Parent Directory
|
Revision Log
|
Patch
| revision 403 by ph10, Sat Mar 21 17:33:11 2009 UTC | revision 538 by ph10, Wed Jun 9 19:30:57 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) 1997-2009 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 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 245 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 268 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 284 argument of match(), which never changes | Line 295 argument of match(), which never changes |
| 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 = (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 303 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 322 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 333 typedef struct heapframe { | Line 347 typedef struct heapframe { |
| 347 | ||
| 348 | /* Function local variables */ | /* Function local variables */ |
| 349 | ||
| 350 | const uschar *Xcallpat; | USPTR Xcallpat; |
| 351 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 352 | const uschar *Xcharptr; | USPTR Xcharptr; |
| 353 | #endif | #endif |
| 354 | const uschar *Xdata; | USPTR Xdata; |
| 355 | const uschar *Xnext; | USPTR Xnext; |
| 356 | const uschar *Xpp; | USPTR Xpp; |
| 357 | const uschar *Xprev; | USPTR Xprev; |
| 358 | const uschar *Xsaved_eptr; | USPTR Xsaved_eptr; |
| 359 | ||
| 360 | recursion_info Xnew_recursive; | recursion_info Xnew_recursive; |
| 361 | ||
| # | Line 398 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 && eptr > mstart)\ | |
| 426 | {\ | |
| 427 | md->hitend = TRUE;\ | |
| 428 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL);\ | |
| 429 | } | |
| 430 | ||
| 431 | #define SCHECK_PARTIAL()\ | |
| 432 | if (md->partial != 0 && eptr > mstart)\ | |
| 433 | {\ | |
| 434 | md->hitend = TRUE;\ | |
| 435 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL);\ | |
| 436 | } | |
| 437 | ||
| 438 | ||
| 439 | 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 |
| 440 | 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 |
| 441 | 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 |
| 442 | made performance worse. | made performance worse. |
| 443 | ||
| # | Line 410 Arguments: | Line 446 Arguments: |
| 446 | ecode pointer to current position in compiled code | ecode pointer to current position in compiled code |
| 447 | mstart pointer to the current match start position (can be modified | mstart pointer to the current match start position (can be modified |
| 448 | by encountering \K) | by encountering \K) |
| 449 | markptr pointer to the most recent MARK name, or NULL | |
| 450 | offset_top current top pointer | offset_top current top pointer |
| 451 | md pointer to "static" info for the match | md pointer to "static" info for the match |
| 452 | ims current /i, /m, and /s options | ims current /i, /m, and /s options |
| # | Line 423 Arguments: | Line 460 Arguments: |
| 460 | ||
| 461 | Returns: MATCH_MATCH if matched ) these values are >= 0 | Returns: MATCH_MATCH if matched ) these values are >= 0 |
| 462 | MATCH_NOMATCH if failed to match ) | MATCH_NOMATCH if failed to match ) |
| 463 | a negative MATCH_xxx value for PRUNE, SKIP, etc | |
| 464 | a negative PCRE_ERROR_xxx value if aborted by an error condition | a negative PCRE_ERROR_xxx value if aborted by an error condition |
| 465 | (e.g. stopped by repeated call or recursion limit) | (e.g. stopped by repeated call or recursion limit) |
| 466 | */ | */ |
| 467 | ||
| 468 | static int | static int |
| 469 | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart, | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart, |
| 470 | 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, |
| 471 | int flags, unsigned int rdepth) | eptrblock *eptrb, int flags, unsigned int rdepth) |
| 472 | { | { |
| 473 | /* 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, |
| 474 | 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 451 heap whenever RMATCH() does a "recursion | Line 489 heap whenever RMATCH() does a "recursion |
| 489 | ||
| 490 | #ifdef NO_RECURSE | #ifdef NO_RECURSE |
| 491 | heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe)); | heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe)); |
| 492 | if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY); | |
| 493 | frame->Xprevframe = NULL; /* Marks the top level */ | frame->Xprevframe = NULL; /* Marks the top level */ |
| 494 | ||
| 495 | /* Copy in the original argument variables */ | /* Copy in the original argument variables */ |
| # | Line 458 frame->Xprevframe = NULL; /* | Line 497 frame->Xprevframe = NULL; /* |
| 497 | frame->Xeptr = eptr; | frame->Xeptr = eptr; |
| 498 | frame->Xecode = ecode; | frame->Xecode = ecode; |
| 499 | frame->Xmstart = mstart; | frame->Xmstart = mstart; |
| 500 | frame->Xmarkptr = markptr; | |
| 501 | frame->Xoffset_top = offset_top; | frame->Xoffset_top = offset_top; |
| 502 | frame->Xims = ims; | frame->Xims = ims; |
| 503 | frame->Xeptrb = eptrb; | frame->Xeptrb = eptrb; |
| # | Line 473 HEAP_RECURSE: | Line 513 HEAP_RECURSE: |
| 513 | #define eptr frame->Xeptr | #define eptr frame->Xeptr |
| 514 | #define ecode frame->Xecode | #define ecode frame->Xecode |
| 515 | #define mstart frame->Xmstart | #define mstart frame->Xmstart |
| 516 | #define markptr frame->Xmarkptr | |
| 517 | #define offset_top frame->Xoffset_top | #define offset_top frame->Xoffset_top |
| 518 | #define ims frame->Xims | #define ims frame->Xims |
| 519 | #define eptrb frame->Xeptrb | #define eptrb frame->Xeptrb |
| # | Line 600 TAIL_RECURSE: | Line 641 TAIL_RECURSE: |
| 641 | /* 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 |
| 642 | 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 |
| 643 | 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() |
| 644 | and a "return", respectively (possibly with some debugging if DEBUG is | and a "return", respectively (possibly with some debugging if PCRE_DEBUG is |
| 645 | 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 |
| 646 | 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, |
| 647 | however, impact performance when true recursion is being used. */ | however, impact performance when true recursion is being used. */ |
| # | Line 641 for (;;) | Line 682 for (;;) |
| 682 | { | { |
| 683 | minimize = possessive = FALSE; | minimize = possessive = FALSE; |
| 684 | op = *ecode; | op = *ecode; |
| /* 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; | ||
| 685 | ||
| 686 | switch(op) | switch(op) |
| 687 | { | { |
| 688 | case OP_MARK: | |
| 689 | markptr = ecode + 2; | |
| 690 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 691 | ims, eptrb, flags, RM55); | |
| 692 | ||
| 693 | /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an | |
| 694 | argument, and we must check whether that argument matches this MARK's | |
| 695 | argument. It is passed back in md->start_match_ptr (an overloading of that | |
| 696 | variable). If it does match, we reset that variable to the current subject | |
| 697 | position and return MATCH_SKIP. Otherwise, pass back the return code | |
| 698 | unaltered. */ | |
| 699 | ||
| 700 | if (rrc == MATCH_SKIP_ARG && | |
| 701 | strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0) | |
| 702 | { | |
| 703 | md->start_match_ptr = eptr; | |
| 704 | RRETURN(MATCH_SKIP); | |
| 705 | } | |
| 706 | ||
| 707 | if (md->mark == NULL) md->mark = markptr; | |
| 708 | RRETURN(rrc); | |
| 709 | ||
| 710 | case OP_FAIL: | case OP_FAIL: |
| 711 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 712 | ||
| 713 | case OP_COMMIT: | |
| 714 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | |
| 715 | ims, eptrb, flags, RM52); | |
| 716 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 717 | MRRETURN(MATCH_COMMIT); | |
| 718 | ||
| 719 | case OP_PRUNE: | case OP_PRUNE: |
| 720 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 721 | ims, eptrb, flags, RM51); | ims, eptrb, flags, RM51); |
| 722 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 723 | RRETURN(MATCH_PRUNE); | MRRETURN(MATCH_PRUNE); |
| 724 | ||
| 725 | case OP_COMMIT: | case OP_PRUNE_ARG: |
| 726 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, |
| 727 | ims, eptrb, flags, RM52); | ims, eptrb, flags, RM56); |
| 728 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 729 | RRETURN(MATCH_COMMIT); | md->mark = ecode + 2; |
| 730 | RRETURN(MATCH_PRUNE); | |
| 731 | ||
| 732 | case OP_SKIP: | case OP_SKIP: |
| 733 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 734 | ims, eptrb, flags, RM53); | ims, eptrb, flags, RM53); |
| 735 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 736 | md->start_match_ptr = eptr; /* Pass back current position */ | md->start_match_ptr = eptr; /* Pass back current position */ |
| 737 | RRETURN(MATCH_SKIP); | MRRETURN(MATCH_SKIP); |
| 738 | ||
| 739 | case OP_SKIP_ARG: | |
| 740 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 741 | ims, eptrb, flags, RM57); | |
| 742 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 743 | ||
| 744 | /* Pass back the current skip name by overloading md->start_match_ptr and | |
| 745 | returning the special MATCH_SKIP_ARG return code. This will either be | |
| 746 | caught by a matching MARK, or get to the top, where it is treated the same | |
| 747 | as PRUNE. */ | |
| 748 | ||
| 749 | md->start_match_ptr = ecode + 2; | |
| 750 | RRETURN(MATCH_SKIP_ARG); | |
| 751 | ||
| 752 | case OP_THEN: | case OP_THEN: |
| 753 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 754 | ims, eptrb, flags, RM54); | ims, eptrb, flags, RM54); |
| 755 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 756 | MRRETURN(MATCH_THEN); | |
| 757 | ||
| 758 | case OP_THEN_ARG: | |
| 759 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 760 | ims, eptrb, flags, RM58); | |
| 761 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 762 | md->mark = ecode + 2; | |
| 763 | RRETURN(MATCH_THEN); | RRETURN(MATCH_THEN); |
| 764 | ||
| 765 | /* 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 699 for (;;) | Line 781 for (;;) |
| 781 | number = GET2(ecode, 1+LINK_SIZE); | number = GET2(ecode, 1+LINK_SIZE); |
| 782 | offset = number << 1; | offset = number << 1; |
| 783 | ||
| 784 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 785 | printf("start bracket %d\n", number); | printf("start bracket %d\n", number); |
| 786 | printf("subject="); | printf("subject="); |
| 787 | pchars(eptr, 16, TRUE, md); | pchars(eptr, 16, TRUE, md); |
| # | Line 714 for (;;) | Line 796 for (;;) |
| 796 | save_capture_last = md->capture_last; | save_capture_last = md->capture_last; |
| 797 | ||
| 798 | DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); | DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); |
| 799 | md->offset_vector[md->offset_end - number] = eptr - md->start_subject; | md->offset_vector[md->offset_end - number] = |
| 800 | (int)(eptr - md->start_subject); | |
| 801 | ||
| 802 | flags = (op == OP_SCBRA)? match_cbegroup : 0; | flags = (op == OP_SCBRA)? match_cbegroup : 0; |
| 803 | do | do |
| # | Line 733 for (;;) | Line 816 for (;;) |
| 816 | md->offset_vector[offset+1] = save_offset2; | md->offset_vector[offset+1] = save_offset2; |
| 817 | md->offset_vector[md->offset_end - number] = save_offset3; | md->offset_vector[md->offset_end - number] = save_offset3; |
| 818 | ||
| 819 | if (rrc != MATCH_THEN) md->mark = markptr; | |
| 820 | RRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 821 | } | } |
| 822 | ||
| # | Line 772 for (;;) | Line 856 for (;;) |
| 856 | ||
| 857 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, |
| 858 | eptrb, flags, RM48); | eptrb, flags, RM48); |
| 859 | if (rrc == MATCH_NOMATCH) md->mark = markptr; | |
| 860 | RRETURN(rrc); | RRETURN(rrc); |
| 861 | } | } |
| 862 | ||
| # | Line 794 for (;;) | Line 879 for (;;) |
| 879 | case OP_COND: | case OP_COND: |
| 880 | case OP_SCOND: | case OP_SCOND: |
| 881 | codelink= GET(ecode, 1); | codelink= GET(ecode, 1); |
| 882 | ||
| 883 | /* Because of the way auto-callout works during compile, a callout item is | /* Because of the way auto-callout works during compile, a callout item is |
| 884 | inserted between OP_COND and an assertion condition. */ | inserted between OP_COND and an assertion condition. */ |
| 885 | ||
| # | Line 807 for (;;) | Line 892 for (;;) |
| 892 | cb.callout_number = ecode[LINK_SIZE+2]; | cb.callout_number = ecode[LINK_SIZE+2]; |
| 893 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
| 894 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
| 895 | cb.subject_length = md->end_subject - md->start_subject; | cb.subject_length = (int)(md->end_subject - md->start_subject); |
| 896 | cb.start_match = mstart - md->start_subject; | cb.start_match = (int)(mstart - md->start_subject); |
| 897 | cb.current_position = eptr - md->start_subject; | cb.current_position = (int)(eptr - md->start_subject); |
| 898 | cb.pattern_position = GET(ecode, LINK_SIZE + 3); | cb.pattern_position = GET(ecode, LINK_SIZE + 3); |
| 899 | cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE); | cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE); |
| 900 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 901 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 902 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 903 | if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH); | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
| 904 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 905 | } | } |
| 906 | ecode += _pcre_OP_lengths[OP_CALLOUT]; | ecode += _pcre_OP_lengths[OP_CALLOUT]; |
| 907 | } | } |
| 908 | ||
| 909 | condcode = ecode[LINK_SIZE+1]; | condcode = ecode[LINK_SIZE+1]; |
| 910 | ||
| 911 | /* Now see what the actual condition is */ | /* Now see what the actual condition is */ |
| 912 | ||
| 913 | if (condcode == OP_RREF) /* Recursion test */ | if (condcode == OP_RREF || condcode == OP_NRREF) /* Recursion test */ |
| 914 | { | { |
| 915 | offset = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | if (md->recursive == NULL) /* Not recursing => FALSE */ |
| 916 | condition = md->recursive != NULL && | { |
| 917 | (offset == RREF_ANY || offset == md->recursive->group_num); | condition = FALSE; |
| 918 | ecode += condition? 3 : GET(ecode, 1); | ecode += GET(ecode, 1); |
| 919 | } | |
| 920 | else | |
| 921 | { | |
| 922 | int recno = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | |
| 923 | condition = (recno == RREF_ANY || recno == md->recursive->group_num); | |
| 924 | ||
| 925 | /* If the test is for recursion into a specific subpattern, and it is | |
| 926 | false, but the test was set up by name, scan the table to see if the | |
| 927 | name refers to any other numbers, and test them. The condition is true | |
| 928 | if any one is set. */ | |
| 929 | ||
| 930 | if (!condition && condcode == OP_NRREF && recno != RREF_ANY) | |
| 931 | { | |
| 932 | uschar *slotA = md->name_table; | |
| 933 | for (i = 0; i < md->name_count; i++) | |
| 934 | { | |
| 935 | if (GET2(slotA, 0) == recno) break; | |
| 936 | slotA += md->name_entry_size; | |
| 937 | } | |
| 938 | ||
| 939 | /* Found a name for the number - there can be only one; duplicate | |
| 940 | names for different numbers are allowed, but not vice versa. First | |
| 941 | scan down for duplicates. */ | |
| 942 | ||
| 943 | if (i < md->name_count) | |
| 944 | { | |
| 945 | uschar *slotB = slotA; | |
| 946 | while (slotB > md->name_table) | |
| 947 | { | |
| 948 | slotB -= md->name_entry_size; | |
| 949 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 950 | { | |
| 951 | condition = GET2(slotB, 0) == md->recursive->group_num; | |
| 952 | if (condition) break; | |
| 953 | } | |
| 954 | else break; | |
| 955 | } | |
| 956 | ||
| 957 | /* Scan up for duplicates */ | |
| 958 | ||
| 959 | if (!condition) | |
| 960 | { | |
| 961 | slotB = slotA; | |
| 962 | for (i++; i < md->name_count; i++) | |
| 963 | { | |
| 964 | slotB += md->name_entry_size; | |
| 965 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 966 | { | |
| 967 | condition = GET2(slotB, 0) == md->recursive->group_num; | |
| 968 | if (condition) break; | |
| 969 | } | |
| 970 | else break; | |
| 971 | } | |
| 972 | } | |
| 973 | } | |
| 974 | } | |
| 975 | ||
| 976 | /* Chose branch according to the condition */ | |
| 977 | ||
| 978 | ecode += condition? 3 : GET(ecode, 1); | |
| 979 | } | |
| 980 | } | } |
| 981 | ||
| 982 | else if (condcode == OP_CREF) /* Group used test */ | else if (condcode == OP_CREF || condcode == OP_NCREF) /* Group used test */ |
| 983 | { | { |
| 984 | offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ | offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ |
| 985 | condition = offset < offset_top && md->offset_vector[offset] >= 0; | condition = offset < offset_top && md->offset_vector[offset] >= 0; |
