Parent Directory
|
Revision Log
|
Patch
| revision 227 by ph10, Tue Aug 21 15:00:15 2007 UTC | revision 579 by ph10, Wed Nov 24 17:39:25 2010 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) 1997-2007 University of Cambridge | Copyright (c) 1997-2010 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 43 pattern matching using an NFA algorithm, | Line 43 pattern matching using an NFA algorithm, |
| 43 | possible. There are also some static supporting functions. */ | possible. There are also some static supporting functions. */ |
| 44 | ||
| 45 | #ifdef HAVE_CONFIG_H | #ifdef HAVE_CONFIG_H |
| 46 | #include <config.h> | #include "config.h" |
| 47 | #endif | #endif |
| 48 | ||
| 49 | #define NLBLOCK md /* Block containing newline information */ | #define NLBLOCK md /* Block containing newline information */ |
| # | Line 71 defined PCRE_ERROR_xxx codes, which are | Line 71 defined PCRE_ERROR_xxx codes, which are |
| 71 | /* Special internal returns from the match() function. Make them sufficiently | /* Special internal returns from the match() function. Make them sufficiently |
| 72 | negative to avoid the external error codes. */ | negative to avoid the external error codes. */ |
| 73 | ||
| 74 | #define MATCH_COMMIT (-999) | #define MATCH_ACCEPT (-999) |
| 75 | #define MATCH_PRUNE (-998) | #define MATCH_COMMIT (-998) |
| 76 | #define MATCH_SKIP (-997) | #define MATCH_PRUNE (-997) |
| 77 | #define MATCH_THEN (-996) | #define MATCH_SKIP (-996) |
| 78 | #define MATCH_SKIP_ARG (-995) | |
| 79 | #define MATCH_THEN (-994) | |
| 80 | ||
| 81 | /* This is a convenience macro for code that occurs many times. */ | |
| 82 | ||
| 83 | #define MRRETURN(ra) \ | |
| 84 | { \ | |
| 85 | md->mark = markptr; \ | |
| 86 | RRETURN(ra); \ | |
| 87 | } | |
| 88 | ||
| 89 | /* 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. |
| 90 | 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, |
| # | Line 89 static const char rep_max[] = { 0, 0, 0, | Line 99 static const char rep_max[] = { 0, 0, 0, |
| 99 | ||
| 100 | ||
| 101 | ||
| 102 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 103 | /************************************************* | /************************************************* |
| 104 | * Debugging function to print chars * | * Debugging function to print chars * |
| 105 | *************************************************/ | *************************************************/ |
| # | Line 141 match_ref(int offset, register USPTR ept | Line 151 match_ref(int offset, register USPTR ept |
| 151 | { | { |
| 152 | USPTR p = md->start_subject + md->offset_vector[offset]; | USPTR p = md->start_subject + md->offset_vector[offset]; |
| 153 | ||
| 154 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 155 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 156 | printf("matching subject <null>"); | printf("matching subject <null>"); |
| 157 | else | else |
| # | Line 158 printf("\n"); | Line 168 printf("\n"); |
| 168 | ||
| 169 | if (length > md->end_subject - eptr) return FALSE; | if (length > md->end_subject - eptr) return FALSE; |
| 170 | ||
| 171 | /* Separate the caselesss case for speed */ | /* Separate the caseless case for speed. In UTF-8 mode we can only do this |
| 172 | properly if Unicode properties are supported. Otherwise, we can check only | |
| 173 | ASCII characters. */ | |
| 174 | ||
| 175 | if ((ims & PCRE_CASELESS) != 0) | if ((ims & PCRE_CASELESS) != 0) |
| 176 | { | { |
| 177 | #ifdef SUPPORT_UTF8 | |
| 178 | #ifdef SUPPORT_UCP | |
| 179 | if (md->utf8) | |
| 180 | { | |
| 181 | USPTR endptr = eptr + length; | |
| 182 | while (eptr < endptr) | |
| 183 | { | |
| 184 | int c, d; | |
| 185 | GETCHARINC(c, eptr); | |
| 186 | GETCHARINC(d, p); | |
| 187 | if (c != d && c != UCD_OTHERCASE(d)) return FALSE; | |
| 188 | } | |
| 189 | } | |
| 190 | else | |
| 191 | #endif | |
| 192 | #endif | |
| 193 | ||
| 194 | /* The same code works when not in UTF-8 mode and in UTF-8 mode when there | |
| 195 | is no UCP support. */ | |
| 196 | ||
| 197 | while (length-- > 0) | while (length-- > 0) |
| 198 | if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; | { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; } |
| 199 | } | } |
| 200 | ||
| 201 | /* In the caseful case, we can just compare the bytes, whether or not we | |
| 202 | are in UTF-8 mode. */ | |
| 203 | ||
| 204 | else | else |
| 205 | { while (length-- > 0) if (*p++ != *eptr++) return FALSE; } | { while (length-- > 0) if (*p++ != *eptr++) return FALSE; } |
| 206 | ||
| # | Line 219 enum { RM1=1, RM2, RM3, RM4, RM5, RM | Line 255 enum { RM1=1, RM2, RM3, RM4, RM5, RM |
| 255 | RM21, RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30, | RM21, RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30, |
| 256 | RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, | RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, |
| 257 | RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, | RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, |
| 258 | RM51, RM52, RM53, RM54 }; | RM51, RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60, |
| 259 | RM61, RM62 }; | |
| 260 | ||
| 261 | /* 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 |
| 262 | 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 |
| 263 | actuall used in this definition. */ | actually used in this definition. */ |
| 264 | ||
| 265 | #ifndef NO_RECURSE | #ifndef NO_RECURSE |
| 266 | #define REGISTER register | #define REGISTER register |
| 267 | ||
| 268 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 269 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ |
| 270 | { \ | { \ |
| 271 | printf("match() called in line %d\n", __LINE__); \ | printf("match() called in line %d\n", __LINE__); \ |
| 272 | rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \ | rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1); \ |
| 273 | printf("to line %d\n", __LINE__); \ | printf("to line %d\n", __LINE__); \ |
| 274 | } | } |
| 275 | #define RRETURN(ra) \ | #define RRETURN(ra) \ |
| # | Line 242 actuall used in this definition. */ | Line 279 actuall used in this definition. */ |
| 279 | } | } |
| 280 | #else | #else |
| 281 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ |
| 282 | rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1) | rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1) |
| 283 | #define RRETURN(ra) return ra | #define RRETURN(ra) return ra |
| 284 | #endif | #endif |
| 285 | ||
| # | Line 257 argument of match(), which never changes | Line 294 argument of match(), which never changes |
| 294 | ||
| 295 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\ |
| 296 | {\ | {\ |
| 297 | heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\ | heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\ |
| 298 | if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\ | |
| 299 | frame->Xwhere = rw; \ | frame->Xwhere = rw; \ |
| 300 | newframe->Xeptr = ra;\ | newframe->Xeptr = ra;\ |
| 301 | newframe->Xecode = rb;\ | newframe->Xecode = rb;\ |
| 302 | newframe->Xmstart = mstart;\ | newframe->Xmstart = mstart;\ |
| 303 | newframe->Xmarkptr = markptr;\ | |
| 304 | newframe->Xoffset_top = rc;\ | newframe->Xoffset_top = rc;\ |
| 305 | newframe->Xims = re;\ | newframe->Xims = re;\ |
| 306 | newframe->Xeptrb = rf;\ | newframe->Xeptrb = rf;\ |
| # | Line 277 argument of match(), which never changes | Line 316 argument of match(), which never changes |
| 316 | ||
| 317 | #define RRETURN(ra)\ | #define RRETURN(ra)\ |
| 318 | {\ | {\ |
| 319 | heapframe *newframe = frame;\ | heapframe *oldframe = frame;\ |
| 320 | frame = newframe->Xprevframe;\ | frame = oldframe->Xprevframe;\ |
| 321 | (pcre_stack_free)(newframe);\ | (pcre_stack_free)(oldframe);\ |
| 322 | if (frame != NULL)\ | if (frame != NULL)\ |
| 323 | {\ | {\ |
| 324 | rrc = ra;\ | rrc = ra;\ |
| # | Line 296 typedef struct heapframe { | Line 335 typedef struct heapframe { |
| 335 | ||
| 336 | /* Function arguments that may change */ | /* Function arguments that may change */ |
| 337 | ||
| 338 | const uschar *Xeptr; | USPTR Xeptr; |
| 339 | const uschar *Xecode; | const uschar *Xecode; |
| 340 | const uschar *Xmstart; | USPTR Xmstart; |
| 341 | USPTR Xmarkptr; | |
| 342 | int Xoffset_top; | int Xoffset_top; |
| 343 | long int Xims; | long int Xims; |
| 344 | eptrblock *Xeptrb; | eptrblock *Xeptrb; |
| # | Line 307 typedef struct heapframe { | Line 347 typedef struct heapframe { |
| 347 | ||
| 348 | /* Function local variables */ | /* Function local variables */ |
| 349 | ||
| 350 | const uschar *Xcallpat; | USPTR Xcallpat; |
| 351 | const uschar *Xcharptr; | #ifdef SUPPORT_UTF8 |
| 352 | const uschar *Xdata; | USPTR Xcharptr; |
| 353 | const uschar *Xnext; | #endif |
| 354 | const uschar *Xpp; | USPTR Xdata; |
| 355 | const uschar *Xprev; | USPTR Xnext; |
| 356 | const uschar *Xsaved_eptr; | USPTR Xpp; |
| 357 | USPTR Xprev; | |
| 358 | USPTR Xsaved_eptr; | |
| 359 | ||
| 360 | recursion_info Xnew_recursive; | recursion_info Xnew_recursive; |
| 361 | ||
| # | Line 334 typedef struct heapframe { | Line 376 typedef struct heapframe { |
| 376 | uschar Xocchars[8]; | uschar Xocchars[8]; |
| 377 | #endif | #endif |
| 378 | ||
| 379 | int Xcodelink; | |
| 380 | int Xctype; | int Xctype; |
| 381 | unsigned int Xfc; | unsigned int Xfc; |
| 382 | int Xfi; | int Xfi; |
| # | Line 369 typedef struct heapframe { | Line 412 typedef struct heapframe { |
| 412 | ||
| 413 | /* This function is called recursively in many circumstances. Whenever it | /* This function is called recursively in many circumstances. Whenever it |
| 414 | returns a negative (error) response, the outer incarnation must also return the | returns a negative (error) response, the outer incarnation must also return the |
| 415 | same response. | same response. */ |
| 416 | ||
| 417 | /* These macros pack up tests that are used for partial matching, and which | |
| 418 | appears several times in the code. We set the "hit end" flag if the pointer is | |
| 419 | at the end of the subject and also past the start of the subject (i.e. | |
| 420 | something has been matched). For hard partial matching, we then return | |
| 421 | immediately. The second one is used when we already know we are past the end of | |
| 422 | the subject. */ | |
| 423 | ||
| 424 | #define CHECK_PARTIAL()\ | |
| 425 | if (md->partial != 0 && eptr >= md->end_subject && \ | |
| 426 | eptr > md->start_used_ptr) \ | |
| 427 | { \ | |
| 428 | md->hitend = TRUE; \ | |
| 429 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \ | |
| 430 | } | |
| 431 | ||
| 432 | #define SCHECK_PARTIAL()\ | |
| 433 | if (md->partial != 0 && eptr > md->start_used_ptr) \ | |
| 434 | { \ | |
| 435 | md->hitend = TRUE; \ | |
| 436 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \ | |
| 437 | } | |
| 438 | ||
| 439 | ||
| 440 | Performance note: It might be tempting to extract commonly used fields from the | /* Performance note: It might be tempting to extract commonly used fields from |
| 441 | md structure (e.g. utf8, end_subject) into individual variables to improve | the md structure (e.g. utf8, end_subject) into individual variables to improve |
| 442 | 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 |
| 443 | made performance worse. | made performance worse. |
| 444 | ||
| # | Line 381 Arguments: | Line 447 Arguments: |
| 447 | ecode pointer to current position in compiled code | ecode pointer to current position in compiled code |
| 448 | mstart pointer to the current match start position (can be modified | mstart pointer to the current match start position (can be modified |
| 449 | by encountering \K) | by encountering \K) |
| 450 | markptr pointer to the most recent MARK name, or NULL | |
| 451 | offset_top current top pointer | offset_top current top pointer |
| 452 | md pointer to "static" info for the match | md pointer to "static" info for the match |
| 453 | ims current /i, /m, and /s options | ims current /i, /m, and /s options |
| # | Line 394 Arguments: | Line 461 Arguments: |
| 461 | ||
| 462 | Returns: MATCH_MATCH if matched ) these values are >= 0 | Returns: MATCH_MATCH if matched ) these values are >= 0 |
| 463 | MATCH_NOMATCH if failed to match ) | MATCH_NOMATCH if failed to match ) |
| 464 | a negative MATCH_xxx value for PRUNE, SKIP, etc | |
| 465 | a negative PCRE_ERROR_xxx value if aborted by an error condition | a negative PCRE_ERROR_xxx value if aborted by an error condition |
| 466 | (e.g. stopped by repeated call or recursion limit) | (e.g. stopped by repeated call or recursion limit) |
| 467 | */ | */ |
| 468 | ||
| 469 | static int | static int |
| 470 | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart, | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart, |
| 471 | int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb, | const uschar *markptr, int offset_top, match_data *md, unsigned long int ims, |
| 472 | int flags, unsigned int rdepth) | eptrblock *eptrb, int flags, unsigned int rdepth) |
| 473 | { | { |
| 474 | /* 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, |
| 475 | so they can be ordinary variables in all cases. Mark some of them with | so they can be ordinary variables in all cases. Mark some of them with |
| # | Line 413 register unsigned int c; /* Character | Line 481 register unsigned int c; /* Character |
| 481 | register BOOL utf8; /* Local copy of UTF-8 flag for speed */ | register BOOL utf8; /* Local copy of UTF-8 flag for speed */ |
| 482 | ||
| 483 | BOOL minimize, possessive; /* Quantifier options */ | BOOL minimize, possessive; /* Quantifier options */ |
| 484 | int condcode; | |
| 485 | ||
| 486 | /* 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 |
| 487 | preserved over calls to RMATCH() are part of a "frame" which is obtained from | preserved over calls to RMATCH() are part of a "frame" which is obtained from |
| # | Line 420 heap storage. Set up the top-level frame | Line 489 heap storage. Set up the top-level frame |
| 489 | heap whenever RMATCH() does a "recursion". See the macro definitions above. */ | heap whenever RMATCH() does a "recursion". See the macro definitions above. */ |
| 490 | ||
| 491 | #ifdef NO_RECURSE | #ifdef NO_RECURSE |
| 492 | heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe)); | heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe)); |
| 493 | if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY); | |
| 494 | frame->Xprevframe = NULL; /* Marks the top level */ | frame->Xprevframe = NULL; /* Marks the top level */ |
| 495 | ||
| 496 | /* Copy in the original argument variables */ | /* Copy in the original argument variables */ |
| # | Line 428 frame->Xprevframe = NULL; /* | Line 498 frame->Xprevframe = NULL; /* |
| 498 | frame->Xeptr = eptr; | frame->Xeptr = eptr; |
| 499 | frame->Xecode = ecode; | frame->Xecode = ecode; |
| 500 | frame->Xmstart = mstart; | frame->Xmstart = mstart; |
| 501 | frame->Xmarkptr = markptr; | |
| 502 | frame->Xoffset_top = offset_top; | frame->Xoffset_top = offset_top; |
| 503 | frame->Xims = ims; | frame->Xims = ims; |
| 504 | frame->Xeptrb = eptrb; | frame->Xeptrb = eptrb; |
| # | Line 443 HEAP_RECURSE: | Line 514 HEAP_RECURSE: |
| 514 | #define eptr frame->Xeptr | #define eptr frame->Xeptr |
| 515 | #define ecode frame->Xecode | #define ecode frame->Xecode |
| 516 | #define mstart frame->Xmstart | #define mstart frame->Xmstart |
| 517 | #define markptr frame->Xmarkptr | |
| 518 | #define offset_top frame->Xoffset_top | #define offset_top frame->Xoffset_top |
| 519 | #define ims frame->Xims | #define ims frame->Xims |
| 520 | #define eptrb frame->Xeptrb | #define eptrb frame->Xeptrb |
| # | Line 455 HEAP_RECURSE: | Line 527 HEAP_RECURSE: |
| 527 | #define charptr frame->Xcharptr | #define charptr frame->Xcharptr |
| 528 | #endif | #endif |
| 529 | #define callpat frame->Xcallpat | #define callpat frame->Xcallpat |
| 530 | #define codelink frame->Xcodelink | |
| 531 | #define data frame->Xdata | #define data frame->Xdata |
| 532 | #define next frame->Xnext | #define next frame->Xnext |
| 533 | #define pp frame->Xpp | #define pp frame->Xpp |
| # | Line 535 int oclength; | Line 608 int oclength; |
| 608 | uschar occhars[8]; | uschar occhars[8]; |
| 609 | #endif | #endif |
| 610 | ||
| 611 | int codelink; | |
| 612 | int ctype; | int ctype; |
| 613 | int length; | int length; |
| 614 | int max; | int max; |
| # | Line 568 TAIL_RECURSE: | Line 642 TAIL_RECURSE: |
| 642 | /* OK, now we can get on with the real code of the function. Recursive calls | /* OK, now we can get on with the real code of the function. Recursive calls |
| 643 | are specified by the macro RMATCH and RRETURN is used to return. When | are specified by the macro RMATCH and RRETURN is used to return. When |
| 644 | NO_RECURSE is *not* defined, these just turn into a recursive call to match() | NO_RECURSE is *not* defined, these just turn into a recursive call to match() |
| 645 | and a "return", respectively (possibly with some debugging if DEBUG is | and a "return", respectively (possibly with some debugging if PCRE_DEBUG is |
| 646 | defined). However, RMATCH isn't like a function call because it's quite a | defined). However, RMATCH isn't like a function call because it's quite a |
| 647 | 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, |
| 648 | however, impact performance when true recursion is being used. */ | however, impact performance when true recursion is being used. */ |
| # | Line 610 for (;;) | Line 684 for (;;) |
| 684 | minimize = possessive = FALSE; | minimize = possessive = FALSE; |
| 685 | op = *ecode; | op = *ecode; |
| 686 | ||
| /* For partial matching, remember if we ever hit the end of the subject after | ||
| matching at least one subject character. */ | ||
| if (md->partial && | ||
| eptr >= md->end_subject && | ||
| eptr > mstart) | ||
| md->hitend = TRUE; | ||
| 687 | switch(op) | switch(op) |
| 688 | { | { |
| 689 | case OP_MARK: | |
| 690 | markptr = ecode + 2; | |
