Parent Directory
|
Revision Log
|
Patch
| revision 366 by ph10, Mon Jul 14 15:45:32 2008 UTC | revision 510 by ph10, Sat Mar 27 17:45:29 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) 1997-2008 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 74 negative to avoid the external error cod | Line 74 negative to avoid the external error cod |
| 74 | #define MATCH_COMMIT (-999) | #define MATCH_COMMIT (-999) |
| 75 | #define MATCH_PRUNE (-998) | #define MATCH_PRUNE (-998) |
| 76 | #define MATCH_SKIP (-997) | #define MATCH_SKIP (-997) |
| 77 | #define MATCH_THEN (-996) | #define MATCH_SKIP_ARG (-996) |
| 78 | #define MATCH_THEN (-995) | |
| 79 | ||
| 80 | /* This is a convenience macro for code that occurs many times. */ | |
| 81 | ||
| 82 | #define MRRETURN(ra) \ | |
| 83 | { \ | |
| 84 | md->mark = markptr; \ | |
| 85 | RRETURN(ra); \ | |
| 86 | } | |
| 87 | ||
| 88 | /* 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. |
| 89 | 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 98 static const char rep_max[] = { 0, 0, 0, |
| 98 | ||
| 99 | ||
| 100 | ||
| 101 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 102 | /************************************************* | /************************************************* |
| 103 | * Debugging function to print chars * | * Debugging function to print chars * |
| 104 | *************************************************/ | *************************************************/ |
| # | Line 141 match_ref(int offset, register USPTR ept | Line 150 match_ref(int offset, register USPTR ept |
| 150 | { | { |
| 151 | USPTR p = md->start_subject + md->offset_vector[offset]; | USPTR p = md->start_subject + md->offset_vector[offset]; |
| 152 | ||
| 153 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 154 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 155 | printf("matching subject <null>"); | printf("matching subject <null>"); |
| 156 | else | else |
| # | Line 249 enum { RM1=1, RM2, RM3, RM4, RM5, RM | Line 258 enum { RM1=1, RM2, RM3, RM4, RM5, RM |
| 258 | ||
| 259 | /* 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 |
| 260 | 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 |
| 261 | actuall used in this definition. */ | actually used in this definition. */ |
| 262 | ||
| 263 | #ifndef NO_RECURSE | #ifndef NO_RECURSE |
| 264 | #define REGISTER register | #define REGISTER register |
| 265 | ||
| 266 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 267 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ |
| 268 | { \ | { \ |
| 269 | printf("match() called in line %d\n", __LINE__); \ | printf("match() called in line %d\n", __LINE__); \ |
| 270 | 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); \ |
| 271 | printf("to line %d\n", __LINE__); \ | printf("to line %d\n", __LINE__); \ |
| 272 | } | } |
| 273 | #define RRETURN(ra) \ | #define RRETURN(ra) \ |
| # | Line 268 actuall used in this definition. */ | Line 277 actuall used in this definition. */ |
| 277 | } | } |
| 278 | #else | #else |
| 279 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ |
| 280 | 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) |
| 281 | #define RRETURN(ra) return ra | #define RRETURN(ra) return ra |
| 282 | #endif | #endif |
| 283 | ||
| # | Line 288 argument of match(), which never changes | Line 297 argument of match(), which never changes |
| 297 | newframe->Xeptr = ra;\ | newframe->Xeptr = ra;\ |
| 298 | newframe->Xecode = rb;\ | newframe->Xecode = rb;\ |
| 299 | newframe->Xmstart = mstart;\ | newframe->Xmstart = mstart;\ |
| 300 | newframe->Xmarkptr = markptr;\ | |
| 301 | newframe->Xoffset_top = rc;\ | newframe->Xoffset_top = rc;\ |
| 302 | newframe->Xims = re;\ | newframe->Xims = re;\ |
| 303 | newframe->Xeptrb = rf;\ | newframe->Xeptrb = rf;\ |
| # | Line 322 typedef struct heapframe { | Line 332 typedef struct heapframe { |
| 332 | ||
| 333 | /* Function arguments that may change */ | /* Function arguments that may change */ |
| 334 | ||
| 335 | const uschar *Xeptr; | USPTR Xeptr; |
| 336 | const uschar *Xecode; | const uschar *Xecode; |
| 337 | const uschar *Xmstart; | USPTR Xmstart; |
| 338 | USPTR Xmarkptr; | |
| 339 | int Xoffset_top; | int Xoffset_top; |
| 340 | long int Xims; | long int Xims; |
| 341 | eptrblock *Xeptrb; | eptrblock *Xeptrb; |
| # | Line 333 typedef struct heapframe { | Line 344 typedef struct heapframe { |
| 344 | ||
| 345 | /* Function local variables */ | /* Function local variables */ |
| 346 | ||
| 347 | const uschar *Xcallpat; | USPTR Xcallpat; |
| 348 | const uschar *Xcharptr; | #ifdef SUPPORT_UTF8 |
| 349 | const uschar *Xdata; | USPTR Xcharptr; |
| 350 | const uschar *Xnext; | #endif |
| 351 | const uschar *Xpp; | USPTR Xdata; |
| 352 | const uschar *Xprev; | USPTR Xnext; |
| 353 | const uschar *Xsaved_eptr; | USPTR Xpp; |
| 354 | USPTR Xprev; | |
| 355 | USPTR Xsaved_eptr; | |
| 356 | ||
| 357 | recursion_info Xnew_recursive; | recursion_info Xnew_recursive; |
| 358 | ||
| # | Line 360 typedef struct heapframe { | Line 373 typedef struct heapframe { |
| 373 | uschar Xocchars[8]; | uschar Xocchars[8]; |
| 374 | #endif | #endif |
| 375 | ||
| 376 | int Xcodelink; | |
| 377 | int Xctype; | int Xctype; |
| 378 | unsigned int Xfc; | unsigned int Xfc; |
| 379 | int Xfi; | int Xfi; |
| # | Line 395 typedef struct heapframe { | Line 409 typedef struct heapframe { |
| 409 | ||
| 410 | /* This function is called recursively in many circumstances. Whenever it | /* This function is called recursively in many circumstances. Whenever it |
| 411 | returns a negative (error) response, the outer incarnation must also return the | returns a negative (error) response, the outer incarnation must also return the |
| 412 | same response. | same response. */ |
| 413 | ||
| 414 | /* These macros pack up tests that are used for partial matching, and which | |
| 415 | appears several times in the code. We set the "hit end" flag if the pointer is | |
| 416 | at the end of the subject and also past the start of the subject (i.e. | |
| 417 | something has been matched). For hard partial matching, we then return | |
| 418 | immediately. The second one is used when we already know we are past the end of | |
| 419 | the subject. */ | |
| 420 | ||
| 421 | #define CHECK_PARTIAL()\ | |
| 422 | if (md->partial != 0 && eptr >= md->end_subject && eptr > mstart)\ | |
| 423 | {\ | |
| 424 | md->hitend = TRUE;\ | |
| 425 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL);\ | |
| 426 | } | |
| 427 | ||
| 428 | #define SCHECK_PARTIAL()\ | |
| 429 | if (md->partial != 0 && eptr > mstart)\ | |
| 430 | {\ | |
| 431 | md->hitend = TRUE;\ | |
| 432 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL);\ | |
| 433 | } | |
| 434 | ||
| 435 | ||
| 436 | 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 |
| 437 | 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 |
| 438 | 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 |
| 439 | made performance worse. | made performance worse. |
| 440 | ||
| # | Line 407 Arguments: | Line 443 Arguments: |
| 443 | ecode pointer to current position in compiled code | ecode pointer to current position in compiled code |
| 444 | mstart pointer to the current match start position (can be modified | mstart pointer to the current match start position (can be modified |
| 445 | by encountering \K) | by encountering \K) |
| 446 | markptr pointer to the most recent MARK name, or NULL | |
| 447 | offset_top current top pointer | offset_top current top pointer |
| 448 | md pointer to "static" info for the match | md pointer to "static" info for the match |
| 449 | ims current /i, /m, and /s options | ims current /i, /m, and /s options |
| # | Line 420 Arguments: | Line 457 Arguments: |
| 457 | ||
| 458 | Returns: MATCH_MATCH if matched ) these values are >= 0 | Returns: MATCH_MATCH if matched ) these values are >= 0 |
| 459 | MATCH_NOMATCH if failed to match ) | MATCH_NOMATCH if failed to match ) |
| 460 | a negative MATCH_xxx value for PRUNE, SKIP, etc | |
| 461 | a negative PCRE_ERROR_xxx value if aborted by an error condition | a negative PCRE_ERROR_xxx value if aborted by an error condition |
| 462 | (e.g. stopped by repeated call or recursion limit) | (e.g. stopped by repeated call or recursion limit) |
| 463 | */ | */ |
| 464 | ||
| 465 | static int | static int |
| 466 | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart, | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart, |
| 467 | 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, |
| 468 | int flags, unsigned int rdepth) | eptrblock *eptrb, int flags, unsigned int rdepth) |
| 469 | { | { |
| 470 | /* 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, |
| 471 | 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 439 register unsigned int c; /* Character | Line 477 register unsigned int c; /* Character |
| 477 | register BOOL utf8; /* Local copy of UTF-8 flag for speed */ | register BOOL utf8; /* Local copy of UTF-8 flag for speed */ |
| 478 | ||
| 479 | BOOL minimize, possessive; /* Quantifier options */ | BOOL minimize, possessive; /* Quantifier options */ |
| 480 | int condcode; | |
| 481 | ||
| 482 | /* When recursion is not being used, all "local" variables that have to be | /* When recursion is not being used, all "local" variables that have to be |
| 483 | preserved over calls to RMATCH() are part of a "frame" which is obtained from | preserved over calls to RMATCH() are part of a "frame" which is obtained from |
| # | Line 454 frame->Xprevframe = NULL; /* | Line 493 frame->Xprevframe = NULL; /* |
| 493 | frame->Xeptr = eptr; | frame->Xeptr = eptr; |
| 494 | frame->Xecode = ecode; | frame->Xecode = ecode; |
| 495 | frame->Xmstart = mstart; | frame->Xmstart = mstart; |
| 496 | frame->Xmarkptr = markptr; | |
| 497 | frame->Xoffset_top = offset_top; | frame->Xoffset_top = offset_top; |
| 498 | frame->Xims = ims; | frame->Xims = ims; |
| 499 | frame->Xeptrb = eptrb; | frame->Xeptrb = eptrb; |
| # | Line 469 HEAP_RECURSE: | Line 509 HEAP_RECURSE: |
| 509 | #define eptr frame->Xeptr | #define eptr frame->Xeptr |
| 510 | #define ecode frame->Xecode | #define ecode frame->Xecode |
| 511 | #define mstart frame->Xmstart | #define mstart frame->Xmstart |
| 512 | #define markptr frame->Xmarkptr | |
| 513 | #define offset_top frame->Xoffset_top | #define offset_top frame->Xoffset_top |
| 514 | #define ims frame->Xims | #define ims frame->Xims |
| 515 | #define eptrb frame->Xeptrb | #define eptrb frame->Xeptrb |
| # | Line 481 HEAP_RECURSE: | Line 522 HEAP_RECURSE: |
| 522 | #define charptr frame->Xcharptr | #define charptr frame->Xcharptr |
| 523 | #endif | #endif |
| 524 | #define callpat frame->Xcallpat | #define callpat frame->Xcallpat |
| 525 | #define codelink frame->Xcodelink | |
| 526 | #define data frame->Xdata | #define data frame->Xdata |
| 527 | #define next frame->Xnext | #define next frame->Xnext |
| 528 | #define pp frame->Xpp | #define pp frame->Xpp |
| # | Line 561 int oclength; | Line 603 int oclength; |
| 603 | uschar occhars[8]; | uschar occhars[8]; |
| 604 | #endif | #endif |
| 605 | ||
| 606 | int codelink; | |
| 607 | int ctype; | int ctype; |
| 608 | int length; | int length; |
| 609 | int max; | int max; |
| # | Line 594 TAIL_RECURSE: | Line 637 TAIL_RECURSE: |
| 637 | /* 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 |
| 638 | 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 |
| 639 | 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() |
| 640 | and a "return", respectively (possibly with some debugging if DEBUG is | and a "return", respectively (possibly with some debugging if PCRE_DEBUG is |
| 641 | 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 |
| 642 | 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, |
| 643 | however, impact performance when true recursion is being used. */ | however, impact performance when true recursion is being used. */ |
| # | Line 636 for (;;) | Line 679 for (;;) |
| 679 | minimize = possessive = FALSE; | minimize = possessive = FALSE; |
| 680 | op = *ecode; | op = *ecode; |
| 681 | ||
| /* 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; | ||
| 682 | switch(op) | switch(op) |
| 683 | { | { |
| 684 | case OP_MARK: | |
| 685 | markptr = ecode + 2; | |
| 686 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 687 | ims, eptrb, flags, RM51); | |
| 688 | ||
| 689 | /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an | |
| 690 | argument, and we must check whether that argument matches this MARK's | |
| 691 | argument. It is passed back in md->start_match_ptr (an overloading of that | |
| 692 | variable). If it does match, we reset that variable to the current subject | |
| 693 | position and return MATCH_SKIP. Otherwise, pass back the return code | |
| 694 | unaltered. */ | |
| 695 | ||
| 696 | if (rrc == MATCH_SKIP_ARG && | |
| 697 | strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0) | |
| 698 | { | |
| 699 | md->start_match_ptr = eptr; | |
| 700 | RRETURN(MATCH_SKIP); | |
| 701 | } | |
| 702 | ||
| 703 | if (md->mark == NULL) md->mark = markptr; | |
| 704 | RRETURN(rrc); | |
| 705 | ||
| 706 | case OP_FAIL: | case OP_FAIL: |
| 707 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 708 | ||
| 709 | case OP_COMMIT: | |
| 710 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | |
| 711 | ims, eptrb, flags, RM52); | |
| 712 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 713 | MRRETURN(MATCH_COMMIT); | |
| 714 | ||
| 715 | case OP_PRUNE: | case OP_PRUNE: |
| 716 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 717 | ims, eptrb, flags, RM51); | ims, eptrb, flags, RM51); |
| 718 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 719 | RRETURN(MATCH_PRUNE); | MRRETURN(MATCH_PRUNE); |
| 720 | ||
| 721 | case OP_COMMIT: | case OP_PRUNE_ARG: |
| 722 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, |
| 723 | ims, eptrb, flags, RM52); | ims, eptrb, flags, RM51); |
| 724 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 725 | RRETURN(MATCH_COMMIT); | md->mark = ecode + 2; |
| 726 | RRETURN(MATCH_PRUNE); | |
| 727 | ||
| 728 | case OP_SKIP: | case OP_SKIP: |
| 729 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 730 | ims, eptrb, flags, RM53); | ims, eptrb, flags, RM53); |
| 731 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 732 | md->start_match_ptr = eptr; /* Pass back current position */ | md->start_match_ptr = eptr; /* Pass back current position */ |
| 733 | RRETURN(MATCH_SKIP); | MRRETURN(MATCH_SKIP); |
| 734 | ||
| 735 | case OP_SKIP_ARG: | |
| 736 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 737 | ims, eptrb, flags, RM53); | |
| 738 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 739 | ||
| 740 | /* Pass back the current skip name by overloading md->start_match_ptr and | |
| 741 | returning the special MATCH_SKIP_ARG return code. This will either be | |
| 742 | caught by a matching MARK, or get to the top, where it is treated the same | |
| 743 | as PRUNE. */ | |
| 744 | ||
| 745 | md->start_match_ptr = ecode + 2; | |
| 746 | RRETURN(MATCH_SKIP_ARG); | |
| 747 | ||
| 748 | case OP_THEN: | case OP_THEN: |
| 749 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 750 | ims, eptrb, flags, RM54); | ims, eptrb, flags, RM54); |
| 751 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 752 | MRRETURN(MATCH_THEN); | |
| 753 | ||
| 754 | case OP_THEN_ARG: | |
| 755 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 756 | ims, eptrb, flags, RM54); | |
| 757 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 758 | md->mark = ecode + 2; | |
| 759 | RRETURN(MATCH_THEN); | RRETURN(MATCH_THEN); |
| 760 | ||
| 761 | /* 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 693 for (;;) | Line 777 for (;;) |
| 777 | number = GET2(ecode, 1+LINK_SIZE); | number = GET2(ecode, 1+LINK_SIZE); |
| 778 | offset = number << 1; | offset = number << 1; |
| 779 | ||
| 780 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 781 | printf("start bracket %d\n", number); | printf("start bracket %d\n", number); |
| 782 | printf("subject="); | printf("subject="); |
| 783 | pchars(eptr, 16, TRUE, md); | pchars(eptr, 16, TRUE, md); |
| # | Line 727 for (;;) | Line 811 for (;;) |
| 811 | md->offset_vector[offset+1] = save_offset2; | md->offset_vector[offset+1] = save_offset2; |
| 812 | md->offset_vector[md->offset_end - number] = save_offset3; | md->offset_vector[md->offset_end - number] = save_offset3; |
| 813 | ||
| 814 | if (rrc != MATCH_THEN) md->mark = markptr; | |
| 815 | RRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 816 | } | } |
| 817 | ||
| # | Line 766 for (;;) | Line 851 for (;;) |
| 851 | ||
| 852 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, |
| 853 | eptrb, flags, RM48); | eptrb, flags, RM48); |
| 854 | RRETURN(rrc); | if (rrc == MATCH_NOMATCH) md->mark = markptr; |
| 855 | RRETURN(rrc); | |
| 856 | } | } |
| 857 | ||
| 858 | /* For non-final alternatives, continue the loop for a NOMATCH result; | /* For non-final alternatives, continue the loop for a NOMATCH result; |
| # | Line 787 for (;;) | Line 873 for (;;) |
| 873 | ||
| 874 | case OP_COND: | case OP_COND: |
| 875 | case OP_SCOND: | case OP_SCOND: |
| 876 | if (ecode[LINK_SIZE+1] == OP_RREF) /* Recursion test */ | codelink= GET(ecode, 1); |
| 877 | ||
| 878 | /* Because of the way auto-callout works during compile, a callout item is | |
| 879 | inserted between OP_COND and an assertion condition. */ | |
| 880 | ||
| 881 | if (ecode[LINK_SIZE+1] == OP_CALLOUT) | |
| 882 | { | { |
| 883 | offset = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | if (pcre_callout != NULL) |
| 884 | condition = md->recursive != NULL && | { |
| 885 | (offset == RREF_ANY || offset == md->recursive->group_num); | pcre_callout_block cb; |
| 886 | ecode += condition? 3 : GET(ecode, 1); | cb.version = 1; /* Version 1 of the callout block */ |
| 887 | cb.callout_number = ecode[LINK_SIZE+2]; | |
| 888 | cb.offset_vector = md->offset_vector; | |
| 889 | cb.subject = (PCRE_SPTR)md->start_subject; | |
| 890 | cb.subject_length = md->end_subject - md->start_subject; | |
| 891 | cb.start_match = mstart - md->start_subject; | |
| 892 | cb.current_position = eptr - md->start_subject; | |