| 986 | ||
| 987 | /* If the numbered capture is unset, but the reference was by name, | |
| 988 | scan the table to see if the name refers to any other numbers, and test | |
| 989 | them. The condition is true if any one is set. This is tediously similar | |
| 990 | to the code above, but not close enough to try to amalgamate. */ | |
| 991 | ||
| 992 | if (!condition && condcode == OP_NCREF) | |
| 993 | { | |
| 994 | int refno = offset >> 1; | |
| 995 | uschar *slotA = md->name_table; | |
| 996 | ||
| 997 | for (i = 0; i < md->name_count; i++) | |
| 998 | { | |
| 999 | if (GET2(slotA, 0) == refno) break; | |
| 1000 | slotA += md->name_entry_size; | |
| 1001 | } | |
| 1002 | ||
| 1003 | /* Found a name for the number - there can be only one; duplicate names | |
| 1004 | for different numbers are allowed, but not vice versa. First scan down | |
| 1005 | for duplicates. */ | |
| 1006 | ||
| 1007 | if (i < md->name_count) | |
| 1008 | { | |
| 1009 | uschar *slotB = slotA; | |
| 1010 | while (slotB > md->name_table) | |
| 1011 | { | |
| 1012 | slotB -= md->name_entry_size; | |
| 1013 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1014 | { | |
| 1015 | offset = GET2(slotB, 0) << 1; | |
| 1016 | condition = offset < offset_top && | |
| 1017 | md->offset_vector[offset] >= 0; | |
| 1018 | if (condition) break; | |
| 1019 | } | |
| 1020 | else break; | |
| 1021 | } | |
| 1022 | ||
| 1023 | /* Scan up for duplicates */ | |
| 1024 | ||
| 1025 | if (!condition) | |
| 1026 | { | |
| 1027 | slotB = slotA; | |
| 1028 | for (i++; i < md->name_count; i++) | |
| 1029 | { | |
| 1030 | slotB += md->name_entry_size; | |
| 1031 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1032 | { | |
| 1033 | offset = GET2(slotB, 0) << 1; | |
| 1034 | condition = offset < offset_top && | |
| 1035 | md->offset_vector[offset] >= 0; | |
| 1036 | if (condition) break; | |
| 1037 | } | |
| 1038 | else break; | |
| 1039 | } | |
| 1040 | } | |
| 1041 | } | |
| 1042 | } | |
| 1043 | ||
| 1044 | /* Chose branch according to the condition */ | |
| 1045 | ||
| 1046 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 3 : GET(ecode, 1); |
| 1047 | } | } |
| 1048 | ||
| # | Line 897 for (;;) | Line 1103 for (;;) |
| 1103 | break; | break; |
| 1104 | ||
| 1105 | ||
| 1106 | /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes, | |
| 1107 | to close any currently open capturing brackets. */ | |
| 1108 | ||
| 1109 | case OP_CLOSE: | |
| 1110 | number = GET2(ecode, 1); | |
| 1111 | offset = number << 1; | |
| 1112 | ||
| 1113 | #ifdef PCRE_DEBUG | |
| 1114 | printf("end bracket %d at *ACCEPT", number); | |
| 1115 | printf("\n"); | |
| 1116 | #endif | |
| 1117 | ||
| 1118 | md->capture_last = number; | |
| 1119 | if (offset >= md->offset_max) md->offset_overflow = TRUE; else | |
| 1120 | { | |
| 1121 | md->offset_vector[offset] = | |
| 1122 | md->offset_vector[md->offset_end - number]; | |
| 1123 | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); | |
| 1124 | if (offset_top <= offset) offset_top = offset + 2; | |
| 1125 | } | |
| 1126 | ecode += 3; | |
| 1127 | break; | |
| 1128 | ||
| 1129 | ||
| 1130 | /* 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 |
| 1131 | recursion, we should restore the offsets appropriately and continue from | recursion, we should restore the offsets appropriately and continue from |
| 1132 | after the call. */ | after the call. */ |
| # | Line 910 for (;;) | Line 1140 for (;;) |
| 1140 | md->recursive = rec->prevrec; | md->recursive = rec->prevrec; |
| 1141 | memmove(md->offset_vector, rec->offset_save, | memmove(md->offset_vector, rec->offset_save, |
| 1142 | rec->saved_max * sizeof(int)); | rec->saved_max * sizeof(int)); |
| 1143 | mstart = rec->save_start; | offset_top = rec->save_offset_top; |
| 1144 | ims = original_ims; | ims = original_ims; |
| 1145 | ecode = rec->after_call; | ecode = rec->after_call; |
| 1146 | break; | break; |
| 1147 | } | } |
| 1148 | ||
| 1149 | /* 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 |
| 1150 | 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 |
| 1151 | the subject. In both cases, backtracking will then try other alternatives, | |
| 1152 | if any. */ | |
| 1153 | ||
| 1154 | if (eptr == mstart && | |
| 1155 | (md->notempty || | |
| 1156 | (md->notempty_atstart && | |
| 1157 | mstart == md->start_subject + md->start_offset))) | |
| 1158 | MRRETURN(MATCH_NOMATCH); | |
| 1159 | ||
| 1160 | /* Otherwise, we have a match. */ | |
| 1161 | ||
| if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH); | ||
| 1162 | md->end_match_ptr = eptr; /* Record where we ended */ | md->end_match_ptr = eptr; /* Record where we ended */ |
| 1163 | md->end_offset_top = offset_top; /* and how many extracts were taken */ | md->end_offset_top = offset_top; /* and how many extracts were taken */ |
| 1164 | md->start_match_ptr = mstart; /* and the start (\K can modify) */ | md->start_match_ptr = mstart; /* and the start (\K can modify) */ |
| 1165 | RRETURN(MATCH_MATCH); | |
| 1166 | /* For some reason, the macros don't work properly if an expression is | |
| 1167 | given as the argument to MRRETURN when the heap is in use. */ | |
| 1168 | ||
| 1169 | rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; | |
| 1170 | MRRETURN(rrc); | |
| 1171 | ||
| 1172 | /* Change option settings */ | /* Change option settings */ |
| 1173 | ||
| # | Line 945 for (;;) | Line 1189 for (;;) |
| 1189 | { | { |
| 1190 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, |
| 1191 | RM4); | RM4); |
| 1192 | if (rrc == MATCH_MATCH) break; | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
| 1193 | { | |
| 1194 | mstart = md->start_match_ptr; /* In case \K reset it */ | |
| 1195 | break; | |
| 1196 | } | |
| 1197 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| 1198 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 1199 | } | } |
| 1200 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| 1201 | if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); | if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH); |
| 1202 | ||
| 1203 | /* If checking an assertion for a condition, return MATCH_MATCH. */ | /* If checking an assertion for a condition, return MATCH_MATCH. */ |
| 1204 | ||
| # | Line 964 for (;;) | Line 1212 for (;;) |
| 1212 | offset_top = md->end_offset_top; | offset_top = md->end_offset_top; |
| 1213 | continue; | continue; |
| 1214 | ||
| 1215 | /* Negative assertion: all branches must fail to match */ | /* Negative assertion: all branches must fail to match. Encountering SKIP, |
| 1216 | PRUNE, or COMMIT means we must assume failure without checking subsequent | |
| 1217 | branches. */ | |
| 1218 | ||
| 1219 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
| 1220 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
| # | Line 972 for (;;) | Line 1222 for (;;) |
| 1222 | { | { |
| 1223 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, |
| 1224 | RM5); | RM5); |
| 1225 | if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH); | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH); |
| 1226 | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) | |
| 1227 | { | |
| 1228 | do ecode += GET(ecode,1); while (*ecode == OP_ALT); | |
| 1229 | break; | |
| 1230 | } | |
| 1231 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| 1232 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
| 1233 | } | } |
| # | Line 996 for (;;) | Line 1251 for (;;) |
| 1251 | while (i-- > 0) | while (i-- > 0) |
| 1252 | { | { |
| 1253 | eptr--; | eptr--; |
| 1254 | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1255 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 1256 | } | } |
| 1257 | } | } |
| # | Line 1007 for (;;) | Line 1262 for (;;) |
| 1262 | ||
| 1263 | { | { |
| 1264 | eptr -= GET(ecode, 1); | eptr -= GET(ecode, 1); |
| 1265 | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1266 | } | } |
| 1267 | ||
| 1268 | /* Skip to next op code */ | /* Save the earliest consulted character, then skip to next op code */ |
| 1269 | ||
| 1270 | if (eptr < md->start_used_ptr) md->start_used_ptr = eptr; | |
| 1271 | ecode += 1 + LINK_SIZE; | ecode += 1 + LINK_SIZE; |
| 1272 | break; | break; |
| 1273 | ||
| # | Line 1027 for (;;) | Line 1283 for (;;) |
| 1283 | cb.callout_number = ecode[1]; | cb.callout_number = ecode[1]; |
| 1284 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
| 1285 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
| 1286 | cb.subject_length = md->end_subject - md->start_subject; | cb.subject_length = (int)(md->end_subject - md->start_subject); |
| 1287 | cb.start_match = mstart - md->start_subject; | cb.start_match = (int)(mstart - md->start_subject); |
| 1288 | cb.current_position = eptr - md->start_subject; | cb.current_position = (int)(eptr - md->start_subject); |
| 1289 | cb.pattern_position = GET(ecode, 2); | cb.pattern_position = GET(ecode, 2); |
| 1290 | cb.next_item_length = GET(ecode, 2 + LINK_SIZE); | cb.next_item_length = GET(ecode, 2 + LINK_SIZE); |
| 1291 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 1292 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 1293 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 1294 | if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH); | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
| 1295 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 1296 | } | } |
| 1297 | ecode += 2 + 2*LINK_SIZE; | ecode += 2 + 2*LINK_SIZE; |
| # | Line 1090 for (;;) | Line 1346 for (;;) |
| 1346 | ||
| 1347 | memcpy(new_recursive.offset_save, md->offset_vector, | memcpy(new_recursive.offset_save, md->offset_vector, |
| 1348 | new_recursive.saved_max * sizeof(int)); | new_recursive.saved_max * sizeof(int)); |
| 1349 | new_recursive.save_start = mstart; | new_recursive.save_offset_top = offset_top; |
| mstart = eptr; | ||
| 1350 | ||
| 1351 | /* 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 |
| 1352 | restore the offset and recursion data. */ | restore the offset and recursion data. */ |
| # | Line 1102 for (;;) | Line 1357 for (;;) |
| 1357 | { | { |
| 1358 | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, |
| 1359 | md, ims, eptrb, flags, RM6); | md, ims, eptrb, flags, RM6); |
| 1360 | if (rrc == MATCH_MATCH) | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
| 1361 | { | { |
| 1362 | DPRINTF(("Recursion matched\n")); | DPRINTF(("Recursion matched\n")); |
| 1363 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1364 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1365 | (pcre_free)(new_recursive.offset_save); | (pcre_free)(new_recursive.offset_save); |
| 1366 | RRETURN(MATCH_MATCH); | MRRETURN(MATCH_MATCH); |
| 1367 | } | } |
| 1368 | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) |
| 1369 | { | { |
| # | Line 1129 for (;;) | Line 1384 for (;;) |
| 1384 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1385 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1386 | (pcre_free)(new_recursive.offset_save); | (pcre_free)(new_recursive.offset_save); |
| 1387 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1388 | } | } |
| 1389 | /* Control never reaches here */ | /* Control never reaches here */ |
| 1390 | ||
| # | Line 1138 for (;;) | Line 1393 for (;;) |
| 1393 | a move back into the brackets. Friedl calls these "atomic" subpatterns. | a move back into the brackets. Friedl calls these "atomic" subpatterns. |
| 1394 | 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 |
| 1395 | 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 |
| 1396 | the end of a normal bracket, leaving the subject pointer. */ | the end of a normal bracket, leaving the subject pointer, but resetting |
| 1397 | the start-of-match value in case it was changed by \K. */ | |
| 1398 | ||
| 1399 | case OP_ONCE: | case OP_ONCE: |
| 1400 | prev = ecode; | prev = ecode; |
| # | Line 1147 for (;;) | Line 1403 for (;;) |
| 1403 | do | do |
| 1404 | { | { |
| 1405 | 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); |
| 1406 | if (rrc == MATCH_MATCH) break; | if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ |
| 1407 | { | |
| 1408 | mstart = md->start_match_ptr; | |
| 1409 | break; | |
| 1410 | } | |
| 1411 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| 1412 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
| 1413 | } | } |
| # | Line 1266 for (;;) | Line 1526 for (;;) |
| 1526 | } | } |
| 1527 | else saved_eptr = NULL; | else saved_eptr = NULL; |
| 1528 | ||
| 1529 | /* 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 |
| 1530 | 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 |
| 1531 | assertions. Do this also for the "once" (atomic) groups. */ | use by positive assertions. We also need to record the match start in case |
| 1532 | it was changed by \K. */ | |
| 1533 | ||
| 1534 | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || |
| 1535 | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || |
| # | Line 1276 for (;;) | Line 1537 for (;;) |
| 1537 | { | { |
| 1538 | md->end_match_ptr = eptr; /* For ONCE */ | md->end_match_ptr = eptr; /* For ONCE */ |
| 1539 | md->end_offset_top = offset_top; | md->end_offset_top = offset_top; |
| 1540 | RRETURN(MATCH_MATCH); | md->start_match_ptr = mstart; |
| 1541 | MRRETURN(MATCH_MATCH); | |
| 1542 | } | } |
| 1543 | ||
| 1544 | /* 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 1290 for (;;) | Line 1552 for (;;) |
| 1552 | number = GET2(prev, 1+LINK_SIZE); | number = GET2(prev, 1+LINK_SIZE); |
| 1553 | offset = number << 1; | offset = number << 1; |
| 1554 | ||
| 1555 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 1556 | printf("end bracket %d", number); | printf("end bracket %d", number); |
| 1557 | printf("\n"); | printf("\n"); |
| 1558 | #endif | #endif |
| # | Line 1300 for (;;) | Line 1562 for (;;) |
| 1562 | { | { |
| 1563 | md->offset_vector[offset] = | md->offset_vector[offset] = |
| 1564 | md->offset_vector[md->offset_end - number]; | md->offset_vector[md->offset_end - number]; |
| 1565 | md->offset_vector[offset+1] = eptr - md->start_subject; | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); |
| 1566 | if (offset_top <= offset) offset_top = offset + 2; | if (offset_top <= offset) offset_top = offset + 2; |
| 1567 | } | } |
| 1568 | ||
| # | Line 1312 for (;;) | Line 1574 for (;;) |
| 1574 | recursion_info *rec = md->recursive; | recursion_info *rec = md->recursive; |
| 1575 | DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); | DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); |
| 1576 | md->recursive = rec->prevrec; | md->recursive = rec->prevrec; |
| mstart = rec->save_start; | ||
| 1577 | memcpy(md->offset_vector, rec->offset_save, | memcpy(md->offset_vector, rec->offset_save, |
| 1578 | rec->saved_max * sizeof(int)); | rec->saved_max * sizeof(int)); |
| 1579 | offset_top = rec->save_offset_top; | |
| 1580 | ecode = rec->after_call; | ecode = rec->after_call; |
| 1581 | ims = original_ims; | ims = original_ims; |
| 1582 | break; | break; |
| # | Line 1371 for (;;) | Line 1633 for (;;) |
| 1633 | /* Start of subject unless notbol, or after internal newline if multiline */ | /* Start of subject unless notbol, or after internal newline if multiline */ |
| 1634 | ||
| 1635 | case OP_CIRC: | case OP_CIRC: |
| 1636 | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1637 | if ((ims & PCRE_MULTILINE) != 0) | if ((ims & PCRE_MULTILINE) != 0) |
| 1638 | { | { |
| 1639 | if (eptr != md->start_subject && | if (eptr != md->start_subject && |
| 1640 | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) |
| 1641 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1642 | ecode++; | ecode++; |
| 1643 | break; | break; |
| 1644 | } | } |
| # | Line 1385 for (;;) | Line 1647 for (;;) |
| 1647 | /* Start of subject assertion */ | /* Start of subject assertion */ |
| 1648 | ||
| 1649 | case OP_SOD: | case OP_SOD: |
| 1650 | if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1651 | ecode++; | ecode++; |
| 1652 | break; | break; |
| 1653 | ||
| 1654 | /* Start of match assertion */ | /* Start of match assertion */ |
| 1655 | ||
| 1656 | case OP_SOM: | case OP_SOM: |
| 1657 | if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH); |
| 1658 | ecode++; | ecode++; |
| 1659 | break; | break; |
| 1660 | ||
| # | Line 1410 for (;;) | Line 1672 for (;;) |
| 1672 | if ((ims & PCRE_MULTILINE) != 0) | if ((ims & PCRE_MULTILINE) != 0) |
| 1673 | { | { |
| 1674 | if (eptr < md->end_subject) | if (eptr < md->end_subject) |
| 1675 | { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); } | { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); } |
| 1676 | else | else |
| 1677 | { if (md->noteol) RRETURN(MATCH_NOMATCH); } | { if (md->noteol) MRRETURN(MATCH_NOMATCH); } |
