Parent Directory
|
Revision Log
|
Patch
| revision 835 by ph10, Wed Dec 28 16:10:09 2011 UTC | revision 836 by ph10, Wed Dec 28 17:16:11 2011 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 816 for (;;) | Line 824 for (;;) |
| 824 | if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ | if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ |
| 825 | { | { |
| 826 | mstart = md->start_match_ptr; | mstart = md->start_match_ptr; |
| markptr = md->mark; | ||
| 827 | break; | break; |
| 828 | } | } |
| 829 | if (rrc == MATCH_THEN) | if (rrc == MATCH_THEN) |
| # | Line 916 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 954 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 1004 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 1028 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 1042 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 1071 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 1094 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 1130 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 1162 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 1217 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; | pcre_callout_block cb; |
| 1227 | cb.version = 2; /* Version 1 of the callout block */ | cb.version = 2; /* Version 1 of the callout block */ |
| # | Line 1232 for (;;) | Line 1236 for (;;) |
| 1236 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 1237 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 1238 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 1239 | cb.mark = markptr; | cb.mark = md->nomatch_mark; |
| 1240 | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); | if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH); |
| 1241 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 1242 | } | } |
| 1243 | ecode += _pcre_OP_lengths[OP_CALLOUT]; | ecode += PRIV(OP_lengths)[OP_CALLOUT]; |
| 1244 | } | } |
| 1245 | ||
| 1246 | condcode = ecode[LINK_SIZE+1]; | condcode = ecode[LINK_SIZE+1]; |
| # | Line 1262 for (;;) | Line 1266 for (;;) |
| 1266 | ||
| 1267 | if (!condition && condcode == OP_NRREF) | if (!condition && condcode == OP_NRREF) |
| 1268 | { | { |
| 1269 | uschar *slotA = md->name_table; | pcre_uchar *slotA = md->name_table; |
| 1270 | for (i = 0; i < md->name_count; i++) | for (i = 0; i < md->name_count; i++) |
| 1271 | { | { |
| 1272 | if (GET2(slotA, 0) == recno) break; | if (GET2(slotA, 0) == recno) break; |
| # | Line 1275 for (;;) | Line 1279 for (;;) |
| 1279 | ||
| 1280 | if (i < md->name_count) | if (i < md->name_count) |
| 1281 | { | { |
| 1282 | uschar *slotB = slotA; | pcre_uchar *slotB = slotA; |
| 1283 | while (slotB > md->name_table) | while (slotB > md->name_table) |
| 1284 | { | { |
| 1285 | slotB -= md->name_entry_size; | slotB -= md->name_entry_size; |
| 1286 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
| 1287 | { | { |
| 1288 | condition = GET2(slotB, 0) == md->recursive->group_num; | condition = GET2(slotB, 0) == md->recursive->group_num; |
| 1289 | if (condition) break; | if (condition) break; |
| # | Line 1295 for (;;) | Line 1299 for (;;) |
| 1299 | for (i++; i < md->name_count; i++) | for (i++; i < md->name_count; i++) |
| 1300 | { | { |
| 1301 | slotB += md->name_entry_size; | slotB += md->name_entry_size; |
| 1302 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
| 1303 | { | { |
| 1304 | condition = GET2(slotB, 0) == md->recursive->group_num; | condition = GET2(slotB, 0) == md->recursive->group_num; |
| 1305 | if (condition) break; | if (condition) break; |
| # | Line 1308 for (;;) | Line 1312 for (;;) |
| 1312 | ||
| 1313 | /* Chose branch according to the condition */ | /* Chose branch according to the condition */ |
| 1314 | ||
| 1315 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1); |
| 1316 | } | } |
| 1317 | } | } |
| 1318 | ||
| # | Line 1325 for (;;) | Line 1329 for (;;) |
| 1329 | if (!condition && condcode == OP_NCREF) | if (!condition && condcode == OP_NCREF) |
| 1330 | { | { |
| 1331 | int refno = offset >> 1; | int refno = offset >> 1; |
| 1332 | uschar *slotA = md->name_table; | pcre_uchar *slotA = md->name_table; |
| 1333 | ||
| 1334 | for (i = 0; i < md->name_count; i++) | for (i = 0; i < md->name_count; i++) |
| 1335 | { | { |
| # | Line 1339 for (;;) | Line 1343 for (;;) |
| 1343 | ||
| 1344 | if (i < md->name_count) | if (i < md->name_count) |
| 1345 | { | { |
| 1346 | uschar *slotB = slotA; | pcre_uchar *slotB = slotA; |
| 1347 | while (slotB > md->name_table) | while (slotB > md->name_table) |
| 1348 | { | { |
| 1349 | slotB -= md->name_entry_size; | slotB -= md->name_entry_size; |
| 1350 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
| 1351 | { | { |
| 1352 | offset = GET2(slotB, 0) << 1; | offset = GET2(slotB, 0) << 1; |
| 1353 | condition = offset < offset_top && | condition = offset < offset_top && |
| # | Line 1361 for (;;) | Line 1365 for (;;) |
| 1365 | for (i++; i < md->name_count; i++) | for (i++; i < md->name_count; i++) |
| 1366 | { | { |
| 1367 | slotB += md->name_entry_size; | slotB += md->name_entry_size; |
| 1368 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | if (STRCMP_UC_UC(slotA + IMM2_SIZE, slotB + IMM2_SIZE) == 0) |
| 1369 | { | { |
| 1370 | offset = GET2(slotB, 0) << 1; | offset = GET2(slotB, 0) << 1; |
| 1371 | condition = offset < offset_top && | condition = offset < offset_top && |
| # | Line 1376 for (;;) | Line 1380 for (;;) |
| 1380 | ||
| 1381 | /* Chose branch according to the condition */ | /* Chose branch according to the condition */ |
| 1382 | ||
| 1383 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 1 + IMM2_SIZE : GET(ecode, 1); |
| 1384 | } | } |
| 1385 | ||
| 1386 | else if (condcode == OP_DEF) /* DEFINE - always false */ | else if (condcode == OP_DEF) /* DEFINE - always false */ |
| # | Line 1468 for (;;) | Line 1472 for (;;) |
| 1472 | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); |
| 1473 | if (offset_top <= offset) offset_top = offset + 2; | if (offset_top <= offset) offset_top = offset + 2; |
| 1474 | } | } |
| 1475 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 1476 | break; | break; |
| 1477 | ||
| 1478 | ||
| # | Line 1488 for (;;) | Line 1492 for (;;) |
| 1492 | (md->notempty || | (md->notempty || |
| 1493 | (md->notempty_atstart && | (md->notempty_atstart && |
| 1494 | mstart == md->start_subject + md->start_offset))) | mstart == md->start_subject + md->start_offset))) |
| 1495 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 1496 | ||
| 1497 | /* Otherwise, we have a match. */ | /* Otherwise, we have a match. */ |
| 1498 | ||
| # | Line 1497 for (;;) | Line 1501 for (;;) |
| 1501 | md->start_match_ptr = mstart; /* and the start (\K can modify) */ | md->start_match_ptr = mstart; /* and the start (\K can modify) */ |
| 1502 | ||
| 1503 | /* 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 |
| 1504 | given as the argument to MRRETURN when the heap is in use. */ | given as the argument to RRETURN when the heap is in use. */ |
| 1505 | ||
| 1506 | rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; | rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; |
| 1507 | MRRETURN(rrc); | RRETURN(rrc); |
| 1508 | ||
| 1509 | /* Assertion brackets. Check the alternative branches in turn - the | /* Assertion brackets. Check the alternative branches in turn - the |
| 1510 | 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 1528 for (;;) | Line 1532 for (;;) |
| 1532 | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
| 1533 | { | { |
| 1534 | mstart = md->start_match_ptr; /* In case \K reset it */ | mstart = md->start_match_ptr; /* In case \K reset it */ |
| markptr = md->mark; | ||
| 1535 | break; | break; |
| 1536 | } | } |
| 1537 | ||
| # | Line 1540 for (;;) | Line 1543 for (;;) |
| 1543 | } | } |
| 1544 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| 1545 | ||
| 1546 | if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH); | if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); |
| 1547 | ||
| 1548 | /* If checking an assertion for a condition, return MATCH_MATCH. */ | /* If checking an assertion for a condition, return MATCH_MATCH. */ |
| 1549 | ||
| # | Line 1570 for (;;) | Line 1573 for (;;) |
| 1573 | do | do |
| 1574 | { | { |
| 1575 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5); | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, NULL, RM5); |
| 1576 | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH); | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) RRETURN(MATCH_NOMATCH); |
| 1577 | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) |
| 1578 | { | { |
| 1579 | do ecode += GET(ecode,1); while (*ecode == OP_ALT); | do ecode += GET(ecode,1); while (*ecode == OP_ALT); |
| # | Line 1596 for (;;) | Line 1599 for (;;) |
| 1599 | back a number of characters, not bytes. */ | back a number of characters, not bytes. */ |
| 1600 | ||
| 1601 | case OP_REVERSE: | case OP_REVERSE: |
| 1602 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 1603 | if (utf8) | if (utf) |
| 1604 | { | { |
| 1605 | i = GET(ecode, 1); | i = GET(ecode, 1); |
| 1606 | while (i-- > 0) | while (i-- > 0) |
| 1607 | { | { |
| 1608 | eptr--; | eptr--; |
| 1609 | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); |
| 1610 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 1611 | } | } |
| 1612 | } | } |
| # | Line 1614 for (;;) | Line 1617 for (;;) |
| 1617 | ||
| 1618 | { | { |
| 1619 | eptr -= GET(ecode, 1); | eptr -= GET(ecode, 1); |
| 1620 | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); |
| 1621 | } | } |
| 1622 | ||
| 1623 | /* Save the earliest consulted character, then skip to next op code */ | /* Save the earliest consulted character, then skip to next op code */ |
| # | Line 1628 for (;;) | Line 1631 for (;;) |
| 1631 | function is able to force a failure. */ | function is able to force a failure. */ |
| 1632 | ||
| 1633 | case OP_CALLOUT: | case OP_CALLOUT: |
| 1634 | if (pcre_callout != NULL) | if (PUBL(callout) != NULL) |
| 1635 | { | { |
| 1636 | pcre_callout_block cb; | pcre_callout_block cb; |
| 1637 | cb.version = 2; /* Version 1 of the callout block */ | cb.version = 2; /* Version 1 of the callout block */ |
| # | Line 1643 for (;;) | Line 1646 for (;;) |
| 1646 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 1647 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 1648 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 1649 | cb.mark = markptr; | cb.mark = md->nomatch_mark; |
| 1650 | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); | if ((rrc = (*PUBL(callout))(&cb)) > 0) RRETURN(MATCH_NOMATCH); |
| 1651 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 1652 | } | } |
| 1653 | ecode += 2 + 2*LINK_SIZE; | ecode += 2 + 2*LINK_SIZE; |
| # | Line 1703 for (;;) | Line 1706 for (;;) |
| 1706 | else | else |
| 1707 | { | { |
| 1708 | new_recursive.offset_save = | new_recursive.offset_save = |
| 1709 | (int *)(pcre_malloc)(new_recursive.saved_max * sizeof(int)); | (int *)(PUBL(malloc))(new_recursive.saved_max * sizeof(int)); |
| 1710 | if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); | if (new_recursive.offset_save == NULL) RRETURN(PCRE_ERROR_NOMEMORY); |
| 1711 | } | } |
| 1712 | memcpy(new_recursive.offset_save, md->offset_vector, | memcpy(new_recursive.offset_save, md->offset_vector, |
| # | Line 1718 for (;;) | Line 1721 for (;;) |
| 1721 | do | do |
| 1722 | { | { |
| 1723 | if (cbegroup) md->match_function_type = MATCH_CBEGROUP; | if (cbegroup) md->match_function_type = MATCH_CBEGROUP; |
| 1724 | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, | RMATCH(eptr, callpat + PRIV(OP_lengths)[*callpat], offset_top, |
| 1725 | md, eptrb, RM6); | md, eptrb, RM6); |
| 1726 | memcpy(md->offset_vector, new_recursive.offset_save, | memcpy(md->offset_vector, new_recursive.offset_save, |
| 1727 | new_recursive.saved_max * sizeof(int)); | new_recursive.saved_max * sizeof(int)); |
| # | Line 1727 for (;;) | Line 1730 for (;;) |
| 1730 | { | { |
| 1731 | DPRINTF(("Recursion matched\n")); | DPRINTF(("Recursion matched\n")); |
| 1732 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1733 | (pcre_free)(new_recursive.offset_save); | (PUBL(free))(new_recursive.offset_save); |
| 1734 | ||
| 1735 | /* 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 |
| 1736 | 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 1745 for (;;) | Line 1748 for (;;) |
| 1748 | { | { |
| 1749 | DPRINTF(("Recursion gave error %d\n", rrc)); | DPRINTF(("Recursion gave error %d\n", rrc)); |
| 1750 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1751 | (pcre_free)(new_recursive.offset_save); | (PUBL(free))(new_recursive.offset_save); |
| 1752 | RRETURN(rrc); | RRETURN(rrc); |
| 1753 | } | } |
| 1754 | ||
| # | Line 1757 for (;;) | Line 1760 for (;;) |
| 1760 | DPRINTF(("Recursion didn't match\n")); | DPRINTF(("Recursion didn't match\n")); |
| 1761 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1762 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1763 | (pcre_free)(new_recursive.offset_save); | (PUBL(free))(new_recursive.offset_save); |
| 1764 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 1765 | } | } |
| 1766 | ||
| 1767 | RECURSION_MATCHED: | RECURSION_MATCHED: |
| # | Line 1838 for (;;) | Line 1841 for (;;) |
| 1841 | md->end_match_ptr = eptr; /* For ONCE_NC */ | md->end_match_ptr = eptr; /* For ONCE_NC */ |
| 1842 | md->end_offset_top = offset_top; | md->end_offset_top = offset_top; |
| 1843 | md->start_match_ptr = mstart; | md->start_match_ptr = mstart; |
| 1844 | MRRETURN(MATCH_MATCH); /* Sets md->mark */ | RRETURN(MATCH_MATCH); /* Sets md->mark */ |
| 1845 | } | } |
| 1846 | ||
| 1847 | /* 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 1980 for (;;) | Line 1983 for (;;) |
| 1983 | /* Not multiline mode: start of subject assertion, unless notbol. */ | /* Not multiline mode: start of subject assertion, unless notbol. */ |