| 893 | cb.pattern_position = GET(ecode, LINK_SIZE + 3); | |
| 894 | cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE); | |
| 895 | cb.capture_top = offset_top/2; | |
| 896 | cb.capture_last = md->capture_last; | |
| 897 | cb.callout_data = md->callout_data; | |
| 898 | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); | |
| 899 | if (rrc < 0) RRETURN(rrc); | |
| 900 | } | |
| 901 | ecode += _pcre_OP_lengths[OP_CALLOUT]; | |
| 902 | } | |
| 903 | ||
| 904 | condcode = ecode[LINK_SIZE+1]; | |
| 905 | ||
| 906 | /* Now see what the actual condition is */ | |
| 907 | ||
| 908 | if (condcode == OP_RREF || condcode == OP_NRREF) /* Recursion test */ | |
| 909 | { | |
| 910 | if (md->recursive == NULL) /* Not recursing => FALSE */ | |
| 911 | { | |
| 912 | condition = FALSE; | |
| 913 | ecode += GET(ecode, 1); | |
| 914 | } | |
| 915 | else | |
| 916 | { | |
| 917 | int recno = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | |
| 918 | condition = (recno == RREF_ANY || recno == md->recursive->group_num); | |
| 919 | ||
| 920 | /* If the test is for recursion into a specific subpattern, and it is | |
| 921 | false, but the test was set up by name, scan the table to see if the | |
| 922 | name refers to any other numbers, and test them. The condition is true | |
| 923 | if any one is set. */ | |
| 924 | ||
| 925 | if (!condition && condcode == OP_NRREF && recno != RREF_ANY) | |
| 926 | { | |
| 927 | uschar *slotA = md->name_table; | |
| 928 | for (i = 0; i < md->name_count; i++) | |
| 929 | { | |
| 930 | if (GET2(slotA, 0) == recno) break; | |
| 931 | slotA += md->name_entry_size; | |
| 932 | } | |
| 933 | ||
| 934 | /* Found a name for the number - there can be only one; duplicate | |
| 935 | names for different numbers are allowed, but not vice versa. First | |
| 936 | scan down for duplicates. */ | |
| 937 | ||
| 938 | if (i < md->name_count) | |
| 939 | { | |
| 940 | uschar *slotB = slotA; | |
| 941 | while (slotB > md->name_table) | |
| 942 | { | |
| 943 | slotB -= md->name_entry_size; | |
| 944 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 945 | { | |
| 946 | condition = GET2(slotB, 0) == md->recursive->group_num; | |
| 947 | if (condition) break; | |
| 948 | } | |
| 949 | else break; | |
| 950 | } | |
| 951 | ||
| 952 | /* Scan up for duplicates */ | |
| 953 | ||
| 954 | if (!condition) | |
| 955 | { | |
| 956 | slotB = slotA; | |
| 957 | for (i++; i < md->name_count; i++) | |
| 958 | { | |
| 959 | slotB += md->name_entry_size; | |
| 960 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 961 | { | |
| 962 | condition = GET2(slotB, 0) == md->recursive->group_num; | |
| 963 | if (condition) break; | |
| 964 | } | |
| 965 | else break; | |
| 966 | } | |
| 967 | } | |
| 968 | } | |
| 969 | } | |
| 970 | ||
| 971 | /* Chose branch according to the condition */ | |
| 972 | ||
| 973 | ecode += condition? 3 : GET(ecode, 1); | |
| 974 | } | |
| 975 | } | } |
| 976 | ||
| 977 | else if (ecode[LINK_SIZE+1] == OP_CREF) /* Group used test */ | else if (condcode == OP_CREF || condcode == OP_NCREF) /* Group used test */ |
| 978 | { | { |
| 979 | offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ | offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ |
| 980 | condition = offset < offset_top && md->offset_vector[offset] >= 0; | condition = offset < offset_top && md->offset_vector[offset] >= 0; |
| 981 | ||
| 982 | /* If the numbered capture is unset, but the reference was by name, | |
| 983 | scan the table to see if the name refers to any other numbers, and test | |
| 984 | them. The condition is true if any one is set. This is tediously similar | |
| 985 | to the code above, but not close enough to try to amalgamate. */ | |
| 986 | ||
| 987 | if (!condition && condcode == OP_NCREF) | |
| 988 | { | |
| 989 | int refno = offset >> 1; | |
| 990 | uschar *slotA = md->name_table; | |
| 991 | ||
| 992 | for (i = 0; i < md->name_count; i++) | |
| 993 | { | |
| 994 | if (GET2(slotA, 0) == refno) break; | |
| 995 | slotA += md->name_entry_size; | |
| 996 | } | |
| 997 | ||
| 998 | /* Found a name for the number - there can be only one; duplicate names | |
| 999 | for different numbers are allowed, but not vice versa. First scan down | |
| 1000 | for duplicates. */ | |
| 1001 | ||
| 1002 | if (i < md->name_count) | |
| 1003 | { | |
| 1004 | uschar *slotB = slotA; | |
| 1005 | while (slotB > md->name_table) | |
| 1006 | { | |
| 1007 | slotB -= md->name_entry_size; | |
| 1008 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1009 | { | |
| 1010 | offset = GET2(slotB, 0) << 1; | |
| 1011 | condition = offset < offset_top && | |
| 1012 | md->offset_vector[offset] >= 0; | |
| 1013 | if (condition) break; | |
| 1014 | } | |
| 1015 | else break; | |
| 1016 | } | |
| 1017 | ||
| 1018 | /* Scan up for duplicates */ | |
| 1019 | ||
| 1020 | if (!condition) | |
| 1021 | { | |
| 1022 | slotB = slotA; | |
| 1023 | for (i++; i < md->name_count; i++) | |
| 1024 | { | |
| 1025 | slotB += md->name_entry_size; | |
| 1026 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1027 | { | |
| 1028 | offset = GET2(slotB, 0) << 1; | |
| 1029 | condition = offset < offset_top && | |
| 1030 | md->offset_vector[offset] >= 0; | |
| 1031 | if (condition) break; | |
| 1032 | } | |
| 1033 | else break; | |
| 1034 | } | |
| 1035 | } | |
| 1036 | } | |
| 1037 | } | |
| 1038 | ||
| 1039 | /* Chose branch according to the condition */ | |
| 1040 | ||
| 1041 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 3 : GET(ecode, 1); |
| 1042 | } | } |
| 1043 | ||
| 1044 | else if (ecode[LINK_SIZE+1] == OP_DEF) /* DEFINE - always false */ | else if (condcode == OP_DEF) /* DEFINE - always false */ |
| 1045 | { | { |
| 1046 | condition = FALSE; | condition = FALSE; |
| 1047 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| # | Line 829 for (;;) | Line 1068 for (;;) |
| 1068 | else | else |
| 1069 | { | { |
| 1070 | condition = FALSE; | condition = FALSE; |
| 1071 | ecode += GET(ecode, 1); | ecode += codelink; |
| 1072 | } | } |
| 1073 | } | } |
| 1074 | ||
| # | Line 852 for (;;) | Line 1091 for (;;) |
| 1091 | goto TAIL_RECURSE; | goto TAIL_RECURSE; |
| 1092 | } | } |
| 1093 | } | } |
| 1094 | else /* Condition false & no 2nd alternative */ | else /* Condition false & no alternative */ |
| 1095 | { | { |
| 1096 | ecode += 1 + LINK_SIZE; | ecode += 1 + LINK_SIZE; |
| 1097 | } | } |
| 1098 | break; | break; |
| 1099 | ||
| 1100 | ||
| 1101 | /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes, | |
| 1102 | to close any currently open capturing brackets. */ | |
| 1103 | ||
| 1104 | case OP_CLOSE: | |
| 1105 | number = GET2(ecode, 1); | |
| 1106 | offset = number << 1; | |
| 1107 | ||
| 1108 | #ifdef PCRE_DEBUG | |
| 1109 | printf("end bracket %d at *ACCEPT", number); | |
| 1110 | printf("\n"); | |
| 1111 | #endif | |
| 1112 | ||
| 1113 | md->capture_last = number; | |
| 1114 | if (offset >= md->offset_max) md->offset_overflow = TRUE; else | |
| 1115 | { | |
| 1116 | md->offset_vector[offset] = | |
| 1117 | md->offset_vector[md->offset_end - number]; | |
| 1118 | md->offset_vector[offset+1] = eptr - md->start_subject; | |
| 1119 | if (offset_top <= offset) offset_top = offset + 2; | |
| 1120 | } | |
| 1121 | ecode += 3; | |
| 1122 | break; | |
| 1123 | ||
| 1124 | ||
| 1125 | /* 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 |
| 1126 | recursion, we should restore the offsets appropriately and continue from | recursion, we should restore the offsets appropriately and continue from |
| 1127 | after the call. */ | after the call. */ |
| # | Line 872 for (;;) | Line 1135 for (;;) |
| 1135 | md->recursive = rec->prevrec; | md->recursive = rec->prevrec; |
| 1136 | memmove(md->offset_vector, rec->offset_save, | memmove(md->offset_vector, rec->offset_save, |
| 1137 | rec->saved_max * sizeof(int)); | rec->saved_max * sizeof(int)); |
| 1138 | mstart = rec->save_start; | offset_top = rec->save_offset_top; |
| 1139 | ims = original_ims; | ims = original_ims; |
| 1140 | ecode = rec->after_call; | ecode = rec->after_call; |
| 1141 | break; | break; |
| 1142 | } | } |
| 1143 | ||
| 1144 | /* 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 |
| 1145 | 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 |
| 1146 | the subject. In both cases, backtracking will then try other alternatives, | |
| 1147 | if any. */ | |
| 1148 | ||
| 1149 | if (eptr == mstart && | |
| 1150 | (md->notempty || | |
| 1151 | (md->notempty_atstart && | |
| 1152 | mstart == md->start_subject + md->start_offset))) | |
| 1153 | MRRETURN(MATCH_NOMATCH); | |
| 1154 | ||
| 1155 | /* Otherwise, we have a match. */ | |
| 1156 | ||
| if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH); | ||
| 1157 | md->end_match_ptr = eptr; /* Record where we ended */ | md->end_match_ptr = eptr; /* Record where we ended */ |
| 1158 | md->end_offset_top = offset_top; /* and how many extracts were taken */ | md->end_offset_top = offset_top; /* and how many extracts were taken */ |
| 1159 | md->start_match_ptr = mstart; /* and the start (\K can modify) */ | md->start_match_ptr = mstart; /* and the start (\K can modify) */ |
| 1160 | RRETURN(MATCH_MATCH); | MRRETURN(MATCH_MATCH); |
| 1161 | ||
| 1162 | /* Change option settings */ | /* Change option settings */ |
| 1163 | ||
| # | Line 907 for (;;) | Line 1179 for (;;) |
| 1179 | { | { |
| 1180 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, |
| 1181 | RM4); | RM4); |
| 1182 | if (rrc == MATCH_MATCH) break; | if (rrc == MATCH_MATCH) |
| 1183 | { | |
| 1184 | mstart = md->start_match_ptr; /* In case \K reset it */ | |
| 1185 | break; | |
| 1186 | } | |
| 1187 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| 1188 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 1189 | } | } |
| 1190 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| 1191 | if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); | if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH); |
| 1192 | ||
| 1193 | /* If checking an assertion for a condition, return MATCH_MATCH. */ | /* If checking an assertion for a condition, return MATCH_MATCH. */ |
| 1194 | ||
| # | Line 926 for (;;) | Line 1202 for (;;) |
| 1202 | offset_top = md->end_offset_top; | offset_top = md->end_offset_top; |
| 1203 | continue; | continue; |
| 1204 | ||
| 1205 | /* Negative assertion: all branches must fail to match */ | /* Negative assertion: all branches must fail to match. Encountering SKIP, |
| 1206 | PRUNE, or COMMIT means we must assume failure without checking subsequent | |
| 1207 | branches. */ | |
| 1208 | ||
| 1209 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
| 1210 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
| # | Line 934 for (;;) | Line 1212 for (;;) |
| 1212 | { | { |
| 1213 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, |
| 1214 | RM5); | RM5); |
| 1215 | if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH); | if (rrc == MATCH_MATCH) MRRETURN(MATCH_NOMATCH); |
| 1216 | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) | |
| 1217 | { | |
| 1218 | do ecode += GET(ecode,1); while (*ecode == OP_ALT); | |
| 1219 | break; | |
| 1220 | } | |
| 1221 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| 1222 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
| 1223 | } | } |
| # | Line 958 for (;;) | Line 1241 for (;;) |
| 1241 | while (i-- > 0) | while (i-- > 0) |
| 1242 | { | { |
| 1243 | eptr--; | eptr--; |
| 1244 | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1245 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 1246 | } | } |
| 1247 | } | } |
| # | Line 969 for (;;) | Line 1252 for (;;) |
| 1252 | ||
| 1253 | { | { |
| 1254 | eptr -= GET(ecode, 1); | eptr -= GET(ecode, 1); |
| 1255 | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1256 | } | } |
| 1257 | ||
| 1258 | /* Skip to next op code */ | /* Save the earliest consulted character, then skip to next op code */ |
| 1259 | ||
| 1260 | if (eptr < md->start_used_ptr) md->start_used_ptr = eptr; | |
| 1261 | ecode += 1 + LINK_SIZE; | ecode += 1 + LINK_SIZE; |
| 1262 | break; | break; |
| 1263 | ||
| # | Line 997 for (;;) | Line 1281 for (;;) |
| 1281 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 1282 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 1283 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 1284 | if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH); | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
| 1285 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 1286 | } | } |
| 1287 | ecode += 2 + 2*LINK_SIZE; | ecode += 2 + 2*LINK_SIZE; |
| # | Line 1052 for (;;) | Line 1336 for (;;) |
| 1336 | ||
| 1337 | memcpy(new_recursive.offset_save, md->offset_vector, | memcpy(new_recursive.offset_save, md->offset_vector, |
| 1338 | new_recursive.saved_max * sizeof(int)); | new_recursive.saved_max * sizeof(int)); |
| 1339 | new_recursive.save_start = mstart; | new_recursive.save_offset_top = offset_top; |
| mstart = eptr; | ||
| 1340 | ||
| 1341 | /* 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 |
| 1342 | restore the offset and recursion data. */ | restore the offset and recursion data. */ |
| # | Line 1070 for (;;) | Line 1353 for (;;) |
| 1353 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1354 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1355 | (pcre_free)(new_recursive.offset_save); | (pcre_free)(new_recursive.offset_save); |
| 1356 | RRETURN(MATCH_MATCH); | MRRETURN(MATCH_MATCH); |
| 1357 | } | } |
| 1358 | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) |
| 1359 | { | { |
| 1360 | DPRINTF(("Recursion gave error %d\n", rrc)); | DPRINTF(("Recursion gave error %d\n", rrc)); |
| 1361 | if (new_recursive.offset_save != stacksave) | |
| 1362 | (pcre_free)(new_recursive.offset_save); | |
| 1363 | RRETURN(rrc); | RRETURN(rrc); |
| 1364 | } | } |
| 1365 | ||
| # | Line 1089 for (;;) | Line 1374 for (;;) |
| 1374 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1375 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1376 | (pcre_free)(new_recursive.offset_save); | (pcre_free)(new_recursive.offset_save); |
| 1377 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1378 | } | } |
| 1379 | /* Control never reaches here */ | /* Control never reaches here */ |
| 1380 | ||
| # | Line 1098 for (;;) | Line 1383 for (;;) |
| 1383 | a move back into the brackets. Friedl calls these "atomic" subpatterns. | a move back into the brackets. Friedl calls these "atomic" subpatterns. |
| 1384 | 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 |
| 1385 | 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 |
| 1386 | the end of a normal bracket, leaving the subject pointer. */ | the end of a normal bracket, leaving the subject pointer, but resetting |
| 1387 | the start-of-match value in case it was changed by \K. */ | |
| 1388 | ||
| 1389 | case OP_ONCE: | case OP_ONCE: |
| 1390 | prev = ecode; | prev = ecode; |
| # | Line 1107 for (;;) | Line 1393 for (;;) |
| 1393 | do | do |
| 1394 | { | { |
| 1395 | 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); |
| 1396 | if (rrc == MATCH_MATCH) break; | if (rrc == MATCH_MATCH) |
| 1397 | { | |
| 1398 | mstart = md->start_match_ptr; | |
| 1399 | break; | |
| 1400 | } | |
| 1401 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| 1402 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
| 1403 | } | } |
| # | Line 1226 for (;;) | Line 1516 for (;;) |
| 1516 | } | } |
| 1517 | else saved_eptr = NULL; | else saved_eptr = NULL; |
| 1518 | ||
| 1519 | /* 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 |
| 1520 | 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 |
| 1521 | assertions. Do this also for the "once" (atomic) groups. */ | use by positive assertions. We also need to record the match start in case |
| 1522 | it was changed by \K. */ | |
| 1523 | ||
| 1524 | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || |
| 1525 | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || |
| # | Line 1236 for (;;) | Line 1527 for (;;) |
| 1527 | { | { |
| 1528 | md->end_match_ptr = eptr; /* For ONCE */ | md->end_match_ptr = eptr; /* For ONCE */ |
| 1529 | md->end_offset_top = offset_top; | md->end_offset_top = offset_top; |
| 1530 | RRETURN(MATCH_MATCH); | md->start_match_ptr = mstart; |
| 1531 | MRRETURN(MATCH_MATCH); | |
| 1532 | } | } |
| 1533 | ||
| 1534 | /* 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 1250 for (;;) | Line 1542 for (;;) |
| 1542 | number = GET2(prev, 1+LINK_SIZE); | number = GET2(prev, 1+LINK_SIZE); |
| 1543 | offset = number << 1; | offset = number << 1; |
| 1544 | ||
| 1545 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 1546 | printf("end bracket %d", number); | printf("end bracket %d", number); |
| 1547 | printf("\n"); | printf("\n"); |
| 1548 | #endif | #endif |
| # | Line 1272 for (;;) | Line 1564 for (;;) |
| 1564 | recursion_info *rec = md->recursive; | recursion_info *rec = md->recursive; |
| 1565 | DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); | DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); |
| 1566 | md->recursive = rec->prevrec; | md->recursive = rec->prevrec; |
| mstart = rec->save_start; | ||
| 1567 | memcpy(md->offset_vector, rec->offset_save, | memcpy(md->offset_vector, rec->offset_save, |
| 1568 | rec->saved_max * sizeof(int)); | rec->saved_max * sizeof(int)); |
| 1569 | offset_top = rec->save_offset_top; | |
| 1570 | ecode = rec->after_call; | ecode = rec->after_call; |
| 1571 | ims = original_ims; | ims = original_ims; |
| 1572 | break; | break; |
| # | Line 1331 for (;;) | Line 1623 for (;;) |
| 1623 | /* Start of subject unless notbol, or after internal newline if multiline */ | /* Start of subject unless notbol, or after internal newline if multiline */ |
| 1624 | ||
| 1625 | case OP_CIRC: | case OP_CIRC: |
| 1626 | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1627 | if ((ims & PCRE_MULTILINE) != 0) | if ((ims & PCRE_MULTILINE) != 0) |
| 1628 | { | { |
| 1629 | if (eptr != md->start_subject && | if (eptr != md->start_subject && |
| 1630 | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) |
| 1631 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1632 | ecode++; | ecode++; |
