Parent Directory
|
Revision Log
|
Patch
| revision 742 by zherczeg, Sun Nov 6 08:05:33 2011 UTC | revision 852 by zherczeg, Thu Jan 5 19:18:12 2012 UTC | |
|---|---|---|
| # | Line 6 | Line 6 |
| 6 | and semantics are as close as possible to those of the Perl 5 language. | and semantics are as close as possible to those of the Perl 5 language. |
| 7 | ||
| 8 | Written by Philip Hazel | Written by Philip Hazel |
| 9 | Copyright (c) 1997-2011 University of Cambridge | Copyright (c) 1997-2012 University of Cambridge |
| 10 | ||
| 11 | ----------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
| 12 | Redistribution and use in source and binary forms, with or without | Redistribution and use in source and binary forms, with or without |
| # | Line 82 negative to avoid the external error cod | Line 82 negative to avoid the external error cod |
| 82 | #define MATCH_SKIP_ARG (-993) | #define MATCH_SKIP_ARG (-993) |
| 83 | #define MATCH_THEN (-992) | #define MATCH_THEN (-992) |
| 84 | ||
| /* This is a convenience macro for code that occurs many times. */ | ||
| #define MRRETURN(ra) \ | ||
| { \ | ||
| md->mark = markptr; \ | ||
| RRETURN(ra); \ | ||
| } | ||
| 85 | /* Maximum number of ints of offset to save on the stack for recursive calls. | /* Maximum number of ints of offset to save on the stack for recursive calls. |
| 86 | If the offset vector is bigger, malloc is used. This should be a multiple of 3, | If the offset vector is bigger, malloc is used. This should be a multiple of 3, |
| 87 | because the offset vector is always a multiple of 3 long. */ | because the offset vector is always a multiple of 3 long. */ |
| # | Line 121 Returns: nothing | Line 113 Returns: nothing |
| 113 | */ | */ |
| 114 | ||
| 115 | static void | static void |
| 116 | pchars(const uschar *p, int length, BOOL is_subject, match_data *md) | pchars(const pcre_uchar *p, int length, BOOL is_subject, match_data *md) |
| 117 | { | { |
| 118 | unsigned int c; | unsigned int c; |
| 119 | if (is_subject && length > md->end_subject - p) length = md->end_subject - p; | if (is_subject && length > md->end_subject - p) length = md->end_subject - p; |
| # | Line 152 Returns: < 0 if not matched, otherw | Line 144 Returns: < 0 if not matched, otherw |
| 144 | */ | */ |
| 145 | ||
| 146 | static int | static int |
| 147 | match_ref(int offset, register USPTR eptr, int length, match_data *md, | match_ref(int offset, register PCRE_PUCHAR eptr, int length, match_data *md, |
| 148 | BOOL caseless) | BOOL caseless) |
| 149 | { | { |
| 150 | USPTR eptr_start = eptr; | PCRE_PUCHAR eptr_start = eptr; |
| 151 | register USPTR p = md->start_subject + md->offset_vector[offset]; | register PCRE_PUCHAR p = md->start_subject + md->offset_vector[offset]; |
| 152 | ||
| 153 | #ifdef PCRE_DEBUG | #ifdef PCRE_DEBUG |
| 154 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| # | Line 181 ASCII characters. */ | Line 173 ASCII characters. */ |
| 173 | ||
| 174 | if (caseless) | if (caseless) |
| 175 | { | { |
| 176 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 177 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 178 | if (md->utf8) | if (md->utf) |
| 179 | { | { |
| 180 | /* Match characters up to the end of the reference. NOTE: the number of | /* Match characters up to the end of the reference. NOTE: the number of |
| 181 | bytes matched may differ, because there are some characters whose upper and | bytes matched may differ, because there are some characters whose upper and |
| # | Line 193 if (caseless) | Line 185 if (caseless) |
| 185 | the latter. It is important, therefore, to check the length along the | the latter. It is important, therefore, to check the length along the |
| 186 | reference, not along the subject (earlier code did this wrong). */ | reference, not along the subject (earlier code did this wrong). */ |
| 187 | ||
| 188 | USPTR endptr = p + length; | PCRE_PUCHAR endptr = p + length; |
| 189 | while (p < endptr) | while (p < endptr) |
| 190 | { | { |
| 191 | int c, d; | int c, d; |
| # | Line 212 if (caseless) | Line 204 if (caseless) |
| 204 | { | { |
| 205 | if (eptr + length > md->end_subject) return -1; | if (eptr + length > md->end_subject) return -1; |
| 206 | while (length-- > 0) | while (length-- > 0) |
| 207 | { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; } | { |
| 208 | if (TABLE_GET(*p, md->lcc, *p) != TABLE_GET(*eptr, md->lcc, *eptr)) return -1; | |
| 209 | p++; | |
| 210 | eptr++; | |
| 211 | } | |
| 212 | } | } |
| 213 | } | } |
| 214 | ||
| # | Line 225 else | Line 221 else |
| 221 | while (length-- > 0) if (*p++ != *eptr++) return -1; | while (length-- > 0) if (*p++ != *eptr++) return -1; |
| 222 | } | } |
| 223 | ||
| 224 | return eptr - eptr_start; | return (int)(eptr - eptr_start); |
| 225 | } | } |
| 226 | ||
| 227 | ||
| # | Line 290 actually used in this definition. */ | Line 286 actually used in this definition. */ |
| 286 | #define RMATCH(ra,rb,rc,rd,re,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rw) \ |
| 287 | { \ | { \ |
| 288 | printf("match() called in line %d\n", __LINE__); \ | printf("match() called in line %d\n", __LINE__); \ |
| 289 | rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1); \ | rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1); \ |
| 290 | printf("to line %d\n", __LINE__); \ | printf("to line %d\n", __LINE__); \ |
| 291 | } | } |
| 292 | #define RRETURN(ra) \ | #define RRETURN(ra) \ |
| # | Line 300 actually used in this definition. */ | Line 296 actually used in this definition. */ |
| 296 | } | } |
| 297 | #else | #else |
| 298 | #define RMATCH(ra,rb,rc,rd,re,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rw) \ |
| 299 | rrc = match(ra,rb,mstart,markptr,rc,rd,re,rdepth+1) | rrc = match(ra,rb,mstart,rc,rd,re,rdepth+1) |
| 300 | #define RRETURN(ra) return ra | #define RRETURN(ra) return ra |
| 301 | #endif | #endif |
| 302 | ||
| # | Line 315 argument of match(), which never changes | Line 311 argument of match(), which never changes |
| 311 | ||
| 312 | #define RMATCH(ra,rb,rc,rd,re,rw)\ | #define RMATCH(ra,rb,rc,rd,re,rw)\ |
| 313 | {\ | {\ |
| 314 | heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\ | heapframe *newframe = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe));\ |
| 315 | if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\ | if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\ |
| 316 | frame->Xwhere = rw; \ | frame->Xwhere = rw; \ |
| 317 | newframe->Xeptr = ra;\ | newframe->Xeptr = ra;\ |
| 318 | newframe->Xecode = rb;\ | newframe->Xecode = rb;\ |
| 319 | newframe->Xmstart = mstart;\ | newframe->Xmstart = mstart;\ |
| newframe->Xmarkptr = markptr;\ | ||
| 320 | newframe->Xoffset_top = rc;\ | newframe->Xoffset_top = rc;\ |
| 321 | newframe->Xeptrb = re;\ | newframe->Xeptrb = re;\ |
| 322 | newframe->Xrdepth = frame->Xrdepth + 1;\ | newframe->Xrdepth = frame->Xrdepth + 1;\ |
| # | Line 337 argument of match(), which never changes | Line 332 argument of match(), which never changes |
| 332 | {\ | {\ |
| 333 | heapframe *oldframe = frame;\ | heapframe *oldframe = frame;\ |
| 334 | frame = oldframe->Xprevframe;\ | frame = oldframe->Xprevframe;\ |
| 335 | (pcre_stack_free)(oldframe);\ | (PUBL(stack_free))(oldframe);\ |
| 336 | if (frame != NULL)\ | if (frame != NULL)\ |
| 337 | {\ | {\ |
| 338 | rrc = ra;\ | rrc = ra;\ |
| # | Line 354 typedef struct heapframe { | Line 349 typedef struct heapframe { |
| 349 | ||
| 350 | /* Function arguments that may change */ | /* Function arguments that may change */ |
| 351 | ||
| 352 | USPTR Xeptr; | PCRE_PUCHAR Xeptr; |
| 353 | const uschar *Xecode; | const pcre_uchar *Xecode; |
| 354 | USPTR Xmstart; | PCRE_PUCHAR Xmstart; |
| USPTR Xmarkptr; | ||
| 355 | int Xoffset_top; | int Xoffset_top; |
| 356 | eptrblock *Xeptrb; | eptrblock *Xeptrb; |
| 357 | unsigned int Xrdepth; | unsigned int Xrdepth; |
| 358 | ||
| 359 | /* Function local variables */ | /* Function local variables */ |
| 360 | ||
| 361 | USPTR Xcallpat; | PCRE_PUCHAR Xcallpat; |
| 362 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 363 | USPTR Xcharptr; | PCRE_PUCHAR Xcharptr; |
| 364 | #endif | #endif |
| 365 | USPTR Xdata; | PCRE_PUCHAR Xdata; |
| 366 | USPTR Xnext; | PCRE_PUCHAR Xnext; |
| 367 | USPTR Xpp; | PCRE_PUCHAR Xpp; |
| 368 | USPTR Xprev; | PCRE_PUCHAR Xprev; |
| 369 | USPTR Xsaved_eptr; | PCRE_PUCHAR Xsaved_eptr; |
| 370 | ||
| 371 | recursion_info Xnew_recursive; | recursion_info Xnew_recursive; |
| 372 | ||
| # | Line 385 typedef struct heapframe { | Line 379 typedef struct heapframe { |
| 379 | int Xprop_value; | int Xprop_value; |
| 380 | int Xprop_fail_result; | int Xprop_fail_result; |
| 381 | int Xoclength; | int Xoclength; |
| 382 | uschar Xocchars[8]; | pcre_uchar Xocchars[6]; |
| 383 | #endif | #endif |
| 384 | ||
| 385 | int Xcodelink; | int Xcodelink; |
| # | Line 427 returns a negative (error) response, the | Line 421 returns a negative (error) response, the |
| 421 | same response. */ | same response. */ |
| 422 | ||
| 423 | /* These macros pack up tests that are used for partial matching, and which | /* These macros pack up tests that are used for partial matching, and which |
| 424 | appears several times in the code. We set the "hit end" flag if the pointer is | appear several times in the code. We set the "hit end" flag if the pointer is |
| 425 | at the end of the subject and also past the start of the subject (i.e. | at the end of the subject and also past the start of the subject (i.e. |
| 426 | something has been matched). For hard partial matching, we then return | something has been matched). For hard partial matching, we then return |
| 427 | immediately. The second one is used when we already know we are past the end of | immediately. The second one is used when we already know we are past the end of |
| # | Line 438 the subject. */ | Line 432 the subject. */ |
| 432 | eptr > md->start_used_ptr) \ | eptr > md->start_used_ptr) \ |
| 433 | { \ | { \ |
| 434 | md->hitend = TRUE; \ | md->hitend = TRUE; \ |
| 435 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \ | if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \ |
| 436 | } | } |
| 437 | ||
| 438 | #define SCHECK_PARTIAL()\ | #define SCHECK_PARTIAL()\ |
| 439 | if (md->partial != 0 && eptr > md->start_used_ptr) \ | if (md->partial != 0 && eptr > md->start_used_ptr) \ |
| 440 | { \ | { \ |
| 441 | md->hitend = TRUE; \ | md->hitend = TRUE; \ |
| 442 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \ | if (md->partial > 1) RRETURN(PCRE_ERROR_PARTIAL); \ |
| 443 | } | } |
| 444 | ||
| 445 | ||
| 446 | /* Performance note: It might be tempting to extract commonly used fields from | /* Performance note: It might be tempting to extract commonly used fields from |
| 447 | the md structure (e.g. utf8, end_subject) into individual variables to improve | the md structure (e.g. utf, end_subject) into individual variables to improve |
| 448 | 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 |
| 449 | made performance worse. | made performance worse. |
| 450 | ||
| # | Line 459 Arguments: | Line 453 Arguments: |
| 453 | ecode pointer to current position in compiled code | ecode pointer to current position in compiled code |
| 454 | mstart pointer to the current match start position (can be modified | mstart pointer to the current match start position (can be modified |
| 455 | by encountering \K) | by encountering \K) |
| markptr pointer to the most recent MARK name, or NULL | ||
| 456 | offset_top current top pointer | offset_top current top pointer |
| 457 | md pointer to "static" info for the match | md pointer to "static" info for the match |
| 458 | eptrb pointer to chain of blocks containing eptr at start of | eptrb pointer to chain of blocks containing eptr at start of |
| # | Line 474 Returns: MATCH_MATCH if matched | Line 467 Returns: MATCH_MATCH if matched |
| 467 | */ | */ |
| 468 | ||
| 469 | static int | static int |
| 470 | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart, | match(REGISTER PCRE_PUCHAR eptr, REGISTER const pcre_uchar *ecode, |
| 471 | const uschar *markptr, int offset_top, match_data *md, eptrblock *eptrb, | PCRE_PUCHAR mstart, int offset_top, match_data *md, eptrblock *eptrb, |
| 472 | unsigned int rdepth) | unsigned int rdepth) |
| 473 | { | { |
| 474 | /* These variables do not need to be preserved over recursion in this function, | /* These variables do not need to be preserved over recursion in this function, |
| # | Line 485 so they can be ordinary variables in all | Line 478 so they can be ordinary variables in all |
| 478 | register int rrc; /* Returns from recursive calls */ | register int rrc; /* Returns from recursive calls */ |
| 479 | register int i; /* Used for loops not involving calls to RMATCH() */ | register int i; /* Used for loops not involving calls to RMATCH() */ |
| 480 | register unsigned int c; /* Character values not kept over RMATCH() calls */ | register unsigned int c; /* Character values not kept over RMATCH() calls */ |
| 481 | register BOOL utf8; /* Local copy of UTF-8 flag for speed */ | register BOOL utf; /* Local copy of UTF flag for speed */ |
| 482 | ||
| 483 | BOOL minimize, possessive; /* Quantifier options */ | BOOL minimize, possessive; /* Quantifier options */ |
| 484 | BOOL caseless; | BOOL caseless; |
| # | Line 497 heap storage. Set up the top-level frame | Line 490 heap storage. Set up the top-level frame |
| 490 | heap whenever RMATCH() does a "recursion". See the macro definitions above. */ | heap whenever RMATCH() does a "recursion". See the macro definitions above. */ |
| 491 | ||
| 492 | #ifdef NO_RECURSE | #ifdef NO_RECURSE |
| 493 | heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe)); | heapframe *frame = (heapframe *)(PUBL(stack_malloc))(sizeof(heapframe)); |
| 494 | if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY); | if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY); |
| 495 | frame->Xprevframe = NULL; /* Marks the top level */ | frame->Xprevframe = NULL; /* Marks the top level */ |
| 496 | ||
| # | Line 506 frame->Xprevframe = NULL; /* | Line 499 frame->Xprevframe = NULL; /* |
| 499 | frame->Xeptr = eptr; | frame->Xeptr = eptr; |
| 500 | frame->Xecode = ecode; | frame->Xecode = ecode; |
| 501 | frame->Xmstart = mstart; | frame->Xmstart = mstart; |
| frame->Xmarkptr = markptr; | ||
| 502 | frame->Xoffset_top = offset_top; | frame->Xoffset_top = offset_top; |
| 503 | frame->Xeptrb = eptrb; | frame->Xeptrb = eptrb; |
| 504 | frame->Xrdepth = rdepth; | frame->Xrdepth = rdepth; |
| # | Line 520 HEAP_RECURSE: | Line 512 HEAP_RECURSE: |
| 512 | #define eptr frame->Xeptr | #define eptr frame->Xeptr |
| 513 | #define ecode frame->Xecode | #define ecode frame->Xecode |
| 514 | #define mstart frame->Xmstart | #define mstart frame->Xmstart |
| #define markptr frame->Xmarkptr | ||
| 515 | #define offset_top frame->Xoffset_top | #define offset_top frame->Xoffset_top |
| 516 | #define eptrb frame->Xeptrb | #define eptrb frame->Xeptrb |
| 517 | #define rdepth frame->Xrdepth | #define rdepth frame->Xrdepth |
| 518 | ||
| 519 | /* Ditto for the local variables */ | /* Ditto for the local variables */ |
| 520 | ||
| 521 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 522 | #define charptr frame->Xcharptr | #define charptr frame->Xcharptr |
| 523 | #endif | #endif |
| 524 | #define callpat frame->Xcallpat | #define callpat frame->Xcallpat |
| # | Line 585 declarations can be cut out in a block. | Line 576 declarations can be cut out in a block. |
| 576 | below are for variables that do not have to be preserved over a recursive call | below are for variables that do not have to be preserved over a recursive call |
| 577 | to RMATCH(). */ | to RMATCH(). */ |
| 578 | ||
| 579 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 580 | const uschar *charptr; | const pcre_uchar *charptr; |
| 581 | #endif | #endif |
| 582 | const uschar *callpat; | const pcre_uchar *callpat; |
| 583 | const uschar *data; | const pcre_uchar *data; |
| 584 | const uschar *next; | const pcre_uchar *next; |
| 585 | USPTR pp; | PCRE_PUCHAR pp; |
| 586 | const uschar *prev; | const pcre_uchar *prev; |
| 587 | USPTR saved_eptr; | PCRE_PUCHAR saved_eptr; |
| 588 | ||
| 589 | recursion_info new_recursive; | recursion_info new_recursive; |
| 590 | ||
| # | Line 606 int prop_type; | Line 597 int prop_type; |
| 597 | int prop_value; | int prop_value; |
| 598 | int prop_fail_result; | int prop_fail_result; |
| 599 | int oclength; | int oclength; |
| 600 | uschar occhars[8]; | pcre_uchar occhars[6]; |
| 601 | #endif | #endif |
| 602 | ||
| 603 | int codelink; | int codelink; |
| # | Line 634 the alternative names that are used. */ | Line 625 the alternative names that are used. */ |
| 625 | #define code_offset codelink | #define code_offset codelink |
| 626 | #define condassert condition | #define condassert condition |
| 627 | #define matched_once prev_is_word | #define matched_once prev_is_word |
| 628 | #define foc number | |
| 629 | ||
| 630 | /* These statements are here to stop the compiler complaining about unitialized | /* These statements are here to stop the compiler complaining about unitialized |
| 631 | variables. */ | variables. */ |
| # | Line 659 defined). However, RMATCH isn't like a f | Line 651 defined). However, RMATCH isn't like a f |
| 651 | 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, |
| 652 | however, impact performance when true recursion is being used. */ | however, impact performance when true recursion is being used. */ |
| 653 | ||
| 654 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 655 | utf8 = md->utf8; /* Local copy of the flag */ | utf = md->utf; /* Local copy of the flag */ |
| 656 | #else | #else |
| 657 | utf8 = FALSE; | utf = FALSE; |
| 658 | #endif | #endif |
| 659 | ||
| 660 | /* First check that we haven't called match() too many times, or that we | /* First check that we haven't called match() too many times, or that we |
| # | Line 701 for (;;) | Line 693 for (;;) |
| 693 | switch(op) | switch(op) |
| 694 | { | { |
| 695 | case OP_MARK: | case OP_MARK: |
| 696 | markptr = ecode + 2; | md->nomatch_mark = ecode + 2; |
| 697 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | md->mark = NULL; /* In case previously set by assertion */ |
| 698 | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, | |
| 699 | eptrb, RM55); | eptrb, RM55); |
| 700 | if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && | |
| 701 | md->mark == NULL) md->mark = ecode + 2; | |
| 702 | ||
| 703 | /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an | /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an |
| 704 | argument, and we must check whether that argument matches this MARK's | argument, and we must check whether that argument matches this MARK's |
| # | Line 712 for (;;) | Line 707 for (;;) |
| 707 | position and return MATCH_SKIP. Otherwise, pass back the return code | position and return MATCH_SKIP. Otherwise, pass back the return code |
| 708 | unaltered. */ | unaltered. */ |
| 709 | ||
| 710 | if (rrc == MATCH_SKIP_ARG && | else if (rrc == MATCH_SKIP_ARG && |
| 711 | strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0) | STRCMP_UC_UC(ecode + 2, md->start_match_ptr) == 0) |
| 712 | { | { |
| 713 | md->start_match_ptr = eptr; | md->start_match_ptr = eptr; |
| 714 | RRETURN(MATCH_SKIP); | RRETURN(MATCH_SKIP); |
| 715 | } | } |
| if (md->mark == NULL) md->mark = markptr; | ||
| 716 | RRETURN(rrc); | RRETURN(rrc); |
| 717 | ||
| 718 | case OP_FAIL: | case OP_FAIL: |
| 719 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 720 | ||
| 721 | /* COMMIT overrides PRUNE, SKIP, and THEN */ | /* COMMIT overrides PRUNE, SKIP, and THEN */ |