| 1678 | ecode++; | ecode++; |
| 1679 | break; | break; |
| 1680 | } | } |
| 1681 | else | else |
| 1682 | { | { |
| 1683 | if (md->noteol) RRETURN(MATCH_NOMATCH); | if (md->noteol) MRRETURN(MATCH_NOMATCH); |
| 1684 | if (!md->endonly) | if (!md->endonly) |
| 1685 | { | { |
| 1686 | if (eptr != md->end_subject && | if (eptr != md->end_subject && |
| 1687 | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) |
| 1688 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1689 | ecode++; | ecode++; |
| 1690 | break; | break; |
| 1691 | } | } |
| # | Line 1433 for (;;) | Line 1695 for (;;) |
| 1695 | /* End of subject assertion (\z) */ | /* End of subject assertion (\z) */ |
| 1696 | ||
| 1697 | case OP_EOD: | case OP_EOD: |
| 1698 | if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH); |
| 1699 | ecode++; | ecode++; |
| 1700 | break; | break; |
| 1701 | ||
| # | Line 1442 for (;;) | Line 1704 for (;;) |
| 1704 | case OP_EODN: | case OP_EODN: |
| 1705 | if (eptr != md->end_subject && | if (eptr != md->end_subject && |
| 1706 | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) |
| 1707 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1708 | ecode++; | ecode++; |
| 1709 | break; | break; |
| 1710 | ||
| # | Line 1454 for (;;) | Line 1716 for (;;) |
| 1716 | ||
| 1717 | /* Find out if the previous and current characters are "word" characters. | /* Find out if the previous and current characters are "word" characters. |
| 1718 | 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 |
| 1719 | be "non-word" characters. */ | be "non-word" characters. Remember the earliest consulted character for |
| 1720 | partial matching. */ | |
| 1721 | ||
| 1722 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 1723 | if (utf8) | if (utf8) |
| 1724 | { | { |
| 1725 | /* Get status of previous character */ | |
| 1726 | ||
| 1727 | if (eptr == md->start_subject) prev_is_word = FALSE; else | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 1728 | { | { |
| 1729 | const uschar *lastptr = eptr - 1; | USPTR lastptr = eptr - 1; |
| 1730 | while((*lastptr & 0xc0) == 0x80) lastptr--; | while((*lastptr & 0xc0) == 0x80) lastptr--; |
| 1731 | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; | |
| 1732 | GETCHAR(c, lastptr); | GETCHAR(c, lastptr); |
| 1733 | #ifdef SUPPORT_UCP | |
| 1734 | if (md->use_ucp) | |
| 1735 | { | |
| 1736 | if (c == '_') prev_is_word = TRUE; else | |
| 1737 | { | |
| 1738 | int cat = UCD_CATEGORY(c); | |
| 1739 | prev_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1740 | } | |
| 1741 | } | |
| 1742 | else | |
| 1743 | #endif | |
| 1744 | prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; | prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; |
| 1745 | } | } |
| 1746 | if (eptr >= md->end_subject) cur_is_word = FALSE; else | |
| 1747 | /* Get status of next character */ | |
| 1748 | ||
| 1749 | if (eptr >= md->end_subject) | |
| 1750 | { | |
| 1751 | SCHECK_PARTIAL(); | |
| 1752 | cur_is_word = FALSE; | |
| 1753 | } | |
| 1754 | else | |
| 1755 | { | { |
| 1756 | GETCHAR(c, eptr); | GETCHAR(c, eptr); |
| 1757 | #ifdef SUPPORT_UCP | |
| 1758 | if (md->use_ucp) | |
| 1759 | { | |
| 1760 | if (c == '_') cur_is_word = TRUE; else | |
| 1761 | { | |
| 1762 | int cat = UCD_CATEGORY(c); | |
| 1763 | cur_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1764 | } | |
| 1765 | } | |
| 1766 | else | |
| 1767 | #endif | |
| 1768 | cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; | cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; |
| 1769 | } | } |
| 1770 | } | } |
| 1771 | else | else |
| 1772 | #endif | #endif |
| 1773 | ||
| 1774 | /* More streamlined when not in UTF-8 mode */ | /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for |
| 1775 | consistency with the behaviour of \w we do use it in this case. */ | |
| 1776 | ||
| 1777 | { | { |
| 1778 | prev_is_word = (eptr != md->start_subject) && | /* Get status of previous character */ |
| 1779 | ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
| 1780 | cur_is_word = (eptr < md->end_subject) && | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 1781 | ((md->ctypes[*eptr] & ctype_word) != 0); | { |
| 1782 | if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1; | |
| 1783 | #ifdef SUPPORT_UCP | |
| 1784 | if (md->use_ucp) | |
| 1785 | { | |
| 1786 | c = eptr[-1]; | |
| 1787 | if (c == '_') prev_is_word = TRUE; else | |
| 1788 | { | |
| 1789 | int cat = UCD_CATEGORY(c); | |
| 1790 | prev_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1791 | } | |
| 1792 | } | |
| 1793 | else | |
| 1794 | #endif | |
| 1795 | prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
| 1796 | } | |
| 1797 | ||
| 1798 | /* Get status of next character */ | |
| 1799 | ||
| 1800 | if (eptr >= md->end_subject) | |
| 1801 | { | |
| 1802 | SCHECK_PARTIAL(); | |
| 1803 | cur_is_word = FALSE; | |
| 1804 | } | |
| 1805 | else | |
| 1806 | #ifdef SUPPORT_UCP | |
| 1807 | if (md->use_ucp) | |
| 1808 | { | |
| 1809 | c = *eptr; | |
| 1810 | if (c == '_') cur_is_word = TRUE; else | |
| 1811 | { | |
| 1812 | int cat = UCD_CATEGORY(c); | |
| 1813 | cur_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1814 | } | |
| 1815 | } | |
| 1816 | else | |
| 1817 | #endif | |
| 1818 | cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0); | |
| 1819 | } | } |
| 1820 | ||
| 1821 | /* Now see if the situation is what we want */ | /* Now see if the situation is what we want */ |
| 1822 | ||
| 1823 | if ((*ecode++ == OP_WORD_BOUNDARY)? | if ((*ecode++ == OP_WORD_BOUNDARY)? |
| 1824 | cur_is_word == prev_is_word : cur_is_word != prev_is_word) | cur_is_word == prev_is_word : cur_is_word != prev_is_word) |
| 1825 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1826 | } | } |
| 1827 | break; | break; |
| 1828 | ||
| 1829 | /* Match a single character type; inline for speed */ | /* Match a single character type; inline for speed */ |
| 1830 | ||
| 1831 | case OP_ANY: | case OP_ANY: |
| 1832 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); |
| 1833 | /* Fall through */ | /* Fall through */ |
| 1834 | ||
| 1835 | case OP_ALLANY: | case OP_ALLANY: |
| 1836 | if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr++ >= md->end_subject) |
| 1837 | { | |
| 1838 | SCHECK_PARTIAL(); | |
| 1839 | MRRETURN(MATCH_NOMATCH); | |
| 1840 | } | |
| 1841 | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 1842 | ecode++; | ecode++; |
| 1843 | break; | break; |
| # | Line 1508 for (;;) | Line 1846 for (;;) |
| 1846 | any byte, even newline, independent of the setting of PCRE_DOTALL. */ | any byte, even newline, independent of the setting of PCRE_DOTALL. */ |
| 1847 | ||
| 1848 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 1849 | if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr++ >= md->end_subject) |
| 1850 | { | |
| 1851 | SCHECK_PARTIAL(); | |
| 1852 | MRRETURN(MATCH_NOMATCH); | |
| 1853 | } | |
| 1854 | ecode++; | ecode++; |
| 1855 | break; | break; |
| 1856 | ||
| 1857 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 1858 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1859 | { | |
| 1860 | SCHECK_PARTIAL(); | |
| 1861 | MRRETURN(MATCH_NOMATCH); | |
| 1862 | } | |
| 1863 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1864 | if ( | if ( |
| 1865 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1521 for (;;) | Line 1867 for (;;) |
| 1867 | #endif | #endif |
| 1868 | (md->ctypes[c] & ctype_digit) != 0 | (md->ctypes[c] & ctype_digit) != 0 |
| 1869 | ) | ) |
| 1870 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1871 | ecode++; | ecode++; |
| 1872 | break; | break; |
| 1873 | ||
| 1874 | case OP_DIGIT: | case OP_DIGIT: |
| 1875 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1876 | { | |
| 1877 | SCHECK_PARTIAL(); | |
| 1878 | MRRETURN(MATCH_NOMATCH); | |
| 1879 | } | |
| 1880 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1881 | if ( | if ( |
| 1882 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1534 for (;;) | Line 1884 for (;;) |
| 1884 | #endif | #endif |
| 1885 | (md->ctypes[c] & ctype_digit) == 0 | (md->ctypes[c] & ctype_digit) == 0 |
| 1886 | ) | ) |
| 1887 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1888 | ecode++; | ecode++; |
| 1889 | break; | break; |
| 1890 | ||
| 1891 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 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 1547 for (;;) | Line 1901 for (;;) |
| 1901 | #endif | #endif |
| 1902 | (md->ctypes[c] & ctype_space) != 0 | (md->ctypes[c] & ctype_space) != 0 |
| 1903 | ) | ) |
| 1904 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1905 | ecode++; | ecode++; |
| 1906 | break; | break; |
| 1907 | ||
| 1908 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 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 1560 for (;;) | Line 1918 for (;;) |
| 1918 | #endif | #endif |
| 1919 | (md->ctypes[c] & ctype_space) == 0 | (md->ctypes[c] & ctype_space) == 0 |
| 1920 | ) | ) |
| 1921 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1922 | ecode++; | ecode++; |
| 1923 | break; | break; |
| 1924 | ||
| 1925 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 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 1573 for (;;) | Line 1935 for (;;) |
| 1935 | #endif | #endif |
| 1936 | (md->ctypes[c] & ctype_word) != 0 | (md->ctypes[c] & ctype_word) != 0 |
| 1937 | ) | ) |
| 1938 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1939 | ecode++; | ecode++; |
| 1940 | break; | break; |
| 1941 | ||
| 1942 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 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 1586 for (;;) | Line 1952 for (;;) |
| 1952 | #endif | #endif |
| 1953 | (md->ctypes[c] & ctype_word) == 0 | (md->ctypes[c] & ctype_word) == 0 |
| 1954 | ) | ) |
| 1955 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1956 | ecode++; | ecode++; |
| 1957 | break; | break; |
| 1958 | ||
| 1959 | case OP_ANYNL: | case OP_ANYNL: |
| 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 | switch(c) | switch(c) |
| 1967 | { | { |
| 1968 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 1969 | case 0x000d: | case 0x000d: |
| 1970 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 1971 | break; | break; |
| # | Line 1608 for (;;) | Line 1978 for (;;) |
| 1978 | case 0x0085: | case 0x0085: |
| 1979 | case 0x2028: | case 0x2028: |
| 1980 | case 0x2029: | case 0x2029: |
| 1981 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 1982 | break; | break; |
| 1983 | } | } |
| 1984 | ecode++; | ecode++; |
| 1985 | break; | break; |
| 1986 | ||
| 1987 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 1988 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1989 | { | |
| 1990 | SCHECK_PARTIAL(); | |
| 1991 | MRRETURN(MATCH_NOMATCH); | |
| 1992 | } | |
| 1993 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1994 | switch(c) | switch(c) |
| 1995 | { | { |
| # | Line 1639 for (;;) | Line 2013 for (;;) |
| 2013 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 2014 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 2015 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 2016 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2017 | } | } |
| 2018 | ecode++; | ecode++; |
| 2019 | break; | break; |
| 2020 | ||
| 2021 | case OP_HSPACE: | case OP_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 | { | { |
| 2030 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2031 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 2032 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 2033 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 1675 for (;;) | Line 2053 for (;;) |
| 2053 | break; | break; |
| 2054 | ||
| 2055 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 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 | { | { |
| # | Line 1687 for (;;) | Line 2069 for (;;) |
| 2069 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 2070 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 2071 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 2072 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2073 | } | } |
| 2074 | ecode++; | ecode++; |
| 2075 | break; | break; |
| 2076 | ||
| 2077 | case OP_VSPACE: | case OP_VSPACE: |
| 2078 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2079 | { | |
| 2080 | SCHECK_PARTIAL(); | |
| 2081 | MRRETURN(MATCH_NOMATCH); | |
| 2082 | } | |
| 2083 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2084 | switch(c) | switch(c) |
| 2085 | { | { |
| 2086 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2087 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 2088 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 2089 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 1716 for (;;) | Line 2102 for (;;) |
| 2102 | ||
| 2103 | case OP_PROP: | case OP_PROP: |
| 2104 | case OP_NOTPROP: | case OP_NOTPROP: |
| 2105 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2106 | { | |
| 2107 | SCHECK_PARTIAL(); | |
| 2108 | MRRETURN(MATCH_NOMATCH); | |
| 2109 | } | |
| 2110 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2111 | { | { |
| 2112 | const ucd_record *prop = GET_UCD(c); | const ucd_record *prop = GET_UCD(c); |
| # | Line 1724 for (;;) | Line 2114 for (;;) |
| 2114 | switch(ecode[1]) | switch(ecode[1]) |
| 2115 | { | { |
| 2116 | case PT_ANY: | case PT_ANY: |
| 2117 | if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); | if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH); |
| 2118 | break; | break; |
| 2119 | ||
| 2120 | case PT_LAMP: | case PT_LAMP: |
| 2121 | if ((prop->chartype == ucp_Lu || | if ((prop->chartype == ucp_Lu || |
| 2122 | prop->chartype == ucp_Ll || | prop->chartype == ucp_Ll || |
| 2123 | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) |
| 2124 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2125 | break; | break; |
| 2126 | ||
| 2127 | case PT_GC: | case PT_GC: |
| 2128 | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) |
| 2129 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2130 | break; | break; |
| 2131 | ||
| 2132 | case PT_PC: | case PT_PC: |
| 2133 | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) |
| 2134 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2135 | break; | break; |
| 2136 | ||
| 2137 | case PT_SC: | case PT_SC: |
| 2138 | if ((ecode[2] != prop->script) == (op == OP_PROP)) | if ((ecode[2] != prop->script) == (op == OP_PROP)) |
| 2139 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2140 | break; | break; |
| 2141 | ||
| 2142 | /* These are specials */ | |
| 2143 | ||
| 2144 | case PT_ALNUM: | |
| 2145 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | |
| 2146 | _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) | |
| 2147 | MRRETURN(MATCH_NOMATCH); | |
| 2148 | break; | |
| 2149 | ||
| 2150 | case PT_SPACE: /* Perl space */ | |
| 2151 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | |
| 2152 | c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) | |
| 2153 | == (op == OP_NOTPROP)) | |
| 2154 | MRRETURN(MATCH_NOMATCH); | |
| 2155 | break; | |
| 2156 | ||
| 2157 | case PT_PXSPACE: /* POSIX space */ | |
| 2158 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | |
| 2159 | c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || | |
| 2160 | c == CHAR_FF || c == CHAR_CR) | |
| 2161 | == (op == OP_NOTPROP)) | |
| 2162 | MRRETURN(MATCH_NOMATCH); | |
| 2163 | break; | |
| 2164 | ||
| 2165 | case PT_WORD: | |
| 2166 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | |
| 2167 | _pcre_ucp_gentype[prop->chartype] == ucp_N || | |
| 2168 | c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) | |
| 2169 | MRRETURN(MATCH_NOMATCH); | |
| 2170 | break; | |
| 2171 | ||
| 2172 | /* This should never occur */ | |
| 2173 | ||
| 2174 | default: | default: |
| 2175 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 2176 | } | } |
| # | Line 1761 for (;;) | Line 2183 for (;;) |
| 2183 | is in the binary; otherwise a compile-time error occurs. */ | is in the binary; otherwise a compile-time error occurs. */ |
| 2184 | ||
| 2185 | case OP_EXTUNI: | case OP_EXTUNI: |
| 2186 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2187 | { | |
| 2188 | SCHECK_PARTIAL(); | |
| 2189 | MRRETURN(MATCH_NOMATCH); | |
| 2190 | } | |
| 2191 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2192 | { | { |
| 2193 | int category = UCD_CATEGORY(c); | int category = UCD_CATEGORY(c); |
| 2194 | if (category == ucp_M) RRETURN(MATCH_NOMATCH); | if (category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 2195 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 2196 | { | { |
| 2197 | int len = 1; | int len = 1; |
| # | Line 1810 for (;;) | Line 2236 for (;;) |
| 2236 | referenced subpattern. */ | referenced subpattern. */ |
| 2237 | ||
| 2238 | if (offset >= offset_top || md->offset_vector[offset] < 0) | if (offset >= offset_top || md->offset_vector[offset] < 0) |