| 1633 | break; | break; |
| 1634 | } | } |
| # | Line 1345 for (;;) | Line 1637 for (;;) |
| 1637 | /* Start of subject assertion */ | /* Start of subject assertion */ |
| 1638 | ||
| 1639 | case OP_SOD: | case OP_SOD: |
| 1640 | if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1641 | ecode++; | ecode++; |
| 1642 | break; | break; |
| 1643 | ||
| 1644 | /* Start of match assertion */ | /* Start of match assertion */ |
| 1645 | ||
| 1646 | case OP_SOM: | case OP_SOM: |
| 1647 | if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH); |
| 1648 | ecode++; | ecode++; |
| 1649 | break; | break; |
| 1650 | ||
| # | Line 1370 for (;;) | Line 1662 for (;;) |
| 1662 | if ((ims & PCRE_MULTILINE) != 0) | if ((ims & PCRE_MULTILINE) != 0) |
| 1663 | { | { |
| 1664 | if (eptr < md->end_subject) | if (eptr < md->end_subject) |
| 1665 | { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); } | { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); } |
| 1666 | else | else |
| 1667 | { if (md->noteol) RRETURN(MATCH_NOMATCH); } | { if (md->noteol) MRRETURN(MATCH_NOMATCH); } |
| 1668 | ecode++; | ecode++; |
| 1669 | break; | break; |
| 1670 | } | } |
| 1671 | else | else |
| 1672 | { | { |
| 1673 | if (md->noteol) RRETURN(MATCH_NOMATCH); | if (md->noteol) MRRETURN(MATCH_NOMATCH); |
| 1674 | if (!md->endonly) | if (!md->endonly) |
| 1675 | { | { |
| 1676 | if (eptr != md->end_subject && | if (eptr != md->end_subject && |
| 1677 | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) |
| 1678 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1679 | ecode++; | ecode++; |
| 1680 | break; | break; |
| 1681 | } | } |
| # | Line 1393 for (;;) | Line 1685 for (;;) |
| 1685 | /* End of subject assertion (\z) */ | /* End of subject assertion (\z) */ |
| 1686 | ||
| 1687 | case OP_EOD: | case OP_EOD: |
| 1688 | if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH); |
| 1689 | ecode++; | ecode++; |
| 1690 | break; | break; |
| 1691 | ||
| # | Line 1402 for (;;) | Line 1694 for (;;) |
| 1694 | case OP_EODN: | case OP_EODN: |
| 1695 | if (eptr != md->end_subject && | if (eptr != md->end_subject && |
| 1696 | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) |
| 1697 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1698 | ecode++; | ecode++; |
| 1699 | break; | break; |
| 1700 | ||
| # | Line 1414 for (;;) | Line 1706 for (;;) |
| 1706 | ||
| 1707 | /* Find out if the previous and current characters are "word" characters. | /* Find out if the previous and current characters are "word" characters. |
| 1708 | 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 |
| 1709 | be "non-word" characters. */ | be "non-word" characters. Remember the earliest consulted character for |
| 1710 | partial matching. */ | |
| 1711 | ||
| 1712 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 1713 | if (utf8) | if (utf8) |
| 1714 | { | { |
| 1715 | if (eptr == md->start_subject) prev_is_word = FALSE; else | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 1716 | { | { |
| 1717 | const uschar *lastptr = eptr - 1; | USPTR lastptr = eptr - 1; |
| 1718 | while((*lastptr & 0xc0) == 0x80) lastptr--; | while((*lastptr & 0xc0) == 0x80) lastptr--; |
| 1719 | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; | |
| 1720 | GETCHAR(c, lastptr); | GETCHAR(c, lastptr); |
| 1721 | prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; | prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; |
| 1722 | } | } |
| 1723 | if (eptr >= md->end_subject) cur_is_word = FALSE; else | if (eptr >= md->end_subject) |
| 1724 | { | |
| 1725 | SCHECK_PARTIAL(); | |
| 1726 | cur_is_word = FALSE; | |
| 1727 | } | |
| 1728 | else | |
| 1729 | { | { |
| 1730 | GETCHAR(c, eptr); | GETCHAR(c, eptr); |
| 1731 | cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; | cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; |
| # | Line 1435 for (;;) | Line 1734 for (;;) |
| 1734 | else | else |
| 1735 | #endif | #endif |
| 1736 | ||
| 1737 | /* More streamlined when not in UTF-8 mode */ | /* Not in UTF-8 mode */ |
| 1738 | ||
| 1739 | { | { |
| 1740 | prev_is_word = (eptr != md->start_subject) && | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 1741 | ((md->ctypes[eptr[-1]] & ctype_word) != 0); | { |
| 1742 | cur_is_word = (eptr < md->end_subject) && | if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1; |
| 1743 | ((md->ctypes[*eptr] & ctype_word) != 0); | prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0); |
| 1744 | } | |
| 1745 | if (eptr >= md->end_subject) | |
| 1746 | { | |
| 1747 | SCHECK_PARTIAL(); | |
| 1748 | cur_is_word = FALSE; | |
| 1749 | } | |
| 1750 | else cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0); | |
| 1751 | } | } |
| 1752 | ||
| 1753 | /* Now see if the situation is what we want */ | /* Now see if the situation is what we want */ |
| 1754 | ||
| 1755 | if ((*ecode++ == OP_WORD_BOUNDARY)? | if ((*ecode++ == OP_WORD_BOUNDARY)? |
| 1756 | cur_is_word == prev_is_word : cur_is_word != prev_is_word) | cur_is_word == prev_is_word : cur_is_word != prev_is_word) |
| 1757 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1758 | } | } |
| 1759 | break; | break; |
| 1760 | ||
| 1761 | /* Match a single character type; inline for speed */ | /* Match a single character type; inline for speed */ |
| 1762 | ||
| 1763 | case OP_ANY: | case OP_ANY: |
| 1764 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); |
| 1765 | /* Fall through */ | /* Fall through */ |
| 1766 | ||
| 1767 | case OP_ALLANY: | case OP_ALLANY: |
| 1768 | if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr++ >= md->end_subject) |
| 1769 | { | |
| 1770 | SCHECK_PARTIAL(); | |
| 1771 | MRRETURN(MATCH_NOMATCH); | |
| 1772 | } | |
| 1773 | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 1774 | ecode++; | ecode++; |
| 1775 | break; | break; |
| # | Line 1468 for (;;) | Line 1778 for (;;) |
| 1778 | any byte, even newline, independent of the setting of PCRE_DOTALL. */ | any byte, even newline, independent of the setting of PCRE_DOTALL. */ |
| 1779 | ||
| 1780 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 1781 | if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr++ >= md->end_subject) |
| 1782 | { | |
| 1783 | SCHECK_PARTIAL(); | |
| 1784 | MRRETURN(MATCH_NOMATCH); | |
| 1785 | } | |
| 1786 | ecode++; | ecode++; |
| 1787 | break; | break; |
| 1788 | ||
| 1789 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 1790 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1791 | { | |
| 1792 | SCHECK_PARTIAL(); | |
| 1793 | MRRETURN(MATCH_NOMATCH); | |
| 1794 | } | |
| 1795 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1796 | if ( | if ( |
| 1797 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1481 for (;;) | Line 1799 for (;;) |
| 1799 | #endif | #endif |
| 1800 | (md->ctypes[c] & ctype_digit) != 0 | (md->ctypes[c] & ctype_digit) != 0 |
| 1801 | ) | ) |
| 1802 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1803 | ecode++; | ecode++; |
| 1804 | break; | break; |
| 1805 | ||
| 1806 | case OP_DIGIT: | case OP_DIGIT: |
| 1807 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1808 | { | |
| 1809 | SCHECK_PARTIAL(); | |
| 1810 | MRRETURN(MATCH_NOMATCH); | |
| 1811 | } | |
| 1812 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1813 | if ( | if ( |
| 1814 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1494 for (;;) | Line 1816 for (;;) |
| 1816 | #endif | #endif |
| 1817 | (md->ctypes[c] & ctype_digit) == 0 | (md->ctypes[c] & ctype_digit) == 0 |
| 1818 | ) | ) |
| 1819 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1820 | ecode++; | ecode++; |
| 1821 | break; | break; |
| 1822 | ||
| 1823 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 1824 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1825 | { | |
| 1826 | SCHECK_PARTIAL(); | |
| 1827 | MRRETURN(MATCH_NOMATCH); | |
| 1828 | } | |
| 1829 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1830 | if ( | if ( |
| 1831 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1507 for (;;) | Line 1833 for (;;) |
| 1833 | #endif | #endif |
| 1834 | (md->ctypes[c] & ctype_space) != 0 | (md->ctypes[c] & ctype_space) != 0 |
| 1835 | ) | ) |
| 1836 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1837 | ecode++; | ecode++; |
| 1838 | break; | break; |
| 1839 | ||
| 1840 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 1841 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1842 | { | |
| 1843 | SCHECK_PARTIAL(); | |
| 1844 | MRRETURN(MATCH_NOMATCH); | |
| 1845 | } | |
| 1846 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1847 | if ( | if ( |
| 1848 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1520 for (;;) | Line 1850 for (;;) |
| 1850 | #endif | #endif |
| 1851 | (md->ctypes[c] & ctype_space) == 0 | (md->ctypes[c] & ctype_space) == 0 |
| 1852 | ) | ) |
| 1853 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1854 | ecode++; | ecode++; |
| 1855 | break; | break; |
| 1856 | ||
| 1857 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 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 1533 for (;;) | Line 1867 for (;;) |
| 1867 | #endif | #endif |
| 1868 | (md->ctypes[c] & ctype_word) != 0 | (md->ctypes[c] & ctype_word) != 0 |
| 1869 | ) | ) |
| 1870 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1871 | ecode++; | ecode++; |
| 1872 | break; | break; |
| 1873 | ||
| 1874 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 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 1546 for (;;) | Line 1884 for (;;) |
| 1884 | #endif | #endif |
| 1885 | (md->ctypes[c] & ctype_word) == 0 | (md->ctypes[c] & ctype_word) == 0 |
| 1886 | ) | ) |
| 1887 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1888 | ecode++; | ecode++; |
| 1889 | break; | break; |
| 1890 | ||
| 1891 | case OP_ANYNL: | case OP_ANYNL: |
| 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 | switch(c) | switch(c) |
| 1899 | { | { |
| 1900 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 1901 | case 0x000d: | case 0x000d: |
| 1902 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 1903 | break; | break; |
| # | Line 1568 for (;;) | Line 1910 for (;;) |
| 1910 | case 0x0085: | case 0x0085: |
| 1911 | case 0x2028: | case 0x2028: |
| 1912 | case 0x2029: | case 0x2029: |
| 1913 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 1914 | break; | break; |
| 1915 | } | } |
| 1916 | ecode++; | ecode++; |
| 1917 | break; | break; |
| 1918 | ||
| 1919 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 1920 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1921 | { | |
| 1922 | SCHECK_PARTIAL(); | |
| 1923 | MRRETURN(MATCH_NOMATCH); | |
| 1924 | } | |
| 1925 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1926 | switch(c) | switch(c) |
| 1927 | { | { |
| # | Line 1599 for (;;) | Line 1945 for (;;) |
| 1945 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 1946 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 1947 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 1948 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1949 | } | } |
| 1950 | ecode++; | ecode++; |
| 1951 | break; | break; |
| 1952 | ||
| 1953 | case OP_HSPACE: | case OP_HSPACE: |
| 1954 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1955 | { | |
| 1956 | SCHECK_PARTIAL(); | |
| 1957 | MRRETURN(MATCH_NOMATCH); | |
| 1958 | } | |
| 1959 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1960 | switch(c) | switch(c) |
| 1961 | { | { |
| 1962 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 1963 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 1964 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 1965 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 1635 for (;;) | Line 1985 for (;;) |
| 1985 | break; | break; |
| 1986 | ||
| 1987 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 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 1647 for (;;) | Line 2001 for (;;) |
| 2001 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 2002 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 2003 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 2004 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2005 | } | } |
| 2006 | ecode++; | ecode++; |
| 2007 | break; | break; |
| 2008 | ||
| 2009 | case OP_VSPACE: | case OP_VSPACE: |
| 2010 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2011 | { | |
| 2012 | SCHECK_PARTIAL(); | |
| 2013 | MRRETURN(MATCH_NOMATCH); | |
| 2014 | } | |
| 2015 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2016 | switch(c) | switch(c) |
| 2017 | { | { |
| 2018 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2019 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 2020 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 2021 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 1676 for (;;) | Line 2034 for (;;) |
| 2034 | ||
| 2035 | case OP_PROP: | case OP_PROP: |
| 2036 | case OP_NOTPROP: | case OP_NOTPROP: |
| 2037 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2038 | { | |
| 2039 | SCHECK_PARTIAL(); | |
| 2040 | MRRETURN(MATCH_NOMATCH); | |
| 2041 | } | |
| 2042 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2043 | { | { |
| 2044 | const ucd_record * prop = GET_UCD(c); | const ucd_record *prop = GET_UCD(c); |
| 2045 | ||
| 2046 | switch(ecode[1]) | switch(ecode[1]) |
| 2047 | { | { |
| 2048 | case PT_ANY: | case PT_ANY: |
| 2049 | if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); | if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH); |
| 2050 | break; | break; |
| 2051 | ||
| 2052 | case PT_LAMP: | case PT_LAMP: |
| 2053 | if ((prop->chartype == ucp_Lu || | if ((prop->chartype == ucp_Lu || |
| 2054 | prop->chartype == ucp_Ll || | prop->chartype == ucp_Ll || |
| 2055 | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) |
| 2056 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2057 | break; | break; |
| 2058 | ||
| 2059 | case PT_GC: | case PT_GC: |
| 2060 | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) |
| 2061 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2062 | break; | break; |
| 2063 | ||
| 2064 | case PT_PC: | case PT_PC: |
| 2065 | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) |
| 2066 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2067 | break; | break; |
| 2068 | ||
| 2069 | case PT_SC: | case PT_SC: |
| 2070 | if ((ecode[2] != prop->script) == (op == OP_PROP)) | if ((ecode[2] != prop->script) == (op == OP_PROP)) |
| 2071 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2072 | break; | break; |
| 2073 | ||
| 2074 | default: | default: |
| # | Line 1721 for (;;) | Line 2083 for (;;) |
| 2083 | is in the binary; otherwise a compile-time error occurs. */ | is in the binary; otherwise a compile-time error occurs. */ |
| 2084 | ||
| 2085 | case OP_EXTUNI: | case OP_EXTUNI: |
| 2086 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2087 | { | |
| 2088 | SCHECK_PARTIAL(); | |
| 2089 | MRRETURN(MATCH_NOMATCH); | |
| 2090 | } | |
| 2091 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2092 | { | { |
| 2093 | int category = UCD_CATEGORY(c); | int category = UCD_CATEGORY(c); |
| 2094 | if (category == ucp_M) RRETURN(MATCH_NOMATCH); | if (category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 2095 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 2096 | { | { |
| 2097 | int len = 1; | int len = 1; |
| # | Line 1801 for (;;) | Line 2167 for (;;) |
| 2167 | break; | break; |
| 2168 | ||
| 2169 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| 2170 | if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); | if (!match_ref(offset, eptr, length, md, ims)) |
| 2171 | { | |
| 2172 | CHECK_PARTIAL(); | |
| 2173 | MRRETURN(MATCH_NOMATCH); | |
| 2174 | } | |
| 2175 | eptr += length; | eptr += length; |
| 2176 | continue; /* With the main loop */ | continue; /* With the main loop */ |
| 2177 | } | } |
| # | Line 1817 for (;;) | Line 2187 for (;;) |
| 2187 | ||
| 2188 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2189 | { | { |
| 2190 | if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); | if (!match_ref(offset, eptr, length, md, ims)) |
| 2191 | { | |
| 2192 | CHECK_PARTIAL(); | |
| 2193 | MRRETURN(MATCH_NOMATCH); | |
| 2194 | } | |
| 2195 | eptr += length; | eptr += length; |
| 2196 | } | } |
| 2197 | ||
| # | Line 1834 for (;;) | Line 2208 for (;;) |
| 2208 | { | { |
| 2209 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14); |
| 2210 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2211 | if (fi >= max || !match_ref(offset, eptr, length, md, ims)) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2212 | RRETURN(MATCH_NOMATCH); | if (!match_ref(offset, eptr, length, md, ims)) |
| 2213 | { | |
| 2214 | CHECK_PARTIAL(); | |
| 2215 | MRRETURN(MATCH_NOMATCH); | |
| 2216 | } | |
| 2217 | eptr += length; | eptr += length; |
| 2218 | } | } |
| 2219 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 1848 for (;;) | Line 2226 for (;;) |
| 2226 | pp = eptr; | pp = eptr; |
| 2227 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2228 | { | { |
| 2229 | if (!match_ref(offset, eptr, length, md, ims)) break; | if (!match_ref(offset, eptr, length, md, ims)) |
| 2230 | { | |
| 2231 | CHECK_PARTIAL(); | |
| 2232 | break; | |
| 2233 | } | |
| 2234 | eptr += length; | eptr += length; |
| 2235 | } | } |
| 2236 | while (eptr >= pp) | while (eptr >= pp) |
| # | Line 1857 for (;;) | Line 2239 for (;;) |
| 2239 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2240 | eptr -= length; | eptr -= length; |
| 2241 | } | } |
| 2242 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2243 | } | } |
| 2244 | } | } |
| 2245 | /* Control never gets here */ | /* Control never gets here */ |
| 2246 | ||
| 2247 | /* Match a bit-mapped character class, possibly repeatedly. This op code is | /* Match a bit-mapped character class, possibly repeatedly. This op code is |
| 2248 | 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, |
| 2249 | 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 1918 for (;;) | Line 2298 for (;;) |
| 2298 | { | { |
| 2299 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2300 | { | { |
| 2301 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2302 | { | |
| 2303 | SCHECK_PARTIAL(); | |
| 2304 | MRRETURN(MATCH_NOMATCH); | |