| 722 | ||
| 723 | case OP_COMMIT: | case OP_COMMIT: |
| 724 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
| 725 | eptrb, RM52); | eptrb, RM52); |
| 726 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && |
| 727 | rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG && | rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG && |
| 728 | rrc != MATCH_THEN) | rrc != MATCH_THEN) |
| 729 | RRETURN(rrc); | RRETURN(rrc); |
| 730 | MRRETURN(MATCH_COMMIT); | RRETURN(MATCH_COMMIT); |
| 731 | ||
| 732 | /* PRUNE overrides THEN */ | /* PRUNE overrides THEN */ |
| 733 | ||
| 734 | case OP_PRUNE: | case OP_PRUNE: |
| 735 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
| 736 | eptrb, RM51); | eptrb, RM51); |
| 737 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| 738 | MRRETURN(MATCH_PRUNE); | RRETURN(MATCH_PRUNE); |
| 739 | ||
| 740 | case OP_PRUNE_ARG: | case OP_PRUNE_ARG: |
| 741 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | md->nomatch_mark = ecode + 2; |
| 742 | md->mark = NULL; /* In case previously set by assertion */ | |
| 743 | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, | |
| 744 | eptrb, RM56); | eptrb, RM56); |
| 745 | if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && | |
| 746 | md->mark == NULL) md->mark = ecode + 2; | |
| 747 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| md->mark = ecode + 2; | ||
| 748 | RRETURN(MATCH_PRUNE); | RRETURN(MATCH_PRUNE); |
| 749 | ||
| 750 | /* SKIP overrides PRUNE and THEN */ | /* SKIP overrides PRUNE and THEN */ |
| 751 | ||
| 752 | case OP_SKIP: | case OP_SKIP: |
| 753 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
| 754 | eptrb, RM53); | eptrb, RM53); |
| 755 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) |
| 756 | RRETURN(rrc); | RRETURN(rrc); |
| 757 | md->start_match_ptr = eptr; /* Pass back current position */ | md->start_match_ptr = eptr; /* Pass back current position */ |
| 758 | MRRETURN(MATCH_SKIP); | RRETURN(MATCH_SKIP); |
| 759 | ||
| 760 | /* Note that, for Perl compatibility, SKIP with an argument does NOT set | |
| 761 | nomatch_mark. There is a flag that disables this opcode when re-matching a | |
| 762 | pattern that ended with a SKIP for which there was not a matching MARK. */ | |
| 763 | ||
| 764 | case OP_SKIP_ARG: | case OP_SKIP_ARG: |
| 765 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | if (md->ignore_skip_arg) |
| 766 | { | |
| 767 | ecode += PRIV(OP_lengths)[*ecode] + ecode[1]; | |
| 768 | break; | |
| 769 | } | |
| 770 | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, md, | |
| 771 | eptrb, RM57); | eptrb, RM57); |
| 772 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) |
| 773 | RRETURN(rrc); | RRETURN(rrc); |
| 774 | ||
| 775 | /* Pass back the current skip name by overloading md->start_match_ptr and | /* Pass back the current skip name by overloading md->start_match_ptr and |
| 776 | returning the special MATCH_SKIP_ARG return code. This will either be | returning the special MATCH_SKIP_ARG return code. This will either be |
| 777 | caught by a matching MARK, or get to the top, where it is treated the same | caught by a matching MARK, or get to the top, where it causes a rematch |
| 778 | as PRUNE. */ | with the md->ignore_skip_arg flag set. */ |
| 779 | ||
| 780 | md->start_match_ptr = ecode + 2; | md->start_match_ptr = ecode + 2; |
| 781 | RRETURN(MATCH_SKIP_ARG); | RRETURN(MATCH_SKIP_ARG); |
| # | Line 780 for (;;) | Line 785 for (;;) |
| 785 | match pointer to do this. */ | match pointer to do this. */ |
| 786 | ||
| 787 | case OP_THEN: | case OP_THEN: |
| 788 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
| 789 | eptrb, RM54); | eptrb, RM54); |
| 790 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 791 | md->start_match_ptr = ecode; | md->start_match_ptr = ecode; |
| 792 | MRRETURN(MATCH_THEN); | RRETURN(MATCH_THEN); |
| 793 | ||
| 794 | case OP_THEN_ARG: | case OP_THEN_ARG: |
| 795 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, | md->nomatch_mark = ecode + 2; |
| 796 | md->mark = NULL; /* In case previously set by assertion */ | |
| 797 | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode] + ecode[1], offset_top, | |
| 798 | md, eptrb, RM58); | md, eptrb, RM58); |
| 799 | if ((rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) && | |
| 800 | md->mark == NULL) md->mark = ecode + 2; | |
| 801 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 802 | md->start_match_ptr = ecode; | md->start_match_ptr = ecode; |
| md->mark = ecode + 2; | ||
| 803 | RRETURN(MATCH_THEN); | RRETURN(MATCH_THEN); |
| 804 | ||
| 805 | /* Handle an atomic group that does not contain any capturing parentheses. | /* Handle an atomic group that does not contain any capturing parentheses. |
| # | Line 915 for (;;) | Line 923 for (;;) |
| 923 | for (;;) | for (;;) |
| 924 | { | { |
| 925 | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
| 926 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
| 927 | eptrb, RM1); | eptrb, RM1); |
| 928 | if (rrc == MATCH_ONCE) break; /* Backing up through an atomic group */ | if (rrc == MATCH_ONCE) break; /* Backing up through an atomic group */ |
| 929 | ||
| # | Line 953 for (;;) | Line 961 for (;;) |
| 961 | ||
| 962 | /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */ | /* At this point, rrc will be one of MATCH_ONCE or MATCH_NOMATCH. */ |
| 963 | ||
| if (md->mark == NULL) md->mark = markptr; | ||
| 964 | RRETURN(rrc); | RRETURN(rrc); |
| 965 | } | } |
| 966 | ||
| # | Line 1003 for (;;) | Line 1010 for (;;) |
| 1010 | ||
| 1011 | else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT) | else if (!md->hasthen && ecode[GET(ecode, 1)] != OP_ALT) |
| 1012 | { | { |
| 1013 | ecode += _pcre_OP_lengths[*ecode]; | ecode += PRIV(OP_lengths)[*ecode]; |
| 1014 | goto TAIL_RECURSE; | goto TAIL_RECURSE; |
| 1015 | } | } |
| 1016 | ||
| 1017 | /* In all other cases, we have to make another call to match(). */ | /* In all other cases, we have to make another call to match(). */ |
| 1018 | ||
| 1019 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, eptrb, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, eptrb, |
| 1020 | RM2); | RM2); |
| 1021 | ||
| 1022 | /* See comment in the code for capturing groups above about handling | /* See comment in the code for capturing groups above about handling |
| # | Line 1027 for (;;) | Line 1034 for (;;) |
| 1034 | { | { |
| 1035 | if (rrc == MATCH_ONCE) | if (rrc == MATCH_ONCE) |
| 1036 | { | { |
| 1037 | const uschar *scode = ecode; | const pcre_uchar *scode = ecode; |
| 1038 | if (*scode != OP_ONCE) /* If not at start, find it */ | if (*scode != OP_ONCE) /* If not at start, find it */ |
| 1039 | { | { |
| 1040 | while (*scode == OP_ALT) scode += GET(scode, 1); | while (*scode == OP_ALT) scode += GET(scode, 1); |
| # | Line 1041 for (;;) | Line 1048 for (;;) |
| 1048 | if (*ecode != OP_ALT) break; | if (*ecode != OP_ALT) break; |
| 1049 | } | } |
| 1050 | ||
| if (md->mark == NULL) md->mark = markptr; | ||
| 1051 | RRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 1052 | ||
| 1053 | /* Handle possessive capturing brackets with an unlimited repeat. We come | /* Handle possessive capturing brackets with an unlimited repeat. We come |
| # | Line 1070 for (;;) | Line 1076 for (;;) |
| 1076 | if (offset < md->offset_max) | if (offset < md->offset_max) |
| 1077 | { | { |
| 1078 | matched_once = FALSE; | matched_once = FALSE; |
| 1079 | code_offset = ecode - md->start_code; | code_offset = (int)(ecode - md->start_code); |
| 1080 | ||
| 1081 | save_offset1 = md->offset_vector[offset]; | save_offset1 = md->offset_vector[offset]; |
| 1082 | save_offset2 = md->offset_vector[offset+1]; | save_offset2 = md->offset_vector[offset+1]; |
| # | Line 1093 for (;;) | Line 1099 for (;;) |
| 1099 | md->offset_vector[md->offset_end - number] = | md->offset_vector[md->offset_end - number] = |
| 1100 | (int)(eptr - md->start_subject); | (int)(eptr - md->start_subject); |
| 1101 | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
| 1102 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
| 1103 | eptrb, RM63); | eptrb, RM63); |
| 1104 | if (rrc == MATCH_KETRPOS) | if (rrc == MATCH_KETRPOS) |
| 1105 | { | { |
| # | Line 1129 for (;;) | Line 1135 for (;;) |
| 1135 | md->offset_vector[md->offset_end - number] = save_offset3; | md->offset_vector[md->offset_end - number] = save_offset3; |
| 1136 | } | } |
| 1137 | ||
| if (md->mark == NULL) md->mark = markptr; | ||
| 1138 | if (allow_zero || matched_once) | if (allow_zero || matched_once) |
| 1139 | { | { |
| 1140 | ecode += 1 + LINK_SIZE; | ecode += 1 + LINK_SIZE; |
| # | Line 1161 for (;;) | Line 1166 for (;;) |
| 1166 | ||
| 1167 | POSSESSIVE_NON_CAPTURE: | POSSESSIVE_NON_CAPTURE: |
| 1168 | matched_once = FALSE; | matched_once = FALSE; |
| 1169 | code_offset = ecode - md->start_code; | code_offset = (int)(ecode - md->start_code); |
| 1170 | ||
| 1171 | for (;;) | for (;;) |
| 1172 | { | { |
| 1173 | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; | if (op >= OP_SBRA) md->match_function_type = MATCH_CBEGROUP; |
| 1174 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + PRIV(OP_lengths)[*ecode], offset_top, md, |
| 1175 | eptrb, RM48); | eptrb, RM48); |
| 1176 | if (rrc == MATCH_KETRPOS) | if (rrc == MATCH_KETRPOS) |
| 1177 | { | { |
| # | Line 1216 for (;;) | Line 1221 for (;;) |
| 1221 | ||
| 1222 | if (ecode[LINK_SIZE+1] == OP_CALLOUT) | if (ecode[LINK_SIZE+1] == OP_CALLOUT) |
| 1223 | { | { |
| 1224 | if (pcre_callout != NULL) | if (PUBL(callout) != NULL) |
| 1225 | { | { |
| 1226 | pcre_callout_block cb; | PUBL(callout_block) cb; |
| 1227 | cb.version = 2; /* Version 1 of the callout block */ | cb.version = 2; /* Version 1 of the callout block */ |
| 1228 | cb.callout_number = ecode[LINK_SIZE+2]; | cb.callout_number = ecode[LINK_SIZE+2]; |
| 1229 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
| 1230 | #ifdef COMPILE_PCRE8 | |
| 1231 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
| 1232 | #else | |
| 1233 | cb.subject = (PCRE_SPTR16)md->start_subject; | |
| 1234 | #endif | |
| 1235 | cb.subject_length = (int)(md->end_subject - md->start_subject); | cb.subject_length = (int)(md->end_subject - md->start_subject); |
| 1236 | cb.start_match = (int)(mstart - md->start_subject); | cb.start_match = (int)(mstart - md->start_subject); |
| 1237 | cb.current_position = (int)(eptr - md->start_subject); | cb.current_position = (int)(eptr - md->start_subject); |
| # | Line 1231 for (;;) | Line 1240 for (;;) |
| 1240 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 1241 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 1242 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 1243 | cb.mark = markptr; | cb.mark = md->nomatch_mark; |
| 1244 | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); | if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH); |
| 1245 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 1246 | } | } |
| 1247 | ecode += _pcre_OP_lengths[OP_CALLOUT]; | ecode += PRIV(OP_lengths)[OP_CALLOUT]; |
| 1248 | } | } |
| 1249 | ||
| 1250 | condcode = ecode[LINK_SIZE+1]; | condcode = ecode[LINK_SIZE+1]; |
| # | Line 1252 for (;;) | Line 1261 for (;;) |
| 1261 | else | else |
| 1262 | { | { |
| 1263 | int recno = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | int recno = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ |
| 1264 | condition = (recno == RREF_ANY || recno == md->recursive->group_num); | condition = (recno == RREF_ANY || recno == md->recursive->group_num); |
| 1265 | ||
| 1266 | /* If the test is for recursion into a specific subpattern, and it is | /* If the test is for recursion into a specific subpattern, and it is |
| 1267 | false, but the test was set up by name, scan the table to see if the | false, but the test was set up by name, scan the table to see if the |
| 1268 | name refers to any other numbers, and test them. The condition is true | name refers to any other numbers, and test them. The condition is true |
| 1269 | if any one is set. */ | if any one is set. */ |
| 1270 | ||
| 1271 | if (!condition && condcode == OP_NRREF && recno != RREF_ANY) | if (!condition && condcode == OP_NRREF) |
| 1272 | { | { |
| 1273 | uschar *slotA = md->name_table; | pcre_uchar *slotA = md->name_table; |
| 1274 | for (i = 0; i < md->name_count; i++) | for (i = 0; i < md->name_count; i++) |
| 1275 | { | { |
| 1276 | if (GET2(slotA, 0) == recno) break; | if (GET2(slotA, 0) == recno) break; |
| # | Line 1274 for (;;) | Line 1283 for (;;) |
| 1283 | ||
| 1284 | if (i < md->name_count) | if (i < md->name_count) |
| 1285 | { | { |
| 1286 | uschar *slotB = slotA; | pcre_uchar *slotB = slotA; |
| 1287 | while (slotB > md->name_table) | while (slotB > md->name_table) |
| 1288 | { | { |
| 1289 | slotB -= md->name_entry_size; | slotB -= md->name_entry_size; |
| 1290 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
| 1291 | { | { |
| 1292 | condition = GET2(slotB, 0) == md->recursive->group_num; | condition = GET2(slotB, 0) == md->recursive->group_num; |
| 1293 | if (condition) break; | if (condition) break; |
| # | Line 1294 for (;;) | Line 1303 for (;;) |
| 1303 | for (i++; i < md->name_count; i++) | for (i++; i < md->name_count; i++) |
| 1304 | { | { |
| 1305 | slotB += md->name_entry_size; | slotB += md->name_entry_size; |
| 1306 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
| 1307 | { | { |
| 1308 | condition = GET2(slotB, 0) == md->recursive->group_num; | condition = GET2(slotB, 0) == md->recursive->group_num; |
| 1309 | if (condition) break; | if (condition) break; |
| # | Line 1307 for (;;) | Line 1316 for (;;) |
| 1316 | ||
| 1317 | /* Chose branch according to the condition */ | /* Chose branch according to the condition */ |
| 1318 | ||
| 1319 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1); |
| 1320 | } | } |
| 1321 | } | } |
| 1322 | ||
| # | Line 1324 for (;;) | Line 1333 for (;;) |
| 1333 | if (!condition && condcode == OP_NCREF) | if (!condition && condcode == OP_NCREF) |
| 1334 | { | { |
| 1335 | int refno = offset >> 1; | int refno = offset >> 1; |
| 1336 | uschar *slotA = md->name_table; | pcre_uchar *slotA = md->name_table; |
| 1337 | ||
| 1338 | for (i = 0; i < md->name_count; i++) | for (i = 0; i < md->name_count; i++) |
| 1339 | { | { |
| # | Line 1338 for (;;) | Line 1347 for (;;) |
| 1347 | ||
| 1348 | if (i < md->name_count) | if (i < md->name_count) |
| 1349 | { | { |
| 1350 | uschar *slotB = slotA; | pcre_uchar *slotB = slotA; |
| 1351 | while (slotB > md->name_table) | while (slotB > md->name_table) |
| 1352 | { | { |
| 1353 | slotB -= md->name_entry_size; | slotB -= md->name_entry_size; |
| 1354 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
| 1355 | { | { |
| 1356 | offset = GET2(slotB, 0) << 1; | offset = GET2(slotB, 0) << 1; |
| 1357 | condition = offset < offset_top && | condition = offset < offset_top && |
| # | Line 1360 for (;;) | Line 1369 for (;;) |
| 1369 | for (i++; i < md->name_count; i++) | for (i++; i < md->name_count; i++) |
| 1370 | { | { |
| 1371 | slotB += md->name_entry_size; | slotB += md->name_entry_size; |
| 1372 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
| 1373 | { | { |
| 1374 | offset = GET2(slotB, 0) << 1; | offset = GET2(slotB, 0) << 1; |
| 1375 | condition = offset < offset_top && | condition = offset < offset_top && |
| # | Line 1375 for (;;) | Line 1384 for (;;) |
| 1384 | ||
| 1385 | /* Chose branch according to the condition */ | /* Chose branch according to the condition */ |
| 1386 | ||
| 1387 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1); |
| 1388 | } | } |
| 1389 | ||
| 1390 | else if (condcode == OP_DEF) /* DEFINE - always false */ | else if (condcode == OP_DEF) /* DEFINE - always false */ |
| # | Line 1467 for (;;) | Line 1476 for (;;) |
| 1476 | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); |
| 1477 | if (offset_top <= offset) offset_top = offset + 2; | if (offset_top <= offset) offset_top = offset + 2; |
| 1478 | } | } |
| 1479 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 1480 | break; | break; |
| 1481 | ||
| 1482 | ||
| # | Line 1487 for (;;) | Line 1496 for (;;) |
| 1496 | (md->notempty || | (md->notempty || |
| 1497 | (md->notempty_atstart && | (md->notempty_atstart && |
| 1498 | mstart == md->start_subject + md->start_offset))) | mstart == md->start_subject + md->start_offset))) |
| 1499 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 1500 | ||
| 1501 | /* Otherwise, we have a match. */ | /* Otherwise, we have a match. */ |
| 1502 | ||
| # | Line 1496 for (;;) | Line 1505 for (;;) |
| 1505 | md->start_match_ptr = mstart; /* and the start (\K can modify) */ | md->start_match_ptr = mstart; /* and the start (\K can modify) */ |
| 1506 | ||
| 1507 | /* For some reason, the macros don't work properly if an expression is | /* For some reason, the macros don't work properly if an expression is |
| 1508 | given as the argument to MRRETURN when the heap is in use. */ | given as the argument to RRETURN when the heap is in use. */ |
| 1509 | ||
| 1510 | rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; | rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; |
| 1511 | MRRETURN(rrc); | RRETURN(rrc); |
| 1512 | ||
| 1513 | /* Assertion brackets. Check the alternative branches in turn - the | /* Assertion brackets. Check the alternative branches in turn - the |
| 1514 | matching won't pass the KET for an assertion. If any one branch matches, | matching won't pass the KET for an assertion. If any one branch matches, |
| # | Line 1527 for (;;) | Line 1536 for (;;) |
| 1536 | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
| 1537 | { | { |
| 1538 | mstart = md->start_match_ptr; /* In case \K reset it */ | mstart = md->start_match_ptr; /* In case \K reset it */ |
| markptr = md->mark; | ||
| 1539 | break; | break; |
| 1540 | } | } |
| 1541 | ||
| # | Line 1539 for (;;) | Line 1547 for (;;) |
| 1547 | } | } |
| 1548 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| 1549 | ||
| 1550 | if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH); | if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); |
| 1551 | ||
| 1552 | /* If checking an assertion for a condition, return MATCH_MATCH. */ | /* If checking an assertion for a condition, return MATCH_MATCH. */ |
| 1553 | ||
| # | Line 1569 for (;;) | Line 1577 for (;;) |
| 1577 | do | do |
| 1578 | { | { |
| 1579 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5); | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5); |
| 1580 | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH); | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH); |
| 1581 | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) |
| 1582 | { | { |
| 1583 | do ecode += GET(ecode,1); while (*ecode == OP_ALT); | do ecode += GET(ecode,1); while (*ecode == OP_ALT); |
| # | Line 1595 for (;;) | Line 1603 for (;;) |
| 1603 | back a number of characters, not bytes. */ | back a number of characters, not bytes. */ |
| 1604 | ||
| 1605 | case OP_REVERSE: | case OP_REVERSE: |
| 1606 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1607 | if (utf8) | if (utf) |
| 1608 | { | { |
| 1609 | i = GET(ecode, 1); | i = GET(ecode, 1); |
| 1610 | while (i-- > 0) | while (i-- > 0) |
| 1611 | { | { |
| 1612 | eptr--; | eptr--; |
| 1613 | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); |
| 1614 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 1615 | } | } |
| 1616 | } | } |
| # | Line 1613 for (;;) | Line 1621 for (;;) |
| 1621 | ||
| 1622 | { | { |
| 1623 | eptr -= GET(ecode, 1); | eptr -= GET(ecode, 1); |
| 1624 | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); |