| 2239 | length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1; | length = (md->jscript_compat)? 0 : (int)(md->end_subject - eptr + 1); |
| 2240 | else | else |
| 2241 | length = md->offset_vector[offset+1] - md->offset_vector[offset]; | length = md->offset_vector[offset+1] - md->offset_vector[offset]; |
| 2242 | ||
| # | Line 1841 for (;;) | Line 2267 for (;;) |
| 2267 | break; | break; |
| 2268 | ||
| 2269 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| 2270 | if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); | if (!match_ref(offset, eptr, length, md, ims)) |
| 2271 | { | |
| 2272 | CHECK_PARTIAL(); | |
| 2273 | MRRETURN(MATCH_NOMATCH); | |
| 2274 | } | |
| 2275 | eptr += length; | eptr += length; |
| 2276 | continue; /* With the main loop */ | continue; /* With the main loop */ |
| 2277 | } | } |
| # | Line 1857 for (;;) | Line 2287 for (;;) |
| 2287 | ||
| 2288 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2289 | { | { |
| 2290 | if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); | if (!match_ref(offset, eptr, length, md, ims)) |
| 2291 | { | |
| 2292 | CHECK_PARTIAL(); | |
| 2293 | MRRETURN(MATCH_NOMATCH); | |
| 2294 | } | |
| 2295 | eptr += length; | eptr += length; |
| 2296 | } | } |
| 2297 | ||
| # | Line 1874 for (;;) | Line 2308 for (;;) |
| 2308 | { | { |
| 2309 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14); |
| 2310 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2311 | if (fi >= max || !match_ref(offset, eptr, length, md, ims)) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2312 | RRETURN(MATCH_NOMATCH); | if (!match_ref(offset, eptr, length, md, ims)) |
| 2313 | { | |
| 2314 | CHECK_PARTIAL(); | |
| 2315 | MRRETURN(MATCH_NOMATCH); | |
| 2316 | } | |
| 2317 | eptr += length; | eptr += length; |
| 2318 | } | } |
| 2319 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 1888 for (;;) | Line 2326 for (;;) |
| 2326 | pp = eptr; | pp = eptr; |
| 2327 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2328 | { | { |
| 2329 | if (!match_ref(offset, eptr, length, md, ims)) break; | if (!match_ref(offset, eptr, length, md, ims)) |
| 2330 | { | |
| 2331 | CHECK_PARTIAL(); | |
| 2332 | break; | |
| 2333 | } | |
| 2334 | eptr += length; | eptr += length; |
| 2335 | } | } |
| 2336 | while (eptr >= pp) | while (eptr >= pp) |
| # | Line 1897 for (;;) | Line 2339 for (;;) |
| 2339 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2340 | eptr -= length; | eptr -= length; |
| 2341 | } | } |
| 2342 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2343 | } | } |
| 2344 | } | } |
| 2345 | /* Control never gets here */ | /* Control never gets here */ |
| 2346 | ||
| 2347 | /* Match a bit-mapped character class, possibly repeatedly. This op code is | /* Match a bit-mapped character class, possibly repeatedly. This op code is |
| 2348 | 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, |
| 2349 | 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 1958 for (;;) | Line 2398 for (;;) |
| 2398 | { | { |
| 2399 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2400 | { | { |
| 2401 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2402 | { | |
| 2403 | SCHECK_PARTIAL(); | |
| 2404 | MRRETURN(MATCH_NOMATCH); | |
| 2405 | } | |
| 2406 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2407 | if (c > 255) | if (c > 255) |
| 2408 | { | { |
| 2409 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); |
| 2410 | } | } |
| 2411 | else | else |
| 2412 | { | { |
| 2413 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2414 | } | } |
| 2415 | } | } |
| 2416 | } | } |
| # | Line 1976 for (;;) | Line 2420 for (;;) |
| 2420 | { | { |
| 2421 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2422 | { | { |
| 2423 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2424 | { | |
| 2425 | SCHECK_PARTIAL(); | |
| 2426 | MRRETURN(MATCH_NOMATCH); | |
| 2427 | } | |
| 2428 | c = *eptr++; | c = *eptr++; |
| 2429 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2430 | } | } |
| 2431 | } | } |
| 2432 | ||
| # | Line 2000 for (;;) | Line 2448 for (;;) |
| 2448 | { | { |
| 2449 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16); |
| 2450 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2451 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2452 | if (eptr >= md->end_subject) | |
| 2453 | { | |
| 2454 | SCHECK_PARTIAL(); | |
| 2455 | MRRETURN(MATCH_NOMATCH); | |
| 2456 | } | |
| 2457 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2458 | if (c > 255) | if (c > 255) |
| 2459 | { | { |
| 2460 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); |
| 2461 | } | } |
| 2462 | else | else |
| 2463 | { | { |
| 2464 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2465 | } | } |
| 2466 | } | } |
| 2467 | } | } |
| # | Line 2020 for (;;) | Line 2473 for (;;) |
| 2473 | { | { |
| 2474 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17); |
| 2475 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2476 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2477 | if (eptr >= md->end_subject) | |
| 2478 | { | |
| 2479 | SCHECK_PARTIAL(); | |
| 2480 | MRRETURN(MATCH_NOMATCH); | |
| 2481 | } | |
| 2482 | c = *eptr++; | c = *eptr++; |
| 2483 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2484 | } | } |
| 2485 | } | } |
| 2486 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2041 for (;;) | Line 2499 for (;;) |
| 2499 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2500 | { | { |
| 2501 | int len = 1; | int len = 1; |
| 2502 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2503 | { | |
| 2504 | SCHECK_PARTIAL(); | |
| 2505 | break; | |
| 2506 | } | |
| 2507 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 2508 | if (c > 255) | if (c > 255) |
| 2509 | { | { |
| # | Line 2067 for (;;) | Line 2529 for (;;) |
| 2529 | { | { |
| 2530 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2531 | { | { |
| 2532 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2533 | { | |
| 2534 | SCHECK_PARTIAL(); | |
| 2535 | break; | |
| 2536 | } | |
| 2537 | c = *eptr; | c = *eptr; |
| 2538 | if ((data[c/8] & (1 << (c&7))) == 0) break; | if ((data[c/8] & (1 << (c&7))) == 0) break; |
| 2539 | eptr++; | eptr++; |
| # | Line 2080 for (;;) | Line 2546 for (;;) |
| 2546 | } | } |
| 2547 | } | } |
| 2548 | ||
| 2549 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2550 | } | } |
| 2551 | } | } |
| 2552 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2129 for (;;) | Line 2595 for (;;) |
| 2595 | ||
| 2596 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2597 | { | { |
| 2598 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2599 | { | |
| 2600 | SCHECK_PARTIAL(); | |
| 2601 | MRRETURN(MATCH_NOMATCH); | |
| 2602 | } | |
| 2603 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2604 | if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); |
| 2605 | } | } |
| 2606 | ||
| 2607 | /* 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 2148 for (;;) | Line 2618 for (;;) |
| 2618 | { | { |
| 2619 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20); |
| 2620 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2621 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2622 | if (eptr >= md->end_subject) | |
| 2623 | { | |
| 2624 | SCHECK_PARTIAL(); | |
| 2625 | MRRETURN(MATCH_NOMATCH); | |
| 2626 | } | |
| 2627 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2628 | if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); |
| 2629 | } | } |
| 2630 | /* Control never gets here */ | /* Control never gets here */ |
| 2631 | } | } |
| # | Line 2163 for (;;) | Line 2638 for (;;) |
| 2638 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2639 | { | { |
| 2640 | int len = 1; | int len = 1; |
| 2641 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2642 | { | |
| 2643 | SCHECK_PARTIAL(); | |
| 2644 | break; | |
| 2645 | } | |
| 2646 | GETCHARLENTEST(c, eptr, len); | GETCHARLENTEST(c, eptr, len); |
| 2647 | if (!_pcre_xclass(c, data)) break; | if (!_pcre_xclass(c, data)) break; |
| 2648 | eptr += len; | eptr += len; |
| # | Line 2175 for (;;) | Line 2654 for (;;) |
| 2654 | if (eptr-- == pp) break; /* Stop if tried at original pos */ | if (eptr-- == pp) break; /* Stop if tried at original pos */ |
| 2655 | if (utf8) BACKCHAR(eptr); | if (utf8) BACKCHAR(eptr); |
| 2656 | } | } |
| 2657 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2658 | } | } |
| 2659 | ||
| 2660 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2191 for (;;) | Line 2670 for (;;) |
| 2670 | length = 1; | length = 1; |
| 2671 | ecode++; | ecode++; |
| 2672 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 2673 | if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | if (length > md->end_subject - eptr) |
| 2674 | while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 2675 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | |
| 2676 | MRRETURN(MATCH_NOMATCH); | |
| 2677 | } | |
| 2678 | while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2679 | } | } |
| 2680 | else | else |
| 2681 | #endif | #endif |
| 2682 | ||
| 2683 | /* Non-UTF-8 mode */ | /* Non-UTF-8 mode */ |
| 2684 | { | { |
| 2685 | if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); | if (md->end_subject - eptr < 1) |
| 2686 | if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 2687 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | |
| 2688 | MRRETURN(MATCH_NOMATCH); | |
| 2689 | } | |
| 2690 | if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2691 | ecode += 2; | ecode += 2; |
| 2692 | } | } |
| 2693 | break; | break; |
| # | Line 2215 for (;;) | Line 2702 for (;;) |
| 2702 | ecode++; | ecode++; |
| 2703 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 2704 | ||
| 2705 | if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | if (length > md->end_subject - eptr) |
| 2706 | { | |
| 2707 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | |
| 2708 | MRRETURN(MATCH_NOMATCH); | |
| 2709 | } | |
| 2710 | ||
| 2711 | /* 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 |
| 2712 | can use the fast lookup table. */ | can use the fast lookup table. */ |
| 2713 | ||
| 2714 | if (fc < 128) | if (fc < 128) |
| 2715 | { | { |
| 2716 | if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); |
| 2717 | } | } |
| 2718 | ||
| 2719 | /* Otherwise we must pick up the subject character */ | /* Otherwise we must pick up the subject character */ |
| # | Line 2241 for (;;) | Line 2732 for (;;) |
| 2732 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2733 | if (dc != UCD_OTHERCASE(fc)) | if (dc != UCD_OTHERCASE(fc)) |
| 2734 | #endif | #endif |
| 2735 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2736 | } | } |
| 2737 | } | } |
| 2738 | } | } |
| # | Line 2250 for (;;) | Line 2741 for (;;) |
| 2741 | ||
| 2742 | /* Non-UTF-8 mode */ | /* Non-UTF-8 mode */ |
| 2743 | { | { |
| 2744 | if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); | if (md->end_subject - eptr < 1) |
| 2745 | if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 2746 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | |
| 2747 | MRRETURN(MATCH_NOMATCH); | |
| 2748 | } | |
| 2749 | if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2750 | ecode += 2; | ecode += 2; |
| 2751 | } | } |
| 2752 | break; | break; |
| # | Line 2304 for (;;) | Line 2799 for (;;) |
| 2799 | case OP_MINQUERY: | case OP_MINQUERY: |
| 2800 | c = *ecode++ - OP_STAR; | c = *ecode++ - OP_STAR; |
| 2801 | minimize = (c & 1) != 0; | minimize = (c & 1) != 0; |
| 2802 | ||
| 2803 | min = rep_min[c]; /* Pick up values from tables; */ | min = rep_min[c]; /* Pick up values from tables; */ |
| 2804 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 2805 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2806 | ||
| 2807 | /* 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. */ | ||
| 2808 | ||
| 2809 | REPEATCHAR: | REPEATCHAR: |
| 2810 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 2319 for (;;) | Line 2813 for (;;) |
| 2813 | length = 1; | length = 1; |
| 2814 | charptr = ecode; | charptr = ecode; |
| 2815 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 2816 | ecode += length; | ecode += length; |
| 2817 | ||
| 2818 | /* Handle multibyte character matching specially here. There is | /* Handle multibyte character matching specially here. There is |
| # | Line 2337 for (;;) | Line 2830 for (;;) |
| 2830 | ||
| 2831 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2832 | { | { |
| 2833 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | if (eptr <= md->end_subject - length && |
| 2834 | memcmp(eptr, charptr, length) == 0) eptr += length; | |
| 2835 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2836 | /* Need braces because of following else */ | else if (oclength > 0 && |
| 2837 | else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } | eptr <= md->end_subject - oclength && |
| 2838 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2839 | #endif /* SUPPORT_UCP */ | |
| 2840 | else | else |
| 2841 | { | { |
| 2842 | if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2843 | eptr += oclength; | MRRETURN(MATCH_NOMATCH); |
| 2844 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else { RRETURN(MATCH_NOMATCH); } | ||
| #endif /* SUPPORT_UCP */ | ||
| 2845 | } | } |
| 2846 | ||
| 2847 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2359 for (;;) | Line 2852 for (;;) |
| 2852 | { | { |
| 2853 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22); |
| 2854 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2855 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2856 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | if (eptr <= md->end_subject - length && |
| 2857 | memcmp(eptr, charptr, length) == 0) eptr += length; | |
| 2858 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2859 | /* Need braces because of following else */ | else if (oclength > 0 && |
| 2860 | else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } | eptr <= md->end_subject - oclength && |
| 2861 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2862 | #endif /* SUPPORT_UCP */ | |
| 2863 | else | else |
| 2864 | { | { |
| 2865 | if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2866 | eptr += oclength; | MRRETURN(MATCH_NOMATCH); |
| 2867 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else { RRETURN (MATCH_NOMATCH); } | ||
| #endif /* SUPPORT_UCP */ | ||
| 2868 | } | } |
| 2869 | /* Control never gets here */ | /* Control never gets here */ |
| 2870 | } | } |
| # | Line 2381 for (;;) | Line 2874 for (;;) |
| 2874 | pp = eptr; | pp = eptr; |
| 2875 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2876 | { | { |
| 2877 | if (eptr > md->end_subject - length) break; | if (eptr <= md->end_subject - length && |
| 2878 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, length) == 0) eptr += length; |
| 2879 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2880 | else if (oclength == 0) break; | else if (oclength > 0 && |
| 2881 | eptr <= md->end_subject - oclength && | |
| 2882 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2883 | #endif /* SUPPORT_UCP */ | |
| 2884 | else | else |
| 2885 | { | { |
| 2886 | if (memcmp(eptr, occhars, oclength) != 0) break; | CHECK_PARTIAL(); |
| 2887 | eptr += oclength; | break; |
| 2888 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else break; | ||
| #endif /* SUPPORT_UCP */ | ||
| 2889 | } | } |
| 2890 | ||
| 2891 | if (possessive) continue; | if (possessive) continue; |
| 2892 | ||
| 2893 | for(;;) | for(;;) |
| 2894 | { | { |
| 2895 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23); |
| 2896 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2897 | if (eptr == pp) RRETURN(MATCH_NOMATCH); | if (eptr == pp) { MRRETURN(MATCH_NOMATCH); } |
| 2898 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2899 | eptr--; | eptr--; |
| 2900 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 2901 | #else /* without SUPPORT_UCP */ | #else /* without SUPPORT_UCP */ |
| 2902 | eptr -= length; | eptr -= length; |
| 2903 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 2904 | } | } |
| 2905 | } | } |
| 2906 | /* Control never gets here */ | /* Control never gets here */ |
| 2907 | } | } |
| # | Line 2420 for (;;) | Line 2914 for (;;) |
| 2914 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF8 */ |
| 2915 | ||
| 2916 | /* When not in UTF-8 mode, load a single-byte character. */ | /* When not in UTF-8 mode, load a single-byte character. */ |
| 2917 | { | |
| 2918 | if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | fc = *ecode++; |
| fc = *ecode++; | ||
| } | ||
| 2919 | ||
| 2920 | /* 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 |