| 2305 | } | |
| 2306 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2307 | if (c > 255) | if (c > 255) |
| 2308 | { | { |
| 2309 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); |
| 2310 | } | } |
| 2311 | else | else |
| 2312 | { | { |
| 2313 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2314 | } | } |
| 2315 | } | } |
| 2316 | } | } |
| # | Line 1936 for (;;) | Line 2320 for (;;) |
| 2320 | { | { |
| 2321 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2322 | { | { |
| 2323 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2324 | { | |
| 2325 | SCHECK_PARTIAL(); | |
| 2326 | MRRETURN(MATCH_NOMATCH); | |
| 2327 | } | |
| 2328 | c = *eptr++; | c = *eptr++; |
| 2329 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2330 | } | } |
| 2331 | } | } |
| 2332 | ||
| # | Line 1960 for (;;) | Line 2348 for (;;) |
| 2348 | { | { |
| 2349 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16); |
| 2350 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2351 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2352 | if (eptr >= md->end_subject) | |
| 2353 | { | |
| 2354 | SCHECK_PARTIAL(); | |
| 2355 | MRRETURN(MATCH_NOMATCH); | |
| 2356 | } | |
| 2357 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2358 | if (c > 255) | if (c > 255) |
| 2359 | { | { |
| 2360 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); |
| 2361 | } | } |
| 2362 | else | else |
| 2363 | { | { |
| 2364 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2365 | } | } |
| 2366 | } | } |
| 2367 | } | } |
| # | Line 1980 for (;;) | Line 2373 for (;;) |
| 2373 | { | { |
| 2374 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17); |
| 2375 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2376 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2377 | if (eptr >= md->end_subject) | |
| 2378 | { | |
| 2379 | SCHECK_PARTIAL(); | |
| 2380 | MRRETURN(MATCH_NOMATCH); | |
| 2381 | } | |
| 2382 | c = *eptr++; | c = *eptr++; |
| 2383 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2384 | } | } |
| 2385 | } | } |
| 2386 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2001 for (;;) | Line 2399 for (;;) |
| 2399 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2400 | { | { |
| 2401 | int len = 1; | int len = 1; |
| 2402 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2403 | { | |
| 2404 | SCHECK_PARTIAL(); | |
| 2405 | break; | |
| 2406 | } | |
| 2407 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 2408 | if (c > 255) | if (c > 255) |
| 2409 | { | { |
| # | Line 2027 for (;;) | Line 2429 for (;;) |
| 2429 | { | { |
| 2430 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2431 | { | { |
| 2432 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2433 | { | |
| 2434 | SCHECK_PARTIAL(); | |
| 2435 | break; | |
| 2436 | } | |
| 2437 | c = *eptr; | c = *eptr; |
| 2438 | if ((data[c/8] & (1 << (c&7))) == 0) break; | if ((data[c/8] & (1 << (c&7))) == 0) break; |
| 2439 | eptr++; | eptr++; |
| # | Line 2040 for (;;) | Line 2446 for (;;) |
| 2446 | } | } |
| 2447 | } | } |
| 2448 | ||
| 2449 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2450 | } | } |
| 2451 | } | } |
| 2452 | /* Control never gets here */ | /* Control never gets here */ |
| 2453 | ||
| 2454 | ||
| 2455 | /* Match an extended character class. This opcode is encountered only | /* Match an extended character class. This opcode is encountered only |
| 2456 | in UTF-8 mode, because that's the only time it is compiled. */ | when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8 |
| 2457 | mode, because Unicode properties are supported in non-UTF-8 mode. */ | |
| 2458 | ||
| 2459 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 2460 | case OP_XCLASS: | case OP_XCLASS: |
| # | Line 2088 for (;;) | Line 2495 for (;;) |
| 2495 | ||
| 2496 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2497 | { | { |
| 2498 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2499 | GETCHARINC(c, eptr); | { |
| 2500 | if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 2501 | MRRETURN(MATCH_NOMATCH); | |
| 2502 | } | |
| 2503 | GETCHARINCTEST(c, eptr); | |
| 2504 | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); | |
| 2505 | } | } |
| 2506 | ||
| 2507 | /* 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 2107 for (;;) | Line 2518 for (;;) |
| 2518 | { | { |
| 2519 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20); |
| 2520 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2521 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2522 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 2523 | if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); | { |
| 2524 | SCHECK_PARTIAL(); | |
| 2525 | MRRETURN(MATCH_NOMATCH); | |
| 2526 | } | |
| 2527 | GETCHARINCTEST(c, eptr); | |
| 2528 | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); | |
| 2529 | } | } |
| 2530 | /* Control never gets here */ | /* Control never gets here */ |
| 2531 | } | } |
| # | Line 2122 for (;;) | Line 2538 for (;;) |
| 2538 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2539 | { | { |
| 2540 | int len = 1; | int len = 1; |
| 2541 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2542 | GETCHARLEN(c, eptr, len); | { |
| 2543 | SCHECK_PARTIAL(); | |
| 2544 | break; | |
| 2545 | } | |
| 2546 | GETCHARLENTEST(c, eptr, len); | |
| 2547 | if (!_pcre_xclass(c, data)) break; | if (!_pcre_xclass(c, data)) break; |
| 2548 | eptr += len; | eptr += len; |
| 2549 | } | } |
| # | Line 2134 for (;;) | Line 2554 for (;;) |
| 2554 | if (eptr-- == pp) break; /* Stop if tried at original pos */ | if (eptr-- == pp) break; /* Stop if tried at original pos */ |
| 2555 | if (utf8) BACKCHAR(eptr); | if (utf8) BACKCHAR(eptr); |
| 2556 | } | } |
| 2557 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2558 | } | } |
| 2559 | ||
| 2560 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2150 for (;;) | Line 2570 for (;;) |
| 2570 | length = 1; | length = 1; |
| 2571 | ecode++; | ecode++; |
| 2572 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 2573 | if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | if (length > md->end_subject - eptr) |
| 2574 | while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 2575 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | |
| 2576 | MRRETURN(MATCH_NOMATCH); | |
| 2577 | } | |
| 2578 | while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2579 | } | } |
| 2580 | else | else |
| 2581 | #endif | #endif |
| 2582 | ||
| 2583 | /* Non-UTF-8 mode */ | /* Non-UTF-8 mode */ |
| 2584 | { | { |
| 2585 | if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); | if (md->end_subject - eptr < 1) |
| 2586 | if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 2587 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | |
| 2588 | MRRETURN(MATCH_NOMATCH); | |
| 2589 | } | |
| 2590 | if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2591 | ecode += 2; | ecode += 2; |
| 2592 | } | } |
| 2593 | break; | break; |
| # | Line 2174 for (;;) | Line 2602 for (;;) |
| 2602 | ecode++; | ecode++; |
| 2603 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 2604 | ||
| 2605 | if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | if (length > md->end_subject - eptr) |
| 2606 | { | |
| 2607 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | |
| 2608 | MRRETURN(MATCH_NOMATCH); | |
| 2609 | } | |
| 2610 | ||
| 2611 | /* 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 |
| 2612 | can use the fast lookup table. */ | can use the fast lookup table. */ |
| 2613 | ||
| 2614 | if (fc < 128) | if (fc < 128) |
| 2615 | { | { |
| 2616 | if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); |
| 2617 | } | } |
| 2618 | ||
| 2619 | /* Otherwise we must pick up the subject character */ | /* Otherwise we must pick up the subject character */ |
| # | Line 2200 for (;;) | Line 2632 for (;;) |
| 2632 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2633 | if (dc != UCD_OTHERCASE(fc)) | if (dc != UCD_OTHERCASE(fc)) |
| 2634 | #endif | #endif |
| 2635 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2636 | } | } |
| 2637 | } | } |
| 2638 | } | } |
| # | Line 2209 for (;;) | Line 2641 for (;;) |
| 2641 | ||
| 2642 | /* Non-UTF-8 mode */ | /* Non-UTF-8 mode */ |
| 2643 | { | { |
| 2644 | if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); | if (md->end_subject - eptr < 1) |
| 2645 | if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 2646 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | |
| 2647 | MRRETURN(MATCH_NOMATCH); | |
| 2648 | } | |
| 2649 | if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2650 | ecode += 2; | ecode += 2; |
| 2651 | } | } |
| 2652 | break; | break; |
| # | Line 2263 for (;;) | Line 2699 for (;;) |
| 2699 | case OP_MINQUERY: | case OP_MINQUERY: |
| 2700 | c = *ecode++ - OP_STAR; | c = *ecode++ - OP_STAR; |
| 2701 | minimize = (c & 1) != 0; | minimize = (c & 1) != 0; |
| 2702 | ||
| 2703 | min = rep_min[c]; /* Pick up values from tables; */ | min = rep_min[c]; /* Pick up values from tables; */ |
| 2704 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 2705 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2706 | ||
| 2707 | /* 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. */ | ||
| 2708 | ||
| 2709 | REPEATCHAR: | REPEATCHAR: |
| 2710 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 2278 for (;;) | Line 2713 for (;;) |
| 2713 | length = 1; | length = 1; |
| 2714 | charptr = ecode; | charptr = ecode; |
| 2715 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 2716 | ecode += length; | ecode += length; |
| 2717 | ||
| 2718 | /* Handle multibyte character matching specially here. There is | /* Handle multibyte character matching specially here. There is |
| # | Line 2296 for (;;) | Line 2730 for (;;) |
| 2730 | ||
| 2731 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2732 | { | { |
| 2733 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | if (eptr <= md->end_subject - length && |
| 2734 | memcmp(eptr, charptr, length) == 0) eptr += length; | |
| 2735 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2736 | /* Need braces because of following else */ | else if (oclength > 0 && |
| 2737 | else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } | eptr <= md->end_subject - oclength && |
| 2738 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2739 | #endif /* SUPPORT_UCP */ | |
| 2740 | else | else |
| 2741 | { | { |
| 2742 | if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2743 | eptr += oclength; | MRRETURN(MATCH_NOMATCH); |
| 2744 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else { RRETURN(MATCH_NOMATCH); } | ||
| #endif /* SUPPORT_UCP */ | ||
| 2745 | } | } |
| 2746 | ||
| 2747 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2318 for (;;) | Line 2752 for (;;) |
| 2752 | { | { |
| 2753 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22); |
| 2754 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2755 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2756 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | if (eptr <= md->end_subject - length && |
| 2757 | memcmp(eptr, charptr, length) == 0) eptr += length; | |
| 2758 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2759 | /* Need braces because of following else */ | else if (oclength > 0 && |
| 2760 | else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } | eptr <= md->end_subject - oclength && |
| 2761 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2762 | #endif /* SUPPORT_UCP */ | |
| 2763 | else | else |
| 2764 | { | { |
| 2765 | if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2766 | eptr += oclength; | MRRETURN(MATCH_NOMATCH); |
| 2767 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else { RRETURN (MATCH_NOMATCH); } | ||
| #endif /* SUPPORT_UCP */ | ||
| 2768 | } | } |
| 2769 | /* Control never gets here */ | /* Control never gets here */ |
| 2770 | } | } |
| # | Line 2340 for (;;) | Line 2774 for (;;) |
| 2774 | pp = eptr; | pp = eptr; |
| 2775 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2776 | { | { |
| 2777 | if (eptr > md->end_subject - length) break; | if (eptr <= md->end_subject - length && |
| 2778 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, length) == 0) eptr += length; |
| 2779 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2780 | else if (oclength == 0) break; | else if (oclength > 0 && |
| 2781 | eptr <= md->end_subject - oclength && | |
| 2782 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2783 | #endif /* SUPPORT_UCP */ | |
| 2784 | else | else |
| 2785 | { | { |
| 2786 | if (memcmp(eptr, occhars, oclength) != 0) break; | CHECK_PARTIAL(); |
| 2787 | eptr += oclength; | break; |
| 2788 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else break; | ||
| #endif /* SUPPORT_UCP */ | ||
| 2789 | } | } |
| 2790 | ||
| 2791 | if (possessive) continue; | if (possessive) continue; |
| 2792 | ||
| 2793 | for(;;) | for(;;) |
| 2794 | { | { |
| 2795 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23); |
| 2796 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2797 | if (eptr == pp) RRETURN(MATCH_NOMATCH); | if (eptr == pp) { MRRETURN(MATCH_NOMATCH); } |
| 2798 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2799 | eptr--; | eptr--; |
| 2800 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 2801 | #else /* without SUPPORT_UCP */ | #else /* without SUPPORT_UCP */ |
| 2802 | eptr -= length; | eptr -= length; |
| 2803 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 2804 | } | } |
| 2805 | } | } |
| 2806 | /* Control never gets here */ | /* Control never gets here */ |
| 2807 | } | } |
| # | Line 2379 for (;;) | Line 2814 for (;;) |
| 2814 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF8 */ |
| 2815 | ||
| 2816 | /* When not in UTF-8 mode, load a single-byte character. */ | /* When not in UTF-8 mode, load a single-byte character. */ |
| 2817 | { | |
| 2818 | if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | fc = *ecode++; |
| fc = *ecode++; | ||
| } | ||
| 2819 | ||
| 2820 | /* 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 |
| 2821 | 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 2400 for (;;) | Line 2833 for (;;) |
| 2833 | { | { |
| 2834 | fc = md->lcc[fc]; | fc = md->lcc[fc]; |
| 2835 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2836 | if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 2837 | if (eptr >= md->end_subject) | |
| 2838 | { | |
| 2839 | SCHECK_PARTIAL(); | |
| 2840 | MRRETURN(MATCH_NOMATCH); | |
| 2841 | } | |
| 2842 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2843 | } | |
| 2844 | if (min == max) continue; | if (min == max) continue; |
| 2845 | if (minimize) | if (minimize) |
| 2846 | { | { |
| # | Line 2408 for (;;) | Line 2848 for (;;) |
| 2848 | { | { |
| 2849 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24); |
| 2850 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2851 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2852 | fc != md->lcc[*eptr++]) | if (eptr >= md->end_subject) |
| 2853 | RRETURN(MATCH_NOMATCH); | { |
| 2854 | SCHECK_PARTIAL(); | |
| 2855 | MRRETURN(MATCH_NOMATCH); | |
| 2856 | } | |
| 2857 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2858 | } | } |
| 2859 | /* Control never gets here */ | /* Control never gets here */ |
| 2860 | } | } |
| # | Line 2419 for (;;) | Line 2863 for (;;) |
| 2863 | pp = eptr; | pp = eptr; |
| 2864 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2865 | { | { |
| 2866 | if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break; | if (eptr >= md->end_subject) |
| 2867 | { | |
| 2868 | SCHECK_PARTIAL(); | |
| 2869 | break; | |
| 2870 | } | |
| 2871 | if (fc != md->lcc[*eptr]) break; | |
| 2872 | eptr++; | eptr++; |
| 2873 | } | } |
| 2874 | ||
| 2875 | if (possessive) continue; | if (possessive) continue; |
| 2876 | ||
| 2877 | while (eptr >= pp) | while (eptr >= pp) |
| 2878 | { | { |
| 2879 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25); |
| 2880 | eptr--; | eptr--; |
| 2881 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2882 | } | } |
| 2883 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2884 | } | } |
| 2885 | /* Control never gets here */ | /* Control never gets here */ |
| 2886 | } | } |
| # | Line 2438 for (;;) | Line 2889 for (;;) |
| 2889 | ||
| 2890 | else | else |
| 2891 | { | { |
| 2892 | for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH); | for (i = 1; i <= min; i++) |
| 2893 | { | |
| 2894 | if (eptr >= md->end_subject) | |
| 2895 | { | |
| 2896 | SCHECK_PARTIAL(); | |
| 2897 | MRRETURN(MATCH_NOMATCH); | |
| 2898 | } | |
| 2899 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2900 | } | |
| 2901 | ||
| 2902 | if (min == max) continue; | if (min == max) continue; |
| 2903 | ||
| 2904 | if (minimize) | if (minimize) |
| 2905 | { | { |
| 2906 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 2907 | { | { |
| 2908 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26); |
| 2909 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2910 | if (fi >= max || eptr >= md->end_subject || fc != *eptr++) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2911 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2912 | { | |
| 2913 | SCHECK_PARTIAL(); | |
| 2914 | MRRETURN(MATCH_NOMATCH); | |
| 2915 | } | |
| 2916 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2917 | } | } |
| 2918 | /* Control never gets here */ | /* Control never gets here */ |
| 2919 | } | } |
| # | Line 2456 for (;;) | Line 2922 for (;;) |
| 2922 | pp = eptr; | pp = eptr; |
| 2923 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2924 | { | { |
| 2925 | if (eptr >= md->end_subject || fc != *eptr) break; | if (eptr >= md->end_subject) |
| 2926 | { | |
| 2927 | SCHECK_PARTIAL(); | |
| 2928 | break; | |
| 2929 | } | |
| 2930 | if (fc != *eptr) break; | |
| 2931 | eptr++; | eptr++; |
| 2932 | } | } |
| 2933 | if (possessive) continue; | if (possessive) continue; |
| 2934 | ||
| 2935 | while (eptr >= pp) | while (eptr >= pp) |
| 2936 | { | { |
| 2937 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27); |
| 2938 | eptr--; | eptr--; |
| 2939 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2940 | } | } |