| 691 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 692 | ims, eptrb, flags, RM55); | |
| 693 | ||
| 694 | /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an | |
| 695 | argument, and we must check whether that argument matches this MARK's | |
| 696 | argument. It is passed back in md->start_match_ptr (an overloading of that | |
| 697 | variable). If it does match, we reset that variable to the current subject | |
| 698 | position and return MATCH_SKIP. Otherwise, pass back the return code | |
| 699 | unaltered. */ | |
| 700 | ||
| 701 | if (rrc == MATCH_SKIP_ARG && | |
| 702 | strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0) | |
| 703 | { | |
| 704 | md->start_match_ptr = eptr; | |
| 705 | RRETURN(MATCH_SKIP); | |
| 706 | } | |
| 707 | ||
| 708 | if (md->mark == NULL) md->mark = markptr; | |
| 709 | RRETURN(rrc); | |
| 710 | ||
| 711 | case OP_FAIL: | case OP_FAIL: |
| 712 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 713 | ||
| 714 | /* COMMIT overrides PRUNE, SKIP, and THEN */ | |
| 715 | ||
| 716 | case OP_COMMIT: | |
| 717 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | |
| 718 | ims, eptrb, flags, RM52); | |
| 719 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && | |
| 720 | rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG && | |
| 721 | rrc != MATCH_THEN) | |
| 722 | RRETURN(rrc); | |
| 723 | MRRETURN(MATCH_COMMIT); | |
| 724 | ||
| 725 | /* PRUNE overrides THEN */ | |
| 726 | ||
| 727 | case OP_PRUNE: | case OP_PRUNE: |
| 728 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 729 | ims, eptrb, flags, RM51); | ims, eptrb, flags, RM51); |
| 730 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| 731 | MRRETURN(MATCH_PRUNE); | |
| 732 | ||
| 733 | case OP_PRUNE_ARG: | |
| 734 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 735 | ims, eptrb, flags, RM56); | |
| 736 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | |
| 737 | md->mark = ecode + 2; | |
| 738 | RRETURN(MATCH_PRUNE); | RRETURN(MATCH_PRUNE); |
| 739 | ||
| 740 | case OP_COMMIT: | /* SKIP overrides PRUNE and THEN */ |
| RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | ||
| ims, eptrb, flags, RM52); | ||
| if (rrc != MATCH_NOMATCH) RRETURN(rrc); | ||
| RRETURN(MATCH_COMMIT); | ||
| 741 | ||
| 742 | case OP_SKIP: | case OP_SKIP: |
| 743 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 744 | ims, eptrb, flags, RM53); | ims, eptrb, flags, RM53); |
| 745 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) |
| 746 | RRETURN(rrc); | |
| 747 | md->start_match_ptr = eptr; /* Pass back current position */ | md->start_match_ptr = eptr; /* Pass back current position */ |
| 748 | RRETURN(MATCH_SKIP); | MRRETURN(MATCH_SKIP); |
| 749 | ||
| 750 | case OP_SKIP_ARG: | |
| 751 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 752 | ims, eptrb, flags, RM57); | |
| 753 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) | |
| 754 | RRETURN(rrc); | |
| 755 | ||
| 756 | /* Pass back the current skip name by overloading md->start_match_ptr and | |
| 757 | returning the special MATCH_SKIP_ARG return code. This will either be | |
| 758 | caught by a matching MARK, or get to the top, where it is treated the same | |
| 759 | as PRUNE. */ | |
| 760 | ||
| 761 | md->start_match_ptr = ecode + 2; | |
| 762 | RRETURN(MATCH_SKIP_ARG); | |
| 763 | ||
| 764 | /* For THEN (and THEN_ARG) we pass back the address of the bracket or | |
| 765 | the alt that is at the start of the current branch. This makes it possible | |
| 766 | to skip back past alternatives that precede the THEN within the current | |
| 767 | branch. */ | |
| 768 | ||
| 769 | case OP_THEN: | case OP_THEN: |
| 770 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 771 | ims, eptrb, flags, RM54); | ims, eptrb, flags, RM54); |
| 772 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 773 | md->start_match_ptr = ecode - GET(ecode, 1); | |
| 774 | MRRETURN(MATCH_THEN); | |
| 775 | ||
| 776 | case OP_THEN_ARG: | |
| 777 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE], | |
| 778 | offset_top, md, ims, eptrb, flags, RM58); | |
| 779 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 780 | md->start_match_ptr = ecode - GET(ecode, 1); | |
| 781 | md->mark = ecode + LINK_SIZE + 2; | |
| 782 | RRETURN(MATCH_THEN); | RRETURN(MATCH_THEN); |
| 783 | ||
| 784 | /* Handle a capturing bracket. If there is space in the offset vector, save | /* Handle a capturing bracket. If there is space in the offset vector, save |
| # | Line 667 for (;;) | Line 800 for (;;) |
| 800 | number = GET2(ecode, 1+LINK_SIZE); | number = GET2(ecode, 1+LINK_SIZE); |
| 801 | offset = number << 1; | offset = number << 1; |
| 802 | ||
| 803 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 804 | printf("start bracket %d\n", number); | printf("start bracket %d\n", number); |
| 805 | printf("subject="); | printf("subject="); |
| 806 | pchars(eptr, 16, TRUE, md); | pchars(eptr, 16, TRUE, md); |
| # | Line 682 for (;;) | Line 815 for (;;) |
| 815 | save_capture_last = md->capture_last; | save_capture_last = md->capture_last; |
| 816 | ||
| 817 | DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); | DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); |
| 818 | md->offset_vector[md->offset_end - number] = eptr - md->start_subject; | md->offset_vector[md->offset_end - number] = |
| 819 | (int)(eptr - md->start_subject); | |
| 820 | ||
| 821 | flags = (op == OP_SCBRA)? match_cbegroup : 0; | flags = (op == OP_SCBRA)? match_cbegroup : 0; |
| 822 | do | do |
| 823 | { | { |
| 824 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 825 | ims, eptrb, flags, RM1); | ims, eptrb, flags, RM1); |
| 826 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && |
| 827 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 828 | RRETURN(rrc); | |
| 829 | md->capture_last = save_capture_last; | md->capture_last = save_capture_last; |
| 830 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 831 | } | } |
| # | Line 701 for (;;) | Line 837 for (;;) |
| 837 | md->offset_vector[offset+1] = save_offset2; | md->offset_vector[offset+1] = save_offset2; |
| 838 | md->offset_vector[md->offset_end - number] = save_offset3; | md->offset_vector[md->offset_end - number] = save_offset3; |
| 839 | ||
| 840 | if (rrc != MATCH_THEN) md->mark = markptr; | |
| 841 | RRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 842 | } | } |
| 843 | ||
| # | Line 740 for (;;) | Line 877 for (;;) |
| 877 | ||
| 878 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, |
| 879 | eptrb, flags, RM48); | eptrb, flags, RM48); |
| 880 | if (rrc == MATCH_NOMATCH) md->mark = markptr; | |
| 881 | RRETURN(rrc); | RRETURN(rrc); |
| 882 | } | } |
| 883 | ||
| # | Line 748 for (;;) | Line 886 for (;;) |
| 886 | ||
| 887 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, |
| 888 | eptrb, flags, RM2); | eptrb, flags, RM2); |
| 889 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && |
| 890 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 891 | RRETURN(rrc); | |
| 892 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 893 | } | } |
| 894 | /* Control never reaches here. */ | /* Control never reaches here. */ |
| # | Line 761 for (;;) | Line 901 for (;;) |
| 901 | ||
| 902 | case OP_COND: | case OP_COND: |
| 903 | case OP_SCOND: | case OP_SCOND: |
| 904 | if (ecode[LINK_SIZE+1] == OP_RREF) /* Recursion test */ | codelink= GET(ecode, 1); |
| 905 | ||
| 906 | /* Because of the way auto-callout works during compile, a callout item is | |
| 907 | inserted between OP_COND and an assertion condition. */ | |
| 908 | ||
| 909 | if (ecode[LINK_SIZE+1] == OP_CALLOUT) | |
| 910 | { | { |
| 911 | offset = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | if (pcre_callout != NULL) |
| 912 | condition = md->recursive != NULL && | { |
| 913 | (offset == RREF_ANY || offset == md->recursive->group_num); | pcre_callout_block cb; |
| 914 | ecode += condition? 3 : GET(ecode, 1); | cb.version = 1; /* Version 1 of the callout block */ |
| 915 | cb.callout_number = ecode[LINK_SIZE+2]; | |
| 916 | cb.offset_vector = md->offset_vector; | |
| 917 | cb.subject = (PCRE_SPTR)md->start_subject; | |
| 918 | cb.subject_length = (int)(md->end_subject - md->start_subject); | |
| 919 | cb.start_match = (int)(mstart - md->start_subject); | |
| 920 | cb.current_position = (int)(eptr - md->start_subject); | |
| 921 | cb.pattern_position = GET(ecode, LINK_SIZE + 3); | |
| 922 | cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE); | |
| 923 | cb.capture_top = offset_top/2; | |
| 924 | cb.capture_last = md->capture_last; | |
| 925 | cb.callout_data = md->callout_data; | |
| 926 | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); | |
| 927 | if (rrc < 0) RRETURN(rrc); | |
| 928 | } | |
| 929 | ecode += _pcre_OP_lengths[OP_CALLOUT]; | |
| 930 | } | |
| 931 | ||
| 932 | condcode = ecode[LINK_SIZE+1]; | |
| 933 | ||
| 934 | /* Now see what the actual condition is */ | |
| 935 | ||
| 936 | if (condcode == OP_RREF || condcode == OP_NRREF) /* Recursion test */ | |
| 937 | { | |
| 938 | if (md->recursive == NULL) /* Not recursing => FALSE */ | |
| 939 | { | |
| 940 | condition = FALSE; | |
| 941 | ecode += GET(ecode, 1); | |
| 942 | } | |
| 943 | else | |
| 944 | { | |
| 945 | int recno = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | |
| 946 | condition = (recno == RREF_ANY || recno == md->recursive->group_num); | |
| 947 | ||
| 948 | /* If the test is for recursion into a specific subpattern, and it is | |
| 949 | false, but the test was set up by name, scan the table to see if the | |
| 950 | name refers to any other numbers, and test them. The condition is true | |
| 951 | if any one is set. */ | |
| 952 | ||
| 953 | if (!condition && condcode == OP_NRREF && recno != RREF_ANY) | |
| 954 | { | |
| 955 | uschar *slotA = md->name_table; | |
| 956 | for (i = 0; i < md->name_count; i++) | |
| 957 | { | |
| 958 | if (GET2(slotA, 0) == recno) break; | |
| 959 | slotA += md->name_entry_size; | |
| 960 | } | |
| 961 | ||
| 962 | /* Found a name for the number - there can be only one; duplicate | |
| 963 | names for different numbers are allowed, but not vice versa. First | |
| 964 | scan down for duplicates. */ | |
| 965 | ||
| 966 | if (i < md->name_count) | |
| 967 | { | |
| 968 | uschar *slotB = slotA; | |
| 969 | while (slotB > md->name_table) | |
| 970 | { | |
| 971 | slotB -= md->name_entry_size; | |
| 972 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 973 | { | |
| 974 | condition = GET2(slotB, 0) == md->recursive->group_num; | |
| 975 | if (condition) break; | |
| 976 | } | |
| 977 | else break; | |
| 978 | } | |
| 979 | ||
| 980 | /* Scan up for duplicates */ | |
| 981 | ||
| 982 | if (!condition) | |
| 983 | { | |
| 984 | slotB = slotA; | |
| 985 | for (i++; i < md->name_count; i++) | |
| 986 | { | |
| 987 | slotB += md->name_entry_size; | |
| 988 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 989 | { | |
| 990 | condition = GET2(slotB, 0) == md->recursive->group_num; | |
| 991 | if (condition) break; | |
| 992 | } | |
| 993 | else break; | |
| 994 | } | |
| 995 | } | |
| 996 | } | |
| 997 | } | |
| 998 | ||
| 999 | /* Chose branch according to the condition */ | |
| 1000 | ||
| 1001 | ecode += condition? 3 : GET(ecode, 1); | |
| 1002 | } | |
| 1003 | } | } |
| 1004 | ||
| 1005 | else if (ecode[LINK_SIZE+1] == OP_CREF) /* Group used test */ | else if (condcode == OP_CREF || condcode == OP_NCREF) /* Group used test */ |
| 1006 | { | { |
| 1007 | offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ | offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ |
| 1008 | condition = offset < offset_top && md->offset_vector[offset] >= 0; | condition = offset < offset_top && md->offset_vector[offset] >= 0; |
| 1009 | ||
| 1010 | /* If the numbered capture is unset, but the reference was by name, | |
| 1011 | scan the table to see if the name refers to any other numbers, and test | |
| 1012 | them. The condition is true if any one is set. This is tediously similar | |
| 1013 | to the code above, but not close enough to try to amalgamate. */ | |
| 1014 | ||
| 1015 | if (!condition && condcode == OP_NCREF) | |
| 1016 | { | |
| 1017 | int refno = offset >> 1; | |
| 1018 | uschar *slotA = md->name_table; | |
| 1019 | ||
| 1020 | for (i = 0; i < md->name_count; i++) | |
| 1021 | { | |
| 1022 | if (GET2(slotA, 0) == refno) break; | |
| 1023 | slotA += md->name_entry_size; | |
| 1024 | } | |
| 1025 | ||
| 1026 | /* Found a name for the number - there can be only one; duplicate names | |
| 1027 | for different numbers are allowed, but not vice versa. First scan down | |
| 1028 | for duplicates. */ | |
| 1029 | ||
| 1030 | if (i < md->name_count) | |
| 1031 | { | |
| 1032 | uschar *slotB = slotA; | |
| 1033 | while (slotB > md->name_table) | |
| 1034 | { | |
| 1035 | slotB -= md->name_entry_size; | |
| 1036 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1037 | { | |
| 1038 | offset = GET2(slotB, 0) << 1; | |
| 1039 | condition = offset < offset_top && | |
| 1040 | md->offset_vector[offset] >= 0; | |
| 1041 | if (condition) break; | |
| 1042 | } | |
| 1043 | else break; | |
| 1044 | } | |
| 1045 | ||
| 1046 | /* Scan up for duplicates */ | |
| 1047 | ||
| 1048 | if (!condition) | |
| 1049 | { | |
| 1050 | slotB = slotA; | |
| 1051 | for (i++; i < md->name_count; i++) | |
| 1052 | { | |
| 1053 | slotB += md->name_entry_size; | |
| 1054 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1055 | { | |
| 1056 | offset = GET2(slotB, 0) << 1; | |
| 1057 | condition = offset < offset_top && | |
| 1058 | md->offset_vector[offset] >= 0; | |
| 1059 | if (condition) break; | |
| 1060 | } | |
| 1061 | else break; | |
| 1062 | } | |
| 1063 | } | |
| 1064 | } | |
| 1065 | } | |
| 1066 | ||
| 1067 | /* Chose branch according to the condition */ | |
| 1068 | ||
| 1069 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 3 : GET(ecode, 1); |
| 1070 | } | } |
| 1071 | ||
| 1072 | else if (ecode[LINK_SIZE+1] == OP_DEF) /* DEFINE - always false */ | else if (condcode == OP_DEF) /* DEFINE - always false */ |
| 1073 | { | { |
| 1074 | condition = FALSE; | condition = FALSE; |
| 1075 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| # | Line 796 for (;;) | Line 1089 for (;;) |
| 1089 | ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); | ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); |
| 1090 | while (*ecode == OP_ALT) ecode += GET(ecode, 1); | while (*ecode == OP_ALT) ecode += GET(ecode, 1); |
| 1091 | } | } |
| 1092 | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) | else if (rrc != MATCH_NOMATCH && |
| 1093 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1094 | { | { |
| 1095 | RRETURN(rrc); /* Need braces because of following else */ | RRETURN(rrc); /* Need braces because of following else */ |
| 1096 | } | } |
| 1097 | else | else |
| 1098 | { | { |
| 1099 | condition = FALSE; | condition = FALSE; |
| 1100 | ecode += GET(ecode, 1); | ecode += codelink; |
| 1101 | } | } |
| 1102 | } | } |
| 1103 | ||
| # | Line 826 for (;;) | Line 1120 for (;;) |
| 1120 | goto TAIL_RECURSE; | goto TAIL_RECURSE; |
| 1121 | } | } |
| 1122 | } | } |
| 1123 | else /* Condition false & no 2nd alternative */ | else /* Condition false & no alternative */ |
| 1124 | { | { |
| 1125 | ecode += 1 + LINK_SIZE; | ecode += 1 + LINK_SIZE; |
| 1126 | } | } |
| 1127 | break; | break; |
| 1128 | ||
| 1129 | ||
| 1130 | /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes, | |
| 1131 | to close any currently open capturing brackets. */ | |
| 1132 | ||
| 1133 | case OP_CLOSE: | |
| 1134 | number = GET2(ecode, 1); | |
| 1135 | offset = number << 1; | |
| 1136 | ||
| 1137 | #ifdef PCRE_DEBUG | |
| 1138 | printf("end bracket %d at *ACCEPT", number); | |
| 1139 | printf("\n"); | |
| 1140 | #endif | |
| 1141 | ||
| 1142 | md->capture_last = number; | |
| 1143 | if (offset >= md->offset_max) md->offset_overflow = TRUE; else | |
| 1144 | { | |
| 1145 | md->offset_vector[offset] = | |
| 1146 | md->offset_vector[md->offset_end - number]; | |
| 1147 | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); | |
| 1148 | if (offset_top <= offset) offset_top = offset + 2; | |
| 1149 | } | |
| 1150 | ecode += 3; | |
| 1151 | break; | |
| 1152 | ||
| 1153 | ||
| 1154 | /* End of the pattern, either real or forced. If we are in a top-level | /* End of the pattern, either real or forced. If we are in a top-level |
| 1155 | recursion, we should restore the offsets appropriately and continue from | recursion, we should restore the offsets appropriately and continue from |
| 1156 | after the call. */ | after the call. */ |
| # | Line 846 for (;;) | Line 1164 for (;;) |
| 1164 | md->recursive = rec->prevrec; | md->recursive = rec->prevrec; |
| 1165 | memmove(md->offset_vector, rec->offset_save, | memmove(md->offset_vector, rec->offset_save, |
| 1166 | rec->saved_max * sizeof(int)); | rec->saved_max * sizeof(int)); |
| 1167 | mstart = rec->save_start; | offset_top = rec->save_offset_top; |
| 1168 | ims = original_ims; | ims = original_ims; |
| 1169 | ecode = rec->after_call; | ecode = rec->after_call; |
| 1170 | break; | break; |
| 1171 | } | } |
| 1172 | ||
| 1173 | /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty | /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is |
| 1174 | string - backtracking will then try other alternatives, if any. */ | set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of |
| 1175 | the subject. In both cases, backtracking will then try other alternatives, | |
| 1176 | if any. */ | |
| 1177 | ||
| 1178 | if (eptr == mstart && | |
| 1179 | (md->notempty || | |
| 1180 | (md->notempty_atstart && | |
| 1181 | mstart == md->start_subject + md->start_offset))) | |
| 1182 | MRRETURN(MATCH_NOMATCH); | |
| 1183 | ||
| 1184 | /* Otherwise, we have a match. */ | |
| 1185 | ||
| if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH); | ||
| 1186 | md->end_match_ptr = eptr; /* Record where we ended */ | md->end_match_ptr = eptr; /* Record where we ended */ |
