Parent Directory
|
Revision Log
|
Patch
| revision 406 by ph10, Mon Mar 23 12:05:43 2009 UTC | revision 595 by ph10, Mon May 2 10:33:29 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) 1997-2009 University of Cambridge | Copyright (c) 1997-2010 University of Cambridge |
| 10 | ||
| 11 | ----------------------------------------------------------------------------- | ----------------------------------------------------------------------------- |
| 12 | Redistribution and use in source and binary forms, with or without | Redistribution and use in source and binary forms, with or without |
| # | Line 71 defined PCRE_ERROR_xxx codes, which are | Line 71 defined PCRE_ERROR_xxx codes, which are |
| 71 | /* Special internal returns from the match() function. Make them sufficiently | /* Special internal returns from the match() function. Make them sufficiently |
| 72 | negative to avoid the external error codes. */ | negative to avoid the external error codes. */ |
| 73 | ||
| 74 | #define MATCH_COMMIT (-999) | #define MATCH_ACCEPT (-999) |
| 75 | #define MATCH_PRUNE (-998) | #define MATCH_COMMIT (-998) |
| 76 | #define MATCH_SKIP (-997) | #define MATCH_PRUNE (-997) |
| 77 | #define MATCH_THEN (-996) | #define MATCH_SKIP (-996) |
| 78 | #define MATCH_SKIP_ARG (-995) | |
| 79 | #define MATCH_THEN (-994) | |
| 80 | ||
| 81 | /* This is a convenience macro for code that occurs many times. */ | |
| 82 | ||
| 83 | #define MRRETURN(ra) \ | |
| 84 | { \ | |
| 85 | md->mark = markptr; \ | |
| 86 | RRETURN(ra); \ | |
| 87 | } | |
| 88 | ||
| 89 | /* Maximum number of ints of offset to save on the stack for recursive calls. | /* Maximum number of ints of offset to save on the stack for recursive calls. |
| 90 | If the offset vector is bigger, malloc is used. This should be a multiple of 3, | If the offset vector is bigger, malloc is used. This should be a multiple of 3, |
| # | Line 89 static const char rep_max[] = { 0, 0, 0, | Line 99 static const char rep_max[] = { 0, 0, 0, |
| 99 | ||
| 100 | ||
| 101 | ||
| 102 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 103 | /************************************************* | /************************************************* |
| 104 | * Debugging function to print chars * | * Debugging function to print chars * |
| 105 | *************************************************/ | *************************************************/ |
| # | Line 122 while (length-- > 0) | Line 132 while (length-- > 0) |
| 132 | * Match a back-reference * | * Match a back-reference * |
| 133 | *************************************************/ | *************************************************/ |
| 134 | ||
| 135 | /* If a back reference hasn't been set, the length that is passed is greater | /* Normally, if a back reference hasn't been set, the length that is passed is |
| 136 | than the number of characters left in the string, so the match fails. | negative, so the match always fails. However, in JavaScript compatibility mode, |
| 137 | the length passed is zero. Note that in caseless UTF-8 mode, the number of | |
| 138 | subject bytes matched may be different to the number of reference bytes. | |
| 139 | ||
| 140 | Arguments: | Arguments: |
| 141 | offset index into the offset vector | offset index into the offset vector |
| 142 | eptr points into the subject | eptr pointer into the subject |
| 143 | length length to be matched | length length of reference to be matched (number of bytes) |
| 144 | md points to match data block | md points to match data block |
| 145 | ims the ims flags | ims the ims flags |
| 146 | ||
| 147 | Returns: TRUE if matched | Returns: < 0 if not matched, otherwise the number of subject bytes matched |
| 148 | */ | */ |
| 149 | ||
| 150 | static BOOL | static int |
| 151 | match_ref(int offset, register USPTR eptr, int length, match_data *md, | match_ref(int offset, register USPTR eptr, int length, match_data *md, |
| 152 | unsigned long int ims) | unsigned long int ims) |
| 153 | { | { |
| 154 | USPTR p = md->start_subject + md->offset_vector[offset]; | USPTR eptr_start = eptr; |
| 155 | register USPTR p = md->start_subject + md->offset_vector[offset]; | |
| 156 | ||
| 157 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 158 | if (eptr >= md->end_subject) | if (eptr >= md->end_subject) |
| 159 | printf("matching subject <null>"); | printf("matching subject <null>"); |
| 160 | else | else |
| # | Line 154 pchars(p, length, FALSE, md); | Line 167 pchars(p, length, FALSE, md); |
| 167 | printf("\n"); | printf("\n"); |
| 168 | #endif | #endif |
| 169 | ||
| 170 | /* Always fail if not enough characters left */ | /* Always fail if reference not set (and not JavaScript compatible). */ |
| 171 | ||
| 172 | if (length > md->end_subject - eptr) return FALSE; | if (length < 0) return -1; |
| 173 | ||
| 174 | /* Separate the caseless case for speed. In UTF-8 mode we can only do this | /* Separate the caseless case for speed. In UTF-8 mode we can only do this |
| 175 | properly if Unicode properties are supported. Otherwise, we can check only | properly if Unicode properties are supported. Otherwise, we can check only |
| # | Line 168 if ((ims & PCRE_CASELESS) != 0) | Line 181 if ((ims & PCRE_CASELESS) != 0) |
| 181 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 182 | if (md->utf8) | if (md->utf8) |
| 183 | { | { |
| 184 | USPTR endptr = eptr + length; | /* Match characters up to the end of the reference. NOTE: the number of |
| 185 | while (eptr < endptr) | bytes matched may differ, because there are some characters whose upper and |
| 186 | lower case versions code as different numbers of bytes. For example, U+023A | |
| 187 | (2 bytes in UTF-8) is the upper case version of U+2C65 (3 bytes in UTF-8); | |
| 188 | a sequence of 3 of the former uses 6 bytes, as does a sequence of two of | |
| 189 | the latter. It is important, therefore, to check the length along the | |
| 190 | reference, not along the subject (earlier code did this wrong). */ | |
| 191 | ||
| 192 | USPTR endptr = p + length; | |
| 193 | while (p < endptr) | |
| 194 | { | { |
| 195 | int c, d; | int c, d; |
| 196 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 197 | GETCHARINC(d, p); | GETCHARINC(d, p); |
| 198 | if (c != d && c != UCD_OTHERCASE(d)) return FALSE; | if (c != d && c != UCD_OTHERCASE(d)) return -1; |
| 199 | } | } |
| 200 | } | } |
| 201 | else | else |
| # | Line 185 if ((ims & PCRE_CASELESS) != 0) | Line 206 if ((ims & PCRE_CASELESS) != 0) |
| 206 | is no UCP support. */ | is no UCP support. */ |
| 207 | ||
| 208 | while (length-- > 0) | while (length-- > 0) |
| 209 | { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; } | { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; } |
| 210 | } | } |
| 211 | ||