| 1625 | } | } |
| 1626 | ||
| 1627 | /* Save the earliest consulted character, then skip to next op code */ | /* Save the earliest consulted character, then skip to next op code */ |
| # | Line 1627 for (;;) | Line 1635 for (;;) |
| 1635 | function is able to force a failure. */ | function is able to force a failure. */ |
| 1636 | ||
| 1637 | case OP_CALLOUT: | case OP_CALLOUT: |
| 1638 | if (pcre_callout != NULL) | if (PUBL(callout) != NULL) |
| 1639 | { | { |
| 1640 | pcre_callout_block cb; | PUBL(callout_block) cb; |
| 1641 | cb.version = 2; /* Version 1 of the callout block */ | cb.version = 2; /* Version 1 of the callout block */ |
| 1642 | cb.callout_number = ecode[1]; | cb.callout_number = ecode[1]; |
| 1643 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
| 1644 | #ifdef COMPILE_PCRE8 | |
| 1645 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
| 1646 | #else | |
| 1647 | cb.subject = (PCRE_SPTR16)md->start_subject; | |
| 1648 | #endif | |
| 1649 | cb.subject_length = (int)(md->end_subject - md->start_subject); | cb.subject_length = (int)(md->end_subject - md->start_subject); |
| 1650 | cb.start_match = (int)(mstart - md->start_subject); | cb.start_match = (int)(mstart - md->start_subject); |
| 1651 | cb.current_position = (int)(eptr - md->start_subject); | cb.current_position = (int)(eptr - md->start_subject); |
| # | Line 1642 for (;;) | Line 1654 for (;;) |
| 1654 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 1655 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 1656 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 1657 | cb.mark = markptr; | cb.mark = md->nomatch_mark; |
| 1658 | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); | if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH); |
| 1659 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 1660 | } | } |
| 1661 | ecode += 2 + 2*LINK_SIZE; | ecode += 2 + 2*LINK_SIZE; |
| # | Line 1702 for (;;) | Line 1714 for (;;) |
| 1714 | else | else |
| 1715 | { | { |
| 1716 | new_recursive.offset_save = | new_recursive.offset_save = |
| 1717 | (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int)); | (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int)); |
| 1718 | if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); | if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); |
| 1719 | } | } |
| 1720 | memcpy(new_recursive.offset_save, md->offset_vector, | memcpy(new_recursive.offset_save, md->offset_vector, |
| # | Line 1717 for (;;) | Line 1729 for (;;) |
| 1729 | do | do |
| 1730 | { | { |
| 1731 | if (cbegroup) md->match_function_type = MATCH_CBEGROUP; | if (cbegroup) md->match_function_type = MATCH_CBEGROUP; |
| 1732 | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, | RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top, |
| 1733 | md, eptrb, RM6); | md, eptrb, RM6); |
| 1734 | memcpy(md->offset_vector, new_recursive.offset_save, | memcpy(md->offset_vector, new_recursive.offset_save, |
| 1735 | new_recursive.saved_max * sizeof(int)); | new_recursive.saved_max * sizeof(int)); |
| # | Line 1726 for (;;) | Line 1738 for (;;) |
| 1738 | { | { |
| 1739 | DPRINTF(("Recursion matched\n")); | DPRINTF(("Recursion matched\n")); |
| 1740 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1741 | (pcre_free)(new_recursive.offset_save); | (PUBL(free))(new_recursive.offset_save); |
| 1742 | ||
| 1743 | /* Set where we got to in the subject, and reset the start in case | /* Set where we got to in the subject, and reset the start in case |
| 1744 | it was changed by \K. This *is* propagated back out of a recursion, | it was changed by \K. This *is* propagated back out of a recursion, |
| # | Line 1744 for (;;) | Line 1756 for (;;) |
| 1756 | { | { |
| 1757 | DPRINTF(("Recursion gave error %d\n", rrc)); | DPRINTF(("Recursion gave error %d\n", rrc)); |
| 1758 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1759 | (pcre_free)(new_recursive.offset_save); | (PUBL(free))(new_recursive.offset_save); |
| 1760 | RRETURN(rrc); | RRETURN(rrc); |
| 1761 | } | } |
| 1762 | ||
| # | Line 1756 for (;;) | Line 1768 for (;;) |
| 1768 | DPRINTF(("Recursion didn't match\n")); | DPRINTF(("Recursion didn't match\n")); |
| 1769 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1770 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1771 | (pcre_free)(new_recursive.offset_save); | (PUBL(free))(new_recursive.offset_save); |
| 1772 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 1773 | } | } |
| 1774 | ||
| 1775 | RECURSION_MATCHED: | RECURSION_MATCHED: |
| # | Line 1837 for (;;) | Line 1849 for (;;) |
| 1849 | md->end_match_ptr = eptr; /* For ONCE_NC */ | md->end_match_ptr = eptr; /* For ONCE_NC */ |
| 1850 | md->end_offset_top = offset_top; | md->end_offset_top = offset_top; |
| 1851 | md->start_match_ptr = mstart; | md->start_match_ptr = mstart; |
| 1852 | MRRETURN(MATCH_MATCH); /* Sets md->mark */ | RRETURN(MATCH_MATCH); /* Sets md->mark */ |
| 1853 | } | } |
| 1854 | ||
| 1855 | /* 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 1979 for (;;) | Line 1991 for (;;) |
| 1991 | /* Not multiline mode: start of subject assertion, unless notbol. */ | /* Not multiline mode: start of subject assertion, unless notbol. */ |
| 1992 | ||
| 1993 | case OP_CIRC: | case OP_CIRC: |
| 1994 | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); |
| 1995 | ||
| 1996 | /* Start of subject assertion */ | /* Start of subject assertion */ |
| 1997 | ||
| 1998 | case OP_SOD: | case OP_SOD: |
| 1999 | if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); |
| 2000 | ecode++; | ecode++; |
| 2001 | break; | break; |
| 2002 | ||
| 2003 | /* Multiline mode: start of subject unless notbol, or after any newline. */ | /* Multiline mode: start of subject unless notbol, or after any newline. */ |
| 2004 | ||
| 2005 | case OP_CIRCM: | case OP_CIRCM: |
| 2006 | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); |
| 2007 | if (eptr != md->start_subject && | if (eptr != md->start_subject && |
| 2008 | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) |
| 2009 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2010 | ecode++; | ecode++; |
| 2011 | break; | break; |
| 2012 | ||
| 2013 | /* Start of match assertion */ | /* Start of match assertion */ |
| 2014 | ||
| 2015 | case OP_SOM: | case OP_SOM: |
| 2016 | if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); |
| 2017 | ecode++; | ecode++; |
| 2018 | break; | break; |
| 2019 | ||
| # | Line 2017 for (;;) | Line 2029 for (;;) |
| 2029 | ||
| 2030 | case OP_DOLLM: | case OP_DOLLM: |
| 2031 | if (eptr < md->end_subject) | if (eptr < md->end_subject) |
| 2032 | { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); } | { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); } |
| 2033 | else | else |
| 2034 | { | { |
| 2035 | if (md->noteol) MRRETURN(MATCH_NOMATCH); | if (md->noteol) RRETURN(MATCH_NOMATCH); |
| 2036 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2037 | } | } |
| 2038 | ecode++; | ecode++; |
| # | Line 2030 for (;;) | Line 2042 for (;;) |
| 2042 | subject unless noteol is set. */ | subject unless noteol is set. */ |
| 2043 | ||
| 2044 | case OP_DOLL: | case OP_DOLL: |
| 2045 | if (md->noteol) MRRETURN(MATCH_NOMATCH); | if (md->noteol) RRETURN(MATCH_NOMATCH); |
| 2046 | if (!md->endonly) goto ASSERT_NL_OR_EOS; | if (!md->endonly) goto ASSERT_NL_OR_EOS; |
| 2047 | ||
| 2048 | /* ... else fall through for endonly */ | /* ... else fall through for endonly */ |
| # | Line 2038 for (;;) | Line 2050 for (;;) |
| 2050 | /* End of subject assertion (\z) */ | /* End of subject assertion (\z) */ |
| 2051 | ||
| 2052 | case OP_EOD: | case OP_EOD: |
| 2053 | if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH); | if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); |
| 2054 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2055 | ecode++; | ecode++; |
| 2056 | break; | break; |
| # | Line 2049 for (;;) | Line 2061 for (;;) |
| 2061 | ASSERT_NL_OR_EOS: | ASSERT_NL_OR_EOS: |
| 2062 | if (eptr < md->end_subject && | if (eptr < md->end_subject && |
| 2063 | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) |
| 2064 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2065 | ||
| 2066 | /* Either at end of string or \n before end. */ | /* Either at end of string or \n before end. */ |
| 2067 | ||
| # | Line 2068 for (;;) | Line 2080 for (;;) |
| 2080 | be "non-word" characters. Remember the earliest consulted character for | be "non-word" characters. Remember the earliest consulted character for |
| 2081 | partial matching. */ | partial matching. */ |
| 2082 | ||
| 2083 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2084 | if (utf8) | if (utf) |
| 2085 | { | { |
| 2086 | /* Get status of previous character */ | /* Get status of previous character */ |
| 2087 | ||
| 2088 | if (eptr == md->start_subject) prev_is_word = FALSE; else | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 2089 | { | { |
| 2090 | USPTR lastptr = eptr - 1; | PCRE_PUCHAR lastptr = eptr - 1; |
| 2091 | while((*lastptr & 0xc0) == 0x80) lastptr--; | BACKCHAR(lastptr); |
| 2092 | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; |
| 2093 | GETCHAR(c, lastptr); | GETCHAR(c, lastptr); |
| 2094 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| # | Line 2141 for (;;) | Line 2153 for (;;) |
| 2153 | } | } |
| 2154 | else | else |
| 2155 | #endif | #endif |
| 2156 | prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0); | prev_is_word = MAX_255(eptr[-1]) |
| 2157 | && ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
| 2158 | } | } |
| 2159 | ||
| 2160 | /* Get status of next character */ | /* Get status of next character */ |
| # | Line 2164 for (;;) | Line 2177 for (;;) |
| 2177 | } | } |
| 2178 | else | else |
| 2179 | #endif | #endif |
| 2180 | cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0); | cur_is_word = MAX_255(*eptr) |
| 2181 | && ((md->ctypes[*eptr] & ctype_word) != 0); | |
| 2182 | } | } |
| 2183 | ||
| 2184 | /* Now see if the situation is what we want */ | /* Now see if the situation is what we want */ |
| 2185 | ||
| 2186 | if ((*ecode++ == OP_WORD_BOUNDARY)? | if ((*ecode++ == OP_WORD_BOUNDARY)? |
| 2187 | cur_is_word == prev_is_word : cur_is_word != prev_is_word) | cur_is_word == prev_is_word : cur_is_word != prev_is_word) |
| 2188 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2189 | } | } |
| 2190 | break; | break; |
| 2191 | ||
| 2192 | /* Match a single character type; inline for speed */ | /* Match a single character type; inline for speed */ |
| 2193 | ||
| 2194 | case OP_ANY: | case OP_ANY: |
| 2195 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); |
| 2196 | /* Fall through */ | /* Fall through */ |
| 2197 | ||
| 2198 | case OP_ALLANY: | case OP_ALLANY: |
| 2199 | if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ | if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ |
| 2200 | { /* not be updated before SCHECK_PARTIAL. */ | { /* not be updated before SCHECK_PARTIAL. */ |
| 2201 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2202 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2203 | } | } |
| 2204 | eptr++; | eptr++; |
| 2205 | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | #ifdef SUPPORT_UTF |
| 2206 | if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); | |
| 2207 | #endif | |
| 2208 | ecode++; | ecode++; |
| 2209 | break; | break; |
| 2210 | ||
| # | Line 2199 for (;;) | Line 2215 for (;;) |
| 2215 | if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ | if (eptr >= md->end_subject) /* DO NOT merge the eptr++ here; it must */ |
| 2216 | { /* not be updated before SCHECK_PARTIAL. */ | { /* not be updated before SCHECK_PARTIAL. */ |
| 2217 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2218 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2219 | } | } |
| 2220 | eptr++; | eptr++; |
| 2221 | ecode++; | ecode++; |
| # | Line 2209 for (;;) | Line 2225 for (;;) |
| 2225 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2226 | { | { |
| 2227 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2228 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2229 | } | } |
| 2230 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2231 | if ( | if ( |
| 2232 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2233 | c < 256 && | c < 256 && |
| 2234 | #endif | #endif |
| 2235 | (md->ctypes[c] & ctype_digit) != 0 | (md->ctypes[c] & ctype_digit) != 0 |
| 2236 | ) | ) |
| 2237 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2238 | ecode++; | ecode++; |
| 2239 | break; | break; |
| 2240 | ||
| # | Line 2226 for (;;) | Line 2242 for (;;) |
| 2242 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2243 | { | { |
| 2244 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2245 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2246 | } | } |
| 2247 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2248 | if ( | if ( |
| 2249 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2250 | c >= 256 || | c > 255 || |
| 2251 | #endif | #endif |
| 2252 | (md->ctypes[c] & ctype_digit) == 0 | (md->ctypes[c] & ctype_digit) == 0 |
| 2253 | ) | ) |
| 2254 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2255 | ecode++; | ecode++; |
| 2256 | break; | break; |
| 2257 | ||
| # | Line 2243 for (;;) | Line 2259 for (;;) |
| 2259 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2260 | { | { |
| 2261 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2262 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2263 | } | } |
| 2264 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2265 | if ( | if ( |
| 2266 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2267 | c < 256 && | c < 256 && |
| 2268 | #endif | #endif |
| 2269 | (md->ctypes[c] & ctype_space) != 0 | (md->ctypes[c] & ctype_space) != 0 |
| 2270 | ) | ) |
| 2271 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2272 | ecode++; | ecode++; |
| 2273 | break; | break; |
| 2274 | ||
| # | Line 2260 for (;;) | Line 2276 for (;;) |
| 2276 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2277 | { | { |
| 2278 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2279 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2280 | } | } |
| 2281 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2282 | if ( | if ( |
| 2283 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2284 | c >= 256 || | c > 255 || |
| 2285 | #endif | #endif |
| 2286 | (md->ctypes[c] & ctype_space) == 0 | (md->ctypes[c] & ctype_space) == 0 |
| 2287 | ) | ) |
| 2288 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2289 | ecode++; | ecode++; |
| 2290 | break; | break; |
| 2291 | ||
| # | Line 2277 for (;;) | Line 2293 for (;;) |
| 2293 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2294 | { | { |
| 2295 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2296 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2297 | } | } |
| 2298 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2299 | if ( | if ( |
| 2300 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2301 | c < 256 && | c < 256 && |
| 2302 | #endif | #endif |
| 2303 | (md->ctypes[c] & ctype_word) != 0 | (md->ctypes[c] & ctype_word) != 0 |
| 2304 | ) | ) |
| 2305 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2306 | ecode++; | ecode++; |
| 2307 | break; | break; |
| 2308 | ||
| # | Line 2294 for (;;) | Line 2310 for (;;) |
| 2310 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2311 | { | { |
| 2312 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2313 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2314 | } | } |
| 2315 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2316 | if ( | if ( |
| 2317 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2318 | c >= 256 || | c > 255 || |
| 2319 | #endif | #endif |
| 2320 | (md->ctypes[c] & ctype_word) == 0 | (md->ctypes[c] & ctype_word) == 0 |
| 2321 | ) | ) |
| 2322 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2323 | ecode++; | ecode++; |
| 2324 | break; | break; |
| 2325 | ||
| # | Line 2311 for (;;) | Line 2327 for (;;) |
| 2327 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2328 | { | { |
| 2329 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2330 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2331 | } | } |
| 2332 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2333 | switch(c) | switch(c) |
| 2334 | { | { |
| 2335 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 2336 | ||
| 2337 | case 0x000d: | case 0x000d: |
| 2338 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| # | Line 2330 for (;;) | Line 2346 for (;;) |
| 2346 | case 0x0085: | case 0x0085: |
| 2347 | case 0x2028: | case 0x2028: |
| 2348 | case 0x2029: | case 0x2029: |
| 2349 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); |
| 2350 | break; | break; |
| 2351 | } | } |
| 2352 | ecode++; | ecode++; |
| # | Line 2340 for (;;) | Line 2356 for (;;) |
| 2356 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2357 | { | { |
| 2358 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2359 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2360 | } | } |
| 2361 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2362 | switch(c) | switch(c) |
| # | Line 2365 for (;;) | Line 2381 for (;;) |
| 2381 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 2382 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 2383 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 2384 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2385 | } | } |
| 2386 | ecode++; | ecode++; |
| 2387 | break; | break; |
| # | Line 2374 for (;;) | Line 2390 for (;;) |
| 2390 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2391 | { | { |
| 2392 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2393 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2394 | } | } |
| 2395 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2396 | switch(c) | switch(c) |
| 2397 | { | { |
| 2398 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 2399 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 2400 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 2401 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 2408 for (;;) | Line 2424 for (;;) |
| 2424 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2425 | { | { |
| 2426 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2427 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2428 | } | } |
| 2429 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2430 | switch(c) | switch(c) |
| # | Line 2421 for (;;) | Line 2437 for (;;) |
| 2437 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 2438 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 2439 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 2440 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2441 | } | } |
| 2442 | ecode++; | ecode++; |
| 2443 | break; | break; |
| # | Line 2430 for (;;) | Line 2446 for (;;) |
| 2446 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2447 | { | { |
| 2448 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2449 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2450 | } | } |
| 2451 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2452 | switch(c) | switch(c) |
| 2453 | { | { |
| 2454 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 2455 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 2456 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 2457 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 2457 for (;;) | Line 2473 for (;;) |
| 2473 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2474 | { | { |
| 2475 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2476 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2477 | } | } |