| 1187 | md->end_offset_top = offset_top; /* and how many extracts were taken */ | md->end_offset_top = offset_top; /* and how many extracts were taken */ |
| 1188 | md->start_match_ptr = mstart; /* and the start (\K can modify) */ | md->start_match_ptr = mstart; /* and the start (\K can modify) */ |
| 1189 | RRETURN(MATCH_MATCH); | |
| 1190 | /* For some reason, the macros don't work properly if an expression is | |
| 1191 | given as the argument to MRRETURN when the heap is in use. */ | |
| 1192 | ||
| 1193 | rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; | |
| 1194 | MRRETURN(rrc); | |
| 1195 | ||
| 1196 | /* Change option settings */ | /* Change option settings */ |
| 1197 | ||
| # | Line 881 for (;;) | Line 1213 for (;;) |
| 1213 | { | { |
| 1214 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, |
| 1215 | RM4); | RM4); |
| 1216 | if (rrc == MATCH_MATCH) break; | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
| 1217 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | { |
| 1218 | mstart = md->start_match_ptr; /* In case \K reset it */ | |
| 1219 | break; | |
| 1220 | } | |
| 1221 | if (rrc != MATCH_NOMATCH && | |
| 1222 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1223 | RRETURN(rrc); | |
| 1224 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 1225 | } | } |
| 1226 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| 1227 | if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); | if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH); |
| 1228 | ||
| 1229 | /* If checking an assertion for a condition, return MATCH_MATCH. */ | /* If checking an assertion for a condition, return MATCH_MATCH. */ |
| 1230 | ||
| # | Line 900 for (;;) | Line 1238 for (;;) |
| 1238 | offset_top = md->end_offset_top; | offset_top = md->end_offset_top; |
| 1239 | continue; | continue; |
| 1240 | ||
| 1241 | /* Negative assertion: all branches must fail to match */ | /* Negative assertion: all branches must fail to match. Encountering SKIP, |
| 1242 | PRUNE, or COMMIT means we must assume failure without checking subsequent | |
| 1243 | branches. */ | |
| 1244 | ||
| 1245 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
| 1246 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
| # | Line 908 for (;;) | Line 1248 for (;;) |
| 1248 | { | { |
| 1249 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, |
| 1250 | RM5); | RM5); |
| 1251 | if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH); | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH); |
| 1252 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) |
| 1253 | { | |
| 1254 | do ecode += GET(ecode,1); while (*ecode == OP_ALT); | |
| 1255 | break; | |
| 1256 | } | |
| 1257 | if (rrc != MATCH_NOMATCH && | |
| 1258 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1259 | RRETURN(rrc); | |
| 1260 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
| 1261 | } | } |
| 1262 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| # | Line 932 for (;;) | Line 1279 for (;;) |
| 1279 | while (i-- > 0) | while (i-- > 0) |
| 1280 | { | { |
| 1281 | eptr--; | eptr--; |
| 1282 | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1283 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 1284 | } | } |
| 1285 | } | } |
| # | Line 943 for (;;) | Line 1290 for (;;) |
| 1290 | ||
| 1291 | { | { |
| 1292 | eptr -= GET(ecode, 1); | eptr -= GET(ecode, 1); |
| 1293 | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1294 | } | } |
| 1295 | ||
| 1296 | /* Skip to next op code */ | /* Save the earliest consulted character, then skip to next op code */ |
| 1297 | ||
| 1298 | if (eptr < md->start_used_ptr) md->start_used_ptr = eptr; | |
| 1299 | ecode += 1 + LINK_SIZE; | ecode += 1 + LINK_SIZE; |
| 1300 | break; | break; |
| 1301 | ||
| # | Line 963 for (;;) | Line 1311 for (;;) |
| 1311 | cb.callout_number = ecode[1]; | cb.callout_number = ecode[1]; |
| 1312 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
| 1313 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
| 1314 | cb.subject_length = md->end_subject - md->start_subject; | cb.subject_length = (int)(md->end_subject - md->start_subject); |
| 1315 | cb.start_match = mstart - md->start_subject; | cb.start_match = (int)(mstart - md->start_subject); |
| 1316 | cb.current_position = eptr - md->start_subject; | cb.current_position = (int)(eptr - md->start_subject); |
| 1317 | cb.pattern_position = GET(ecode, 2); | cb.pattern_position = GET(ecode, 2); |
| 1318 | cb.next_item_length = GET(ecode, 2 + LINK_SIZE); | cb.next_item_length = GET(ecode, 2 + LINK_SIZE); |
| 1319 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 1320 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 1321 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 1322 | if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH); | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
| 1323 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 1324 | } | } |
| 1325 | ecode += 2 + 2*LINK_SIZE; | ecode += 2 + 2*LINK_SIZE; |
| # | Line 1026 for (;;) | Line 1374 for (;;) |
| 1374 | ||
| 1375 | memcpy(new_recursive.offset_save, md->offset_vector, | memcpy(new_recursive.offset_save, md->offset_vector, |
| 1376 | new_recursive.saved_max * sizeof(int)); | new_recursive.saved_max * sizeof(int)); |
| 1377 | new_recursive.save_start = mstart; | new_recursive.save_offset_top = offset_top; |
| mstart = eptr; | ||
| 1378 | ||
| 1379 | /* OK, now we can do the recursion. For each top-level alternative we | /* OK, now we can do the recursion. For each top-level alternative we |
| 1380 | restore the offset and recursion data. */ | restore the offset and recursion data. */ |
| # | Line 1038 for (;;) | Line 1385 for (;;) |
| 1385 | { | { |
| 1386 | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, |
| 1387 | md, ims, eptrb, flags, RM6); | md, ims, eptrb, flags, RM6); |
| 1388 | if (rrc == MATCH_MATCH) | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
| 1389 | { | { |
| 1390 | DPRINTF(("Recursion matched\n")); | DPRINTF(("Recursion matched\n")); |
| 1391 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1392 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1393 | (pcre_free)(new_recursive.offset_save); | (pcre_free)(new_recursive.offset_save); |
| 1394 | RRETURN(MATCH_MATCH); | MRRETURN(MATCH_MATCH); |
| 1395 | } | } |
| 1396 | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) | else if (rrc != MATCH_NOMATCH && |
| 1397 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1398 | { | { |
| 1399 | DPRINTF(("Recursion gave error %d\n", rrc)); | DPRINTF(("Recursion gave error %d\n", rrc)); |
| 1400 | if (new_recursive.offset_save != stacksave) | |
| 1401 | (pcre_free)(new_recursive.offset_save); | |
| 1402 | RRETURN(rrc); | RRETURN(rrc); |
| 1403 | } | } |
| 1404 | ||
| # | Line 1063 for (;;) | Line 1413 for (;;) |
| 1413 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1414 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1415 | (pcre_free)(new_recursive.offset_save); | (pcre_free)(new_recursive.offset_save); |
| 1416 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1417 | } | } |
| 1418 | /* Control never reaches here */ | /* Control never reaches here */ |
| 1419 | ||
| # | Line 1072 for (;;) | Line 1422 for (;;) |
| 1422 | a move back into the brackets. Friedl calls these "atomic" subpatterns. | a move back into the brackets. Friedl calls these "atomic" subpatterns. |
| 1423 | Check the alternative branches in turn - the matching won't pass the KET | Check the alternative branches in turn - the matching won't pass the KET |
| 1424 | for this kind of subpattern. If any one branch matches, we carry on as at | for this kind of subpattern. If any one branch matches, we carry on as at |
| 1425 | the end of a normal bracket, leaving the subject pointer. */ | the end of a normal bracket, leaving the subject pointer, but resetting |
| 1426 | the start-of-match value in case it was changed by \K. */ | |
| 1427 | ||
| 1428 | case OP_ONCE: | case OP_ONCE: |
| 1429 | prev = ecode; | prev = ecode; |
| # | Line 1081 for (;;) | Line 1432 for (;;) |
| 1432 | do | do |
| 1433 | { | { |
| 1434 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7); | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7); |
| 1435 | if (rrc == MATCH_MATCH) break; | if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ |
| 1436 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | { |
| 1437 | mstart = md->start_match_ptr; | |
| 1438 | break; | |
| 1439 | } | |
| 1440 | if (rrc != MATCH_NOMATCH && | |
| 1441 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1442 | RRETURN(rrc); | |
| 1443 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
| 1444 | } | } |
| 1445 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| # | Line 1148 for (;;) | Line 1505 for (;;) |
| 1505 | do ecode += GET(ecode,1); while (*ecode == OP_ALT); | do ecode += GET(ecode,1); while (*ecode == OP_ALT); |
| 1506 | break; | break; |
| 1507 | ||
| 1508 | /* BRAZERO and BRAMINZERO occur just before a bracket group, indicating | /* BRAZERO, BRAMINZERO and SKIPZERO occur just before a bracket group, |
| 1509 | that it may occur zero times. It may repeat infinitely, or not at all - | indicating that it may occur zero times. It may repeat infinitely, or not |
| 1510 | i.e. it could be ()* or ()? in the pattern. Brackets with fixed upper | at all - i.e. it could be ()* or ()? or even (){0} in the pattern. Brackets |
| 1511 | repeat limits are compiled as a number of copies, with the optional ones | with fixed upper repeat limits are compiled as a number of copies, with the |
| 1512 | preceded by BRAZERO or BRAMINZERO. */ | optional ones preceded by BRAZERO or BRAMINZERO. */ |
| 1513 | ||
| 1514 | case OP_BRAZERO: | case OP_BRAZERO: |
| 1515 | { | { |
| # | Line 1174 for (;;) | Line 1531 for (;;) |
| 1531 | } | } |
| 1532 | break; | break; |
| 1533 | ||
| 1534 | case OP_SKIPZERO: | |
| 1535 | { | |
| 1536 | next = ecode+1; | |
| 1537 | do next += GET(next,1); while (*next == OP_ALT); | |
| 1538 | ecode = next + 1 + LINK_SIZE; | |
| 1539 | } | |
| 1540 | break; | |
| 1541 | ||
| 1542 | /* End of a group, repeated or non-repeating. */ | /* End of a group, repeated or non-repeating. */ |
| 1543 | ||
| 1544 | case OP_KET: | case OP_KET: |
| # | Line 1192 for (;;) | Line 1557 for (;;) |
| 1557 | } | } |
| 1558 | else saved_eptr = NULL; | else saved_eptr = NULL; |
| 1559 | ||
| 1560 | /* 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 an atomic group, stop |
| 1561 | MATCH_MATCH, but record the current high water mark for use by positive | matching and return MATCH_MATCH, but record the current high water mark for |
| 1562 | assertions. Do this also for the "once" (atomic) groups. */ | use by positive assertions. We also need to record the match start in case |
| 1563 | it was changed by \K. */ | |
| 1564 | ||
| 1565 | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || |
| 1566 | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || |
| # | Line 1202 for (;;) | Line 1568 for (;;) |
| 1568 | { | { |
| 1569 | md->end_match_ptr = eptr; /* For ONCE */ | md->end_match_ptr = eptr; /* For ONCE */ |
| 1570 | md->end_offset_top = offset_top; | md->end_offset_top = offset_top; |
| 1571 | RRETURN(MATCH_MATCH); | md->start_match_ptr = mstart; |
| 1572 | MRRETURN(MATCH_MATCH); | |
| 1573 | } | } |
| 1574 | ||
| 1575 | /* 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 1216 for (;;) | Line 1583 for (;;) |
| 1583 | number = GET2(prev, 1+LINK_SIZE); | number = GET2(prev, 1+LINK_SIZE); |
| 1584 | offset = number << 1; | offset = number << 1; |
| 1585 | ||
| 1586 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 1587 | printf("end bracket %d", number); | printf("end bracket %d", number); |
| 1588 | printf("\n"); | printf("\n"); |
| 1589 | #endif | #endif |
| # | Line 1226 for (;;) | Line 1593 for (;;) |
| 1593 | { | { |
| 1594 | md->offset_vector[offset] = | md->offset_vector[offset] = |
| 1595 | md->offset_vector[md->offset_end - number]; | md->offset_vector[md->offset_end - number]; |
| 1596 | md->offset_vector[offset+1] = eptr - md->start_subject; | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); |
| 1597 | if (offset_top <= offset) offset_top = offset + 2; | if (offset_top <= offset) offset_top = offset + 2; |
| 1598 | } | } |
| 1599 | ||
| # | Line 1238 for (;;) | Line 1605 for (;;) |
| 1605 | recursion_info *rec = md->recursive; | recursion_info *rec = md->recursive; |
| 1606 | DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); | DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); |
| 1607 | md->recursive = rec->prevrec; | md->recursive = rec->prevrec; |
| mstart = rec->save_start; | ||
| 1608 | memcpy(md->offset_vector, rec->offset_save, | memcpy(md->offset_vector, rec->offset_save, |
| 1609 | rec->saved_max * sizeof(int)); | rec->saved_max * sizeof(int)); |
| 1610 | offset_top = rec->save_offset_top; | |
| 1611 | ecode = rec->after_call; | ecode = rec->after_call; |
| 1612 | ims = original_ims; | ims = original_ims; |
| 1613 | break; | break; |
| # | Line 1297 for (;;) | Line 1664 for (;;) |
| 1664 | /* Start of subject unless notbol, or after internal newline if multiline */ | /* Start of subject unless notbol, or after internal newline if multiline */ |
| 1665 | ||
| 1666 | case OP_CIRC: | case OP_CIRC: |
| 1667 | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1668 | if ((ims & PCRE_MULTILINE) != 0) | if ((ims & PCRE_MULTILINE) != 0) |
| 1669 | { | { |
| 1670 | if (eptr != md->start_subject && | if (eptr != md->start_subject && |
| 1671 | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) |
| 1672 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1673 | ecode++; | ecode++; |
| 1674 | break; | break; |
| 1675 | } | } |
| # | Line 1311 for (;;) | Line 1678 for (;;) |
| 1678 | /* Start of subject assertion */ | /* Start of subject assertion */ |
| 1679 | ||
| 1680 | case OP_SOD: | case OP_SOD: |
| 1681 | if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1682 | ecode++; | ecode++; |
| 1683 | break; | break; |
| 1684 | ||
| 1685 | /* Start of match assertion */ | /* Start of match assertion */ |
| 1686 | ||
| 1687 | case OP_SOM: | case OP_SOM: |
| 1688 | if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH); |
| 1689 | ecode++; | ecode++; |
| 1690 | break; | break; |
| 1691 | ||
| # | Line 1336 for (;;) | Line 1703 for (;;) |
| 1703 | if ((ims & PCRE_MULTILINE) != 0) | if ((ims & PCRE_MULTILINE) != 0) |
| 1704 | { | { |
| 1705 | if (eptr < md->end_subject) | if (eptr < md->end_subject) |
| 1706 | { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); } | { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); } |
| 1707 | else | else |
| 1708 | { if (md->noteol) RRETURN(MATCH_NOMATCH); } | { |
| 1709 | if (md->noteol) MRRETURN(MATCH_NOMATCH); | |
| 1710 | SCHECK_PARTIAL(); | |
| 1711 | } | |
| 1712 | ecode++; | ecode++; |
| 1713 | break; | break; |
| 1714 | } | } |
| 1715 | else | else /* Not multiline */ |
| 1716 | { | { |
| 1717 | if (md->noteol) RRETURN(MATCH_NOMATCH); | if (md->noteol) MRRETURN(MATCH_NOMATCH); |
| 1718 | if (!md->endonly) | if (!md->endonly) goto ASSERT_NL_OR_EOS; |
| { | ||
| if (eptr != md->end_subject && | ||
| (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | ||
| RRETURN(MATCH_NOMATCH); | ||
| ecode++; | ||
| break; | ||
| } | ||
| 1719 | } | } |
| 1720 | ||
| 1721 | /* ... else fall through for endonly */ | /* ... else fall through for endonly */ |
| 1722 | ||
| 1723 | /* End of subject assertion (\z) */ | /* End of subject assertion (\z) */ |
| 1724 | ||
| 1725 | case OP_EOD: | case OP_EOD: |
| 1726 | if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH); |
| 1727 | SCHECK_PARTIAL(); | |
| 1728 | ecode++; | ecode++; |
| 1729 | break; | break; |
| 1730 | ||
| 1731 | /* End of subject or ending \n assertion (\Z) */ | /* End of subject or ending \n assertion (\Z) */ |
| 1732 | ||
| 1733 | case OP_EODN: | case OP_EODN: |
| 1734 | if (eptr != md->end_subject && | ASSERT_NL_OR_EOS: |
| 1735 | if (eptr < md->end_subject && | |
| 1736 | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) |
| 1737 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1738 | ||
| 1739 | /* Either at end of string or \n before end. */ | |
| 1740 | ||
| 1741 | SCHECK_PARTIAL(); | |
| 1742 | ecode++; | ecode++; |
| 1743 | break; | break; |
| 1744 | ||
| # | Line 1380 for (;;) | Line 1750 for (;;) |
| 1750 | ||
| 1751 | /* Find out if the previous and current characters are "word" characters. | /* Find out if the previous and current characters are "word" characters. |
| 1752 | It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to | It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to |
| 1753 | be "non-word" characters. */ | be "non-word" characters. Remember the earliest consulted character for |
| 1754 | partial matching. */ | |
| 1755 | ||
| 1756 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 1757 | if (utf8) | if (utf8) |
| 1758 | { | { |
| 1759 | /* Get status of previous character */ | |
| 1760 | ||
| 1761 | if (eptr == md->start_subject) prev_is_word = FALSE; else | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 1762 | { | { |
| 1763 | const uschar *lastptr = eptr - 1; | USPTR lastptr = eptr - 1; |
| 1764 | while((*lastptr & 0xc0) == 0x80) lastptr--; | while((*lastptr & 0xc0) == 0x80) lastptr--; |
| 1765 | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; | |
| 1766 | GETCHAR(c, lastptr); | GETCHAR(c, lastptr); |
| 1767 | #ifdef SUPPORT_UCP | |
| 1768 | if (md->use_ucp) | |
| 1769 | { | |
| 1770 | if (c == '_') prev_is_word = TRUE; else | |
| 1771 | { | |
| 1772 | int cat = UCD_CATEGORY(c); | |
| 1773 | prev_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1774 | } | |
| 1775 | } | |
| 1776 | else | |
| 1777 | #endif | |
| 1778 | prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; | prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; |
| 1779 | } | } |
| 1780 | if (eptr >= md->end_subject) cur_is_word = FALSE; else | |
| 1781 | /* Get status of next character */ | |
| 1782 | ||