| 212 | /* In the caseful case, we can just compare the bytes, whether or not we | /* In the caseful case, we can just compare the bytes, whether or not we |
| 213 | are in UTF-8 mode. */ | are in UTF-8 mode. */ |
| 214 | ||
| 215 | else | else |
| 216 | { while (length-- > 0) if (*p++ != *eptr++) return FALSE; } | { while (length-- > 0) if (*p++ != *eptr++) return -1; } |
| 217 | ||
| 218 | return TRUE; | return eptr - eptr_start; |
| 219 | } | } |
| 220 | ||
| 221 | ||
| # | Line 245 enum { RM1=1, RM2, RM3, RM4, RM5, RM | Line 266 enum { RM1=1, RM2, RM3, RM4, RM5, RM |
| 266 | RM21, RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30, | RM21, RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30, |
| 267 | RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, | RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, |
| 268 | RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, | RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, |
| 269 | RM51, RM52, RM53, RM54 }; | RM51, RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60, |
| 270 | RM61, RM62 }; | |
| 271 | ||
| 272 | /* These versions of the macros use the stack, as normal. There are debugging | /* These versions of the macros use the stack, as normal. There are debugging |
| 273 | versions and production versions. Note that the "rw" argument of RMATCH isn't | versions and production versions. Note that the "rw" argument of RMATCH isn't |
| 274 | actuall used in this definition. */ | actually used in this definition. */ |
| 275 | ||
| 276 | #ifndef NO_RECURSE | #ifndef NO_RECURSE |
| 277 | #define REGISTER register | #define REGISTER register |
| 278 | ||
| 279 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 280 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ |
| 281 | { \ | { \ |
| 282 | printf("match() called in line %d\n", __LINE__); \ | printf("match() called in line %d\n", __LINE__); \ |
| 283 | rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \ | rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1); \ |
| 284 | printf("to line %d\n", __LINE__); \ | printf("to line %d\n", __LINE__); \ |
| 285 | } | } |
| 286 | #define RRETURN(ra) \ | #define RRETURN(ra) \ |
| # | Line 268 actuall used in this definition. */ | Line 290 actuall used in this definition. */ |
| 290 | } | } |
| 291 | #else | #else |
| 292 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ |
| 293 | rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1) | rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1) |
| 294 | #define RRETURN(ra) return ra | #define RRETURN(ra) return ra |
| 295 | #endif | #endif |
| 296 | ||
| # | Line 283 argument of match(), which never changes | Line 305 argument of match(), which never changes |
| 305 | ||
| 306 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\ |
| 307 | {\ | {\ |
| 308 | heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\ | heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\ |
| 309 | if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\ | |
| 310 | frame->Xwhere = rw; \ | frame->Xwhere = rw; \ |
| 311 | newframe->Xeptr = ra;\ | newframe->Xeptr = ra;\ |
| 312 | newframe->Xecode = rb;\ | newframe->Xecode = rb;\ |
| 313 | newframe->Xmstart = mstart;\ | newframe->Xmstart = mstart;\ |
| 314 | newframe->Xmarkptr = markptr;\ | |
| 315 | newframe->Xoffset_top = rc;\ | newframe->Xoffset_top = rc;\ |
| 316 | newframe->Xims = re;\ | newframe->Xims = re;\ |
| 317 | newframe->Xeptrb = rf;\ | newframe->Xeptrb = rf;\ |
| # | Line 303 argument of match(), which never changes | Line 327 argument of match(), which never changes |
| 327 | ||
| 328 | #define RRETURN(ra)\ | #define RRETURN(ra)\ |
| 329 | {\ | {\ |
| 330 | heapframe *newframe = frame;\ | heapframe *oldframe = frame;\ |
| 331 | frame = newframe->Xprevframe;\ | frame = oldframe->Xprevframe;\ |
| 332 | (pcre_stack_free)(newframe);\ | (pcre_stack_free)(oldframe);\ |
| 333 | if (frame != NULL)\ | if (frame != NULL)\ |
| 334 | {\ | {\ |
| 335 | rrc = ra;\ | rrc = ra;\ |
| # | Line 322 typedef struct heapframe { | Line 346 typedef struct heapframe { |
| 346 | ||
| 347 | /* Function arguments that may change */ | /* Function arguments that may change */ |
| 348 | ||
| 349 | const uschar *Xeptr; | USPTR Xeptr; |
| 350 | const uschar *Xecode; | const uschar *Xecode; |
| 351 | const uschar *Xmstart; | USPTR Xmstart; |
| 352 | USPTR Xmarkptr; | |
| 353 | int Xoffset_top; | int Xoffset_top; |
| 354 | long int Xims; | long int Xims; |
| 355 | eptrblock *Xeptrb; | eptrblock *Xeptrb; |
| # | Line 333 typedef struct heapframe { | Line 358 typedef struct heapframe { |
| 358 | ||
| 359 | /* Function local variables */ | /* Function local variables */ |
| 360 | ||
| 361 | const uschar *Xcallpat; | USPTR Xcallpat; |
| 362 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 363 | const uschar *Xcharptr; | USPTR Xcharptr; |
| 364 | #endif | #endif |
| 365 | const uschar *Xdata; | USPTR Xdata; |
| 366 | const uschar *Xnext; | USPTR Xnext; |
| 367 | const uschar *Xpp; | USPTR Xpp; |
| 368 | const uschar *Xprev; | USPTR Xprev; |
| 369 | const uschar *Xsaved_eptr; | USPTR Xsaved_eptr; |
| 370 | ||
| 371 | recursion_info Xnew_recursive; | recursion_info Xnew_recursive; |
| 372 | ||
| # | Line 398 typedef struct heapframe { | Line 423 typedef struct heapframe { |
| 423 | ||
| 424 | /* This function is called recursively in many circumstances. Whenever it | /* This function is called recursively in many circumstances. Whenever it |
| 425 | returns a negative (error) response, the outer incarnation must also return the | returns a negative (error) response, the outer incarnation must also return the |
| 426 | same response. | same response. */ |
| 427 | ||
| 428 | /* These macros pack up tests that are used for partial matching, and which | |
| 429 | appears several times in the code. We set the "hit end" flag if the pointer is | |
| 430 | at the end of the subject and also past the start of the subject (i.e. | |
| 431 | something has been matched). For hard partial matching, we then return | |
| 432 | immediately. The second one is used when we already know we are past the end of | |
| 433 | the subject. */ | |
| 434 | ||
| 435 | #define CHECK_PARTIAL()\ | |
| 436 | if (md->partial != 0 && eptr >= md->end_subject && \ | |
| 437 | eptr > md->start_used_ptr) \ | |
| 438 | { \ | |
| 439 | md->hitend = TRUE; \ | |
| 440 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \ | |
| 441 | } | |
| 442 | ||
| 443 | #define SCHECK_PARTIAL()\ | |
| 444 | if (md->partial != 0 && eptr > md->start_used_ptr) \ | |
| 445 | { \ | |
| 446 | md->hitend = TRUE; \ | |
| 447 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \ | |
| 448 | } | |
| 449 | ||
| 450 | ||
| 451 | Performance note: It might be tempting to extract commonly used fields from the | /* Performance note: It might be tempting to extract commonly used fields from |
