_KET || eptr == saved_eptr) | |
892 | { | |
893 | ecode += 1+LINK_SIZE; | |
894 | break; | |
895 | } | |
896 | ||
897 | /* The repeating kets try the rest of the pattern or restart from the | |
898 | preceding bracket, in the appropriate order. The second "call" of match() | |
899 | uses tail recursion, to avoid using another stack frame. */ | |
900 | ||
901 | if (*ecode == OP_KETRMIN) | |
902 | { | |
903 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM65); | |
904 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
905 | ecode = prev; | |
906 | goto TAIL_RECURSE; | |
907 | } | |
908 | else /* OP_KETRMAX */ | |
909 | { | |
910 | md->match_function_type = MATCH_CBEGROUP; | |
911 | RMATCH(eptr, prev, offset_top, md, eptrb, RM66); | |
912 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
913 | ecode += 1 + LINK_SIZE; | |
914 | goto TAIL_RECURSE; | |
915 | } | |
916 | /* Control never gets here */ | |
917 | ||
918 | /* Handle a capturing bracket, other than those that are possessive with an | /* Handle a capturing bracket, other than those that are possessive with an |
919 | unlimited repeat. If there is space in the offset vector, save the current | unlimited repeat. If there is space in the offset vector, save the current |
920 | subject position in the working slot at the top of the vector. We mustn't | subject position in the working slot at the top of the vector. We mustn't |
# | Line 827 for (;;) | Line 947 for (;;) |
947 | save_offset2 = md->offset_vector[offset+1]; | save_offset2 = md->offset_vector[offset+1]; |
948 | save_offset3 = md->offset_vector[md->offset_end - number]; | save_offset3 = md->offset_vector[md->offset_end - number]; |
949 | save_capture_last = md->capture_last; | save_capture_last = md->capture_last; |
950 | save_mark = md->mark; | |
951 | ||
952 | DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); | DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); |
953 | md->offset_vector[md->offset_end - number] = | md->offset_vector[md->offset_end - number] = |
# | Line 835 for (;;) | Line 956 for (;;) |
956 | for (;;) | for (;;) |
957 | { | { |
958 | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
959 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
960 | eptrb, RM1); | eptrb, RM1); |
961 | if (rrc == MATCH_ONCE) break; /* Backing up through an atomic group */ | if (rrc == MATCH_ONCE) break; /* Backing up through an atomic group */ |
962 | if (rrc != MATCH_NOMATCH && | |
963 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | /* If we backed up to a THEN, check whether it is within the current |
964 | RRETURN(rrc); | branch by comparing the address of the THEN that is passed back with |
965 | the end of the branch. If it is within the current branch, and the | |
966 | branch is one of two or more alternatives (it either starts or ends | |
967 | with OP_ALT), we have reached the limit of THEN's action, so convert | |
968 | the return code to NOMATCH, which will cause normal backtracking to | |
969 | happen from now on. Otherwise, THEN is passed back to an outer | |
970 | alternative. This implements Perl's treatment of parenthesized groups, | |
971 | where a group not containing | does not affect the current alternative, | |
972 | that is, (X) is NOT the same as (X|(*F)). */ | |
973 | ||
974 | if (rrc == MATCH_THEN) | |
975 | { | |
976 | next = ecode + GET(ecode,1); | |
977 | if (md->start_match_ptr < next && | |
978 | (*ecode == OP_ALT || *next == OP_ALT)) | |
979 | rrc = MATCH_NOMATCH; | |
980 | } | |
981 | ||
982 | /* Anything other than NOMATCH is passed back. */ | |
983 | ||
984 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
985 | md->capture_last = save_capture_last; | md->capture_last = save_capture_last; |
986 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
987 | md->mark = save_mark; | |
988 | if (*ecode != OP_ALT) break; | if (*ecode != OP_ALT) break; |
989 | } | } |
990 | ||
# | Line 851 for (;;) | Line 993 for (;;) |
993 | md->offset_vector[offset+1] = save_offset2; | md->offset_vector[offset+1] = save_offset2; |
994 | md->offset_vector[md->offset_end - number] = save_offset3; | md->offset_vector[md->offset_end - number] = save_offset3; |
995 | ||
996 | /* At this point, rrc will be one of MATCH_ONCE, MATCH_NOMATCH, or | /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */ |
MATCH_THEN. */ | ||
997 | ||
998 | if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr; | RRETURN(rrc); |
RRETURN(((rrc == MATCH_ONCE)? MATCH_ONCE:MATCH_NOMATCH)); | ||
999 | } | } |
1000 | ||
1001 | /* FALL THROUGH ... Insufficient room for saving captured contents. Treat | /* FALL THROUGH ... Insufficient room for saving captured contents. Treat |
# | Line 870 for (;;) | Line 1010 for (;;) |
1010 | /* VVVVVVVVVVVVVVVVVVVVVVVVV */ | /* VVVVVVVVVVVVVVVVVVVVVVVVV */ |
1011 | ||
1012 | /* Non-capturing or atomic group, except for possessive with unlimited | /* Non-capturing or atomic group, except for possessive with unlimited |
1013 | repeat. Loop for all the alternatives. When we get to the final alternative | repeat and ONCE group with no captures. Loop for all the alternatives. |
1014 | within the brackets, we used to return the result of a recursive call to | |
1015 | match() whatever happened so it was possible to reduce stack usage by | When we get to the final alternative within the brackets, we used to return |
1016 | turning this into a tail recursion, except in the case of a possibly empty | the result of a recursive call to match() whatever happened so it was |
1017 | group. However, now that there is the possiblity of (*THEN) occurring in | possible to reduce stack usage by turning this into a tail recursion, |
1018 | the final alternative, this optimization is no longer possible. | except in the case of a possibly empty group. However, now that there is |
1019 | the possiblity of (*THEN) occurring in the final alternative, this | |
1020 | optimization is no longer always possible. | |
1021 | ||
1022 | We can optimize if we know there are no (*THEN)s in the pattern; at present | |
1023 | this is the best that can be done. | |
1024 | ||
1025 | MATCH_ONCE is returned when the end of an atomic group is successfully | MATCH_ONCE is returned when the end of an atomic group is successfully |
1026 | reached, but subsequent matching fails. It passes back up the tree (causing | reached, but subsequent matching fails. It passes back up the tree (causing |
# | Line 892 for (;;) | Line 1037 for (;;) |
1037 | for (;;) | for (;;) |
1038 | { | { |
1039 | if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP; | if (op >= OP_SBRA || op == OP_ONCE) md->match_function_type = MATCH_CBEGROUP; |
1040 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb, | |
1041 | /* If this is not a possibly empty group, and there are no (*THEN)s in | |
1042 | the pattern, and this is the final alternative, optimize as described | |
1043 | above. */ | |
1044 | ||
1045 | else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT) | |
1046 | { | |
1047 | ecode += PRIV(OP_lengths)[*ecode]; | |
1048 | goto TAIL_RECURSE; | |
1049 | } | |
1050 | ||
1051 | /* In all other cases, we have to make another call to match(). */ | |
1052 | ||
1053 | save_mark = md->mark; | |
1054 | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb, | |
1055 | RM2); | RM2); |
1056 | if (rrc != MATCH_NOMATCH && | |
1057 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | /* See comment in the code for capturing groups above about handling |
1058 | THEN. */ | |
1059 | ||
1060 | if (rrc == MATCH_THEN) | |
1061 | { | |
1062 | next = ecode + GET(ecode,1); | |
1063 | if (md->start_match_ptr < next && | |
1064 | (*ecode == OP_ALT || *next == OP_ALT)) | |
1065 | rrc = MATCH_NOMATCH; | |
1066 | } | |
1067 | ||
1068 | if (rrc != MATCH_NOMATCH) | |
1069 | { | { |
1070 | if (rrc == MATCH_ONCE) | if (rrc == MATCH_ONCE) |
1071 | { | { |
1072 | const uschar *scode = ecode; | const pcre_uchar *scode = ecode; |
1073 | if (*scode != OP_ONCE) /* If not at start, find it */ | if (*scode != OP_ONCE) /* If not at start, find it */ |
1074 | { | { |
1075 | while (*scode == OP_ALT) scode += GET(scode, 1); | while (*scode == OP_ALT) scode += GET(scode, 1); |
# | Line 910 for (;;) | Line 1080 for (;;) |
1080 | RRETURN(rrc); | RRETURN(rrc); |
1081 | } | } |
1082 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
1083 | md->mark = save_mark; | |
1084 | if (*ecode != OP_ALT) break; | if (*ecode != OP_ALT) break; |
1085 | } | } |
1086 | if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr; | |
1087 | RRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
1088 | ||
1089 | /* Handle possessive capturing brackets with an unlimited repeat. We come | /* Handle possessive capturing brackets with an unlimited repeat. We come |
# | Line 941 for (;;) | Line 1112 for (;;) |
1112 | if (offset < md->offset_max) | if (offset < md->offset_max) |
1113 | { | { |
1114 | matched_once = FALSE; | matched_once = FALSE; |
1115 | code_offset = ecode - md->start_code; | code_offset = (int)(ecode - md->start_code); |
1116 | ||
1117 | save_offset1 = md->offset_vector[offset]; | save_offset1 = md->offset_vector[offset]; |
1118 | save_offset2 = md->offset_vector[offset+1]; | save_offset2 = md->offset_vector[offset+1]; |
# | Line 964 for (;;) | Line 1135 for (;;) |
1135 | md->offset_vector[md->offset_end - number] = | md->offset_vector[md->offset_end - number] = |
1136 | (int)(eptr - md->start_subject); | (int)(eptr - md->start_subject); |
1137 | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
1138 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
1139 | eptrb, RM63); | eptrb, RM63); |
1140 | if (rrc == MATCH_KETRPOS) | if (rrc == MATCH_KETRPOS) |
1141 | { | { |
# | Line 975 for (;;) | Line 1146 for (;;) |
1146 | matched_once = TRUE; | matched_once = TRUE; |
1147 | continue; | continue; |
1148 | } | } |
1149 | if (rrc != MATCH_NOMATCH && | |
1150 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | /* See comment in the code for capturing groups above about handling |
1151 | RRETURN(rrc); | THEN. */ |
1152 | ||
1153 | if (rrc == MATCH_THEN) | |
1154 | { | |
1155 | next = ecode + GET(ecode,1); | |
1156 | if (md->start_match_ptr < next && | |
1157 | (*ecode == OP_ALT || *next == OP_ALT)) | |
1158 | rrc = MATCH_NOMATCH; | |
1159 | } | |
1160 | ||
1161 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
1162 | md->capture_last = save_capture_last; | md->capture_last = save_capture_last; |
1163 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
1164 | if (*ecode != OP_ALT) break; | if (*ecode != OP_ALT) break; |
# | Line 990 for (;;) | Line 1171 for (;;) |
1171 | md->offset_vector[md->offset_end - number] = save_offset3; | md->offset_vector[md->offset_end - number] = save_offset3; |
1172 | } | } |
1173 | ||
if (rrc != MATCH_THEN && md->mark == NULL) md->mark = markptr; | ||
1174 | if (allow_zero || matched_once) | if (allow_zero || matched_once) |
1175 | { | { |
1176 | ecode += 1 + LINK_SIZE; | ecode += 1 + LINK_SIZE; |
# | Line 1022 for (;;) | Line 1202 for (;;) |
1202 | ||
1203 | POSSESSIVE_NON_CAPTURE: | POSSESSIVE_NON_CAPTURE: |
1204 | matched_once = FALSE; | matched_once = FALSE; |
1205 | code_offset = ecode - md->start_code; | code_offset = (int)(ecode - md->start_code); |
1206 | ||
1207 | for (;;) | for (;;) |
1208 | { | { |
1209 | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
1210 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
1211 | eptrb, RM48); | eptrb, RM48); |
1212 | if (rrc == MATCH_KETRPOS) | if (rrc == MATCH_KETRPOS) |
1213 | { | { |
# | Line 1037 for (;;) | Line 1217 for (;;) |
1217 | matched_once = TRUE; | matched_once = TRUE; |
1218 | continue; | continue; |
1219 | } | } |
1220 | if (rrc != MATCH_NOMATCH && | |
1221 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | /* See comment in the code for capturing groups above about handling |
1222 | RRETURN(rrc); | THEN. */ |
1223 | ||
1224 | if (rrc == MATCH_THEN) | |
1225 | { | |
1226 | next = ecode + GET(ecode,1); | |
1227 | if (md->start_match_ptr < next && | |
1228 | (*ecode == OP_ALT || *next == OP_ALT)) | |
1229 | rrc = MATCH_NOMATCH; | |
1230 | } | |
1231 | ||
1232 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
1233 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
1234 | if (*ecode != OP_ALT) break; | if (*ecode != OP_ALT) break; |
1235 | } | } |
# | Line 1067 for (;;) | Line 1257 for (;;) |
1257 | ||
1258 | if (ecode[LINK_SIZE+1] == OP_CALLOUT) | if (ecode[LINK_SIZE+1] == OP_CALLOUT) |
1259 | { | { |
1260 | if (pcre_callout != NULL) | if (PUBL(callout) != NULL) |
1261 | { | { |
1262 | pcre_callout_block cb; | PUBL(callout_block) cb; |
1263 | cb.version = 1; /* Version 1 of the callout block */ | cb.version = 2; /* Version 1 of the callout block */ |
1264 | cb.callout_number = ecode[LINK_SIZE+2]; | cb.callout_number = ecode[LINK_SIZE+2]; |
1265 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
1266 | #ifdef COMPILE_PCRE8 | |
1267 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
1268 | #else | |
1269 | cb.subject = (PCRE_SPTR16)md->start_subject; | |
1270 | #endif | |
1271 | cb.subject_length = (int)(md->end_subject - md->start_subject); | cb.subject_length = (int)(md->end_subject - md->start_subject); |
1272 | cb.start_match = (int)(mstart - md->start_subject); | cb.start_match = (int)(mstart - md->start_subject); |
1273 | cb.current_position = (int)(eptr - md->start_subject); | cb.current_position = (int)(eptr - md->start_subject); |
# | Line 1082 for (;;) | Line 1276 for (;;) |
1276 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
1277 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
1278 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
1279 | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); | cb.mark = md->nomatch_mark; |
1280 | if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH); | |
1281 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
1282 | } | } |
1283 | ecode += _pcre_OP_lengths[OP_CALLOUT]; | ecode += PRIV(OP_lengths)[OP_CALLOUT]; |
1284 | } | } |
1285 | ||
1286 | condcode = ecode[LINK_SIZE+1]; | condcode = ecode[LINK_SIZE+1]; |
# | Line 1102 for (;;) | Line 1297 for (;;) |
1297 | else | else |
1298 | { | { |
1299 | int recno = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | int recno = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ |
1300 | condition = (recno == RREF_ANY || recno == md->recursive->group_num); | condition = (recno == RREF_ANY || recno == md->recursive->group_num); |
1301 | ||
1302 | /* If the test is for recursion into a specific subpattern, and it is | /* If the test is for recursion into a specific subpattern, and it is |
1303 | false, but the test was set up by name, scan the table to see if the | false, but the test was set up by name, scan the table to see if the |
1304 | name refers to any other numbers, and test them. The condition is true | name refers to any other numbers, and test them. The condition is true |
1305 | if any one is set. */ | if any one is set. */ |
1306 | ||
1307 | if (!condition && condcode == OP_NRREF && recno != RREF_ANY) | if (!condition && condcode == OP_NRREF) |
1308 | { | { |
1309 | uschar *slotA = md->name_table; | pcre_uchar *slotA = md->name_table; |
1310 | for (i = 0; i < md->name_count; i++) | for (i = 0; i < md->name_count; i++) |
1311 | { | { |
1312 | if (GET2(slotA, 0) == recno) break; | if (GET2(slotA, 0) == recno) break; |
# | Line 1124 for (;;) | Line 1319 for (;;) |
1319 | ||
1320 | if (i < md->name_count) | if (i < md->name_count) |
1321 | { | { |
1322 | uschar *slotB = slotA; | pcre_uchar *slotB = slotA; |
1323 | while (slotB > md->name_table) | while (slotB > md->name_table) |
1324 | { | { |
1325 | slotB -= md->name_entry_size; | slotB -= md->name_entry_size; |