| 2921 | 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 2441 for (;;) | Line 2933 for (;;) |
| 2933 | { | { |
| 2934 | fc = md->lcc[fc]; | fc = md->lcc[fc]; |
| 2935 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2936 | if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 2937 | if (eptr >= md->end_subject) | |
| 2938 | { | |
| 2939 | SCHECK_PARTIAL(); | |
| 2940 | MRRETURN(MATCH_NOMATCH); | |
| 2941 | } | |
| 2942 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2943 | } | |
| 2944 | if (min == max) continue; | if (min == max) continue; |
| 2945 | if (minimize) | if (minimize) |
| 2946 | { | { |
| # | Line 2449 for (;;) | Line 2948 for (;;) |
| 2948 | { | { |
| 2949 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24); |
| 2950 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2951 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2952 | fc != md->lcc[*eptr++]) | if (eptr >= md->end_subject) |
| 2953 | RRETURN(MATCH_NOMATCH); | { |
| 2954 | SCHECK_PARTIAL(); | |
| 2955 | MRRETURN(MATCH_NOMATCH); | |
| 2956 | } | |
| 2957 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2958 | } | } |
| 2959 | /* Control never gets here */ | /* Control never gets here */ |
| 2960 | } | } |
| # | Line 2460 for (;;) | Line 2963 for (;;) |
| 2963 | pp = eptr; | pp = eptr; |
| 2964 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2965 | { | { |
| 2966 | if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break; | if (eptr >= md->end_subject) |
| 2967 | { | |
| 2968 | SCHECK_PARTIAL(); | |
| 2969 | break; | |
| 2970 | } | |
| 2971 | if (fc != md->lcc[*eptr]) break; | |
| 2972 | eptr++; | eptr++; |
| 2973 | } | } |
| 2974 | ||
| 2975 | if (possessive) continue; | if (possessive) continue; |
| 2976 | ||
| 2977 | while (eptr >= pp) | while (eptr >= pp) |
| 2978 | { | { |
| 2979 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25); |
| 2980 | eptr--; | eptr--; |
| 2981 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2982 | } | } |
| 2983 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2984 | } | } |
| 2985 | /* Control never gets here */ | /* Control never gets here */ |
| 2986 | } | } |
| # | Line 2479 for (;;) | Line 2989 for (;;) |
| 2989 | ||
| 2990 | else | else |
| 2991 | { | { |
| 2992 | for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH); | for (i = 1; i <= min; i++) |
| 2993 | { | |
| 2994 | if (eptr >= md->end_subject) | |
| 2995 | { | |
| 2996 | SCHECK_PARTIAL(); | |
| 2997 | MRRETURN(MATCH_NOMATCH); | |
| 2998 | } | |
| 2999 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3000 | } | |
| 3001 | ||
| 3002 | if (min == max) continue; | if (min == max) continue; |
| 3003 | ||
| 3004 | if (minimize) | if (minimize) |
| 3005 | { | { |
| 3006 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3007 | { | { |
| 3008 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26); |
| 3009 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3010 | if (fi >= max || eptr >= md->end_subject || fc != *eptr++) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3011 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3012 | { | |
| 3013 | SCHECK_PARTIAL(); | |
| 3014 | MRRETURN(MATCH_NOMATCH); | |
| 3015 | } | |
| 3016 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3017 | } | } |
| 3018 | /* Control never gets here */ | /* Control never gets here */ |
| 3019 | } | } |
| # | Line 2497 for (;;) | Line 3022 for (;;) |
| 3022 | pp = eptr; | pp = eptr; |
| 3023 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3024 | { | { |
| 3025 | if (eptr >= md->end_subject || fc != *eptr) break; | if (eptr >= md->end_subject) |
| 3026 | { | |
| 3027 | SCHECK_PARTIAL(); | |
| 3028 | break; | |
| 3029 | } | |
| 3030 | if (fc != *eptr) break; | |
| 3031 | eptr++; | eptr++; |
| 3032 | } | } |
| 3033 | if (possessive) continue; | if (possessive) continue; |
| 3034 | ||
| 3035 | while (eptr >= pp) | while (eptr >= pp) |
| 3036 | { | { |
| 3037 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27); |
| 3038 | eptr--; | eptr--; |
| 3039 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3040 | } | } |
| 3041 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3042 | } | } |
| 3043 | } | } |
| 3044 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2516 for (;;) | Line 3047 for (;;) |
| 3047 | checking can be multibyte. */ | checking can be multibyte. */ |
| 3048 | ||
| 3049 | case OP_NOT: | case OP_NOT: |
| 3050 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3051 | { | |
| 3052 | SCHECK_PARTIAL(); | |
| 3053 | MRRETURN(MATCH_NOMATCH); | |
| 3054 | } | |
| 3055 | ecode++; | ecode++; |
| 3056 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3057 | if ((ims & PCRE_CASELESS) != 0) | if ((ims & PCRE_CASELESS) != 0) |
| # | Line 2525 for (;;) | Line 3060 for (;;) |
| 3060 | if (c < 256) | if (c < 256) |
| 3061 | #endif | #endif |
| 3062 | c = md->lcc[c]; | c = md->lcc[c]; |
| 3063 | if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH); | if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH); |
| 3064 | } | } |
| 3065 | else | else |
| 3066 | { | { |
| 3067 | if (*ecode++ == c) RRETURN(MATCH_NOMATCH); | if (*ecode++ == c) MRRETURN(MATCH_NOMATCH); |
| 3068 | } | } |
| 3069 | break; | break; |
| 3070 | ||
| # | Line 2593 for (;;) | Line 3128 for (;;) |
| 3128 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 3129 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 3130 | ||
| 3131 | /* 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. */ | ||
| 3132 | ||
| 3133 | REPEATNOTCHAR: | REPEATNOTCHAR: |
| if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 3134 | fc = *ecode++; | fc = *ecode++; |
| 3135 | ||
| 3136 | /* 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 2623 for (;;) | Line 3155 for (;;) |
| 3155 | register unsigned int d; | register unsigned int d; |
| 3156 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3157 | { | { |
| 3158 | if (eptr >= md->end_subject) | |
| 3159 | { | |
| 3160 | SCHECK_PARTIAL(); | |
| 3161 | MRRETURN(MATCH_NOMATCH); | |
| 3162 | } | |
| 3163 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3164 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3165 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3166 | } | } |
| 3167 | } | } |
| 3168 | else | else |
| # | Line 2634 for (;;) | Line 3171 for (;;) |
| 3171 | /* Not UTF-8 mode */ | /* Not UTF-8 mode */ |
| 3172 | { | { |
| 3173 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3174 | if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 3175 | if (eptr >= md->end_subject) | |
| 3176 | { | |
| 3177 | SCHECK_PARTIAL(); | |
| 3178 | MRRETURN(MATCH_NOMATCH); | |
| 3179 | } | |
| 3180 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3181 | } | |
| 3182 | } | } |
| 3183 | ||
| 3184 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2650 for (;;) | Line 3194 for (;;) |
| 3194 | { | { |
| 3195 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28); |
| 3196 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3197 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3198 | if (eptr >= md->end_subject) | |
| 3199 | { | |
| 3200 | SCHECK_PARTIAL(); | |
| 3201 | MRRETURN(MATCH_NOMATCH); | |
| 3202 | } | |
| 3203 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3204 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3205 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3206 | } | } |
| 3207 | } | } |
| 3208 | else | else |
| # | Line 2665 for (;;) | Line 3213 for (;;) |
| 3213 | { | { |
| 3214 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29); |
| 3215 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3216 | if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++]) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3217 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3218 | { | |
| 3219 | SCHECK_PARTIAL(); | |
| 3220 | MRRETURN(MATCH_NOMATCH); | |
| 3221 | } | |
| 3222 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3223 | } | } |
| 3224 | } | } |
| 3225 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2686 for (;;) | Line 3239 for (;;) |
| 3239 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3240 | { | { |
| 3241 | int len = 1; | int len = 1; |
| 3242 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 3243 | { | |
| 3244 | SCHECK_PARTIAL(); | |
| 3245 | break; | |
| 3246 | } | |
| 3247 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3248 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3249 | if (fc == d) break; | if (fc == d) break; |
| # | Line 2707 for (;;) | Line 3264 for (;;) |
| 3264 | { | { |
| 3265 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3266 | { | { |
| 3267 | if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break; | if (eptr >= md->end_subject) |
| 3268 | { | |
| 3269 | SCHECK_PARTIAL(); | |
| 3270 | break; | |
| 3271 | } | |
| 3272 | if (fc == md->lcc[*eptr]) break; | |
| 3273 | eptr++; | eptr++; |
| 3274 | } | } |
| 3275 | if (possessive) continue; | if (possessive) continue; |
| # | Line 2719 for (;;) | Line 3281 for (;;) |
| 3281 | } | } |
| 3282 | } | } |
| 3283 | ||
| 3284 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3285 | } | } |
| 3286 | /* Control never gets here */ | /* Control never gets here */ |
| 3287 | } | } |
| # | Line 2735 for (;;) | Line 3297 for (;;) |
| 3297 | register unsigned int d; | register unsigned int d; |
| 3298 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3299 | { | { |
| 3300 | if (eptr >= md->end_subject) | |
| 3301 | { | |
| 3302 | SCHECK_PARTIAL(); | |
| 3303 | MRRETURN(MATCH_NOMATCH); | |
| 3304 | } | |
| 3305 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3306 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3307 | } | } |
| 3308 | } | } |
| 3309 | else | else |
| # | Line 2744 for (;;) | Line 3311 for (;;) |
| 3311 | /* Not UTF-8 mode */ | /* Not UTF-8 mode */ |
| 3312 | { | { |
| 3313 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3314 | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 3315 | if (eptr >= md->end_subject) | |
| 3316 | { | |
| 3317 | SCHECK_PARTIAL(); | |
| 3318 | MRRETURN(MATCH_NOMATCH); | |
| 3319 | } | |
| 3320 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3321 | } | |
| 3322 | } | } |
| 3323 | ||
| 3324 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2760 for (;;) | Line 3334 for (;;) |
| 3334 | { | { |
| 3335 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32); |
| 3336 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3337 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3338 | if (eptr >= md->end_subject) | |
| 3339 | { | |
| 3340 | SCHECK_PARTIAL(); | |
| 3341 | MRRETURN(MATCH_NOMATCH); | |
| 3342 | } | |
| 3343 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3344 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3345 | } | } |
| 3346 | } | } |
| 3347 | else | else |
| # | Line 2773 for (;;) | Line 3352 for (;;) |
| 3352 | { | { |
| 3353 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33); |
| 3354 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3355 | if (fi >= max || eptr >= md->end_subject || fc == *eptr++) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3356 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3357 | { | |
| 3358 | SCHECK_PARTIAL(); | |
| 3359 | MRRETURN(MATCH_NOMATCH); | |
| 3360 | } | |
| 3361 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3362 | } | } |
| 3363 | } | } |
| 3364 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2794 for (;;) | Line 3378 for (;;) |
| 3378 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3379 | { | { |
| 3380 | int len = 1; | int len = 1; |
| 3381 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 3382 | { | |
| 3383 | SCHECK_PARTIAL(); | |
| 3384 | break; | |
| 3385 | } | |
| 3386 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3387 | if (fc == d) break; | if (fc == d) break; |
| 3388 | eptr += len; | eptr += len; |
| # | Line 2814 for (;;) | Line 3402 for (;;) |
| 3402 | { | { |
| 3403 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3404 | { | { |
| 3405 | if (eptr >= md->end_subject || fc == *eptr) break; | if (eptr >= md->end_subject) |
| 3406 | { | |
| 3407 | SCHECK_PARTIAL(); | |
| 3408 | break; | |
| 3409 | } | |
| 3410 | if (fc == *eptr) break; | |
| 3411 | eptr++; | eptr++; |
| 3412 | } | } |
| 3413 | if (possessive) continue; | if (possessive) continue; |
| # | Line 2826 for (;;) | Line 3419 for (;;) |
| 3419 | } | } |
| 3420 | } | } |
| 3421 | ||
| 3422 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3423 | } | } |
| 3424 | } | } |
| 3425 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2908 for (;;) | Line 3501 for (;;) |
| 3501 | ||
| 3502 | /* First, ensure the minimum number of matches are present. Use inline | /* First, ensure the minimum number of matches are present. Use inline |
| 3503 | 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 |
| 3504 | (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 | ||
| 3505 | 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 |
| 3506 | and single-bytes. */ | and single-bytes. */ |
| 3507 | ||
| if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 3508 | if (min > 0) | if (min > 0) |
| 3509 | { | { |
| 3510 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| # | Line 2923 for (;;) | Line 3513 for (;;) |
| 3513 | switch(prop_type) | switch(prop_type) |
| 3514 | { | { |
| 3515 | case PT_ANY: | case PT_ANY: |
| 3516 | if (prop_fail_result) RRETURN(MATCH_NOMATCH); | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); |
| 3517 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3518 | { | { |
| 3519 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3520 | { | |
| 3521 | SCHECK_PARTIAL(); | |
| 3522 | MRRETURN(MATCH_NOMATCH); | |
| 3523 | } | |
| 3524 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3525 | } | } |
| 3526 | break; | break; |
| # | Line 2934 for (;;) | Line 3528 for (;;) |
| 3528 | case PT_LAMP: | case PT_LAMP: |
| 3529 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3530 | { | { |
| 3531 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3532 | { | |
| 3533 | SCHECK_PARTIAL(); | |
| 3534 | MRRETURN(MATCH_NOMATCH); | |
| 3535 | } | |
| 3536 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3537 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 3538 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 3539 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| 3540 | prop_chartype == ucp_Lt) == prop_fail_result) | prop_chartype == ucp_Lt) == prop_fail_result) |
| 3541 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3542 | } | } |
| 3543 | break; | break; |
| 3544 | ||
| 3545 | case PT_GC: | case PT_GC: |
| 3546 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3547 | { | { |
| 3548 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3549 | { | |
| 3550 | SCHECK_PARTIAL(); | |
| 3551 | MRRETURN(MATCH_NOMATCH); | |
| 3552 | } | |
| 3553 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3554 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3555 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| 3556 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3557 | } | } |
| 3558 | break; | break; |
| 3559 | ||
| 3560 | case PT_PC: | case PT_PC: |
| 3561 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3562 | { | { |
| 3563 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3564 | { | |
| 3565 | SCHECK_PARTIAL(); | |
| 3566 | MRRETURN(MATCH_NOMATCH); | |
| 3567 | } | |
| 3568 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3569 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 3570 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| 3571 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3572 | } | } |
| 3573 | break; | break; |
| 3574 | ||
| 3575 | case PT_SC: | case PT_SC: |
| 3576 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3577 | { | { |
| 3578 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3579 | { | |
| 3580 | SCHECK_PARTIAL(); | |
| 3581 | MRRETURN(MATCH_NOMATCH); | |
| 3582 | } | |
| 3583 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3584 | prop_script = UCD_SCRIPT(c); | prop_script = UCD_SCRIPT(c); |
| 3585 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| 3586 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3587 | } | |
| 3588 | break; | |
| 3589 | ||
| 3590 | case PT_ALNUM: | |
| 3591 | for (i = 1; i <= min; i++) | |
| 3592 | { | |
| 3593 | if (eptr >= md->end_subject) | |
| 3594 | { | |
| 3595 | SCHECK_PARTIAL(); | |
| 3596 | MRRETURN(MATCH_NOMATCH); | |
| 3597 | } | |
| 3598 | GETCHARINCTEST(c, eptr); | |
| 3599 | prop_category = UCD_CATEGORY(c); | |
| 3600 | if ((prop_category == ucp_L || prop_category == ucp_N) | |
| 3601 | == prop_fail_result) | |
| 3602 | MRRETURN(MATCH_NOMATCH); | |
| 3603 | } | |
| 3604 | break; | |
| 3605 | ||
| 3606 | case PT_SPACE: /* Perl space */ | |