| 452 | md structure (e.g. utf8, end_subject) into individual variables to improve | the md structure (e.g. utf8, end_subject) into individual variables to improve |
| 453 | 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 |
| 454 | made performance worse. | made performance worse. |
| 455 | ||
| # | Line 410 Arguments: | Line 458 Arguments: |
| 458 | ecode pointer to current position in compiled code | ecode pointer to current position in compiled code |
| 459 | mstart pointer to the current match start position (can be modified | mstart pointer to the current match start position (can be modified |
| 460 | by encountering \K) | by encountering \K) |
| 461 | markptr pointer to the most recent MARK name, or NULL | |
| 462 | offset_top current top pointer | offset_top current top pointer |
| 463 | md pointer to "static" info for the match | md pointer to "static" info for the match |
| 464 | ims current /i, /m, and /s options | ims current /i, /m, and /s options |
| # | Line 423 Arguments: | Line 472 Arguments: |
| 472 | ||
| 473 | Returns: MATCH_MATCH if matched ) these values are >= 0 | Returns: MATCH_MATCH if matched ) these values are >= 0 |
| 474 | MATCH_NOMATCH if failed to match ) | MATCH_NOMATCH if failed to match ) |
| 475 | a negative MATCH_xxx value for PRUNE, SKIP, etc | |
| 476 | a negative PCRE_ERROR_xxx value if aborted by an error condition | a negative PCRE_ERROR_xxx value if aborted by an error condition |
| 477 | (e.g. stopped by repeated call or recursion limit) | (e.g. stopped by repeated call or recursion limit) |
| 478 | */ | */ |
| 479 | ||
| 480 | static int | static int |
| 481 | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart, | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart, |
| 482 | int offset_top, match_data *md, unsigned long int ims, eptrblock *eptrb, | const uschar *markptr, int offset_top, match_data *md, unsigned long int ims, |
| 483 | int flags, unsigned int rdepth) | eptrblock *eptrb, int flags, unsigned int rdepth) |
| 484 | { | { |
| 485 | /* 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, |
| 486 | so they can be ordinary variables in all cases. Mark some of them with | so they can be ordinary variables in all cases. Mark some of them with |
| # | Line 450 heap storage. Set up the top-level frame | Line 500 heap storage. Set up the top-level frame |
| 500 | heap whenever RMATCH() does a "recursion". See the macro definitions above. */ | heap whenever RMATCH() does a "recursion". See the macro definitions above. */ |
| 501 | ||
| 502 | #ifdef NO_RECURSE | #ifdef NO_RECURSE |
| 503 | heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe)); | heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe)); |
| 504 | if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY); | |
| 505 | frame->Xprevframe = NULL; /* Marks the top level */ | frame->Xprevframe = NULL; /* Marks the top level */ |
| 506 | ||
| 507 | /* Copy in the original argument variables */ | /* Copy in the original argument variables */ |
| # | Line 458 frame->Xprevframe = NULL; /* | Line 509 frame->Xprevframe = NULL; /* |
| 509 | frame->Xeptr = eptr; | frame->Xeptr = eptr; |
| 510 | frame->Xecode = ecode; | frame->Xecode = ecode; |
| 511 | frame->Xmstart = mstart; | frame->Xmstart = mstart; |
| 512 | frame->Xmarkptr = markptr; | |
| 513 | frame->Xoffset_top = offset_top; | frame->Xoffset_top = offset_top; |
| 514 | frame->Xims = ims; | frame->Xims = ims; |
| 515 | frame->Xeptrb = eptrb; | frame->Xeptrb = eptrb; |
| # | Line 473 HEAP_RECURSE: | Line 525 HEAP_RECURSE: |
| 525 | #define eptr frame->Xeptr | #define eptr frame->Xeptr |
| 526 | #define ecode frame->Xecode | #define ecode frame->Xecode |
| 527 | #define mstart frame->Xmstart | #define mstart frame->Xmstart |
| 528 | #define markptr frame->Xmarkptr | |
| 529 | #define offset_top frame->Xoffset_top | #define offset_top frame->Xoffset_top |
| 530 | #define ims frame->Xims | #define ims frame->Xims |
| 531 | #define eptrb frame->Xeptrb | #define eptrb frame->Xeptrb |
| # | Line 600 TAIL_RECURSE: | Line 653 TAIL_RECURSE: |
| 653 | /* OK, now we can get on with the real code of the function. Recursive calls | /* OK, now we can get on with the real code of the function. Recursive calls |
| 654 | are specified by the macro RMATCH and RRETURN is used to return. When | are specified by the macro RMATCH and RRETURN is used to return. When |
| 655 | NO_RECURSE is *not* defined, these just turn into a recursive call to match() | NO_RECURSE is *not* defined, these just turn into a recursive call to match() |
| 656 | and a "return", respectively (possibly with some debugging if DEBUG is | and a "return", respectively (possibly with some debugging if PCRE_DEBUG is |
| 657 | defined). However, RMATCH isn't like a function call because it's quite a | defined). However, RMATCH isn't like a function call because it's quite a |
| 658 | 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, |
| 659 | however, impact performance when true recursion is being used. */ | however, impact performance when true recursion is being used. */ |
| # | Line 642 for (;;) | Line 695 for (;;) |
| 695 | minimize = possessive = FALSE; | minimize = possessive = FALSE; |
| 696 | op = *ecode; | op = *ecode; |
| 697 | ||
| /* For partial matching, remember if we ever hit the end of the subject after | ||
| matching at least one subject character. */ | ||
| if (md->partial && | ||
| eptr >= md->end_subject && | ||
| eptr > mstart) | ||
| md->hitend = TRUE; | ||
| 698 | switch(op) | switch(op) |
| 699 | { | { |
| 700 | case OP_MARK: | |
| 701 | markptr = ecode + 2; | |
| 702 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 703 | ims, eptrb, flags, RM55); | |
| 704 | ||
| 705 | /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an | |
| 706 | argument, and we must check whether that argument matches this MARK's | |
| 707 | argument. It is passed back in md->start_match_ptr (an overloading of that | |
| 708 | variable). If it does match, we reset that variable to the current subject | |
| 709 | position and return MATCH_SKIP. Otherwise, pass back the return code | |
| 710 | unaltered. */ | |
| 711 | ||
| 712 | if (rrc == MATCH_SKIP_ARG && | |
| 713 | strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0) | |
| 714 | { | |
| 715 | md->start_match_ptr = eptr; | |
| 716 | RRETURN(MATCH_SKIP); | |
| 717 | } | |
| 718 | ||
| 719 | if (md->mark == NULL) md->mark = markptr; | |
| 720 | RRETURN(rrc); | |
| 721 | ||
| 722 | case OP_FAIL: | case OP_FAIL: |
| 723 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 724 | ||
| 725 | /* COMMIT overrides PRUNE, SKIP, and THEN */ | |
| 726 | ||
| 727 | case OP_COMMIT: | |
| 728 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | |
| 729 | ims, eptrb, flags, RM52); | |
| 730 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && | |
| 731 | rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG && | |
| 732 | rrc != MATCH_THEN) | |
| 733 | RRETURN(rrc); | |
| 734 | MRRETURN(MATCH_COMMIT); | |
| 735 | ||
| 736 | /* PRUNE overrides THEN */ | |
| 737 | ||
| 738 | case OP_PRUNE: | case OP_PRUNE: |
| 739 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 740 | ims, eptrb, flags, RM51); | ims, eptrb, flags, RM51); |
| 741 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| 742 | MRRETURN(MATCH_PRUNE); | |
| 743 | ||
| 744 | case OP_PRUNE_ARG: | |
| 745 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 746 | ims, eptrb, flags, RM56); | |
| 747 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | |
| 748 | md->mark = ecode + 2; | |
| 749 | RRETURN(MATCH_PRUNE); | RRETURN(MATCH_PRUNE); |
| 750 | ||
| 751 | case OP_COMMIT: | /* SKIP overrides PRUNE and THEN */ |
| RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | ||
| ims, eptrb, flags, RM52); | ||
| if (rrc != MATCH_NOMATCH) RRETURN(rrc); | ||
| RRETURN(MATCH_COMMIT); | ||
| 752 | ||
| 753 | case OP_SKIP: | case OP_SKIP: |
| 754 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 755 | ims, eptrb, flags, RM53); | ims, eptrb, flags, RM53); |
| 756 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) |
| 757 | RRETURN(rrc); | |
| 758 | md->start_match_ptr = eptr; /* Pass back current position */ | md->start_match_ptr = eptr; /* Pass back current position */ |
| 759 | RRETURN(MATCH_SKIP); | MRRETURN(MATCH_SKIP); |
| 760 | ||
| 761 | case OP_SKIP_ARG: | |
| 762 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 763 | ims, eptrb, flags, RM57); | |
| 764 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) | |
| 765 | RRETURN(rrc); | |
| 766 | ||
| 767 | /* Pass back the current skip name by overloading md->start_match_ptr and | |
| 768 | returning the special MATCH_SKIP_ARG return code. This will either be | |
| 769 | caught by a matching MARK, or get to the top, where it is treated the same | |
| 770 | as PRUNE. */ | |
| 771 | ||
| 772 | md->start_match_ptr = ecode + 2; | |
| 773 | RRETURN(MATCH_SKIP_ARG); | |
| 774 | ||
| 775 | /* For THEN (and THEN_ARG) we pass back the address of the bracket or | |
| 776 | the alt that is at the start of the current branch. This makes it possible | |
| 777 | to skip back past alternatives that precede the THEN within the current | |
| 778 | branch. */ | |
| 779 | ||
| 780 | case OP_THEN: | case OP_THEN: |
| 781 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 782 | ims, eptrb, flags, RM54); | ims, eptrb, flags, RM54); |
| 783 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 784 | md->start_match_ptr = ecode - GET(ecode, 1); | |
| 785 | MRRETURN(MATCH_THEN); | |
| 786 | ||
| 787 | case OP_THEN_ARG: | |
| 788 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE], | |
| 789 | offset_top, md, ims, eptrb, flags, RM58); | |
| 790 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 791 | md->start_match_ptr = ecode - GET(ecode, 1); | |
| 792 | md->mark = ecode + LINK_SIZE + 2; | |
| 793 | RRETURN(MATCH_THEN); | RRETURN(MATCH_THEN); |
| 794 | ||
| 795 | /* Handle a capturing bracket. If there is space in the offset vector, save | /* Handle a capturing bracket. If there is space in the offset vector, save |
| # | Line 699 for (;;) | Line 811 for (;;) |
| 811 | number = GET2(ecode, 1+LINK_SIZE); | number = GET2(ecode, 1+LINK_SIZE); |
| 812 | offset = number << 1; | offset = number << 1; |
| 813 | ||
| 814 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 815 | printf("start bracket %d\n", number); | printf("start bracket %d\n", number); |
| 816 | printf("subject="); | printf("subject="); |
| 817 | pchars(eptr, 16, TRUE, md); | pchars(eptr, 16, TRUE, md); |
| # | Line 714 for (;;) | Line 826 for (;;) |
| 826 | save_capture_last = md->capture_last; | save_capture_last = md->capture_last; |
| 827 | ||
| 828 | DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); | DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); |
| 829 | md->offset_vector[md->offset_end - number] = eptr - md->start_subject; | md->offset_vector[md->offset_end - number] = |
| 830 | (int)(eptr - md->start_subject); | |
| 831 | ||
| 832 | flags = (op == OP_SCBRA)? match_cbegroup : 0; | flags = (op == OP_SCBRA)? match_cbegroup : 0; |
| 833 | do | do |
| 834 | { | { |
| 835 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 836 | ims, eptrb, flags, RM1); | ims, eptrb, flags, RM1); |
| 837 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && |
| 838 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 839 | RRETURN(rrc); | |
| 840 | md->capture_last = save_capture_last; | md->capture_last = save_capture_last; |
| 841 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 842 | } | } |
| # | Line 733 for (;;) | Line 848 for (;;) |
| 848 | md->offset_vector[offset+1] = save_offset2; | md->offset_vector[offset+1] = save_offset2; |
| 849 | md->offset_vector[md->offset_end - number] = save_offset3; | md->offset_vector[md->offset_end - number] = save_offset3; |
| 850 | ||
| 851 | if (rrc != MATCH_THEN) md->mark = markptr; | |
| 852 | RRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 853 | } | } |
| 854 | ||
| # | Line 772 for (;;) | Line 888 for (;;) |
| 888 | ||
| 889 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, |
| 890 | eptrb, flags, RM48); | eptrb, flags, RM48); |
| 891 | if (rrc == MATCH_NOMATCH) md->mark = markptr; | |
| 892 | RRETURN(rrc); | RRETURN(rrc); |
| 893 | } | } |
| 894 | ||
| # | Line 780 for (;;) | Line 897 for (;;) |
| 897 | ||
| 898 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, |
| 899 | eptrb, flags, RM2); | eptrb, flags, RM2); |
| 900 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && |
| 901 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 902 | RRETURN(rrc); | |
| 903 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 904 | } | } |
| 905 | /* Control never reaches here. */ | /* Control never reaches here. */ |
| # | Line 807 for (;;) | Line 926 for (;;) |
| 926 | cb.callout_number = ecode[LINK_SIZE+2]; | cb.callout_number = ecode[LINK_SIZE+2]; |
| 927 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
| 928 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
| 929 | cb.subject_length = md->end_subject - md->start_subject; | cb.subject_length = (int)(md->end_subject - md->start_subject); |
| 930 | cb.start_match = mstart - md->start_subject; | cb.start_match = (int)(mstart - md->start_subject); |
| 931 | cb.current_position = eptr - md->start_subject; | cb.current_position = (int)(eptr - md->start_subject); |
| 932 | cb.pattern_position = GET(ecode, LINK_SIZE + 3); | cb.pattern_position = GET(ecode, LINK_SIZE + 3); |
| 933 | cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE); | cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE); |