1326 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
1327 | { | { |
1328 | condition = GET2(slotB, 0) == md->recursive->group_num; | condition = GET2(slotB, 0) == md->recursive->group_num; |
1329 | if (condition) break; | if (condition) break; |
# | Line 1144 for (;;) | Line 1339 for (;;) |
1339 | for (i++; i < md->name_count; i++) | for (i++; i < md->name_count; i++) |
1340 | { | { |
1341 | slotB += md->name_entry_size; | slotB += md->name_entry_size; |
1342 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
1343 | { | { |
1344 | condition = GET2(slotB, 0) == md->recursive->group_num; | condition = GET2(slotB, 0) == md->recursive->group_num; |
1345 | if (condition) break; | if (condition) break; |
# | Line 1157 for (;;) | Line 1352 for (;;) |
1352 | ||
1353 | /* Chose branch according to the condition */ | /* Chose branch according to the condition */ |
1354 | ||
1355 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1); |
1356 | } | } |
1357 | } | } |
1358 | ||
# | Line 1174 for (;;) | Line 1369 for (;;) |
1369 | if (!condition && condcode == OP_NCREF) | if (!condition && condcode == OP_NCREF) |
1370 | { | { |
1371 | int refno = offset >> 1; | int refno = offset >> 1; |
1372 | uschar *slotA = md->name_table; | pcre_uchar *slotA = md->name_table; |
1373 | ||
1374 | for (i = 0; i < md->name_count; i++) | for (i = 0; i < md->name_count; i++) |
1375 | { | { |
# | Line 1188 for (;;) | Line 1383 for (;;) |
1383 | ||
1384 | if (i < md->name_count) | if (i < md->name_count) |
1385 | { | { |
1386 | uschar *slotB = slotA; | pcre_uchar *slotB = slotA; |
1387 | while (slotB > md->name_table) | while (slotB > md->name_table) |
1388 | { | { |
1389 | slotB -= md->name_entry_size; | slotB -= md->name_entry_size; |
1390 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
1391 | { | { |
1392 | offset = GET2(slotB, 0) << 1; | offset = GET2(slotB, 0) << 1; |
1393 | condition = offset < offset_top && | condition = offset < offset_top && |
# | Line 1210 for (;;) | Line 1405 for (;;) |
1405 | for (i++; i < md->name_count; i++) | for (i++; i < md->name_count; i++) |
1406 | { | { |
1407 | slotB += md->name_entry_size; | slotB += md->name_entry_size; |
1408 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
1409 | { | { |
1410 | offset = GET2(slotB, 0) << 1; | offset = GET2(slotB, 0) << 1; |
1411 | condition = offset < offset_top && | condition = offset < offset_top && |
# | Line 1225 for (;;) | Line 1420 for (;;) |
1420 | ||
1421 | /* Chose branch according to the condition */ | /* Chose branch according to the condition */ |
1422 | ||
1423 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1); |
1424 | } | } |
1425 | ||
1426 | else if (condcode == OP_DEF) /* DEFINE - always false */ | else if (condcode == OP_DEF) /* DEFINE - always false */ |
# | Line 1250 for (;;) | Line 1445 for (;;) |
1445 | ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); | ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); |
1446 | while (*ecode == OP_ALT) ecode += GET(ecode, 1); | while (*ecode == OP_ALT) ecode += GET(ecode, 1); |
1447 | } | } |
1448 | else if (rrc != MATCH_NOMATCH && | |
1449 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | /* PCRE doesn't allow the effect of (*THEN) to escape beyond an |
1450 | assertion; it is therefore treated as NOMATCH. */ | |
1451 | ||
1452 | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) | |
1453 | { | { |
1454 | RRETURN(rrc); /* Need braces because of following else */ | RRETURN(rrc); /* Need braces because of following else */ |
1455 | } | } |
# | Line 1262 for (;;) | Line 1460 for (;;) |
1460 | } | } |
1461 | } | } |
1462 | ||
1463 | /* We are now at the branch that is to be obeyed. As there is only one, | /* We are now at the branch that is to be obeyed. As there is only one, can |
1464 | we used to use tail recursion to avoid using another stack frame, except | use tail recursion to avoid using another stack frame, except when there is |
1465 | when there was unlimited repeat of a possibly empty group. However, that | unlimited repeat of a possibly empty group. In the latter case, a recursive |
1466 | strategy no longer works because of the possibilty of (*THEN) being | call to match() is always required, unless the second alternative doesn't |
1467 | encountered in the branch. A recursive call to match() is always required, | exist, in which case we can just plough on. Note that, for compatibility |
1468 | unless the second alternative doesn't exist, in which case we can just | with Perl, the | in a conditional group is NOT treated as creating two |
1469 | plough on. */ | alternatives. If a THEN is encountered in the branch, it propagates out to |
1470 | the enclosing alternative (unless nested in a deeper set of alternatives, | |
1471 | of course). */ | |
1472 | ||
1473 | if (condition || *ecode == OP_ALT) | if (condition || *ecode == OP_ALT) |
1474 | { | { |
1475 | if (op == OP_SCOND) md->match_function_type = MATCH_CBEGROUP; | if (op != OP_SCOND) |
1476 | { | |
1477 | ecode += 1 + LINK_SIZE; | |
1478 | goto TAIL_RECURSE; | |
1479 | } | |
1480 | ||
1481 | md->match_function_type = MATCH_CBEGROUP; | |
1482 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49); | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, eptrb, RM49); |
if (rrc == MATCH_THEN && md->start_match_ptr == ecode) | ||
rrc = MATCH_NOMATCH; | ||
1483 | RRETURN(rrc); | RRETURN(rrc); |
1484 | } | } |
1485 | else /* Condition false & no alternative */ | |
1486 | /* Condition false & no alternative; continue after the group. */ | |
1487 | ||
1488 | else | |
1489 | { | { |
1490 | ecode += 1 + LINK_SIZE; | ecode += 1 + LINK_SIZE; |
1491 | } | } |
# | Line 1305 for (;;) | Line 1512 for (;;) |
1512 | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); |
1513 | if (offset_top <= offset) offset_top = offset + 2; | if (offset_top <= offset) offset_top = offset + 2; |
1514 | } | } |
1515 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
1516 | break; | break; |
1517 | ||
1518 | ||
# | Line 1325 for (;;) | Line 1532 for (;;) |
1532 | (md->notempty || | (md->notempty || |
1533 | (md->notempty_atstart && | (md->notempty_atstart && |
1534 | mstart == md->start_subject + md->start_offset))) | mstart == md->start_subject + md->start_offset))) |
1535 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
1536 | ||
1537 | /* Otherwise, we have a match. */ | /* Otherwise, we have a match. */ |
1538 | ||
# | Line 1334 for (;;) | Line 1541 for (;;) |
1541 | md->start_match_ptr = mstart; /* and the start (\K can modify) */ | md->start_match_ptr = mstart; /* and the start (\K can modify) */ |
1542 | ||
1543 | /* For some reason, the macros don't work properly if an expression is | /* For some reason, the macros don't work properly if an expression is |
1544 | given as the argument to MRRETURN when the heap is in use. */ | given as the argument to RRETURN when the heap is in use. */ |
1545 | ||
1546 | rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; | rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; |
1547 | MRRETURN(rrc); | RRETURN(rrc); |
1548 | ||
1549 | /* Assertion brackets. Check the alternative branches in turn - the | /* Assertion brackets. Check the alternative branches in turn - the |
1550 | matching won't pass the KET for an assertion. If any one branch matches, | matching won't pass the KET for an assertion. If any one branch matches, |
# | Line 1352 for (;;) | Line 1559 for (;;) |
1559 | ||
1560 | case OP_ASSERT: | case OP_ASSERT: |
1561 | case OP_ASSERTBACK: | case OP_ASSERTBACK: |
1562 | save_mark = md->mark; | |
1563 | if (md->match_function_type == MATCH_CONDASSERT) | if (md->match_function_type == MATCH_CONDASSERT) |
1564 | { | { |
1565 | condassert = TRUE; | condassert = TRUE; |
# | Line 1365 for (;;) | Line 1573 for (;;) |
1573 | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
1574 | { | { |
1575 | mstart = md->start_match_ptr; /* In case \K reset it */ | mstart = md->start_match_ptr; /* In case \K reset it */ |
markptr = md->mark; | ||
1576 | break; | break; |
1577 | } | } |
1578 | if (rrc != MATCH_NOMATCH && | |
1579 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | /* PCRE does not allow THEN to escape beyond an assertion; it is treated |
1580 | RRETURN(rrc); | as NOMATCH. */ |
1581 | ||
1582 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | |
1583 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
1584 | md->mark = save_mark; | |
1585 | } | } |
1586 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
1587 | ||
1588 | if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH); | if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); |
1589 | ||
1590 | /* If checking an assertion for a condition, return MATCH_MATCH. */ | /* If checking an assertion for a condition, return MATCH_MATCH. */ |
1591 | ||
# | Line 1395 for (;;) | Line 1605 for (;;) |
1605 | ||
1606 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
1607 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
1608 | save_mark = md->mark; | |
1609 | if (md->match_function_type == MATCH_CONDASSERT) | if (md->match_function_type == MATCH_CONDASSERT) |
1610 | { | { |
1611 | condassert = TRUE; | condassert = TRUE; |
# | Line 1405 for (;;) | Line 1616 for (;;) |
1616 | do | do |
1617 | { | { |
1618 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5); | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5); |
1619 | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH); | md->mark = save_mark; |
1620 | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH); | |
1621 | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) |
1622 | { | { |
1623 | do ecode += GET(ecode,1); while (*ecode == OP_ALT); | do ecode += GET(ecode,1); while (*ecode == OP_ALT); |
1624 | break; | break; |
1625 | } | } |
1626 | if (rrc != MATCH_NOMATCH && | |
1627 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | /* PCRE does not allow THEN to escape beyond an assertion; it is treated |
1628 | RRETURN(rrc); | as NOMATCH. */ |
1629 | ||
1630 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | |
1631 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
1632 | } | } |
1633 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
# | Line 1429 for (;;) | Line 1643 for (;;) |
1643 | back a number of characters, not bytes. */ | back a number of characters, not bytes. */ |
1644 | ||
1645 | case OP_REVERSE: | case OP_REVERSE: |
1646 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
1647 | if (utf8) | if (utf) |
1648 | { | { |
1649 | i = GET(ecode, 1); | i = GET(ecode, 1); |
1650 | while (i-- > 0) | while (i-- > 0) |
1651 | { | { |
1652 | eptr--; | eptr--; |
1653 | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); |
1654 | BACKCHAR(eptr); | BACKCHAR(eptr); |
1655 | } | } |
1656 | } | } |
# | Line 1447 for (;;) | Line 1661 for (;;) |
1661 | ||
1662 | { | { |
1663 | eptr -= GET(ecode, 1); | eptr -= GET(ecode, 1); |
1664 | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); |
1665 | } | } |
1666 | ||
1667 | /* Save the earliest consulted character, then skip to next op code */ | /* Save the earliest consulted character, then skip to next op code */ |
# | Line 1461 for (;;) | Line 1675 for (;;) |
1675 | function is able to force a failure. */ | function is able to force a failure. */ |
1676 | ||
1677 | case OP_CALLOUT: | case OP_CALLOUT: |
1678 | if (pcre_callout != NULL) | if (PUBL(callout) != NULL) |
1679 | { | { |
1680 | pcre_callout_block cb; | PUBL(callout_block) cb; |
1681 | cb.version = 1; /* Version 1 of the callout block */ | cb.version = 2; /* Version 1 of the callout block */ |
1682 | cb.callout_number = ecode[1]; | cb.callout_number = ecode[1]; |
1683 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
1684 | #ifdef COMPILE_PCRE8 | |
1685 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
1686 | #else | |
1687 | cb.subject = (PCRE_SPTR16)md->start_subject; | |
1688 | #endif | |
1689 | cb.subject_length = (int)(md->end_subject - md->start_subject); | cb.subject_length = (int)(md->end_subject - md->start_subject); |
1690 | cb.start_match = (int)(mstart - md->start_subject); | cb.start_match = (int)(mstart - md->start_subject); |
1691 | cb.current_position = (int)(eptr - md->start_subject); | cb.current_position = (int)(eptr - md->start_subject); |
# | Line 1476 for (;;) | Line 1694 for (;;) |
1694 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
1695 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
1696 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
1697 | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); | cb.mark = md->nomatch_mark; |
1698 | if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH); | |
1699 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
1700 | } | } |
1701 | ecode += 2 + 2*LINK_SIZE; | ecode += 2 + 2*LINK_SIZE; |
# | Line 1501 for (;;) | Line 1720 for (;;) |
1720 | ||
1721 | case OP_RECURSE: | case OP_RECURSE: |
1722 | { | { |
1723 | recursion_info *ri; | |
1724 | int recno; | |
1725 | ||
1726 | callpat = md->start_code + GET(ecode, 1); | callpat = md->start_code + GET(ecode, 1); |
1727 | new_recursive.group_num = (callpat == md->start_code)? 0 : | recno = (callpat == md->start_code)? 0 : |
1728 | GET2(callpat, 1 + LINK_SIZE); | GET2(callpat, 1 + LINK_SIZE); |
1729 | ||
1730 | /* Check for repeating a recursion without advancing the subject pointer. | |
1731 | This should catch convoluted mutual recursions. (Some simple cases are | |
1732 | caught at compile time.) */ | |
1733 | ||
1734 | for (ri = md->recursive; ri != NULL; ri = ri->prevrec) | |
1735 | if (recno == ri->group_num && eptr == ri->subject_position) | |
1736 | RRETURN(PCRE_ERROR_RECURSELOOP); | |
1737 | ||
1738 | /* Add to "recursing stack" */ | /* Add to "recursing stack" */ |
1739 | ||
1740 | new_recursive.group_num = recno; | |
1741 | new_recursive.subject_position = eptr; | |
1742 | new_recursive.prevrec = md->recursive; | new_recursive.prevrec = md->recursive; |
1743 | md->recursive = &new_recursive; | md->recursive = &new_recursive; |
1744 | ||
# | Line 1522 for (;;) | Line 1754 for (;;) |
1754 | else | else |
1755 | { | { |
1756 | new_recursive.offset_save = | new_recursive.offset_save = |
1757 | (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int)); | (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int)); |
1758 | if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); | if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); |
1759 | } | } |
1760 | memcpy(new_recursive.offset_save, md->offset_vector, | memcpy(new_recursive.offset_save, md->offset_vector, |
# | Line 1537 for (;;) | Line 1769 for (;;) |
1769 | do | do |
1770 | { | { |
1771 | if (cbegroup) md->match_function_type = MATCH_CBEGROUP; | if (cbegroup) md->match_function_type = MATCH_CBEGROUP; |
1772 | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, | RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top, |
1773 | md, eptrb, RM6); | md, eptrb, RM6); |
1774 | memcpy(md->offset_vector, new_recursive.offset_save, | memcpy(md->offset_vector, new_recursive.offset_save, |
1775 | new_recursive.saved_max * sizeof(int)); | new_recursive.saved_max * sizeof(int)); |
1776 | md->recursive = new_recursive.prevrec; | |
1777 | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
1778 | { | { |
1779 | DPRINTF(("Recursion matched\n")); | DPRINTF(("Recursion matched\n")); |
md->recursive = new_recursive.prevrec; | ||
1780 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
1781 | (pcre_free)(new_recursive.offset_save); | (PUBL(free))(new_recursive.offset_save); |
1782 | ||
1783 | /* Set where we got to in the subject, and reset the start in case | /* Set where we got to in the subject, and reset the start in case |
1784 | it was changed by \K. This *is* propagated back out of a recursion, | it was changed by \K. This *is* propagated back out of a recursion, |