| 2941 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2942 | } | } |
| 2943 | } | } |
| 2944 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2475 for (;;) | Line 2947 for (;;) |
| 2947 | checking can be multibyte. */ | checking can be multibyte. */ |
| 2948 | ||
| 2949 | case OP_NOT: | case OP_NOT: |
| 2950 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2951 | { | |
| 2952 | SCHECK_PARTIAL(); | |
| 2953 | MRRETURN(MATCH_NOMATCH); | |
| 2954 | } | |
| 2955 | ecode++; | ecode++; |
| 2956 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2957 | if ((ims & PCRE_CASELESS) != 0) | if ((ims & PCRE_CASELESS) != 0) |
| # | Line 2484 for (;;) | Line 2960 for (;;) |
| 2960 | if (c < 256) | if (c < 256) |
| 2961 | #endif | #endif |
| 2962 | c = md->lcc[c]; | c = md->lcc[c]; |
| 2963 | if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH); | if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH); |
| 2964 | } | } |
| 2965 | else | else |
| 2966 | { | { |
| 2967 | if (*ecode++ == c) RRETURN(MATCH_NOMATCH); | if (*ecode++ == c) MRRETURN(MATCH_NOMATCH); |
| 2968 | } | } |
| 2969 | break; | break; |
| 2970 | ||
| # | Line 2552 for (;;) | Line 3028 for (;;) |
| 3028 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 3029 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 3030 | ||
| 3031 | /* 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. */ | ||
| 3032 | ||
| 3033 | REPEATNOTCHAR: | REPEATNOTCHAR: |
| if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 3034 | fc = *ecode++; | fc = *ecode++; |
| 3035 | ||
| 3036 | /* 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 2582 for (;;) | Line 3055 for (;;) |
| 3055 | register unsigned int d; | register unsigned int d; |
| 3056 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3057 | { | { |
| 3058 | if (eptr >= md->end_subject) | |
| 3059 | { | |
| 3060 | SCHECK_PARTIAL(); | |
| 3061 | MRRETURN(MATCH_NOMATCH); | |
| 3062 | } | |
| 3063 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3064 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3065 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3066 | } | } |
| 3067 | } | } |
| 3068 | else | else |
| # | Line 2593 for (;;) | Line 3071 for (;;) |
| 3071 | /* Not UTF-8 mode */ | /* Not UTF-8 mode */ |
| 3072 | { | { |
| 3073 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3074 | if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 3075 | if (eptr >= md->end_subject) | |
| 3076 | { | |
| 3077 | SCHECK_PARTIAL(); | |
| 3078 | MRRETURN(MATCH_NOMATCH); | |
| 3079 | } | |
| 3080 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3081 | } | |
| 3082 | } | } |
| 3083 | ||
| 3084 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2609 for (;;) | Line 3094 for (;;) |
| 3094 | { | { |
| 3095 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28); |
| 3096 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3097 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3098 | if (eptr >= md->end_subject) | |
| 3099 | { | |
| 3100 | SCHECK_PARTIAL(); | |
| 3101 | MRRETURN(MATCH_NOMATCH); | |
| 3102 | } | |
| 3103 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3104 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3105 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3106 | } | } |
| 3107 | } | } |
| 3108 | else | else |
| # | Line 2624 for (;;) | Line 3113 for (;;) |
| 3113 | { | { |
| 3114 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29); |
| 3115 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3116 | if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++]) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3117 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3118 | { | |
| 3119 | SCHECK_PARTIAL(); | |
| 3120 | MRRETURN(MATCH_NOMATCH); | |
| 3121 | } | |
| 3122 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3123 | } | } |
| 3124 | } | } |
| 3125 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2645 for (;;) | Line 3139 for (;;) |
| 3139 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3140 | { | { |
| 3141 | int len = 1; | int len = 1; |
| 3142 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 3143 | { | |
| 3144 | SCHECK_PARTIAL(); | |
| 3145 | break; | |
| 3146 | } | |
| 3147 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3148 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3149 | if (fc == d) break; | if (fc == d) break; |
| # | Line 2666 for (;;) | Line 3164 for (;;) |
| 3164 | { | { |
| 3165 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3166 | { | { |
| 3167 | if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break; | if (eptr >= md->end_subject) |
| 3168 | { | |
| 3169 | SCHECK_PARTIAL(); | |
| 3170 | break; | |
| 3171 | } | |
| 3172 | if (fc == md->lcc[*eptr]) break; | |
| 3173 | eptr++; | eptr++; |
| 3174 | } | } |
| 3175 | if (possessive) continue; | if (possessive) continue; |
| # | Line 2678 for (;;) | Line 3181 for (;;) |
| 3181 | } | } |
| 3182 | } | } |
| 3183 | ||
| 3184 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3185 | } | } |
| 3186 | /* Control never gets here */ | /* Control never gets here */ |
| 3187 | } | } |
| # | Line 2694 for (;;) | Line 3197 for (;;) |
| 3197 | register unsigned int d; | register unsigned int d; |
| 3198 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3199 | { | { |
| 3200 | if (eptr >= md->end_subject) | |
| 3201 | { | |
| 3202 | SCHECK_PARTIAL(); | |
| 3203 | MRRETURN(MATCH_NOMATCH); | |
| 3204 | } | |
| 3205 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3206 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3207 | } | } |
| 3208 | } | } |
| 3209 | else | else |
| # | Line 2703 for (;;) | Line 3211 for (;;) |
| 3211 | /* Not UTF-8 mode */ | /* Not UTF-8 mode */ |
| 3212 | { | { |
| 3213 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3214 | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 3215 | if (eptr >= md->end_subject) | |
| 3216 | { | |
| 3217 | SCHECK_PARTIAL(); | |
| 3218 | MRRETURN(MATCH_NOMATCH); | |
| 3219 | } | |
| 3220 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3221 | } | |
| 3222 | } | } |
| 3223 | ||
| 3224 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2719 for (;;) | Line 3234 for (;;) |
| 3234 | { | { |
| 3235 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32); |
| 3236 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3237 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3238 | if (eptr >= md->end_subject) | |
| 3239 | { | |
| 3240 | SCHECK_PARTIAL(); | |
| 3241 | MRRETURN(MATCH_NOMATCH); | |
| 3242 | } | |
| 3243 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3244 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3245 | } | } |
| 3246 | } | } |
| 3247 | else | else |
| # | Line 2732 for (;;) | Line 3252 for (;;) |
| 3252 | { | { |
| 3253 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33); |
| 3254 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3255 | if (fi >= max || eptr >= md->end_subject || fc == *eptr++) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3256 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3257 | { | |
| 3258 | SCHECK_PARTIAL(); | |
| 3259 | MRRETURN(MATCH_NOMATCH); | |
| 3260 | } | |
| 3261 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3262 | } | } |
| 3263 | } | } |
| 3264 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2753 for (;;) | Line 3278 for (;;) |
| 3278 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3279 | { | { |
| 3280 | int len = 1; | int len = 1; |
| 3281 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 3282 | { | |
| 3283 | SCHECK_PARTIAL(); | |
| 3284 | break; | |
| 3285 | } | |
| 3286 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3287 | if (fc == d) break; | if (fc == d) break; |
| 3288 | eptr += len; | eptr += len; |
| # | Line 2773 for (;;) | Line 3302 for (;;) |
| 3302 | { | { |
| 3303 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3304 | { | { |
| 3305 | if (eptr >= md->end_subject || fc == *eptr) break; | if (eptr >= md->end_subject) |
| 3306 | { | |
| 3307 | SCHECK_PARTIAL(); | |
| 3308 | break; | |
| 3309 | } | |
| 3310 | if (fc == *eptr) break; | |
| 3311 | eptr++; | eptr++; |
| 3312 | } | } |
| 3313 | if (possessive) continue; | if (possessive) continue; |
| # | Line 2785 for (;;) | Line 3319 for (;;) |
| 3319 | } | } |
| 3320 | } | } |
| 3321 | ||
| 3322 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3323 | } | } |
| 3324 | } | } |
| 3325 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2867 for (;;) | Line 3401 for (;;) |
| 3401 | ||
| 3402 | /* First, ensure the minimum number of matches are present. Use inline | /* First, ensure the minimum number of matches are present. Use inline |
| 3403 | 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 |
| 3404 | (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 | ||
| 3405 | 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 |
| 3406 | and single-bytes. */ | and single-bytes. */ |
| 3407 | ||
| if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 3408 | if (min > 0) | if (min > 0) |
| 3409 | { | { |
| 3410 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| # | Line 2882 for (;;) | Line 3413 for (;;) |
| 3413 | switch(prop_type) | switch(prop_type) |
| 3414 | { | { |
| 3415 | case PT_ANY: | case PT_ANY: |
| 3416 | if (prop_fail_result) RRETURN(MATCH_NOMATCH); | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); |
| 3417 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3418 | { | { |
| 3419 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3420 | { | |
| 3421 | SCHECK_PARTIAL(); | |
| 3422 | MRRETURN(MATCH_NOMATCH); | |
| 3423 | } | |
| 3424 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3425 | } | } |
| 3426 | break; | break; |
| # | Line 2893 for (;;) | Line 3428 for (;;) |
| 3428 | case PT_LAMP: | case PT_LAMP: |
| 3429 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3430 | { | { |
| 3431 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3432 | { | |
| 3433 | SCHECK_PARTIAL(); | |
| 3434 | MRRETURN(MATCH_NOMATCH); | |
| 3435 | } | |
| 3436 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3437 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 3438 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 3439 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| 3440 | prop_chartype == ucp_Lt) == prop_fail_result) | prop_chartype == ucp_Lt) == prop_fail_result) |
| 3441 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3442 | } | } |
| 3443 | break; | break; |
| 3444 | ||
| 3445 | case PT_GC: | case PT_GC: |
| 3446 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3447 | { | { |
| 3448 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3449 | { | |
| 3450 | SCHECK_PARTIAL(); | |
| 3451 | MRRETURN(MATCH_NOMATCH); | |
| 3452 | } | |
| 3453 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3454 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3455 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| 3456 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3457 | } | } |
| 3458 | break; | break; |
| 3459 | ||
| 3460 | case PT_PC: | case PT_PC: |
| 3461 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3462 | { | { |
| 3463 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3464 | { | |
| 3465 | SCHECK_PARTIAL(); | |
| 3466 | MRRETURN(MATCH_NOMATCH); | |
| 3467 | } | |
| 3468 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3469 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 3470 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| 3471 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3472 | } | } |
| 3473 | break; | break; |
| 3474 | ||
| 3475 | case PT_SC: | case PT_SC: |
| 3476 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3477 | { | { |
| 3478 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3479 | { | |
| 3480 | SCHECK_PARTIAL(); | |
| 3481 | MRRETURN(MATCH_NOMATCH); | |
| 3482 | } | |
| 3483 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3484 | prop_script = UCD_SCRIPT(c); | prop_script = UCD_SCRIPT(c); |
| 3485 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| 3486 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3487 | } | } |
| 3488 | break; | break; |
| 3489 | ||
| # | Line 2948 for (;;) | Line 3499 for (;;) |
| 3499 | { | { |
| 3500 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3501 | { | { |
| 3502 | if (eptr >= md->end_subject) | |
| 3503 | { | |
| 3504 | SCHECK_PARTIAL(); | |
| 3505 | MRRETURN(MATCH_NOMATCH); | |
| 3506 | } | |
| 3507 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3508 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3509 | if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); | if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 3510 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 3511 | { | { |
| 3512 | int len = 1; | int len = 1; |
| 3513 | if (!utf8) c = *eptr; else | if (!utf8) c = *eptr; |
| 3514 | { | else { GETCHARLEN(c, eptr, len); } |
| GETCHARLEN(c, eptr, len); | ||
| } | ||
| 3515 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3516 | if (prop_category != ucp_M) break; | if (prop_category != ucp_M) break; |
| 3517 | eptr += len; | eptr += len; |
| # | Line 2976 for (;;) | Line 3530 for (;;) |
| 3530 | case OP_ANY: | case OP_ANY: |
| 3531 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3532 | { | { |
| 3533 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) | if (eptr >= md->end_subject) |
| 3534 | RRETURN(MATCH_NOMATCH); | { |
| 3535 | SCHECK_PARTIAL(); | |
| 3536 | MRRETURN(MATCH_NOMATCH); | |
| 3537 | } | |
| 3538 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | |
| 3539 | eptr++; | eptr++; |
| 3540 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 3541 | } | } |
| # | Line 2986 for (;;) | Line 3544 for (;;) |
| 3544 | case OP_ALLANY: | case OP_ALLANY: |
| 3545 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3546 | { | { |
| 3547 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3548 | { | |
| 3549 | SCHECK_PARTIAL(); | |
| 3550 | MRRETURN(MATCH_NOMATCH); | |
| 3551 | } | |
| 3552 | eptr++; | eptr++; |
| 3553 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 3554 | } | } |
| 3555 | break; | break; |
| 3556 | ||
| 3557 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 3558 | if (eptr > md->end_subject - min) MRRETURN(MATCH_NOMATCH); | |
| 3559 | eptr += min; | eptr += min; |
| 3560 | break; | break; |
| 3561 | ||
| 3562 | case OP_ANYNL: | case OP_ANYNL: |
| 3563 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3564 | { | { |
| 3565 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3566 | { | |
| 3567 | SCHECK_PARTIAL(); | |
| 3568 | MRRETURN(MATCH_NOMATCH); | |
| 3569 | } | |
| 3570 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3571 | switch(c) | switch(c) |
| 3572 | { | { |
| 3573 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3574 | case 0x000d: | case 0x000d: |
| 3575 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 3576 | break; | break; |
| # | Line 3016 for (;;) | Line 3583 for (;;) |
| 3583 | case 0x0085: | case 0x0085: |
| 3584 | case 0x2028: | case 0x2028: |
| 3585 | case 0x2029: | case 0x2029: |
| 3586 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 3587 | break; | break; |
| 3588 | } | } |
| 3589 | } | } |
| # | Line 3025 for (;;) | Line 3592 for (;;) |
| 3592 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 3593 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3594 | { | { |
| 3595 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3596 | { | |
| 3597 | SCHECK_PARTIAL(); | |
| 3598 | MRRETURN(MATCH_NOMATCH); | |
| 3599 | } | |
| 3600 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3601 | switch(c) | switch(c) |
| 3602 | { | { |
| # | Line 3049 for (;;) | Line 3620 for (;;) |
| 3620 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 3621 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 3622 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 3623 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3624 | } | } |
| 3625 | } | } |
| 3626 | break; | break; |
| # | Line 3057 for (;;) | Line 3628 for (;;) |
| 3628 | case OP_HSPACE: | case OP_HSPACE: |
| 3629 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3630 | { | { |
| 3631 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3632 | { | |
| 3633 | SCHECK_PARTIAL(); | |
| 3634 | MRRETURN(MATCH_NOMATCH); | |
| 3635 | } | |
| 3636 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3637 | switch(c) | switch(c) |
| 3638 | { | { |
| 3639 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3640 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 3641 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 3642 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3089 for (;;) | Line 3664 for (;;) |
| 3664 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 3665 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3666 | { | { |
| 3667 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3668 | { | |
| 3669 | SCHECK_PARTIAL(); | |
| 3670 | MRRETURN(MATCH_NOMATCH); | |
| 3671 | } | |
| 3672 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3673 | switch(c) | switch(c) |
| 3674 | { | { |
| # | Line 3101 for (;;) | Line 3680 for (;;) |
| 3680 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 3681 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 3682 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 3683 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3684 | } | } |
| 3685 | } | } |
| 3686 | break; | break; |
| # | Line 3109 for (;;) | Line 3688 for (;;) |
| 3688 | case OP_VSPACE: | case OP_VSPACE: |
| 3689 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3690 | { | { |
| 3691 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3692 | { | |
| 3693 | SCHECK_PARTIAL(); | |
| 3694 | MRRETURN(MATCH_NOMATCH); | |
| 3695 | } | |
| 3696 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3697 | switch(c) | switch(c) |
| 3698 | { | { |
| 3699 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3700 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 3701 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 3702 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3129 for (;;) | Line 3712 for (;;) |