| 1783 | if (eptr >= md->end_subject) | |
| 1784 | { | |
| 1785 | SCHECK_PARTIAL(); | |
| 1786 | cur_is_word = FALSE; | |
| 1787 | } | |
| 1788 | else | |
| 1789 | { | { |
| 1790 | GETCHAR(c, eptr); | GETCHAR(c, eptr); |
| 1791 | #ifdef SUPPORT_UCP | |
| 1792 | if (md->use_ucp) | |
| 1793 | { | |
| 1794 | if (c == '_') cur_is_word = TRUE; else | |
| 1795 | { | |
| 1796 | int cat = UCD_CATEGORY(c); | |
| 1797 | cur_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1798 | } | |
| 1799 | } | |
| 1800 | else | |
| 1801 | #endif | |
| 1802 | cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; | cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; |
| 1803 | } | } |
| 1804 | } | } |
| 1805 | else | else |
| 1806 | #endif | #endif |
| 1807 | ||
| 1808 | /* More streamlined when not in UTF-8 mode */ | /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for |
| 1809 | consistency with the behaviour of \w we do use it in this case. */ | |
| 1810 | ||
| 1811 | { | { |
| 1812 | prev_is_word = (eptr != md->start_subject) && | /* Get status of previous character */ |
| 1813 | ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
| 1814 | cur_is_word = (eptr < md->end_subject) && | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 1815 | ((md->ctypes[*eptr] & ctype_word) != 0); | { |
| 1816 | if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1; | |
| 1817 | #ifdef SUPPORT_UCP | |
| 1818 | if (md->use_ucp) | |
| 1819 | { | |
| 1820 | c = eptr[-1]; | |
| 1821 | if (c == '_') prev_is_word = TRUE; else | |
| 1822 | { | |
| 1823 | int cat = UCD_CATEGORY(c); | |
| 1824 | prev_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1825 | } | |
| 1826 | } | |
| 1827 | else | |
| 1828 | #endif | |
| 1829 | prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
| 1830 | } | |
| 1831 | ||
| 1832 | /* Get status of next character */ | |
| 1833 | ||
| 1834 | if (eptr >= md->end_subject) | |
| 1835 | { | |
| 1836 | SCHECK_PARTIAL(); | |
| 1837 | cur_is_word = FALSE; | |
| 1838 | } | |
| 1839 | else | |
| 1840 | #ifdef SUPPORT_UCP | |
| 1841 | if (md->use_ucp) | |
| 1842 | { | |
| 1843 | c = *eptr; | |
| 1844 | if (c == '_') cur_is_word = TRUE; else | |
| 1845 | { | |
| 1846 | int cat = UCD_CATEGORY(c); | |
| 1847 | cur_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1848 | } | |
| 1849 | } | |
| 1850 | else | |
| 1851 | #endif | |
| 1852 | cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0); | |
| 1853 | } | } |
| 1854 | ||
| 1855 | /* Now see if the situation is what we want */ | /* Now see if the situation is what we want */ |
| 1856 | ||
| 1857 | if ((*ecode++ == OP_WORD_BOUNDARY)? | if ((*ecode++ == OP_WORD_BOUNDARY)? |
| 1858 | cur_is_word == prev_is_word : cur_is_word != prev_is_word) | cur_is_word == prev_is_word : cur_is_word != prev_is_word) |
| 1859 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1860 | } | } |
| 1861 | break; | break; |
| 1862 | ||
| 1863 | /* Match a single character type; inline for speed */ | /* Match a single character type; inline for speed */ |
| 1864 | ||
| 1865 | case OP_ANY: | case OP_ANY: |
| 1866 | if ((ims & PCRE_DOTALL) == 0) | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); |
| 1867 | /* Fall through */ | |
| 1868 | ||
| 1869 | case OP_ALLANY: | |
| 1870 | if (eptr++ >= md->end_subject) | |
| 1871 | { | { |
| 1872 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 1873 | MRRETURN(MATCH_NOMATCH); | |
| 1874 | } | } |
| 1875 | if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| if (utf8) | ||
| while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | ||
| 1876 | ecode++; | ecode++; |
| 1877 | break; | break; |
| 1878 | ||
| # | Line 1435 for (;;) | Line 1880 for (;;) |
| 1880 | any byte, even newline, independent of the setting of PCRE_DOTALL. */ | any byte, even newline, independent of the setting of PCRE_DOTALL. */ |
| 1881 | ||
| 1882 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 1883 | if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr++ >= md->end_subject) |
| 1884 | { | |
| 1885 | SCHECK_PARTIAL(); | |
| 1886 | MRRETURN(MATCH_NOMATCH); | |
| 1887 | } | |
| 1888 | ecode++; | ecode++; |
| 1889 | break; | break; |
| 1890 | ||
| 1891 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 1892 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1893 | { | |
| 1894 | SCHECK_PARTIAL(); | |
| 1895 | MRRETURN(MATCH_NOMATCH); | |
| 1896 | } | |
| 1897 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1898 | if ( | if ( |
| 1899 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1448 for (;;) | Line 1901 for (;;) |
| 1901 | #endif | #endif |
| 1902 | (md->ctypes[c] & ctype_digit) != 0 | (md->ctypes[c] & ctype_digit) != 0 |
| 1903 | ) | ) |
| 1904 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1905 | ecode++; | ecode++; |
| 1906 | break; | break; |
| 1907 | ||
| 1908 | case OP_DIGIT: | case OP_DIGIT: |
| 1909 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1910 | { | |
| 1911 | SCHECK_PARTIAL(); | |
| 1912 | MRRETURN(MATCH_NOMATCH); | |
| 1913 | } | |
| 1914 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1915 | if ( | if ( |
| 1916 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1461 for (;;) | Line 1918 for (;;) |
| 1918 | #endif | #endif |
| 1919 | (md->ctypes[c] & ctype_digit) == 0 | (md->ctypes[c] & ctype_digit) == 0 |
| 1920 | ) | ) |
| 1921 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1922 | ecode++; | ecode++; |
| 1923 | break; | break; |
| 1924 | ||
| 1925 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 1926 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1927 | { | |
| 1928 | SCHECK_PARTIAL(); | |
| 1929 | MRRETURN(MATCH_NOMATCH); | |
| 1930 | } | |
| 1931 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1932 | if ( | if ( |
| 1933 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1474 for (;;) | Line 1935 for (;;) |
| 1935 | #endif | #endif |
| 1936 | (md->ctypes[c] & ctype_space) != 0 | (md->ctypes[c] & ctype_space) != 0 |
| 1937 | ) | ) |
| 1938 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1939 | ecode++; | ecode++; |
| 1940 | break; | break; |
| 1941 | ||
| 1942 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 1943 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1944 | { | |
| 1945 | SCHECK_PARTIAL(); | |
| 1946 | MRRETURN(MATCH_NOMATCH); | |
| 1947 | } | |
| 1948 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1949 | if ( | if ( |
| 1950 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1487 for (;;) | Line 1952 for (;;) |
| 1952 | #endif | #endif |
| 1953 | (md->ctypes[c] & ctype_space) == 0 | (md->ctypes[c] & ctype_space) == 0 |
| 1954 | ) | ) |
| 1955 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1956 | ecode++; | ecode++; |
| 1957 | break; | break; |
| 1958 | ||
| 1959 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 1960 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1961 | { | |
| 1962 | SCHECK_PARTIAL(); | |
| 1963 | MRRETURN(MATCH_NOMATCH); | |
| 1964 | } | |
| 1965 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1966 | if ( | if ( |
| 1967 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1500 for (;;) | Line 1969 for (;;) |
| 1969 | #endif | #endif |
| 1970 | (md->ctypes[c] & ctype_word) != 0 | (md->ctypes[c] & ctype_word) != 0 |
| 1971 | ) | ) |
| 1972 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1973 | ecode++; | ecode++; |
| 1974 | break; | break; |
| 1975 | ||
| 1976 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 1977 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1978 | { | |
| 1979 | SCHECK_PARTIAL(); | |
| 1980 | MRRETURN(MATCH_NOMATCH); | |
| 1981 | } | |
| 1982 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1983 | if ( | if ( |
| 1984 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1513 for (;;) | Line 1986 for (;;) |
| 1986 | #endif | #endif |
| 1987 | (md->ctypes[c] & ctype_word) == 0 | (md->ctypes[c] & ctype_word) == 0 |
| 1988 | ) | ) |
| 1989 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1990 | ecode++; | ecode++; |
| 1991 | break; | break; |
| 1992 | ||
| 1993 | case OP_ANYNL: | case OP_ANYNL: |
| 1994 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1995 | { | |
| 1996 | SCHECK_PARTIAL(); | |
| 1997 | MRRETURN(MATCH_NOMATCH); | |
| 1998 | } | |
| 1999 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2000 | switch(c) | switch(c) |
| 2001 | { | { |
| 2002 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2003 | case 0x000d: | case 0x000d: |
| 2004 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 2005 | break; | break; |
| 2006 | ||
| 2007 | case 0x000a: | case 0x000a: |
| 2008 | break; | |
| 2009 | ||
| 2010 | case 0x000b: | case 0x000b: |
| 2011 | case 0x000c: | case 0x000c: |
| 2012 | case 0x0085: | case 0x0085: |
| 2013 | case 0x2028: | case 0x2028: |
| 2014 | case 0x2029: | case 0x2029: |
| 2015 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | |
| 2016 | break; | break; |
| 2017 | } | } |
| 2018 | ecode++; | ecode++; |
| 2019 | break; | break; |
| 2020 | ||
| 2021 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 2022 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2023 | { | |
| 2024 | SCHECK_PARTIAL(); | |
| 2025 | MRRETURN(MATCH_NOMATCH); | |
| 2026 | } | |
| 2027 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2028 | switch(c) | switch(c) |
| 2029 | { | { |
| # | Line 1562 for (;;) | Line 2047 for (;;) |
| 2047 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 2048 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 2049 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 2050 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2051 | } | } |
| 2052 | ecode++; | ecode++; |
| 2053 | break; | break; |
| 2054 | ||
| 2055 | case OP_HSPACE: | case OP_HSPACE: |
| 2056 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2057 | { | |
| 2058 | SCHECK_PARTIAL(); | |
| 2059 | MRRETURN(MATCH_NOMATCH); | |
| 2060 | } | |
| 2061 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2062 | switch(c) | switch(c) |
| 2063 | { | { |
| 2064 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2065 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 2066 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 2067 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 1598 for (;;) | Line 2087 for (;;) |
| 2087 | break; | break; |
| 2088 | ||
| 2089 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 2090 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2091 | { | |
| 2092 | SCHECK_PARTIAL(); | |
| 2093 | MRRETURN(MATCH_NOMATCH); | |
| 2094 | } | |
| 2095 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2096 | switch(c) | switch(c) |
| 2097 | { | { |
| # | Line 1610 for (;;) | Line 2103 for (;;) |
| 2103 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 2104 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 2105 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 2106 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2107 | } | } |
| 2108 | ecode++; | ecode++; |
| 2109 | break; | break; |
| 2110 | ||
| 2111 | case OP_VSPACE: | case OP_VSPACE: |
| 2112 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2113 | { | |
| 2114 | SCHECK_PARTIAL(); | |
| 2115 | MRRETURN(MATCH_NOMATCH); | |
| 2116 | } | |
| 2117 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2118 | switch(c) | switch(c) |
| 2119 | { | { |
| 2120 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2121 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 2122 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 2123 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 1639 for (;;) | Line 2136 for (;;) |
| 2136 | ||
| 2137 | case OP_PROP: | case OP_PROP: |
| 2138 | case OP_NOTPROP: | case OP_NOTPROP: |
| 2139 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2140 | { | |
| 2141 | SCHECK_PARTIAL(); | |
| 2142 | MRRETURN(MATCH_NOMATCH); | |
| 2143 | } | |
| 2144 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2145 | { | { |
| 2146 | int chartype, script; | const ucd_record *prop = GET_UCD(c); |
| int category = _pcre_ucp_findprop(c, &chartype, &script); | ||
| 2147 | ||
| 2148 | switch(ecode[1]) | switch(ecode[1]) |
| 2149 | { | { |
| 2150 | case PT_ANY: | case PT_ANY: |
| 2151 | if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); | if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH); |
| 2152 | break; | break; |
| 2153 | ||
| 2154 | case PT_LAMP: | case PT_LAMP: |
| 2155 | if ((chartype == ucp_Lu || | if ((prop->chartype == ucp_Lu || |
| 2156 | chartype == ucp_Ll || | prop->chartype == ucp_Ll || |
| 2157 | chartype == ucp_Lt) == (op == OP_NOTPROP)) | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) |
| 2158 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2159 | break; | break; |
| 2160 | ||
| 2161 | case PT_GC: | case PT_GC: |
| 2162 | if ((ecode[2] != category) == (op == OP_PROP)) | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) |
| 2163 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2164 | break; | break; |
| 2165 | ||
| 2166 | case PT_PC: | case PT_PC: |
| 2167 | if ((ecode[2] != chartype) == (op == OP_PROP)) | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) |
| 2168 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2169 | break; | break; |
| 2170 | ||
| 2171 | case PT_SC: | case PT_SC: |
| 2172 | if ((ecode[2] != script) == (op == OP_PROP)) | if ((ecode[2] != prop->script) == (op == OP_PROP)) |
| 2173 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2174 | break; | |
| 2175 | ||
| 2176 | /* These are specials */ | |
| 2177 | ||
| 2178 | case PT_ALNUM: | |
| 2179 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | |
| 2180 | _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) | |
| 2181 | MRRETURN(MATCH_NOMATCH); | |
| 2182 | break; | break; |
| 2183 | ||
| 2184 | case PT_SPACE: /* Perl space */ | |
| 2185 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | |
| 2186 | c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) | |
| 2187 | == (op == OP_NOTPROP)) | |
| 2188 | MRRETURN(MATCH_NOMATCH); | |
| 2189 | break; | |
| 2190 | ||
| 2191 | case PT_PXSPACE: /* POSIX space */ | |
| 2192 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | |
| 2193 | c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || | |
| 2194 | c == CHAR_FF || c == CHAR_CR) | |
| 2195 | == (op == OP_NOTPROP)) | |
| 2196 | MRRETURN(MATCH_NOMATCH); | |
| 2197 | break; | |
| 2198 | ||
| 2199 | case PT_WORD: | |
| 2200 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | |
| 2201 | _pcre_ucp_gentype[prop->chartype] == ucp_N || | |
| 2202 | c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) | |
| 2203 | MRRETURN(MATCH_NOMATCH); | |
| 2204 | break; | |
| 2205 | ||
| 2206 | /* This should never occur */ | |
| 2207 | ||
| 2208 | default: | default: |
| 2209 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 2210 | } | } |
| # | Line 1685 for (;;) | Line 2217 for (;;) |
| 2217 | is in the binary; otherwise a compile-time error occurs. */ | is in the binary; otherwise a compile-time error occurs. */ |
| 2218 | ||
| 2219 | case OP_EXTUNI: | case OP_EXTUNI: |
| 2220 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| GETCHARINCTEST(c, eptr); | ||
| 2221 | { | { |
| 2222 | int chartype, script; | SCHECK_PARTIAL(); |
| 2223 | int category = _pcre_ucp_findprop(c, &chartype, &script); | MRRETURN(MATCH_NOMATCH); |
| 2224 | if (category == ucp_M) RRETURN(MATCH_NOMATCH); | } |
| 2225 | GETCHARINCTEST(c, eptr); | |
| 2226 | { | |
| 2227 | int category = UCD_CATEGORY(c); | |
| 2228 | if (category == ucp_M) MRRETURN(MATCH_NOMATCH); | |
| 2229 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 2230 | { | { |
| 2231 | int len = 1; | int len = 1; |
| # | Line 1698 for (;;) | Line 2233 for (;;) |
| 2233 | { | { |
| 2234 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 2235 | } | } |
| 2236 | category = _pcre_ucp_findprop(c, &chartype, &script); | category = UCD_CATEGORY(c); |
| 2237 | if (category != ucp_M) break; | if (category != ucp_M) break; |
| 2238 | eptr += len; | eptr += len; |
| 2239 | } | } |
| # | Line 1719 for (;;) | Line 2254 for (;;) |
| 2254 | case OP_REF: | case OP_REF: |
| 2255 | { | { |
| 2256 | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ |
| 2257 | ecode += 3; /* Advance past item */ | ecode += 3; |
| 2258 | ||
| 2259 | /* If the reference is unset, there are two possibilities: | |
| 2260 | ||
| 2261 | (a) In the default, Perl-compatible state, set the length to be longer | |
| 2262 | than the amount of subject left; this ensures that every attempt at a | |
| 2263 | match fails. We can't just fail here, because of the possibility of | |
| 2264 | quantifiers with zero minima. | |
| 2265 | ||
| 2266 | (b) If the JavaScript compatibility flag is set, set the length to zero | |
| 2267 | so that the back reference matches an empty string. | |
| 2268 | ||
| 2269 | Otherwise, set the length to the length of what was matched by the | |
| 2270 | referenced subpattern. */ | |
| 2271 | ||
| 2272 | /* If the reference is unset, set the length to be longer than the amount | if (offset >= offset_top || md->offset_vector[offset] < 0) |
| 2273 | of subject left; this ensures that every attempt at a match fails. We | length = (md->jscript_compat)? 0 : (int)(md->end_subject - eptr + 1); |
| 2274 | can't just fail here, because of the possibility of quantifiers with zero | else |
| 2275 | minima. */ | length = md->offset_vector[offset+1] - md->offset_vector[offset]; |
| length = (offset >= offset_top || md->offset_vector[offset] < 0)? | ||
| md->end_subject - eptr + 1 : | ||
| md->offset_vector[offset+1] - md->offset_vector[offset]; | ||
| 2276 | ||
| 2277 | /* Set up for repetition, or handle the non-repeated case */ | /* Set up for repetition, or handle the non-repeated case */ |
| 2278 | ||
| # | Line 1757 for (;;) | Line 2301 for (;;) |
| 2301 | break; | break; |
| 2302 | ||
| 2303 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| 2304 | if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); | if (!match_ref(offset, eptr, length, md, ims)) |
| 2305 | { | |