# | Line 1556 for (;;) | Line 1788 for (;;) |
1788 | mstart = md->start_match_ptr; | mstart = md->start_match_ptr; |
1789 | goto RECURSION_MATCHED; /* Exit loop; end processing */ | goto RECURSION_MATCHED; /* Exit loop; end processing */ |
1790 | } | } |
1791 | else if (rrc != MATCH_NOMATCH && | |
1792 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | /* PCRE does not allow THEN to escape beyond a recursion; it is treated |
1793 | as NOMATCH. */ | |
1794 | ||
1795 | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) | |
1796 | { | { |
1797 | DPRINTF(("Recursion gave error %d\n", rrc)); | DPRINTF(("Recursion gave error %d\n", rrc)); |
1798 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
1799 | (pcre_free)(new_recursive.offset_save); | (PUBL(free))(new_recursive.offset_save); |
1800 | RRETURN(rrc); | RRETURN(rrc); |
1801 | } | } |
1802 | ||
# | Line 1573 for (;;) | Line 1808 for (;;) |
1808 | DPRINTF(("Recursion didn't match\n")); | DPRINTF(("Recursion didn't match\n")); |
1809 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
1810 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
1811 | (pcre_free)(new_recursive.offset_save); | (PUBL(free))(new_recursive.offset_save); |
1812 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
1813 | } | } |
1814 | ||
1815 | RECURSION_MATCHED: | RECURSION_MATCHED: |
# | Line 1643 for (;;) | Line 1878 for (;;) |
1878 | } | } |
1879 | else saved_eptr = NULL; | else saved_eptr = NULL; |
1880 | ||
1881 | /* If we are at the end of an assertion group, stop matching and return | /* If we are at the end of an assertion group or a non-capturing atomic |
1882 | MATCH_MATCH, but record the current high water mark for use by positive | group, stop matching and return MATCH_MATCH, but record the current high |
1883 | assertions. We also need to record the match start in case it was changed | water mark for use by positive assertions. We also need to record the match |
1884 | by \K. */ | start in case it was changed by \K. */ |
1885 | ||
1886 | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || | if ((*prev >= OP_ASSERT && *prev <= OP_ASSERTBACK_NOT) || |
1887 | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT) | *prev == OP_ONCE_NC) |
1888 | { | { |
1889 | md->end_match_ptr = eptr; /* For ONCE */ | md->end_match_ptr = eptr; /* For ONCE_NC */ |
1890 | md->end_offset_top = offset_top; | md->end_offset_top = offset_top; |
1891 | md->start_match_ptr = mstart; | md->start_match_ptr = mstart; |
1892 | MRRETURN(MATCH_MATCH); /* Sets md->mark */ | RRETURN(MATCH_MATCH); /* Sets md->mark */ |
1893 | } | } |
1894 | ||
1895 | /* For capturing groups we have to check the group number back at the start | /* For capturing groups we have to check the group number back at the start |
# | Line 1719 for (;;) | Line 1954 for (;;) |
1954 | /* For an ordinary non-repeating ket, just continue at this level. This | /* For an ordinary non-repeating ket, just continue at this level. This |
1955 | also happens for a repeating ket if no characters were matched in the | also happens for a repeating ket if no characters were matched in the |
1956 | group. This is the forcible breaking of infinite loops as implemented in | group. This is the forcible breaking of infinite loops as implemented in |
1957 | Perl 5.005. For a non-repeating atomic group, establish a backup point by | Perl 5.005. For a non-repeating atomic group that includes captures, |
1958 | processing the rest of the pattern at a lower level. If this results in a | establish a backup point by processing the rest of the pattern at a lower |
1959 | NOMATCH return, pass MATCH_ONCE back to the original OP_ONCE level, thereby | level. If this results in a NOMATCH return, pass MATCH_ONCE back to the |
1960 | bypassing intermediate backup points, but resetting any captures that | original OP_ONCE level, thereby bypassing intermediate backup points, but |
1961 | happened along the way. */ | resetting any captures that happened along the way. */ |
1962 | ||
1963 | if (*ecode == OP_KET || eptr == saved_eptr) | if (*ecode == OP_KET || eptr == saved_eptr) |
1964 | { | { |
# | Line 1796 for (;;) | Line 2031 for (;;) |
2031 | /* Not multiline mode: start of subject assertion, unless notbol. */ | /* Not multiline mode: start of subject assertion, unless notbol. */ |
2032 | ||
2033 | case OP_CIRC: | case OP_CIRC: |
2034 | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); |
2035 | ||
2036 | /* Start of subject assertion */ | /* Start of subject assertion */ |
2037 | ||
2038 | case OP_SOD: | case OP_SOD: |
2039 | if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); |
2040 | ecode++; | ecode++; |
2041 | break; | break; |
2042 | ||
2043 | /* Multiline mode: start of subject unless notbol, or after any newline. */ | /* Multiline mode: start of subject unless notbol, or after any newline. */ |
2044 | ||
2045 | case OP_CIRCM: | case OP_CIRCM: |
2046 | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); |
2047 | if (eptr != md->start_subject && | if (eptr != md->start_subject && |
2048 | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) |
2049 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2050 | ecode++; | ecode++; |
2051 | break; | break; |
2052 | ||
2053 | /* Start of match assertion */ | /* Start of match assertion */ |
2054 | ||
2055 | case OP_SOM: | case OP_SOM: |
2056 | if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); |
2057 | ecode++; | ecode++; |
2058 | break; | break; |
2059 | ||
# | Line 1834 for (;;) | Line 2069 for (;;) |
2069 | ||
2070 | case OP_DOLLM: | case OP_DOLLM: |
2071 | if (eptr < md->end_subject) | if (eptr < md->end_subject) |
2072 | { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); } | { |
2073 | if (!IS_NEWLINE(eptr)) | |
2074 | { | |
2075 | if (md->partial != 0 && | |
2076 | eptr + 1 >= md->end_subject && | |
2077 | NLBLOCK->nltype == NLTYPE_FIXED && | |
2078 | NLBLOCK->nllen == 2 && | |
2079 | *eptr == NLBLOCK->nl[0]) | |
2080 | { | |
2081 | md->hitend = TRUE; | |
2082 | if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); | |
2083 | } | |
2084 | RRETURN(MATCH_NOMATCH); | |
2085 | } | |
2086 | } | |
2087 | else | else |
2088 | { | { |
2089 | if (md->noteol) MRRETURN(MATCH_NOMATCH); | if (md->noteol) RRETURN(MATCH_NOMATCH); |
2090 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2091 | } | } |
2092 | ecode++; | ecode++; |
# | Line 1847 for (;;) | Line 2096 for (;;) |
2096 | subject unless noteol is set. */ | subject unless noteol is set. */ |
2097 | ||
2098 | case OP_DOLL: | case OP_DOLL: |
2099 | if (md->noteol) MRRETURN(MATCH_NOMATCH); | if (md->noteol) RRETURN(MATCH_NOMATCH); |
2100 | if (!md->endonly) goto ASSERT_NL_OR_EOS; | if (!md->endonly) goto ASSERT_NL_OR_EOS; |
2101 | ||
2102 | /* ... else fall through for endonly */ | /* ... else fall through for endonly */ |
# | Line 1855 for (;;) | Line 2104 for (;;) |
2104 | /* End of subject assertion (\z) */ | /* End of subject assertion (\z) */ |
2105 | ||
2106 | case OP_EOD: | case OP_EOD: |
2107 | if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH); | if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); |
2108 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2109 | ecode++; | ecode++; |
2110 | break; | break; |
# | Line 1866 for (;;) | Line 2115 for (;;) |
2115 | ASSERT_NL_OR_EOS: | ASSERT_NL_OR_EOS: |
2116 | if (eptr < md->end_subject && | if (eptr < md->end_subject && |
2117 | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) |
2118 | MRRETURN(MATCH_NOMATCH); | { |
2119 | if (md->partial != 0 && | |
2120 | eptr + 1 >= md->end_subject && | |
2121 | NLBLOCK->nltype == NLTYPE_FIXED && | |
2122 | NLBLOCK->nllen == 2 && | |
2123 | *eptr == NLBLOCK->nl[0]) | |
2124 | { | |
2125 | md->hitend = TRUE; | |
2126 | if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); | |
2127 | } | |
2128 | RRETURN(MATCH_NOMATCH); | |
2129 | } | |
2130 | ||
2131 | /* Either at end of string or \n before end. */ | /* Either at end of string or \n before end. */ |
2132 | ||
# | Line 1885 for (;;) | Line 2145 for (;;) |
2145 | be "non-word" characters. Remember the earliest consulted character for | be "non-word" characters. Remember the earliest consulted character for |
2146 | partial matching. */ | partial matching. */ |
2147 | ||
2148 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
2149 | if (utf8) | if (utf) |
2150 | { | { |
2151 | /* Get status of previous character */ | /* Get status of previous character */ |
2152 | ||
2153 | if (eptr == md->start_subject) prev_is_word = FALSE; else | if (eptr == md->start_subject) prev_is_word = FALSE; else |
2154 | { | { |
2155 | USPTR lastptr = eptr - 1; | PCRE_PUCHAR lastptr = eptr - 1; |
2156 | while((*lastptr & 0xc0) == 0x80) lastptr--; | BACKCHAR(lastptr); |
2157 | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; |
2158 | GETCHAR(c, lastptr); | GETCHAR(c, lastptr); |
2159 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
# | Line 1958 for (;;) | Line 2218 for (;;) |
2218 | } | } |
2219 | else | else |
2220 | #endif | #endif |
2221 | prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0); | prev_is_word = MAX_255(eptr[-1]) |
2222 | && ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
2223 | } | } |
2224 | ||
2225 | /* Get status of next character */ | /* Get status of next character */ |
# | Line 1981 for (;;) | Line 2242 for (;;) |
2242 | } | } |
2243 | else | else |
2244 | #endif | #endif |
2245 | cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0); | cur_is_word = MAX_255(*eptr) |
2246 | && ((md->ctypes[*eptr] & ctype_word) != 0); | |
2247 | } | } |
2248 | ||
2249 | /* Now see if the situation is what we want */ | /* Now see if the situation is what we want */ |
2250 | ||
2251 | if ((*ecode++ == OP_WORD_BOUNDARY)? | if ((*ecode++ == OP_WORD_BOUNDARY)? |
2252 | cur_is_word == prev_is_word : cur_is_word != prev_is_word) | cur_is_word == prev_is_word : cur_is_word != prev_is_word) |
2253 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2254 | } | } |
2255 | break; | break; |
2256 | ||
2257 | /* Match a single character type; inline for speed */ | /* Match any single character type except newline; have to take care with |
2258 | CRLF newlines and partial matching. */ | |
2259 | ||
2260 | case OP_ANY: | case OP_ANY: |
2261 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); |
2262 | if (md->partial != 0 && | |
2263 | eptr + 1 >= md->end_subject && | |
2264 | NLBLOCK->nltype == NLTYPE_FIXED && | |
2265 | NLBLOCK->nllen == 2 && | |
2266 | *eptr == NLBLOCK->nl[0]) | |
2267 | { | |
2268 | md->hitend = TRUE; | |
2269 | if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); | |
2270 | } | |
2271 | ||
2272 | /* Fall through */ | /* Fall through */ |
2273 | ||
2274 | /* Match any single character whatsoever. */ | |
2275 | ||
2276 | case OP_ALLANY: | case OP_ALLANY: |
2277 | if (eptr++ >= md->end_subject) | if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ |
2278 | { | { /* not be updated before SCHECK_PARTIAL. */ |
2279 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2280 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2281 | } | } |
2282 | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | eptr++; |
2283 | #ifdef SUPPORT_UTF | |
2284 | if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); | |
2285 | #endif | |
2286 | ecode++; | ecode++; |
2287 | break; | break; |
2288 | ||
# | Line 2012 for (;;) | Line 2290 for (;;) |
2290 | any byte, even newline, independent of the setting of PCRE_DOTALL. */ | any byte, even newline, independent of the setting of PCRE_DOTALL. */ |
2291 | ||
2292 | case OP_ANYBYTE: | case OP_ANYBYTE: |
2293 | if (eptr++ >= md->end_subject) | if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ |
2294 | { | { /* not be updated before SCHECK_PARTIAL. */ |
2295 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2296 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2297 | } | } |
2298 | eptr++; | |
2299 | ecode++; | ecode++; |
2300 | break; | break; |
2301 | ||
# | Line 2024 for (;;) | Line 2303 for (;;) |
2303 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2304 | { | { |
2305 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2306 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2307 | } | } |
2308 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2309 | if ( | if ( |
2310 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
2311 | c < 256 && | c < 256 && |
2312 | #endif | #endif |
2313 | (md->ctypes[c] & ctype_digit) != 0 | (md->ctypes[c] & ctype_digit) != 0 |
2314 | ) | ) |
2315 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2316 | ecode++; | ecode++; |
2317 | break; | break; |
2318 | ||
# | Line 2041 for (;;) | Line 2320 for (;;) |
2320 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2321 | { | { |
2322 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2323 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2324 | } | } |
2325 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2326 | if ( | if ( |
2327 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
2328 | c >= 256 || | c > 255 || |
2329 | #endif | #endif |
2330 | (md->ctypes[c] & ctype_digit) == 0 | (md->ctypes[c] & ctype_digit) == 0 |
2331 | ) | ) |
2332 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2333 | ecode++; | ecode++; |
2334 | break; | break; |
2335 | ||
# | Line 2058 for (;;) | Line 2337 for (;;) |
2337 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2338 | { | { |
2339 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2340 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2341 | } | } |
2342 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2343 | if ( | if ( |
2344 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
2345 | c < 256 && | c < 256 && |
2346 | #endif | #endif |
2347 | (md->ctypes[c] & ctype_space) != 0 | (md->ctypes[c] & ctype_space) != 0 |
2348 | ) | ) |
2349 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2350 | ecode++; | ecode++; |
2351 | break; | break; |
2352 | ||
# | Line 2075 for (;;) | Line 2354 for (;;) |
2354 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2355 | { | { |
2356 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2357 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2358 | } | } |
2359 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2360 | if ( | if ( |
2361 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
2362 | c >= 256 || | c > 255 || |
2363 | #endif | #endif |
2364 | (md->ctypes[c] & ctype_space) == 0 | (md->ctypes[c] & ctype_space) == 0 |
2365 | ) | ) |
2366 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2367 | ecode++; | ecode++; |
2368 | break; | break; |
2369 | ||
# | Line 2092 for (;;) | Line 2371 for (;;) |
2371 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2372 | { | { |
2373 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2374 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2375 | } | } |
2376 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2377 | if ( | if ( |
2378 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
2379 | c < 256 && | c < 256 && |
2380 | #endif | #endif |
2381 | (md->ctypes[c] & ctype_word) != 0 | (md->ctypes[c] & ctype_word) != 0 |
2382 | ) | ) |
2383 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2384 | ecode++; | ecode++; |
2385 | break; | break; |
2386 | ||
# | Line 2109 for (;;) | Line 2388 for (;;) |
2388 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2389 | { | { |
2390 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2391 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2392 | } | } |
2393 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2394 | if ( | if ( |
2395 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