| 2478 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2479 | { | { |
| # | Line 2466 for (;;) | Line 2482 for (;;) |
| 2482 | switch(ecode[1]) | switch(ecode[1]) |
| 2483 | { | { |
| 2484 | case PT_ANY: | case PT_ANY: |
| 2485 | if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH); | if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); |
| 2486 | break; | break; |
| 2487 | ||
| 2488 | case PT_LAMP: | case PT_LAMP: |
| 2489 | if ((prop->chartype == ucp_Lu || | if ((prop->chartype == ucp_Lu || |
| 2490 | prop->chartype == ucp_Ll || | prop->chartype == ucp_Ll || |
| 2491 | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) |
| 2492 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2493 | break; | break; |
| 2494 | ||
| 2495 | case PT_GC: | case PT_GC: |
| 2496 | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) | if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP)) |
| 2497 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2498 | break; | break; |
| 2499 | ||
| 2500 | case PT_PC: | case PT_PC: |
| 2501 | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) |
| 2502 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2503 | break; | break; |
| 2504 | ||
| 2505 | case PT_SC: | case PT_SC: |
| 2506 | if ((ecode[2] != prop->script) == (op == OP_PROP)) | if ((ecode[2] != prop->script) == (op == OP_PROP)) |
| 2507 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2508 | break; | break; |
| 2509 | ||
| 2510 | /* These are specials */ | /* These are specials */ |
| 2511 | ||
| 2512 | case PT_ALNUM: | case PT_ALNUM: |
| 2513 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || |
| 2514 | _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) | PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) |
| 2515 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2516 | break; | break; |
| 2517 | ||
| 2518 | case PT_SPACE: /* Perl space */ | case PT_SPACE: /* Perl space */ |
| 2519 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z || |
| 2520 | c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) | c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) |
| 2521 | == (op == OP_NOTPROP)) | == (op == OP_NOTPROP)) |
| 2522 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2523 | break; | break; |
| 2524 | ||
| 2525 | case PT_PXSPACE: /* POSIX space */ | case PT_PXSPACE: /* POSIX space */ |
| 2526 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z || |
| 2527 | c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || | c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || |
| 2528 | c == CHAR_FF || c == CHAR_CR) | c == CHAR_FF || c == CHAR_CR) |
| 2529 | == (op == OP_NOTPROP)) | == (op == OP_NOTPROP)) |
| 2530 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2531 | break; | break; |
| 2532 | ||
| 2533 | case PT_WORD: | case PT_WORD: |
| 2534 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || |
| 2535 | _pcre_ucp_gentype[prop->chartype] == ucp_N || | PRIV(ucp_gentype)[prop->chartype] == ucp_N || |
| 2536 | c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) | c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) |
| 2537 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2538 | break; | break; |
| 2539 | ||
| 2540 | /* This should never occur */ | /* This should never occur */ |
| # | Line 2538 for (;;) | Line 2554 for (;;) |
| 2554 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2555 | { | { |
| 2556 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2557 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2558 | } | } |
| 2559 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2560 | if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); | if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); |
| 2561 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 2562 | { | { |
| 2563 | int len = 1; | int len = 1; |
| 2564 | if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } | if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
| 2565 | if (UCD_CATEGORY(c) != ucp_M) break; | if (UCD_CATEGORY(c) != ucp_M) break; |
| 2566 | eptr += len; | eptr += len; |
| 2567 | } | } |
| # | Line 2566 for (;;) | Line 2582 for (;;) |
| 2582 | case OP_REFI: | case OP_REFI: |
| 2583 | caseless = op == OP_REFI; | caseless = op == OP_REFI; |
| 2584 | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ |
| 2585 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 2586 | ||
| 2587 | /* If the reference is unset, there are two possibilities: | /* If the reference is unset, there are two possibilities: |
| 2588 | ||
| # | Line 2606 for (;;) | Line 2622 for (;;) |
| 2622 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
| 2623 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
| 2624 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
| 2625 | max = GET2(ecode, 3); | max = GET2(ecode, 1 + IMM2_SIZE); |
| 2626 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2627 | ecode += 5; | ecode += 1 + 2 * IMM2_SIZE; |
| 2628 | break; | break; |
| 2629 | ||
| 2630 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| 2631 | if ((length = match_ref(offset, eptr, length, md, caseless)) < 0) | if ((length = match_ref(offset, eptr, length, md, caseless)) < 0) |
| 2632 | { | { |
| 2633 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
| 2634 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2635 | } | } |
| 2636 | eptr += length; | eptr += length; |
| 2637 | continue; /* With the main loop */ | continue; /* With the main loop */ |
| 2638 | } | } |
| 2639 | ||
| 2640 | /* Handle repeated back references. If the length of the reference is | /* Handle repeated back references. If the length of the reference is |
| 2641 | zero, just continue with the main loop. */ | zero, just continue with the main loop. If the length is negative, it |
| 2642 | means the reference is unset in non-Java-compatible mode. If the minimum is | |
| 2643 | zero, we can continue at the same level without recursion. For any other | |
| 2644 | minimum, carrying on will result in NOMATCH. */ | |
| 2645 | ||
| 2646 | if (length == 0) continue; | if (length == 0) continue; |
| 2647 | if (length < 0 && min == 0) continue; | |
| 2648 | ||
| 2649 | /* First, ensure the minimum number of matches are present. We get back | /* First, ensure the minimum number of matches are present. We get back |
| 2650 | the length of the reference string explicitly rather than passing the | the length of the reference string explicitly rather than passing the |
| # | Line 2636 for (;;) | Line 2656 for (;;) |
| 2656 | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
| 2657 | { | { |
| 2658 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
| 2659 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2660 | } | } |
| 2661 | eptr += slength; | eptr += slength; |
| 2662 | } | } |
| # | Line 2655 for (;;) | Line 2675 for (;;) |
| 2675 | int slength; | int slength; |
| 2676 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM14); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM14); |
| 2677 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2678 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 2679 | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
| 2680 | { | { |
| 2681 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
| 2682 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2683 | } | } |
| 2684 | eptr += slength; | eptr += slength; |
| 2685 | } | } |
| # | Line 2687 for (;;) | Line 2707 for (;;) |
| 2707 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2708 | eptr -= length; | eptr -= length; |
| 2709 | } | } |
| 2710 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2711 | } | } |
| 2712 | /* Control never gets here */ | /* Control never gets here */ |
| 2713 | ||
| # | Line 2705 for (;;) | Line 2725 for (;;) |
| 2725 | case OP_NCLASS: | case OP_NCLASS: |
| 2726 | case OP_CLASS: | case OP_CLASS: |
| 2727 | { | { |
| 2728 | /* The data variable is saved across frames, so the byte map needs to | |
| 2729 | be stored there. */ | |
| 2730 | #define BYTE_MAP ((pcre_uint8 *)data) | |
| 2731 | data = ecode + 1; /* Save for matching */ | data = ecode + 1; /* Save for matching */ |
| 2732 | ecode += 33; /* Advance past the item */ | ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */ |
| 2733 | ||
| 2734 | switch (*ecode) | switch (*ecode) |
| 2735 | { | { |
| # | Line 2727 for (;;) | Line 2750 for (;;) |
| 2750 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
| 2751 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
| 2752 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
| 2753 | max = GET2(ecode, 3); | max = GET2(ecode, 1 + IMM2_SIZE); |
| 2754 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2755 | ecode += 5; | ecode += 1 + 2 * IMM2_SIZE; |
| 2756 | break; | break; |
| 2757 | ||
| 2758 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| # | Line 2739 for (;;) | Line 2762 for (;;) |
| 2762 | ||
| 2763 | /* First, ensure the minimum number of matches are present. */ | /* First, ensure the minimum number of matches are present. */ |
| 2764 | ||
| 2765 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2766 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 2767 | { | { |
| 2768 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2769 | { | { |
| 2770 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2771 | { | { |
| 2772 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2773 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2774 | } | } |
| 2775 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2776 | if (c > 255) | if (c > 255) |
| 2777 | { | { |
| 2778 | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); |
| 2779 | } | } |
| 2780 | else | else |
| 2781 | { | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); |
| if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | ||
| } | ||
| 2782 | } | } |
| 2783 | } | } |
| 2784 | else | else |
| 2785 | #endif | #endif |
| 2786 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 2787 | { | { |
| 2788 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2789 | { | { |
| 2790 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2791 | { | { |
| 2792 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2793 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2794 | } | } |
| 2795 | c = *eptr++; | c = *eptr++; |
| 2796 | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | #ifndef COMPILE_PCRE8 |
| 2797 | if (c > 255) | |
| 2798 | { | |
| 2799 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | |
| 2800 | } | |
| 2801 | else | |
| 2802 | #endif | |
| 2803 | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | |
| 2804 | } | } |
| 2805 | } | } |
| 2806 | ||
| # | Line 2787 for (;;) | Line 2814 for (;;) |
| 2814 | ||
| 2815 | if (minimize) | if (minimize) |
| 2816 | { | { |
| 2817 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2818 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 2819 | { | { |
| 2820 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 2821 | { | { |
| 2822 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM16); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM16); |
| 2823 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2824 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 2825 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2826 | { | { |
| 2827 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2828 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2829 | } | } |
| 2830 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2831 | if (c > 255) | if (c > 255) |
| 2832 | { | { |
| 2833 | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); |
| 2834 | } | } |
| 2835 | else | else |
| 2836 | { | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); |
| if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | ||
| } | ||
| 2837 | } | } |
| 2838 | } | } |
| 2839 | else | else |
| 2840 | #endif | #endif |
| 2841 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 2842 | { | { |
| 2843 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 2844 | { | { |
| 2845 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM17); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM17); |
| 2846 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2847 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 2848 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2849 | { | { |
| 2850 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2851 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2852 | } | } |
| 2853 | c = *eptr++; | c = *eptr++; |
| 2854 | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | #ifndef COMPILE_PCRE8 |
| 2855 | if (c > 255) | |
| 2856 | { | |
| 2857 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | |
| 2858 | } | |
| 2859 | else | |
| 2860 | #endif | |
| 2861 | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | |
| 2862 | } | } |
| 2863 | } | } |
| 2864 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2839 for (;;) | Line 2870 for (;;) |
| 2870 | { | { |
| 2871 | pp = eptr; | pp = eptr; |
| 2872 | ||
| 2873 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2874 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 2875 | { | { |
| 2876 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2877 | { | { |
| # | Line 2857 for (;;) | Line 2887 for (;;) |
| 2887 | if (op == OP_CLASS) break; | if (op == OP_CLASS) break; |
| 2888 | } | } |
| 2889 | else | else |
| 2890 | { | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; |
| if ((data[c/8] & (1 << (c&7))) == 0) break; | ||
| } | ||
| 2891 | eptr += len; | eptr += len; |
| 2892 | } | } |
| 2893 | for (;;) | for (;;) |
| # | Line 2872 for (;;) | Line 2900 for (;;) |
| 2900 | } | } |
| 2901 | else | else |
| 2902 | #endif | #endif |
| 2903 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 2904 | { | { |
| 2905 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2906 | { | { |
| # | Line 2882 for (;;) | Line 2910 for (;;) |
| 2910 | break; | break; |
| 2911 | } | } |
| 2912 | c = *eptr; | c = *eptr; |
| 2913 | if ((data[c/8] & (1 << (c&7))) == 0) break; | #ifndef COMPILE_PCRE8 |
| 2914 | if (c > 255) | |
| 2915 | { | |
| 2916 | if (op == OP_CLASS) break; | |
| 2917 | } | |
| 2918 | else | |
| 2919 | #endif | |
| 2920 | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; | |
| 2921 | eptr++; | eptr++; |
| 2922 | } | } |
| 2923 | while (eptr >= pp) | while (eptr >= pp) |
| # | Line 2893 for (;;) | Line 2928 for (;;) |
| 2928 | } | } |
| 2929 | } | } |
| 2930 | ||
| 2931 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2932 | } | } |
| 2933 | #undef BYTE_MAP | |
| 2934 | } | } |
| 2935 | /* Control never gets here */ | /* Control never gets here */ |
| 2936 | ||
| # | Line 2903 for (;;) | Line 2939 for (;;) |
| 2939 | when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8 | when UTF-8 mode mode is supported. Nevertheless, we may not be in UTF-8 |
| 2940 | mode, because Unicode properties are supported in non-UTF-8 mode. */ | mode, because Unicode properties are supported in non-UTF-8 mode. */ |
| 2941 | ||
| 2942 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
| 2943 | case OP_XCLASS: | case OP_XCLASS: |
| 2944 | { | { |
| 2945 | data = ecode + 1 + LINK_SIZE; /* Save for matching */ | data = ecode + 1 + LINK_SIZE; /* Save for matching */ |
| # | Line 2928 for (;;) | Line 2964 for (;;) |
| 2964 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
| 2965 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
| 2966 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
| 2967 | max = GET2(ecode, 3); | max = GET2(ecode, 1 + IMM2_SIZE); |
| 2968 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2969 | ecode += 5; | ecode += 1 + 2 * IMM2_SIZE; |
| 2970 | break; | break; |
| 2971 | ||
| 2972 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| # | Line 2945 for (;;) | Line 2981 for (;;) |
| 2981 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2982 | { | { |
| 2983 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2984 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2985 | } | } |
| 2986 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2987 | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); | if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); |
| 2988 | } | } |
| 2989 | ||
| 2990 | /* 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 2965 for (;;) | Line 3001 for (;;) |
| 3001 | { | { |
| 3002 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM20); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM20); |
| 3003 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3004 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3005 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3006 | { | { |
| 3007 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3008 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3009 | } | } |
| 3010 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3011 | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); | if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); |
| 3012 | } | } |
| 3013 | /* Control never gets here */ | /* Control never gets here */ |
| 3014 | } | } |
| # | Line 2990 for (;;) | Line 3026 for (;;) |
| 3026 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3027 | break; | break; |
| 3028 | } | } |
| 3029 | #ifdef SUPPORT_UTF | |
| 3030 | GETCHARLENTEST(c, eptr, len); | GETCHARLENTEST(c, eptr, len); |
| 3031 | if (!_pcre_xclass(c, data)) break; | #else |
| 3032 | c = *eptr; | |
| 3033 | #endif | |
| 3034 | if (!PRIV(xclass)(c, data, utf)) break; | |
| 3035 | eptr += len; | eptr += len; |
| 3036 | } | } |
| 3037 | for(;;) | for(;;) |
| # | Line 2999 for (;;) | Line 3039 for (;;) |
| 3039 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM21); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM21); |
| 3040 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3041 | if (eptr-- == pp) break; /* Stop if tried at original pos */ | if (eptr-- == pp) break; /* Stop if tried at original pos */ |
| 3042 | if (utf8) BACKCHAR(eptr); | #ifdef SUPPORT_UTF |
| 3043 | if (utf) BACKCHAR(eptr); | |
| 3044 | #endif | |
| 3045 | } | } |
| 3046 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3047 | } | } |
| 3048 | ||
| 3049 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3011 for (;;) | Line 3053 for (;;) |
| 3053 | /* Match a single character, casefully */ | /* Match a single character, casefully */ |
| 3054 | ||
| 3055 | case OP_CHAR: | case OP_CHAR: |
| 3056 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3057 | if (utf8) | if (utf) |
| 3058 | { | { |
| 3059 | length = 1; | length = 1; |
| 3060 | ecode++; | ecode++; |
| # | Line 3020 for (;;) | Line 3062 for (;;) |
| 3062 | if (length > md->end_subject - eptr) | if (length > md->end_subject - eptr) |
| 3063 | { | { |
| 3064 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ |
| 3065 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3066 | } | } |
| 3067 | while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH); | while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH); |
| 3068 | } | } |
| 3069 | else | else |
| 3070 | #endif | #endif |
| 3071 | /* Not UTF mode */ | |
| /* Non-UTF-8 mode */ | ||
| 3072 | { | { |
| 3073 | if (md->end_subject - eptr < 1) | if (md->end_subject - eptr < 1) |
| 3074 | { | { |
| 3075 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ |
| 3076 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3077 | } | } |
| 3078 | if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH); | if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); |
| 3079 | ecode += 2; | ecode += 2; |
| 3080 | } | } |
| 3081 | break; | break; |
| 3082 | ||