| 934 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 935 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 936 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 937 | if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH); | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
| 938 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 939 | } | } |
| 940 | ecode += _pcre_OP_lengths[OP_CALLOUT]; | ecode += _pcre_OP_lengths[OP_CALLOUT]; |
| # | Line 825 for (;;) | Line 944 for (;;) |
| 944 | ||
| 945 | /* Now see what the actual condition is */ | /* Now see what the actual condition is */ |
| 946 | ||
| 947 | if (condcode == OP_RREF) /* Recursion test */ | if (condcode == OP_RREF || condcode == OP_NRREF) /* Recursion test */ |
| 948 | { | { |
| 949 | offset = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | if (md->recursive == NULL) /* Not recursing => FALSE */ |
| 950 | condition = md->recursive != NULL && | { |
| 951 | (offset == RREF_ANY || offset == md->recursive->group_num); | condition = FALSE; |
| 952 | ecode += condition? 3 : GET(ecode, 1); | ecode += GET(ecode, 1); |
| 953 | } | |
| 954 | else | |
| 955 | { | |
| 956 | int recno = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | |
| 957 | condition = (recno == RREF_ANY || recno == md->recursive->group_num); | |
| 958 | ||
| 959 | /* If the test is for recursion into a specific subpattern, and it is | |
| 960 | false, but the test was set up by name, scan the table to see if the | |
| 961 | name refers to any other numbers, and test them. The condition is true | |
| 962 | if any one is set. */ | |
| 963 | ||
| 964 | if (!condition && condcode == OP_NRREF && recno != RREF_ANY) | |
| 965 | { | |
| 966 | uschar *slotA = md->name_table; | |
| 967 | for (i = 0; i < md->name_count; i++) | |
| 968 | { | |
| 969 | if (GET2(slotA, 0) == recno) break; | |
| 970 | slotA += md->name_entry_size; | |
| 971 | } | |
| 972 | ||
| 973 | /* Found a name for the number - there can be only one; duplicate | |
| 974 | names for different numbers are allowed, but not vice versa. First | |
| 975 | scan down for duplicates. */ | |
| 976 | ||
| 977 | if (i < md->name_count) | |
| 978 | { | |
| 979 | uschar *slotB = slotA; | |
| 980 | while (slotB > md->name_table) | |
| 981 | { | |
| 982 | slotB -= md->name_entry_size; | |
| 983 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 984 | { | |
| 985 | condition = GET2(slotB, 0) == md->recursive->group_num; | |
| 986 | if (condition) break; | |
| 987 | } | |
| 988 | else break; | |
| 989 | } | |
| 990 | ||
| 991 | /* Scan up for duplicates */ | |
| 992 | ||
| 993 | if (!condition) | |
| 994 | { | |
| 995 | slotB = slotA; | |
| 996 | for (i++; i < md->name_count; i++) | |
| 997 | { | |
| 998 | slotB += md->name_entry_size; | |
| 999 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1000 | { | |
| 1001 | condition = GET2(slotB, 0) == md->recursive->group_num; | |
| 1002 | if (condition) break; | |
| 1003 | } | |
| 1004 | else break; | |
| 1005 | } | |
| 1006 | } | |
| 1007 | } | |
| 1008 | } | |
| 1009 | ||
| 1010 | /* Chose branch according to the condition */ | |
| 1011 | ||
| 1012 | ecode += condition? 3 : GET(ecode, 1); | |
| 1013 | } | |
| 1014 | } | } |
| 1015 | ||
| 1016 | else if (condcode == OP_CREF) /* Group used test */ | else if (condcode == OP_CREF || condcode == OP_NCREF) /* Group used test */ |
| 1017 | { | { |
| 1018 | offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ | offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ |
| 1019 | condition = offset < offset_top && md->offset_vector[offset] >= 0; | condition = offset < offset_top && md->offset_vector[offset] >= 0; |
| 1020 | ||
| 1021 | /* If the numbered capture is unset, but the reference was by name, | |
| 1022 | scan the table to see if the name refers to any other numbers, and test | |
| 1023 | them. The condition is true if any one is set. This is tediously similar | |
| 1024 | to the code above, but not close enough to try to amalgamate. */ | |
| 1025 | ||
| 1026 | if (!condition && condcode == OP_NCREF) | |
| 1027 | { | |
| 1028 | int refno = offset >> 1; | |
| 1029 | uschar *slotA = md->name_table; | |
| 1030 | ||
| 1031 | for (i = 0; i < md->name_count; i++) | |
| 1032 | { | |
| 1033 | if (GET2(slotA, 0) == refno) break; | |
| 1034 | slotA += md->name_entry_size; | |
| 1035 | } | |
| 1036 | ||
| 1037 | /* Found a name for the number - there can be only one; duplicate names | |
| 1038 | for different numbers are allowed, but not vice versa. First scan down | |
| 1039 | for duplicates. */ | |
| 1040 | ||
| 1041 | if (i < md->name_count) | |
| 1042 | { | |
| 1043 | uschar *slotB = slotA; | |
| 1044 | while (slotB > md->name_table) | |
| 1045 | { | |
| 1046 | slotB -= md->name_entry_size; | |
| 1047 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1048 | { | |
| 1049 | offset = GET2(slotB, 0) << 1; | |
| 1050 | condition = offset < offset_top && | |
| 1051 | md->offset_vector[offset] >= 0; | |
| 1052 | if (condition) break; | |
| 1053 | } | |
| 1054 | else break; | |
| 1055 | } | |
| 1056 | ||
| 1057 | /* Scan up for duplicates */ | |
| 1058 | ||
| 1059 | if (!condition) | |
| 1060 | { | |
| 1061 | slotB = slotA; | |
| 1062 | for (i++; i < md->name_count; i++) | |
| 1063 | { | |
| 1064 | slotB += md->name_entry_size; | |
| 1065 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1066 | { | |
| 1067 | offset = GET2(slotB, 0) << 1; | |
| 1068 | condition = offset < offset_top && | |
| 1069 | md->offset_vector[offset] >= 0; | |
| 1070 | if (condition) break; | |
| 1071 | } | |
| 1072 | else break; | |
| 1073 | } | |
| 1074 | } | |
| 1075 | } | |
| 1076 | } | |
| 1077 | ||
| 1078 | /* Chose branch according to the condition */ | |
| 1079 | ||
| 1080 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 3 : GET(ecode, 1); |
| 1081 | } | } |
| 1082 | ||
| # | Line 860 for (;;) | Line 1100 for (;;) |
| 1100 | ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); | ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); |
| 1101 | while (*ecode == OP_ALT) ecode += GET(ecode, 1); | while (*ecode == OP_ALT) ecode += GET(ecode, 1); |
| 1102 | } | } |
| 1103 | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) | else if (rrc != MATCH_NOMATCH && |
| 1104 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1105 | { | { |
| 1106 | RRETURN(rrc); /* Need braces because of following else */ | RRETURN(rrc); /* Need braces because of following else */ |
| 1107 | } | } |
| # | Line 897 for (;;) | Line 1138 for (;;) |
| 1138 | break; | break; |
| 1139 | ||
| 1140 | ||
| 1141 | /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes, | |
| 1142 | to close any currently open capturing brackets. */ | |
| 1143 | ||
| 1144 | case OP_CLOSE: | |
| 1145 | number = GET2(ecode, 1); | |
| 1146 | offset = number << 1; | |
| 1147 | ||