| 3712 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 3713 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3714 | { | { |
| 3715 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3716 | { | |
| 3717 | SCHECK_PARTIAL(); | |
| 3718 | MRRETURN(MATCH_NOMATCH); | |
| 3719 | } | |
| 3720 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3721 | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) |
| 3722 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3723 | } | } |
| 3724 | break; | break; |
| 3725 | ||
| 3726 | case OP_DIGIT: | case OP_DIGIT: |
| 3727 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3728 | { | { |
| 3729 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3730 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | { |
| 3731 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3732 | MRRETURN(MATCH_NOMATCH); | |
| 3733 | } | |
| 3734 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | |
| 3735 | MRRETURN(MATCH_NOMATCH); | |
| 3736 | /* 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 */ |
| 3737 | } | } |
| 3738 | break; | break; |
| # | Line 3149 for (;;) | Line 3740 for (;;) |
| 3740 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 3741 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3742 | { | { |
| 3743 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3744 | (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0)) | { |
| 3745 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3746 | MRRETURN(MATCH_NOMATCH); | |
| 3747 | } | |
| 3748 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) | |
| 3749 | MRRETURN(MATCH_NOMATCH); | |
| 3750 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); |
| 3751 | } | } |
| 3752 | break; | break; |
| # | Line 3159 for (;;) | Line 3754 for (;;) |
| 3754 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 3755 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3756 | { | { |
| 3757 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3758 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | { |
| 3759 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3760 | MRRETURN(MATCH_NOMATCH); | |
| 3761 | } | |
| 3762 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | |
| 3763 | MRRETURN(MATCH_NOMATCH); | |
| 3764 | /* 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 */ |
| 3765 | } | } |
| 3766 | break; | break; |
| # | Line 3169 for (;;) | Line 3768 for (;;) |
| 3768 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 3769 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3770 | { | { |
| 3771 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3772 | (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0)) | { |
| 3773 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3774 | MRRETURN(MATCH_NOMATCH); | |
| 3775 | } | |
| 3776 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) | |
| 3777 | MRRETURN(MATCH_NOMATCH); | |
| 3778 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); |
| 3779 | } | } |
| 3780 | break; | break; |
| # | Line 3179 for (;;) | Line 3782 for (;;) |
| 3782 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 3783 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3784 | { | { |
| 3785 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3786 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | { |
| 3787 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3788 | MRRETURN(MATCH_NOMATCH); | |
| 3789 | } | |
| 3790 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | |
| 3791 | MRRETURN(MATCH_NOMATCH); | |
| 3792 | /* 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 */ |
| 3793 | } | } |
| 3794 | break; | break; |
| # | Line 3194 for (;;) | Line 3801 for (;;) |
| 3801 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF8 */ |
| 3802 | ||
| 3803 | /* 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 |
| 3804 | 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. */ | ||
| 3805 | ||
| 3806 | switch(ctype) | switch(ctype) |
| 3807 | { | { |
| 3808 | case OP_ANY: | case OP_ANY: |
| 3809 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3810 | { | { |
| 3811 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3812 | { | |
| 3813 | SCHECK_PARTIAL(); | |
| 3814 | MRRETURN(MATCH_NOMATCH); | |
| 3815 | } | |
| 3816 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | |
| 3817 | eptr++; | eptr++; |
| 3818 | } | } |
| 3819 | break; | break; |
| 3820 | ||
| 3821 | case OP_ALLANY: | case OP_ALLANY: |
| 3822 | if (eptr > md->end_subject - min) | |
| 3823 | { | |
| 3824 | SCHECK_PARTIAL(); | |
| 3825 | MRRETURN(MATCH_NOMATCH); | |
| 3826 | } | |
| 3827 | eptr += min; | eptr += min; |
| 3828 | break; | break; |
| 3829 | ||
| 3830 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 3831 | if (eptr > md->end_subject - min) | |
| 3832 | { | |
| 3833 | SCHECK_PARTIAL(); | |
| 3834 | MRRETURN(MATCH_NOMATCH); | |
| 3835 | } | |
| 3836 | eptr += min; | eptr += min; |
| 3837 | break; | break; |
| 3838 | ||
| /* Because of the CRLF case, we can't assume the minimum number of | ||
| bytes are present in this case. */ | ||
| 3839 | case OP_ANYNL: | case OP_ANYNL: |
| 3840 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3841 | { | { |
| 3842 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3843 | { | |
| 3844 | SCHECK_PARTIAL(); | |
| 3845 | MRRETURN(MATCH_NOMATCH); | |
| 3846 | } | |
| 3847 | switch(*eptr++) | switch(*eptr++) |
| 3848 | { | { |
| 3849 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3850 | case 0x000d: | case 0x000d: |
| 3851 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 3852 | break; | break; |
| # | Line 3234 for (;;) | Line 3856 for (;;) |
| 3856 | case 0x000b: | case 0x000b: |
| 3857 | case 0x000c: | case 0x000c: |
| 3858 | case 0x0085: | case 0x0085: |
| 3859 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 3860 | break; | break; |
| 3861 | } | } |
| 3862 | } | } |
| # | Line 3243 for (;;) | Line 3865 for (;;) |
| 3865 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 3866 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3867 | { | { |
| 3868 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3869 | { | |
| 3870 | SCHECK_PARTIAL(); | |
| 3871 | MRRETURN(MATCH_NOMATCH); | |
| 3872 | } | |
| 3873 | switch(*eptr++) | switch(*eptr++) |
| 3874 | { | { |
| 3875 | default: break; | default: break; |
| 3876 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 3877 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 3878 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 3879 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3880 | } | } |
| 3881 | } | } |
| 3882 | break; | break; |
| # | Line 3258 for (;;) | Line 3884 for (;;) |
| 3884 | case OP_HSPACE: | case OP_HSPACE: |
| 3885 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3886 | { | { |
| 3887 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3888 | { | |
| 3889 | SCHECK_PARTIAL(); | |
| 3890 | MRRETURN(MATCH_NOMATCH); | |
| 3891 | } | |
| 3892 | switch(*eptr++) | switch(*eptr++) |
| 3893 | { | { |
| 3894 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3895 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 3896 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 3897 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3273 for (;;) | Line 3903 for (;;) |
| 3903 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 3904 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3905 | { | { |
| 3906 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3907 | { | |
| 3908 | SCHECK_PARTIAL(); | |
| 3909 | MRRETURN(MATCH_NOMATCH); | |
| 3910 | } | |
| 3911 | switch(*eptr++) | switch(*eptr++) |
| 3912 | { | { |
| 3913 | default: break; | default: break; |
| # | Line 3282 for (;;) | Line 3916 for (;;) |
| 3916 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 3917 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 3918 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 3919 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3920 | } | } |
| 3921 | } | } |
| 3922 | break; | break; |
| # | Line 3290 for (;;) | Line 3924 for (;;) |
| 3924 | case OP_VSPACE: | case OP_VSPACE: |
| 3925 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3926 | { | { |
| 3927 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3928 | { | |
| 3929 | SCHECK_PARTIAL(); | |
| 3930 | MRRETURN(MATCH_NOMATCH); | |
| 3931 | } | |
| 3932 | switch(*eptr++) | switch(*eptr++) |
| 3933 | { | { |
| 3934 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3935 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 3936 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 3937 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3306 for (;;) | Line 3944 for (;;) |
| 3944 | ||
| 3945 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 3946 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3947 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); | { |
| 3948 | if (eptr >= md->end_subject) | |
| 3949 | { | |
| 3950 | SCHECK_PARTIAL(); | |
| 3951 | MRRETURN(MATCH_NOMATCH); | |
| 3952 | } | |
| 3953 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); | |
| 3954 | } | |
| 3955 | break; | break; |
| 3956 | ||
| 3957 | case OP_DIGIT: | case OP_DIGIT: |
| 3958 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3959 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); | { |
| 3960 | if (eptr >= md->end_subject) | |
| 3961 | { | |
| 3962 | SCHECK_PARTIAL(); | |
| 3963 | MRRETURN(MATCH_NOMATCH); | |
| 3964 | } | |
| 3965 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); | |
| 3966 | } | |
| 3967 | break; | break; |
| 3968 | ||
| 3969 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 3970 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3971 | if ((md->ctypes[*eptr++] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); | { |
| 3972 | if (eptr >= md->end_subject) | |
| 3973 | { | |
| 3974 | SCHECK_PARTIAL(); | |
| 3975 | MRRETURN(MATCH_NOMATCH); | |
| 3976 | } | |
| 3977 | if ((md->ctypes[*eptr++] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); | |
| 3978 | } | |
| 3979 | break; | break; |
| 3980 | ||
| 3981 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 3982 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3983 | if ((md->ctypes[*eptr++] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); | { |
| 3984 | if (eptr >= md->end_subject) | |
| 3985 | { | |
| 3986 | SCHECK_PARTIAL(); | |
| 3987 | MRRETURN(MATCH_NOMATCH); | |
| 3988 | } | |
| 3989 | if ((md->ctypes[*eptr++] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); | |
| 3990 | } | |
| 3991 | break; | break; |
| 3992 | ||
| 3993 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 3994 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3995 | { | |
| 3996 | if (eptr >= md->end_subject) | |
| 3997 | { | |
| 3998 | SCHECK_PARTIAL(); | |
| 3999 | MRRETURN(MATCH_NOMATCH); | |
| 4000 | } | |
| 4001 | if ((md->ctypes[*eptr++] & ctype_word) != 0) | if ((md->ctypes[*eptr++] & ctype_word) != 0) |
| 4002 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4003 | } | |
| 4004 | break; | break; |
| 4005 | ||
| 4006 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4007 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4008 | { | |
| 4009 | if (eptr >= md->end_subject) | |
| 4010 | { | |
| 4011 | SCHECK_PARTIAL(); | |
| 4012 | MRRETURN(MATCH_NOMATCH); | |
| 4013 | } | |
| 4014 | if ((md->ctypes[*eptr++] & ctype_word) == 0) | if ((md->ctypes[*eptr++] & ctype_word) == 0) |
| 4015 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4016 | } | |
| 4017 | break; | break; |
| 4018 | ||
| 4019 | default: | default: |
| # | Line 3361 for (;;) | Line 4041 for (;;) |
| 4041 | { | { |
| 4042 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36); |
| 4043 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4044 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4045 | if (eptr >= md->end_subject) | |
| 4046 | { | |
| 4047 | SCHECK_PARTIAL(); | |
| 4048 | MRRETURN(MATCH_NOMATCH); | |
| 4049 | } | |
| 4050 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4051 | if (prop_fail_result) RRETURN(MATCH_NOMATCH); | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); |
| 4052 | } | } |
| 4053 | /* Control never gets here */ | /* Control never gets here */ |
| 4054 | ||
| # | Line 3372 for (;;) | Line 4057 for (;;) |
| 4057 | { | { |
| 4058 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37); |
| 4059 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4060 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4061 | if (eptr >= md->end_subject) | |
| 4062 | { | |
| 4063 | SCHECK_PARTIAL(); | |
| 4064 | MRRETURN(MATCH_NOMATCH); | |
| 4065 | } | |
| 4066 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4067 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4068 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 4069 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| 4070 | prop_chartype == ucp_Lt) == prop_fail_result) | prop_chartype == ucp_Lt) == prop_fail_result) |
| 4071 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4072 | } | } |
| 4073 | /* Control never gets here */ | /* Control never gets here */ |
| 4074 | ||
| # | Line 3387 for (;;) | Line 4077 for (;;) |
| 4077 | { | { |
| 4078 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38); |
| 4079 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4080 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4081 | if (eptr >= md->end_subject) | |
| 4082 | { | |
| 4083 | SCHECK_PARTIAL(); | |
| 4084 | MRRETURN(MATCH_NOMATCH); | |
| 4085 | } | |
| 4086 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4087 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4088 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| 4089 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4090 | } | } |
| 4091 | /* Control never gets here */ | /* Control never gets here */ |
| 4092 | ||
| # | Line 3400 for (;;) | Line 4095 for (;;) |
| 4095 | { | { |
| 4096 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39); |
| 4097 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4098 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4099 | if (eptr >= md->end_subject) | |
| 4100 | { | |
| 4101 | SCHECK_PARTIAL(); | |
| 4102 | MRRETURN(MATCH_NOMATCH); | |
| 4103 | } | |
| 4104 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4105 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4106 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| 4107 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4108 | } | } |
| 4109 | /* Control never gets here */ | /* Control never gets here */ |
| 4110 | ||
| # | Line 3413 for (;;) | Line 4113 for (;;) |
| 4113 | { | { |
| 4114 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40); |
| 4115 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4116 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4117 | if (eptr >= md->end_subject) | |
| 4118 | { | |
| 4119 | SCHECK_PARTIAL(); | |
| 4120 | MRRETURN(MATCH_NOMATCH); | |
| 4121 | } | |
| 4122 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4123 | prop_script = UCD_SCRIPT(c); | prop_script = UCD_SCRIPT(c); |
| 4124 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| 4125 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4126 | } | } |
| 4127 | /* Control never gets here */ | /* Control never gets here */ |
| 4128 | ||
| # | Line 3435 for (;;) | Line 4140 for (;;) |
| 4140 | { | { |
| 4141 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41); |
| 4142 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4143 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4144 | if (eptr >= md->end_subject) | |
| 4145 | { | |
| 4146 | SCHECK_PARTIAL(); | |
| 4147 | MRRETURN(MATCH_NOMATCH); | |
| 4148 | } | |
| 4149 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4150 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4151 | if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); | if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 4152 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 4153 | { | { |
| 4154 | int len = 1; | int len = 1; |
| 4155 | if (!utf8) c = *eptr; else | if (!utf8) c = *eptr; |
| 4156 | { | else { GETCHARLEN(c, eptr, len); } |
| GETCHARLEN(c, eptr, len); | ||
| } | ||
| 4157 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4158 | if (prop_category != ucp_M) break; | if (prop_category != ucp_M) break; |
| 4159 | eptr += len; | eptr += len; |
| # | Line 3464 for (;;) | Line 4172 for (;;) |
| 4172 | { | { |
| 4173 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42); |
| 4174 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4175 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4176 | (ctype == OP_ANY && IS_NEWLINE(eptr))) | if (eptr >= md->end_subject) |
| 4177 | RRETURN(MATCH_NOMATCH); | { |
| 4178 | SCHECK_PARTIAL(); | |
| 4179 | MRRETURN(MATCH_NOMATCH); | |
| 4180 | } | |
| 4181 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | |
| 4182 | MRRETURN(MATCH_NOMATCH); | |
| 4183 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4184 | switch(ctype) | switch(ctype) |
| 4185 | { | { |
| # | Line 3479 for (;;) | Line 4191 for (;;) |
| 4191 | case OP_ANYNL: | case OP_ANYNL: |
| 4192 | switch(c) | switch(c) |
| 4193 | { | { |
| 4194 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4195 | case 0x000d: | case 0x000d: |
| 4196 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4197 | break; | break; |
| # | Line 3491 for (;;) | Line 4203 for (;;) |
| 4203 | case 0x0085: | case 0x0085: |
| 4204 | case 0x2028: | case 0x2028: |
| 4205 | case 0x2029: | case 0x2029: |
| 4206 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4207 | break; | break; |
| 4208 | } | } |
| 4209 | break; | break; |
| # | Line 3519 for (;;) | Line 4231 for (;;) |
| 4231 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 4232 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 4233 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 4234 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4235 | } | } |
| 4236 | break; | break; |
| 4237 | ||
| 4238 | case OP_HSPACE: | case OP_HSPACE: |
| 4239 | switch(c) | switch(c) |
| 4240 | { | { |