2396 | c >= 256 || | c > 255 || |
2397 | #endif | #endif |
2398 | (md->ctypes[c] & ctype_word) == 0 | (md->ctypes[c] & ctype_word) == 0 |
2399 | ) | ) |
2400 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2401 | ecode++; | ecode++; |
2402 | break; | break; |
2403 | ||
# | Line 2126 for (;;) | Line 2405 for (;;) |
2405 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2406 | { | { |
2407 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2408 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2409 | } | } |
2410 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2411 | switch(c) | switch(c) |
2412 | { | { |
2413 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
2414 | ||
2415 | case 0x000d: | case 0x000d: |
2416 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr >= md->end_subject) |
2417 | { | |
2418 | SCHECK_PARTIAL(); | |
2419 | } | |
2420 | else if (*eptr == 0x0a) eptr++; | |
2421 | break; | break; |
2422 | ||
2423 | case 0x000a: | case 0x000a: |
# | Line 2145 for (;;) | Line 2428 for (;;) |
2428 | case 0x0085: | case 0x0085: |
2429 | case 0x2028: | case 0x2028: |
2430 | case 0x2029: | case 0x2029: |
2431 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); |
2432 | break; | break; |
2433 | } | } |
2434 | ecode++; | ecode++; |
# | Line 2155 for (;;) | Line 2438 for (;;) |
2438 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2439 | { | { |
2440 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2441 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2442 | } | } |
2443 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2444 | switch(c) | switch(c) |
# | Line 2180 for (;;) | Line 2463 for (;;) |
2463 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
2464 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
2465 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
2466 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2467 | } | } |
2468 | ecode++; | ecode++; |
2469 | break; | break; |
# | Line 2189 for (;;) | Line 2472 for (;;) |
2472 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2473 | { | { |
2474 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2475 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2476 | } | } |
2477 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2478 | switch(c) | switch(c) |
2479 | { | { |
2480 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
2481 | case 0x09: /* HT */ | case 0x09: /* HT */ |
2482 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
2483 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
# | Line 2223 for (;;) | Line 2506 for (;;) |
2506 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2507 | { | { |
2508 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2509 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2510 | } | } |
2511 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2512 | switch(c) | switch(c) |
# | Line 2236 for (;;) | Line 2519 for (;;) |
2519 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
2520 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
2521 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
2522 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2523 | } | } |
2524 | ecode++; | ecode++; |
2525 | break; | break; |
# | Line 2245 for (;;) | Line 2528 for (;;) |
2528 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2529 | { | { |
2530 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2531 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2532 | } | } |
2533 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2534 | switch(c) | switch(c) |
2535 | { | { |
2536 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
2537 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
2538 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
2539 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
# | Line 2272 for (;;) | Line 2555 for (;;) |
2555 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2556 | { | { |
2557 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2558 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2559 | } | } |
2560 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2561 | { | { |
# | Line 2281 for (;;) | Line 2564 for (;;) |
2564 | switch(ecode[1]) | switch(ecode[1]) |
2565 | { | { |
2566 | case PT_ANY: | case PT_ANY: |
2567 | if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH); | if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); |
2568 | break; | break; |
2569 | ||
2570 | case PT_LAMP: | case PT_LAMP: |
2571 | if ((prop->chartype == ucp_Lu || | if ((prop->chartype == ucp_Lu || |
2572 | prop->chartype == ucp_Ll || | prop->chartype == ucp_Ll || |
2573 | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) |
2574 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2575 | break; | break; |
2576 | ||
2577 | case PT_GC: | case PT_GC: |
2578 | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) | if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP)) |
2579 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2580 | break; | break; |
2581 | ||
2582 | case PT_PC: | case PT_PC: |
2583 | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) |
2584 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2585 | break; | break; |
2586 | ||
2587 | case PT_SC: | case PT_SC: |
2588 | if ((ecode[2] != prop->script) == (op == OP_PROP)) | if ((ecode[2] != prop->script) == (op == OP_PROP)) |
2589 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2590 | break; | break; |
2591 | ||
2592 | /* These are specials */ | /* These are specials */ |
2593 | ||
2594 | case PT_ALNUM: | case PT_ALNUM: |
2595 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || |
2596 | _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) | PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) |
2597 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2598 | break; | break; |
2599 | ||
2600 | case PT_SPACE: /* Perl space */ | case PT_SPACE: /* Perl space */ |
2601 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z || |
2602 | c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) | c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) |
2603 | == (op == OP_NOTPROP)) | == (op == OP_NOTPROP)) |
2604 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2605 | break; | break; |
2606 | ||
2607 | case PT_PXSPACE: /* POSIX space */ | case PT_PXSPACE: /* POSIX space */ |
2608 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z || |
2609 | c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || | c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || |
2610 | c == CHAR_FF || c == CHAR_CR) | c == CHAR_FF || c == CHAR_CR) |
2611 | == (op == OP_NOTPROP)) | == (op == OP_NOTPROP)) |
2612 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2613 | break; | break; |
2614 | ||
2615 | case PT_WORD: | case PT_WORD: |
2616 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || |
2617 | _pcre_ucp_gentype[prop->chartype] == ucp_N || | PRIV(ucp_gentype)[prop->chartype] == ucp_N || |
2618 | c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) | c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) |
2619 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2620 | break; | break; |
2621 | ||
2622 | /* This should never occur */ | /* This should never occur */ |
# | Line 2353 for (;;) | Line 2636 for (;;) |
2636 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2637 | { | { |
2638 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2639 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2640 | } | } |
2641 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
2642 | if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); | if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); |
2643 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
2644 | { | { |
2645 | int len = 1; | int len = 1; |
2646 | if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } | if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
2647 | if (UCD_CATEGORY(c) != ucp_M) break; | if (UCD_CATEGORY(c) != ucp_M) break; |
2648 | eptr += len; | eptr += len; |
2649 | } | } |
2650 | CHECK_PARTIAL(); | |
2651 | ecode++; | ecode++; |
2652 | break; | break; |
2653 | #endif | #endif |
# | Line 2381 for (;;) | Line 2665 for (;;) |
2665 | case OP_REFI: | case OP_REFI: |
2666 | caseless = op == OP_REFI; | caseless = op == OP_REFI; |
2667 | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ |
2668 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
2669 | ||
2670 | /* If the reference is unset, there are two possibilities: | /* If the reference is unset, there are two possibilities: |
2671 | ||
# | Line 2421 for (;;) | Line 2705 for (;;) |
2705 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
2706 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
2707 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
2708 | max = GET2(ecode, 3); | max = GET2(ecode, 1 + IMM2_SIZE); |
2709 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
2710 | ecode += 5; | ecode += 1 + 2 * IMM2_SIZE; |
2711 | break; | break; |
2712 | ||
2713 | default: /* No repeat follows */ | default: /* No repeat follows */ |
2714 | if ((length = match_ref(offset, eptr, length, md, caseless)) < 0) | if ((length = match_ref(offset, eptr, length, md, caseless)) < 0) |
2715 | { | { |
2716 | if (length == -2) eptr = md->end_subject; /* Partial match */ | |
2717 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
2718 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2719 | } | } |
2720 | eptr += length; | eptr += length; |
2721 | continue; /* With the main loop */ | continue; /* With the main loop */ |
2722 | } | } |
2723 | ||
2724 | /* Handle repeated back references. If the length of the reference is | /* Handle repeated back references. If the length of the reference is |
2725 | zero, just continue with the main loop. */ | zero, just continue with the main loop. If the length is negative, it |
2726 | means the reference is unset in non-Java-compatible mode. If the minimum is | |
2727 | zero, we can continue at the same level without recursion. For any other | |
2728 | minimum, carrying on will result in NOMATCH. */ | |
2729 | ||
2730 | if (length == 0) continue; | if (length == 0) continue; |
2731 | if (length < 0 && min == 0) continue; | |
2732 | ||
2733 | /* First, ensure the minimum number of matches are present. We get back | /* First, ensure the minimum number of matches are present. We get back |
2734 | the length of the reference string explicitly rather than passing the | the length of the reference string explicitly rather than passing the |
# | Line 2450 for (;;) | Line 2739 for (;;) |
2739 | int slength; | int slength; |
2740 | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
2741 | { | { |
2742 | if (slength == -2) eptr = md->end_subject; /* Partial match */ | |
2743 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
2744 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2745 | } | } |
2746 | eptr += slength; | eptr += slength; |
2747 | } | } |
# | Line 2470 for (;;) | Line 2760 for (;;) |
2760 | int slength; | int slength; |
2761 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM14); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM14); |
2762 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
2763 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
2764 | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
2765 | { | { |
2766 | if (slength == -2) eptr = md->end_subject; /* Partial match */ | |
2767 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
2768 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2769 | } | } |
2770 | eptr += slength; | eptr += slength; |
2771 | } | } |
# | Line 2491 for (;;) | Line 2782 for (;;) |
2782 | int slength; | int slength; |
2783 | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
2784 | { | { |
2785 | CHECK_PARTIAL(); | /* Can't use CHECK_PARTIAL because we don't want to update eptr in |
2786 | the soft partial matching case. */ | |
2787 | ||
2788 | if (slength == -2 && md->partial != 0 && | |
2789 | md->end_subject > md->start_used_ptr) | |
2790 | { | |
2791 | md->hitend = TRUE; | |
2792 | if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); | |
2793 | } | |
2794 | break; | break; |
2795 | } | } |
2796 | eptr += slength; | eptr += slength; |
2797 | } | } |
2798 | ||
2799 | while (eptr >= pp) | while (eptr >= pp) |
2800 | { | { |
2801 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM15); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM15); |
2802 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
2803 | eptr -= length; | eptr -= length; |
2804 | } | } |
2805 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2806 | } | } |
2807 | /* Control never gets here */ | /* Control never gets here */ |
2808 | ||
# | Line 2520 for (;;) | Line 2820 for (;;) |
2820 | case OP_NCLASS: | case OP_NCLASS: |
2821 | case OP_CLASS: | case OP_CLASS: |
2822 | { | { |
2823 | /* The data variable is saved across frames, so the byte map needs to | |
2824 | be stored there. */ | |
2825 | #define BYTE_MAP ((pcre_uint8 *)data) | |
2826 | data = ecode + 1; /* Save for matching */ | data = ecode + 1; /* Save for matching */ |
2827 | ecode += 33; /* Advance past the item */ | ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */ |
2828 | ||
2829 | switch (*ecode) | switch (*ecode) |
2830 | { | { |
# | Line 2542 for (;;) | Line 2845 for (;;) |
2845 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
2846 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
2847 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
2848 | max = GET2(ecode, 3); | max = GET2(ecode, 1 + IMM2_SIZE); |
2849 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
2850 | ecode += 5; | ecode += 1 + 2 * IMM2_SIZE; |
2851 | break; | break; |
2852 | ||
2853 | default: /* No repeat follows */ | default: /* No repeat follows */ |
# | Line 2554 for (;;) | Line 2857 for (;;) |
2857 | ||
2858 | /* First, ensure the minimum number of matches are present. */ | /* First, ensure the minimum number of matches are present. */ |
2859 | ||
2860 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
2861 | /* UTF-8 mode */ | if (utf) |
if (utf8) | ||
2862 | { | { |
2863 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
2864 | { | { |
2865 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2866 | { | { |
2867 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2868 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2869 | } | } |
2870 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
2871 | if (c > 255) | if (c > 255) |
2872 | { | { |
2873 | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); |
2874 | } | } |
2875 | else | else |
2876 | { | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); |
if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | ||
} | ||
2877 | } | } |
2878 | } | } |
2879 | else | else |
2880 | #endif | #endif |
2881 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
2882 | { | { |
2883 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
2884 | { | { |
2885 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2886 | { | { |
2887 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2888 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2889 | } | } |
2890 | c = *eptr++; | c = *eptr++; |
2891 | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | #ifndef COMPILE_PCRE8 |
2892 | if (c > 255) | |
2893 | { | |
2894 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | |
2895 | } | |
2896 | else | |
2897 | #endif | |
2898 | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | |
2899 | } | } |
2900 | } | } |
2901 | ||
# | Line 2602 for (;;) | Line 2909 for (;;) |
2909 | ||
2910 | if (minimize) | if (minimize) |
2911 | { | { |
2912 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