| 3083 | /* Match a single character, caselessly */ | /* Match a single character, caselessly. If we are at the end of the |
| 3084 | subject, give up immediately. */ | |
| 3085 | ||
| 3086 | case OP_CHARI: | case OP_CHARI: |
| 3087 | #ifdef SUPPORT_UTF8 | if (eptr >= md->end_subject) |
| 3088 | if (utf8) | { |
| 3089 | SCHECK_PARTIAL(); | |
| 3090 | RRETURN(MATCH_NOMATCH); | |
| 3091 | } | |
| 3092 | ||
| 3093 | #ifdef SUPPORT_UTF | |
| 3094 | if (utf) | |
| 3095 | { | { |
| 3096 | length = 1; | length = 1; |
| 3097 | ecode++; | ecode++; |
| 3098 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 3099 | ||
| if (length > md->end_subject - eptr) | ||
| { | ||
| CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | ||
| MRRETURN(MATCH_NOMATCH); | ||
| } | ||
| 3100 | /* 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 |
| 3101 | can use the fast lookup table. */ | we know that its other case must also be one byte long, so we can use the |
| 3102 | fast lookup table. We know that there is at least one byte left in the | |
| 3103 | subject. */ | |
| 3104 | ||
| 3105 | if (fc < 128) | if (fc < 128) |
| 3106 | { | { |
| 3107 | if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (md->lcc[fc] |
| 3108 | != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH); | |
| 3109 | ecode++; | |
| 3110 | eptr++; | |
| 3111 | } | } |
| 3112 | ||
| 3113 | /* Otherwise we must pick up the subject character */ | /* Otherwise we must pick up the subject character. Note that we cannot |
| 3114 | use the value of "length" to check for sufficient bytes left, because the | |
| 3115 | other case of the character may have more or fewer bytes. */ | |
| 3116 | ||
| 3117 | else | else |
| 3118 | { | { |
| # | Line 3079 for (;;) | Line 3128 for (;;) |
| 3128 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3129 | if (dc != UCD_OTHERCASE(fc)) | if (dc != UCD_OTHERCASE(fc)) |
| 3130 | #endif | #endif |
| 3131 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3132 | } | } |
| 3133 | } | } |
| 3134 | } | } |
| 3135 | else | else |
| 3136 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF */ |
| 3137 | ||
| 3138 | /* Non-UTF-8 mode */ | /* Not UTF mode */ |
| 3139 | { | { |
| 3140 | if (md->end_subject - eptr < 1) | if (TABLE_GET(ecode[1], md->lcc, ecode[1]) |
| 3141 | { | != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH); |
| 3142 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | eptr++; |
| MRRETURN(MATCH_NOMATCH); | ||
| } | ||
| if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | ||
| 3143 | ecode += 2; | ecode += 2; |
| 3144 | } | } |
| 3145 | break; | break; |
| # | Line 3103 for (;;) | Line 3149 for (;;) |
| 3149 | case OP_EXACT: | case OP_EXACT: |
| 3150 | case OP_EXACTI: | case OP_EXACTI: |
| 3151 | min = max = GET2(ecode, 1); | min = max = GET2(ecode, 1); |
| 3152 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3153 | goto REPEATCHAR; | goto REPEATCHAR; |
| 3154 | ||
| 3155 | case OP_POSUPTO: | case OP_POSUPTO: |
| # | Line 3118 for (;;) | Line 3164 for (;;) |
| 3164 | min = 0; | min = 0; |
| 3165 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
| 3166 | minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI; | minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI; |
| 3167 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3168 | goto REPEATCHAR; | goto REPEATCHAR; |
| 3169 | ||
| 3170 | case OP_POSSTAR: | case OP_POSSTAR: |
| # | Line 3166 for (;;) | Line 3212 for (;;) |
| 3212 | /* Common code for all repeated single-character matches. */ | /* Common code for all repeated single-character matches. */ |
| 3213 | ||
| 3214 | REPEATCHAR: | REPEATCHAR: |
| 3215 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3216 | if (utf8) | if (utf) |
| 3217 | { | { |
| 3218 | length = 1; | length = 1; |
| 3219 | charptr = ecode; | charptr = ecode; |
| # | Line 3183 for (;;) | Line 3229 for (;;) |
| 3229 | unsigned int othercase; | unsigned int othercase; |
| 3230 | if (op >= OP_STARI && /* Caseless */ | if (op >= OP_STARI && /* Caseless */ |
| 3231 | (othercase = UCD_OTHERCASE(fc)) != fc) | (othercase = UCD_OTHERCASE(fc)) != fc) |
| 3232 | oclength = _pcre_ord2utf8(othercase, occhars); | oclength = PRIV(ord2utf)(othercase, occhars); |
| 3233 | else oclength = 0; | else oclength = 0; |
| 3234 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 3235 | ||
| 3236 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3237 | { | { |
| 3238 | if (eptr <= md->end_subject - length && | if (eptr <= md->end_subject - length && |
| 3239 | memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; |
| 3240 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3241 | else if (oclength > 0 && | else if (oclength > 0 && |
| 3242 | eptr <= md->end_subject - oclength && | eptr <= md->end_subject - oclength && |
| 3243 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; |
| 3244 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 3245 | else | else |
| 3246 | { | { |
| 3247 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
| 3248 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3249 | } | } |
| 3250 | } | } |
| 3251 | ||
| # | Line 3211 for (;;) | Line 3257 for (;;) |
| 3257 | { | { |
| 3258 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM22); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM22); |
| 3259 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3260 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3261 | if (eptr <= md->end_subject - length && | if (eptr <= md->end_subject - length && |
| 3262 | memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; |
| 3263 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3264 | else if (oclength > 0 && | else if (oclength > 0 && |
| 3265 | eptr <= md->end_subject - oclength && | eptr <= md->end_subject - oclength && |
| 3266 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; |
| 3267 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 3268 | else | else |
| 3269 | { | { |
| 3270 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
| 3271 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3272 | } | } |
| 3273 | } | } |
| 3274 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3234 for (;;) | Line 3280 for (;;) |
| 3280 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3281 | { | { |
| 3282 | if (eptr <= md->end_subject - length && | if (eptr <= md->end_subject - length && |
| 3283 | memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; |
| 3284 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3285 | else if (oclength > 0 && | else if (oclength > 0 && |
| 3286 | eptr <= md->end_subject - oclength && | eptr <= md->end_subject - oclength && |
| 3287 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; |
| 3288 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 3289 | else | else |
| 3290 | { | { |
| # | Line 3253 for (;;) | Line 3299 for (;;) |
| 3299 | { | { |
| 3300 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM23); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM23); |
| 3301 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3302 | if (eptr == pp) { MRRETURN(MATCH_NOMATCH); } | if (eptr == pp) { RRETURN(MATCH_NOMATCH); } |
| 3303 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3304 | eptr--; | eptr--; |
| 3305 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| # | Line 3270 for (;;) | Line 3316 for (;;) |
| 3316 | value of fc will always be < 128. */ | value of fc will always be < 128. */ |
| 3317 | } | } |
| 3318 | else | else |
| 3319 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF */ |
| 3320 | /* When not in UTF-8 mode, load a single-byte character. */ | |
| 3321 | /* When not in UTF-8 mode, load a single-byte character. */ | fc = *ecode++; |
| fc = *ecode++; | ||
| 3322 | ||
| 3323 | /* The value of fc at this point is always less than 256, though we may or | /* The value of fc at this point is always one character, though we may |
| 3324 | may not be in UTF-8 mode. The code is duplicated for the caseless and | or may not be in UTF mode. The code is duplicated for the caseless and |
| 3325 | caseful cases, for speed, since matching characters is likely to be quite | caseful cases, for speed, since matching characters is likely to be quite |
| 3326 | common. First, ensure the minimum number of matches are present. If min = | common. First, ensure the minimum number of matches are present. If min = |
| 3327 | max, continue at the same level without recursing. Otherwise, if | max, continue at the same level without recursing. Otherwise, if |
| # | Line 3290 for (;;) | Line 3334 for (;;) |
| 3334 | ||
| 3335 | if (op >= OP_STARI) /* Caseless */ | if (op >= OP_STARI) /* Caseless */ |
| 3336 | { | { |
| 3337 | fc = md->lcc[fc]; | #ifdef COMPILE_PCRE8 |
| 3338 | /* fc must be < 128 if UTF is enabled. */ | |
| 3339 | foc = md->fcc[fc]; | |
| 3340 | #else | |
| 3341 | #ifdef SUPPORT_UTF | |
| 3342 | #ifdef SUPPORT_UCP | |
| 3343 | if (utf && fc > 127) | |
| 3344 | foc = UCD_OTHERCASE(fc); | |
| 3345 | #else | |
| 3346 | if (utf && fc > 127) | |
| 3347 | foc = fc; | |
| 3348 | #endif /* SUPPORT_UCP */ | |
| 3349 | else | |
| 3350 | #endif /* SUPPORT_UTF */ | |
| 3351 | foc = TABLE_GET(fc, md->fcc, fc); | |
| 3352 | #endif /* COMPILE_PCRE8 */ | |
| 3353 | ||
| 3354 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3355 | { | { |
| 3356 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3357 | { | { |
| 3358 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3359 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3360 | } | } |
| 3361 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH); |
| 3362 | eptr++; | |
| 3363 | } | } |
| 3364 | if (min == max) continue; | if (min == max) continue; |
| 3365 | if (minimize) | if (minimize) |
| # | Line 3307 for (;;) | Line 3368 for (;;) |
| 3368 | { | { |
| 3369 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM24); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM24); |
| 3370 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3371 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3372 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3373 | { | { |
| 3374 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3375 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3376 | } | } |
| 3377 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH); |
| 3378 | eptr++; | |
| 3379 | } | } |
| 3380 | /* Control never gets here */ | /* Control never gets here */ |
| 3381 | } | } |
| # | Line 3327 for (;;) | Line 3389 for (;;) |
| 3389 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3390 | break; | break; |
| 3391 | } | } |
| 3392 | if (fc != md->lcc[*eptr]) break; | if (fc != *eptr && foc != *eptr) break; |
| 3393 | eptr++; | eptr++; |
| 3394 | } | } |
| 3395 | ||
| # | Line 3339 for (;;) | Line 3401 for (;;) |
| 3401 | eptr--; | eptr--; |
| 3402 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3403 | } | } |
| 3404 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3405 | } | } |
| 3406 | /* Control never gets here */ | /* Control never gets here */ |
| 3407 | } | } |
| # | Line 3353 for (;;) | Line 3415 for (;;) |
| 3415 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3416 | { | { |
| 3417 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3418 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3419 | } | } |
| 3420 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr++) RRETURN(MATCH_NOMATCH); |
| 3421 | } | } |
| 3422 | ||
| 3423 | if (min == max) continue; | if (min == max) continue; |
| # | Line 3366 for (;;) | Line 3428 for (;;) |
| 3428 | { | { |
| 3429 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM26); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM26); |
| 3430 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3431 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3432 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3433 | { | { |
| 3434 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3435 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3436 | } | } |
| 3437 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr++) RRETURN(MATCH_NOMATCH); |
| 3438 | } | } |
| 3439 | /* Control never gets here */ | /* Control never gets here */ |
| 3440 | } | } |
| # | Line 3397 for (;;) | Line 3459 for (;;) |
| 3459 | eptr--; | eptr--; |
| 3460 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3461 | } | } |
| 3462 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3463 | } | } |
| 3464 | } | } |
| 3465 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3410 for (;;) | Line 3472 for (;;) |
| 3472 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3473 | { | { |
| 3474 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3475 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3476 | } | } |
| 3477 | ecode++; | ecode++; |
| 3478 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3479 | if (op == OP_NOTI) /* The caseless case */ | if (op == OP_NOTI) /* The caseless case */ |
| 3480 | { | { |
| 3481 | #ifdef SUPPORT_UTF8 | register int ch, och; |
| 3482 | if (c < 256) | ch = *ecode++; |
| 3483 | #endif | #ifdef COMPILE_PCRE8 |
| 3484 | c = md->lcc[c]; | /* ch must be < 128 if UTF is enabled. */ |
| 3485 | if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH); | och = md->fcc[ch]; |
| 3486 | #else | |
| 3487 | #ifdef SUPPORT_UTF | |
| 3488 | #ifdef SUPPORT_UCP | |
| 3489 | if (utf && ch > 127) | |
| 3490 | och = UCD_OTHERCASE(ch); | |
| 3491 | #else | |
| 3492 | if (utf && ch > 127) | |
| 3493 | och = ch; | |
| 3494 | #endif /* SUPPORT_UCP */ | |
| 3495 | else | |
| 3496 | #endif /* SUPPORT_UTF */ | |
| 3497 | och = TABLE_GET(ch, md->fcc, ch); | |
| 3498 | #endif /* COMPILE_PCRE8 */ | |
| 3499 | if (ch == c || och == c) RRETURN(MATCH_NOMATCH); | |
| 3500 | } | } |
| 3501 | else /* Caseful */ | else /* Caseful */ |
| 3502 | { | { |
| 3503 | if (*ecode++ == c) MRRETURN(MATCH_NOMATCH); | if (*ecode++ == c) RRETURN(MATCH_NOMATCH); |
| 3504 | } | } |
| 3505 | break; | break; |
| 3506 | ||
| # | Line 3438 for (;;) | Line 3514 for (;;) |
| 3514 | case OP_NOTEXACT: | case OP_NOTEXACT: |
| 3515 | case OP_NOTEXACTI: | case OP_NOTEXACTI: |
| 3516 | min = max = GET2(ecode, 1); | min = max = GET2(ecode, 1); |
| 3517 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3518 | goto REPEATNOTCHAR; | goto REPEATNOTCHAR; |
| 3519 | ||
| 3520 | case OP_NOTUPTO: | case OP_NOTUPTO: |
| # | Line 3448 for (;;) | Line 3524 for (;;) |
| 3524 | min = 0; | min = 0; |
| 3525 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
| 3526 | minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI; | minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI; |
| 3527 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3528 | goto REPEATNOTCHAR; | goto REPEATNOTCHAR; |
| 3529 | ||
| 3530 | case OP_NOTPOSSTAR: | case OP_NOTPOSSTAR: |
| # | Line 3480 for (;;) | Line 3556 for (;;) |
| 3556 | possessive = TRUE; | possessive = TRUE; |
| 3557 | min = 0; | min = 0; |
| 3558 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
| 3559 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3560 | goto REPEATNOTCHAR; | goto REPEATNOTCHAR; |
| 3561 | ||
| 3562 | case OP_NOTSTAR: | case OP_NOTSTAR: |
| # | Line 3519 for (;;) | Line 3595 for (;;) |
| 3595 | ||
| 3596 | if (op >= OP_NOTSTARI) /* Caseless */ | if (op >= OP_NOTSTARI) /* Caseless */ |
| 3597 | { | { |
| 3598 | fc = md->lcc[fc]; | #ifdef COMPILE_PCRE8 |
| 3599 | /* fc must be < 128 if UTF is enabled. */ | |
| 3600 | foc = md->fcc[fc]; | |
| 3601 | #else | |
| 3602 | #ifdef SUPPORT_UTF | |
| 3603 | #ifdef SUPPORT_UCP | |
| 3604 | if (utf && fc > 127) | |
| 3605 | foc = UCD_OTHERCASE(fc); | |
| 3606 | #else | |
| 3607 | if (utf && fc > 127) | |
| 3608 | foc = fc; | |
| 3609 | #endif /* SUPPORT_UCP */ | |
| 3610 | else | |
| 3611 | #endif /* SUPPORT_UTF */ | |
| 3612 | foc = TABLE_GET(fc, md->fcc, fc); | |
| 3613 | #endif /* COMPILE_PCRE8 */ | |
| 3614 | ||
| 3615 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3616 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3617 | { | { |
| 3618 | register unsigned int d; | register unsigned int d; |
| 3619 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| # | Line 3531 for (;;) | Line 3621 for (;;) |
| 3621 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3622 | { | { |
| 3623 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3624 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3625 | } | } |
| 3626 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3627 | if (d < 256) d = md->lcc[d]; | if (fc == d || foc == d) RRETURN(MATCH_NOMATCH); |
| if (fc == d) MRRETURN(MATCH_NOMATCH); | ||
| 3628 | } | } |
| 3629 | } | } |
| 3630 | else | else |
| 3631 | #endif | #endif |
| 3632 | /* Not UTF mode */ | |
| /* Not UTF-8 mode */ | ||
| 3633 | { | { |
| 3634 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3635 | { | { |
| 3636 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3637 | { | { |
| 3638 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3639 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3640 | } | } |
| 3641 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); |
| 3642 | eptr++; | |
| 3643 | } | } |
| 3644 | } | } |
| 3645 | ||
| # | Line 3558 for (;;) | Line 3647 for (;;) |
| 3647 | ||
| 3648 | if (minimize) | if (minimize) |
| 3649 | { | { |
| 3650 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3651 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3652 | { | { |
| 3653 | register unsigned int d; | register unsigned int d; |
| 3654 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3655 | { | { |
| 3656 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM28); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM28); |
| 3657 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3658 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3659 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3660 | { | { |
| 3661 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3662 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3663 | } | } |
| 3664 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3665 | if (d < 256) d = md->lcc[d]; | if (fc == d || foc == d) RRETURN(MATCH_NOMATCH); |
| if (fc == d) MRRETURN(MATCH_NOMATCH); | ||
| 3666 | } | } |
| 3667 | } | } |
| 3668 | else | else |
| 3669 | #endif | #endif |
| 3670 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 3671 | { | { |
| 3672 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3673 | { | { |
| 3674 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM29); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM29); |
| 3675 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3676 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3677 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3678 | { | { |
| 3679 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3680 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3681 | } | } |
| 3682 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); |
| 3683 | eptr++; | |
| 3684 | } | } |
| 3685 | } | } |
| 3686 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3604 for (;;) | Line 3692 for (;;) |
| 3692 | { | { |
| 3693 | pp = eptr; | pp = eptr; |
| 3694 | ||
| 3695 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3696 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3697 | { | { |
| 3698 | register unsigned int d; | register unsigned int d; |
| 3699 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| # | Line 3618 for (;;) | Line 3705 for (;;) |
| 3705 | break; | break; |
| 3706 | } | } |
| 3707 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3708 | if (d < 256) d = md->lcc[d]; | if (fc == d || foc == d) break; |