| 1984 | ||
| 1985 | case OP_CIRC: | case OP_CIRC: |
| 1986 | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); |
| 1987 | ||
| 1988 | /* Start of subject assertion */ | /* Start of subject assertion */ |
| 1989 | ||
| 1990 | case OP_SOD: | case OP_SOD: |
| 1991 | if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); |
| 1992 | ecode++; | ecode++; |
| 1993 | break; | break; |
| 1994 | ||
| 1995 | /* Multiline mode: start of subject unless notbol, or after any newline. */ | /* Multiline mode: start of subject unless notbol, or after any newline. */ |
| 1996 | ||
| 1997 | case OP_CIRCM: | case OP_CIRCM: |
| 1998 | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); |
| 1999 | if (eptr != md->start_subject && | if (eptr != md->start_subject && |
| 2000 | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) |
| 2001 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2002 | ecode++; | ecode++; |
| 2003 | break; | break; |
| 2004 | ||
| 2005 | /* Start of match assertion */ | /* Start of match assertion */ |
| 2006 | ||
| 2007 | case OP_SOM: | case OP_SOM: |
| 2008 | if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); |
| 2009 | ecode++; | ecode++; |
| 2010 | break; | break; |
| 2011 | ||
| # | Line 2018 for (;;) | Line 2021 for (;;) |
| 2021 | ||
| 2022 | case OP_DOLLM: | case OP_DOLLM: |
| 2023 | if (eptr < md->end_subject) | if (eptr < md->end_subject) |
| 2024 | { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); } | { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); } |
| 2025 | else | else |
| 2026 | { | { |
| 2027 | if (md->noteol) MRRETURN(MATCH_NOMATCH); | if (md->noteol) RRETURN(MATCH_NOMATCH); |
| 2028 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2029 | } | } |
| 2030 | ecode++; | ecode++; |
| # | Line 2031 for (;;) | Line 2034 for (;;) |
| 2034 | subject unless noteol is set. */ | subject unless noteol is set. */ |
| 2035 | ||
| 2036 | case OP_DOLL: | case OP_DOLL: |
| 2037 | if (md->noteol) MRRETURN(MATCH_NOMATCH); | if (md->noteol) RRETURN(MATCH_NOMATCH); |
| 2038 | if (!md->endonly) goto ASSERT_NL_OR_EOS; | if (!md->endonly) goto ASSERT_NL_OR_EOS; |
| 2039 | ||
| 2040 | /* ... else fall through for endonly */ | /* ... else fall through for endonly */ |
| # | Line 2039 for (;;) | Line 2042 for (;;) |
| 2042 | /* End of subject assertion (\z) */ | /* End of subject assertion (\z) */ |
| 2043 | ||
| 2044 | case OP_EOD: | case OP_EOD: |
| 2045 | if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH); | if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); |
| 2046 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2047 | ecode++; | ecode++; |
| 2048 | break; | break; |
| # | Line 2050 for (;;) | Line 2053 for (;;) |
| 2053 | ASSERT_NL_OR_EOS: | ASSERT_NL_OR_EOS: |
| 2054 | if (eptr < md->end_subject && | if (eptr < md->end_subject && |
| 2055 | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) |
| 2056 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2057 | ||
| 2058 | /* Either at end of string or \n before end. */ | /* Either at end of string or \n before end. */ |
| 2059 | ||
| # | Line 2069 for (;;) | Line 2072 for (;;) |
| 2072 | be "non-word" characters. Remember the earliest consulted character for | be "non-word" characters. Remember the earliest consulted character for |
| 2073 | partial matching. */ | partial matching. */ |
| 2074 | ||
| 2075 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2076 | if (utf8) | if (utf) |
| 2077 | { | { |
| 2078 | /* Get status of previous character */ | /* Get status of previous character */ |
| 2079 | ||
| 2080 | if (eptr == md->start_subject) prev_is_word = FALSE; else | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 2081 | { | { |
| 2082 | USPTR lastptr = eptr - 1; | PCRE_PUCHAR lastptr = eptr - 1; |
| 2083 | while((*lastptr & 0xc0) == 0x80) lastptr--; | BACKCHAR(lastptr); |
| 2084 | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; |
| 2085 | GETCHAR(c, lastptr); | GETCHAR(c, lastptr); |
| 2086 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| # | Line 2142 for (;;) | Line 2145 for (;;) |
| 2145 | } | } |
| 2146 | else | else |
| 2147 | #endif | #endif |
| 2148 | prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0); | prev_is_word = MAX_255(eptr[-1]) |
| 2149 | && ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
| 2150 | } | } |
| 2151 | ||
| 2152 | /* Get status of next character */ | /* Get status of next character */ |
| # | Line 2165 for (;;) | Line 2169 for (;;) |
| 2169 | } | } |
| 2170 | else | else |
| 2171 | #endif | #endif |
| 2172 | cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0); | cur_is_word = MAX_255(*eptr) |
| 2173 | && ((md->ctypes[*eptr] & ctype_word) != 0); | |
| 2174 | } | } |
| 2175 | ||
| 2176 | /* Now see if the situation is what we want */ | /* Now see if the situation is what we want */ |
| 2177 | ||
| 2178 | if ((*ecode++ == OP_WORD_BOUNDARY)? | if ((*ecode++ == OP_WORD_BOUNDARY)? |
| 2179 | cur_is_word == prev_is_word : cur_is_word != prev_is_word) | cur_is_word == prev_is_word : cur_is_word != prev_is_word) |
| 2180 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2181 | } | } |
| 2182 | break; | break; |
| 2183 | ||
| 2184 | /* Match a single character type; inline for speed */ | /* Match a single character type; inline for speed */ |
| 2185 | ||
| 2186 | case OP_ANY: | case OP_ANY: |
| 2187 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); |
| 2188 | /* Fall through */ | /* Fall through */ |
| 2189 | ||
| 2190 | case OP_ALLANY: | case OP_ALLANY: |
| 2191 | 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 */ |
| 2192 | { /* not be updated before SCHECK_PARTIAL. */ | { /* not be updated before SCHECK_PARTIAL. */ |
| 2193 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2194 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2195 | } | } |
| 2196 | eptr++; | eptr++; |
| 2197 | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | #ifdef SUPPORT_UTF |
| 2198 | if (utf) ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); | |
| 2199 | #endif | |
| 2200 | ecode++; | ecode++; |
| 2201 | break; | break; |
| 2202 | ||
| # | Line 2200 for (;;) | Line 2207 for (;;) |
| 2207 | 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 */ |
| 2208 | { /* not be updated before SCHECK_PARTIAL. */ | { /* not be updated before SCHECK_PARTIAL. */ |
| 2209 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2210 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2211 | } | } |
| 2212 | eptr++; | eptr++; |
| 2213 | ecode++; | ecode++; |
| # | Line 2210 for (;;) | Line 2217 for (;;) |
| 2217 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2218 | { | { |
| 2219 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2220 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2221 | } | } |
| 2222 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2223 | if ( | if ( |
| 2224 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2225 | c < 256 && | c < 256 && |
| 2226 | #endif | #endif |
| 2227 | (md->ctypes[c] & ctype_digit) != 0 | (md->ctypes[c] & ctype_digit) != 0 |
| 2228 | ) | ) |
| 2229 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2230 | ecode++; | ecode++; |
| 2231 | break; | break; |
| 2232 | ||
| # | Line 2227 for (;;) | Line 2234 for (;;) |
| 2234 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2235 | { | { |
| 2236 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2237 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2238 | } | } |
| 2239 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2240 | if ( | if ( |
| 2241 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2242 | c >= 256 || | c > 255 || |
| 2243 | #endif | #endif |
| 2244 | (md->ctypes[c] & ctype_digit) == 0 | (md->ctypes[c] & ctype_digit) == 0 |
| 2245 | ) | ) |
| 2246 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2247 | ecode++; | ecode++; |
| 2248 | break; | break; |
| 2249 | ||
| # | Line 2244 for (;;) | Line 2251 for (;;) |
| 2251 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2252 | { | { |
| 2253 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2254 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2255 | } | } |
| 2256 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2257 | if ( | if ( |
| 2258 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2259 | c < 256 && | c < 256 && |
| 2260 | #endif | #endif |
| 2261 | (md->ctypes[c] & ctype_space) != 0 | (md->ctypes[c] & ctype_space) != 0 |
| 2262 | ) | ) |
| 2263 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2264 | ecode++; | ecode++; |
| 2265 | break; | break; |
| 2266 | ||
| # | Line 2261 for (;;) | Line 2268 for (;;) |
| 2268 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2269 | { | { |
| 2270 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2271 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2272 | } | } |
| 2273 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2274 | if ( | if ( |
| 2275 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2276 | c >= 256 || | c > 255 || |
| 2277 | #endif | #endif |
| 2278 | (md->ctypes[c] & ctype_space) == 0 | (md->ctypes[c] & ctype_space) == 0 |
| 2279 | ) | ) |
| 2280 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2281 | ecode++; | ecode++; |
| 2282 | break; | break; |
| 2283 | ||
| # | Line 2278 for (;;) | Line 2285 for (;;) |
| 2285 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2286 | { | { |
| 2287 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2288 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2289 | } | } |
| 2290 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2291 | if ( | if ( |
| 2292 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2293 | c < 256 && | c < 256 && |
| 2294 | #endif | #endif |
| 2295 | (md->ctypes[c] & ctype_word) != 0 | (md->ctypes[c] & ctype_word) != 0 |
| 2296 | ) | ) |
| 2297 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2298 | ecode++; | ecode++; |
| 2299 | break; | break; |
| 2300 | ||
| # | Line 2295 for (;;) | Line 2302 for (;;) |
| 2302 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2303 | { | { |
| 2304 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2305 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2306 | } | } |
| 2307 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2308 | if ( | if ( |
| 2309 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8) |
| 2310 | c >= 256 || | c > 255 || |
| 2311 | #endif | #endif |
| 2312 | (md->ctypes[c] & ctype_word) == 0 | (md->ctypes[c] & ctype_word) == 0 |
| 2313 | ) | ) |
| 2314 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2315 | ecode++; | ecode++; |
| 2316 | break; | break; |
| 2317 | ||
| # | Line 2312 for (;;) | Line 2319 for (;;) |
| 2319 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2320 | { | { |
| 2321 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2322 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2323 | } | } |
| 2324 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2325 | switch(c) | switch(c) |
| 2326 | { | { |
| 2327 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 2328 | ||
| 2329 | case 0x000d: | case 0x000d: |
| 2330 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| # | Line 2331 for (;;) | Line 2338 for (;;) |
| 2338 | case 0x0085: | case 0x0085: |
| 2339 | case 0x2028: | case 0x2028: |
| 2340 | case 0x2029: | case 0x2029: |
| 2341 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); |
| 2342 | break; | break; |
| 2343 | } | } |
| 2344 | ecode++; | ecode++; |
| # | Line 2341 for (;;) | Line 2348 for (;;) |
| 2348 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2349 | { | { |
| 2350 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2351 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2352 | } | } |
| 2353 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2354 | switch(c) | switch(c) |
| # | Line 2366 for (;;) | Line 2373 for (;;) |
| 2373 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 2374 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 2375 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 2376 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2377 | } | } |
| 2378 | ecode++; | ecode++; |
| 2379 | break; | break; |
| # | Line 2375 for (;;) | Line 2382 for (;;) |
| 2382 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2383 | { | { |
| 2384 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2385 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2386 | } | } |
| 2387 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2388 | switch(c) | switch(c) |
| 2389 | { | { |
| 2390 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 2391 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 2392 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 2393 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 2409 for (;;) | Line 2416 for (;;) |
| 2416 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2417 | { | { |
| 2418 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2419 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2420 | } | } |
| 2421 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2422 | switch(c) | switch(c) |
| # | Line 2422 for (;;) | Line 2429 for (;;) |
| 2429 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 2430 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 2431 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 2432 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2433 | } | } |
| 2434 | ecode++; | ecode++; |
| 2435 | break; | break; |
| # | Line 2431 for (;;) | Line 2438 for (;;) |
| 2438 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2439 | { | { |
| 2440 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2441 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2442 | } | } |
| 2443 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2444 | switch(c) | switch(c) |
| 2445 | { | { |