2913 | /* UTF-8 mode */ | if (utf) |
if (utf8) | ||
2914 | { | { |
2915 | for (fi = min;; fi++) | for (fi = min;; fi++) |
2916 | { | { |
2917 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM16); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM16); |
2918 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
2919 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
2920 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2921 | { | { |
2922 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2923 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2924 | } | } |
2925 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
2926 | if (c > 255) | if (c > 255) |
2927 | { | { |
2928 | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); |
2929 | } | } |
2930 | else | else |
2931 | { | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); |
if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | ||
} | ||
2932 | } | } |
2933 | } | } |
2934 | else | else |
2935 | #endif | #endif |
2936 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
2937 | { | { |
2938 | for (fi = min;; fi++) | for (fi = min;; fi++) |
2939 | { | { |
2940 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM17); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM17); |
2941 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
2942 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
2943 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
2944 | { | { |
2945 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
2946 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
2947 | } | } |
2948 | c = *eptr++; | c = *eptr++; |
2949 | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | #ifndef COMPILE_PCRE8 |
2950 | if (c > 255) | |
2951 | { | |
2952 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | |
2953 | } | |
2954 | else | |
2955 | #endif | |
2956 | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | |
2957 | } | } |
2958 | } | } |
2959 | /* Control never gets here */ | /* Control never gets here */ |
# | Line 2654 for (;;) | Line 2965 for (;;) |
2965 | { | { |
2966 | pp = eptr; | pp = eptr; |
2967 | ||
2968 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
2969 | /* UTF-8 mode */ | if (utf) |
if (utf8) | ||
2970 | { | { |
2971 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
2972 | { | { |
# | Line 2672 for (;;) | Line 2982 for (;;) |
2982 | if (op == OP_CLASS) break; | if (op == OP_CLASS) break; |
2983 | } | } |
2984 | else | else |
2985 | { | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; |
if ((data[c/8] & (1 << (c&7))) == 0) break; | ||
} | ||
2986 | eptr += len; | eptr += len; |
2987 | } | } |
2988 | for (;;) | for (;;) |
# | Line 2687 for (;;) | Line 2995 for (;;) |
2995 | } | } |
2996 | else | else |
2997 | #endif | #endif |
2998 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
2999 | { | { |
3000 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
3001 | { | { |
# | Line 2697 for (;;) | Line 3005 for (;;) |
3005 | break; | break; |
3006 | } | } |
3007 | c = *eptr; | c = *eptr; |
3008 | if ((data[c/8] & (1 << (c&7))) == 0) break; | #ifndef COMPILE_PCRE8 |
3009 | if (c > 255) | |
3010 | { | |
3011 | if (op == OP_CLASS) break; | |
3012 | } | |
3013 | else | |
3014 | #endif | |
3015 | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; | |
3016 | eptr++; | eptr++; |
3017 | } | } |
3018 | while (eptr >= pp) | while (eptr >= pp) |
# | Line 2708 for (;;) | Line 3023 for (;;) |
3023 | } | } |
3024 | } | } |
3025 | ||
3026 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3027 | } | } |
3028 | #undef BYTE_MAP | |
3029 | } | } |
3030 | /* Control never gets here */ | /* Control never gets here */ |
3031 | ||
# | Line 2718 for (;;) | Line 3034 for (;;) |
3034 | when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8 | when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8 |
3035 | mode, because Unicode properties are supported in non-UTF-8 mode. */ | mode, because Unicode properties are supported in non-UTF-8 mode. */ |
3036 | ||
3037 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
3038 | case OP_XCLASS: | case OP_XCLASS: |
3039 | { | { |
3040 | data = ecode + 1 + LINK_SIZE; /* Save for matching */ | data = ecode + 1 + LINK_SIZE; /* Save for matching */ |
# | Line 2743 for (;;) | Line 3059 for (;;) |
3059 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
3060 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
3061 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
3062 | max = GET2(ecode, 3); | max = GET2(ecode, 1 + IMM2_SIZE); |
3063 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
3064 | ecode += 5; | ecode += 1 + 2 * IMM2_SIZE; |
3065 | break; | break; |
3066 | ||
3067 | default: /* No repeat follows */ | default: /* No repeat follows */ |
# | Line 2760 for (;;) | Line 3076 for (;;) |
3076 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3077 | { | { |
3078 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3079 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3080 | } | } |
3081 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
3082 | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); | if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); |
3083 | } | } |
3084 | ||
3085 | /* If max == min we can continue with the main loop without the | /* If max == min we can continue with the main loop without the |
# | Line 2780 for (;;) | Line 3096 for (;;) |
3096 | { | { |
3097 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM20); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM20); |
3098 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3099 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
3100 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3101 | { | { |
3102 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3103 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3104 | } | } |
3105 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
3106 | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); | if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); |
3107 | } | } |
3108 | /* Control never gets here */ | /* Control never gets here */ |
3109 | } | } |
# | Line 2805 for (;;) | Line 3121 for (;;) |
3121 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3122 | break; | break; |
3123 | } | } |
3124 | #ifdef SUPPORT_UTF | |
3125 | GETCHARLENTEST(c, eptr, len); | GETCHARLENTEST(c, eptr, len); |
3126 | if (!_pcre_xclass(c, data)) break; | #else |
3127 | c = *eptr; | |
3128 | #endif | |
3129 | if (!PRIV(xclass)(c, data, utf)) break; | |
3130 | eptr += len; | eptr += len; |
3131 | } | } |
3132 | for(;;) | for(;;) |
# | Line 2814 for (;;) | Line 3134 for (;;) |
3134 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM21); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM21); |
3135 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3136 | if (eptr-- == pp) break; /* Stop if tried at original pos */ | if (eptr-- == pp) break; /* Stop if tried at original pos */ |
3137 | if (utf8) BACKCHAR(eptr); | #ifdef SUPPORT_UTF |
3138 | if (utf) BACKCHAR(eptr); | |
3139 | #endif | |
3140 | } | } |
3141 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3142 | } | } |
3143 | ||
3144 | /* Control never gets here */ | /* Control never gets here */ |
# | Line 2826 for (;;) | Line 3148 for (;;) |
3148 | /* Match a single character, casefully */ | /* Match a single character, casefully */ |
3149 | ||
3150 | case OP_CHAR: | case OP_CHAR: |
3151 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
3152 | if (utf8) | if (utf) |
3153 | { | { |
3154 | length = 1; | length = 1; |
3155 | ecode++; | ecode++; |
# | Line 2835 for (;;) | Line 3157 for (;;) |
3157 | if (length > md->end_subject - eptr) | if (length > md->end_subject - eptr) |
3158 | { | { |
3159 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ |
3160 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3161 | } | } |
3162 | while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH); | while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH); |
3163 | } | } |
3164 | else | else |
3165 | #endif | #endif |
3166 | /* Not UTF mode */ | |
/* Non-UTF-8 mode */ | ||
3167 | { | { |
3168 | if (md->end_subject - eptr < 1) | if (md->end_subject - eptr < 1) |
3169 | { | { |
3170 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ |
3171 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3172 | } | } |
3173 | if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH); | if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); |
3174 | ecode += 2; | ecode += 2; |
3175 | } | } |
3176 | break; | break; |
3177 | ||
3178 | /* Match a single character, caselessly */ | /* Match a single character, caselessly. If we are at the end of the |
3179 | subject, give up immediately. */ | |
3180 | ||
3181 | case OP_CHARI: | case OP_CHARI: |
3182 | #ifdef SUPPORT_UTF8 | if (eptr >= md->end_subject) |
3183 | if (utf8) | { |
3184 | SCHECK_PARTIAL(); | |
3185 | RRETURN(MATCH_NOMATCH); | |
3186 | } | |
3187 | ||
3188 | #ifdef SUPPORT_UTF | |
3189 | if (utf) | |
3190 | { | { |
3191 | length = 1; | length = 1; |
3192 | ecode++; | ecode++; |
3193 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
3194 | ||
if (length > md->end_subject - eptr) | ||
{ | ||
CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | ||
MRRETURN(MATCH_NOMATCH); | ||
} | ||
3195 | /* If the pattern character's value is < 128, we have only one byte, and | /* If the pattern character's value is < 128, we have only one byte, and |
3196 | can use the fast lookup table. */ | we know that its other case must also be one byte long, so we can use the |
3197 | fast lookup table. We know that there is at least one byte left in the | |
3198 | subject. */ | |
3199 | ||
3200 | if (fc < 128) | if (fc < 128) |
3201 | { | { |
3202 | if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (md->lcc[fc] |
3203 | != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH); | |
3204 | ecode++; | |
3205 | eptr++; | |
3206 | } | } |
3207 | ||
3208 | /* Otherwise we must pick up the subject character */ | /* Otherwise we must pick up the subject character. Note that we cannot |
3209 | use the value of "length" to check for sufficient bytes left, because the | |
3210 | other case of the character may have more or fewer bytes. */ | |
3211 | ||
3212 | else | else |
3213 | { | { |
# | Line 2894 for (;;) | Line 3223 for (;;) |
3223 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
3224 | if (dc != UCD_OTHERCASE(fc)) | if (dc != UCD_OTHERCASE(fc)) |
3225 | #endif | #endif |
3226 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3227 | } | } |
3228 | } | } |
3229 | } | } |
3230 | else | else |
3231 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF */ |
3232 | ||
3233 | /* Non-UTF-8 mode */ | /* Not UTF mode */ |
3234 | { | { |
3235 | if (md->end_subject - eptr < 1) | if (TABLE_GET(ecode[1], md->lcc, ecode[1]) |
3236 | { | != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH); |
3237 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | eptr++; |
MRRETURN(MATCH_NOMATCH); | ||
} | ||
if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | ||
3238 | ecode += 2; | ecode += 2; |
3239 | } | } |
3240 | break; | break; |
# | Line 2918 for (;;) | Line 3244 for (;;) |
3244 | case OP_EXACT: | case OP_EXACT: |
3245 | case OP_EXACTI: | case OP_EXACTI: |
3246 | min = max = GET2(ecode, 1); | min = max = GET2(ecode, 1); |
3247 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
3248 | goto REPEATCHAR; | goto REPEATCHAR; |
3249 | ||
3250 | case OP_POSUPTO: | case OP_POSUPTO: |
# | Line 2933 for (;;) | Line 3259 for (;;) |
3259 | min = 0; | min = 0; |
3260 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
3261 | minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI; | minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI; |
3262 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
3263 | goto REPEATCHAR; | goto REPEATCHAR; |
3264 | ||
3265 | case OP_POSSTAR: | case OP_POSSTAR: |
# | Line 2981 for (;;) | Line 3307 for (;;) |
3307 | /* Common code for all repeated single-character matches. */ | /* Common code for all repeated single-character matches. */ |
3308 | ||
3309 | REPEATCHAR: | REPEATCHAR: |
3310 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
3311 | if (utf8) | if (utf) |
3312 | { | { |
3313 | length = 1; | length = 1; |
3314 | charptr = ecode; | charptr = ecode; |
# | Line 2998 for (;;) | Line 3324 for (;;) |
3324 | unsigned int othercase; | unsigned int othercase; |
3325 | if (op >= OP_STARI && /* Caseless */ | if (op >= OP_STARI && /* Caseless */ |
3326 | (othercase = UCD_OTHERCASE(fc)) != fc) | (othercase = UCD_OTHERCASE(fc)) != fc) |
3327 | oclength = _pcre_ord2utf8(othercase, occhars); | oclength = PRIV(ord2utf)(othercase, occhars); |
3328 | else oclength = 0; | else oclength = 0; |
3329 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
3330 | ||
3331 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
3332 | { | { |
3333 | if (eptr <= md->end_subject - length && | if (eptr <= md->end_subject - length && |
3334 | memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; |
3335 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
3336 | else if (oclength > 0 && | else if (oclength > 0 && |
3337 | eptr <= md->end_subject - oclength && | eptr <= md->end_subject - oclength && |
3338 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; |
3339 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
3340 | else | else |
3341 | { | { |
3342 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
3343 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3344 | } | } |
3345 | } | } |
3346 | ||
# | Line 3026 for (;;) | Line 3352 for (;;) |
3352 | { | { |
3353 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM22); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM22); |
3354 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3355 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
3356 | if (eptr <= md->end_subject - length && | if (eptr <= md->end_subject - length && |
3357 | memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; |
3358 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
3359 | else if (oclength > 0 && | else if (oclength > 0 && |
3360 | eptr <= md->end_subject - oclength && | eptr <= md->end_subject - oclength && |
3361 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; |
3362 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
3363 | else | else |
3364 | { | { |
3365 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
3366 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3367 | } | } |
3368 | } | } |
3369 | /* Control never gets here */ | /* Control never gets here */ |
# | Line 3049 for (;;) | Line 3375 for (;;) |
3375 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
3376 | { | { |
3377 | if (eptr <= md->end_subject - length && | if (eptr <= md->end_subject - length && |
3378 | memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; |
3379 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
3380 | else if (oclength > 0 && | else if (oclength > 0 && |
3381 | eptr <= md->end_subject - oclength && | eptr <= md->end_subject - oclength && |
3382 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; |
3383 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
3384 | else | else |
3385 | { | { |
# | Line 3068 for (;;) | Line 3394 for (;;) |
3394 | { | { |
3395 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM23); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM23); |
3396 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3397 | if (eptr == pp) { MRRETURN(MATCH_NOMATCH); } | if (eptr == pp) { RRETURN(MATCH_NOMATCH); } |
3398 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
3399 | eptr--; | eptr--; |
3400 | BACKCHAR(eptr); | BACKCHAR(eptr); |