| if (fc == d) break; | ||
| 3709 | eptr += len; | eptr += len; |
| 3710 | } | } |
| 3711 | if (possessive) continue; | if (possessive) continue; |
| # | Line 3633 for (;;) | Line 3719 for (;;) |
| 3719 | } | } |
| 3720 | else | else |
| 3721 | #endif | #endif |
| 3722 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 3723 | { | { |
| 3724 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3725 | { | { |
| # | Line 3642 for (;;) | Line 3728 for (;;) |
| 3728 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3729 | break; | break; |
| 3730 | } | } |
| 3731 | if (fc == md->lcc[*eptr]) break; | if (fc == *eptr || foc == *eptr) break; |
| 3732 | eptr++; | eptr++; |
| 3733 | } | } |
| 3734 | if (possessive) continue; | if (possessive) continue; |
| # | Line 3654 for (;;) | Line 3740 for (;;) |
| 3740 | } | } |
| 3741 | } | } |
| 3742 | ||
| 3743 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3744 | } | } |
| 3745 | /* Control never gets here */ | /* Control never gets here */ |
| 3746 | } | } |
| # | Line 3663 for (;;) | Line 3749 for (;;) |
| 3749 | ||
| 3750 | else | else |
| 3751 | { | { |
| 3752 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3753 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3754 | { | { |
| 3755 | register unsigned int d; | register unsigned int d; |
| 3756 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| # | Line 3673 for (;;) | Line 3758 for (;;) |
| 3758 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3759 | { | { |
| 3760 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3761 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3762 | } | } |
| 3763 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3764 | if (fc == d) MRRETURN(MATCH_NOMATCH); | if (fc == d) RRETURN(MATCH_NOMATCH); |
| 3765 | } | } |
| 3766 | } | } |
| 3767 | else | else |
| 3768 | #endif | #endif |
| 3769 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 3770 | { | { |
| 3771 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3772 | { | { |
| 3773 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3774 | { | { |
| 3775 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3776 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3777 | } | } |
| 3778 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); |
| 3779 | } | } |
| 3780 | } | } |
| 3781 | ||
| # | Line 3698 for (;;) | Line 3783 for (;;) |
| 3783 | ||
| 3784 | if (minimize) | if (minimize) |
| 3785 | { | { |
| 3786 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3787 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3788 | { | { |
| 3789 | register unsigned int d; | register unsigned int d; |
| 3790 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3791 | { | { |
| 3792 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM32); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM32); |
| 3793 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3794 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3795 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3796 | { | { |
| 3797 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3798 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3799 | } | } |
| 3800 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3801 | if (fc == d) MRRETURN(MATCH_NOMATCH); | if (fc == d) RRETURN(MATCH_NOMATCH); |
| 3802 | } | } |
| 3803 | } | } |
| 3804 | else | else |
| 3805 | #endif | #endif |
| 3806 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 3807 | { | { |
| 3808 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3809 | { | { |
| 3810 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM33); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM33); |
| 3811 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3812 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3813 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3814 | { | { |
| 3815 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3816 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3817 | } | } |
| 3818 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); |
| 3819 | } | } |
| 3820 | } | } |
| 3821 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3743 for (;;) | Line 3827 for (;;) |
| 3827 | { | { |
| 3828 | pp = eptr; | pp = eptr; |
| 3829 | ||
| 3830 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3831 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3832 | { | { |
| 3833 | register unsigned int d; | register unsigned int d; |
| 3834 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| # | Line 3771 for (;;) | Line 3854 for (;;) |
| 3854 | } | } |
| 3855 | else | else |
| 3856 | #endif | #endif |
| 3857 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 3858 | { | { |
| 3859 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3860 | { | { |
| # | Line 3792 for (;;) | Line 3875 for (;;) |
| 3875 | } | } |
| 3876 | } | } |
| 3877 | ||
| 3878 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3879 | } | } |
| 3880 | } | } |
| 3881 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3804 for (;;) | Line 3887 for (;;) |
| 3887 | case OP_TYPEEXACT: | case OP_TYPEEXACT: |
| 3888 | min = max = GET2(ecode, 1); | min = max = GET2(ecode, 1); |
| 3889 | minimize = TRUE; | minimize = TRUE; |
| 3890 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3891 | goto REPEATTYPE; | goto REPEATTYPE; |
| 3892 | ||
| 3893 | case OP_TYPEUPTO: | case OP_TYPEUPTO: |
| # | Line 3812 for (;;) | Line 3895 for (;;) |
| 3895 | min = 0; | min = 0; |
| 3896 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
| 3897 | minimize = *ecode == OP_TYPEMINUPTO; | minimize = *ecode == OP_TYPEMINUPTO; |
| 3898 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3899 | goto REPEATTYPE; | goto REPEATTYPE; |
| 3900 | ||
| 3901 | case OP_TYPEPOSSTAR: | case OP_TYPEPOSSTAR: |
| # | Line 3840 for (;;) | Line 3923 for (;;) |
| 3923 | possessive = TRUE; | possessive = TRUE; |
| 3924 | min = 0; | min = 0; |
| 3925 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
| 3926 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3927 | goto REPEATTYPE; | goto REPEATTYPE; |
| 3928 | ||
| 3929 | case OP_TYPESTAR: | case OP_TYPESTAR: |
| # | Line 3886 for (;;) | Line 3969 for (;;) |
| 3969 | switch(prop_type) | switch(prop_type) |
| 3970 | { | { |
| 3971 | case PT_ANY: | case PT_ANY: |
| 3972 | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); | if (prop_fail_result) RRETURN(MATCH_NOMATCH); |
| 3973 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3974 | { | { |
| 3975 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3976 | { | { |
| 3977 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3978 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3979 | } | } |
| 3980 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3981 | } | } |
| # | Line 3905 for (;;) | Line 3988 for (;;) |
| 3988 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3989 | { | { |
| 3990 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3991 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3992 | } | } |
| 3993 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3994 | chartype = UCD_CHARTYPE(c); | chartype = UCD_CHARTYPE(c); |
| 3995 | if ((chartype == ucp_Lu || | if ((chartype == ucp_Lu || |
| 3996 | chartype == ucp_Ll || | chartype == ucp_Ll || |
| 3997 | chartype == ucp_Lt) == prop_fail_result) | chartype == ucp_Lt) == prop_fail_result) |
| 3998 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3999 | } | } |
| 4000 | break; | break; |
| 4001 | ||
| # | Line 3922 for (;;) | Line 4005 for (;;) |
| 4005 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4006 | { | { |
| 4007 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4008 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4009 | } | } |
| 4010 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4011 | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) |
| 4012 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4013 | } | } |
| 4014 | break; | break; |
| 4015 | ||
| # | Line 3936 for (;;) | Line 4019 for (;;) |
| 4019 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4020 | { | { |
| 4021 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4022 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4023 | } | } |
| 4024 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4025 | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) |
| 4026 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4027 | } | } |
| 4028 | break; | break; |
| 4029 | ||
| # | Line 3950 for (;;) | Line 4033 for (;;) |
| 4033 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4034 | { | { |
| 4035 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4036 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4037 | } | } |
| 4038 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4039 | if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) | if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) |
| 4040 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4041 | } | } |
| 4042 | break; | break; |
| 4043 | ||
| # | Line 3965 for (;;) | Line 4048 for (;;) |
| 4048 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4049 | { | { |
| 4050 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4051 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4052 | } | } |
| 4053 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4054 | category = UCD_CATEGORY(c); | category = UCD_CATEGORY(c); |
| 4055 | if ((category == ucp_L || category == ucp_N) == prop_fail_result) | if ((category == ucp_L || category == ucp_N) == prop_fail_result) |
| 4056 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4057 | } | } |
| 4058 | break; | break; |
| 4059 | ||
| # | Line 3980 for (;;) | Line 4063 for (;;) |
| 4063 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4064 | { | { |
| 4065 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4066 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4067 | } | } |
| 4068 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4069 | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
| 4070 | c == CHAR_FF || c == CHAR_CR) | c == CHAR_FF || c == CHAR_CR) |
| 4071 | == prop_fail_result) | == prop_fail_result) |
| 4072 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4073 | } | } |
| 4074 | break; | break; |
| 4075 | ||
| # | Line 3996 for (;;) | Line 4079 for (;;) |
| 4079 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4080 | { | { |
| 4081 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4082 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4083 | } | } |
| 4084 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4085 | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
| 4086 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
| 4087 | == prop_fail_result) | == prop_fail_result) |
| 4088 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4089 | } | } |
| 4090 | break; | break; |
| 4091 | ||
| # | Line 4013 for (;;) | Line 4096 for (;;) |
| 4096 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4097 | { | { |
| 4098 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4099 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4100 | } | } |
| 4101 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4102 | category = UCD_CATEGORY(c); | category = UCD_CATEGORY(c); |
| 4103 | if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE) | if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE) |
| 4104 | == prop_fail_result) | == prop_fail_result) |
| 4105 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4106 | } | } |
| 4107 | break; | break; |
| 4108 | ||
| # | Line 4040 for (;;) | Line 4123 for (;;) |
| 4123 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4124 | { | { |
| 4125 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4126 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4127 | } | } |
| 4128 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4129 | if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); | if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); |
| 4130 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 4131 | { | { |
| 4132 | int len = 1; | int len = 1; |
| 4133 | if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } | if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
| 4134 | if (UCD_CATEGORY(c) != ucp_M) break; | if (UCD_CATEGORY(c) != ucp_M) break; |
| 4135 | eptr += len; | eptr += len; |
| 4136 | } | } |
| # | Line 4059 for (;;) | Line 4142 for (;;) |
| 4142 | ||
| 4143 | /* Handle all other cases when the coding is UTF-8 */ | /* Handle all other cases when the coding is UTF-8 */ |
| 4144 | ||
| 4145 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 4146 | if (utf8) switch(ctype) | if (utf) switch(ctype) |
| 4147 | { | { |
| 4148 | case OP_ANY: | case OP_ANY: |
| 4149 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| # | Line 4068 for (;;) | Line 4151 for (;;) |
| 4151 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4152 | { | { |
| 4153 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4154 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4155 | } | } |
| 4156 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); |
| 4157 | eptr++; | eptr++; |
| 4158 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); |
| 4159 | } | } |
| 4160 | break; | break; |
| 4161 | ||
| # | Line 4082 for (;;) | Line 4165 for (;;) |
| 4165 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4166 | { | { |
| 4167 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4168 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4169 | } | } |
| 4170 | eptr++; | eptr++; |
| 4171 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); |
| 4172 | } | } |
| 4173 | break; | break; |
| 4174 | ||
| 4175 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 4176 | if (eptr > md->end_subject - min) MRRETURN(MATCH_NOMATCH); | if (eptr > md->end_subject - min) RRETURN(MATCH_NOMATCH); |
| 4177 | eptr += min; | eptr += min; |
| 4178 | break; | break; |
| 4179 | ||
| # | Line 4100 for (;;) | Line 4183 for (;;) |
| 4183 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4184 | { | { |
| 4185 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4186 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4187 | } | } |
| 4188 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4189 | switch(c) | switch(c) |
| 4190 | { | { |
| 4191 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4192 | ||
| 4193 | case 0x000d: | case 0x000d: |
| 4194 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| # | Line 4119 for (;;) | Line 4202 for (;;) |
| 4202 | case 0x0085: | case 0x0085: |
| 4203 | case 0x2028: | case 0x2028: |
| 4204 | case 0x2029: | case 0x2029: |
| 4205 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); |
| 4206 | break; | break; |
| 4207 | } | } |
| 4208 | } | } |
| # | Line 4131 for (;;) | Line 4214 for (;;) |
| 4214 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4215 | { | { |
| 4216 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4217 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4218 | } | } |
| 4219 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4220 | switch(c) | switch(c) |
| # | Line 4156 for (;;) | Line 4239 for (;;) |
| 4239 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 4240 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 4241 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 4242 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4243 | } | } |
| 4244 | } | } |
| 4245 | break; | break; |
| # | Line 4167 for (;;) | Line 4250 for (;;) |
| 4250 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4251 | { | { |
| 4252 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4253 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4254 | } | } |
| 4255 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4256 | switch(c) | switch(c) |
| 4257 | { | { |
| 4258 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4259 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4260 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4261 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 4203 for (;;) | Line 4286 for (;;) |
| 4286 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4287 | { | { |
| 4288 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4289 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4290 | } | } |
| 4291 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4292 | switch(c) | switch(c) |
| # | Line 4216 for (;;) | Line 4299 for (;;) |
| 4299 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4300 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 4301 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 4302 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4303 | } | } |
| 4304 | } | } |
| 4305 | break; | break; |
| # | Line 4227 for (;;) | Line 4310 for (;;) |
| 4310 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4311 | { | { |
| 4312 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4313 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4314 | } | } |
| 4315 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4316 | switch(c) | switch(c) |
| 4317 | { | { |
| 4318 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4319 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4320 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4321 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 4251 for (;;) | Line 4334 for (;;) |
| 4334 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4335 | { | { |
| 4336 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4337 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4338 | } | } |
| 4339 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4340 | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) |
| 4341 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4342 | } | } |
| 4343 | break; | break; |
| 4344 | ||
| # | Line 4265 for (;;) | Line 4348 for (;;) |
| 4348 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4349 | { | { |
| 4350 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4351 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4352 | } | } |
| 4353 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_digit) == 0) |
| 4354 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4355 | eptr++; | |
| 4356 | /* 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 */ |
| 4357 | } | } |
| 4358 | break; | break; |
| # | Line 4279 for (;;) | Line 4363 for (;;) |
| 4363 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4364 | { | { |
| 4365 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4366 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4367 | } | } |
| 4368 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) |
| 4369 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4370 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | eptr++; |
| 4371 | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); | |
| 4372 | } | } |
| 4373 | break; | break; |
| 4374 | ||
| # | Line 4293 for (;;) | Line 4378 for (;;) |
| 4378 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4379 | { | { |
| 4380 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4381 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4382 | } | } |
| 4383 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_space) == 0) |
| 4384 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4385 | eptr++; | |
| 4386 | /* 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 */ |
| 4387 | } | } |
| 4388 | break; | break; |
| # | Line 4307 for (;;) | Line 4393 for (;;) |
| 4393 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4394 | { | { |