| 2446 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 2447 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 2448 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 2449 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 2458 for (;;) | Line 2465 for (;;) |
| 2465 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2466 | { | { |
| 2467 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2468 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2469 | } | } |
| 2470 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2471 | { | { |
| # | Line 2467 for (;;) | Line 2474 for (;;) |
| 2474 | switch(ecode[1]) | switch(ecode[1]) |
| 2475 | { | { |
| 2476 | case PT_ANY: | case PT_ANY: |
| 2477 | if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH); | if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); |
| 2478 | break; | break; |
| 2479 | ||
| 2480 | case PT_LAMP: | case PT_LAMP: |
| 2481 | if ((prop->chartype == ucp_Lu || | if ((prop->chartype == ucp_Lu || |
| 2482 | prop->chartype == ucp_Ll || | prop->chartype == ucp_Ll || |
| 2483 | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) |
| 2484 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2485 | break; | break; |
| 2486 | ||
| 2487 | case PT_GC: | case PT_GC: |
| 2488 | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) | if ((ecode[2] != PRIV(ucp_gentype)[prop->chartype]) == (op == OP_PROP)) |
| 2489 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2490 | break; | break; |
| 2491 | ||
| 2492 | case PT_PC: | case PT_PC: |
| 2493 | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) |
| 2494 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2495 | break; | break; |
| 2496 | ||
| 2497 | case PT_SC: | case PT_SC: |
| 2498 | if ((ecode[2] != prop->script) == (op == OP_PROP)) | if ((ecode[2] != prop->script) == (op == OP_PROP)) |
| 2499 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2500 | break; | break; |
| 2501 | ||
| 2502 | /* These are specials */ | /* These are specials */ |
| 2503 | ||
| 2504 | case PT_ALNUM: | case PT_ALNUM: |
| 2505 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || |
| 2506 | _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) | PRIV(ucp_gentype)[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) |
| 2507 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2508 | break; | break; |
| 2509 | ||
| 2510 | case PT_SPACE: /* Perl space */ | case PT_SPACE: /* Perl space */ |
| 2511 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z || |
| 2512 | c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) | c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) |
| 2513 | == (op == OP_NOTPROP)) | == (op == OP_NOTPROP)) |
| 2514 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2515 | break; | break; |
| 2516 | ||
| 2517 | case PT_PXSPACE: /* POSIX space */ | case PT_PXSPACE: /* POSIX space */ |
| 2518 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_Z || |
| 2519 | c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || | c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || |
| 2520 | c == CHAR_FF || c == CHAR_CR) | 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_WORD: | case PT_WORD: |
| 2526 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | if ((PRIV(ucp_gentype)[prop->chartype] == ucp_L || |
| 2527 | _pcre_ucp_gentype[prop->chartype] == ucp_N || | PRIV(ucp_gentype)[prop->chartype] == ucp_N || |
| 2528 | c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) | c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) |
| 2529 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2530 | break; | break; |
| 2531 | ||
| 2532 | /* This should never occur */ | /* This should never occur */ |
| # | Line 2539 for (;;) | Line 2546 for (;;) |
| 2546 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2547 | { | { |
| 2548 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2549 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2550 | } | } |
| 2551 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2552 | if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); | if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); |
| 2553 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 2554 | { | { |
| 2555 | int len = 1; | int len = 1; |
| 2556 | if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } | if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
| 2557 | if (UCD_CATEGORY(c) != ucp_M) break; | if (UCD_CATEGORY(c) != ucp_M) break; |
| 2558 | eptr += len; | eptr += len; |
| 2559 | } | } |
| # | Line 2567 for (;;) | Line 2574 for (;;) |
| 2574 | case OP_REFI: | case OP_REFI: |
| 2575 | caseless = op == OP_REFI; | caseless = op == OP_REFI; |
| 2576 | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ |
| 2577 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 2578 | ||
| 2579 | /* If the reference is unset, there are two possibilities: | /* If the reference is unset, there are two possibilities: |
| 2580 | ||
| # | Line 2607 for (;;) | Line 2614 for (;;) |
| 2614 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
| 2615 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
| 2616 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
| 2617 | max = GET2(ecode, 3); | max = GET2(ecode, 1 + IMM2_SIZE); |
| 2618 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2619 | ecode += 5; | ecode += 1 + 2 * IMM2_SIZE; |
| 2620 | break; | break; |
| 2621 | ||
| 2622 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| 2623 | if ((length = match_ref(offset, eptr, length, md, caseless)) < 0) | if ((length = match_ref(offset, eptr, length, md, caseless)) < 0) |
| 2624 | { | { |
| 2625 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
| 2626 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2627 | } | } |
| 2628 | eptr += length; | eptr += length; |
| 2629 | continue; /* With the main loop */ | continue; /* With the main loop */ |
| 2630 | } | } |
| 2631 | ||
| 2632 | /* Handle repeated back references. If the length of the reference is | /* Handle repeated back references. If the length of the reference is |
| 2633 | zero, just continue with the main loop. */ | zero, just continue with the main loop. If the length is negative, it |
| 2634 | means the reference is unset in non-Java-compatible mode. If the minimum is | |
| 2635 | zero, we can continue at the same level without recursion. For any other | |
| 2636 | minimum, carrying on will result in NOMATCH. */ | |
| 2637 | ||
| 2638 | if (length == 0) continue; | if (length == 0) continue; |
| 2639 | if (length < 0 && min == 0) continue; | |
| 2640 | ||
| 2641 | /* First, ensure the minimum number of matches are present. We get back | /* First, ensure the minimum number of matches are present. We get back |
| 2642 | the length of the reference string explicitly rather than passing the | the length of the reference string explicitly rather than passing the |
| # | Line 2637 for (;;) | Line 2648 for (;;) |
| 2648 | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
| 2649 | { | { |
| 2650 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
| 2651 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2652 | } | } |
| 2653 | eptr += slength; | eptr += slength; |
| 2654 | } | } |
| # | Line 2656 for (;;) | Line 2667 for (;;) |
| 2667 | int slength; | int slength; |
| 2668 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM14); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM14); |
| 2669 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2670 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 2671 | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) | if ((slength = match_ref(offset, eptr, length, md, caseless)) < 0) |
| 2672 | { | { |
| 2673 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
| 2674 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2675 | } | } |
| 2676 | eptr += slength; | eptr += slength; |
| 2677 | } | } |
| # | Line 2688 for (;;) | Line 2699 for (;;) |
| 2699 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2700 | eptr -= length; | eptr -= length; |
| 2701 | } | } |
| 2702 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2703 | } | } |
| 2704 | /* Control never gets here */ | /* Control never gets here */ |
| 2705 | ||
| # | Line 2706 for (;;) | Line 2717 for (;;) |
| 2717 | case OP_NCLASS: | case OP_NCLASS: |
| 2718 | case OP_CLASS: | case OP_CLASS: |
| 2719 | { | { |
| 2720 | /* The data variable is saved across frames, so the byte map needs to | |
| 2721 | be stored there. */ | |
| 2722 | #define BYTE_MAP ((pcre_uint8 *)data) | |
| 2723 | data = ecode + 1; /* Save for matching */ | data = ecode + 1; /* Save for matching */ |
| 2724 | ecode += 33; /* Advance past the item */ | ecode += 1 + (32 / sizeof(pcre_uchar)); /* Advance past the item */ |
| 2725 | ||
| 2726 | switch (*ecode) | switch (*ecode) |
| 2727 | { | { |
| # | Line 2728 for (;;) | Line 2742 for (;;) |
| 2742 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
| 2743 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
| 2744 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
| 2745 | max = GET2(ecode, 3); | max = GET2(ecode, 1 + IMM2_SIZE); |
| 2746 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2747 | ecode += 5; | ecode += 1 + 2 * IMM2_SIZE; |
| 2748 | break; | break; |
| 2749 | ||
| 2750 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| # | Line 2740 for (;;) | Line 2754 for (;;) |
| 2754 | ||
| 2755 | /* First, ensure the minimum number of matches are present. */ | /* First, ensure the minimum number of matches are present. */ |
| 2756 | ||
| 2757 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2758 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 2759 | { | { |
| 2760 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2761 | { | { |
| 2762 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2763 | { | { |
| 2764 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2765 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2766 | } | } |
| 2767 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2768 | if (c > 255) | if (c > 255) |
| 2769 | { | { |
| 2770 | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); |
| 2771 | } | } |
| 2772 | else | else |
| 2773 | { | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); |
| if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | ||
| } | ||
| 2774 | } | } |
| 2775 | } | } |
| 2776 | else | else |
| 2777 | #endif | #endif |
| 2778 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 2779 | { | { |
| 2780 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2781 | { | { |
| 2782 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2783 | { | { |
| 2784 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2785 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2786 | } | } |
| 2787 | c = *eptr++; | c = *eptr++; |
| 2788 | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | #ifndef COMPILE_PCRE8 |
| 2789 | if (c > 255) | |
| 2790 | { | |
| 2791 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | |
| 2792 | } | |
| 2793 | else | |
| 2794 | #endif | |
| 2795 | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | |
| 2796 | } | } |
| 2797 | } | } |
| 2798 | ||
| # | Line 2788 for (;;) | Line 2806 for (;;) |
| 2806 | ||
| 2807 | if (minimize) | if (minimize) |
| 2808 | { | { |
| 2809 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2810 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 2811 | { | { |
| 2812 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 2813 | { | { |
| 2814 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM16); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM16); |
| 2815 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2816 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 2817 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2818 | { | { |
| 2819 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2820 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2821 | } | } |
| 2822 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2823 | if (c > 255) | if (c > 255) |
| 2824 | { | { |
| 2825 | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); |
| 2826 | } | } |
| 2827 | else | else |
| 2828 | { | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); |
| if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | ||
| } | ||
| 2829 | } | } |
| 2830 | } | } |
| 2831 | else | else |
| 2832 | #endif | #endif |
| 2833 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 2834 | { | { |
| 2835 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 2836 | { | { |
| 2837 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM17); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM17); |
| 2838 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2839 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 2840 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2841 | { | { |
| 2842 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2843 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2844 | } | } |
| 2845 | c = *eptr++; | c = *eptr++; |
| 2846 | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); | #ifndef COMPILE_PCRE8 |
| 2847 | if (c > 255) | |
| 2848 | { | |
| 2849 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | |
| 2850 | } | |
| 2851 | else | |
| 2852 | #endif | |
| 2853 | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | |
| 2854 | } | } |
| 2855 | } | } |
| 2856 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2840 for (;;) | Line 2862 for (;;) |
| 2862 | { | { |
| 2863 | pp = eptr; | pp = eptr; |
| 2864 | ||
| 2865 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 2866 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 2867 | { | { |
| 2868 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2869 | { | { |
| # | Line 2858 for (;;) | Line 2879 for (;;) |
| 2879 | if (op == OP_CLASS) break; | if (op == OP_CLASS) break; |
| 2880 | } | } |
| 2881 | else | else |