# | Line 3085 for (;;) | Line 3411 for (;;) |
3411 | value of fc will always be < 128. */ | value of fc will always be < 128. */ |
3412 | } | } |
3413 | else | else |
3414 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF */ |
3415 | /* When not in UTF-8 mode, load a single-byte character. */ | |
3416 | /* When not in UTF-8 mode, load a single-byte character. */ | fc = *ecode++; |
fc = *ecode++; | ||
3417 | ||
3418 | /* The value of fc at this point is always less than 256, though we may or | /* The value of fc at this point is always one character, though we may |
3419 | may not be in UTF-8 mode. The code is duplicated for the caseless and | or may not be in UTF mode. The code is duplicated for the caseless and |
3420 | caseful cases, for speed, since matching characters is likely to be quite | caseful cases, for speed, since matching characters is likely to be quite |
3421 | common. First, ensure the minimum number of matches are present. If min = | common. First, ensure the minimum number of matches are present. If min = |
3422 | max, continue at the same level without recursing. Otherwise, if | max, continue at the same level without recursing. Otherwise, if |
# | Line 3105 for (;;) | Line 3429 for (;;) |
3429 | ||
3430 | if (op >= OP_STARI) /* Caseless */ | if (op >= OP_STARI) /* Caseless */ |
3431 | { | { |
3432 | fc = md->lcc[fc]; | #ifdef COMPILE_PCRE8 |
3433 | /* fc must be < 128 if UTF is enabled. */ | |
3434 | foc = md->fcc[fc]; | |
3435 | #else | |
3436 | #ifdef SUPPORT_UTF | |
3437 | #ifdef SUPPORT_UCP | |
3438 | if (utf && fc > 127) | |
3439 | foc = UCD_OTHERCASE(fc); | |
3440 | #else | |
3441 | if (utf && fc > 127) | |
3442 | foc = fc; | |
3443 | #endif /* SUPPORT_UCP */ | |
3444 | else | |
3445 | #endif /* SUPPORT_UTF */ | |
3446 | foc = TABLE_GET(fc, md->fcc, fc); | |
3447 | #endif /* COMPILE_PCRE8 */ | |
3448 | ||
3449 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
3450 | { | { |
3451 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3452 | { | { |
3453 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3454 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3455 | } | } |
3456 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH); |
3457 | eptr++; | |
3458 | } | } |
3459 | if (min == max) continue; | if (min == max) continue; |
3460 | if (minimize) | if (minimize) |
# | Line 3122 for (;;) | Line 3463 for (;;) |
3463 | { | { |
3464 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM24); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM24); |
3465 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3466 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
3467 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3468 | { | { |
3469 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3470 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3471 | } | } |
3472 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH); |
3473 | eptr++; | |
3474 | } | } |
3475 | /* Control never gets here */ | /* Control never gets here */ |
3476 | } | } |
# | Line 3142 for (;;) | Line 3484 for (;;) |
3484 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3485 | break; | break; |
3486 | } | } |
3487 | if (fc != md->lcc[*eptr]) break; | if (fc != *eptr && foc != *eptr) break; |
3488 | eptr++; | eptr++; |
3489 | } | } |
3490 | ||
# | Line 3154 for (;;) | Line 3496 for (;;) |
3496 | eptr--; | eptr--; |
3497 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3498 | } | } |
3499 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3500 | } | } |
3501 | /* Control never gets here */ | /* Control never gets here */ |
3502 | } | } |
# | Line 3168 for (;;) | Line 3510 for (;;) |
3510 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3511 | { | { |
3512 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3513 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3514 | } | } |
3515 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr++) RRETURN(MATCH_NOMATCH); |
3516 | } | } |
3517 | ||
3518 | if (min == max) continue; | if (min == max) continue; |
# | Line 3181 for (;;) | Line 3523 for (;;) |
3523 | { | { |
3524 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM26); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM26); |
3525 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3526 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
3527 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3528 | { | { |
3529 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3530 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3531 | } | } |
3532 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr++) RRETURN(MATCH_NOMATCH); |
3533 | } | } |
3534 | /* Control never gets here */ | /* Control never gets here */ |
3535 | } | } |
# | Line 3212 for (;;) | Line 3554 for (;;) |
3554 | eptr--; | eptr--; |
3555 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3556 | } | } |
3557 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3558 | } | } |
3559 | } | } |
3560 | /* Control never gets here */ | /* Control never gets here */ |
# | Line 3225 for (;;) | Line 3567 for (;;) |
3567 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3568 | { | { |
3569 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3570 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3571 | } | } |
3572 | ecode++; | #ifdef SUPPORT_UTF |
3573 | GETCHARINCTEST(c, eptr); | if (utf) |
if (op == OP_NOTI) /* The caseless case */ | ||
3574 | { | { |
3575 | #ifdef SUPPORT_UTF8 | register unsigned int ch, och; |
3576 | if (c < 256) | |
3577 | #endif | ecode++; |
3578 | c = md->lcc[c]; | GETCHARINC(ch, ecode); |
3579 | if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH); | GETCHARINC(c, eptr); |
3580 | ||
3581 | if (op == OP_NOT) | |
3582 | { | |
3583 | if (ch == c) RRETURN(MATCH_NOMATCH); | |
3584 | } | |
3585 | else | |
3586 | { | |
3587 | #ifdef SUPPORT_UCP | |
3588 | if (ch > 127) | |
3589 | och = UCD_OTHERCASE(ch); | |
3590 | #else | |
3591 | if (ch > 127) | |
3592 | och = ch; | |
3593 | #endif /* SUPPORT_UCP */ | |
3594 | else | |
3595 | och = TABLE_GET(ch, md->fcc, ch); | |
3596 | if (ch == c || och == c) RRETURN(MATCH_NOMATCH); | |
3597 | } | |
3598 | } | } |
3599 | else /* Caseful */ | else |
3600 | #endif | |
3601 | { | { |
3602 | if (*ecode++ == c) MRRETURN(MATCH_NOMATCH); | register unsigned int ch = ecode[1]; |
3603 | c = *eptr++; | |
3604 | if (ch == c || (op == OP_NOTI && TABLE_GET(ch, md->fcc, ch) == c)) | |
3605 | RRETURN(MATCH_NOMATCH); | |
3606 | ecode += 2; | |
3607 | } | } |
3608 | break; | break; |
3609 | ||
# | Line 3253 for (;;) | Line 3617 for (;;) |
3617 | case OP_NOTEXACT: | case OP_NOTEXACT: |
3618 | case OP_NOTEXACTI: | case OP_NOTEXACTI: |
3619 | min = max = GET2(ecode, 1); | min = max = GET2(ecode, 1); |
3620 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
3621 | goto REPEATNOTCHAR; | goto REPEATNOTCHAR; |
3622 | ||
3623 | case OP_NOTUPTO: | case OP_NOTUPTO: |
# | Line 3263 for (;;) | Line 3627 for (;;) |
3627 | min = 0; | min = 0; |
3628 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
3629 | minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI; | minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI; |
3630 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
3631 | goto REPEATNOTCHAR; | goto REPEATNOTCHAR; |
3632 | ||
3633 | case OP_NOTPOSSTAR: | case OP_NOTPOSSTAR: |
# | Line 3295 for (;;) | Line 3659 for (;;) |
3659 | possessive = TRUE; | possessive = TRUE; |
3660 | min = 0; | min = 0; |
3661 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
3662 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
3663 | goto REPEATNOTCHAR; | goto REPEATNOTCHAR; |
3664 | ||
3665 | case OP_NOTSTAR: | case OP_NOTSTAR: |
# | Line 3319 for (;;) | Line 3683 for (;;) |
3683 | /* Common code for all repeated single-byte matches. */ | /* Common code for all repeated single-byte matches. */ |
3684 | ||
3685 | REPEATNOTCHAR: | REPEATNOTCHAR: |
3686 | fc = *ecode++; | GETCHARINCTEST(fc, ecode); |
3687 | ||
3688 | /* The code is duplicated for the caseless and caseful cases, for speed, | /* The code is duplicated for the caseless and caseful cases, for speed, |
3689 | since matching characters is likely to be quite common. First, ensure the | since matching characters is likely to be quite common. First, ensure the |
# | Line 3334 for (;;) | Line 3698 for (;;) |
3698 | ||
3699 | if (op >= OP_NOTSTARI) /* Caseless */ | if (op >= OP_NOTSTARI) /* Caseless */ |
3700 | { | { |
3701 | fc = md->lcc[fc]; | #ifdef SUPPORT_UTF |
3702 | #ifdef SUPPORT_UCP | |
3703 | if (utf && fc > 127) | |
3704 | foc = UCD_OTHERCASE(fc); | |
3705 | #else | |
3706 | if (utf && fc > 127) | |
3707 | foc = fc; | |
3708 | #endif /* SUPPORT_UCP */ | |
3709 | else | |
3710 | #endif /* SUPPORT_UTF */ | |
3711 | foc = TABLE_GET(fc, md->fcc, fc); | |
3712 | ||
3713 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
3714 | /* UTF-8 mode */ | if (utf) |
if (utf8) | ||
3715 | { | { |
3716 | register unsigned int d; | register unsigned int d; |
3717 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
# | Line 3346 for (;;) | Line 3719 for (;;) |
3719 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3720 | { | { |
3721 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3722 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3723 | } | } |
3724 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
3725 | if (d < 256) d = md->lcc[d]; | if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH); |
if (fc == d) MRRETURN(MATCH_NOMATCH); | ||
3726 | } | } |
3727 | } | } |
3728 | else | else |
3729 | #endif | #endif |
3730 | /* Not UTF mode */ | |
/* Not UTF-8 mode */ | ||
3731 | { | { |
3732 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
3733 | { | { |
3734 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3735 | { | { |
3736 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3737 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3738 | } | } |
3739 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); |
3740 | eptr++; | |
3741 | } | } |
3742 | } | } |
3743 | ||
# | Line 3373 for (;;) | Line 3745 for (;;) |
3745 | ||
3746 | if (minimize) | if (minimize) |
3747 | { | { |
3748 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
3749 | /* UTF-8 mode */ | if (utf) |
if (utf8) | ||
3750 | { | { |
3751 | register unsigned int d; | register unsigned int d; |
3752 | for (fi = min;; fi++) | for (fi = min;; fi++) |
3753 | { | { |
3754 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM28); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM28); |
3755 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3756 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
3757 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3758 | { | { |
3759 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3760 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3761 | } | } |
3762 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
3763 | if (d < 256) d = md->lcc[d]; | if (fc == d || (unsigned int)foc == d) RRETURN(MATCH_NOMATCH); |
if (fc == d) MRRETURN(MATCH_NOMATCH); | ||
3764 | } | } |
3765 | } | } |
3766 | else | else |
3767 | #endif | #endif |
3768 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
3769 | { | { |
3770 | for (fi = min;; fi++) | for (fi = min;; fi++) |
3771 | { | { |
3772 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM29); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM29); |
3773 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3774 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
3775 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3776 | { | { |
3777 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3778 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3779 | } | } |
3780 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); |
3781 | eptr++; | |
3782 | } | } |
3783 | } | } |
3784 | /* Control never gets here */ | /* Control never gets here */ |
# | Line 3419 for (;;) | Line 3790 for (;;) |
3790 | { | { |
3791 | pp = eptr; | pp = eptr; |
3792 | ||
3793 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
3794 | /* UTF-8 mode */ | if (utf) |
if (utf8) | ||
3795 | { | { |
3796 | register unsigned int d; | register unsigned int d; |
3797 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
# | Line 3433 for (;;) | Line 3803 for (;;) |
3803 | break; | break; |
3804 | } | } |
3805 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
3806 | if (d < 256) d = md->lcc[d]; | if (fc == d || (unsigned int)foc == d) break; |
if (fc == d) break; | ||
3807 | eptr += len; | eptr += len; |
3808 | } | } |
3809 | if (possessive) continue; | if (possessive) continue; |
3810 | for(;;) | for(;;) |
3811 | { | { |
3812 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM30); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM30); |
3813 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
# | Line 3448 for (;;) | Line 3817 for (;;) |
3817 | } | } |
3818 | else | else |
3819 | #endif | #endif |
3820 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
3821 | { | { |
3822 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
3823 | { | { |
# | Line 3457 for (;;) | Line 3826 for (;;) |
3826 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3827 | break; | break; |
3828 | } | } |
3829 | if (fc == md->lcc[*eptr]) break; | if (fc == *eptr || foc == *eptr) break; |
3830 | eptr++; | eptr++; |
3831 | } | } |
3832 | if (possessive) continue; | if (possessive) continue; |
# | Line 3469 for (;;) | Line 3838 for (;;) |
3838 | } | } |
3839 | } | } |
3840 | ||
3841 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3842 | } | } |
3843 | /* Control never gets here */ | /* Control never gets here */ |
3844 | } | } |
# | Line 3478 for (;;) | Line 3847 for (;;) |
3847 | ||
3848 | else | else |
3849 | { | { |
3850 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
3851 | /* UTF-8 mode */ | if (utf) |
if (utf8) | ||
3852 | { | { |
3853 | register unsigned int d; | register unsigned int d; |
3854 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
# | Line 3488 for (;;) | Line 3856 for (;;) |
3856 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3857 | { | { |
3858 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3859 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3860 | } | } |
3861 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
3862 | if (fc == d) MRRETURN(MATCH_NOMATCH); | if (fc == d) RRETURN(MATCH_NOMATCH); |
3863 | } | } |
3864 | } | } |
3865 | else | else |
3866 | #endif | #endif |
3867 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
3868 | { | { |
3869 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
3870 | { | { |
3871 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3872 | { | { |
3873 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3874 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3875 | } | } |
3876 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); |
3877 | } | } |
3878 | } | } |
3879 | ||
# | Line 3513 for (;;) | Line 3881 for (;;) |
3881 | ||
3882 | if (minimize) | if (minimize) |
3883 | { | { |
3884 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
3885 | /* UTF-8 mode */ | if (utf) |
if (utf8) | ||
3886 | { | { |
3887 | register unsigned int d; | register unsigned int d; |
3888 | for (fi = min;; fi++) | for (fi = min;; fi++) |
3889 | { | { |
3890 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM32); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM32); |
3891 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3892 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
3893 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3894 | { | { |
3895 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3896 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3897 | } | } |