| 2306 | CHECK_PARTIAL(); | |
| 2307 | MRRETURN(MATCH_NOMATCH); | |
| 2308 | } | |
| 2309 | eptr += length; | eptr += length; |
| 2310 | continue; /* With the main loop */ | continue; /* With the main loop */ |
| 2311 | } | } |
| # | Line 1773 for (;;) | Line 2321 for (;;) |
| 2321 | ||
| 2322 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2323 | { | { |
| 2324 | if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); | if (!match_ref(offset, eptr, length, md, ims)) |
| 2325 | { | |
| 2326 | CHECK_PARTIAL(); | |
| 2327 | MRRETURN(MATCH_NOMATCH); | |
| 2328 | } | |
| 2329 | eptr += length; | eptr += length; |
| 2330 | } | } |
| 2331 | ||
| # | Line 1790 for (;;) | Line 2342 for (;;) |
| 2342 | { | { |
| 2343 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14); |
| 2344 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2345 | if (fi >= max || !match_ref(offset, eptr, length, md, ims)) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2346 | RRETURN(MATCH_NOMATCH); | if (!match_ref(offset, eptr, length, md, ims)) |
| 2347 | { | |
| 2348 | CHECK_PARTIAL(); | |
| 2349 | MRRETURN(MATCH_NOMATCH); | |
| 2350 | } | |
| 2351 | eptr += length; | eptr += length; |
| 2352 | } | } |
| 2353 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 1804 for (;;) | Line 2360 for (;;) |
| 2360 | pp = eptr; | pp = eptr; |
| 2361 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2362 | { | { |
| 2363 | if (!match_ref(offset, eptr, length, md, ims)) break; | if (!match_ref(offset, eptr, length, md, ims)) |
| 2364 | { | |
| 2365 | CHECK_PARTIAL(); | |
| 2366 | break; | |
| 2367 | } | |
| 2368 | eptr += length; | eptr += length; |
| 2369 | } | } |
| 2370 | while (eptr >= pp) | while (eptr >= pp) |
| # | Line 1813 for (;;) | Line 2373 for (;;) |
| 2373 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2374 | eptr -= length; | eptr -= length; |
| 2375 | } | } |
| 2376 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2377 | } | } |
| 2378 | } | } |
| 2379 | /* Control never gets here */ | /* Control never gets here */ |
| 2380 | ||
| 2381 | /* Match a bit-mapped character class, possibly repeatedly. This op code is | /* Match a bit-mapped character class, possibly repeatedly. This op code is |
| 2382 | used when all the characters in the class have values in the range 0-255, | used when all the characters in the class have values in the range 0-255, |
| 2383 | and either the matching is caseful, or the characters are in the range | and either the matching is caseful, or the characters are in the range |
| # | Line 1874 for (;;) | Line 2432 for (;;) |
| 2432 | { | { |
| 2433 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2434 | { | { |
| 2435 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2436 | { | |
| 2437 | SCHECK_PARTIAL(); | |
| 2438 | MRRETURN(MATCH_NOMATCH); | |
| 2439 | } | |
| 2440 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2441 | if (c > 255) | if (c > 255) |
| 2442 | { | { |
| 2443 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); |
| 2444 | } | } |
| 2445 | else | else |
| 2446 | { | { |
| 2447 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2448 | } | } |
| 2449 | } | } |
| 2450 | } | } |
| # | Line 1892 for (;;) | Line 2454 for (;;) |
| 2454 | { | { |
| 2455 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2456 | { | { |
| 2457 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2458 | { | |
| 2459 | SCHECK_PARTIAL(); | |
| 2460 | MRRETURN(MATCH_NOMATCH); | |
| 2461 | } | |
| 2462 | c = *eptr++; | c = *eptr++; |
| 2463 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2464 | } | } |
| 2465 | } | } |
| 2466 | ||
| # | Line 1916 for (;;) | Line 2482 for (;;) |
| 2482 | { | { |
| 2483 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16); |
| 2484 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2485 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2486 | if (eptr >= md->end_subject) | |
| 2487 | { | |
| 2488 | SCHECK_PARTIAL(); | |
| 2489 | MRRETURN(MATCH_NOMATCH); | |
| 2490 | } | |
| 2491 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2492 | if (c > 255) | if (c > 255) |
| 2493 | { | { |
| 2494 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); |
| 2495 | } | } |
| 2496 | else | else |
| 2497 | { | { |
| 2498 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2499 | } | } |
| 2500 | } | } |
| 2501 | } | } |
| # | Line 1936 for (;;) | Line 2507 for (;;) |
| 2507 | { | { |
| 2508 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17); |
| 2509 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2510 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2511 | if (eptr >= md->end_subject) | |
| 2512 | { | |
| 2513 | SCHECK_PARTIAL(); | |
| 2514 | MRRETURN(MATCH_NOMATCH); | |
| 2515 | } | |
| 2516 | c = *eptr++; | c = *eptr++; |
| 2517 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2518 | } | } |
| 2519 | } | } |
| 2520 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 1957 for (;;) | Line 2533 for (;;) |
| 2533 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2534 | { | { |
| 2535 | int len = 1; | int len = 1; |
| 2536 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2537 | { | |
| 2538 | SCHECK_PARTIAL(); | |
| 2539 | break; | |
| 2540 | } | |
| 2541 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 2542 | if (c > 255) | if (c > 255) |
| 2543 | { | { |
| # | Line 1983 for (;;) | Line 2563 for (;;) |
| 2563 | { | { |
| 2564 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2565 | { | { |
| 2566 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2567 | { | |
| 2568 | SCHECK_PARTIAL(); | |
| 2569 | break; | |
| 2570 | } | |
| 2571 | c = *eptr; | c = *eptr; |
| 2572 | if ((data[c/8] & (1 << (c&7))) == 0) break; | if ((data[c/8] & (1 << (c&7))) == 0) break; |
| 2573 | eptr++; | eptr++; |
| # | Line 1996 for (;;) | Line 2580 for (;;) |
| 2580 | } | } |
| 2581 | } | } |
| 2582 | ||
| 2583 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2584 | } | } |
| 2585 | } | } |
| 2586 | /* Control never gets here */ | /* Control never gets here */ |
| 2587 | ||
| 2588 | ||
| 2589 | /* Match an extended character class. This opcode is encountered only | /* Match an extended character class. This opcode is encountered only |
| 2590 | in UTF-8 mode, because that's the only time it is compiled. */ | when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8 |
| 2591 | mode, because Unicode properties are supported in non-UTF-8 mode. */ | |
| 2592 | ||
| 2593 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 2594 | case OP_XCLASS: | case OP_XCLASS: |
| # | Line 2044 for (;;) | Line 2629 for (;;) |
| 2629 | ||
| 2630 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2631 | { | { |
| 2632 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2633 | GETCHARINC(c, eptr); | { |
| 2634 | if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 2635 | MRRETURN(MATCH_NOMATCH); | |
| 2636 | } | |
| 2637 | GETCHARINCTEST(c, eptr); | |
| 2638 | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); | |
| 2639 | } | } |
| 2640 | ||
| 2641 | /* 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 2063 for (;;) | Line 2652 for (;;) |
| 2652 | { | { |
| 2653 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20); |
| 2654 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2655 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2656 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 2657 | if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); | { |
| 2658 | SCHECK_PARTIAL(); | |
| 2659 | MRRETURN(MATCH_NOMATCH); | |
| 2660 | } | |
| 2661 | GETCHARINCTEST(c, eptr); | |
| 2662 | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); | |
| 2663 | } | } |
| 2664 | /* Control never gets here */ | /* Control never gets here */ |
| 2665 | } | } |
| # | Line 2078 for (;;) | Line 2672 for (;;) |
| 2672 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2673 | { | { |
| 2674 | int len = 1; | int len = 1; |
| 2675 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2676 | GETCHARLEN(c, eptr, len); | { |
| 2677 | SCHECK_PARTIAL(); | |
| 2678 | break; | |
| 2679 | } | |
| 2680 | GETCHARLENTEST(c, eptr, len); | |
| 2681 | if (!_pcre_xclass(c, data)) break; | if (!_pcre_xclass(c, data)) break; |
| 2682 | eptr += len; | eptr += len; |
| 2683 | } | } |
| # | Line 2090 for (;;) | Line 2688 for (;;) |
| 2688 | if (eptr-- == pp) break; /* Stop if tried at original pos */ | if (eptr-- == pp) break; /* Stop if tried at original pos */ |
| 2689 | if (utf8) BACKCHAR(eptr); | if (utf8) BACKCHAR(eptr); |
| 2690 | } | } |
| 2691 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2692 | } | } |
| 2693 | ||
| 2694 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2106 for (;;) | Line 2704 for (;;) |
| 2704 | length = 1; | length = 1; |
| 2705 | ecode++; | ecode++; |
| 2706 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 2707 | if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | if (length > md->end_subject - eptr) |
| 2708 | while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 2709 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | |
| 2710 | MRRETURN(MATCH_NOMATCH); | |
| 2711 | } | |
| 2712 | while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2713 | } | } |
| 2714 | else | else |
| 2715 | #endif | #endif |
| 2716 | ||
| 2717 | /* Non-UTF-8 mode */ | /* Non-UTF-8 mode */ |
| 2718 | { | { |
| 2719 | if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); | if (md->end_subject - eptr < 1) |
| 2720 | if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 2721 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | |
| 2722 | MRRETURN(MATCH_NOMATCH); | |
| 2723 | } | |
| 2724 | if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2725 | ecode += 2; | ecode += 2; |
| 2726 | } | } |
| 2727 | break; | break; |
| # | Line 2130 for (;;) | Line 2736 for (;;) |
| 2736 | ecode++; | ecode++; |
| 2737 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 2738 | ||
| 2739 | if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | if (length > md->end_subject - eptr) |
| 2740 | { | |
| 2741 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | |
| 2742 | MRRETURN(MATCH_NOMATCH); | |
| 2743 | } | |
| 2744 | ||
| 2745 | /* 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 |
| 2746 | can use the fast lookup table. */ | can use the fast lookup table. */ |
| 2747 | ||
| 2748 | if (fc < 128) | if (fc < 128) |
| 2749 | { | { |
| 2750 | if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); |
| 2751 | } | } |
| 2752 | ||
| 2753 | /* Otherwise we must pick up the subject character */ | /* Otherwise we must pick up the subject character */ |
| # | Line 2154 for (;;) | Line 2764 for (;;) |
| 2764 | if (fc != dc) | if (fc != dc) |
| 2765 | { | { |
| 2766 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2767 | if (dc != _pcre_ucp_othercase(fc)) | if (dc != UCD_OTHERCASE(fc)) |
| 2768 | #endif | #endif |
| 2769 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2770 | } | } |
| 2771 | } | } |
| 2772 | } | } |
| # | Line 2165 for (;;) | Line 2775 for (;;) |
| 2775 | ||
| 2776 | /* Non-UTF-8 mode */ | /* Non-UTF-8 mode */ |
| 2777 | { | { |
| 2778 | if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); | if (md->end_subject - eptr < 1) |
| 2779 | if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 2780 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | |
| 2781 | MRRETURN(MATCH_NOMATCH); | |
| 2782 | } | |
| 2783 | if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2784 | ecode += 2; | ecode += 2; |
| 2785 | } | } |
| 2786 | break; | break; |
| # | Line 2219 for (;;) | Line 2833 for (;;) |
| 2833 | case OP_MINQUERY: | case OP_MINQUERY: |
| 2834 | c = *ecode++ - OP_STAR; | c = *ecode++ - OP_STAR; |
| 2835 | minimize = (c & 1) != 0; | minimize = (c & 1) != 0; |
| 2836 | ||
| 2837 | min = rep_min[c]; /* Pick up values from tables; */ | min = rep_min[c]; /* Pick up values from tables; */ |
| 2838 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 2839 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2840 | ||
| 2841 | /* Common code for all repeated single-character matches. We can give | /* Common code for all repeated single-character matches. */ |
| up quickly if there are fewer than the minimum number of characters left in | ||
| the subject. */ | ||
| 2842 | ||
| 2843 | REPEATCHAR: | REPEATCHAR: |
| 2844 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 2234 for (;;) | Line 2847 for (;;) |
| 2847 | length = 1; | length = 1; |
| 2848 | charptr = ecode; | charptr = ecode; |
| 2849 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 2850 | ecode += length; | ecode += length; |
| 2851 | ||
| 2852 | /* Handle multibyte character matching specially here. There is | /* Handle multibyte character matching specially here. There is |
| # | Line 2245 for (;;) | Line 2857 for (;;) |
| 2857 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2858 | unsigned int othercase; | unsigned int othercase; |
| 2859 | if ((ims & PCRE_CASELESS) != 0 && | if ((ims & PCRE_CASELESS) != 0 && |
| 2860 | (othercase = _pcre_ucp_othercase(fc)) != NOTACHAR) | (othercase = UCD_OTHERCASE(fc)) != fc) |
| 2861 | oclength = _pcre_ord2utf8(othercase, occhars); | oclength = _pcre_ord2utf8(othercase, occhars); |
| 2862 | else oclength = 0; | else oclength = 0; |
| 2863 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 2864 | ||
| 2865 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2866 | { | { |
| 2867 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | if (eptr <= md->end_subject - length && |
| 2868 | memcmp(eptr, charptr, length) == 0) eptr += length; | |
| 2869 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2870 | /* Need braces because of following else */ | else if (oclength > 0 && |
| 2871 | else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } | eptr <= md->end_subject - oclength && |
| 2872 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2873 | #endif /* SUPPORT_UCP */ | |
| 2874 | else | else |
| 2875 | { | { |
| 2876 | if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2877 | eptr += oclength; | MRRETURN(MATCH_NOMATCH); |
| 2878 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else { RRETURN(MATCH_NOMATCH); } | ||
| #endif /* SUPPORT_UCP */ | ||
| 2879 | } | } |
| 2880 | ||
| 2881 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2274 for (;;) | Line 2886 for (;;) |
| 2886 | { | { |
| 2887 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22); |
| 2888 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2889 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2890 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | if (eptr <= md->end_subject - length && |
| 2891 | memcmp(eptr, charptr, length) == 0) eptr += length; | |
| 2892 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2893 | /* Need braces because of following else */ | else if (oclength > 0 && |
| 2894 | else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } | eptr <= md->end_subject - oclength && |
| 2895 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2896 | #endif /* SUPPORT_UCP */ | |
| 2897 | else | else |
| 2898 | { | { |
| 2899 | if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2900 | eptr += oclength; | MRRETURN(MATCH_NOMATCH); |
| 2901 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else { RRETURN (MATCH_NOMATCH); } | ||
| #endif /* SUPPORT_UCP */ | ||
| 2902 | } | } |
| 2903 | /* Control never gets here */ | /* Control never gets here */ |
| 2904 | } | } |
| # | Line 2296 for (;;) | Line 2908 for (;;) |
| 2908 | pp = eptr; | pp = eptr; |
| 2909 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2910 | { | { |
| 2911 | if (eptr > md->end_subject - length) break; | if (eptr <= md->end_subject - length && |
| 2912 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, length) == 0) eptr += length; |
| 2913 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2914 | else if (oclength == 0) break; | else if (oclength > 0 && |
| 2915 | eptr <= md->end_subject - oclength && | |
| 2916 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2917 | #endif /* SUPPORT_UCP */ | |
| 2918 | else | else |
| 2919 | { | { |
| 2920 | if (memcmp(eptr, occhars, oclength) != 0) break; | CHECK_PARTIAL(); |
| 2921 | eptr += oclength; | break; |
| 2922 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else break; | ||
| #endif /* SUPPORT_UCP */ | ||
| 2923 | } | } |
| 2924 | ||
| 2925 | if (possessive) continue; | if (possessive) continue; |
| 2926 | ||
| 2927 | for(;;) | for(;;) |
| 2928 | { | { |
| 2929 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23); |
| 2930 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2931 | if (eptr == pp) RRETURN(MATCH_NOMATCH); | if (eptr == pp) { MRRETURN(MATCH_NOMATCH); } |
| 2932 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2933 | eptr--; | eptr--; |
| 2934 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 2935 | #else /* without SUPPORT_UCP */ | #else /* without SUPPORT_UCP */ |
| 2936 | eptr -= length; | eptr -= length; |
| 2937 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 2938 | } | } |
| 2939 | } | } |
| 2940 | /* Control never gets here */ | /* Control never gets here */ |
| 2941 | } | } |
| # | Line 2335 for (;;) | Line 2948 for (;;) |
| 2948 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF8 */ |
| 2949 | ||
| 2950 | /* When not in UTF-8 mode, load a single-byte character. */ | /* When not in UTF-8 mode, load a single-byte character. */ |
| 2951 | { | |
| 2952 | if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | fc = *ecode++; |
| fc = *ecode++; | ||
| } | ||
| 2953 | ||
| 2954 | /* 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 less than 256, though we may or |
| 2955 | may not be in UTF-8 mode. The code is duplicated for the caseless and | may not be in UTF-8 mode. The code is duplicated for the caseless and |
| # | Line 2356 for (;;) | Line 2967 for (;;) |
| 2967 | { | { |
| 2968 | fc = md->lcc[fc]; | fc = md->lcc[fc]; |