| 2882 | { | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; |
| if ((data[c/8] & (1 << (c&7))) == 0) break; | ||
| } | ||
| 2883 | eptr += len; | eptr += len; |
| 2884 | } | } |
| 2885 | for (;;) | for (;;) |
| # | Line 2873 for (;;) | Line 2892 for (;;) |
| 2892 | } | } |
| 2893 | else | else |
| 2894 | #endif | #endif |
| 2895 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 2896 | { | { |
| 2897 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2898 | { | { |
| # | Line 2883 for (;;) | Line 2902 for (;;) |
| 2902 | break; | break; |
| 2903 | } | } |
| 2904 | c = *eptr; | c = *eptr; |
| 2905 | if ((data[c/8] & (1 << (c&7))) == 0) break; | #ifndef COMPILE_PCRE8 |
| 2906 | if (c > 255) | |
| 2907 | { | |
| 2908 | if (op == OP_CLASS) break; | |
| 2909 | } | |
| 2910 | else | |
| 2911 | #endif | |
| 2912 | if ((BYTE_MAP[c/8] & (1 << (c&7))) == 0) break; | |
| 2913 | eptr++; | eptr++; |
| 2914 | } | } |
| 2915 | while (eptr >= pp) | while (eptr >= pp) |
| # | Line 2894 for (;;) | Line 2920 for (;;) |
| 2920 | } | } |
| 2921 | } | } |
| 2922 | ||
| 2923 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2924 | } | } |
| 2925 | #undef BYTE_MAP | |
| 2926 | } | } |
| 2927 | /* Control never gets here */ | /* Control never gets here */ |
| 2928 | ||
| # | Line 2904 for (;;) | Line 2931 for (;;) |
| 2931 | 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 |
| 2932 | mode, because Unicode properties are supported in non-UTF-8 mode. */ | mode, because Unicode properties are supported in non-UTF-8 mode. */ |
| 2933 | ||
| 2934 | #ifdef SUPPORT_UTF8 | #if defined SUPPORT_UTF || !defined COMPILE_PCRE8 |
| 2935 | case OP_XCLASS: | case OP_XCLASS: |
| 2936 | { | { |
| 2937 | data = ecode + 1 + LINK_SIZE; /* Save for matching */ | data = ecode + 1 + LINK_SIZE; /* Save for matching */ |
| # | Line 2929 for (;;) | Line 2956 for (;;) |
| 2956 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
| 2957 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
| 2958 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
| 2959 | max = GET2(ecode, 3); | max = GET2(ecode, 1 + IMM2_SIZE); |
| 2960 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2961 | ecode += 5; | ecode += 1 + 2 * IMM2_SIZE; |
| 2962 | break; | break; |
| 2963 | ||
| 2964 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| # | Line 2946 for (;;) | Line 2973 for (;;) |
| 2973 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2974 | { | { |
| 2975 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 2976 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 2977 | } | } |
| 2978 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2979 | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); | if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); |
| 2980 | } | } |
| 2981 | ||
| 2982 | /* 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 2966 for (;;) | Line 2993 for (;;) |
| 2993 | { | { |
| 2994 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM20); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM20); |
| 2995 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2996 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 2997 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 2998 | { | { |
| 2999 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3000 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3001 | } | } |
| 3002 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3003 | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); | if (!PRIV(xclass)(c, data, utf)) RRETURN(MATCH_NOMATCH); |
| 3004 | } | } |
| 3005 | /* Control never gets here */ | /* Control never gets here */ |
| 3006 | } | } |
| # | Line 2991 for (;;) | Line 3018 for (;;) |
| 3018 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3019 | break; | break; |
| 3020 | } | } |
| 3021 | #ifdef SUPPORT_UTF | |
| 3022 | GETCHARLENTEST(c, eptr, len); | GETCHARLENTEST(c, eptr, len); |
| 3023 | if (!_pcre_xclass(c, data)) break; | #else |
| 3024 | c = *eptr; | |
| 3025 | #endif | |
| 3026 | if (!PRIV(xclass)(c, data, utf)) break; | |
| 3027 | eptr += len; | eptr += len; |
| 3028 | } | } |
| 3029 | for(;;) | for(;;) |
| # | Line 3000 for (;;) | Line 3031 for (;;) |
| 3031 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM21); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM21); |
| 3032 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3033 | if (eptr-- == pp) break; /* Stop if tried at original pos */ | if (eptr-- == pp) break; /* Stop if tried at original pos */ |
| 3034 | if (utf8) BACKCHAR(eptr); | #ifdef SUPPORT_UTF |
| 3035 | if (utf) BACKCHAR(eptr); | |
| 3036 | #endif | |
| 3037 | } | } |
| 3038 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3039 | } | } |
| 3040 | ||
| 3041 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3012 for (;;) | Line 3045 for (;;) |
| 3045 | /* Match a single character, casefully */ | /* Match a single character, casefully */ |
| 3046 | ||
| 3047 | case OP_CHAR: | case OP_CHAR: |
| 3048 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3049 | if (utf8) | if (utf) |
| 3050 | { | { |
| 3051 | length = 1; | length = 1; |
| 3052 | ecode++; | ecode++; |
| # | Line 3021 for (;;) | Line 3054 for (;;) |
| 3054 | if (length > md->end_subject - eptr) | if (length > md->end_subject - eptr) |
| 3055 | { | { |
| 3056 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ |
| 3057 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3058 | } | } |
| 3059 | while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH); | while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH); |
| 3060 | } | } |
| 3061 | else | else |
| 3062 | #endif | #endif |
| 3063 | /* Not UTF mode */ | |
| /* Non-UTF-8 mode */ | ||
| 3064 | { | { |
| 3065 | if (md->end_subject - eptr < 1) | if (md->end_subject - eptr < 1) |
| 3066 | { | { |
| 3067 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ |
| 3068 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3069 | } | } |
| 3070 | if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH); | if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); |
| 3071 | ecode += 2; | ecode += 2; |
| 3072 | } | } |
| 3073 | break; | break; |
| 3074 | ||
| 3075 | /* Match a single character, caselessly */ | /* Match a single character, caselessly. If we are at the end of the |
| 3076 | subject, give up immediately. */ | |
| 3077 | ||
| 3078 | case OP_CHARI: | case OP_CHARI: |
| 3079 | #ifdef SUPPORT_UTF8 | if (eptr >= md->end_subject) |
| 3080 | if (utf8) | { |
| 3081 | SCHECK_PARTIAL(); | |
| 3082 | RRETURN(MATCH_NOMATCH); | |
| 3083 | } | |
| 3084 | ||
| 3085 | #ifdef SUPPORT_UTF | |
| 3086 | if (utf) | |
| 3087 | { | { |
| 3088 | length = 1; | length = 1; |
| 3089 | ecode++; | ecode++; |
| 3090 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 3091 | ||
| if (length > md->end_subject - eptr) | ||
| { | ||
| CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | ||
| MRRETURN(MATCH_NOMATCH); | ||
| } | ||
| 3092 | /* 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 |
| 3093 | can use the fast lookup table. */ | we know that its other case must also be one byte long, so we can use the |
| 3094 | fast lookup table. We know that there is at least one byte left in the | |
| 3095 | subject. */ | |
| 3096 | ||
| 3097 | if (fc < 128) | if (fc < 128) |
| 3098 | { | { |
| 3099 | if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (md->lcc[fc] |
| 3100 | != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH); | |
| 3101 | ecode++; | |
| 3102 | eptr++; | |
| 3103 | } | } |
| 3104 | ||
| 3105 | /* Otherwise we must pick up the subject character */ | /* Otherwise we must pick up the subject character. Note that we cannot |
| 3106 | use the value of "length" to check for sufficient bytes left, because the | |
| 3107 | other case of the character may have more or fewer bytes. */ | |
| 3108 | ||
| 3109 | else | else |
| 3110 | { | { |
| # | Line 3080 for (;;) | Line 3120 for (;;) |
| 3120 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3121 | if (dc != UCD_OTHERCASE(fc)) | if (dc != UCD_OTHERCASE(fc)) |
| 3122 | #endif | #endif |
| 3123 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3124 | } | } |
| 3125 | } | } |
| 3126 | } | } |
| 3127 | else | else |
| 3128 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF */ |
| 3129 | ||
| 3130 | /* Non-UTF-8 mode */ | /* Not UTF mode */ |
| 3131 | { | { |
| 3132 | if (md->end_subject - eptr < 1) | if (TABLE_GET(ecode[1], md->lcc, ecode[1]) |
| 3133 | { | != TABLE_GET(*eptr, md->lcc, *eptr)) RRETURN(MATCH_NOMATCH); |
| 3134 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | eptr++; |
| MRRETURN(MATCH_NOMATCH); | ||
| } | ||
| if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | ||
| 3135 | ecode += 2; | ecode += 2; |
| 3136 | } | } |
| 3137 | break; | break; |
| # | Line 3104 for (;;) | Line 3141 for (;;) |
| 3141 | case OP_EXACT: | case OP_EXACT: |
| 3142 | case OP_EXACTI: | case OP_EXACTI: |
| 3143 | min = max = GET2(ecode, 1); | min = max = GET2(ecode, 1); |
| 3144 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3145 | goto REPEATCHAR; | goto REPEATCHAR; |
| 3146 | ||
| 3147 | case OP_POSUPTO: | case OP_POSUPTO: |
| # | Line 3119 for (;;) | Line 3156 for (;;) |
| 3156 | min = 0; | min = 0; |
| 3157 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
| 3158 | minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI; | minimize = *ecode == OP_MINUPTO || *ecode == OP_MINUPTOI; |
| 3159 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3160 | goto REPEATCHAR; | goto REPEATCHAR; |
| 3161 | ||
| 3162 | case OP_POSSTAR: | case OP_POSSTAR: |
| # | Line 3167 for (;;) | Line 3204 for (;;) |
| 3204 | /* Common code for all repeated single-character matches. */ | /* Common code for all repeated single-character matches. */ |
| 3205 | ||
| 3206 | REPEATCHAR: | REPEATCHAR: |
| 3207 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3208 | if (utf8) | if (utf) |
| 3209 | { | { |
| 3210 | length = 1; | length = 1; |
| 3211 | charptr = ecode; | charptr = ecode; |
| # | Line 3184 for (;;) | Line 3221 for (;;) |
| 3221 | unsigned int othercase; | unsigned int othercase; |
| 3222 | if (op >= OP_STARI && /* Caseless */ | if (op >= OP_STARI && /* Caseless */ |
| 3223 | (othercase = UCD_OTHERCASE(fc)) != fc) | (othercase = UCD_OTHERCASE(fc)) != fc) |
| 3224 | oclength = _pcre_ord2utf8(othercase, occhars); | oclength = PRIV(ord2utf)(othercase, occhars); |
| 3225 | else oclength = 0; | else oclength = 0; |
| 3226 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 3227 | ||
| 3228 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3229 | { | { |
| 3230 | if (eptr <= md->end_subject - length && | if (eptr <= md->end_subject - length && |
| 3231 | memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; |
| 3232 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3233 | else if (oclength > 0 && | else if (oclength > 0 && |
| 3234 | eptr <= md->end_subject - oclength && | eptr <= md->end_subject - oclength && |
| 3235 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; |
| 3236 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 3237 | else | else |
| 3238 | { | { |
| 3239 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
| 3240 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3241 | } | } |
| 3242 | } | } |
| 3243 | ||
| # | Line 3212 for (;;) | Line 3249 for (;;) |
| 3249 | { | { |
| 3250 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM22); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM22); |
| 3251 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3252 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3253 | if (eptr <= md->end_subject - length && | if (eptr <= md->end_subject - length && |
| 3254 | memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; |
| 3255 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3256 | else if (oclength > 0 && | else if (oclength > 0 && |
| 3257 | eptr <= md->end_subject - oclength && | eptr <= md->end_subject - oclength && |
| 3258 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; |
| 3259 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 3260 | else | else |
| 3261 | { | { |
| 3262 | CHECK_PARTIAL(); | CHECK_PARTIAL(); |
| 3263 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3264 | } | } |
| 3265 | } | } |
| 3266 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3235 for (;;) | Line 3272 for (;;) |
| 3272 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3273 | { | { |
| 3274 | if (eptr <= md->end_subject - length && | if (eptr <= md->end_subject - length && |
| 3275 | memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, IN_UCHARS(length)) == 0) eptr += length; |
| 3276 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3277 | else if (oclength > 0 && | else if (oclength > 0 && |
| 3278 | eptr <= md->end_subject - oclength && | eptr <= md->end_subject - oclength && |
| 3279 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | memcmp(eptr, occhars, IN_UCHARS(oclength)) == 0) eptr += oclength; |
| 3280 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 3281 | else | else |
| 3282 | { | { |
| # | Line 3254 for (;;) | Line 3291 for (;;) |
| 3291 | { | { |
| 3292 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM23); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM23); |
| 3293 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3294 | if (eptr == pp) { MRRETURN(MATCH_NOMATCH); } | if (eptr == pp) { RRETURN(MATCH_NOMATCH); } |
| 3295 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 3296 | eptr--; | eptr--; |
| 3297 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| # | Line 3271 for (;;) | Line 3308 for (;;) |
| 3308 | value of fc will always be < 128. */ | value of fc will always be < 128. */ |
| 3309 | } | } |
| 3310 | else | else |
| 3311 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF */ |
| 3312 | /* When not in UTF-8 mode, load a single-byte character. */ | |
| 3313 | /* When not in UTF-8 mode, load a single-byte character. */ | fc = *ecode++; |
| 3314 | ||
| 3315 | fc = *ecode++; | /* The value of fc at this point is always one character, though we may |
| 3316 | or may not be in UTF mode. The code is duplicated for the caseless and | |
| /* The value of fc at this point is always less than 256, though we may or | ||
| may not be in UTF-8 mode. The code is duplicated for the caseless and | ||
| 3317 | caseful cases, for speed, since matching characters is likely to be quite | caseful cases, for speed, since matching characters is likely to be quite |
| 3318 | common. First, ensure the minimum number of matches are present. If min = | common. First, ensure the minimum number of matches are present. If min = |
| 3319 | max, continue at the same level without recursing. Otherwise, if | max, continue at the same level without recursing. Otherwise, if |
| # | Line 3291 for (;;) | Line 3326 for (;;) |
| 3326 | ||
| 3327 | if (op >= OP_STARI) /* Caseless */ | if (op >= OP_STARI) /* Caseless */ |
| 3328 | { | { |
| 3329 | fc = md->lcc[fc]; | #ifdef COMPILE_PCRE8 |
| 3330 | /* fc must be < 128 if UTF is enabled. */ | |
| 3331 | foc = md->fcc[fc]; | |
| 3332 | #else | |
| 3333 | #ifdef SUPPORT_UTF | |
| 3334 | #ifdef SUPPORT_UCP | |
| 3335 | if (utf && fc > 127) | |
| 3336 | foc = UCD_OTHERCASE(fc); | |
| 3337 | #else | |
| 3338 | if (utf && fc > 127) | |
| 3339 | foc = fc; | |
| 3340 | #endif /* SUPPORT_UCP */ | |
| 3341 | else | |
| 3342 | #endif /* SUPPORT_UTF */ | |
| 3343 | foc = TABLE_GET(fc, md->fcc, fc); | |
| 3344 | #endif /* COMPILE_PCRE8 */ | |
| 3345 | ||
| 3346 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3347 | { | { |
| 3348 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3349 | { | { |
| 3350 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3351 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3352 | } | } |
| 3353 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH); |
| 3354 | eptr++; | |
| 3355 | } | } |
| 3356 | if (min == max) continue; | if (min == max) continue; |
| 3357 | if (minimize) | if (minimize) |
| # | Line 3308 for (;;) | Line 3360 for (;;) |
| 3360 | { | { |
| 3361 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM24); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM24); |
| 3362 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3363 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3364 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3365 | { | { |
| 3366 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3367 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3368 | } | } |
| 3369 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr && foc != *eptr) RRETURN(MATCH_NOMATCH); |
| 3370 | eptr++; | |
| 3371 | } | } |
| 3372 | /* Control never gets here */ | /* Control never gets here */ |
| 3373 | } | } |
| # | Line 3328 for (;;) | Line 3381 for (;;) |
| 3381 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3382 | break; | break; |
| 3383 | } | } |
| 3384 | if (fc != md->lcc[*eptr]) break; | if (fc != *eptr && foc != *eptr) break; |
| 3385 | eptr++; | eptr++; |
| 3386 | } | } |
| 3387 | ||
| # | Line 3340 for (;;) | Line 3393 for (;;) |
| 3393 | eptr--; | eptr--; |
| 3394 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3395 | } | } |
| 3396 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3397 | } | } |
| 3398 | /* Control never gets here */ | /* Control never gets here */ |
| 3399 | } | } |
| # | Line 3354 for (;;) | Line 3407 for (;;) |
| 3407 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3408 | { | { |
| 3409 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3410 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3411 | } | } |
| 3412 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr++) RRETURN(MATCH_NOMATCH); |
| 3413 | } | } |
| 3414 | ||
| 3415 | if (min == max) continue; | if (min == max) continue; |
| # | Line 3367 for (;;) | Line 3420 for (;;) |
| 3420 | { | { |
| 3421 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM26); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM26); |
| 3422 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3423 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3424 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3425 | { | { |
| 3426 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3427 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3428 | } | } |
| 3429 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc != *eptr++) RRETURN(MATCH_NOMATCH); |
| 3430 | } | } |
| 3431 | /* Control never gets here */ | /* Control never gets here */ |
| 3432 | } | } |
| # | Line 3398 for (;;) | Line 3451 for (;;) |
| 3451 | eptr--; | eptr--; |
| 3452 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3453 | } | } |
| 3454 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3455 | } | } |
| 3456 | } | } |
| 3457 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3411 for (;;) | Line 3464 for (;;) |
| 3464 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3465 | { | { |
| 3466 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3467 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3468 | } | } |
| 3469 | ecode++; | ecode++; |
| 3470 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3471 | if (op == OP_NOTI) /* The caseless case */ | if (op == OP_NOTI) /* The caseless case */ |
| 3472 | { | { |
| 3473 | #ifdef SUPPORT_UTF8 | register int ch, och; |
| 3474 | if (c < 256) | ch = *ecode++; |
| 3475 | #endif | #ifdef COMPILE_PCRE8 |
| 3476 | c = md->lcc[c]; | /* ch must be < 128 if UTF is enabled. */ |
| 3477 | if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH); | och = md->fcc[ch]; |
| 3478 | #else | |
| 3479 | #ifdef SUPPORT_UTF | |
| 3480 | #ifdef SUPPORT_UCP | |
| 3481 | if (utf && ch > 127) | |
| 3482 | och = UCD_OTHERCASE(ch); | |
| 3483 | #else | |
| 3484 | if (utf && ch > 127) | |
| 3485 | och = ch; | |
| 3486 | #endif /* SUPPORT_UCP */ | |
| 3487 | else | |
| 3488 | #endif /* SUPPORT_UTF */ | |
| 3489 | och = TABLE_GET(ch, md->fcc, ch); | |
| 3490 | #endif /* COMPILE_PCRE8 */ | |
| 3491 | if (ch == c || och == c) RRETURN(MATCH_NOMATCH); | |
| 3492 | } | } |
| 3493 | else /* Caseful */ | else /* Caseful */ |
| 3494 | { | { |
| 3495 | if (*ecode++ == c) MRRETURN(MATCH_NOMATCH); | if (*ecode++ == c) RRETURN(MATCH_NOMATCH); |
| 3496 | } | } |
| 3497 | break; | break; |
| 3498 | ||
| # | Line 3439 for (;;) | Line 3506 for (;;) |
| 3506 | case OP_NOTEXACT: | case OP_NOTEXACT: |
| 3507 | case OP_NOTEXACTI: | case OP_NOTEXACTI: |
| 3508 | min = max = GET2(ecode, 1); | min = max = GET2(ecode, 1); |
| 3509 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3510 | goto REPEATNOTCHAR; | goto REPEATNOTCHAR; |
| 3511 | ||
| 3512 | case OP_NOTUPTO: | case OP_NOTUPTO: |
| # | Line 3449 for (;;) | Line 3516 for (;;) |
| 3516 | min = 0; | min = 0; |
| 3517 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
| 3518 | minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI; | minimize = *ecode == OP_NOTMINUPTO || *ecode == OP_NOTMINUPTOI; |
| 3519 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3520 | goto REPEATNOTCHAR; | goto REPEATNOTCHAR; |
| 3521 | ||
| 3522 | case OP_NOTPOSSTAR: | case OP_NOTPOSSTAR: |
| # | Line 3481 for (;;) | Line 3548 for (;;) |
| 3548 | possessive = TRUE; | possessive = TRUE; |
| 3549 | min = 0; | min = 0; |
| 3550 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
| 3551 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3552 | goto REPEATNOTCHAR; | goto REPEATNOTCHAR; |
| 3553 | ||
| 3554 | case OP_NOTSTAR: | case OP_NOTSTAR: |
| # | Line 3520 for (;;) | Line 3587 for (;;) |
| 3587 | ||
| 3588 | if (op >= OP_NOTSTARI) /* Caseless */ | if (op >= OP_NOTSTARI) /* Caseless */ |
| 3589 | { | { |
| 3590 | fc = md->lcc[fc]; | #ifdef COMPILE_PCRE8 |
| 3591 | /* fc must be < 128 if UTF is enabled. */ | |
| 3592 | foc = md->fcc[fc]; | |
| 3593 | #else | |
| 3594 | #ifdef SUPPORT_UTF | |
| 3595 | #ifdef SUPPORT_UCP | |
| 3596 | if (utf && fc > 127) | |
| 3597 | foc = UCD_OTHERCASE(fc); | |
| 3598 | #else | |
| 3599 | if (utf && fc > 127) | |
| 3600 | foc = fc; | |
| 3601 | #endif /* SUPPORT_UCP */ | |
| 3602 | else | |
| 3603 | #endif /* SUPPORT_UTF */ | |
| 3604 | foc = TABLE_GET(fc, md->fcc, fc); | |
| 3605 | #endif /* COMPILE_PCRE8 */ | |
| 3606 | ||
| 3607 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3608 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3609 | { | { |
| 3610 | register unsigned int d; | register unsigned int d; |
| 3611 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| # | Line 3532 for (;;) | Line 3613 for (;;) |
| 3613 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3614 | { | { |
| 3615 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3616 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3617 | } | } |
| 3618 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3619 | if (d < 256) d = md->lcc[d]; | if (fc == d || foc == d) RRETURN(MATCH_NOMATCH); |
| if (fc == d) MRRETURN(MATCH_NOMATCH); | ||
| 3620 | } | } |
| 3621 | } | } |
| 3622 | else | else |
| 3623 | #endif | #endif |
| 3624 | /* Not UTF mode */ | |
| /* Not UTF-8 mode */ | ||
| 3625 | { | { |
| 3626 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3627 | { | { |
| 3628 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3629 | { | { |
| 3630 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3631 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3632 | } | } |
| 3633 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); |
| 3634 | eptr++; | |
| 3635 | } | } |
| 3636 | } | } |
| 3637 | ||
| # | Line 3559 for (;;) | Line 3639 for (;;) |
| 3639 | ||
| 3640 | if (minimize) | if (minimize) |
| 3641 | { | { |
| 3642 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3643 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3644 | { | { |
| 3645 | register unsigned int d; | register unsigned int d; |
| 3646 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3647 | { | { |
| 3648 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM28); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM28); |
| 3649 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3650 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3651 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3652 | { | { |
| 3653 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3654 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3655 | } | } |
| 3656 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3657 | if (d < 256) d = md->lcc[d]; | if (fc == d || foc == d) RRETURN(MATCH_NOMATCH); |
| if (fc == d) MRRETURN(MATCH_NOMATCH); | ||
| 3658 | } | } |
| 3659 | } | } |
| 3660 | else | else |
| 3661 | #endif | #endif |
| 3662 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 3663 | { | { |
| 3664 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3665 | { | { |
| 3666 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM29); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM29); |
| 3667 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3668 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3669 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3670 | { | { |
| 3671 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3672 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3673 | } | } |
| 3674 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr || foc == *eptr) RRETURN(MATCH_NOMATCH); |
| 3675 | eptr++; | |
| 3676 | } | } |
| 3677 | } | } |
| 3678 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3605 for (;;) | Line 3684 for (;;) |
| 3684 | { | { |
| 3685 | pp = eptr; | pp = eptr; |
| 3686 | ||
| 3687 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3688 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3689 | { | { |
| 3690 | register unsigned int d; | register unsigned int d; |
| 3691 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| # | Line 3619 for (;;) | Line 3697 for (;;) |
| 3697 | break; | break; |
| 3698 | } | } |
| 3699 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3700 | if (d < 256) d = md->lcc[d]; | if (fc == d || foc == d) break; |
| if (fc == d) break; | ||
| 3701 | eptr += len; | eptr += len; |
| 3702 | } | } |
| 3703 | if (possessive) continue; | if (possessive) continue; |
| # | Line 3634 for (;;) | Line 3711 for (;;) |
| 3711 | } | } |
| 3712 | else | else |
| 3713 | #endif | #endif |
| 3714 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 3715 | { | { |
| 3716 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3717 | { | { |
| # | Line 3643 for (;;) | Line 3720 for (;;) |
| 3720 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3721 | break; | break; |
| 3722 | } | } |
| 3723 | if (fc == md->lcc[*eptr]) break; | if (fc == *eptr || foc == *eptr) break; |
| 3724 | eptr++; | eptr++; |
| 3725 | } | } |
| 3726 | if (possessive) continue; | if (possessive) continue; |
| # | Line 3655 for (;;) | Line 3732 for (;;) |
| 3732 | } | } |
| 3733 | } | } |
| 3734 | ||
| 3735 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3736 | } | } |
| 3737 | /* Control never gets here */ | /* Control never gets here */ |
| 3738 | } | } |
| # | Line 3664 for (;;) | Line 3741 for (;;) |
| 3741 | ||
| 3742 | else | else |
| 3743 | { | { |
| 3744 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3745 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3746 | { | { |
| 3747 | register unsigned int d; | register unsigned int d; |
| 3748 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| # | Line 3674 for (;;) | Line 3750 for (;;) |
| 3750 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3751 | { | { |
| 3752 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3753 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3754 | } | } |