| 3607 | for (i = 1; i <= min; i++) | |
| 3608 | { | |
| 3609 | if (eptr >= md->end_subject) | |
| 3610 | { | |
| 3611 | SCHECK_PARTIAL(); | |
| 3612 | MRRETURN(MATCH_NOMATCH); | |
| 3613 | } | |
| 3614 | GETCHARINCTEST(c, eptr); | |
| 3615 | prop_category = UCD_CATEGORY(c); | |
| 3616 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 3617 | c == CHAR_FF || c == CHAR_CR) | |
| 3618 | == prop_fail_result) | |
| 3619 | MRRETURN(MATCH_NOMATCH); | |
| 3620 | } | |
| 3621 | break; | |
| 3622 | ||
| 3623 | case PT_PXSPACE: /* POSIX space */ | |
| 3624 | for (i = 1; i <= min; i++) | |
| 3625 | { | |
| 3626 | if (eptr >= md->end_subject) | |
| 3627 | { | |
| 3628 | SCHECK_PARTIAL(); | |
| 3629 | MRRETURN(MATCH_NOMATCH); | |
| 3630 | } | |
| 3631 | GETCHARINCTEST(c, eptr); | |
| 3632 | prop_category = UCD_CATEGORY(c); | |
| 3633 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 3634 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | |
| 3635 | == prop_fail_result) | |
| 3636 | MRRETURN(MATCH_NOMATCH); | |
| 3637 | } | |
| 3638 | break; | |
| 3639 | ||
| 3640 | case PT_WORD: | |
| 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_L || prop_category == ucp_N || | |
| 3651 | c == CHAR_UNDERSCORE) | |
| 3652 | == prop_fail_result) | |
| 3653 | MRRETURN(MATCH_NOMATCH); | |
| 3654 | } | } |
| 3655 | break; | break; |
| 3656 | ||
| 3657 | /* This should not occur */ | |
| 3658 | ||
| 3659 | default: | default: |
| 3660 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 3661 | } | } |
| # | Line 2989 for (;;) | Line 3668 for (;;) |
| 3668 | { | { |
| 3669 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3670 | { | { |
| 3671 | if (eptr >= md->end_subject) | |
| 3672 | { | |
| 3673 | SCHECK_PARTIAL(); | |
| 3674 | MRRETURN(MATCH_NOMATCH); | |
| 3675 | } | |
| 3676 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3677 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3678 | if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); | if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 3679 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 3680 | { | { |
| 3681 | int len = 1; | int len = 1; |
| 3682 | if (!utf8) c = *eptr; else | if (!utf8) c = *eptr; |
| 3683 | { | else { GETCHARLEN(c, eptr, len); } |
| GETCHARLEN(c, eptr, len); | ||
| } | ||
| 3684 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3685 | if (prop_category != ucp_M) break; | if (prop_category != ucp_M) break; |
| 3686 | eptr += len; | eptr += len; |
| # | Line 3017 for (;;) | Line 3699 for (;;) |
| 3699 | case OP_ANY: | case OP_ANY: |
| 3700 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3701 | { | { |
| 3702 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) | if (eptr >= md->end_subject) |
| 3703 | RRETURN(MATCH_NOMATCH); | { |
| 3704 | SCHECK_PARTIAL(); | |
| 3705 | MRRETURN(MATCH_NOMATCH); | |
| 3706 | } | |
| 3707 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | |
| 3708 | eptr++; | eptr++; |
| 3709 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 3710 | } | } |
| # | Line 3027 for (;;) | Line 3713 for (;;) |
| 3713 | case OP_ALLANY: | case OP_ALLANY: |
| 3714 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3715 | { | { |
| 3716 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3717 | { | |
| 3718 | SCHECK_PARTIAL(); | |
| 3719 | MRRETURN(MATCH_NOMATCH); | |
| 3720 | } | |
| 3721 | eptr++; | eptr++; |
| 3722 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 3723 | } | } |
| 3724 | break; | break; |
| 3725 | ||
| 3726 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 3727 | if (eptr > md->end_subject - min) MRRETURN(MATCH_NOMATCH); | |
| 3728 | eptr += min; | eptr += min; |
| 3729 | break; | break; |
| 3730 | ||
| 3731 | case OP_ANYNL: | case OP_ANYNL: |
| 3732 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3733 | { | { |
| 3734 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3735 | { | |
| 3736 | SCHECK_PARTIAL(); | |
| 3737 | MRRETURN(MATCH_NOMATCH); | |
| 3738 | } | |
| 3739 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3740 | switch(c) | switch(c) |
| 3741 | { | { |
| 3742 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3743 | case 0x000d: | case 0x000d: |
| 3744 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 3745 | break; | break; |
| # | Line 3057 for (;;) | Line 3752 for (;;) |
| 3752 | case 0x0085: | case 0x0085: |
| 3753 | case 0x2028: | case 0x2028: |
| 3754 | case 0x2029: | case 0x2029: |
| 3755 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 3756 | break; | break; |
| 3757 | } | } |
| 3758 | } | } |
| # | Line 3066 for (;;) | Line 3761 for (;;) |
| 3761 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 3762 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3763 | { | { |
| 3764 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3765 | { | |
| 3766 | SCHECK_PARTIAL(); | |
| 3767 | MRRETURN(MATCH_NOMATCH); | |
| 3768 | } | |
| 3769 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3770 | switch(c) | switch(c) |
| 3771 | { | { |
| # | Line 3090 for (;;) | Line 3789 for (;;) |
| 3789 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 3790 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 3791 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 3792 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3793 | } | } |
| 3794 | } | } |
| 3795 | break; | break; |
| # | Line 3098 for (;;) | Line 3797 for (;;) |
| 3797 | case OP_HSPACE: | case OP_HSPACE: |
| 3798 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3799 | { | { |
| 3800 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3801 | { | |
| 3802 | SCHECK_PARTIAL(); | |
| 3803 | MRRETURN(MATCH_NOMATCH); | |
| 3804 | } | |
| 3805 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3806 | switch(c) | switch(c) |
| 3807 | { | { |
| 3808 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3809 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 3810 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 3811 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3130 for (;;) | Line 3833 for (;;) |
| 3833 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 3834 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3835 | { | { |
| 3836 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3837 | { | |
| 3838 | SCHECK_PARTIAL(); | |
| 3839 | MRRETURN(MATCH_NOMATCH); | |
| 3840 | } | |
| 3841 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3842 | switch(c) | switch(c) |
| 3843 | { | { |
| # | Line 3142 for (;;) | Line 3849 for (;;) |
| 3849 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 3850 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 3851 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 3852 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3853 | } | } |
| 3854 | } | } |
| 3855 | break; | break; |
| # | Line 3150 for (;;) | Line 3857 for (;;) |
| 3857 | case OP_VSPACE: | case OP_VSPACE: |
| 3858 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3859 | { | { |
| 3860 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3861 | { | |
| 3862 | SCHECK_PARTIAL(); | |
| 3863 | MRRETURN(MATCH_NOMATCH); | |
| 3864 | } | |
| 3865 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3866 | switch(c) | switch(c) |
| 3867 | { | { |
| 3868 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3869 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 3870 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 3871 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3170 for (;;) | Line 3881 for (;;) |
| 3881 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 3882 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3883 | { | { |
| 3884 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3885 | { | |
| 3886 | SCHECK_PARTIAL(); | |
| 3887 | MRRETURN(MATCH_NOMATCH); | |
| 3888 | } | |
| 3889 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3890 | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) |
| 3891 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3892 | } | } |
| 3893 | break; | break; |
| 3894 | ||
| 3895 | case OP_DIGIT: | case OP_DIGIT: |
| 3896 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3897 | { | { |
| 3898 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3899 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | { |
| 3900 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3901 | MRRETURN(MATCH_NOMATCH); | |
| 3902 | } | |
| 3903 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | |
| 3904 | MRRETURN(MATCH_NOMATCH); | |
| 3905 | /* 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 */ |
| 3906 | } | } |
| 3907 | break; | break; |
| # | Line 3190 for (;;) | Line 3909 for (;;) |
| 3909 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 3910 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3911 | { | { |
| 3912 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3913 | (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0)) | { |
| 3914 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3915 | MRRETURN(MATCH_NOMATCH); | |
| 3916 | } | |
| 3917 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) | |
| 3918 | MRRETURN(MATCH_NOMATCH); | |
| 3919 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); |
| 3920 | } | } |
| 3921 | break; | break; |
| # | Line 3200 for (;;) | Line 3923 for (;;) |
| 3923 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 3924 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3925 | { | { |
| 3926 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3927 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | { |
| 3928 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3929 | MRRETURN(MATCH_NOMATCH); | |
| 3930 | } | |
| 3931 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | |
| 3932 | MRRETURN(MATCH_NOMATCH); | |
| 3933 | /* 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 */ |
| 3934 | } | } |
| 3935 | break; | break; |
| # | Line 3210 for (;;) | Line 3937 for (;;) |
| 3937 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 3938 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3939 | { | { |
| 3940 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3941 | (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0)) | { |
| 3942 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3943 | MRRETURN(MATCH_NOMATCH); | |
| 3944 | } | |
| 3945 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) | |
| 3946 | MRRETURN(MATCH_NOMATCH); | |
| 3947 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); |
| 3948 | } | } |
| 3949 | break; | break; |
| # | Line 3220 for (;;) | Line 3951 for (;;) |
| 3951 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 3952 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3953 | { | { |
| 3954 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3955 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | { |
| 3956 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3957 | MRRETURN(MATCH_NOMATCH); | |
| 3958 | } | |
| 3959 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | |
| 3960 | MRRETURN(MATCH_NOMATCH); | |
| 3961 | /* 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 */ |
| 3962 | } | } |
| 3963 | break; | break; |
| # | Line 3235 for (;;) | Line 3970 for (;;) |
| 3970 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF8 */ |
| 3971 | ||
| 3972 | /* 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 |
| 3973 | 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. */ | ||
| 3974 | ||
| 3975 | switch(ctype) | switch(ctype) |
| 3976 | { | { |
| 3977 | case OP_ANY: | case OP_ANY: |
| 3978 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3979 | { | { |
| 3980 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3981 | { | |
| 3982 | SCHECK_PARTIAL(); | |
| 3983 | MRRETURN(MATCH_NOMATCH); | |
| 3984 | } | |
| 3985 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | |
| 3986 | eptr++; | eptr++; |
| 3987 | } | } |
| 3988 | break; | break; |
| 3989 | ||
| 3990 | case OP_ALLANY: | case OP_ALLANY: |
| 3991 | if (eptr > md->end_subject - min) | |
| 3992 | { | |
| 3993 | SCHECK_PARTIAL(); | |
| 3994 | MRRETURN(MATCH_NOMATCH); | |
| 3995 | } | |
| 3996 | eptr += min; | eptr += min; |
| 3997 | break; | break; |
| 3998 | ||
| 3999 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 4000 | if (eptr > md->end_subject - min) | |
| 4001 | { | |
| 4002 | SCHECK_PARTIAL(); | |
| 4003 | MRRETURN(MATCH_NOMATCH); | |
| 4004 | } | |
| 4005 | eptr += min; | eptr += min; |
| 4006 | break; | break; |
| 4007 | ||
| /* Because of the CRLF case, we can't assume the minimum number of | ||
| bytes are present in this case. */ | ||
| 4008 | case OP_ANYNL: | case OP_ANYNL: |
| 4009 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4010 | { | { |
| 4011 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4012 | { | |
| 4013 | SCHECK_PARTIAL(); | |
| 4014 | MRRETURN(MATCH_NOMATCH); | |
| 4015 | } | |
| 4016 | switch(*eptr++) | switch(*eptr++) |
| 4017 | { | { |
| 4018 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4019 | case 0x000d: | case 0x000d: |
| 4020 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4021 | break; | break; |
| # | Line 3275 for (;;) | Line 4025 for (;;) |
| 4025 | case 0x000b: | case 0x000b: |
| 4026 | case 0x000c: | case 0x000c: |
| 4027 | case 0x0085: | case 0x0085: |
| 4028 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4029 | break; | break; |
| 4030 | } | } |
| 4031 | } | } |
| # | Line 3284 for (;;) | Line 4034 for (;;) |
| 4034 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 4035 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4036 | { | { |
| 4037 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4038 | { | |
| 4039 | SCHECK_PARTIAL(); | |
| 4040 | MRRETURN(MATCH_NOMATCH); | |
| 4041 | } | |
| 4042 | switch(*eptr++) | switch(*eptr++) |
| 4043 | { | { |
| 4044 | default: break; | default: break; |
| 4045 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4046 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4047 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4048 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4049 | } | } |
| 4050 | } | } |
| 4051 | break; | break; |
| # | Line 3299 for (;;) | Line 4053 for (;;) |
| 4053 | case OP_HSPACE: | case OP_HSPACE: |
| 4054 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4055 | { | { |
| 4056 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4057 | { | |
| 4058 | SCHECK_PARTIAL(); | |
| 4059 | MRRETURN(MATCH_NOMATCH); | |
| 4060 | } | |
| 4061 | switch(*eptr++) | switch(*eptr++) |
| 4062 | { | { |
| 4063 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4064 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4065 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4066 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3314 for (;;) | Line 4072 for (;;) |
| 4072 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 4073 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4074 | { | { |
| 4075 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4076 | { | |
| 4077 | SCHECK_PARTIAL(); | |
| 4078 | MRRETURN(MATCH_NOMATCH); | |
| 4079 | } | |
| 4080 | switch(*eptr++) | switch(*eptr++) |
| 4081 | { | { |
| 4082 | default: break; | default: break; |
| # | Line 3323 for (;;) | Line 4085 for (;;) |
| 4085 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4086 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4087 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4088 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4089 | } | } |
| 4090 | } | } |
| 4091 | break; | break; |
| # | Line 3331 for (;;) | Line 4093 for (;;) |
| 4093 | case OP_VSPACE: | case OP_VSPACE: |
| 4094 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4095 | { | { |
| 4096 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4097 | { | |
| 4098 | SCHECK_PARTIAL(); | |
| 4099 | MRRETURN(MATCH_NOMATCH); | |
| 4100 | } | |
| 4101 | switch(*eptr++) | switch(*eptr++) |
| 4102 | { | { |
| 4103 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4104 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4105 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4106 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3347 for (;;) | Line 4113 for (;;) |
| 4113 | ||
| 4114 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4115 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4116 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); | { |
| 4117 | if (eptr >= md->end_subject) | |
| 4118 | { | |
| 4119 | SCHECK_PARTIAL(); | |
| 4120 | MRRETURN(MATCH_NOMATCH); | |
| 4121 | } | |
| 4122 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); | |
| 4123 | } | |
| 4124 | break; | break; |
| 4125 | ||
| 4126 | case OP_DIGIT: | case OP_DIGIT: |
| 4127 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4128 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); | { |