| 4241 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4242 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4243 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4244 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3561 for (;;) | Line 4273 for (;;) |
| 4273 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4274 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 4275 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 4276 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4277 | } | } |
| 4278 | break; | break; |
| 4279 | ||
| 4280 | case OP_VSPACE: | case OP_VSPACE: |
| 4281 | switch(c) | switch(c) |
| 4282 | { | { |
| 4283 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4284 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4285 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4286 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3582 for (;;) | Line 4294 for (;;) |
| 4294 | ||
| 4295 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4296 | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) |
| 4297 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4298 | break; | break; |
| 4299 | ||
| 4300 | case OP_DIGIT: | case OP_DIGIT: |
| 4301 | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) |
| 4302 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4303 | break; | break; |
| 4304 | ||
| 4305 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4306 | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) |
| 4307 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4308 | break; | break; |
| 4309 | ||
| 4310 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4311 | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) |
| 4312 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4313 | break; | break; |
| 4314 | ||
| 4315 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4316 | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) |
| 4317 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4318 | break; | break; |
| 4319 | ||
| 4320 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4321 | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) |
| 4322 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4323 | break; | break; |
| 4324 | ||
| 4325 | default: | default: |
| # | Line 3623 for (;;) | Line 4335 for (;;) |
| 4335 | { | { |
| 4336 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43); |
| 4337 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4338 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4339 | (ctype == OP_ANY && IS_NEWLINE(eptr))) | if (eptr >= md->end_subject) |
| 4340 | RRETURN(MATCH_NOMATCH); | { |
| 4341 | SCHECK_PARTIAL(); | |
| 4342 | MRRETURN(MATCH_NOMATCH); | |
| 4343 | } | |
| 4344 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | |
| 4345 | MRRETURN(MATCH_NOMATCH); | |
| 4346 | c = *eptr++; | c = *eptr++; |
| 4347 | switch(ctype) | switch(ctype) |
| 4348 | { | { |
| # | Line 3638 for (;;) | Line 4354 for (;;) |
| 4354 | case OP_ANYNL: | case OP_ANYNL: |
| 4355 | switch(c) | switch(c) |
| 4356 | { | { |
| 4357 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4358 | case 0x000d: | case 0x000d: |
| 4359 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4360 | break; | break; |
| # | Line 3649 for (;;) | Line 4365 for (;;) |
| 4365 | case 0x000b: | case 0x000b: |
| 4366 | case 0x000c: | case 0x000c: |
| 4367 | case 0x0085: | case 0x0085: |
| 4368 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4369 | break; | break; |
| 4370 | } | } |
| 4371 | break; | break; |
| # | Line 3661 for (;;) | Line 4377 for (;;) |
| 4377 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4378 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4379 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4380 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4381 | } | } |
| 4382 | break; | break; |
| 4383 | ||
| 4384 | case OP_HSPACE: | case OP_HSPACE: |
| 4385 | switch(c) | switch(c) |
| 4386 | { | { |
| 4387 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4388 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4389 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4390 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3685 for (;;) | Line 4401 for (;;) |
| 4401 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4402 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4403 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4404 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4405 | } | } |
| 4406 | break; | break; |
| 4407 | ||
| 4408 | case OP_VSPACE: | case OP_VSPACE: |
| 4409 | switch(c) | switch(c) |
| 4410 | { | { |
| 4411 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4412 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4413 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4414 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3703 for (;;) | Line 4419 for (;;) |
| 4419 | break; | break; |
| 4420 | ||
| 4421 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4422 | if ((md->ctypes[c] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); |
| 4423 | break; | break; |
| 4424 | ||
| 4425 | case OP_DIGIT: | case OP_DIGIT: |
| 4426 | if ((md->ctypes[c] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); |
| 4427 | break; | break; |
| 4428 | ||
| 4429 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4430 | if ((md->ctypes[c] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); |
| 4431 | break; | break; |
| 4432 | ||
| 4433 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4434 | if ((md->ctypes[c] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); |
| 4435 | break; | break; |
| 4436 | ||
| 4437 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4438 | if ((md->ctypes[c] & ctype_word) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_word) != 0) MRRETURN(MATCH_NOMATCH); |
| 4439 | break; | break; |
| 4440 | ||
| 4441 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4442 | if ((md->ctypes[c] & ctype_word) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_word) == 0) MRRETURN(MATCH_NOMATCH); |
| 4443 | break; | break; |
| 4444 | ||
| 4445 | default: | default: |
| # | Line 3751 for (;;) | Line 4467 for (;;) |
| 4467 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4468 | { | { |
| 4469 | int len = 1; | int len = 1; |
| 4470 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4471 | { | |
| 4472 | SCHECK_PARTIAL(); | |
| 4473 | break; | |
| 4474 | } | |
| 4475 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4476 | if (prop_fail_result) break; | if (prop_fail_result) break; |
| 4477 | eptr+= len; | eptr+= len; |
| # | Line 3762 for (;;) | Line 4482 for (;;) |
| 4482 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4483 | { | { |
| 4484 | int len = 1; | int len = 1; |
| 4485 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4486 | { | |
| 4487 | SCHECK_PARTIAL(); | |
| 4488 | break; | |
| 4489 | } | |
| 4490 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4491 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4492 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| # | Line 3777 for (;;) | Line 4501 for (;;) |
| 4501 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4502 | { | { |
| 4503 | int len = 1; | int len = 1; |
| 4504 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4505 | { | |
| 4506 | SCHECK_PARTIAL(); | |
| 4507 | break; | |
| 4508 | } | |
| 4509 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4510 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4511 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| # | Line 3790 for (;;) | Line 4518 for (;;) |
| 4518 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4519 | { | { |
| 4520 | int len = 1; | int len = 1; |
| 4521 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4522 | { | |
| 4523 | SCHECK_PARTIAL(); | |
| 4524 | break; | |
| 4525 | } | |
| 4526 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4527 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4528 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| # | Line 3803 for (;;) | Line 4535 for (;;) |
| 4535 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4536 | { | { |
| 4537 | int len = 1; | int len = 1; |
| 4538 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4539 | { | |
| 4540 | SCHECK_PARTIAL(); | |
| 4541 | break; | |
| 4542 | } | |
| 4543 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4544 | prop_script = UCD_SCRIPT(c); | prop_script = UCD_SCRIPT(c); |
| 4545 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| # | Line 3832 for (;;) | Line 4568 for (;;) |
| 4568 | { | { |
| 4569 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4570 | { | { |
| 4571 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4572 | { | |
| 4573 | SCHECK_PARTIAL(); | |
| 4574 | break; | |
| 4575 | } | |
| 4576 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4577 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4578 | if (prop_category == ucp_M) break; | if (prop_category == ucp_M) break; |
| # | Line 3852 for (;;) | Line 4592 for (;;) |
| 4592 | /* eptr is now past the end of the maximum run */ | /* eptr is now past the end of the maximum run */ |
| 4593 | ||
| 4594 | if (possessive) continue; | if (possessive) continue; |
| 4595 | ||
| 4596 | for(;;) | for(;;) |
| 4597 | { | { |
| 4598 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45); |
| # | Line 3887 for (;;) | Line 4628 for (;;) |
| 4628 | { | { |
| 4629 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4630 | { | { |
| 4631 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break; | if (eptr >= md->end_subject) |
| 4632 | { | |
| 4633 | SCHECK_PARTIAL(); | |
| 4634 | break; | |
| 4635 | } | |
| 4636 | if (IS_NEWLINE(eptr)) break; | |
| 4637 | eptr++; | eptr++; |
| 4638 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 4639 | } | } |
| # | Line 3899 for (;;) | Line 4645 for (;;) |
| 4645 | { | { |
| 4646 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4647 | { | { |
| 4648 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break; | if (eptr >= md->end_subject) |
| 4649 | { | |
| 4650 | SCHECK_PARTIAL(); | |
| 4651 | break; | |
| 4652 | } | |
| 4653 | if (IS_NEWLINE(eptr)) break; | |
| 4654 | eptr++; | eptr++; |
| 4655 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 4656 | } | } |
| # | Line 3911 for (;;) | Line 4662 for (;;) |
| 4662 | { | { |
| 4663 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4664 | { | { |
| 4665 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4666 | { | |
| 4667 | SCHECK_PARTIAL(); | |
| 4668 | break; | |
| 4669 | } | |
| 4670 | eptr++; | eptr++; |
| 4671 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 4672 | } | } |
| # | Line 3924 for (;;) | Line 4679 for (;;) |
| 4679 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 4680 | c = max - min; | c = max - min; |
| 4681 | if (c > (unsigned int)(md->end_subject - eptr)) | if (c > (unsigned int)(md->end_subject - eptr)) |
| 4682 | c = md->end_subject - eptr; | { |
| 4683 | eptr += c; | eptr = md->end_subject; |
| 4684 | SCHECK_PARTIAL(); | |
| 4685 | } | |
| 4686 | else eptr += c; | |
| 4687 | break; | break; |
| 4688 | ||
| 4689 | case OP_ANYNL: | case OP_ANYNL: |
| 4690 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4691 | { | { |
| 4692 | int len = 1; | int len = 1; |
| 4693 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4694 | { | |
| 4695 | SCHECK_PARTIAL(); | |
| 4696 | break; | |
| 4697 | } | |
| 4698 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4699 | if (c == 0x000d) | if (c == 0x000d) |
| 4700 | { | { |
| # | Line 3957 for (;;) | Line 4719 for (;;) |
| 4719 | { | { |
| 4720 | BOOL gotspace; | BOOL gotspace; |
| 4721 | int len = 1; | int len = 1; |
| 4722 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4723 | { | |
| 4724 | SCHECK_PARTIAL(); | |
| 4725 | break; | |
| 4726 | } | |
| 4727 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4728 | switch(c) | switch(c) |
| 4729 | { | { |
| # | Line 3995 for (;;) | Line 4761 for (;;) |
| 4761 | { | { |
| 4762 | BOOL gotspace; | BOOL gotspace; |
| 4763 | int len = 1; | int len = 1; |
| 4764 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4765 | { | |
| 4766 | SCHECK_PARTIAL(); | |
| 4767 | break; | |
| 4768 | } | |
| 4769 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4770 | switch(c) | switch(c) |
| 4771 | { | { |
| # | Line 4019 for (;;) | Line 4789 for (;;) |
| 4789 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4790 | { | { |
| 4791 | int len = 1; | int len = 1; |
| 4792 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4793 | { | |
| 4794 | SCHECK_PARTIAL(); | |
| 4795 | break; | |
| 4796 | } | |
| 4797 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4798 | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) break; | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) break; |
| 4799 | eptr+= len; | eptr+= len; |
| # | Line 4030 for (;;) | Line 4804 for (;;) |
| 4804 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4805 | { | { |
| 4806 | int len = 1; | int len = 1; |
| 4807 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4808 | { | |
| 4809 | SCHECK_PARTIAL(); | |
| 4810 | break; | |
| 4811 | } | |
| 4812 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4813 | if (c >= 256 ||(md->ctypes[c] & ctype_digit) == 0) break; | if (c >= 256 ||(md->ctypes[c] & ctype_digit) == 0) break; |
| 4814 | eptr+= len; | eptr+= len; |
| # | Line 4041 for (;;) | Line 4819 for (;;) |
| 4819 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4820 | { | { |
| 4821 | int len = 1; | int len = 1; |
| 4822 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4823 | { | |
| 4824 | SCHECK_PARTIAL(); | |
| 4825 | break; | |
| 4826 | } | |
| 4827 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4828 | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) break; | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) break; |
| 4829 | eptr+= len; | eptr+= len; |
| # | Line 4052 for (;;) | Line 4834 for (;;) |
| 4834 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4835 | { | { |
| 4836 | int len = 1; | int len = 1; |
| 4837 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4838 | { | |
| 4839 | SCHECK_PARTIAL(); | |
| 4840 | break; | |
| 4841 | } | |
| 4842 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4843 | if (c >= 256 ||(md->ctypes[c] & ctype_space) == 0) break; | if (c >= 256 ||(md->ctypes[c] & ctype_space) == 0) break; |
| 4844 | eptr+= len; | eptr+= len; |
| # | Line 4063 for (;;) | Line 4849 for (;;) |
| 4849 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4850 | { | { |
| 4851 | int len = 1; | int len = 1; |
| 4852 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4853 | { | |
| 4854 | SCHECK_PARTIAL(); | |
| 4855 | break; | |
| 4856 | } | |
| 4857 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4858 | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) break; | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) break; |
| 4859 | eptr+= len; | eptr+= len; |
| # | Line 4074 for (;;) | Line 4864 for (;;) |
| 4864 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4865 | { | { |
| 4866 | int len = 1; | int len = 1; |
| 4867 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4868 | { | |
| 4869 | SCHECK_PARTIAL(); | |
| 4870 | break; | |
| 4871 | } | |
| 4872 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 4873 | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) break; | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) break; |
| 4874 | eptr+= len; | eptr+= len; |
| # | Line 4106 for (;;) | Line 4900 for (;;) |
| 4900 | case OP_ANY: | case OP_ANY: |
| 4901 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4902 | { | { |
| 4903 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break; | if (eptr >= md->end_subject) |
| 4904 | { | |
| 4905 | SCHECK_PARTIAL(); | |
| 4906 | break; | |
| 4907 | } | |
| 4908 | if (IS_NEWLINE(eptr)) break; | |
| 4909 | eptr++; | eptr++; |
| 4910 | } | } |
| 4911 | break; | break; |
| # | Line 4115 for (;;) | Line 4914 for (;;) |
| 4914 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 4915 | c = max - min; | c = max - min; |
| 4916 | if (c > (unsigned int)(md->end_subject - eptr)) | if (c > (unsigned int)(md->end_subject - eptr)) |
| 4917 | c = md->end_subject - eptr; | { |
| 4918 | eptr += c; | eptr = md->end_subject; |
| 4919 | SCHECK_PARTIAL(); | |
| 4920 | } | |
| 4921 | else eptr += c; | |
| 4922 | break; | break; |
| 4923 | ||
| 4924 | case OP_ANYNL: | case OP_ANYNL: |
| 4925 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4926 | { | { |
| 4927 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4928 | { | |
| 4929 | SCHECK_PARTIAL(); | |
| 4930 | break; | |
| 4931 | } | |
| 4932 | c = *eptr; | c = *eptr; |
| 4933 | if (c == 0x000d) | if (c == 0x000d) |
| 4934 | { | { |
| # | Line 4143 for (;;) | Line 4949 for (;;) |
| 4949 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 4950 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4951 | { | { |
| 4952 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4953 | { | |
| 4954 | SCHECK_PARTIAL(); | |
| 4955 | break; | |
| 4956 | } | |
| 4957 | c = *eptr; | c = *eptr; |
| 4958 | if (c == 0x09 || c == 0x20 || c == 0xa0) break; | if (c == 0x09 || c == 0x20 || c == 0xa0) break; |
| 4959 | eptr++; | eptr++; |
| # | Line 4153 for (;;) | Line 4963 for (;;) |
| 4963 | case OP_HSPACE: | case OP_HSPACE: |
| 4964 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4965 | { | { |
| 4966 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4967 | { | |
| 4968 | SCHECK_PARTIAL(); | |
| 4969 | break; | |
| 4970 | } | |
| 4971 | c = *eptr; | c = *eptr; |
| 4972 | if (c != 0x09 && c != 0x20 && c != 0xa0) break; | if (c != 0x09 && c != 0x20 && c != 0xa0) break; |
| 4973 | eptr++; | eptr++; |
| # | Line 4163 for (;;) | Line 4977 for (;;) |
| 4977 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 4978 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4979 | { | { |
| 4980 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4981 | { | |
| 4982 | SCHECK_PARTIAL(); | |
| 4983 | break; | |
| 4984 | } | |
| 4985 | c = *eptr; | c = *eptr; |
| 4986 | if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85) | if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85) |