| 3755 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3756 | if (fc == d) MRRETURN(MATCH_NOMATCH); | if (fc == d) RRETURN(MATCH_NOMATCH); |
| 3757 | } | } |
| 3758 | } | } |
| 3759 | else | else |
| 3760 | #endif | #endif |
| 3761 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 3762 | { | { |
| 3763 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3764 | { | { |
| 3765 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3766 | { | { |
| 3767 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3768 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3769 | } | } |
| 3770 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); |
| 3771 | } | } |
| 3772 | } | } |
| 3773 | ||
| # | Line 3699 for (;;) | Line 3775 for (;;) |
| 3775 | ||
| 3776 | if (minimize) | if (minimize) |
| 3777 | { | { |
| 3778 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3779 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3780 | { | { |
| 3781 | register unsigned int d; | register unsigned int d; |
| 3782 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3783 | { | { |
| 3784 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM32); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM32); |
| 3785 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3786 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3787 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3788 | { | { |
| 3789 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3790 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3791 | } | } |
| 3792 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3793 | if (fc == d) MRRETURN(MATCH_NOMATCH); | if (fc == d) RRETURN(MATCH_NOMATCH); |
| 3794 | } | } |
| 3795 | } | } |
| 3796 | else | else |
| 3797 | #endif | #endif |
| 3798 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 3799 | { | { |
| 3800 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3801 | { | { |
| 3802 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM33); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM33); |
| 3803 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3804 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 3805 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3806 | { | { |
| 3807 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3808 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3809 | } | } |
| 3810 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); |
| 3811 | } | } |
| 3812 | } | } |
| 3813 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3744 for (;;) | Line 3819 for (;;) |
| 3819 | { | { |
| 3820 | pp = eptr; | pp = eptr; |
| 3821 | ||
| 3822 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 3823 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 3824 | { | { |
| 3825 | register unsigned int d; | register unsigned int d; |
| 3826 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| # | Line 3772 for (;;) | Line 3846 for (;;) |
| 3846 | } | } |
| 3847 | else | else |
| 3848 | #endif | #endif |
| 3849 | /* Not UTF-8 mode */ | /* Not UTF mode */ |
| 3850 | { | { |
| 3851 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3852 | { | { |
| # | Line 3793 for (;;) | Line 3867 for (;;) |
| 3867 | } | } |
| 3868 | } | } |
| 3869 | ||
| 3870 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3871 | } | } |
| 3872 | } | } |
| 3873 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 3805 for (;;) | Line 3879 for (;;) |
| 3879 | case OP_TYPEEXACT: | case OP_TYPEEXACT: |
| 3880 | min = max = GET2(ecode, 1); | min = max = GET2(ecode, 1); |
| 3881 | minimize = TRUE; | minimize = TRUE; |
| 3882 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3883 | goto REPEATTYPE; | goto REPEATTYPE; |
| 3884 | ||
| 3885 | case OP_TYPEUPTO: | case OP_TYPEUPTO: |
| # | Line 3813 for (;;) | Line 3887 for (;;) |
| 3887 | min = 0; | min = 0; |
| 3888 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
| 3889 | minimize = *ecode == OP_TYPEMINUPTO; | minimize = *ecode == OP_TYPEMINUPTO; |
| 3890 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3891 | goto REPEATTYPE; | goto REPEATTYPE; |
| 3892 | ||
| 3893 | case OP_TYPEPOSSTAR: | case OP_TYPEPOSSTAR: |
| # | Line 3841 for (;;) | Line 3915 for (;;) |
| 3915 | possessive = TRUE; | possessive = TRUE; |
| 3916 | min = 0; | min = 0; |
| 3917 | max = GET2(ecode, 1); | max = GET2(ecode, 1); |
| 3918 | ecode += 3; | ecode += 1 + IMM2_SIZE; |
| 3919 | goto REPEATTYPE; | goto REPEATTYPE; |
| 3920 | ||
| 3921 | case OP_TYPESTAR: | case OP_TYPESTAR: |
| # | Line 3887 for (;;) | Line 3961 for (;;) |
| 3961 | switch(prop_type) | switch(prop_type) |
| 3962 | { | { |
| 3963 | case PT_ANY: | case PT_ANY: |
| 3964 | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); | if (prop_fail_result) RRETURN(MATCH_NOMATCH); |
| 3965 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3966 | { | { |
| 3967 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3968 | { | { |
| 3969 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3970 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3971 | } | } |
| 3972 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3973 | } | } |
| # | Line 3906 for (;;) | Line 3980 for (;;) |
| 3980 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3981 | { | { |
| 3982 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 3983 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3984 | } | } |
| 3985 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3986 | chartype = UCD_CHARTYPE(c); | chartype = UCD_CHARTYPE(c); |
| 3987 | if ((chartype == ucp_Lu || | if ((chartype == ucp_Lu || |
| 3988 | chartype == ucp_Ll || | chartype == ucp_Ll || |
| 3989 | chartype == ucp_Lt) == prop_fail_result) | chartype == ucp_Lt) == prop_fail_result) |
| 3990 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 3991 | } | } |
| 3992 | break; | break; |
| 3993 | ||
| # | Line 3923 for (;;) | Line 3997 for (;;) |
| 3997 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 3998 | { | { |
| 3999 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4000 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4001 | } | } |
| 4002 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4003 | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) |
| 4004 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4005 | } | } |
| 4006 | break; | break; |
| 4007 | ||
| # | Line 3937 for (;;) | Line 4011 for (;;) |
| 4011 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4012 | { | { |
| 4013 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4014 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4015 | } | } |
| 4016 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4017 | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) |
| 4018 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4019 | } | } |
| 4020 | break; | break; |
| 4021 | ||
| # | Line 3951 for (;;) | Line 4025 for (;;) |
| 4025 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4026 | { | { |
| 4027 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4028 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4029 | } | } |
| 4030 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4031 | if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) | if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) |
| 4032 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4033 | } | } |
| 4034 | break; | break; |
| 4035 | ||
| # | Line 3966 for (;;) | Line 4040 for (;;) |
| 4040 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4041 | { | { |
| 4042 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4043 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4044 | } | } |
| 4045 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4046 | category = UCD_CATEGORY(c); | category = UCD_CATEGORY(c); |
| 4047 | if ((category == ucp_L || category == ucp_N) == prop_fail_result) | if ((category == ucp_L || category == ucp_N) == prop_fail_result) |
| 4048 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4049 | } | } |
| 4050 | break; | break; |
| 4051 | ||
| # | Line 3981 for (;;) | Line 4055 for (;;) |
| 4055 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4056 | { | { |
| 4057 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4058 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4059 | } | } |
| 4060 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4061 | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
| 4062 | c == CHAR_FF || c == CHAR_CR) | c == CHAR_FF || c == CHAR_CR) |
| 4063 | == prop_fail_result) | == prop_fail_result) |
| 4064 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4065 | } | } |
| 4066 | break; | break; |
| 4067 | ||
| # | Line 3997 for (;;) | Line 4071 for (;;) |
| 4071 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4072 | { | { |
| 4073 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4074 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4075 | } | } |
| 4076 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4077 | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
| 4078 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
| 4079 | == prop_fail_result) | == prop_fail_result) |
| 4080 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4081 | } | } |
| 4082 | break; | break; |
| 4083 | ||
| # | Line 4014 for (;;) | Line 4088 for (;;) |
| 4088 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4089 | { | { |
| 4090 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4091 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4092 | } | } |
| 4093 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4094 | category = UCD_CATEGORY(c); | category = UCD_CATEGORY(c); |
| 4095 | if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE) | if ((category == ucp_L || category == ucp_N || c == CHAR_UNDERSCORE) |
| 4096 | == prop_fail_result) | == prop_fail_result) |
| 4097 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4098 | } | } |
| 4099 | break; | break; |
| 4100 | ||
| # | Line 4041 for (;;) | Line 4115 for (;;) |
| 4115 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4116 | { | { |
| 4117 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4118 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4119 | } | } |
| 4120 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4121 | if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); | if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); |
| 4122 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 4123 | { | { |
| 4124 | int len = 1; | int len = 1; |
| 4125 | if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } | if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
| 4126 | if (UCD_CATEGORY(c) != ucp_M) break; | if (UCD_CATEGORY(c) != ucp_M) break; |
| 4127 | eptr += len; | eptr += len; |
| 4128 | } | } |
| # | Line 4060 for (;;) | Line 4134 for (;;) |
| 4134 | ||
| 4135 | /* Handle all other cases when the coding is UTF-8 */ | /* Handle all other cases when the coding is UTF-8 */ |
| 4136 | ||
| 4137 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 4138 | if (utf8) switch(ctype) | if (utf) switch(ctype) |
| 4139 | { | { |
| 4140 | case OP_ANY: | case OP_ANY: |
| 4141 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| # | Line 4069 for (;;) | Line 4143 for (;;) |
| 4143 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4144 | { | { |
| 4145 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4146 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4147 | } | } |
| 4148 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); |
| 4149 | eptr++; | eptr++; |
| 4150 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); |
| 4151 | } | } |
| 4152 | break; | break; |
| 4153 | ||
| # | Line 4083 for (;;) | Line 4157 for (;;) |
| 4157 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4158 | { | { |
| 4159 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4160 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4161 | } | } |
| 4162 | eptr++; | eptr++; |
| 4163 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); |
| 4164 | } | } |
| 4165 | break; | break; |
| 4166 | ||
| 4167 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 4168 | if (eptr > md->end_subject - min) MRRETURN(MATCH_NOMATCH); | if (eptr > md->end_subject - min) RRETURN(MATCH_NOMATCH); |
| 4169 | eptr += min; | eptr += min; |
| 4170 | break; | break; |
| 4171 | ||
| # | Line 4101 for (;;) | Line 4175 for (;;) |
| 4175 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4176 | { | { |
| 4177 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4178 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4179 | } | } |
| 4180 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4181 | switch(c) | switch(c) |
| 4182 | { | { |
| 4183 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4184 | ||
| 4185 | case 0x000d: | case 0x000d: |
| 4186 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| # | Line 4120 for (;;) | Line 4194 for (;;) |
| 4194 | case 0x0085: | case 0x0085: |
| 4195 | case 0x2028: | case 0x2028: |
| 4196 | case 0x2029: | case 0x2029: |
| 4197 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); |
| 4198 | break; | break; |
| 4199 | } | } |
| 4200 | } | } |
| # | Line 4132 for (;;) | Line 4206 for (;;) |
| 4206 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4207 | { | { |
| 4208 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4209 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4210 | } | } |
| 4211 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4212 | switch(c) | switch(c) |
| # | Line 4157 for (;;) | Line 4231 for (;;) |
| 4231 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 4232 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 4233 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 4234 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4235 | } | } |
| 4236 | } | } |
| 4237 | break; | break; |
| # | Line 4168 for (;;) | Line 4242 for (;;) |
| 4242 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4243 | { | { |