| 4395 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4396 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4397 | } | } |
| 4398 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) |
| 4399 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4400 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | eptr++; |
| 4401 | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); | |
| 4402 | } | } |
| 4403 | break; | break; |
| 4404 | ||
| # | Line 4321 for (;;) | Line 4408 for (;;) |
| 4408 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4409 | { | { |
| 4410 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4411 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4412 | } | } |
| 4413 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_word) == 0) |
| 4414 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4415 | eptr++; | |
| 4416 | /* 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 */ |
| 4417 | } | } |
| 4418 | break; | break; |
| # | Line 4334 for (;;) | Line 4422 for (;;) |
| 4422 | } /* End switch(ctype) */ | } /* End switch(ctype) */ |
| 4423 | ||
| 4424 | else | else |
| 4425 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF */ |
| 4426 | ||
| 4427 | /* 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 |
| 4428 | than OP_PROP and OP_NOTPROP. */ | than OP_PROP and OP_NOTPROP. */ |
| # | Line 4347 for (;;) | Line 4435 for (;;) |
| 4435 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4436 | { | { |
| 4437 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4438 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4439 | } | } |
| 4440 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); |
| 4441 | eptr++; | eptr++; |
| 4442 | } | } |
| 4443 | break; | break; |
| # | Line 4358 for (;;) | Line 4446 for (;;) |
| 4446 | if (eptr > md->end_subject - min) | if (eptr > md->end_subject - min) |
| 4447 | { | { |
| 4448 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4449 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4450 | } | } |
| 4451 | eptr += min; | eptr += min; |
| 4452 | break; | break; |
| # | Line 4367 for (;;) | Line 4455 for (;;) |
| 4455 | if (eptr > md->end_subject - min) | if (eptr > md->end_subject - min) |
| 4456 | { | { |
| 4457 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4458 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4459 | } | } |
| 4460 | eptr += min; | eptr += min; |
| 4461 | break; | break; |
| # | Line 4378 for (;;) | Line 4466 for (;;) |
| 4466 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4467 | { | { |
| 4468 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4469 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4470 | } | } |
| 4471 | switch(*eptr++) | switch(*eptr++) |
| 4472 | { | { |
| 4473 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4474 | ||
| 4475 | case 0x000d: | case 0x000d: |
| 4476 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| # | Line 4394 for (;;) | Line 4482 for (;;) |
| 4482 | case 0x000b: | case 0x000b: |
| 4483 | case 0x000c: | case 0x000c: |
| 4484 | case 0x0085: | case 0x0085: |
| 4485 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
| 4486 | case 0x2028: | |
| 4487 | case 0x2029: | |
| 4488 | #endif | |
| 4489 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | |
| 4490 | break; | break; |
| 4491 | } | } |
| 4492 | } | } |
| # | Line 4406 for (;;) | Line 4498 for (;;) |
| 4498 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4499 | { | { |
| 4500 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4501 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4502 | } | } |
| 4503 | switch(*eptr++) | switch(*eptr++) |
| 4504 | { | { |
| # | Line 4414 for (;;) | Line 4506 for (;;) |
| 4506 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4507 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4508 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4509 | MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
| 4510 | case 0x1680: /* OGHAM SPACE MARK */ | |
| 4511 | case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ | |
| 4512 | case 0x2000: /* EN QUAD */ | |
| 4513 | case 0x2001: /* EM QUAD */ | |
| 4514 | case 0x2002: /* EN SPACE */ | |
| 4515 | case 0x2003: /* EM SPACE */ | |
| 4516 | case 0x2004: /* THREE-PER-EM SPACE */ | |
| 4517 | case 0x2005: /* FOUR-PER-EM SPACE */ | |
| 4518 | case 0x2006: /* SIX-PER-EM SPACE */ | |
| 4519 | case 0x2007: /* FIGURE SPACE */ | |
| 4520 | case 0x2008: /* PUNCTUATION SPACE */ | |
| 4521 | case 0x2009: /* THIN SPACE */ | |
| 4522 | case 0x200A: /* HAIR SPACE */ | |
| 4523 | case 0x202f: /* NARROW NO-BREAK SPACE */ | |
| 4524 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | |
| 4525 | case 0x3000: /* IDEOGRAPHIC SPACE */ | |
| 4526 | #endif | |
| 4527 | RRETURN(MATCH_NOMATCH); | |
| 4528 | } | } |
| 4529 | } | } |
| 4530 | break; | break; |
| # | Line 4425 for (;;) | Line 4535 for (;;) |
| 4535 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4536 | { | { |
| 4537 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4538 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4539 | } | } |
| 4540 | switch(*eptr++) | switch(*eptr++) |
| 4541 | { | { |
| 4542 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4543 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4544 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4545 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4546 | #ifdef COMPILE_PCRE16 | |
| 4547 | case 0x1680: /* OGHAM SPACE MARK */ | |
| 4548 | case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ | |
| 4549 | case 0x2000: /* EN QUAD */ | |
| 4550 | case 0x2001: /* EM QUAD */ | |
| 4551 | case 0x2002: /* EN SPACE */ | |
| 4552 | case 0x2003: /* EM SPACE */ | |
| 4553 | case 0x2004: /* THREE-PER-EM SPACE */ | |
| 4554 | case 0x2005: /* FOUR-PER-EM SPACE */ | |
| 4555 | case 0x2006: /* SIX-PER-EM SPACE */ | |
| 4556 | case 0x2007: /* FIGURE SPACE */ | |
| 4557 | case 0x2008: /* PUNCTUATION SPACE */ | |
| 4558 | case 0x2009: /* THIN SPACE */ | |
| 4559 | case 0x200A: /* HAIR SPACE */ | |
| 4560 | case 0x202f: /* NARROW NO-BREAK SPACE */ | |
| 4561 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | |
| 4562 | case 0x3000: /* IDEOGRAPHIC SPACE */ | |
| 4563 | #endif | |
| 4564 | break; | break; |
| 4565 | } | } |
| 4566 | } | } |
| # | Line 4444 for (;;) | Line 4572 for (;;) |
| 4572 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4573 | { | { |
| 4574 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4575 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4576 | } | } |
| 4577 | switch(*eptr++) | switch(*eptr++) |
| 4578 | { | { |
| # | Line 4454 for (;;) | Line 4582 for (;;) |
| 4582 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4583 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4584 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4585 | MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
| 4586 | case 0x2028: /* LINE SEPARATOR */ | |
| 4587 | case 0x2029: /* PARAGRAPH SEPARATOR */ | |
| 4588 | #endif | |
| 4589 | RRETURN(MATCH_NOMATCH); | |
| 4590 | } | } |
| 4591 | } | } |
| 4592 | break; | break; |
| # | Line 4465 for (;;) | Line 4597 for (;;) |
| 4597 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4598 | { | { |
| 4599 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4600 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4601 | } | } |
| 4602 | switch(*eptr++) | switch(*eptr++) |
| 4603 | { | { |
| 4604 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4605 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4606 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4607 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4608 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4609 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4610 | #ifdef COMPILE_PCRE16 | |
| 4611 | case 0x2028: /* LINE SEPARATOR */ | |
| 4612 | case 0x2029: /* PARAGRAPH SEPARATOR */ | |
| 4613 | #endif | |
| 4614 | break; | break; |
| 4615 | } | } |
| 4616 | } | } |
| # | Line 4486 for (;;) | Line 4622 for (;;) |
| 4622 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4623 | { | { |
| 4624 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4625 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4626 | } | } |
| 4627 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_digit) != 0) |
| 4628 | RRETURN(MATCH_NOMATCH); | |
| 4629 | eptr++; | |
| 4630 | } | } |
| 4631 | break; | break; |
| 4632 | ||
| # | Line 4498 for (;;) | Line 4636 for (;;) |
| 4636 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4637 | { | { |
| 4638 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4639 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4640 | } | } |
| 4641 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_digit) == 0) |
| 4642 | RRETURN(MATCH_NOMATCH); | |
| 4643 | eptr++; | |
| 4644 | } | } |
| 4645 | break; | break; |
| 4646 | ||
| # | Line 4510 for (;;) | Line 4650 for (;;) |
| 4650 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4651 | { | { |
| 4652 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4653 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4654 | } | } |
| 4655 | if ((md->ctypes[*eptr++] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_space) != 0) |
| 4656 | RRETURN(MATCH_NOMATCH); | |
| 4657 | eptr++; | |
| 4658 | } | } |
| 4659 | break; | break; |
| 4660 | ||
| # | Line 4522 for (;;) | Line 4664 for (;;) |
| 4664 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4665 | { | { |
| 4666 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4667 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4668 | } | } |
| 4669 | if ((md->ctypes[*eptr++] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_space) == 0) |
| 4670 | RRETURN(MATCH_NOMATCH); | |
| 4671 | eptr++; | |
| 4672 | } | } |
| 4673 | break; | break; |
| 4674 | ||
| # | Line 4534 for (;;) | Line 4678 for (;;) |
| 4678 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4679 | { | { |
| 4680 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4681 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4682 | } | } |
| 4683 | if ((md->ctypes[*eptr++] & ctype_word) != 0) | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_word) != 0) |
| 4684 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4685 | eptr++; | |
| 4686 | } | } |
| 4687 | break; | break; |
| 4688 | ||
| # | Line 4547 for (;;) | Line 4692 for (;;) |
| 4692 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4693 | { | { |
| 4694 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4695 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4696 | } | } |
| 4697 | if ((md->ctypes[*eptr++] & ctype_word) == 0) | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_word) == 0) |
| 4698 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4699 | eptr++; | |
| 4700 | } | } |
| 4701 | break; | break; |
| 4702 | ||
| # | Line 4579 for (;;) | Line 4725 for (;;) |
| 4725 | { | { |
| 4726 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM36); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM36); |
| 4727 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4728 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4729 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4730 | { | { |
| 4731 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4732 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4733 | } | } |
| 4734 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4735 | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); | if (prop_fail_result) RRETURN(MATCH_NOMATCH); |
| 4736 | } | } |
| 4737 | /* Control never gets here */ | /* Control never gets here */ |
| 4738 | ||
| # | Line 4596 for (;;) | Line 4742 for (;;) |
| 4742 | int chartype; | int chartype; |
| 4743 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM37); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM37); |
| 4744 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4745 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4746 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4747 | { | { |
| 4748 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4749 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4750 | } | } |
| 4751 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4752 | chartype = UCD_CHARTYPE(c); | chartype = UCD_CHARTYPE(c); |
| 4753 | if ((chartype == ucp_Lu || | if ((chartype == ucp_Lu || |
| 4754 | chartype == ucp_Ll || | chartype == ucp_Ll || |
| 4755 | chartype == ucp_Lt) == prop_fail_result) | chartype == ucp_Lt) == prop_fail_result) |
| 4756 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4757 | } | } |
| 4758 | /* Control never gets here */ | /* Control never gets here */ |
| 4759 | ||
| # | Line 4616 for (;;) | Line 4762 for (;;) |
| 4762 | { | { |
| 4763 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM38); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM38); |
| 4764 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4765 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4766 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4767 | { | { |
| 4768 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4769 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4770 | } | } |
| 4771 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4772 | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) |
| 4773 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4774 | } | } |
| 4775 | /* Control never gets here */ | /* Control never gets here */ |
| 4776 | ||
| # | Line 4633 for (;;) | Line 4779 for (;;) |
| 4779 | { | { |
| 4780 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM39); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM39); |
| 4781 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4782 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4783 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4784 | { | { |
| 4785 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4786 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4787 | } | } |
| 4788 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4789 | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) |
| 4790 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4791 | } | } |
| 4792 | /* Control never gets here */ | /* Control never gets here */ |
| 4793 | ||
| # | Line 4650 for (;;) | Line 4796 for (;;) |
| 4796 | { | { |
| 4797 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM40); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM40); |
| 4798 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4799 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4800 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4801 | { | { |
| 4802 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4803 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4804 | } | } |
| 4805 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4806 | if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) | if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) |
| 4807 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4808 | } | } |
| 4809 | /* Control never gets here */ | /* Control never gets here */ |
| 4810 | ||
| # | Line 4668 for (;;) | Line 4814 for (;;) |
| 4814 | int category; | int category; |
| 4815 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM59); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM59); |
| 4816 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4817 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4818 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4819 | { | { |
| 4820 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4821 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4822 | } | } |
| 4823 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4824 | category = UCD_CATEGORY(c); | category = UCD_CATEGORY(c); |
| 4825 | if ((category == ucp_L || category == ucp_N) == prop_fail_result) | if ((category == ucp_L || category == ucp_N) == prop_fail_result) |
| 4826 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4827 | } | } |
| 4828 | /* Control never gets here */ | /* Control never gets here */ |
| 4829 | ||
| # | Line 4686 for (;;) | Line 4832 for (;;) |
| 4832 | { | { |
| 4833 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM60); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM60); |
| 4834 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4835 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4836 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4837 | { | { |
| 4838 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4839 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4840 | } | } |
| 4841 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4842 | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
| 4843 | c == CHAR_FF || c == CHAR_CR) | c == CHAR_FF || c == CHAR_CR) |
| 4844 | == prop_fail_result) | == prop_fail_result) |
| 4845 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4846 | } | } |
| 4847 | /* Control never gets here */ | /* Control never gets here */ |
| 4848 | ||
| # | Line 4705 for (;;) | Line 4851 for (;;) |
| 4851 | { | { |
| 4852 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM61); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM61); |
| 4853 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4854 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4855 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4856 | { | { |
| 4857 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4858 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4859 | } | } |
| 4860 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4861 | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
| 4862 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
| 4863 | == prop_fail_result) | == prop_fail_result) |
| 4864 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4865 | } | } |
| 4866 | /* Control never gets here */ | /* Control never gets here */ |
| 4867 | ||
| # | Line 4725 for (;;) | Line 4871 for (;;) |
| 4871 | int category; | int category; |
| 4872 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM62); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM62); |
| 4873 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4874 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4875 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4876 | { | { |
| 4877 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4878 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4879 | } | } |
| 4880 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4881 | category = UCD_CATEGORY(c); | category = UCD_CATEGORY(c); |
| # | Line 4737 for (;;) | Line 4883 for (;;) |
| 4883 | category == ucp_N || | category == ucp_N || |
| 4884 | c == CHAR_UNDERSCORE) | c == CHAR_UNDERSCORE) |
| 4885 | == prop_fail_result) | == prop_fail_result) |
| 4886 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4887 | } | } |
| 4888 | /* Control never gets here */ | /* Control never gets here */ |
| 4889 | ||
| # | Line 4757 for (;;) | Line 4903 for (;;) |
| 4903 | { | { |
| 4904 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM41); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM41); |
| 4905 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4906 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4907 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4908 | { | { |
| 4909 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4910 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4911 | } | } |
| 4912 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4913 | if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); | if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); |
| 4914 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 4915 | { | { |