3898 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
3899 | if (fc == d) MRRETURN(MATCH_NOMATCH); | if (fc == d) RRETURN(MATCH_NOMATCH); |
3900 | } | } |
3901 | } | } |
3902 | else | else |
3903 | #endif | #endif |
3904 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
3905 | { | { |
3906 | for (fi = min;; fi++) | for (fi = min;; fi++) |
3907 | { | { |
3908 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM33); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM33); |
3909 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
3910 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
3911 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
3912 | { | { |
3913 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
3914 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3915 | } | } |
3916 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); |
3917 | } | } |
3918 | } | } |
3919 | /* Control never gets here */ | /* Control never gets here */ |
# | Line 3558 for (;;) | Line 3925 for (;;) |
3925 | { | { |
3926 | pp = eptr; | pp = eptr; |
3927 | ||
3928 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
3929 | /* UTF-8 mode */ | if (utf) |
if (utf8) | ||
3930 | { | { |
3931 | register unsigned int d; | register unsigned int d; |
3932 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
# | Line 3586 for (;;) | Line 3952 for (;;) |
3952 | } | } |
3953 | else | else |
3954 | #endif | #endif |
3955 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
3956 | { | { |
3957 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
3958 | { | { |
# | Line 3607 for (;;) | Line 3973 for (;;) |
3973 | } | } |
3974 | } | } |
3975 | ||
3976 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
3977 | } | } |
3978 | } | } |
3979 | /* Control never gets here */ | /* Control never gets here */ |
# | Line 3619 for (;;) | Line 3985 for (;;) |
3985 | case OP_TYPEEXACT: | case OP_TYPEEXACT: |
3986 | min = max = GET2(ecode, 1); | min = max = GET2(ecode, 1); |
3987 | minimize = TRUE; | minimize = TRUE; |
3988 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
3989 | goto REPEATTYPE; | goto REPEATTYPE; |
3990 | ||
3991 | case OP_TYPEUPTO: | case OP_TYPEUPTO: |
# | Line 3627 for (;;) | Line 3993 for (;;) |
3993 | min = 0; | min = 0; |
3994 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
3995 | minimize = *ecode == OP_TYPEMINUPTO; | minimize = *ecode == OP_TYPEMINUPTO; |
3996 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
3997 | goto REPEATTYPE; | goto REPEATTYPE; |
3998 | ||
3999 | case OP_TYPEPOSSTAR: | case OP_TYPEPOSSTAR: |
# | Line 3655 for (;;) | Line 4021 for (;;) |
4021 | possessive = TRUE; | possessive = TRUE; |
4022 | min = 0; | min = 0; |
4023 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
4024 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
4025 | goto REPEATTYPE; | goto REPEATTYPE; |
4026 | ||
4027 | case OP_TYPESTAR: | case OP_TYPESTAR: |
# | Line 3701 for (;;) | Line 4067 for (;;) |
4067 | switch(prop_type) | switch(prop_type) |
4068 | { | { |
4069 | case PT_ANY: | case PT_ANY: |
4070 | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); | if (prop_fail_result) RRETURN(MATCH_NOMATCH); |
4071 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
4072 | { | { |
4073 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4074 | { | { |
4075 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4076 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4077 | } | } |
4078 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4079 | } | } |
# | Line 3720 for (;;) | Line 4086 for (;;) |
4086 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4087 | { | { |
4088 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4089 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4090 | } | } |
4091 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4092 | chartype = UCD_CHARTYPE(c); | chartype = UCD_CHARTYPE(c); |
4093 | if ((chartype == ucp_Lu || | if ((chartype == ucp_Lu || |
4094 | chartype == ucp_Ll || | chartype == ucp_Ll || |
4095 | chartype == ucp_Lt) == prop_fail_result) | chartype == ucp_Lt) == prop_fail_result) |
4096 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4097 | } | } |
4098 | break; | break; |
4099 | ||
# | Line 3737 for (;;) | Line 4103 for (;;) |
4103 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4104 | { | { |
4105 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4106 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4107 | } | } |
4108 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4109 | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) |
4110 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4111 | } | } |
4112 | break; | break; |
4113 | ||
# | Line 3751 for (;;) | Line 4117 for (;;) |
4117 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4118 | { | { |
4119 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4120 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4121 | } | } |
4122 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4123 | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) |
4124 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4125 | } | } |
4126 | break; | break; |
4127 | ||
# | Line 3765 for (;;) | Line 4131 for (;;) |
4131 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4132 | { | { |
4133 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4134 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4135 | } | } |
4136 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4137 | if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) | if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) |
4138 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4139 | } | } |
4140 | break; | break; |
4141 | ||
# | Line 3780 for (;;) | Line 4146 for (;;) |
4146 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4147 | { | { |
4148 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4149 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4150 | } | } |
4151 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4152 | category = UCD_CATEGORY(c); | category = UCD_CATEGORY(c); |
4153 | if ((category == ucp_L || category == ucp_N) == prop_fail_result) | if ((category == ucp_L || category == ucp_N) == prop_fail_result) |
4154 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4155 | } | } |
4156 | break; | break; |
4157 | ||
# | Line 3795 for (;;) | Line 4161 for (;;) |
4161 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4162 | { | { |
4163 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4164 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4165 | } | } |
4166 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4167 | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
4168 | c == CHAR_FF || c == CHAR_CR) | c == CHAR_FF || c == CHAR_CR) |
4169 | == prop_fail_result) | == prop_fail_result) |
4170 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4171 | } | } |
4172 | break; | break; |
4173 | ||
# | Line 3811 for (;;) | Line 4177 for (;;) |
4177 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4178 | { | { |
4179 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4180 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4181 | } | } |
4182 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4183 | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
4184 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
4185 | == prop_fail_result) | == prop_fail_result) |
4186 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4187 | } | } |
4188 | break; | break; |
4189 | ||
# | Line 3828 for (;;) | Line 4194 for (;;) |
4194 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4195 | { | { |
4196 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4197 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4198 | } | } |
4199 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4200 | category = UCD_CATEGORY(c); | category = UCD_CATEGORY(c); |
4201 | if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE) | if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE) |
4202 | == prop_fail_result) | == prop_fail_result) |
4203 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4204 | } | } |
4205 | break; | break; |
4206 | ||
# | Line 3855 for (;;) | Line 4221 for (;;) |
4221 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4222 | { | { |
4223 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4224 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4225 | } | } |
4226 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4227 | if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); | if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); |
4228 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
4229 | { | { |
4230 | int len = 1; | int len = 1; |
4231 | if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } | if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
4232 | if (UCD_CATEGORY(c) != ucp_M) break; | if (UCD_CATEGORY(c) != ucp_M) break; |
4233 | eptr += len; | eptr += len; |
4234 | } | } |
4235 | CHECK_PARTIAL(); | |
4236 | } | } |
4237 | } | } |
4238 | ||
# | Line 3874 for (;;) | Line 4241 for (;;) |
4241 | ||
4242 | /* Handle all other cases when the coding is UTF-8 */ | /* Handle all other cases when the coding is UTF-8 */ |
4243 | ||
4244 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
4245 | if (utf8) switch(ctype) | if (utf) switch(ctype) |
4246 | { | { |
4247 | case OP_ANY: | case OP_ANY: |
4248 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
# | Line 3883 for (;;) | Line 4250 for (;;) |
4250 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4251 | { | { |
4252 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4253 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4254 | } | |
4255 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | |
4256 | if (md->partial != 0 && | |
4257 | eptr + 1 >= md->end_subject && | |
4258 | NLBLOCK->nltype == NLTYPE_FIXED && | |
4259 | NLBLOCK->nllen == 2 && | |
4260 | *eptr == NLBLOCK->nl[0]) | |
4261 | { | |
4262 | md->hitend = TRUE; | |
4263 | if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); | |
4264 | } | } |
if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | ||
4265 | eptr++; | eptr++; |
4266 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); |
4267 | } | } |
4268 | break; | break; |
4269 | ||
# | Line 3897 for (;;) | Line 4273 for (;;) |
4273 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4274 | { | { |
4275 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4276 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4277 | } | } |
4278 | eptr++; | eptr++; |
4279 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); |
4280 | } | } |
4281 | break; | break; |
4282 | ||
4283 | case OP_ANYBYTE: | case OP_ANYBYTE: |
4284 | if (eptr > md->end_subject - min) MRRETURN(MATCH_NOMATCH); | if (eptr > md->end_subject - min) RRETURN(MATCH_NOMATCH); |
4285 | eptr += min; | eptr += min; |
4286 | break; | break; |
4287 | ||
# | Line 3915 for (;;) | Line 4291 for (;;) |
4291 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4292 | { | { |
4293 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4294 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4295 | } | } |
4296 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
4297 | switch(c) | switch(c) |
4298 | { | { |
4299 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
4300 | ||
4301 | case 0x000d: | case 0x000d: |
4302 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
# | Line 3934 for (;;) | Line 4310 for (;;) |
4310 | case 0x0085: | case 0x0085: |
4311 | case 0x2028: | case 0x2028: |
4312 | case 0x2029: | case 0x2029: |
4313 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); |
4314 | break; | break; |
4315 | } | } |
4316 | } | } |
# | Line 3946 for (;;) | Line 4322 for (;;) |
4322 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4323 | { | { |
4324 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4325 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4326 | } | } |
4327 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
4328 | switch(c) | switch(c) |
# | Line 3971 for (;;) | Line 4347 for (;;) |
4347 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
4348 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
4349 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
4350 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4351 | } | } |
4352 | } | } |
4353 | break; | break; |
# | Line 3982 for (;;) | Line 4358 for (;;) |
4358 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4359 | { | { |
4360 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4361 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4362 | } | } |
4363 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
4364 | switch(c) | switch(c) |
4365 | { | { |
4366 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
4367 | case 0x09: /* HT */ | case 0x09: /* HT */ |
4368 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
4369 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
# | Line 4018 for (;;) | Line 4394 for (;;) |
4394 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4395 | { | { |
4396 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4397 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4398 | } | } |
4399 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
4400 | switch(c) | switch(c) |
# | Line 4031 for (;;) | Line 4407 for (;;) |
4407 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
4408 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
4409 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
4410 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4411 | } | } |
4412 | } | } |
4413 | break; | break; |
# | Line 4042 for (;;) | Line 4418 for (;;) |
4418 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4419 | { | { |
4420 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4421 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4422 | } | } |
4423 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
4424 | switch(c) | switch(c) |
4425 | { | { |
4426 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
4427 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
4428 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
4429 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
# | Line 4066 for (;;) | Line 4442 for (;;) |
4442 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4443 | { | { |
4444 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4445 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4446 | } | } |
4447 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
4448 | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) |
4449 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4450 | } | } |
4451 | break; | break; |
4452 | ||
# | Line 4080 for (;;) | Line 4456 for (;;) |
4456 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4457 | { | { |
4458 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4459 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4460 | } | } |
4461 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_digit) == 0) |
4462 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4463 | eptr++; | |
4464 | /* No need to skip more bytes - we know it's a 1-byte character */ | /* No need to skip more bytes - we know it's a 1-byte character */ |
4465 | } | } |
4466 | break; | break; |
# | Line 4094 for (;;) | Line 4471 for (;;) |
4471 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4472 | { | { |
4473 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4474 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4475 | } | } |
4476 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) |
4477 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4478 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | eptr++; |