| 2969 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2970 | if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 2971 | if (eptr >= md->end_subject) | |
| 2972 | { | |
| 2973 | SCHECK_PARTIAL(); | |
| 2974 | MRRETURN(MATCH_NOMATCH); | |
| 2975 | } | |
| 2976 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2977 | } | |
| 2978 | if (min == max) continue; | if (min == max) continue; |
| 2979 | if (minimize) | if (minimize) |
| 2980 | { | { |
| # | Line 2364 for (;;) | Line 2982 for (;;) |
| 2982 | { | { |
| 2983 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24); |
| 2984 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2985 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2986 | fc != md->lcc[*eptr++]) | if (eptr >= md->end_subject) |
| 2987 | RRETURN(MATCH_NOMATCH); | { |
| 2988 | SCHECK_PARTIAL(); | |
| 2989 | MRRETURN(MATCH_NOMATCH); | |
| 2990 | } | |
| 2991 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2992 | } | } |
| 2993 | /* Control never gets here */ | /* Control never gets here */ |
| 2994 | } | } |
| # | Line 2375 for (;;) | Line 2997 for (;;) |
| 2997 | pp = eptr; | pp = eptr; |
| 2998 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2999 | { | { |
| 3000 | if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break; | if (eptr >= md->end_subject) |
| 3001 | { | |
| 3002 | SCHECK_PARTIAL(); | |
| 3003 | break; | |
| 3004 | } | |
| 3005 | if (fc != md->lcc[*eptr]) break; | |
| 3006 | eptr++; | eptr++; |
| 3007 | } | } |
| 3008 | ||
| 3009 | if (possessive) continue; | if (possessive) continue; |
| 3010 | ||
| 3011 | while (eptr >= pp) | while (eptr >= pp) |
| 3012 | { | { |
| 3013 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25); |
| 3014 | eptr--; | eptr--; |
| 3015 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3016 | } | } |
| 3017 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3018 | } | } |
| 3019 | /* Control never gets here */ | /* Control never gets here */ |
| 3020 | } | } |
| # | Line 2394 for (;;) | Line 3023 for (;;) |
| 3023 | ||
| 3024 | else | else |
| 3025 | { | { |
| 3026 | for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH); | for (i = 1; i <= min; i++) |
| 3027 | { | |
| 3028 | if (eptr >= md->end_subject) | |
| 3029 | { | |
| 3030 | SCHECK_PARTIAL(); | |
| 3031 | MRRETURN(MATCH_NOMATCH); | |
| 3032 | } | |
| 3033 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3034 | } | |
| 3035 | ||
| 3036 | if (min == max) continue; | if (min == max) continue; |
| 3037 | ||
| 3038 | if (minimize) | if (minimize) |
| 3039 | { | { |
| 3040 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3041 | { | { |
| 3042 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26); |
| 3043 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3044 | if (fi >= max || eptr >= md->end_subject || fc != *eptr++) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3045 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3046 | { | |
| 3047 | SCHECK_PARTIAL(); | |
| 3048 | MRRETURN(MATCH_NOMATCH); | |
| 3049 | } | |
| 3050 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3051 | } | } |
| 3052 | /* Control never gets here */ | /* Control never gets here */ |
| 3053 | } | } |
| # | Line 2412 for (;;) | Line 3056 for (;;) |
| 3056 | pp = eptr; | pp = eptr; |
| 3057 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3058 | { | { |
| 3059 | if (eptr >= md->end_subject || fc != *eptr) break; | if (eptr >= md->end_subject) |
| 3060 | { | |
| 3061 | SCHECK_PARTIAL(); | |
| 3062 | break; | |
| 3063 | } | |
| 3064 | if (fc != *eptr) break; | |
| 3065 | eptr++; | eptr++; |
| 3066 | } | } |
| 3067 | if (possessive) continue; | if (possessive) continue; |
| 3068 | ||
| 3069 | while (eptr >= pp) | while (eptr >= pp) |
| 3070 | { | { |
| 3071 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27); |
| 3072 | eptr--; | eptr--; |
| 3073 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3074 | } | } |
| 3075 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3076 | } | } |
| 3077 | } | } |
| 3078 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2431 for (;;) | Line 3081 for (;;) |
| 3081 | checking can be multibyte. */ | checking can be multibyte. */ |
| 3082 | ||
| 3083 | case OP_NOT: | case OP_NOT: |
| 3084 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3085 | { | |
| 3086 | SCHECK_PARTIAL(); | |
| 3087 | MRRETURN(MATCH_NOMATCH); | |
| 3088 | } | |
| 3089 | ecode++; | ecode++; |
| 3090 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3091 | if ((ims & PCRE_CASELESS) != 0) | if ((ims & PCRE_CASELESS) != 0) |
| # | Line 2440 for (;;) | Line 3094 for (;;) |
| 3094 | if (c < 256) | if (c < 256) |
| 3095 | #endif | #endif |
| 3096 | c = md->lcc[c]; | c = md->lcc[c]; |
| 3097 | if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH); | if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH); |
| 3098 | } | } |
| 3099 | else | else |
| 3100 | { | { |
| 3101 | if (*ecode++ == c) RRETURN(MATCH_NOMATCH); | if (*ecode++ == c) MRRETURN(MATCH_NOMATCH); |
| 3102 | } | } |
| 3103 | break; | break; |
| 3104 | ||
| # | Line 2508 for (;;) | Line 3162 for (;;) |
| 3162 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 3163 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 3164 | ||
| 3165 | /* Common code for all repeated single-byte matches. We can give up quickly | /* Common code for all repeated single-byte matches. */ |
| if there are fewer than the minimum number of bytes left in the | ||
| subject. */ | ||
| 3166 | ||
| 3167 | REPEATNOTCHAR: | REPEATNOTCHAR: |
| if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 3168 | fc = *ecode++; | fc = *ecode++; |
| 3169 | ||
| 3170 | /* The code is duplicated for the caseless and caseful cases, for speed, | /* The code is duplicated for the caseless and caseful cases, for speed, |
| # | Line 2538 for (;;) | Line 3189 for (;;) |
| 3189 | register unsigned int d; | register unsigned int d; |
| 3190 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3191 | { | { |
| 3192 | if (eptr >= md->end_subject) | |
| 3193 | { | |
| 3194 | SCHECK_PARTIAL(); | |
| 3195 | MRRETURN(MATCH_NOMATCH); | |
| 3196 | } | |
| 3197 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3198 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3199 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3200 | } | } |
| 3201 | } | } |
| 3202 | else | else |
| # | Line 2549 for (;;) | Line 3205 for (;;) |
| 3205 | /* Not UTF-8 mode */ | /* Not UTF-8 mode */ |
| 3206 | { | { |
| 3207 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3208 | if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 3209 | if (eptr >= md->end_subject) | |
| 3210 | { | |
| 3211 | SCHECK_PARTIAL(); | |
| 3212 | MRRETURN(MATCH_NOMATCH); | |
| 3213 | } | |
| 3214 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3215 | } | |
| 3216 | } | } |
| 3217 | ||
| 3218 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2565 for (;;) | Line 3228 for (;;) |
| 3228 | { | { |
| 3229 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28); |
| 3230 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3231 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 3232 | if (eptr >= md->end_subject) | |
| 3233 | { | |
| 3234 | SCHECK_PARTIAL(); | |
| 3235 | MRRETURN(MATCH_NOMATCH); | |
| 3236 | } | |
| 3237 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3238 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3239 | if (fi >= max || eptr >= md->end_subject || fc == d) | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| RRETURN(MATCH_NOMATCH); | ||
| 3240 | } | } |
| 3241 | } | } |
| 3242 | else | else |
| # | Line 2579 for (;;) | Line 3247 for (;;) |
| 3247 | { | { |
| 3248 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29); |
| 3249 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3250 | if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++]) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3251 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3252 | { | |
| 3253 | SCHECK_PARTIAL(); | |
| 3254 | MRRETURN(MATCH_NOMATCH); | |
| 3255 | } | |
| 3256 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3257 | } | } |
| 3258 | } | } |
| 3259 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2600 for (;;) | Line 3273 for (;;) |
| 3273 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3274 | { | { |
| 3275 | int len = 1; | int len = 1; |
| 3276 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 3277 | { | |
| 3278 | SCHECK_PARTIAL(); | |
| 3279 | break; | |
| 3280 | } | |
| 3281 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3282 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3283 | if (fc == d) break; | if (fc == d) break; |
| # | Line 2621 for (;;) | Line 3298 for (;;) |
| 3298 | { | { |
| 3299 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3300 | { | { |
| 3301 | if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break; | if (eptr >= md->end_subject) |
| 3302 | { | |
| 3303 | SCHECK_PARTIAL(); | |
| 3304 | break; | |
| 3305 | } | |
| 3306 | if (fc == md->lcc[*eptr]) break; | |
| 3307 | eptr++; | eptr++; |
| 3308 | } | } |
| 3309 | if (possessive) continue; | if (possessive) continue; |
| # | Line 2633 for (;;) | Line 3315 for (;;) |
| 3315 | } | } |
| 3316 | } | } |
| 3317 | ||
| 3318 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3319 | } | } |
| 3320 | /* Control never gets here */ | /* Control never gets here */ |
| 3321 | } | } |
| # | Line 2649 for (;;) | Line 3331 for (;;) |
| 3331 | register unsigned int d; | register unsigned int d; |
| 3332 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3333 | { | { |
| 3334 | if (eptr >= md->end_subject) | |
| 3335 | { | |
| 3336 | SCHECK_PARTIAL(); | |
| 3337 | MRRETURN(MATCH_NOMATCH); | |
| 3338 | } | |
| 3339 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3340 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3341 | } | } |
| 3342 | } | } |
| 3343 | else | else |
| # | Line 2658 for (;;) | Line 3345 for (;;) |
| 3345 | /* Not UTF-8 mode */ | /* Not UTF-8 mode */ |
| 3346 | { | { |
| 3347 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3348 | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 3349 | if (eptr >= md->end_subject) | |
| 3350 | { | |
| 3351 | SCHECK_PARTIAL(); | |
| 3352 | MRRETURN(MATCH_NOMATCH); | |
| 3353 | } | |
| 3354 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3355 | } | |
| 3356 | } | } |
| 3357 | ||
| 3358 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2674 for (;;) | Line 3368 for (;;) |
| 3368 | { | { |
| 3369 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32); |
| 3370 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3371 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 3372 | if (eptr >= md->end_subject) | |
| 3373 | { | |
| 3374 | SCHECK_PARTIAL(); | |
| 3375 | MRRETURN(MATCH_NOMATCH); | |
| 3376 | } | |
| 3377 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3378 | if (fi >= max || eptr >= md->end_subject || fc == d) | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| RRETURN(MATCH_NOMATCH); | ||
| 3379 | } | } |
| 3380 | } | } |
| 3381 | else | else |
| # | Line 2687 for (;;) | Line 3386 for (;;) |
| 3386 | { | { |
| 3387 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33); |
| 3388 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3389 | if (fi >= max || eptr >= md->end_subject || fc == *eptr++) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3390 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3391 | { | |
| 3392 | SCHECK_PARTIAL(); | |
| 3393 | MRRETURN(MATCH_NOMATCH); | |
| 3394 | } | |
| 3395 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3396 | } | } |
| 3397 | } | } |
| 3398 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2708 for (;;) | Line 3412 for (;;) |
| 3412 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3413 | { | { |
| 3414 | int len = 1; | int len = 1; |
| 3415 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 3416 | { | |
| 3417 | SCHECK_PARTIAL(); | |
| 3418 | break; | |
| 3419 | } | |
| 3420 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3421 | if (fc == d) break; | if (fc == d) break; |
| 3422 | eptr += len; | eptr += len; |
| # | Line 2728 for (;;) | Line 3436 for (;;) |
| 3436 | { | { |
| 3437 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3438 | { | { |
| 3439 | if (eptr >= md->end_subject || fc == *eptr) break; | if (eptr >= md->end_subject) |
| 3440 | { | |
| 3441 | SCHECK_PARTIAL(); | |
| 3442 | break; | |
| 3443 | } | |
| 3444 | if (fc == *eptr) break; | |
| 3445 | eptr++; | eptr++; |
| 3446 | } | } |
| 3447 | if (possessive) continue; | if (possessive) continue; |
| # | Line 2740 for (;;) | Line 3453 for (;;) |
| 3453 | } | } |
| 3454 | } | } |
| 3455 | ||
| 3456 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3457 | } | } |
| 3458 | } | } |
| 3459 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2822 for (;;) | Line 3535 for (;;) |
| 3535 | ||
| 3536 | /* First, ensure the minimum number of matches are present. Use inline | /* First, ensure the minimum number of matches are present. Use inline |
| 3537 | code for maximizing the speed, and do the type test once at the start | code for maximizing the speed, and do the type test once at the start |
| 3538 | (i.e. keep it out of the loop). Also we can test that there are at least | (i.e. keep it out of the loop). Separate the UTF-8 code completely as that |
| the minimum number of bytes before we start. This isn't as effective in | ||
| UTF-8 mode, but it does no harm. Separate the UTF-8 code completely as that | ||
| 3539 | is tidier. Also separate the UCP code, which can be the same for both UTF-8 | is tidier. Also separate the UCP code, which can be the same for both UTF-8 |
| 3540 | and single-bytes. */ | and single-bytes. */ |
| 3541 | ||
| if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 3542 | if (min > 0) | if (min > 0) |
| 3543 | { | { |
| 3544 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| # | Line 2837 for (;;) | Line 3547 for (;;) |
| 3547 | switch(prop_type) | switch(prop_type) |
| 3548 | { | { |
| 3549 | case PT_ANY: | case PT_ANY: |
| 3550 | if (prop_fail_result) RRETURN(MATCH_NOMATCH); | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); |
| 3551 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3552 | { | { |
| 3553 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3554 | { | |
| 3555 | SCHECK_PARTIAL(); | |
| 3556 | MRRETURN(MATCH_NOMATCH); | |
| 3557 | } | |
| 3558 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3559 | } | } |
| 3560 | break; | break; |
| # | Line 2848 for (;;) | Line 3562 for (;;) |
| 3562 | case PT_LAMP: | case PT_LAMP: |
| 3563 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3564 | { | { |
| 3565 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3566 | { | |
| 3567 | SCHECK_PARTIAL(); | |
| 3568 | MRRETURN(MATCH_NOMATCH); | |
| 3569 | } | |
| 3570 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3571 | prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | prop_chartype = UCD_CHARTYPE(c); |
| 3572 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 3573 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| 3574 | prop_chartype == ucp_Lt) == prop_fail_result) | prop_chartype == ucp_Lt) == prop_fail_result) |
| 3575 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3576 | } | } |
| 3577 | break; | break; |
| 3578 | ||
| 3579 | case PT_GC: | case PT_GC: |
| 3580 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3581 | { | { |
| 3582 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3583 | { | |
| 3584 | SCHECK_PARTIAL(); | |
| 3585 | MRRETURN(MATCH_NOMATCH); | |
| 3586 | } | |
| 3587 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3588 | prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | prop_category = UCD_CATEGORY(c); |
| 3589 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| 3590 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3591 | } | } |
| 3592 | break; | break; |
| 3593 | ||
| 3594 | case PT_PC: | case PT_PC: |
| 3595 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3596 | { | { |
| 3597 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3598 | { | |
| 3599 | SCHECK_PARTIAL(); | |
| 3600 | MRRETURN(MATCH_NOMATCH); | |
| 3601 | } | |
| 3602 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3603 | prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | prop_chartype = UCD_CHARTYPE(c); |
| 3604 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| 3605 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3606 | } | } |
| 3607 | break; | break; |
| 3608 | ||
| 3609 | case PT_SC: | case PT_SC: |
| 3610 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3611 | { | { |
| 3612 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3613 | { | |
| 3614 | SCHECK_PARTIAL(); | |
| 3615 | MRRETURN(MATCH_NOMATCH); | |
| 3616 | } | |
| 3617 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3618 | prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | prop_script = UCD_SCRIPT(c); |
| 3619 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| 3620 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3621 | } | |
| 3622 | break; | |
| 3623 | ||
| 3624 | case PT_ALNUM: | |
| 3625 | for (i = 1; i <= min; i++) | |
| 3626 | { | |
| 3627 | if (eptr >= md->end_subject) | |
| 3628 | { | |
| 3629 | SCHECK_PARTIAL(); | |
| 3630 | MRRETURN(MATCH_NOMATCH); | |
| 3631 | } | |
| 3632 | GETCHARINCTEST(c, eptr); | |
| 3633 | prop_category = UCD_CATEGORY(c); | |
| 3634 | if ((prop_category == ucp_L || prop_category == ucp_N) | |
| 3635 | == prop_fail_result) | |
| 3636 | MRRETURN(MATCH_NOMATCH); | |
| 3637 | } | |
| 3638 | break; | |
| 3639 | ||
| 3640 | case PT_SPACE: /* Perl space */ | |
| 3641 | for (i = 1; i <= min; i++) | |
| 3642 | { | |
| 3643 | if (eptr >= md->end_subject) | |
| 3644 | { | |
| 3645 | SCHECK_PARTIAL(); | |
| 3646 | MRRETURN(MATCH_NOMATCH); | |
| 3647 | } | |
| 3648 | GETCHARINCTEST(c, eptr); | |
| 3649 | prop_category = UCD_CATEGORY(c); | |
| 3650 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 3651 | c == CHAR_FF || c == CHAR_CR) | |