| 1148 | #ifdef PCRE_DEBUG | |
| 1149 | printf("end bracket %d at *ACCEPT", number); | |
| 1150 | printf("\n"); | |
| 1151 | #endif | |
| 1152 | ||
| 1153 | md->capture_last = number; | |
| 1154 | if (offset >= md->offset_max) md->offset_overflow = TRUE; else | |
| 1155 | { | |
| 1156 | md->offset_vector[offset] = | |
| 1157 | md->offset_vector[md->offset_end - number]; | |
| 1158 | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); | |
| 1159 | if (offset_top <= offset) offset_top = offset + 2; | |
| 1160 | } | |
| 1161 | ecode += 3; | |
| 1162 | break; | |
| 1163 | ||
| 1164 | ||
| 1165 | /* End of the pattern, either real or forced. If we are in a top-level | /* End of the pattern, either real or forced. If we are in a top-level |
| 1166 | recursion, we should restore the offsets appropriately and continue from | recursion, we should restore the offsets appropriately and continue from |
| 1167 | after the call. */ | after the call. */ |
| # | Line 910 for (;;) | Line 1175 for (;;) |
| 1175 | md->recursive = rec->prevrec; | md->recursive = rec->prevrec; |
| 1176 | memmove(md->offset_vector, rec->offset_save, | memmove(md->offset_vector, rec->offset_save, |
| 1177 | rec->saved_max * sizeof(int)); | rec->saved_max * sizeof(int)); |
| 1178 | mstart = rec->save_start; | offset_top = rec->save_offset_top; |
| 1179 | ims = original_ims; | ims = original_ims; |
| 1180 | ecode = rec->after_call; | ecode = rec->after_call; |
| 1181 | break; | break; |
| 1182 | } | } |
| 1183 | ||
| 1184 | /* Otherwise, if PCRE_NOTEMPTY is set, fail if we have matched an empty | /* Otherwise, if we have matched an empty string, fail if PCRE_NOTEMPTY is |
| 1185 | string - backtracking will then try other alternatives, if any. */ | set, or if PCRE_NOTEMPTY_ATSTART is set and we have matched at the start of |
| 1186 | the subject. In both cases, backtracking will then try other alternatives, | |
| 1187 | if any. */ | |
| 1188 | ||
| 1189 | if (eptr == mstart && | |
| 1190 | (md->notempty || | |
| 1191 | (md->notempty_atstart && | |
| 1192 | mstart == md->start_subject + md->start_offset))) | |
| 1193 | MRRETURN(MATCH_NOMATCH); | |
| 1194 | ||
| 1195 | /* Otherwise, we have a match. */ | |
| 1196 | ||
| if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH); | ||
| 1197 | md->end_match_ptr = eptr; /* Record where we ended */ | md->end_match_ptr = eptr; /* Record where we ended */ |
| 1198 | md->end_offset_top = offset_top; /* and how many extracts were taken */ | md->end_offset_top = offset_top; /* and how many extracts were taken */ |
| 1199 | md->start_match_ptr = mstart; /* and the start (\K can modify) */ | md->start_match_ptr = mstart; /* and the start (\K can modify) */ |
| 1200 | RRETURN(MATCH_MATCH); | |
| 1201 | /* For some reason, the macros don't work properly if an expression is | |
| 1202 | given as the argument to MRRETURN when the heap is in use. */ | |
| 1203 | ||
| 1204 | rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; | |
| 1205 | MRRETURN(rrc); | |
| 1206 | ||
| 1207 | /* Change option settings */ | /* Change option settings */ |
| 1208 | ||
| # | Line 945 for (;;) | Line 1224 for (;;) |
| 1224 | { | { |
| 1225 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, |
| 1226 | RM4); | RM4); |
| 1227 | if (rrc == MATCH_MATCH) break; | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
| 1228 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | { |
| 1229 | mstart = md->start_match_ptr; /* In case \K reset it */ | |
| 1230 | break; | |
| 1231 | } | |
| 1232 | if (rrc != MATCH_NOMATCH && | |
| 1233 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1234 | RRETURN(rrc); | |
| 1235 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 1236 | } | } |
| 1237 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| 1238 | if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); | if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH); |
| 1239 | ||
| 1240 | /* If checking an assertion for a condition, return MATCH_MATCH. */ | /* If checking an assertion for a condition, return MATCH_MATCH. */ |
| 1241 | ||
| # | Line 964 for (;;) | Line 1249 for (;;) |
| 1249 | offset_top = md->end_offset_top; | offset_top = md->end_offset_top; |
| 1250 | continue; | continue; |
| 1251 | ||
| 1252 | /* Negative assertion: all branches must fail to match */ | /* Negative assertion: all branches must fail to match. Encountering SKIP, |
| 1253 | PRUNE, or COMMIT means we must assume failure without checking subsequent | |
| 1254 | branches. */ | |
| 1255 | ||
| 1256 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
| 1257 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
| # | Line 972 for (;;) | Line 1259 for (;;) |
| 1259 | { | { |
| 1260 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, |
| 1261 | RM5); | RM5); |
| 1262 | if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH); | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH); |
| 1263 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) |
| 1264 | { | |
| 1265 | do ecode += GET(ecode,1); while (*ecode == OP_ALT); | |
| 1266 | break; | |
| 1267 | } | |
| 1268 | if (rrc != MATCH_NOMATCH && | |
| 1269 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1270 | RRETURN(rrc); | |
| 1271 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
| 1272 | } | } |
| 1273 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| # | Line 996 for (;;) | Line 1290 for (;;) |
| 1290 | while (i-- > 0) | while (i-- > 0) |
| 1291 | { | { |
| 1292 | eptr--; | eptr--; |
| 1293 | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1294 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 1295 | } | } |
| 1296 | } | } |
| # | Line 1007 for (;;) | Line 1301 for (;;) |
| 1301 | ||
| 1302 | { | { |
| 1303 | eptr -= GET(ecode, 1); | eptr -= GET(ecode, 1); |
| 1304 | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1305 | } | } |
| 1306 | ||
| 1307 | /* Skip to next op code */ | /* Save the earliest consulted character, then skip to next op code */ |
| 1308 | ||
| 1309 | if (eptr < md->start_used_ptr) md->start_used_ptr = eptr; | |
| 1310 | ecode += 1 + LINK_SIZE; | ecode += 1 + LINK_SIZE; |
| 1311 | break; | break; |
| 1312 | ||
| # | Line 1027 for (;;) | Line 1322 for (;;) |
| 1322 | cb.callout_number = ecode[1]; | cb.callout_number = ecode[1]; |
| 1323 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
| 1324 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
| 1325 | cb.subject_length = md->end_subject - md->start_subject; | cb.subject_length = (int)(md->end_subject - md->start_subject); |
| 1326 | cb.start_match = mstart - md->start_subject; | cb.start_match = (int)(mstart - md->start_subject); |
| 1327 | cb.current_position = eptr - md->start_subject; | cb.current_position = (int)(eptr - md->start_subject); |
| 1328 | cb.pattern_position = GET(ecode, 2); | cb.pattern_position = GET(ecode, 2); |