| 4129 | if (eptr >= md->end_subject) | |
| 4130 | { | |
| 4131 | SCHECK_PARTIAL(); | |
| 4132 | MRRETURN(MATCH_NOMATCH); | |
| 4133 | } | |
| 4134 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); | |
| 4135 | } | |
| 4136 | break; | break; |
| 4137 | ||
| 4138 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4139 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4140 | if ((md->ctypes[*eptr++] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); | { |
| 4141 | if (eptr >= md->end_subject) | |
| 4142 | { | |
| 4143 | SCHECK_PARTIAL(); | |
| 4144 | MRRETURN(MATCH_NOMATCH); | |
| 4145 | } | |
| 4146 | if ((md->ctypes[*eptr++] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); | |
| 4147 | } | |
| 4148 | break; | break; |
| 4149 | ||
| 4150 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4151 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4152 | if ((md->ctypes[*eptr++] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); | { |
| 4153 | if (eptr >= md->end_subject) | |
| 4154 | { | |
| 4155 | SCHECK_PARTIAL(); | |
| 4156 | MRRETURN(MATCH_NOMATCH); | |
| 4157 | } | |
| 4158 | if ((md->ctypes[*eptr++] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); | |
| 4159 | } | |
| 4160 | break; | break; |
| 4161 | ||
| 4162 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4163 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4164 | { | |
| 4165 | if (eptr >= md->end_subject) | |
| 4166 | { | |
| 4167 | SCHECK_PARTIAL(); | |
| 4168 | MRRETURN(MATCH_NOMATCH); | |
| 4169 | } | |
| 4170 | if ((md->ctypes[*eptr++] & ctype_word) != 0) | if ((md->ctypes[*eptr++] & ctype_word) != 0) |
| 4171 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4172 | } | |
| 4173 | break; | break; |
| 4174 | ||
| 4175 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4176 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4177 | { | |
| 4178 | if (eptr >= md->end_subject) | |
| 4179 | { | |
| 4180 | SCHECK_PARTIAL(); | |
| 4181 | MRRETURN(MATCH_NOMATCH); | |
| 4182 | } | |
| 4183 | if ((md->ctypes[*eptr++] & ctype_word) == 0) | if ((md->ctypes[*eptr++] & ctype_word) == 0) |
| 4184 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4185 | } | |
| 4186 | break; | break; |
| 4187 | ||
| 4188 | default: | default: |
| # | Line 3402 for (;;) | Line 4210 for (;;) |
| 4210 | { | { |
| 4211 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36); |
| 4212 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4213 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4214 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4215 | if (prop_fail_result) RRETURN(MATCH_NOMATCH); | { |
| 4216 | SCHECK_PARTIAL(); | |
| 4217 | MRRETURN(MATCH_NOMATCH); | |
| 4218 | } | |
| 4219 | GETCHARINCTEST(c, eptr); | |
| 4220 | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); | |
| 4221 | } | } |
| 4222 | /* Control never gets here */ | /* Control never gets here */ |
| 4223 | ||
| # | Line 3413 for (;;) | Line 4226 for (;;) |
| 4226 | { | { |
| 4227 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37); |
| 4228 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4229 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4230 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4231 | { | |
| 4232 | SCHECK_PARTIAL(); | |
| 4233 | MRRETURN(MATCH_NOMATCH); | |
| 4234 | } | |
| 4235 | GETCHARINCTEST(c, eptr); | |
| 4236 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4237 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 4238 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| 4239 | prop_chartype == ucp_Lt) == prop_fail_result) | prop_chartype == ucp_Lt) == prop_fail_result) |
| 4240 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4241 | } | } |
| 4242 | /* Control never gets here */ | /* Control never gets here */ |
| 4243 | ||
| # | Line 3428 for (;;) | Line 4246 for (;;) |
| 4246 | { | { |
| 4247 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38); |
| 4248 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4249 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4250 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4251 | { | |
| 4252 | SCHECK_PARTIAL(); | |
| 4253 | MRRETURN(MATCH_NOMATCH); | |
| 4254 | } | |
| 4255 | GETCHARINCTEST(c, eptr); | |
| 4256 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4257 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| 4258 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4259 | } | } |
| 4260 | /* Control never gets here */ | /* Control never gets here */ |
| 4261 | ||
| # | Line 3441 for (;;) | Line 4264 for (;;) |
| 4264 | { | { |
| 4265 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39); |
| 4266 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4267 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4268 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4269 | { | |
| 4270 | SCHECK_PARTIAL(); | |
| 4271 | MRRETURN(MATCH_NOMATCH); | |
| 4272 | } | |
| 4273 | GETCHARINCTEST(c, eptr); | |
| 4274 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4275 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| 4276 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4277 | } | } |
| 4278 | /* Control never gets here */ | /* Control never gets here */ |
| 4279 | ||
| # | Line 3454 for (;;) | Line 4282 for (;;) |
| 4282 | { | { |
| 4283 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40); |
| 4284 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4285 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4286 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4287 | { | |
| 4288 | SCHECK_PARTIAL(); | |
| 4289 | MRRETURN(MATCH_NOMATCH); | |
| 4290 | } | |
| 4291 | GETCHARINCTEST(c, eptr); | |
| 4292 | prop_script = UCD_SCRIPT(c); | prop_script = UCD_SCRIPT(c); |
| 4293 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| 4294 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4295 | } | |
| 4296 | /* Control never gets here */ | |
| 4297 | ||
| 4298 | case PT_ALNUM: | |
| 4299 | for (fi = min;; fi++) | |
| 4300 | { | |
| 4301 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM59); | |
| 4302 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4303 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4304 | if (eptr >= md->end_subject) | |
| 4305 | { | |
| 4306 | SCHECK_PARTIAL(); | |
| 4307 | MRRETURN(MATCH_NOMATCH); | |
| 4308 | } | |
| 4309 | GETCHARINCTEST(c, eptr); | |
| 4310 | prop_category = UCD_CATEGORY(c); | |
| 4311 | if ((prop_category == ucp_L || prop_category == ucp_N) | |
| 4312 | == prop_fail_result) | |
| 4313 | MRRETURN(MATCH_NOMATCH); | |
| 4314 | } | } |
| 4315 | /* Control never gets here */ | /* Control never gets here */ |
| 4316 | ||
| 4317 | case PT_SPACE: /* Perl space */ | |
| 4318 | for (fi = min;; fi++) | |
| 4319 | { | |
| 4320 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM60); | |
| 4321 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4322 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4323 | if (eptr >= md->end_subject) | |
| 4324 | { | |
| 4325 | SCHECK_PARTIAL(); | |
| 4326 | MRRETURN(MATCH_NOMATCH); | |
| 4327 | } | |
| 4328 | GETCHARINCTEST(c, eptr); | |
| 4329 | prop_category = UCD_CATEGORY(c); | |
| 4330 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 4331 | c == CHAR_FF || c == CHAR_CR) | |
| 4332 | == prop_fail_result) | |
| 4333 | MRRETURN(MATCH_NOMATCH); | |
| 4334 | } | |
| 4335 | /* Control never gets here */ | |
| 4336 | ||
| 4337 | case PT_PXSPACE: /* POSIX space */ | |
| 4338 | for (fi = min;; fi++) | |
| 4339 | { | |
| 4340 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM61); | |
| 4341 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4342 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4343 | if (eptr >= md->end_subject) | |
| 4344 | { | |
| 4345 | SCHECK_PARTIAL(); | |
| 4346 | MRRETURN(MATCH_NOMATCH); | |
| 4347 | } | |
| 4348 | GETCHARINCTEST(c, eptr); | |
| 4349 | prop_category = UCD_CATEGORY(c); | |
| 4350 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 4351 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | |
| 4352 | == prop_fail_result) | |
| 4353 | MRRETURN(MATCH_NOMATCH); | |
| 4354 | } | |
| 4355 | /* Control never gets here */ | |
| 4356 | ||
| 4357 | case PT_WORD: | |
| 4358 | for (fi = min;; fi++) | |
| 4359 | { | |
| 4360 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM62); | |
| 4361 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4362 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4363 | if (eptr >= md->end_subject) | |
| 4364 | { | |
| 4365 | SCHECK_PARTIAL(); | |
| 4366 | MRRETURN(MATCH_NOMATCH); | |
| 4367 | } | |
| 4368 | GETCHARINCTEST(c, eptr); | |
| 4369 | prop_category = UCD_CATEGORY(c); | |
| 4370 | if ((prop_category == ucp_L || | |
| 4371 | prop_category == ucp_N || | |
| 4372 | c == CHAR_UNDERSCORE) | |
| 4373 | == prop_fail_result) | |
| 4374 | MRRETURN(MATCH_NOMATCH); | |
| 4375 | } | |
| 4376 | /* Control never gets here */ | |
| 4377 | ||
| 4378 | /* This should never occur */ | |
| 4379 | ||
| 4380 | default: | default: |
| 4381 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 4382 | } | } |
| # | Line 3476 for (;;) | Line 4391 for (;;) |
| 4391 | { | { |
| 4392 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41); |
| 4393 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4394 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4395 | if (eptr >= md->end_subject) | |
| 4396 | { | |
| 4397 | SCHECK_PARTIAL(); | |
| 4398 | MRRETURN(MATCH_NOMATCH); | |
| 4399 | } | |
| 4400 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4401 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4402 | if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); | if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 4403 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 4404 | { | { |
| 4405 | int len = 1; | int len = 1; |
| 4406 | if (!utf8) c = *eptr; else | if (!utf8) c = *eptr; |
| 4407 | { | else { GETCHARLEN(c, eptr, len); } |
| GETCHARLEN(c, eptr, len); | ||
| } | ||
| 4408 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4409 | if (prop_category != ucp_M) break; | if (prop_category != ucp_M) break; |
| 4410 | eptr += len; | eptr += len; |
| # | Line 3505 for (;;) | Line 4423 for (;;) |
| 4423 | { | { |
| 4424 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42); |
| 4425 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4426 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4427 | (ctype == OP_ANY && IS_NEWLINE(eptr))) | if (eptr >= md->end_subject) |
| 4428 | RRETURN(MATCH_NOMATCH); | { |
| 4429 | SCHECK_PARTIAL(); | |
| 4430 | MRRETURN(MATCH_NOMATCH); | |
| 4431 | } | |
| 4432 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | |
| 4433 | MRRETURN(MATCH_NOMATCH); | |
| 4434 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4435 | switch(ctype) | switch(ctype) |
| 4436 | { | { |
| # | Line 3520 for (;;) | Line 4442 for (;;) |
| 4442 | case OP_ANYNL: | case OP_ANYNL: |
| 4443 | switch(c) | switch(c) |
| 4444 | { | { |
| 4445 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4446 | case 0x000d: | case 0x000d: |
| 4447 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4448 | break; | break; |
| # | Line 3532 for (;;) | Line 4454 for (;;) |
| 4454 | case 0x0085: | case 0x0085: |
| 4455 | case 0x2028: | case 0x2028: |
| 4456 | case 0x2029: | case 0x2029: |
| 4457 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4458 | break; | break; |
| 4459 | } | } |
| 4460 | break; | break; |
| # | Line 3560 for (;;) | Line 4482 for (;;) |
| 4482 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 4483 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 4484 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 4485 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4486 | } | } |
| 4487 | break; | break; |
| 4488 | ||
| 4489 | case OP_HSPACE: | case OP_HSPACE: |
| 4490 | switch(c) | switch(c) |
| 4491 | { | { |
| 4492 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4493 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4494 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4495 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3602 for (;;) | Line 4524 for (;;) |
| 4524 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4525 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 4526 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 4527 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4528 | } | } |
| 4529 | break; | break; |
| 4530 | ||
| 4531 | case OP_VSPACE: | case OP_VSPACE: |
| 4532 | switch(c) | switch(c) |
| 4533 | { | { |
| 4534 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4535 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4536 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4537 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3623 for (;;) | Line 4545 for (;;) |
| 4545 | ||
| 4546 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4547 | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) |
| 4548 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4549 | break; | break; |
| 4550 | ||
| 4551 | case OP_DIGIT: | case OP_DIGIT: |
| 4552 | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) |
| 4553 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4554 | break; | break; |
| 4555 | ||
| 4556 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4557 | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) |
| 4558 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4559 | break; | break; |
| 4560 | ||
| 4561 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4562 | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) |
| 4563 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4564 | break; | break; |
| 4565 | ||
| 4566 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4567 | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) |
| 4568 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4569 | break; | break; |
| 4570 | ||
| 4571 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4572 | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) |
| 4573 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4574 | break; | break; |
| 4575 | ||
| 4576 | default: | default: |
| # | Line 3664 for (;;) | Line 4586 for (;;) |
| 4586 | { | { |
| 4587 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43); |
| 4588 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4589 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4590 | (ctype == OP_ANY && IS_NEWLINE(eptr))) | if (eptr >= md->end_subject) |
| 4591 | RRETURN(MATCH_NOMATCH); | { |
| 4592 | SCHECK_PARTIAL(); | |
| 4593 | MRRETURN(MATCH_NOMATCH); | |
| 4594 | } | |
| 4595 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | |
| 4596 | MRRETURN(MATCH_NOMATCH); | |
| 4597 | c = *eptr++; | c = *eptr++; |
| 4598 | switch(ctype) | switch(ctype) |
| 4599 | { | { |
| # | Line 3679 for (;;) | Line 4605 for (;;) |
| 4605 | case OP_ANYNL: | case OP_ANYNL: |
| 4606 | switch(c) | switch(c) |
| 4607 | { | { |
| 4608 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4609 | case 0x000d: | case 0x000d: |
| 4610 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4611 | break; | break; |
| # | Line 3690 for (;;) | Line 4616 for (;;) |
| 4616 | case 0x000b: | case 0x000b: |
| 4617 | case 0x000c: | case 0x000c: |
| 4618 | case 0x0085: | case 0x0085: |
| 4619 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4620 | break; | break; |
| 4621 | } | } |
| 4622 | break; | break; |
| # | Line 3702 for (;;) | Line 4628 for (;;) |
| 4628 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4629 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4630 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4631 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4632 | } | } |
| 4633 | break; | break; |
| 4634 | ||
| 4635 | case OP_HSPACE: | case OP_HSPACE: |
| 4636 | switch(c) | switch(c) |
| 4637 | { | { |
| 4638 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4639 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4640 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4641 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3726 for (;;) | Line 4652 for (;;) |
| 4652 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4653 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4654 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4655 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4656 | } | } |
| 4657 | break; | break; |
| 4658 | ||
| 4659 | case OP_VSPACE: | case OP_VSPACE: |
| 4660 | switch(c) | switch(c) |
| 4661 | { | { |
| 4662 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4663 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4664 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4665 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3744 for (;;) | Line 4670 for (;;) |