| 3652 | == prop_fail_result) | |
| 3653 | MRRETURN(MATCH_NOMATCH); | |
| 3654 | } | |
| 3655 | break; | |
| 3656 | ||
| 3657 | case PT_PXSPACE: /* POSIX space */ | |
| 3658 | for (i = 1; i <= min; i++) | |
| 3659 | { | |
| 3660 | if (eptr >= md->end_subject) | |
| 3661 | { | |
| 3662 | SCHECK_PARTIAL(); | |
| 3663 | MRRETURN(MATCH_NOMATCH); | |
| 3664 | } | |
| 3665 | GETCHARINCTEST(c, eptr); | |
| 3666 | prop_category = UCD_CATEGORY(c); | |
| 3667 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 3668 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | |
| 3669 | == prop_fail_result) | |
| 3670 | MRRETURN(MATCH_NOMATCH); | |
| 3671 | } | |
| 3672 | break; | |
| 3673 | ||
| 3674 | case PT_WORD: | |
| 3675 | for (i = 1; i <= min; i++) | |
| 3676 | { | |
| 3677 | if (eptr >= md->end_subject) | |
| 3678 | { | |
| 3679 | SCHECK_PARTIAL(); | |
| 3680 | MRRETURN(MATCH_NOMATCH); | |
| 3681 | } | |
| 3682 | GETCHARINCTEST(c, eptr); | |
| 3683 | prop_category = UCD_CATEGORY(c); | |
| 3684 | if ((prop_category == ucp_L || prop_category == ucp_N || | |
| 3685 | c == CHAR_UNDERSCORE) | |
| 3686 | == prop_fail_result) | |
| 3687 | MRRETURN(MATCH_NOMATCH); | |
| 3688 | } | } |
| 3689 | break; | break; |
| 3690 | ||
| 3691 | /* This should not occur */ | |
| 3692 | ||
| 3693 | default: | default: |
| 3694 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 3695 | } | } |
| # | Line 2903 for (;;) | Line 3702 for (;;) |
| 3702 | { | { |
| 3703 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3704 | { | { |
| 3705 | if (eptr >= md->end_subject) | |
| 3706 | { | |
| 3707 | SCHECK_PARTIAL(); | |
| 3708 | MRRETURN(MATCH_NOMATCH); | |
| 3709 | } | |
| 3710 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3711 | prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | prop_category = UCD_CATEGORY(c); |
| 3712 | if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); | if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 3713 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 3714 | { | { |
| 3715 | int len = 1; | int len = 1; |
| 3716 | if (!utf8) c = *eptr; else | if (!utf8) c = *eptr; |
| 3717 | { | else { GETCHARLEN(c, eptr, len); } |
| 3718 | GETCHARLEN(c, eptr, len); | prop_category = UCD_CATEGORY(c); |
| } | ||
| prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | ||
| 3719 | if (prop_category != ucp_M) break; | if (prop_category != ucp_M) break; |
| 3720 | eptr += len; | eptr += len; |
| 3721 | } | } |
| # | Line 2931 for (;;) | Line 3733 for (;;) |
| 3733 | case OP_ANY: | case OP_ANY: |
| 3734 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3735 | { | { |
| 3736 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3737 | ((ims & PCRE_DOTALL) == 0 && IS_NEWLINE(eptr))) | { |
| 3738 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3739 | MRRETURN(MATCH_NOMATCH); | |
| 3740 | } | |
| 3741 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | |
| 3742 | eptr++; | |
| 3743 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | |
| 3744 | } | |
| 3745 | break; | |
| 3746 | ||
| 3747 | case OP_ALLANY: | |
| 3748 | for (i = 1; i <= min; i++) | |
| 3749 | { | |
| 3750 | if (eptr >= md->end_subject) | |
| 3751 | { | |
| 3752 | SCHECK_PARTIAL(); | |
| 3753 | MRRETURN(MATCH_NOMATCH); | |
| 3754 | } | |
| 3755 | eptr++; | eptr++; |
| 3756 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 3757 | } | } |
| 3758 | break; | break; |
| 3759 | ||
| 3760 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 3761 | if (eptr > md->end_subject - min) MRRETURN(MATCH_NOMATCH); | |
| 3762 | eptr += min; | eptr += min; |
| 3763 | break; | break; |
| 3764 | ||
| 3765 | case OP_ANYNL: | case OP_ANYNL: |
| 3766 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3767 | { | { |
| 3768 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3769 | { | |
| 3770 | SCHECK_PARTIAL(); | |
| 3771 | MRRETURN(MATCH_NOMATCH); | |
| 3772 | } | |
| 3773 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3774 | switch(c) | switch(c) |
| 3775 | { | { |
| 3776 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3777 | case 0x000d: | case 0x000d: |
| 3778 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 3779 | break; | break; |
| 3780 | ||
| 3781 | case 0x000a: | case 0x000a: |
| 3782 | break; | |
| 3783 | ||
| 3784 | case 0x000b: | case 0x000b: |
| 3785 | case 0x000c: | case 0x000c: |
| 3786 | case 0x0085: | case 0x0085: |
| 3787 | case 0x2028: | case 0x2028: |
| 3788 | case 0x2029: | case 0x2029: |
| 3789 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | |
| 3790 | break; | break; |
| 3791 | } | } |
| 3792 | } | } |
| # | Line 2968 for (;;) | Line 3795 for (;;) |
| 3795 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 3796 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3797 | { | { |
| 3798 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3799 | { | |
| 3800 | SCHECK_PARTIAL(); | |
| 3801 | MRRETURN(MATCH_NOMATCH); | |
| 3802 | } | |
| 3803 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3804 | switch(c) | switch(c) |
| 3805 | { | { |
| # | Line 2992 for (;;) | Line 3823 for (;;) |
| 3823 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 3824 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 3825 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 3826 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3827 | } | } |
| 3828 | } | } |
| 3829 | break; | break; |
| # | Line 3000 for (;;) | Line 3831 for (;;) |
| 3831 | case OP_HSPACE: | case OP_HSPACE: |
| 3832 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3833 | { | { |
| 3834 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3835 | { | |
| 3836 | SCHECK_PARTIAL(); | |
| 3837 | MRRETURN(MATCH_NOMATCH); | |
| 3838 | } | |
| 3839 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3840 | switch(c) | switch(c) |
| 3841 | { | { |
| 3842 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3843 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 3844 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 3845 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3032 for (;;) | Line 3867 for (;;) |
| 3867 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 3868 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3869 | { | { |
| 3870 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3871 | { | |
| 3872 | SCHECK_PARTIAL(); | |
| 3873 | MRRETURN(MATCH_NOMATCH); | |
| 3874 | } | |
| 3875 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3876 | switch(c) | switch(c) |
| 3877 | { | { |
| # | Line 3044 for (;;) | Line 3883 for (;;) |
| 3883 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 3884 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 3885 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 3886 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3887 | } | } |
| 3888 | } | } |
| 3889 | break; | break; |
| # | Line 3052 for (;;) | Line 3891 for (;;) |
| 3891 | case OP_VSPACE: | case OP_VSPACE: |
| 3892 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3893 | { | { |
| 3894 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3895 | { | |
| 3896 | SCHECK_PARTIAL(); | |
| 3897 | MRRETURN(MATCH_NOMATCH); | |
| 3898 | } | |
| 3899 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3900 | switch(c) | switch(c) |
| 3901 | { | { |
| 3902 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3903 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 3904 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 3905 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3072 for (;;) | Line 3915 for (;;) |
| 3915 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 3916 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3917 | { | { |
| 3918 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3919 | { | |
| 3920 | SCHECK_PARTIAL(); | |
| 3921 | MRRETURN(MATCH_NOMATCH); | |
| 3922 | } | |
| 3923 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3924 | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) |
| 3925 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3926 | } | } |
| 3927 | break; | break; |
| 3928 | ||
| 3929 | case OP_DIGIT: | case OP_DIGIT: |
| 3930 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3931 | { | { |
| 3932 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3933 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | { |
| 3934 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3935 | MRRETURN(MATCH_NOMATCH); | |
| 3936 | } | |
| 3937 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | |
| 3938 | MRRETURN(MATCH_NOMATCH); | |
| 3939 | /* 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 */ |
| 3940 | } | } |
| 3941 | break; | break; |
| # | Line 3092 for (;;) | Line 3943 for (;;) |
| 3943 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 3944 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3945 | { | { |
| 3946 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3947 | (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0)) | { |
| 3948 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3949 | MRRETURN(MATCH_NOMATCH); | |
| 3950 | } | |
| 3951 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) | |
| 3952 | MRRETURN(MATCH_NOMATCH); | |
| 3953 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); |
| 3954 | } | } |
| 3955 | break; | break; |
| # | Line 3102 for (;;) | Line 3957 for (;;) |
| 3957 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 3958 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3959 | { | { |
| 3960 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3961 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | { |
| 3962 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3963 | MRRETURN(MATCH_NOMATCH); | |
| 3964 | } | |
| 3965 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | |
| 3966 | MRRETURN(MATCH_NOMATCH); | |
| 3967 | /* 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 */ |
| 3968 | } | } |
| 3969 | break; | break; |
| # | Line 3112 for (;;) | Line 3971 for (;;) |
| 3971 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 3972 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3973 | { | { |
| 3974 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3975 | (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0)) | { |
| 3976 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3977 | MRRETURN(MATCH_NOMATCH); | |
| 3978 | } | |
| 3979 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) | |
| 3980 | MRRETURN(MATCH_NOMATCH); | |
| 3981 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); |
| 3982 | } | } |
| 3983 | break; | break; |
| # | Line 3122 for (;;) | Line 3985 for (;;) |
| 3985 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 3986 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3987 | { | { |
| 3988 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3989 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | { |
| 3990 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3991 | MRRETURN(MATCH_NOMATCH); | |
| 3992 | } | |
| 3993 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | |
| 3994 | MRRETURN(MATCH_NOMATCH); | |
| 3995 | /* 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 */ |
| 3996 | } | } |
| 3997 | break; | break; |
| # | Line 3137 for (;;) | Line 4004 for (;;) |
| 4004 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF8 */ |
| 4005 | ||
| 4006 | /* 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 |
| 4007 | than OP_PROP and OP_NOTPROP. We can assume that there are the minimum | than OP_PROP and OP_NOTPROP. */ |
| number of bytes present, as this was tested above. */ | ||
| 4008 | ||
| 4009 | switch(ctype) | switch(ctype) |
| 4010 | { | { |
| 4011 | case OP_ANY: | case OP_ANY: |
| 4012 | if ((ims & PCRE_DOTALL) == 0) | for (i = 1; i <= min; i++) |
| 4013 | { | { |
| 4014 | for (i = 1; i <= min; i++) | if (eptr >= md->end_subject) |
| 4015 | { | { |
| 4016 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 4017 | eptr++; | MRRETURN(MATCH_NOMATCH); |
| 4018 | } | } |
| 4019 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | |
| 4020 | eptr++; | |
| 4021 | } | } |
| else eptr += min; | ||
| 4022 | break; | break; |
| 4023 | ||
| 4024 | case OP_ANYBYTE: | case OP_ALLANY: |
| 4025 | if (eptr > md->end_subject - min) | |
| 4026 | { | |
| 4027 | SCHECK_PARTIAL(); | |
| 4028 | MRRETURN(MATCH_NOMATCH); | |
| 4029 | } | |
| 4030 | eptr += min; | eptr += min; |
| 4031 | break; | break; |
| 4032 | ||
| 4033 | /* Because of the CRLF case, we can't assume the minimum number of | case OP_ANYBYTE: |
| 4034 | bytes are present in this case. */ | if (eptr > md->end_subject - min) |
| 4035 | { | |
| 4036 | SCHECK_PARTIAL(); | |
| 4037 | MRRETURN(MATCH_NOMATCH); | |
| 4038 | } | |
| 4039 | eptr += min; | |
| 4040 | break; | |
| 4041 | ||
| 4042 | case OP_ANYNL: | case OP_ANYNL: |
| 4043 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4044 | { | { |
| 4045 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4046 | { | |
| 4047 | SCHECK_PARTIAL(); | |
| 4048 | MRRETURN(MATCH_NOMATCH); | |
| 4049 | } | |
| 4050 | switch(*eptr++) | switch(*eptr++) |
| 4051 | { | { |
| 4052 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4053 | case 0x000d: | case 0x000d: |
| 4054 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4055 | break; | break; |
| 4056 | case 0x000a: | case 0x000a: |
| 4057 | break; | |
| 4058 | ||
| 4059 | case 0x000b: | case 0x000b: |
| 4060 | case 0x000c: | case 0x000c: |
| 4061 | case 0x0085: | case 0x0085: |
| 4062 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | |
| 4063 | break; | break; |
| 4064 | } | } |
| 4065 | } | } |
| # | Line 3183 for (;;) | Line 4068 for (;;) |
| 4068 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 4069 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4070 | { | { |
| 4071 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4072 | { | |
| 4073 | SCHECK_PARTIAL(); | |
| 4074 | MRRETURN(MATCH_NOMATCH); | |
| 4075 | } | |
| 4076 | switch(*eptr++) | switch(*eptr++) |
| 4077 | { | { |
| 4078 | default: break; | default: break; |
| 4079 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4080 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4081 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4082 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4083 | } | } |
| 4084 | } | } |
| 4085 | break; | break; |
| # | Line 3198 for (;;) | Line 4087 for (;;) |
| 4087 | case OP_HSPACE: | case OP_HSPACE: |
| 4088 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4089 | { | { |
| 4090 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4091 | { | |
| 4092 | SCHECK_PARTIAL(); | |
| 4093 | MRRETURN(MATCH_NOMATCH); | |
| 4094 | } | |
| 4095 | switch(*eptr++) | switch(*eptr++) |
| 4096 | { | { |
| 4097 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4098 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4099 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4100 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3213 for (;;) | Line 4106 for (;;) |
| 4106 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 4107 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4108 | { | { |
| 4109 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4110 | { | |
| 4111 | SCHECK_PARTIAL(); | |
| 4112 | MRRETURN(MATCH_NOMATCH); | |
| 4113 | } | |
| 4114 | switch(*eptr++) | switch(*eptr++) |
| 4115 | { | { |
| 4116 | default: break; | default: break; |
| # | Line 3222 for (;;) | Line 4119 for (;;) |
| 4119 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4120 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4121 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4122 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4123 | } | } |
| 4124 | } | } |
| 4125 | break; | break; |
| # | Line 3230 for (;;) | Line 4127 for (;;) |
| 4127 | case OP_VSPACE: | case OP_VSPACE: |
| 4128 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4129 | { | { |
| 4130 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4131 | { | |
| 4132 | SCHECK_PARTIAL(); | |
| 4133 | MRRETURN(MATCH_NOMATCH); | |
| 4134 | } | |
| 4135 | switch(*eptr++) | switch(*eptr++) |
| 4136 | { | { |
| 4137 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4138 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4139 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4140 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3246 for (;;) | Line 4147 for (;;) |
| 4147 | ||
| 4148 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4149 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4150 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); | { |
| 4151 | if (eptr >= md->end_subject) | |
| 4152 | { | |
| 4153 | SCHECK_PARTIAL(); | |
| 4154 | MRRETURN(MATCH_NOMATCH); | |
| 4155 | } | |
| 4156 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); | |
| 4157 | } | |
| 4158 | break; | break; |
| 4159 | ||
| 4160 | case OP_DIGIT: | case OP_DIGIT: |
| 4161 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4162 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); | { |
| 4163 | if (eptr >= md->end_subject) | |
| 4164 | { | |
| 4165 | SCHECK_PARTIAL(); | |
| 4166 | MRRETURN(MATCH_NOMATCH); | |
| 4167 | } | |
| 4168 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); | |
| 4169 | } | |
| 4170 | break; | break; |
| 4171 | ||
| 4172 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4173 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4174 | if ((md->ctypes[*eptr++] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); | { |
| 4175 | if (eptr >= md->end_subject) | |
| 4176 | { | |
| 4177 | SCHECK_PARTIAL(); | |
| 4178 | MRRETURN(MATCH_NOMATCH); | |
| 4179 | } | |
| 4180 | if ((md->ctypes[*eptr++] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); | |
| 4181 | } | |
| 4182 | break; | break; |
| 4183 | ||
| 4184 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4185 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4186 | if ((md->ctypes[*eptr++] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); | { |
| 4187 | break; | if (eptr >= md->end_subject) |
| 4188 | { | |