| 1329 | cb.next_item_length = GET(ecode, 2 + LINK_SIZE); | cb.next_item_length = GET(ecode, 2 + LINK_SIZE); |
| 1330 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 1331 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 1332 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 1333 | if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH); | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
| 1334 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 1335 | } | } |
| 1336 | ecode += 2 + 2*LINK_SIZE; | ecode += 2 + 2*LINK_SIZE; |
| # | Line 1090 for (;;) | Line 1385 for (;;) |
| 1385 | ||
| 1386 | memcpy(new_recursive.offset_save, md->offset_vector, | memcpy(new_recursive.offset_save, md->offset_vector, |
| 1387 | new_recursive.saved_max * sizeof(int)); | new_recursive.saved_max * sizeof(int)); |
| 1388 | new_recursive.save_start = mstart; | new_recursive.save_offset_top = offset_top; |
| mstart = eptr; | ||
| 1389 | ||
| 1390 | /* OK, now we can do the recursion. For each top-level alternative we | /* OK, now we can do the recursion. For each top-level alternative we |
| 1391 | restore the offset and recursion data. */ | restore the offset and recursion data. */ |
| # | Line 1102 for (;;) | Line 1396 for (;;) |
| 1396 | { | { |
| 1397 | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, |
| 1398 | md, ims, eptrb, flags, RM6); | md, ims, eptrb, flags, RM6); |
| 1399 | if (rrc == MATCH_MATCH) | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
| 1400 | { | { |
| 1401 | DPRINTF(("Recursion matched\n")); | DPRINTF(("Recursion matched\n")); |
| 1402 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1403 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1404 | (pcre_free)(new_recursive.offset_save); | (pcre_free)(new_recursive.offset_save); |
| 1405 | RRETURN(MATCH_MATCH); | MRRETURN(MATCH_MATCH); |
| 1406 | } | } |
| 1407 | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) | else if (rrc != MATCH_NOMATCH && |
| 1408 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1409 | { | { |
| 1410 | DPRINTF(("Recursion gave error %d\n", rrc)); | DPRINTF(("Recursion gave error %d\n", rrc)); |
| 1411 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| # | Line 1129 for (;;) | Line 1424 for (;;) |
| 1424 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1425 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1426 | (pcre_free)(new_recursive.offset_save); | (pcre_free)(new_recursive.offset_save); |
| 1427 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1428 | } | } |
| 1429 | /* Control never reaches here */ | /* Control never reaches here */ |
| 1430 | ||
| # | Line 1138 for (;;) | Line 1433 for (;;) |
| 1433 | a move back into the brackets. Friedl calls these "atomic" subpatterns. | a move back into the brackets. Friedl calls these "atomic" subpatterns. |
| 1434 | Check the alternative branches in turn - the matching won't pass the KET | Check the alternative branches in turn - the matching won't pass the KET |
| 1435 | for this kind of subpattern. If any one branch matches, we carry on as at | for this kind of subpattern. If any one branch matches, we carry on as at |
| 1436 | the end of a normal bracket, leaving the subject pointer. */ | the end of a normal bracket, leaving the subject pointer, but resetting |
| 1437 | the start-of-match value in case it was changed by \K. */ | |
| 1438 | ||
| 1439 | case OP_ONCE: | case OP_ONCE: |
| 1440 | prev = ecode; | prev = ecode; |
| # | Line 1147 for (;;) | Line 1443 for (;;) |
| 1443 | do | do |
| 1444 | { | { |
| 1445 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7); | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, eptrb, 0, RM7); |
| 1446 | if (rrc == MATCH_MATCH) break; | if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ |
| 1447 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | { |
| 1448 | mstart = md->start_match_ptr; | |
| 1449 | break; | |
| 1450 | } | |
| 1451 | if (rrc != MATCH_NOMATCH && | |
| 1452 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1453 | RRETURN(rrc); | |
| 1454 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
| 1455 | } | } |
| 1456 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| # | Line 1266 for (;;) | Line 1568 for (;;) |
| 1568 | } | } |
| 1569 | else saved_eptr = NULL; | else saved_eptr = NULL; |
| 1570 | ||
| 1571 | /* If we are at the end of an assertion group, stop matching and return | /* If we are at the end of an assertion group or an atomic group, stop |
| 1572 | MATCH_MATCH, but record the current high water mark for use by positive | matching and return MATCH_MATCH, but record the current high water mark for |
| 1573 | assertions. Do this also for the "once" (atomic) groups. */ | use by positive assertions. We also need to record the match start in case |
| 1574 | it was changed by \K. */ | |
| 1575 | ||
| 1576 | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || |
| 1577 | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || |
| # | Line 1276 for (;;) | Line 1579 for (;;) |
| 1579 | { | { |
| 1580 | md->end_match_ptr = eptr; /* For ONCE */ | md->end_match_ptr = eptr; /* For ONCE */ |
| 1581 | md->end_offset_top = offset_top; | md->end_offset_top = offset_top; |
| 1582 | RRETURN(MATCH_MATCH); | md->start_match_ptr = mstart; |
| 1583 | MRRETURN(MATCH_MATCH); | |
| 1584 | } | } |
| 1585 | ||
| 1586 | /* For capturing groups we have to check the group number back at the start | /* For capturing groups we have to check the group number back at the start |
| # | Line 1290 for (;;) | Line 1594 for (;;) |
| 1594 | number = GET2(prev, 1+LINK_SIZE); | number = GET2(prev, 1+LINK_SIZE); |
| 1595 | offset = number << 1; | offset = number << 1; |
| 1596 | ||
| 1597 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 1598 | printf("end bracket %d", number); | printf("end bracket %d", number); |
| 1599 | printf("\n"); | printf("\n"); |
| 1600 | #endif | #endif |
| # | Line 1300 for (;;) | Line 1604 for (;;) |
| 1604 | { | { |
| 1605 | md->offset_vector[offset] = | md->offset_vector[offset] = |
| 1606 | md->offset_vector[md->offset_end - number]; | md->offset_vector[md->offset_end - number]; |
| 1607 | md->offset_vector[offset+1] = eptr - md->start_subject; | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); |
| 1608 | if (offset_top <= offset) offset_top = offset + 2; | if (offset_top <= offset) offset_top = offset + 2; |
| 1609 | } | } |
| 1610 | ||
| # | Line 1312 for (;;) | Line 1616 for (;;) |
| 1616 | recursion_info *rec = md->recursive; | recursion_info *rec = md->recursive; |
| 1617 | DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); | DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); |
| 1618 | md->recursive = rec->prevrec; | md->recursive = rec->prevrec; |
| mstart = rec->save_start; | ||