4479 | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); | |
4480 | } | } |
4481 | break; | break; |
4482 | ||
# | Line 4108 for (;;) | Line 4486 for (;;) |
4486 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4487 | { | { |
4488 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4489 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4490 | } | } |
4491 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_space) == 0) |
4492 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4493 | eptr++; | |
4494 | /* No need to skip more bytes - we know it's a 1-byte character */ | /* No need to skip more bytes - we know it's a 1-byte character */ |
4495 | } | } |
4496 | break; | break; |
# | Line 4122 for (;;) | Line 4501 for (;;) |
4501 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4502 | { | { |
4503 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4504 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4505 | } | } |
4506 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) |
4507 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4508 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | eptr++; |
4509 | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); | |
4510 | } | } |
4511 | break; | break; |
4512 | ||
# | Line 4136 for (;;) | Line 4516 for (;;) |
4516 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4517 | { | { |
4518 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4519 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4520 | } | } |
4521 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_word) == 0) |
4522 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4523 | eptr++; | |
4524 | /* No need to skip more bytes - we know it's a 1-byte character */ | /* No need to skip more bytes - we know it's a 1-byte character */ |
4525 | } | } |
4526 | break; | break; |
# | Line 4149 for (;;) | Line 4530 for (;;) |
4530 | } /* End switch(ctype) */ | } /* End switch(ctype) */ |
4531 | ||
4532 | else | else |
4533 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF */ |
4534 | ||
4535 | /* Code for the non-UTF-8 case for minimum matching of operators other | /* Code for the non-UTF-8 case for minimum matching of operators other |
4536 | than OP_PROP and OP_NOTPROP. */ | than OP_PROP and OP_NOTPROP. */ |
# | Line 4162 for (;;) | Line 4543 for (;;) |
4543 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4544 | { | { |
4545 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4546 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4547 | } | |
4548 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | |
4549 | if (md->partial != 0 && | |
4550 | eptr + 1 >= md->end_subject && | |
4551 | NLBLOCK->nltype == NLTYPE_FIXED && | |
4552 | NLBLOCK->nllen == 2 && | |
4553 | *eptr == NLBLOCK->nl[0]) | |
4554 | { | |
4555 | md->hitend = TRUE; | |
4556 | if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); | |
4557 | } | } |
if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | ||
4558 | eptr++; | eptr++; |
4559 | } | } |
4560 | break; | break; |
# | Line 4173 for (;;) | Line 4563 for (;;) |
4563 | if (eptr > md->end_subject - min) | if (eptr > md->end_subject - min) |
4564 | { | { |
4565 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4566 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4567 | } | } |
4568 | eptr += min; | eptr += min; |
4569 | break; | break; |
# | Line 4182 for (;;) | Line 4572 for (;;) |
4572 | if (eptr > md->end_subject - min) | if (eptr > md->end_subject - min) |
4573 | { | { |
4574 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4575 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4576 | } | } |
4577 | eptr += min; | eptr += min; |
4578 | break; | break; |
# | Line 4193 for (;;) | Line 4583 for (;;) |
4583 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4584 | { | { |
4585 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4586 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4587 | } | } |
4588 | switch(*eptr++) | switch(*eptr++) |
4589 | { | { |
4590 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
4591 | ||
4592 | case 0x000d: | case 0x000d: |
4593 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
# | Line 4209 for (;;) | Line 4599 for (;;) |
4599 | case 0x000b: | case 0x000b: |
4600 | case 0x000c: | case 0x000c: |
4601 | case 0x0085: | case 0x0085: |
4602 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
4603 | case 0x2028: | |
4604 | case 0x2029: | |
4605 | #endif | |
4606 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | |
4607 | break; | break; |
4608 | } | } |
4609 | } | } |
# | Line 4221 for (;;) | Line 4615 for (;;) |
4615 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4616 | { | { |
4617 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4618 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4619 | } | } |
4620 | switch(*eptr++) | switch(*eptr++) |
4621 | { | { |
# | Line 4229 for (;;) | Line 4623 for (;;) |
4623 | case 0x09: /* HT */ | case 0x09: /* HT */ |
4624 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
4625 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
4626 | MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
4627 | case 0x1680: /* OGHAM SPACE MARK */ | |
4628 | case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ | |
4629 | case 0x2000: /* EN QUAD */ | |
4630 | case 0x2001: /* EM QUAD */ | |
4631 | case 0x2002: /* EN SPACE */ | |
4632 | case 0x2003: /* EM SPACE */ | |
4633 | case 0x2004: /* THREE-PER-EM SPACE */ | |
4634 | case 0x2005: /* FOUR-PER-EM SPACE */ | |
4635 | case 0x2006: /* SIX-PER-EM SPACE */ | |
4636 | case 0x2007: /* FIGURE SPACE */ | |
4637 | case 0x2008: /* PUNCTUATION SPACE */ | |
4638 | case 0x2009: /* THIN SPACE */ | |
4639 | case 0x200A: /* HAIR SPACE */ | |
4640 | case 0x202f: /* NARROW NO-BREAK SPACE */ | |
4641 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | |
4642 | case 0x3000: /* IDEOGRAPHIC SPACE */ | |
4643 | #endif | |
4644 | RRETURN(MATCH_NOMATCH); | |
4645 | } | } |
4646 | } | } |
4647 | break; | break; |
# | Line 4240 for (;;) | Line 4652 for (;;) |
4652 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4653 | { | { |
4654 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4655 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4656 | } | } |
4657 | switch(*eptr++) | switch(*eptr++) |
4658 | { | { |
4659 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
4660 | case 0x09: /* HT */ | case 0x09: /* HT */ |
4661 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
4662 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
4663 | #ifdef COMPILE_PCRE16 | |
4664 | case 0x1680: /* OGHAM SPACE MARK */ | |
4665 | case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ | |
4666 | case 0x2000: /* EN QUAD */ | |
4667 | case 0x2001: /* EM QUAD */ | |
4668 | case 0x2002: /* EN SPACE */ | |
4669 | case 0x2003: /* EM SPACE */ | |
4670 | case 0x2004: /* THREE-PER-EM SPACE */ | |
4671 | case 0x2005: /* FOUR-PER-EM SPACE */ | |
4672 | case 0x2006: /* SIX-PER-EM SPACE */ | |
4673 | case 0x2007: /* FIGURE SPACE */ | |
4674 | case 0x2008: /* PUNCTUATION SPACE */ | |
4675 | case 0x2009: /* THIN SPACE */ | |
4676 | case 0x200A: /* HAIR SPACE */ | |
4677 | case 0x202f: /* NARROW NO-BREAK SPACE */ | |
4678 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | |
4679 | case 0x3000: /* IDEOGRAPHIC SPACE */ | |
4680 | #endif | |
4681 | break; | break; |
4682 | } | } |
4683 | } | } |
# | Line 4259 for (;;) | Line 4689 for (;;) |
4689 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4690 | { | { |
4691 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4692 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4693 | } | } |
4694 | switch(*eptr++) | switch(*eptr++) |
4695 | { | { |
# | Line 4269 for (;;) | Line 4699 for (;;) |
4699 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
4700 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
4701 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
4702 | MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
4703 | case 0x2028: /* LINE SEPARATOR */ | |
4704 | case 0x2029: /* PARAGRAPH SEPARATOR */ | |
4705 | #endif | |
4706 | RRETURN(MATCH_NOMATCH); | |
4707 | } | } |
4708 | } | } |
4709 | break; | break; |
# | Line 4280 for (;;) | Line 4714 for (;;) |
4714 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4715 | { | { |
4716 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4717 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4718 | } | } |
4719 | switch(*eptr++) | switch(*eptr++) |
4720 | { | { |
4721 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
4722 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
4723 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
4724 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
4725 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
4726 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
4727 | #ifdef COMPILE_PCRE16 | |
4728 | case 0x2028: /* LINE SEPARATOR */ | |
4729 | case 0x2029: /* PARAGRAPH SEPARATOR */ | |
4730 | #endif | |
4731 | break; | break; |
4732 | } | } |
4733 | } | } |
# | Line 4301 for (;;) | Line 4739 for (;;) |
4739 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4740 | { | { |
4741 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4742 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4743 | } | } |
4744 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_digit) != 0) |
4745 | RRETURN(MATCH_NOMATCH); | |
4746 | eptr++; | |
4747 | } | } |
4748 | break; | break; |
4749 | ||
# | Line 4313 for (;;) | Line 4753 for (;;) |
4753 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4754 | { | { |
4755 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4756 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4757 | } | } |
4758 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_digit) == 0) |
4759 | RRETURN(MATCH_NOMATCH); | |
4760 | eptr++; | |
4761 | } | } |
4762 | break; | break; |
4763 | ||
# | Line 4325 for (;;) | Line 4767 for (;;) |
4767 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4768 | { | { |
4769 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4770 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4771 | } | } |
4772 | if ((md->ctypes[*eptr++] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_space) != 0) |
4773 | RRETURN(MATCH_NOMATCH); | |
4774 | eptr++; | |
4775 | } | } |
4776 | break; | break; |
4777 | ||
# | Line 4337 for (;;) | Line 4781 for (;;) |
4781 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4782 | { | { |
4783 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4784 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4785 | } | } |
4786 | if ((md->ctypes[*eptr++] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_space) == 0) |
4787 | RRETURN(MATCH_NOMATCH); | |
4788 | eptr++; | |
4789 | } | } |
4790 | break; | break; |
4791 | ||
# | Line 4349 for (;;) | Line 4795 for (;;) |
4795 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4796 | { | { |
4797 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4798 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4799 | } | } |
4800 | if ((md->ctypes[*eptr++] & ctype_word) != 0) | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_word) != 0) |
4801 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4802 | eptr++; | |
4803 | } | } |
4804 | break; | break; |
4805 | ||
# | Line 4362 for (;;) | Line 4809 for (;;) |
4809 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4810 | { | { |
4811 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4812 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4813 | } | } |
4814 | if ((md->ctypes[*eptr++] & ctype_word) == 0) | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_word) == 0) |
4815 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4816 | eptr++; | |
4817 | } | } |
4818 | break; | break; |
4819 | ||
# | Line 4394 for (;;) | Line 4842 for (;;) |
4842 | { | { |
4843 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM36); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM36); |
4844 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
4845 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
4846 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4847 | { | { |
4848 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4849 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4850 | } | } |
4851 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4852 | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); | if (prop_fail_result) RRETURN(MATCH_NOMATCH); |
4853 | } | } |
4854 | /* Control never gets here */ | /* Control never gets here */ |
4855 | ||
# | Line 4411 for (;;) | Line 4859 for (;;) |
4859 | int chartype; | int chartype; |
4860 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM37); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM37); |
4861 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
4862 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
4863 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4864 | { | { |
4865 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4866 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4867 | } | } |
4868 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4869 | chartype = UCD_CHARTYPE(c); | chartype = UCD_CHARTYPE(c); |
4870 | if ((chartype == ucp_Lu || | if ((chartype == ucp_Lu || |
4871 | chartype == ucp_Ll || | chartype == ucp_Ll || |
4872 | chartype == ucp_Lt) == prop_fail_result) | chartype == ucp_Lt) == prop_fail_result) |
4873 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4874 | } | } |
4875 | /* Control never gets here */ | /* Control never gets here */ |
4876 | ||
# | Line 4431 for (;;) | Line 4879 for (;;) |
4879 | { | { |
4880 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM38); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM38); |
4881 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
4882 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
4883 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4884 | { | { |
4885 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4886 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4887 | } | } |
4888 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4889 | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) |
4890 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4891 | } | } |
4892 | /* Control never gets here */ | /* Control never gets here */ |
4893 | ||
# | Line 4448 for (;;) | Line 4896 for (;;) |
4896 | { | { |
4897 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM39); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM39); |
4898 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
4899 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
4900 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4901 | { | { |
4902 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4903 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4904 | } | } |
4905 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
4906 | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) |
4907 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4908 | } | } |
4909 | /* Control never gets here */ | /* Control never gets here */ |
4910 | ||
# | Line 4465 for (;;) | Line 4913 for (;;) |
4913 | { | { |
4914 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM40); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM40); |
4915 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
4916 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
4917 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
4918 | { | { |
4919 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
4920 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
4921 | } | } |
4922 | GETCHARINCTEST(c, eptr); |