| 4916 | int len = 1; | int len = 1; |
| 4917 | if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } | if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
| 4918 | if (UCD_CATEGORY(c) != ucp_M) break; | if (UCD_CATEGORY(c) != ucp_M) break; |
| 4919 | eptr += len; | eptr += len; |
| 4920 | } | } |
| # | Line 4777 for (;;) | Line 4923 for (;;) |
| 4923 | else | else |
| 4924 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 4925 | ||
| 4926 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 4927 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 4928 | { | { |
| 4929 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 4930 | { | { |
| 4931 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM42); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM42); |
| 4932 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4933 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4934 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4935 | { | { |
| 4936 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4937 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4938 | } | } |
| 4939 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | if (ctype == OP_ANY && IS_NEWLINE(eptr)) |
| 4940 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4941 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4942 | switch(ctype) | switch(ctype) |
| 4943 | { | { |
| # | Line 4804 for (;;) | Line 4949 for (;;) |
| 4949 | case OP_ANYNL: | case OP_ANYNL: |
| 4950 | switch(c) | switch(c) |
| 4951 | { | { |
| 4952 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4953 | case 0x000d: | case 0x000d: |
| 4954 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4955 | break; | break; |
| # | Line 4816 for (;;) | Line 4961 for (;;) |
| 4961 | case 0x0085: | case 0x0085: |
| 4962 | case 0x2028: | case 0x2028: |
| 4963 | case 0x2029: | case 0x2029: |
| 4964 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); |
| 4965 | break; | break; |
| 4966 | } | } |
| 4967 | break; | break; |
| # | Line 4844 for (;;) | Line 4989 for (;;) |
| 4989 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 4990 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 4991 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 4992 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4993 | } | } |
| 4994 | break; | break; |
| 4995 | ||
| 4996 | case OP_HSPACE: | case OP_HSPACE: |
| 4997 | switch(c) | switch(c) |
| 4998 | { | { |
| 4999 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 5000 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 5001 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 5002 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 4886 for (;;) | Line 5031 for (;;) |
| 5031 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 5032 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 5033 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 5034 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5035 | } | } |
| 5036 | break; | break; |
| 5037 | ||
| 5038 | case OP_VSPACE: | case OP_VSPACE: |
| 5039 | switch(c) | switch(c) |
| 5040 | { | { |
| 5041 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 5042 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 5043 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 5044 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 4907 for (;;) | Line 5052 for (;;) |
| 5052 | ||
| 5053 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 5054 | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) |
| 5055 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5056 | break; | break; |
| 5057 | ||
| 5058 | case OP_DIGIT: | case OP_DIGIT: |
| 5059 | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) |
| 5060 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5061 | break; | break; |
| 5062 | ||
| 5063 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 5064 | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) |
| 5065 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5066 | break; | break; |
| 5067 | ||
| 5068 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 5069 | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) |
| 5070 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5071 | break; | break; |
| 5072 | ||
| 5073 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 5074 | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) |
| 5075 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5076 | break; | break; |
| 5077 | ||
| 5078 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 5079 | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) |
| 5080 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5081 | break; | break; |
| 5082 | ||
| 5083 | default: | default: |
| # | Line 4942 for (;;) | Line 5087 for (;;) |
| 5087 | } | } |
| 5088 | else | else |
| 5089 | #endif | #endif |
| 5090 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 5091 | { | { |
| 5092 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 5093 | { | { |
| 5094 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM43); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM43); |
| 5095 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 5096 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 5097 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 5098 | { | { |
| 5099 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 5100 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5101 | } | } |
| 5102 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | if (ctype == OP_ANY && IS_NEWLINE(eptr)) |
| 5103 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5104 | c = *eptr++; | c = *eptr++; |
| 5105 | switch(ctype) | switch(ctype) |
| 5106 | { | { |
| # | Line 4967 for (;;) | Line 5112 for (;;) |
| 5112 | case OP_ANYNL: | case OP_ANYNL: |
| 5113 | switch(c) | switch(c) |
| 5114 | { | { |
| 5115 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 5116 | case 0x000d: | case 0x000d: |
| 5117 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 5118 | break; | break; |
| # | Line 4978 for (;;) | Line 5123 for (;;) |
| 5123 | case 0x000b: | case 0x000b: |
| 5124 | case 0x000c: | case 0x000c: |
| 5125 | case 0x0085: | case 0x0085: |
| 5126 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
| 5127 | case 0x2028: | |
| 5128 | case 0x2029: | |
| 5129 | #endif | |
| 5130 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | |
| 5131 | break; | break; |
| 5132 | } | } |
| 5133 | break; | break; |
| # | Line 4990 for (;;) | Line 5139 for (;;) |
| 5139 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 5140 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 5141 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 5142 | MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
| 5143 | case 0x1680: /* OGHAM SPACE MARK */ | |
| 5144 | case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ | |
| 5145 | case 0x2000: /* EN QUAD */ | |
| 5146 | case 0x2001: /* EM QUAD */ | |
| 5147 | case 0x2002: /* EN SPACE */ | |
| 5148 | case 0x2003: /* EM SPACE */ | |
| 5149 | case 0x2004: /* THREE-PER-EM SPACE */ | |
| 5150 | case 0x2005: /* FOUR-PER-EM SPACE */ | |
| 5151 | case 0x2006: /* SIX-PER-EM SPACE */ | |
| 5152 | case 0x2007: /* FIGURE SPACE */ | |
| 5153 | case 0x2008: /* PUNCTUATION SPACE */ | |
| 5154 | case 0x2009: /* THIN SPACE */ | |
| 5155 | case 0x200A: /* HAIR SPACE */ | |
| 5156 | case 0x202f: /* NARROW NO-BREAK SPACE */ | |
| 5157 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | |
| 5158 | case 0x3000: /* IDEOGRAPHIC SPACE */ | |
| 5159 | #endif | |
| 5160 | RRETURN(MATCH_NOMATCH); | |
| 5161 | } | } |
| 5162 | break; | break; |
| 5163 | ||
| 5164 | case OP_HSPACE: | case OP_HSPACE: |
| 5165 | switch(c) | switch(c) |
| 5166 | { | { |
| 5167 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 5168 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 5169 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 5170 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 5171 | #ifdef COMPILE_PCRE16 | |
| 5172 | case 0x1680: /* OGHAM SPACE MARK */ | |
| 5173 | case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ | |
| 5174 | case 0x2000: /* EN QUAD */ | |
| 5175 | case 0x2001: /* EM QUAD */ | |
| 5176 | case 0x2002: /* EN SPACE */ | |
| 5177 | case 0x2003: /* EM SPACE */ | |
| 5178 | case 0x2004: /* THREE-PER-EM SPACE */ | |
| 5179 | case 0x2005: /* FOUR-PER-EM SPACE */ | |
| 5180 | case 0x2006: /* SIX-PER-EM SPACE */ | |
| 5181 | case 0x2007: /* FIGURE SPACE */ | |
| 5182 | case 0x2008: /* PUNCTUATION SPACE */ | |
| 5183 | case 0x2009: /* THIN SPACE */ | |
| 5184 | case 0x200A: /* HAIR SPACE */ | |
| 5185 | case 0x202f: /* NARROW NO-BREAK SPACE */ | |
| 5186 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | |
| 5187 | case 0x3000: /* IDEOGRAPHIC SPACE */ | |
| 5188 | #endif | |
| 5189 | break; | break; |
| 5190 | } | } |
| 5191 | break; | break; |
| # | Line 5014 for (;;) | Line 5199 for (;;) |
| 5199 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 5200 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 5201 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 5202 | MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
| 5203 | case 0x2028: /* LINE SEPARATOR */ | |
| 5204 | case 0x2029: /* PARAGRAPH SEPARATOR */ | |
| 5205 | #endif | |
| 5206 | RRETURN(MATCH_NOMATCH); | |
| 5207 | } | } |
| 5208 | break; | break; |
| 5209 | ||
| 5210 | case OP_VSPACE: | case OP_VSPACE: |
| 5211 | switch(c) | switch(c) |
| 5212 | { | { |
| 5213 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 5214 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 5215 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 5216 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 5217 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 5218 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 5219 | #ifdef COMPILE_PCRE16 | |
| 5220 | case 0x2028: /* LINE SEPARATOR */ | |
| 5221 | case 0x2029: /* PARAGRAPH SEPARATOR */ | |
| 5222 | #endif | |
| 5223 | break; | break; |
| 5224 | } | } |
| 5225 | break; | break; |
| 5226 | ||
| 5227 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 5228 | if ((md->ctypes[c] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); | if (MAX_255(c) && (md->ctypes[c] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); |
| 5229 | break; | break; |
| 5230 | ||
| 5231 | case OP_DIGIT: | case OP_DIGIT: |
| 5232 | if ((md->ctypes[c] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); | if (!MAX_255(c) || (md->ctypes[c] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); |
| 5233 | break; | break; |
| 5234 | ||
| 5235 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 5236 | if ((md->ctypes[c] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); | if (MAX_255(c) && (md->ctypes[c] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); |
| 5237 | break; | break; |
| 5238 | ||
| 5239 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 5240 | if ((md->ctypes[c] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); | if (!MAX_255(c) || (md->ctypes[c] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); |
| 5241 | break; | break; |
| 5242 | ||
| 5243 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 5244 | if ((md->ctypes[c] & ctype_word) != 0) MRRETURN(MATCH_NOMATCH); | if (MAX_255(c) && (md->ctypes[c] & ctype_word) != 0) RRETURN(MATCH_NOMATCH); |
| 5245 | break; | break; |
| 5246 | ||
| 5247 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 5248 | if ((md->ctypes[c] & ctype_word) == 0) MRRETURN(MATCH_NOMATCH); | if (!MAX_255(c) || (md->ctypes[c] & ctype_word) == 0) RRETURN(MATCH_NOMATCH); |
| 5249 | break; | break; |
| 5250 | ||
| 5251 | default: | default: |
| # | Line 5241 for (;;) | Line 5434 for (;;) |
| 5434 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM44); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM44); |
| 5435 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 5436 | if (eptr-- == pp) break; /* Stop if tried at original pos */ | if (eptr-- == pp) break; /* Stop if tried at original pos */ |
| 5437 | if (utf8) BACKCHAR(eptr); | if (utf) BACKCHAR(eptr); |
| 5438 | } | } |
| 5439 | } | } |
| 5440 | ||
| # | Line 5258 for (;;) | Line 5451 for (;;) |
| 5451 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 5452 | break; | break; |
| 5453 | } | } |
| 5454 | if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } | if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
| 5455 | if (UCD_CATEGORY(c) == ucp_M) break; | if (UCD_CATEGORY(c) == ucp_M) break; |
| 5456 | eptr += len; | eptr += len; |
| 5457 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 5458 | { | { |
| 5459 | len = 1; | len = 1; |
| 5460 | if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } | if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
| 5461 | if (UCD_CATEGORY(c) != ucp_M) break; | if (UCD_CATEGORY(c) != ucp_M) break; |
| 5462 | eptr += len; | eptr += len; |
| 5463 | } | } |
| # | Line 5281 for (;;) | Line 5474 for (;;) |
| 5474 | if (eptr-- == pp) break; /* Stop if tried at original pos */ | if (eptr-- == pp) break; /* Stop if tried at original pos */ |
| 5475 | for (;;) /* Move back over one extended */ | for (;;) /* Move back over one extended */ |
| 5476 | { | { |
| 5477 | if (!utf8) c = *eptr; else | if (!utf) c = *eptr; else |
| 5478 | { | { |
| 5479 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 5480 | GETCHAR(c, eptr); | GETCHAR(c, eptr); |
| # | Line 5295 for (;;) | Line 5488 for (;;) |
| 5488 | else | else |
| 5489 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 5490 | ||
| 5491 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 5492 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 5493 | { | { |
| 5494 | switch(ctype) | switch(ctype) |
| 5495 | { | { |
| # | Line 5314 for (;;) | Line 5505 for (;;) |
| 5505 | } | } |
| 5506 | if (IS_NEWLINE(eptr)) break; | if (IS_NEWLINE(eptr)) break; |
| 5507 | eptr++; | eptr++; |
| 5508 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); |
| 5509 | } | } |
| 5510 | } | } |
| 5511 | ||
| # | Line 5331 for (;;) | Line 5522 for (;;) |
| 5522 | } | } |
| 5523 | if (IS_NEWLINE(eptr)) break; | if (IS_NEWLINE(eptr)) break; |
| 5524 | eptr++; | eptr++; |
| 5525 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); |
| 5526 | } | } |
| 5527 | } | } |
| 5528 | break; | break; |
| # | Line 5347 for (;;) | Line 5538 for (;;) |
| 5538 | break; | break; |
| 5539 | } | } |
| 5540 | eptr++; | eptr++; |
| 5541 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); |
| 5542 | } | } |
| 5543 | } | } |
| 5544 | else | else |
| # | Line 5580 for (;;) | Line 5771 for (;;) |
| 5771 | } | } |
| 5772 | } | } |
| 5773 | else | else |
| 5774 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF */ |
| 5775 | /* Not UTF mode */ | |
| /* Not UTF-8 mode */ | ||
| 5776 | { | { |
| 5777 | switch(ctype) | switch(ctype) |
| 5778 | { | { |
| # | Line 5626 for (;;) | Line 5816 for (;;) |
| 5816 | } | } |
| 5817 | else | else |
| 5818 | { | { |
| 5819 | if (c != 0x000a && | if (c != 0x000a && (md->bsr_anycrlf || |
| 5820 | (md->bsr_anycrlf || | (c != 0x000b && c != 0x000c && c != 0x0085 |
| 5821 | (c != 0x000b && c != 0x000c && c != 0x0085))) | #ifdef COMPILE_PCRE16 |
| 5822 | break; | && c != 0x2028 && c != 0x2029 |
| 5823 | #endif | |
| 5824 | ))) break; | |
| 5825 | eptr++; | eptr++; |
| 5826 | } | } |
| 5827 | } | } |
| # | Line 5644 for (;;) | Line 5836 for (;;) |
| 5836 | break; | break; |
| 5837 | } | } |
| 5838 | c = *eptr; | c = *eptr; |
| 5839 | if (c == 0x09 || c == 0x20 || c == 0xa0) break; | if (c == 0x09 || c == 0x20 || c == 0xa0 |
| 5840 | #ifdef COMPILE_PCRE16 | |
| 5841 | || c == 0x1680 || c == 0x180e || (c >= 0x2000 && c <= 0x200A) | |
| 5842 | || c == 0x202f || c == 0x205f || c == 0x3000 | |
| 5843 | #endif | |
| 5844 | ) break; | |
| 5845 | eptr++; | eptr++; |
| 5846 | } | } |
| 5847 | break; | break; |
| # | Line 5658 for (;;) | Line 5855 for (;;) |
| 5855 | break; | break; |
| 5856 | } | } |
| 5857 | c = *eptr; | c = *eptr; |
| 5858 | if (c != 0x09 && c != 0x20 && c != 0xa0) break; | if (c != 0x09 && c != 0x20 && c != 0xa0 |
| 5859 | #ifdef COMPILE_PCRE16 | |
| 5860 | && c != 0x1680 && c != 0x180e && (c < 0x2000 || c > 0x200A) | |
| 5861 | && c != 0x202f && c != 0x205f && c != 0x3000 | |
| 5862 | #endif | |
| 5863 | ) break; | |
| 5864 | eptr++; | eptr++; |
| 5865 | } | } |
| 5866 | break; | break; |
| # | Line 5672 for (;;) | Line 5874 for (;;) |
| 5874 | break; | break; |
| 5875 | } | } |
| 5876 | c = *eptr; | c = *eptr; |
| 5877 | if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85) | if (c == 0x0a || c == 0x0b || c == 0x0c || c == 0x0d || c == 0x85 |
| 5878 | break; | #ifdef COMPILE_PCRE16 |
| 5879 | || c == 0x2028 || c == 0x2029 | |
| 5880 | #endif | |
| 5881 | ) break; | |
| 5882 | eptr++; | eptr++; |
| 5883 | } | } |
| 5884 | break; | break; |
| # | Line 5687 for (;;) | Line 5892 for (;;) |
| 5892 | break; | break; |
| 5893 | } | } |
| 5894 | c = *eptr; | c = *eptr; |
| 5895 | if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85) | if (c != 0x0a && c != 0x0b && c != 0x0c && c != 0x0d && c != 0x85 |
| 5896 | break; | #ifdef COMPILE_PCRE16 |
| 5897 | && c != 0x2028 && c != 0x2029 | |
| 5898 | #endif | |
| 5899 | ) break; | |
| 5900 | eptr++; | eptr++; |
| 5901 | } | } |
| 5902 | break; | break; |
| # | Line 5701 for (;;) | Line 5909 for (;;) |
| 5909 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 5910 | break; | break; |
| 5911 | } | } |
| 5912 | if ((md->ctypes[*eptr] & ctype_digit) != 0) break; | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_digit) != 0) break; |
| 5913 | eptr++; | eptr++; |
| 5914 | } | } |
| 5915 | break; | break; |
| # | Line 5714 for (;;) | Line 5922 for (;;) |
| 5922 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 5923 | break; | break; |
| 5924 | } | } |
| 5925 | if ((md->ctypes[*eptr] & ctype_digit) == 0) break; | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_digit) == 0) break; |
| 5926 | eptr++; | eptr++; |
| 5927 | } | } |
| 5928 | break; | break; |
| # | Line 5727 for (;;) | Line 5935 for (;;) |
| 5935 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 5936 | break; | break; |
| 5937 | } | } |
| 5938 | if ((md->ctypes[*eptr] & ctype_space) != 0) break; | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_space) != 0) break; |
| 5939 | eptr++; | eptr++; |
| 5940 | } | } |
| 5941 | break; | break; |
| # | Line 5740 for (;;) | Line 5948 for (;;) |
| 5948 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 5949 | break; | break; |
| 5950 | } | } |
| 5951 | if ((md->ctypes[*eptr] & ctype_space) == 0) break; | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_space) == 0) break; |
| 5952 | eptr++; | eptr++; |
| 5953 | } | } |
| 5954 | break; | break; |
| # | Line 5753 for (;;) | Line 5961 for (;;) |
| 5961 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 5962 | break; | break; |
| 5963 | } | } |
| 5964 | if ((md->ctypes[*eptr] & ctype_word) != 0) break; | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_word) != 0) break; |
| 5965 | eptr++; | eptr++; |
| 5966 | } | } |
| 5967 | break; | break; |
| # | Line 5766 for (;;) | Line 5974 for (;;) |
| 5974 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 5975 | break; | break; |
| 5976 | } | } |
| 5977 | if ((md->ctypes[*eptr] & ctype_word) == 0) break; | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_word) == 0) break; |
| 5978 | eptr++; | eptr++; |
| 5979 | } | } |
| 5980 | break; | break; |
| # | Line 5794 for (;;) | Line 6002 for (;;) |
| 6002 | ||
| 6003 | /* 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 */ |
| 6004 | ||
| 6005 |   |