| 4244 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4245 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4246 | } | } |
| 4247 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4248 | switch(c) | switch(c) |
| 4249 | { | { |
| 4250 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4251 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4252 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4253 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 4204 for (;;) | Line 4278 for (;;) |
| 4278 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4279 | { | { |
| 4280 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4281 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4282 | } | } |
| 4283 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4284 | switch(c) | switch(c) |
| # | Line 4217 for (;;) | Line 4291 for (;;) |
| 4291 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4292 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 4293 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 4294 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4295 | } | } |
| 4296 | } | } |
| 4297 | break; | break; |
| # | Line 4228 for (;;) | Line 4302 for (;;) |
| 4302 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4303 | { | { |
| 4304 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4305 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4306 | } | } |
| 4307 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4308 | switch(c) | switch(c) |
| 4309 | { | { |
| 4310 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4311 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4312 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4313 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 4252 for (;;) | Line 4326 for (;;) |
| 4326 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4327 | { | { |
| 4328 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4329 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4330 | } | } |
| 4331 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4332 | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) |
| 4333 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4334 | } | } |
| 4335 | break; | break; |
| 4336 | ||
| # | Line 4266 for (;;) | Line 4340 for (;;) |
| 4340 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4341 | { | { |
| 4342 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4343 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4344 | } | } |
| 4345 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_digit) == 0) |
| 4346 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4347 | eptr++; | |
| 4348 | /* 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 */ |
| 4349 | } | } |
| 4350 | break; | break; |
| # | Line 4280 for (;;) | Line 4355 for (;;) |
| 4355 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4356 | { | { |
| 4357 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4358 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4359 | } | } |
| 4360 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) |
| 4361 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4362 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | eptr++; |
| 4363 | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); | |
| 4364 | } | } |
| 4365 | break; | break; |
| 4366 | ||
| # | Line 4294 for (;;) | Line 4370 for (;;) |
| 4370 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4371 | { | { |
| 4372 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4373 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4374 | } | } |
| 4375 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_space) == 0) |
| 4376 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4377 | eptr++; | |
| 4378 | /* 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 */ |
| 4379 | } | } |
| 4380 | break; | break; |
| # | Line 4308 for (;;) | Line 4385 for (;;) |
| 4385 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4386 | { | { |
| 4387 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4388 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4389 | } | } |
| 4390 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) |
| 4391 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4392 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | eptr++; |
| 4393 | ACROSSCHAR(eptr < md->end_subject, *eptr, eptr++); | |
| 4394 | } | } |
| 4395 | break; | break; |
| 4396 | ||
| # | Line 4322 for (;;) | Line 4400 for (;;) |
| 4400 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4401 | { | { |
| 4402 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4403 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4404 | } | } |
| 4405 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | if (*eptr >= 128 || (md->ctypes[*eptr] & ctype_word) == 0) |
| 4406 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4407 | eptr++; | |
| 4408 | /* 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 */ |
| 4409 | } | } |
| 4410 | break; | break; |
| # | Line 4335 for (;;) | Line 4414 for (;;) |
| 4414 | } /* End switch(ctype) */ | } /* End switch(ctype) */ |
| 4415 | ||
| 4416 | else | else |
| 4417 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF */ |
| 4418 | ||
| 4419 | /* 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 |
| 4420 | than OP_PROP and OP_NOTPROP. */ | than OP_PROP and OP_NOTPROP. */ |
| # | Line 4348 for (;;) | Line 4427 for (;;) |
| 4427 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4428 | { | { |
| 4429 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4430 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4431 | } | } |
| 4432 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); |
| 4433 | eptr++; | eptr++; |
| 4434 | } | } |
| 4435 | break; | break; |
| # | Line 4359 for (;;) | Line 4438 for (;;) |
| 4438 | if (eptr > md->end_subject - min) | if (eptr > md->end_subject - min) |
| 4439 | { | { |
| 4440 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4441 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4442 | } | } |
| 4443 | eptr += min; | eptr += min; |
| 4444 | break; | break; |
| # | Line 4368 for (;;) | Line 4447 for (;;) |
| 4447 | if (eptr > md->end_subject - min) | if (eptr > md->end_subject - min) |
| 4448 | { | { |
| 4449 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4450 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4451 | } | } |
| 4452 | eptr += min; | eptr += min; |
| 4453 | break; | break; |
| # | Line 4379 for (;;) | Line 4458 for (;;) |
| 4458 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4459 | { | { |
| 4460 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4461 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4462 | } | } |
| 4463 | switch(*eptr++) | switch(*eptr++) |
| 4464 | { | { |
| 4465 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4466 | ||
| 4467 | case 0x000d: | case 0x000d: |
| 4468 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| # | Line 4395 for (;;) | Line 4474 for (;;) |
| 4474 | case 0x000b: | case 0x000b: |
| 4475 | case 0x000c: | case 0x000c: |
| 4476 | case 0x0085: | case 0x0085: |
| 4477 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
| 4478 | case 0x2028: | |
| 4479 | case 0x2029: | |
| 4480 | #endif | |
| 4481 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | |
| 4482 | break; | break; |
| 4483 | } | } |
| 4484 | } | } |
| # | Line 4407 for (;;) | Line 4490 for (;;) |
| 4490 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4491 | { | { |
| 4492 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4493 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4494 | } | } |
| 4495 | switch(*eptr++) | switch(*eptr++) |
| 4496 | { | { |
| # | Line 4415 for (;;) | Line 4498 for (;;) |
| 4498 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4499 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4500 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4501 | MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
| 4502 | case 0x1680: /* OGHAM SPACE MARK */ | |
| 4503 | case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ | |
| 4504 | case 0x2000: /* EN QUAD */ | |
| 4505 | case 0x2001: /* EM QUAD */ | |
| 4506 | case 0x2002: /* EN SPACE */ | |
| 4507 | case 0x2003: /* EM SPACE */ | |
| 4508 | case 0x2004: /* THREE-PER-EM SPACE */ | |
| 4509 | case 0x2005: /* FOUR-PER-EM SPACE */ | |
| 4510 | case 0x2006: /* SIX-PER-EM SPACE */ | |
| 4511 | case 0x2007: /* FIGURE SPACE */ | |
| 4512 | case 0x2008: /* PUNCTUATION SPACE */ | |
| 4513 | case 0x2009: /* THIN SPACE */ | |
| 4514 | case 0x200A: /* HAIR SPACE */ | |
| 4515 | case 0x202f: /* NARROW NO-BREAK SPACE */ | |
| 4516 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | |
| 4517 | case 0x3000: /* IDEOGRAPHIC SPACE */ | |
| 4518 | #endif | |
| 4519 | RRETURN(MATCH_NOMATCH); | |
| 4520 | } | } |
| 4521 | } | } |
| 4522 | break; | break; |
| # | Line 4426 for (;;) | Line 4527 for (;;) |
| 4527 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4528 | { | { |
| 4529 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4530 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4531 | } | } |
| 4532 | switch(*eptr++) | switch(*eptr++) |
| 4533 | { | { |
| 4534 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4535 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4536 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4537 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4538 | #ifdef COMPILE_PCRE16 | |
| 4539 | case 0x1680: /* OGHAM SPACE MARK */ | |
| 4540 | case 0x180e: /* MONGOLIAN VOWEL SEPARATOR */ | |
| 4541 | case 0x2000: /* EN QUAD */ | |
| 4542 | case 0x2001: /* EM QUAD */ | |
| 4543 | case 0x2002: /* EN SPACE */ | |
| 4544 | case 0x2003: /* EM SPACE */ | |
| 4545 | case 0x2004: /* THREE-PER-EM SPACE */ | |
| 4546 | case 0x2005: /* FOUR-PER-EM SPACE */ | |
| 4547 | case 0x2006: /* SIX-PER-EM SPACE */ | |
| 4548 | case 0x2007: /* FIGURE SPACE */ | |
| 4549 | case 0x2008: /* PUNCTUATION SPACE */ | |
| 4550 | case 0x2009: /* THIN SPACE */ | |
| 4551 | case 0x200A: /* HAIR SPACE */ | |
| 4552 | case 0x202f: /* NARROW NO-BREAK SPACE */ | |
| 4553 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | |
| 4554 | case 0x3000: /* IDEOGRAPHIC SPACE */ | |
| 4555 | #endif | |
| 4556 | break; | break; |
| 4557 | } | } |
| 4558 | } | } |
| # | Line 4445 for (;;) | Line 4564 for (;;) |
| 4564 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4565 | { | { |
| 4566 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4567 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4568 | } | } |
| 4569 | switch(*eptr++) | switch(*eptr++) |
| 4570 | { | { |
| # | Line 4455 for (;;) | Line 4574 for (;;) |
| 4574 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4575 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4576 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4577 | MRRETURN(MATCH_NOMATCH); | #ifdef COMPILE_PCRE16 |
| 4578 | case 0x2028: /* LINE SEPARATOR */ | |
| 4579 | case 0x2029: /* PARAGRAPH SEPARATOR */ | |
| 4580 | #endif | |
| 4581 | RRETURN(MATCH_NOMATCH); | |
| 4582 | } | } |
| 4583 | } | } |
| 4584 | break; | break; |
| # | Line 4466 for (;;) | Line 4589 for (;;) |
| 4589 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4590 | { | { |
| 4591 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4592 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4593 | } | } |
| 4594 | switch(*eptr++) | switch(*eptr++) |
| 4595 | { | { |
| 4596 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4597 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4598 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4599 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4600 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4601 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4602 | #ifdef COMPILE_PCRE16 | |
| 4603 | case 0x2028: /* LINE SEPARATOR */ | |
| 4604 | case 0x2029: /* PARAGRAPH SEPARATOR */ | |
| 4605 | #endif | |
| 4606 | break; | break; |
| 4607 | } | } |
| 4608 | } | } |
| # | Line 4487 for (;;) | Line 4614 for (;;) |
| 4614 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4615 | { | { |
| 4616 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4617 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4618 | } | } |
| 4619 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_digit) != 0) |
| 4620 | RRETURN(MATCH_NOMATCH); | |
| 4621 | eptr++; | |
| 4622 | } | } |
| 4623 | break; | break; |
| 4624 | ||
| # | Line 4499 for (;;) | Line 4628 for (;;) |
| 4628 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4629 | { | { |
| 4630 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4631 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4632 | } | } |
| 4633 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_digit) == 0) |
| 4634 | RRETURN(MATCH_NOMATCH); | |
| 4635 | eptr++; | |
| 4636 | } | } |
| 4637 | break; | break; |
| 4638 | ||
| # | Line 4511 for (;;) | Line 4642 for (;;) |
| 4642 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4643 | { | { |
| 4644 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4645 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4646 | } | } |
| 4647 | if ((md->ctypes[*eptr++] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_space) != 0) |
| 4648 | RRETURN(MATCH_NOMATCH); | |
| 4649 | eptr++; | |
| 4650 | } | } |
| 4651 | break; | break; |
| 4652 | ||
| # | Line 4523 for (;;) | Line 4656 for (;;) |
| 4656 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4657 | { | { |
| 4658 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4659 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4660 | } | } |
| 4661 | if ((md->ctypes[*eptr++] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_space) == 0) |
| 4662 | RRETURN(MATCH_NOMATCH); | |
| 4663 | eptr++; | |
| 4664 | } | } |
| 4665 | break; | break; |
| 4666 | ||