| 4987 | break; | break; |
| # | Line 4174 for (;;) | Line 4992 for (;;) |
| 4992 | case OP_VSPACE: | case OP_VSPACE: |
| 4993 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4994 | { | { |
| 4995 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4996 | { | |
| 4997 | SCHECK_PARTIAL(); | |
| 4998 | break; | |
| 4999 | } | |
| 5000 | c = *eptr; | c = *eptr; |
| 5001 | if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85) | if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85) |
| 5002 | break; | break; |
| # | Line 4185 for (;;) | Line 5007 for (;;) |
| 5007 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 5008 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5009 | { | { |
| 5010 | if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_digit) != 0) | if (eptr >= md->end_subject) |
| 5011 | { | |
| 5012 | SCHECK_PARTIAL(); | |
| 5013 | break; | break; |
| 5014 | } | |
| 5015 | if ((md->ctypes[*eptr] & ctype_digit) != 0) break; | |
| 5016 | eptr++; | eptr++; |
| 5017 | } | } |
| 5018 | break; | break; |
| # | Line 4194 for (;;) | Line 5020 for (;;) |
| 5020 | case OP_DIGIT: | case OP_DIGIT: |
| 5021 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5022 | { | { |
| 5023 | if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_digit) == 0) | if (eptr >= md->end_subject) |
| 5024 | { | |
| 5025 | SCHECK_PARTIAL(); | |
| 5026 | break; | break; |
| 5027 | } | |
| 5028 | if ((md->ctypes[*eptr] & ctype_digit) == 0) break; | |
| 5029 | eptr++; | eptr++; |
| 5030 | } | } |
| 5031 | break; | break; |
| # | Line 4203 for (;;) | Line 5033 for (;;) |
| 5033 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 5034 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5035 | { | { |
| 5036 | if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_space) != 0) | if (eptr >= md->end_subject) |
| 5037 | { | |
| 5038 | SCHECK_PARTIAL(); | |
| 5039 | break; | break; |
| 5040 | } | |
| 5041 | if ((md->ctypes[*eptr] & ctype_space) != 0) break; | |
| 5042 | eptr++; | eptr++; |
| 5043 | } | } |
| 5044 | break; | break; |
| # | Line 4212 for (;;) | Line 5046 for (;;) |
| 5046 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 5047 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5048 | { | { |
| 5049 | if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_space) == 0) | if (eptr >= md->end_subject) |
| 5050 | { | |
| 5051 | SCHECK_PARTIAL(); | |
| 5052 | break; | break; |
| 5053 | } | |
| 5054 | if ((md->ctypes[*eptr] & ctype_space) == 0) break; | |
| 5055 | eptr++; | eptr++; |
| 5056 | } | } |
| 5057 | break; | break; |
| # | Line 4221 for (;;) | Line 5059 for (;;) |
| 5059 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 5060 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5061 | { | { |
| 5062 | if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_word) != 0) | if (eptr >= md->end_subject) |
| 5063 | { | |
| 5064 | SCHECK_PARTIAL(); | |
| 5065 | break; | break; |
| 5066 | } | |
| 5067 | if ((md->ctypes[*eptr] & ctype_word) != 0) break; | |
| 5068 | eptr++; | eptr++; |
| 5069 | } | } |
| 5070 | break; | break; |
| # | Line 4230 for (;;) | Line 5072 for (;;) |
| 5072 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 5073 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5074 | { | { |
| 5075 | if (eptr >= md->end_subject || (md->ctypes[*eptr] & ctype_word) == 0) | if (eptr >= md->end_subject) |
| 5076 | { | |
| 5077 | SCHECK_PARTIAL(); | |
| 5078 | break; | break; |
| 5079 | } | |
| 5080 | if ((md->ctypes[*eptr] & ctype_word) == 0) break; | |
| 5081 | eptr++; | eptr++; |
| 5082 | } | } |
| 5083 | break; | break; |
| # | Line 4253 for (;;) | Line 5099 for (;;) |
| 5099 | ||
| 5100 | /* Get here if we can't make it match with any permitted repetitions */ | /* Get here if we can't make it match with any permitted repetitions */ |
| 5101 | ||
| 5102 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 5103 | } | } |
| 5104 | /* Control never gets here */ | /* Control never gets here */ |
| 5105 | ||
| # | Line 4409 const uschar *tables; | Line 5255 const uschar *tables; |
| 5255 | const uschar *start_bits = NULL; | const uschar *start_bits = NULL; |
| 5256 | USPTR start_match = (USPTR)subject + start_offset; | USPTR start_match = (USPTR)subject + start_offset; |
| 5257 | USPTR end_subject; | USPTR end_subject; |
| 5258 | USPTR start_partial = NULL; | |
| 5259 | USPTR req_byte_ptr = start_match - 1; | USPTR req_byte_ptr = start_match - 1; |
| 5260 | ||
| 5261 | pcre_study_data internal_study; | pcre_study_data internal_study; |
| # | Line 4425 if (re == NULL || subject == NULL || | Line 5272 if (re == NULL || subject == NULL || |
| 5272 | (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL; | (offsets == NULL && offsetcount > 0)) return PCRE_ERROR_NULL; |
| 5273 | if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; | if (offsetcount < 0) return PCRE_ERROR_BADCOUNT; |
| 5274 | ||
| 5275 | /* This information is for finding all the numbers associated with a given | |
| 5276 | name, for condition testing. */ | |
| 5277 | ||
| 5278 | md->name_table = (uschar *)re + re->name_table_offset; | |
| 5279 | md->name_count = re->name_count; | |
| 5280 | md->name_entry_size = re->name_entry_size; | |
| 5281 | ||
| 5282 | /* Fish out the optional data from the extra_data structure, first setting | /* Fish out the optional data from the extra_data structure, first setting |
| 5283 | the default values. */ | the default values. */ |
| 5284 | ||
| # | Line 4492 md->jscript_compat = (re->options & PCRE | Line 5346 md->jscript_compat = (re->options & PCRE |
| 5346 | md->notbol = (options & PCRE_NOTBOL) != 0; | md->notbol = (options & PCRE_NOTBOL) != 0; |
| 5347 | md->noteol = (options & PCRE_NOTEOL) != 0; | md->noteol = (options & PCRE_NOTEOL) != 0; |
| 5348 | md->notempty = (options & PCRE_NOTEMPTY) != 0; | md->notempty = (options & PCRE_NOTEMPTY) != 0; |
| 5349 | md->partial = (options & PCRE_PARTIAL) != 0; | md->notempty_atstart = (options & PCRE_NOTEMPTY_ATSTART) != 0; |
| 5350 | md->partial = ((options & PCRE_PARTIAL_HARD) != 0)? 2 : | |
| 5351 | ((options & PCRE_PARTIAL_SOFT) != 0)? 1 : 0; | |
| 5352 | md->hitend = FALSE; | md->hitend = FALSE; |
| 5353 | md->mark = NULL; /* In case never set */ | |
| 5354 | ||
| 5355 | md->recursive = NULL; /* No recursion at top level */ | md->recursive = NULL; /* No recursion at top level */ |
| 5356 | ||
| # | Line 4533 switch ((((options & PCRE_NEWLINE_BITS) | Line 5390 switch ((((options & PCRE_NEWLINE_BITS) |
| 5390 | (pcre_uint32)options) & PCRE_NEWLINE_BITS) | (pcre_uint32)options) & PCRE_NEWLINE_BITS) |
| 5391 | { | { |
| 5392 | case 0: newline = NEWLINE; break; /* Compile-time default */ | case 0: newline = NEWLINE; break; /* Compile-time default */ |
| 5393 | case PCRE_NEWLINE_CR: newline = '\r'; break; | case PCRE_NEWLINE_CR: newline = CHAR_CR; break; |
| 5394 | case PCRE_NEWLINE_LF: newline = '\n'; break; | case PCRE_NEWLINE_LF: newline = CHAR_NL; break; |
| 5395 | case PCRE_NEWLINE_CR+ | case PCRE_NEWLINE_CR+ |
| 5396 | PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break; | PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break; |
| 5397 | case PCRE_NEWLINE_ANY: newline = -1; break; | case PCRE_NEWLINE_ANY: newline = -1; break; |
| 5398 | case PCRE_NEWLINE_ANYCRLF: newline = -2; break; | case PCRE_NEWLINE_ANYCRLF: newline = -2; break; |
| 5399 | default: return PCRE_ERROR_BADNEWLINE; | default: return PCRE_ERROR_BADNEWLINE; |
| # | Line 4566 else | Line 5423 else |
| 5423 | } | } |
| 5424 | } | } |
| 5425 | ||
| 5426 | /* Partial matching is supported only for a restricted set of regexes at the | /* Partial matching was originally supported only for a restricted set of |
| 5427 | moment. */ | regexes; from release 8.00 there are no restrictions, but the bits are still |
| 5428 | defined (though never set). So there's no harm in leaving this code. */ | |
| 5429 | ||
| 5430 | if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0) | if (md->partial && (re->flags & PCRE_NOPARTIAL) != 0) |
| 5431 | return PCRE_ERROR_BADPARTIAL; | return PCRE_ERROR_BADPARTIAL; |
| # | Line 4578 back the character offset. */ | Line 5436 back the character offset. */ |
| 5436 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 5437 | if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0) | if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0) |
| 5438 | { | { |
| 5439 | if (_pcre_valid_utf8((uschar *)subject, length) >= 0) | if (_pcre_valid_utf8((USPTR)subject, length) >= 0) |
| 5440 | return PCRE_ERROR_BADUTF8; | return PCRE_ERROR_BADUTF8; |
| 5441 | if (start_offset > 0 && start_offset < length) | if (start_offset > 0 && start_offset < length) |
| 5442 | { | { |
| 5443 | int tb = ((uschar *)subject)[start_offset]; | int tb = ((USPTR)subject)[start_offset]; |
| 5444 | if (tb > 127) | if (tb > 127) |
| 5445 | { | { |
| 5446 | tb &= 0xc0; | tb &= 0xc0; |
| # | Line 4654 if (!anchored) | Line 5512 if (!anchored) |
| 5512 | } | } |
| 5513 | else | else |
| 5514 | if (!startline && study != NULL && | if (!startline && study != NULL && |
| 5515 | (study->options & PCRE_STUDY_MAPPED) != 0) | (study->flags & PCRE_STUDY_MAPPED) != 0) |
| 5516 | start_bits = study->start_bits; | start_bits = study->start_bits; |
| 5517 | } | } |
| 5518 | ||
| # | Line 4688 for(;;) | Line 5546 for(;;) |
| 5546 | while (iptr < iend) *iptr++ = -1; | while (iptr < iend) *iptr++ = -1; |
| 5547 | } | } |
| 5548 | ||
| 5549 | /* Advance to a unique first char if possible. If firstline is TRUE, the | /* If firstline is TRUE, the start of the match is constrained to the first |
| 5550 | start of the match is constrained to the first line of a multiline string. | line of a multiline string. That is, the match must be before or at the first |
| 5551 | That is, the match must be before or at the first newline. Implement this by | newline. Implement this by temporarily adjusting end_subject so that we stop |
| 5552 | temporarily adjusting end_subject so that we stop scanning at a newline. If | scanning at a newline. If the match fails at the newline, later code breaks |
| 5553 | the match fails at the newline, later code breaks this loop. */ | this loop. */ |
| 5554 | ||
| 5555 | if (firstline) | if (firstline) |
| 5556 | { | { |
| 5557 | USPTR t = start_match; | USPTR t = start_match; |
| 5558 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 5559 | if (utf8) | if (utf8) |
| 5560 | { | { |
| 5561 | while (t < md->end_subject && !IS_NEWLINE(t)) | while (t < md->end_subject && !IS_NEWLINE(t)) |
| 5562 | { | { |
| 5563 | t++; | t++; |
| 5564 | while (t < end_subject && (*t & 0xc0) == 0x80) t++; | while (t < end_subject && (*t & 0xc0) == 0x80) t++; |
| 5565 | } | } |
| 5566 | } | } |
| 5567 | else | else |
| 5568 | #endif | #endif |
| 5569 | while (t < md->end_subject && !IS_NEWLINE(t)) t++; | while (t < md->end_subject && !IS_NEWLINE(t)) t++; |
| 5570 | end_subject = t; | end_subject = t; |
| 5571 | } | } |
| 5572 | ||
| 5573 | /* Now advance to a unique first byte if there is one. */ | /* There are some optimizations that avoid running the match if a known |
| 5574 | starting point is not found, or if a known later character is not present. | |
| 5575 | However, there is an option that disables these, for testing and for ensuring | |
| 5576 | that all callouts do actually occur. */ | |
| 5577 | ||
| 5578 | if (first_byte >= 0) | if ((options & PCRE_NO_START_OPTIMIZE) == 0) |
| 5579 | { | { |
| 5580 | if (first_byte_caseless) | /* Advance to a unique first byte if there is one. */ |
| while (start_match < end_subject && md->lcc[*start_match] != first_byte) | ||
| start_match++; | ||
| else | ||
| while (start_match < end_subject && *start_match != first_byte) | ||
| start_match++; | ||
| } | ||
| 5581 | ||
| 5582 | /* Or to just after a linebreak for a multiline match */ | if (first_byte >= 0) |
| 5583 | { | |
| 5584 | if (first_byte_caseless) | |
| 5585 | while (start_match < end_subject && md->lcc[*start_match] != first_byte) | |
| 5586 | start_match++; | |
| 5587 | else | |
| 5588 | while (start_match < end_subject && *start_match != first_byte) | |
| 5589 | start_match++; | |
| 5590 | } | |
| 5591 | ||
| 5592 | else if (startline) | /* Or to just after a linebreak for a multiline match */ |
| 5593 | { | |
| 5594 | if (start_match > md->start_subject + start_offset) | else if (startline) |
| 5595 | { | { |
| 5596 | #ifdef SUPPORT_UTF8 | if (start_match > md->start_subject + start_offset) |
| if (utf8) | ||
| 5597 | { | { |
| 5598 | while (start_match < end_subject && !WAS_NEWLINE(start_match)) | #ifdef SUPPORT_UTF8 |
| 5599 | if (utf8) | |
| 5600 | { | { |
| 5601 | start_match++; | while (start_match < end_subject && !WAS_NEWLINE(start_match)) |
| 5602 | while(start_match < end_subject && (*start_match & 0xc0) == 0x80) | { |
| 5603 | start_match++; | start_match++; |
| 5604 | } | while(start_match < end_subject && (*start_match & 0xc0) == 0x80) |
| 5605 | start_match++; | |
| 5606 | } | |
| 5607 | } | |
| 5608 | else | |
| 5609 | #endif | |
| 5610 | while (start_match < end_subject && !WAS_NEWLINE(start_match)) | |
| 5611 | start_match++; | |
| 5612 | ||
| 5613 | /* If we have just passed a CR and the newline option is ANY or ANYCRLF, | |
| 5614 | and we are now at a LF, advance the match position by one more character. | |
| 5615 | */ | |
| 5616 | ||
| 5617 | if (start_match[-1] == CHAR_CR && | |
| 5618 | (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) && | |
| 5619 | start_match < end_subject && | |
| 5620 | *start_match == CHAR_NL) | |
| 5621 | start_match++; | |
| 5622 | } | } |
| else | ||
| #endif | ||
| while (start_match < end_subject && !WAS_NEWLINE(start_match)) | ||
| start_match++; | ||
| /* If we have just passed a CR and the newline option is ANY or ANYCRLF, | ||
| and we are now at a LF, advance the match position by one more character. | ||
| */ | ||
| if (start_match[-1] == '\r' && | ||
| (md->nltype == NLTYPE_ANY || md->nltype == NLTYPE_ANYCRLF) && | ||
| start_match < end_subject && | ||
| *start_match == '\n') | ||
| start_match++; | ||
| 5623 | } | } |
| } | ||
| 5624 | ||
| 5625 | /* Or to a non-unique first byte after study */ | /* Or to a non-unique first byte after study */ |
| 5626 | ||
| 5627 | else if (start_bits != NULL) | else if (start_bits != NULL) |
| { | ||
| while (start_match < end_subject) | ||
| 5628 | { | { |
| 5629 | register unsigned int c = *start_match; | while (start_match < end_subject) |
| 5630 | if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++; | { |
| 5631 | else break; | register unsigned int c = *start_match; |
| 5632 | if ((start_bits[c/8] & (1 << (c&7))) == 0) start_match++; | |
| 5633 | else break; | |
| 5634 | } | |
| 5635 | } | } |
| 5636 | } | } /* Starting optimizations */ |
| 5637 | ||
| 5638 | /* Restore fudged end_subject */ | /* Restore fudged end_subject */ |
| 5639 | ||
| 5640 | end_subject = save_end_subject; | end_subject = save_end_subject; |
| 5641 | ||
| 5642 | #ifdef DEBUG /* Sigh. Some compilers never learn. */ | /* The following two optimizations are disabled for partial matching or if |
| 5643 | printf(">>>> Match against: "); | disabling is explicitly requested. */ |
| pchars(start_match, end_subject - start_match, TRUE, md); | ||
| printf("\n"); | ||
| #endif | ||
| 5644 | ||
| 5645 | /* If req_byte is set, we know that that character must appear in the subject | if ((options & PCRE_NO_START_OPTIMIZE) == 0 && !md->partial) |
| for the match to succeed. If the first character is set, req_byte must be | ||
| later in the subject; otherwise the test starts at the match point. This | ||
| optimization can save a huge amount of backtracking in patterns with nested | ||
| unlimited repeats that aren't going to match. Writing separate code for | ||
| cased/caseless versions makes it go faster, as does using an autoincrement | ||
| and backing off on a match. | ||
| HOWEVER: when the subject string is very, very long, searching to its end can | ||
| take a long time, and give bad performance on quite ordinary patterns. This | ||
| showed up when somebody was matching something like /^\d+C/ on a 32-megabyte | ||
| string... so we don't do this when the string is sufficiently long. | ||
| ALSO: this processing is disabled when partial matching is requested. | ||
| */ | ||
| if (req_byte >= 0 && | ||
| end_subject - start_match < REQ_BYTE_MAX && | ||
| !md->partial) | ||
| 5646 | { | { |
| 5647 | register USPTR p = start_match + ((first_byte >= 0)? 1 : 0); | /* If the pattern was studied, a minimum subject length may be set. This is |
| 5648 | a lower bound; no actual string of that length may actually match the | |
| 5649 | pattern. Although the value is, strictly, in characters, we treat it as | |
| 5650 | bytes to avoid spending too much time in this optimization. */ | |
| 5651 | ||
| 5652 | if (study != NULL && (study->flags & PCRE_STUDY_MINLEN) != 0 && | |
| 5653 | (pcre_uint32)(end_subject - start_match) < study->minlength) | |
| 5654 | { | |
| 5655 | rc = MATCH_NOMATCH; | |
| 5656 | break; | |
| 5657 | } | |
| 5658 | ||
| 5659 | /* We don't need to repeat the search if we haven't yet reached the | /* If req_byte is set, we know that that character must appear in the |
| 5660 | place we found it at last time. */ | subject for the match to succeed. If the first character is set, req_byte |
| 5661 | must be later in the subject; otherwise the test starts at the match point. | |
| 5662 | This optimization can save a huge amount of backtracking in patterns with | |
| 5663 | nested unlimited repeats that aren't going to match. Writing separate code | |
| 5664 | for cased/caseless versions makes it go faster, as does using an | |
| 5665 | autoincrement and backing off on a match. | |
| 5666 | ||
| 5667 | if (p > req_byte_ptr) | HOWEVER: when the subject string is very, very long, searching to its end |
| 5668 | can take a long time, and give bad performance on quite ordinary patterns. | |
| 5669 | This showed up when somebody was matching something like /^\d+C/ on a | |
| 5670 | 32-megabyte string... so we don't do this when the string is sufficiently | |
| 5671 | long. */ | |
| 5672 | <