| 4670 | break; | break; |
| 4671 | ||
| 4672 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4673 | if ((md->ctypes[c] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); |
| 4674 | break; | break; |
| 4675 | ||
| 4676 | case OP_DIGIT: | case OP_DIGIT: |
| 4677 | if ((md->ctypes[c] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); |
| 4678 | break; | break; |
| 4679 | ||
| 4680 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4681 | if ((md->ctypes[c] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); |
| 4682 | break; | break; |
| 4683 | ||
| 4684 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4685 | if ((md->ctypes[c] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); |
| 4686 | break; | break; |
| 4687 | ||
| 4688 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4689 | if ((md->ctypes[c] & ctype_word) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_word) != 0) MRRETURN(MATCH_NOMATCH); |
| 4690 | break; | break; |
| 4691 | ||
| 4692 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4693 | if ((md->ctypes[c] & ctype_word) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_word) == 0) MRRETURN(MATCH_NOMATCH); |
| 4694 | break; | break; |
| 4695 | ||
| 4696 | default: | default: |
| # | Line 3792 for (;;) | Line 4718 for (;;) |
| 4718 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4719 | { | { |
| 4720 | int len = 1; | int len = 1; |
| 4721 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4722 | GETCHARLEN(c, eptr, len); | { |
| 4723 | SCHECK_PARTIAL(); | |
| 4724 | break; | |
| 4725 | } | |
| 4726 | GETCHARLENTEST(c, eptr, len); | |
| 4727 | if (prop_fail_result) break; | if (prop_fail_result) break; |
| 4728 | eptr+= len; | eptr+= len; |
| 4729 | } | } |
| # | Line 3803 for (;;) | Line 4733 for (;;) |
| 4733 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4734 | { | { |
| 4735 | int len = 1; | int len = 1; |
| 4736 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4737 | GETCHARLEN(c, eptr, len); | { |
| 4738 | SCHECK_PARTIAL(); | |
| 4739 | break; | |
| 4740 | } | |
| 4741 | GETCHARLENTEST(c, eptr, len); | |
| 4742 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4743 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 4744 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| # | Line 3818 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; |
| 4755 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4756 | GETCHARLEN(c, eptr, len); | { |
| 4757 | SCHECK_PARTIAL(); | |
| 4758 | break; | |
| 4759 | } | |
| 4760 | GETCHARLENTEST(c, eptr, len); | |
| 4761 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4762 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| 4763 | break; | break; |
| # | Line 3831 for (;;) | Line 4769 for (;;) |
| 4769 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4770 | { | { |
| 4771 | int len = 1; | int len = 1; |
| 4772 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4773 | GETCHARLEN(c, eptr, len); | { |
| 4774 | SCHECK_PARTIAL(); | |
| 4775 | break; | |
| 4776 | } | |
| 4777 | GETCHARLENTEST(c, eptr, len); | |
| 4778 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4779 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| 4780 | break; | break; |
| # | Line 3844 for (;;) | Line 4786 for (;;) |
| 4786 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4787 | { | { |
| 4788 | int len = 1; | int len = 1; |
| 4789 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4790 | GETCHARLEN(c, eptr, len); | { |
| 4791 | SCHECK_PARTIAL(); | |
| 4792 | break; | |
| 4793 | } | |
| 4794 | GETCHARLENTEST(c, eptr, len); | |
| 4795 | prop_script = UCD_SCRIPT(c); | prop_script = UCD_SCRIPT(c); |
| 4796 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| 4797 | break; | break; |
| 4798 | eptr+= len; | eptr+= len; |
| 4799 | } | } |
| 4800 | break; | break; |
| 4801 | ||
| 4802 | case PT_ALNUM: | |
| 4803 | for (i = min; i < max; i++) | |
| 4804 | { | |
| 4805 | int len = 1; | |
| 4806 | if (eptr >= md->end_subject) | |
| 4807 | { | |
| 4808 | SCHECK_PARTIAL(); | |
| 4809 | break; | |
| 4810 | } | |
| 4811 | GETCHARLENTEST(c, eptr, len); | |
| 4812 | prop_category = UCD_CATEGORY(c); | |
| 4813 | if ((prop_category == ucp_L || prop_category == ucp_N) | |
| 4814 | == prop_fail_result) | |
| 4815 | break; | |
| 4816 | eptr+= len; | |
| 4817 | } | |
| 4818 | break; | |
| 4819 | ||
| 4820 | case PT_SPACE: /* Perl space */ | |
| 4821 | for (i = min; i < max; i++) | |
| 4822 | { | |
| 4823 | int len = 1; | |
| 4824 | if (eptr >= md->end_subject) | |
| 4825 | { | |
| 4826 | SCHECK_PARTIAL(); | |
| 4827 | break; | |
| 4828 | } | |
| 4829 | GETCHARLENTEST(c, eptr, len); | |
| 4830 | prop_category = UCD_CATEGORY(c); | |
| 4831 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 4832 | c == CHAR_FF || c == CHAR_CR) | |
| 4833 | == prop_fail_result) | |
| 4834 | break; | |
| 4835 | eptr+= len; | |
| 4836 | } | |
| 4837 | break; | |
| 4838 | ||
| 4839 | case PT_PXSPACE: /* POSIX space */ | |
| 4840 | for (i = min; i < max; i++) | |
| 4841 | { | |
| 4842 | int len = 1; | |
| 4843 | if (eptr >= md->end_subject) | |
| 4844 | { | |
| 4845 | SCHECK_PARTIAL(); | |
| 4846 | break; | |
| 4847 | } | |
| 4848 | GETCHARLENTEST(c, eptr, len); | |
| 4849 | prop_category = UCD_CATEGORY(c); | |
| 4850 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 4851 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | |
| 4852 | == prop_fail_result) | |
| 4853 | break; | |
| 4854 | eptr+= len; | |
| 4855 | } | |
| 4856 | break; | |
| 4857 | ||
| 4858 | case PT_WORD: | |
| 4859 | for (i = min; i < max; i++) | |
| 4860 | { | |
| 4861 | int len = 1; | |
| 4862 | if (eptr >= md->end_subject) | |
| 4863 | { | |
| 4864 | SCHECK_PARTIAL(); | |
| 4865 | break; | |
| 4866 | } | |
| 4867 | GETCHARLENTEST(c, eptr, len); | |
| 4868 | prop_category = UCD_CATEGORY(c); | |
| 4869 | if ((prop_category == ucp_L || prop_category == ucp_N || | |
| 4870 | c == CHAR_UNDERSCORE) == prop_fail_result) | |
| 4871 | break; | |
| 4872 | eptr+= len; | |
| 4873 | } | |
| 4874 | break; | |
| 4875 | ||
| 4876 | default: | |
| 4877 | RRETURN(PCRE_ERROR_INTERNAL); | |
| 4878 | } | } |
| 4879 | ||
| 4880 | /* eptr is now past the end of the maximum run */ | /* eptr is now past the end of the maximum run */ |
| # | Line 3873 for (;;) | Line 4896 for (;;) |
| 4896 | { | { |
| 4897 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4898 | { | { |
| 4899 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4900 | { | |
| 4901 | SCHECK_PARTIAL(); | |
| 4902 | break; | |
| 4903 | } | |
| 4904 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4905 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4906 | if (prop_category == ucp_M) break; | if (prop_category == ucp_M) break; |
| # | Line 3893 for (;;) | Line 4920 for (;;) |
| 4920 | /* eptr is now past the end of the maximum run */ | /* eptr is now past the end of the maximum run */ |
| 4921 | ||
| 4922 | if (possessive) continue; | if (possessive) continue; |
| 4923 | ||
| 4924 | for(;;) | for(;;) |
| 4925 | { | { |
| 4926 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45); |
| # | Line 3928 for (;;) | Line 4956 for (;;) |
| 4956 | { | { |
| 4957 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4958 | { | { |
| 4959 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break; | if (eptr >= md->end_subject) |
| 4960 | { | |
| 4961 | SCHECK_PARTIAL(); | |
| 4962 | break; | |
| 4963 | } | |
| 4964 | if (IS_NEWLINE(eptr)) break; | |
| 4965 | eptr++; | eptr++; |
| 4966 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 4967 | } | } |
| # | Line 3940 for (;;) | Line 4973 for (;;) |
| 4973 | { | { |
| 4974 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4975 | { | { |
| 4976 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break; | if (eptr >= md->end_subject) |
| 4977 | { | |
| 4978 | SCHECK_PARTIAL(); | |
| 4979 | break; | |
| 4980 | } | |
| 4981 | if (IS_NEWLINE(eptr)) break; | |
| 4982 | eptr++; | eptr++; |
| 4983 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 4984 | } | } |
| # | Line 3952 for (;;) | Line 4990 for (;;) |
| 4990 | { | { |
| 4991 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4992 | { | { |
| 4993 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4994 | { | |
| 4995 | SCHECK_PARTIAL(); | |
| 4996 | break; | |
| 4997 | } | |
| 4998 | eptr++; | eptr++; |
| 4999 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 5000 | } | } |
| # | Line 3965 for (;;) | Line 5007 for (;;) |
| 5007 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 5008 | c = max - min; | c = max - min; |
| 5009 | if (c > (unsigned int)(md->end_subject - eptr)) | if (c > (unsigned int)(md->end_subject - eptr)) |
| 5010 | c = md->end_subject - eptr; | { |
| 5011 | eptr += c; | eptr = md->end_subject; |
| 5012 | SCHECK_PARTIAL(); | |
| 5013 | } | |
| 5014 | else eptr += c; | |
| 5015 | break; | break; |
| 5016 | ||
| 5017 | case OP_ANYNL: | case OP_ANYNL: |
| 5018 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5019 | { | { |
| 5020 | int len = 1; | int len = 1; |
| 5021 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5022 | { | |
| 5023 | SCHECK_PARTIAL(); | |
| 5024 | break; | |
| 5025 | } | |
| 5026 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 5027 | if (c == 0x000d) | if (c == 0x000d) |
| 5028 | { | { |
| # | Line 3998 for (;;) | Line 5047 for (;;) |
| 5047 | { | { |
| 5048 | BOOL gotspace; | BOOL gotspace; |
| 5049 | int len = 1; | int len = 1; |
| 5050 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5051 | { | |
| 5052 | SCHECK_PARTIAL(); | |
| 5053 | break; | |
| 5054 | } | |
| 5055 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 5056 | switch(c) | switch(c) |
| 5057 | { | { |
| # | Line 4036 for (;;) | Line 5089 for (;;) |
| 5089 | { | { |
| 5090 | BOOL gotspace; | BOOL gotspace; |
| 5091 | int len = 1; | int len = 1; |
| 5092 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5093 | { | |
| 5094 | SCHECK_PARTIAL(); | |
| 5095 | break; | |
| 5096 | } | |
| 5097 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 5098 | switch(c) | switch(c) |
| 5099 | { | { |
| # | Line 4060 for (;;) | Line 5117 for (;;) |
| 5117 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5118 | { | { |
| 5119 | int len = 1; | int len = 1; |
| 5120 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5121 | { | |
| 5122 | SCHECK_PARTIAL(); | |
| 5123 | break; | |
| 5124 | } | |
| 5125 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 5126 | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) break; | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) break; |
| 5127 | eptr+= len; | eptr+= len; |
| # | Line 4071 for (;;) | Line 5132 for (;;) |
| 5132 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5133 | { | { |
| 5134 | int len = 1; | int len = 1; |
| 5135 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5136 | { | |
| 5137 | SCHECK_PARTIAL(); | |
| 5138 | break; | |
| 5139 | } | |
| 5140 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 5141 | if (c >= 256 ||(md->ctypes[c] & ctype_digit) == 0) break; | if (c >= 256 ||(md->ctypes[c] & ctype_digit) == 0) break; |
| 5142 | eptr+= len; | eptr+= len; |
| # | Line 4082 for (;;) | Line 5147 for (;;) |
| 5147 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5148 | { | { |
| 5149 | int len = 1; | int len = 1; |
| 5150 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5151 | { | |
| 5152 | SCHECK_PARTIAL(); | |
| 5153 | break; | |
| 5154 | } | |
| 5155 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 5156 | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) break; | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) break; |
| 5157 | eptr+= len; | eptr+= len; |
| # | Line 4093 for (;;) | Line 5162 for (;;) |
| 5162 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5163 | { | { |
| 5164 | int len = 1; | int len = 1; |
| 5165 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5166 | { | |
| 5167 | SCHECK_PARTIAL(); | |
| 5168 | break; | |
| 5169 | } | |
| 5170 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 5171 | if (c >= 256 ||(md->ctypes[c] & ctype_space) == 0) break; | if (c >= 256 ||(md->ctypes[c] & ctype_space) == 0) break; |
| 5172 | eptr+= len; | eptr+= len; |
| # | Line 4104 for (;;) | Line 5177 for (;;) |
| 5177 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5178 | { | { |
| 5179 | int len = 1; | int len = 1; |
| 5180 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5181 | { | |
| 5182 | SCHECK_PARTIAL(); | |
| 5183 | break; | |
| 5184 | } | |
| 5185 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 5186 | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) break; | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) break; |
| 5187 | eptr+= len; | eptr+= len; |
| # | Line 4115 for (;;) | Line 5192 for (;;) |
| 5192 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5193 | { | { |
| 5194 | int len = 1; | int len = 1; |
| 5195 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5196 | { | |
| 5197 | SCHECK_PARTIAL(); | |
| 5198 | break; | |
| 5199 | } | |
| 5200 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 5201 | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) break; | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) break; |
| 5202 | eptr+= len; | eptr+= len; |
| # | Line 4147 for (;;) | Line 5228 for (;;) |
| 5228 | case OP_ANY: | case OP_ANY: |
| 5229 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5230 | { | { |
| 5231 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break; | if (eptr >= md->end_subject) |
| 5232 | { | |
| 5233 | SCHECK_PARTIAL(); | |
| 5234 | break; | |
| 5235 | } | |
| 5236 | if (IS_NEWLINE(eptr)) break; | |
| 5237 | eptr++; | eptr++; |
| 5238 | } | } |
| 5239 | break; | break; |
| # | Line 4156 for (;;) | Line 5242 for (;;) |
| 5242 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 5243 | c = max - min; | c = max - min; |
| 5244 | if (c > (unsigned int)(md->end_subject - eptr)) | if (c > (unsigned int)(md->end_subject - eptr)) |
| 5245 | c = md->end_subject - eptr; | { |
| 5246 | eptr += c; | eptr = md->end_subject; |
| 5247 | SCHECK_PARTIAL(); | |
| 5248 | } | |
| 5249 | else eptr += c; | |
| 5250 | break; | break; |
| 5251 | ||
| 5252 | case OP_ANYNL: | case OP_ANYNL: |
| 5253 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5254 | { | { |
| 5255 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5256 | { | |
| 5257 | SCHECK_PARTIAL(); | |
| 5258 | break; | |
| 5259 | } | |
| 5260 | c = *eptr; | c = *eptr; |
| 5261 | if (c == 0x000d) | if (c == 0x000d) |
| 5262 | { | { |
| # | Line 4184 for (;;) | Line 5277 for (;;) |
| 5277 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 5278 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5279 | { | { |
| 5280 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5281 | { | |
| 5282 | SCHECK_PARTIAL(); | |
| 5283 | break; | |
| 5284 | } | |
| 5285 | c = *eptr; | c = *eptr; |
| 5286 | if (c == 0x09 || c == 0x20 || c == 0xa0) break; | if (c == 0x09 || c == 0x20 || c == 0xa0) break; |
| 5287 | eptr++; | eptr++; |
| # | Line 4194 for (;;) | Line 5291 for (;;) |
| 5291 | case OP_HSPACE: | case OP_HSPACE: |
| 5292 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5293 | { | { |
| 5294 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5295 | { | |
| 5296 | SCHECK_PARTIAL(); | |
| 5297 | break; | |
| 5298 | } | |
| 5299 | c = *eptr; | c = *eptr; |
| 5300 | if (c != 0x09 && c != 0x20 && c != 0xa0) break; | if (c != 0x09 && c != 0x20 && c != 0xa0) break; |
| 5301 | eptr++; | eptr++; |
| # | Line 4204 for (;;) | Line 5305 for (;;) |
| 5305 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 5306 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5307 | { | { |
| 5308 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5309 | { | |
| 5310 | SCHECK_PARTIAL(); | |
| 5311 | break; | |
| 5312 | } | |
| 5313 | c = *eptr; | c = *eptr; |
| 5314 | if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85) | if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85) |
| 5315 | break; | break; |
| # | Line 4215 for (;;) | Line 5320 for (;;) |
| 5320 | case OP_VSPACE: | case OP_VSPACE: |
| 5321 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5322 | { | { |
| 5323 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5324 | { | |
| 5325 | SCHECK_PARTIAL(); | |
| 5326 | break; | |
| 5327 | } | |
| 5328 | c = *eptr; | c = *eptr; |
| 5329 | if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85) | if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85) |
| 5330 | break; | break; |
| # | Line 4226 for (;;) | Line 5335 for (;;) |
| 5335 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 5336 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5337 | { | { |