| 4189 | case OP_NOT_WORDCHAR: | SCHECK_PARTIAL(); |
| 4190 | MRRETURN(MATCH_NOMATCH); | |
| 4191 | } | |
| 4192 | if ((md->ctypes[*eptr++] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); | |
| 4193 | } | |
| 4194 | break; | |
| 4195 | ||
| 4196 | case OP_NOT_WORDCHAR: | |
| 4197 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4198 | { | |
| 4199 | if (eptr >= md->end_subject) | |
| 4200 | { | |
| 4201 | SCHECK_PARTIAL(); | |
| 4202 | MRRETURN(MATCH_NOMATCH); | |
| 4203 | } | |
| 4204 | if ((md->ctypes[*eptr++] & ctype_word) != 0) | if ((md->ctypes[*eptr++] & ctype_word) != 0) |
| 4205 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4206 | } | |
| 4207 | break; | break; |
| 4208 | ||
| 4209 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4210 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4211 | { | |
| 4212 | if (eptr >= md->end_subject) | |
| 4213 | { | |
| 4214 | SCHECK_PARTIAL(); | |
| 4215 | MRRETURN(MATCH_NOMATCH); | |
| 4216 | } | |
| 4217 | if ((md->ctypes[*eptr++] & ctype_word) == 0) | if ((md->ctypes[*eptr++] & ctype_word) == 0) |
| 4218 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4219 | } | |
| 4220 | break; | break; |
| 4221 | ||
| 4222 | default: | default: |
| # | Line 3301 for (;;) | Line 4244 for (;;) |
| 4244 | { | { |
| 4245 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36); |
| 4246 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4247 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4248 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4249 | if (prop_fail_result) RRETURN(MATCH_NOMATCH); | { |
| 4250 | SCHECK_PARTIAL(); | |
| 4251 | MRRETURN(MATCH_NOMATCH); | |
| 4252 | } | |
| 4253 | GETCHARINCTEST(c, eptr); | |
| 4254 | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); | |
| 4255 | } | } |
| 4256 | /* Control never gets here */ | /* Control never gets here */ |
| 4257 | ||
| # | Line 3312 for (;;) | Line 4260 for (;;) |
| 4260 | { | { |
| 4261 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37); |
| 4262 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4263 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4264 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4265 | prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | { |
| 4266 | SCHECK_PARTIAL(); | |
| 4267 | MRRETURN(MATCH_NOMATCH); | |
| 4268 | } | |
| 4269 | GETCHARINCTEST(c, eptr); | |
| 4270 | prop_chartype = UCD_CHARTYPE(c); | |
| 4271 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 4272 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| 4273 | prop_chartype == ucp_Lt) == prop_fail_result) | prop_chartype == ucp_Lt) == prop_fail_result) |
| 4274 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4275 | } | } |
| 4276 | /* Control never gets here */ | /* Control never gets here */ |
| 4277 | ||
| # | Line 3327 for (;;) | Line 4280 for (;;) |
| 4280 | { | { |
| 4281 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38); |
| 4282 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4283 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4284 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4285 | prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | { |
| 4286 | SCHECK_PARTIAL(); | |
| 4287 | MRRETURN(MATCH_NOMATCH); | |
| 4288 | } | |
| 4289 | GETCHARINCTEST(c, eptr); | |
| 4290 | prop_category = UCD_CATEGORY(c); | |
| 4291 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| 4292 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4293 | } | } |
| 4294 | /* Control never gets here */ | /* Control never gets here */ |
| 4295 | ||
| # | Line 3340 for (;;) | Line 4298 for (;;) |
| 4298 | { | { |
| 4299 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39); |
| 4300 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4301 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4302 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4303 | prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | { |
| 4304 | SCHECK_PARTIAL(); | |
| 4305 | MRRETURN(MATCH_NOMATCH); | |
| 4306 | } | |
| 4307 | GETCHARINCTEST(c, eptr); | |
| 4308 | prop_chartype = UCD_CHARTYPE(c); | |
| 4309 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| 4310 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4311 | } | } |
| 4312 | /* Control never gets here */ | /* Control never gets here */ |
| 4313 | ||
| # | Line 3353 for (;;) | Line 4316 for (;;) |
| 4316 | { | { |
| 4317 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40); |
| 4318 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4319 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4320 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4321 | prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | { |
| 4322 | SCHECK_PARTIAL(); | |
| 4323 | MRRETURN(MATCH_NOMATCH); | |
| 4324 | } | |
| 4325 | GETCHARINCTEST(c, eptr); | |
| 4326 | prop_script = UCD_SCRIPT(c); | |
| 4327 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| 4328 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4329 | } | } |
| 4330 | /* Control never gets here */ | /* Control never gets here */ |
| 4331 | ||
| 4332 | case PT_ALNUM: | |
| 4333 | for (fi = min;; fi++) | |
| 4334 | { | |
| 4335 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM59); | |
| 4336 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4337 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4338 | if (eptr >= md->end_subject) | |
| 4339 | { | |
| 4340 | SCHECK_PARTIAL(); | |
| 4341 | MRRETURN(MATCH_NOMATCH); | |
| 4342 | } | |
| 4343 | GETCHARINCTEST(c, eptr); | |
| 4344 | prop_category = UCD_CATEGORY(c); | |
| 4345 | if ((prop_category == ucp_L || prop_category == ucp_N) | |
| 4346 | == prop_fail_result) | |
| 4347 | MRRETURN(MATCH_NOMATCH); | |
| 4348 | } | |
| 4349 | /* Control never gets here */ | |
| 4350 | ||
| 4351 | case PT_SPACE: /* Perl space */ | |
| 4352 | for (fi = min;; fi++) | |
| 4353 | { | |
| 4354 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM60); | |
| 4355 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4356 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4357 | if (eptr >= md->end_subject) | |
| 4358 | { | |
| 4359 | SCHECK_PARTIAL(); | |
| 4360 | MRRETURN(MATCH_NOMATCH); | |
| 4361 | } | |
| 4362 | GETCHARINCTEST(c, eptr); | |
| 4363 | prop_category = UCD_CATEGORY(c); | |
| 4364 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 4365 | c == CHAR_FF || c == CHAR_CR) | |
| 4366 | == prop_fail_result) | |
| 4367 | MRRETURN(MATCH_NOMATCH); | |
| 4368 | } | |
| 4369 | /* Control never gets here */ | |
| 4370 | ||
| 4371 | case PT_PXSPACE: /* POSIX space */ | |
| 4372 | for (fi = min;; fi++) | |
| 4373 | { | |
| 4374 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM61); | |
| 4375 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4376 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4377 | if (eptr >= md->end_subject) | |
| 4378 | { | |
| 4379 | SCHECK_PARTIAL(); | |
| 4380 | MRRETURN(MATCH_NOMATCH); | |
| 4381 | } | |
| 4382 | GETCHARINCTEST(c, eptr); | |
| 4383 | prop_category = UCD_CATEGORY(c); | |
| 4384 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 4385 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | |
| 4386 | == prop_fail_result) | |
| 4387 | MRRETURN(MATCH_NOMATCH); | |
| 4388 | } | |
| 4389 | /* Control never gets here */ | |
| 4390 | ||
| 4391 | case PT_WORD: | |
| 4392 | for (fi = min;; fi++) | |
| 4393 | { | |
| 4394 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM62); | |
| 4395 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4396 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4397 | if (eptr >= md->end_subject) | |
| 4398 | { | |
| 4399 | SCHECK_PARTIAL(); | |
| 4400 | MRRETURN(MATCH_NOMATCH); | |
| 4401 | } | |
| 4402 | GETCHARINCTEST(c, eptr); | |
| 4403 | prop_category = UCD_CATEGORY(c); | |
| 4404 | if ((prop_category == ucp_L || | |
| 4405 | prop_category == ucp_N || | |
| 4406 | c == CHAR_UNDERSCORE) | |
| 4407 | == prop_fail_result) | |
| 4408 | MRRETURN(MATCH_NOMATCH); | |
| 4409 | } | |
| 4410 | /* Control never gets here */ | |
| 4411 | ||
| 4412 | /* This should never occur */ | |
| 4413 | ||
| 4414 | default: | default: |
| 4415 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 4416 | } | } |
| # | Line 3375 for (;;) | Line 4425 for (;;) |
| 4425 | { | { |
| 4426 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41); |
| 4427 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4428 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4429 | if (eptr >= md->end_subject) | |
| 4430 | { | |
| 4431 | SCHECK_PARTIAL(); | |
| 4432 | MRRETURN(MATCH_NOMATCH); | |
| 4433 | } | |
| 4434 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4435 | prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | prop_category = UCD_CATEGORY(c); |
| 4436 | if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); | if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 4437 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 4438 | { | { |
| 4439 | int len = 1; | int len = 1; |
| 4440 | if (!utf8) c = *eptr; else | if (!utf8) c = *eptr; |
| 4441 | { | else { GETCHARLEN(c, eptr, len); } |
| 4442 | GETCHARLEN(c, eptr, len); | prop_category = UCD_CATEGORY(c); |
| } | ||
| prop_category = _pcre_ucp_findprop(c, &prop_chartype, &prop_script); | ||
| 4443 | if (prop_category != ucp_M) break; | if (prop_category != ucp_M) break; |
| 4444 | eptr += len; | eptr += len; |
| 4445 | } | } |
| # | Line 3404 for (;;) | Line 4457 for (;;) |
| 4457 | { | { |
| 4458 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42); |
| 4459 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4460 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4461 | (ctype == OP_ANY && (ims & PCRE_DOTALL) == 0 && | if (eptr >= md->end_subject) |
| 4462 | IS_NEWLINE(eptr))) | { |
| 4463 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 4464 | MRRETURN(MATCH_NOMATCH); | |
| 4465 | } | |
| 4466 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | |
| 4467 | MRRETURN(MATCH_NOMATCH); | |
| 4468 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4469 | switch(ctype) | switch(ctype) |
| 4470 | { | { |
| 4471 | case OP_ANY: /* This is the DOTALL case */ | case OP_ANY: /* This is the non-NL case */ |
| 4472 | break; | case OP_ALLANY: |
| 4473 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 4474 | break; | break; |
| 4475 | ||
| 4476 | case OP_ANYNL: | case OP_ANYNL: |
| 4477 | switch(c) | switch(c) |
| 4478 | { | { |
| 4479 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4480 | case 0x000d: | case 0x000d: |
| 4481 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4482 | break; | break; |
| 4483 | case 0x000a: | case 0x000a: |
| 4484 | break; | |
| 4485 | ||
| 4486 | case 0x000b: | case 0x000b: |
| 4487 | case 0x000c: | case 0x000c: |
| 4488 | case 0x0085: | case 0x0085: |
| 4489 | case 0x2028: | case 0x2028: |
| 4490 | case 0x2029: | case 0x2029: |
| 4491 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | |
| 4492 | break; | break; |
| 4493 | } | } |
| 4494 | break; | break; |
| # | Line 3458 for (;;) | Line 4516 for (;;) |
| 4516 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 4517 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 4518 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 4519 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4520 | } | } |
| 4521 | break; | break; |
| 4522 | ||
| 4523 | case OP_HSPACE: | case OP_HSPACE: |
| 4524 | switch(c) | switch(c) |
| 4525 | { | { |
| 4526 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4527 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4528 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4529 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3500 for (;;) | Line 4558 for (;;) |
| 4558 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4559 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 4560 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 4561 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4562 | } | } |
| 4563 | break; | break; |
| 4564 | ||
| 4565 | case OP_VSPACE: | case OP_VSPACE: |
| 4566 | switch(c) | switch(c) |
| 4567 | { | { |
| 4568 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4569 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4570 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4571 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3521 for (;;) | Line 4579 for (;;) |
| 4579 | ||
| 4580 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4581 | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) |
| 4582 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4583 | break; | break; |
| 4584 | ||
| 4585 | case OP_DIGIT: | case OP_DIGIT: |
| 4586 | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) |
| 4587 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4588 | break; | break; |
| 4589 | ||
| 4590 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4591 | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) |
| 4592 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4593 | break; | break; |
| 4594 | ||
| 4595 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4596 | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) |
| 4597 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4598 | break; | break; |
| 4599 | ||
| 4600 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4601 | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) |
| 4602 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4603 | break; | break; |
| 4604 | ||
| 4605 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4606 | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) |
| 4607 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4608 | break; | break; |
| 4609 | ||
| 4610 | default: | default: |
| # | Line 3562 for (;;) | Line 4620 for (;;) |
| 4620 | { | { |
| 4621 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43); |
| 4622 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4623 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4624 | ((ims & PCRE_DOTALL) == 0 && IS_NEWLINE(eptr))) | if (eptr >= md->end_subject) |
| 4625 | RRETURN(MATCH_NOMATCH); | { |
| 4626 | SCHECK_PARTIAL(); | |
| 4627 | MRRETURN(MATCH_NOMATCH); | |
| 4628 | } | |
| 4629 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | |
| 4630 | MRRETURN(MATCH_NOMATCH); | |
| 4631 | c = *eptr++; | c = *eptr++; |
| 4632 | switch(ctype) | switch(ctype) |
| 4633 | { | { |
| 4634 | case OP_ANY: /* This is the DOTALL case */ | case OP_ANY: /* This is the non-NL case */ |
| 4635 | break; | case OP_ALLANY: |
| 4636 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 4637 | break; | break; |
| 4638 | ||
| 4639 | case OP_ANYNL: | case OP_ANYNL: |
| 4640 | switch(c) | switch(c) |
| 4641 | { | { |
| 4642 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4643 | case 0x000d: | case 0x000d: |
| 4644 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4645 | break; | break; |
| 4646 | ||
| 4647 | case 0x000a: | case 0x000a: |
| 4648 | break; | |
| 4649 | ||
| 4650 | case 0x000b: | case 0x000b: |
| 4651 | case 0x000c: | case 0x000c: |
| 4652 | case 0x0085: | case 0x0085: |
| 4653 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | |
| 4654 | break; | break; |
| 4655 | } | } |
| 4656 | break; | break; |
| # | Line 3597 for (;;) | Line 4662 for (;;) |
| 4662 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4663 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4664 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4665 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4666 | } | } |
| 4667 | break; | break; |
| 4668 | ||
| 4669 | case OP_HSPACE: | case OP_HSPACE: |
| 4670 | switch(c) | switch(c) |
| 4671 | { | { |
| 4672 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4673 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4674 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4675 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3621 for (;;) | Line 4686 for (;;) |
| 4686 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4687 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4688 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4689 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4690 | } | } |
| 4691 | break; | break; |
| 4692 | ||
| 4693 | case OP_VSPACE: | case OP_VSPACE: |
| 4694 | switch(c) | switch(c) |
| 4695 | { | { |
| 4696 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4697 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4698 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4699 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3639 for (;;) | Line 4704 for (;;) |
| 4704 | break; | break; |
| 4705 | ||
| 4706 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4707 | if ((md->ctypes[c] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); |
| 4708 | break; | break; |
| 4709 | ||
| 4710 | case OP_DIGIT: | case OP_DIGIT: |
| 4711 | if ((md->ctypes[c] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); |
| 4712 | break; | break; |
| 4713 | ||
| 4714 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4715 | if ((md->ctypes[c] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); |
| 4716 | break; | break; |
| 4717 | ||
| 4718 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4719 | if ((md->ctypes[c] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); |
| 4720 | break; | break; |
| 4721 | ||
| 4722 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4723 | if ((md->ctypes[c] & ctype_word) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_word) != 0) MRRETURN(MATCH_NOMATCH); |
| 4724 | break; | break; |
| 4725 | ||
| 4726 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4727 | if ((md->ctypes[c] & ctype_word) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_word) == 0) MRRETURN(MATCH_NOMATCH); |
| 4728 | break; | break; |
| 4729 | ||
| 4730 | default: | default: |
| # | Line 3687 for (;;) | Line 4752 for (;;) |
| 4752 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4753 | { | { |
| 4754 | int len = 1; | int len = 1; |
| 475 |