| # | Line 4535 for (;;) | Line 4670 for (;;) |
| 4670 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4671 | { | { |
| 4672 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4673 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4674 | } | } |
| 4675 | if ((md->ctypes[*eptr++] & ctype_word) != 0) | if (MAX_255(*eptr) && (md->ctypes[*eptr] & ctype_word) != 0) |
| 4676 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4677 | eptr++; | |
| 4678 | } | } |
| 4679 | break; | break; |
| 4680 | ||
| # | Line 4548 for (;;) | Line 4684 for (;;) |
| 4684 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4685 | { | { |
| 4686 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4687 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4688 | } | } |
| 4689 | if ((md->ctypes[*eptr++] & ctype_word) == 0) | if (!MAX_255(*eptr) || (md->ctypes[*eptr] & ctype_word) == 0) |
| 4690 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4691 | eptr++; | |
| 4692 | } | } |
| 4693 | break; | break; |
| 4694 | ||
| # | Line 4580 for (;;) | Line 4717 for (;;) |
| 4717 | { | { |
| 4718 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM36); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM36); |
| 4719 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4720 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4721 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4722 | { | { |
| 4723 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4724 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4725 | } | } |
| 4726 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4727 | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); | if (prop_fail_result) RRETURN(MATCH_NOMATCH); |
| 4728 | } | } |
| 4729 | /* Control never gets here */ | /* Control never gets here */ |
| 4730 | ||
| # | Line 4597 for (;;) | Line 4734 for (;;) |
| 4734 | int chartype; | int chartype; |
| 4735 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM37); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM37); |
| 4736 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4737 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4738 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4739 | { | { |
| 4740 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4741 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4742 | } | } |
| 4743 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4744 | chartype = UCD_CHARTYPE(c); | chartype = UCD_CHARTYPE(c); |
| 4745 | if ((chartype == ucp_Lu || | if ((chartype == ucp_Lu || |
| 4746 | chartype == ucp_Ll || | chartype == ucp_Ll || |
| 4747 | chartype == ucp_Lt) == prop_fail_result) | chartype == ucp_Lt) == prop_fail_result) |
| 4748 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4749 | } | } |
| 4750 | /* Control never gets here */ | /* Control never gets here */ |
| 4751 | ||
| # | Line 4617 for (;;) | Line 4754 for (;;) |
| 4754 | { | { |
| 4755 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM38); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM38); |
| 4756 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4757 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4758 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4759 | { | { |
| 4760 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4761 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4762 | } | } |
| 4763 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4764 | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) | if ((UCD_CATEGORY(c) == prop_value) == prop_fail_result) |
| 4765 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4766 | } | } |
| 4767 | /* Control never gets here */ | /* Control never gets here */ |
| 4768 | ||
| # | Line 4634 for (;;) | Line 4771 for (;;) |
| 4771 | { | { |
| 4772 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM39); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM39); |
| 4773 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4774 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4775 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4776 | { | { |
| 4777 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4778 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4779 | } | } |
| 4780 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4781 | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) | if ((UCD_CHARTYPE(c) == prop_value) == prop_fail_result) |
| 4782 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4783 | } | } |
| 4784 | /* Control never gets here */ | /* Control never gets here */ |
| 4785 | ||
| # | Line 4651 for (;;) | Line 4788 for (;;) |
| 4788 | { | { |
| 4789 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM40); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM40); |
| 4790 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4791 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4792 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4793 | { | { |
| 4794 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4795 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4796 | } | } |
| 4797 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4798 | if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) | if ((UCD_SCRIPT(c) == prop_value) == prop_fail_result) |
| 4799 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4800 | } | } |
| 4801 | /* Control never gets here */ | /* Control never gets here */ |
| 4802 | ||
| # | Line 4669 for (;;) | Line 4806 for (;;) |
| 4806 | int category; | int category; |
| 4807 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM59); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM59); |
| 4808 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4809 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4810 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4811 | { | { |
| 4812 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4813 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4814 | } | } |
| 4815 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4816 | category = UCD_CATEGORY(c); | category = UCD_CATEGORY(c); |
| 4817 | if ((category == ucp_L || category == ucp_N) == prop_fail_result) | if ((category == ucp_L || category == ucp_N) == prop_fail_result) |
| 4818 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4819 | } | } |
| 4820 | /* Control never gets here */ | /* Control never gets here */ |
| 4821 | ||
| # | Line 4687 for (;;) | Line 4824 for (;;) |
| 4824 | { | { |
| 4825 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM60); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM60); |
| 4826 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4827 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4828 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4829 | { | { |
| 4830 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4831 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4832 | } | } |
| 4833 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4834 | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
| 4835 | c == CHAR_FF || c == CHAR_CR) | c == CHAR_FF || c == CHAR_CR) |
| 4836 | == prop_fail_result) | == prop_fail_result) |
| 4837 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4838 | } | } |
| 4839 | /* Control never gets here */ | /* Control never gets here */ |
| 4840 | ||
| # | Line 4706 for (;;) | Line 4843 for (;;) |
| 4843 | { | { |
| 4844 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM61); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM61); |
| 4845 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4846 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4847 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4848 | { | { |
| 4849 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4850 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4851 | } | } |
| 4852 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4853 | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || | if ((UCD_CATEGORY(c) == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
| 4854 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) |
| 4855 | == prop_fail_result) | == prop_fail_result) |
| 4856 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4857 | } | } |
| 4858 | /* Control never gets here */ | /* Control never gets here */ |
| 4859 | ||
| # | Line 4726 for (;;) | Line 4863 for (;;) |
| 4863 | int category; | int category; |
| 4864 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM62); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM62); |
| 4865 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4866 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4867 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4868 | { | { |
| 4869 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4870 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4871 | } | } |
| 4872 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4873 | category = UCD_CATEGORY(c); | category = UCD_CATEGORY(c); |
| # | Line 4738 for (;;) | Line 4875 for (;;) |
| 4875 | category == ucp_N || | category == ucp_N || |
| 4876 | c == CHAR_UNDERSCORE) | c == CHAR_UNDERSCORE) |
| 4877 | == prop_fail_result) | == prop_fail_result) |
| 4878 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4879 | } | } |
| 4880 | /* Control never gets here */ | /* Control never gets here */ |
| 4881 | ||
| # | Line 4758 for (;;) | Line 4895 for (;;) |
| 4895 | { | { |
| 4896 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM41); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM41); |
| 4897 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4898 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4899 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4900 | { | { |
| 4901 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4902 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4903 | } | } |
| 4904 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4905 | if (UCD_CATEGORY(c) == ucp_M) MRRETURN(MATCH_NOMATCH); | if (UCD_CATEGORY(c) == ucp_M) RRETURN(MATCH_NOMATCH); |
| 4906 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 4907 | { | { |
| 4908 | int len = 1; | int len = 1; |
| 4909 | if (!utf8) c = *eptr; else { GETCHARLEN(c, eptr, len); } | if (!utf) c = *eptr; else { GETCHARLEN(c, eptr, len); } |
| 4910 | if (UCD_CATEGORY(c) != ucp_M) break; | if (UCD_CATEGORY(c) != ucp_M) break; |
| 4911 | eptr += len; | eptr += len; |
| 4912 | } | } |
| # | Line 4778 for (;;) | Line 4915 for (;;) |
| 4915 | else | else |
| 4916 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 4917 | ||
| 4918 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF |
| 4919 | /* UTF-8 mode */ | if (utf) |
| if (utf8) | ||
| 4920 | { | { |
| 4921 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 4922 | { | { |
| 4923 | RMATCH(eptr, ecode, offset_top, md, eptrb, RM42); | RMATCH(eptr, ecode, offset_top, md, eptrb, RM42); |
| 4924 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4925 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | if (fi >= max) RRETURN(MATCH_NOMATCH); |
| 4926 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 4927 | { | { |
| 4928 | SCHECK_PARTIAL(); | SCHECK_PARTIAL(); |
| 4929 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4930 | } | } |
| 4931 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | if (ctype == OP_ANY && IS_NEWLINE(eptr)) |
| 4932 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4933 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4934 | switch(ctype) | switch(ctype) |
| 4935 | { | { |
| # | Line 4805 for (;;) | Line 4941 for (;;) |
| 4941 | case OP_ANYNL: | case OP_ANYNL: |
| 4942 | switch(c) | switch(c) |
| 4943 | { | { |
| 4944 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4945 | case 0x000d: | case 0x000d: |
| 4946 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4947 | break; | break; |
| # | Line 4817 for (;;) | Line 4953 for (;;) |
| 4953 | case 0x0085: | case 0x0085: |
| 4954 | case 0x2028: | case 0x2028: |
| 4955 | case 0x2029: | case 0x2029: |
| 4956 | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); |
| 4957 | break; | break; |
| 4958 | } | } |
| 4959 | break; | break; |
| # | Line 4845 for (;;) | Line 4981 for (;;) |
| 4981 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 4982 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 4983 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 4984 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 4985 | } | } |
| 4986 | break; | break; |
| 4987 | ||
| 4988 | case OP_HSPACE: | case OP_HSPACE: |
| 4989 | switch(c) | switch(c) |
| 4990 | { | { |
| 4991 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 4992 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4993 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4994 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 4887 for (;;) | Line 5023 for (;;) |
| 5023 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 5024 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 5025 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 5026 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5027 | } | } |
| 5028 | break; | break; |
| 5029 | ||
| 5030 | case OP_VSPACE: | case OP_VSPACE: |
| 5031 | switch(c) | switch(c) |
| 5032 | { | { |
| 5033 | default: MRRETURN(MATCH_NOMATCH); | default: RRETURN(MATCH_NOMATCH); |
| 5034 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 5035 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 5036 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 4908 for (;;) | Line 5044 for (;;) |
| 5044 | ||
| 5045 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 5046 | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) |
| 5047 | MRRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 5048 | break; | break; |
| 5049 | ||
| 5050 | case OP_DIGIT: | case OP_DIGIT: |
| 5051 | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) |
| 5052 |