| 1619 | memcpy(md->offset_vector, rec->offset_save, | memcpy(md->offset_vector, rec->offset_save, |
| 1620 | rec->saved_max * sizeof(int)); | rec->saved_max * sizeof(int)); |
| 1621 | offset_top = rec->save_offset_top; | |
| 1622 | ecode = rec->after_call; | ecode = rec->after_call; |
| 1623 | ims = original_ims; | ims = original_ims; |
| 1624 | break; | break; |
| # | Line 1371 for (;;) | Line 1675 for (;;) |
| 1675 | /* Start of subject unless notbol, or after internal newline if multiline */ | /* Start of subject unless notbol, or after internal newline if multiline */ |
| 1676 | ||
| 1677 | case OP_CIRC: | case OP_CIRC: |
| 1678 | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1679 | if ((ims & PCRE_MULTILINE) != 0) | if ((ims & PCRE_MULTILINE) != 0) |
| 1680 | { | { |
| 1681 | if (eptr != md->start_subject && | if (eptr != md->start_subject && |
| 1682 | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) |
| 1683 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1684 | ecode++; | ecode++; |
| 1685 | break; | break; |
| 1686 | } | } |
| # | Line 1385 for (;;) | Line 1689 for (;;) |
| 1689 | /* Start of subject assertion */ | /* Start of subject assertion */ |
| 1690 | ||
| 1691 | case OP_SOD: | case OP_SOD: |
| 1692 | if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1693 | ecode++; | ecode++; |
| 1694 | break; | break; |
| 1695 | ||
| 1696 | /* Start of match assertion */ | /* Start of match assertion */ |
| 1697 | ||
| 1698 | case OP_SOM: | case OP_SOM: |
| 1699 | if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH); |
| 1700 | ecode++; | ecode++; |
| 1701 | break; | break; |
| 1702 | ||
| # | Line 1410 for (;;) | Line 1714 for (;;) |
| 1714 | if ((ims & PCRE_MULTILINE) != 0) | if ((ims & PCRE_MULTILINE) != 0) |
| 1715 | { | { |
| 1716 | if (eptr < md->end_subject) | if (eptr < md->end_subject) |
| 1717 | { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); } | { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); } |
| 1718 | else | else |
| 1719 | { if (md->noteol) RRETURN(MATCH_NOMATCH); } | { |
| 1720 | if (md->noteol) MRRETURN(MATCH_NOMATCH); | |
| 1721 | SCHECK_PARTIAL(); | |
| 1722 | } | |
| 1723 | ecode++; | ecode++; |
| 1724 | break; | break; |
| 1725 | } | } |
| 1726 | else | else /* Not multiline */ |
| 1727 | { | { |
| 1728 | if (md->noteol) RRETURN(MATCH_NOMATCH); | if (md->noteol) MRRETURN(MATCH_NOMATCH); |
| 1729 | if (!md->endonly) | if (!md->endonly) goto ASSERT_NL_OR_EOS; |
| { | ||
| if (eptr != md->end_subject && | ||
| (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | ||
| RRETURN(MATCH_NOMATCH); | ||
| ecode++; | ||
| break; | ||
| } | ||
| 1730 | } | } |
| 1731 | ||
| 1732 | /* ... else fall through for endonly */ | /* ... else fall through for endonly */ |
| 1733 | ||
| 1734 | /* End of subject assertion (\z) */ | /* End of subject assertion (\z) */ |
| 1735 | ||
| 1736 | case OP_EOD: | case OP_EOD: |
| 1737 | if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH); |
| 1738 | SCHECK_PARTIAL(); | |
| 1739 | ecode++; | ecode++; |
| 1740 | break; | break; |
| 1741 | ||
| 1742 | /* End of subject or ending \n assertion (\Z) */ | /* End of subject or ending \n assertion (\Z) */ |
| 1743 | ||
| 1744 | case OP_EODN: | case OP_EODN: |
| 1745 | if (eptr != md->end_subject && | ASSERT_NL_OR_EOS: |
| 1746 | if (eptr < md->end_subject && | |
| 1747 | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) |
| 1748 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1749 | ||
| 1750 | /* Either at end of string or \n before end. */ | |
| 1751 | ||
| 1752 | SCHECK_PARTIAL(); | |
| 1753 | ecode++; | ecode++; |
| 1754 | break; | break; |
| 1755 | ||
| # | Line 1454 for (;;) | Line 1761 for (;;) |
| 1761 | ||
| 1762 | /* Find out if the previous and current characters are "word" characters. | /* Find out if the previous and current characters are "word" characters. |
| 1763 | It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to | It takes a bit more work in UTF-8 mode. Characters > 255 are assumed to |
| 1764 | be "non-word" characters. */ | be "non-word" characters. Remember the earliest consulted character for |
| 1765 | partial matching. */ | |
| 1766 | ||
| 1767 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 1768 | if (utf8) | if (utf8) |
| 1769 | { | { |
| 1770 | /* Get status of previous character */ | |
| 1771 | ||
| 1772 | if (eptr == md->start_subject) prev_is_word = FALSE; else | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 1773 | { | { |
| 1774 | const uschar *lastptr = eptr - 1; | USPTR lastptr = eptr - 1; |
| 1775 | while((*lastptr & 0xc0) == 0x80) lastptr--; | while((*lastptr & 0xc0) == 0x80) lastptr--; |
| 1776 | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; | |
| 1777 | GETCHAR(c, lastptr); | GETCHAR(c, lastptr); |
| 1778 | #ifdef SUPPORT_UCP | |
| 1779 | if (md->use_ucp) | |
| 1780 | { | |
| 1781 | if (c == '_') prev_is_word = TRUE; else | |
| 1782 | { | |
| 1783 | int cat = UCD_CATEGORY(c); | |
| 1784 | prev_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1785 | } | |
| 1786 | } | |
| 1787 | else | |
| 1788 | #endif | |
| 1789 | prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; | prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; |
| 1790 | } | } |
| 1791 | if (eptr >= md->end_subject) cur_is_word = FALSE; else | |
| 1792 | /* Get status of next character */ | |
| 1793 | ||
| 1794 | if (eptr >= md->end_subject) | |
| 1795 | { | |
| 1796 | SCHECK_PARTIAL(); | |
| 1797 | cur_is_word = FALSE; | |
| 1798 | } | |
| 1799 | else | |
| 1800 | { | { |
| 1801 | GETCHAR(c, eptr); | GETCHAR(c, eptr); |
| 1802 | #ifdef SUPPORT_UCP | |
| 1803 | if (md->use_ucp) | |
| 1804 | { | |
| 1805 | if (c == '_') cur_is_word = TRUE; else | |
| 1806 | { | |
| 1807 | int cat = UCD_CATEGORY(c); | |
| 1808 | cur_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1809 | } | |
| 1810 | } | |
| 1811 | else | |
| 1812 | #endif | |
| 1813 | cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; | cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; |
| 1814 | } | } |
| 1815 | } | } |
| 1816 | else | else |
| 1817 | #endif | #endif |
| 1818 | ||
| 1819 | /* More streamlined when not in UTF-8 mode */ | /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for |
| 1820 | consistency with the behaviour of \w we do use it in this case. */ | |
| 1821 | ||
| 1822 | { | { |
| 1823 | prev_is_word = (eptr != md->start_subject) && | /* Get status of previous character */ |
| 1824 | ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
| 1825 | cur_is_word = (eptr < md->end_subject) && | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 1826 | ((md->ctypes[*eptr] & ctype_word) != 0); | { |
| 1827 | if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1; | |
| 1828 | #ifdef SUPPORT_UCP | |
| 1829 | if (md->use_ucp) | |
| 1830 | { | |
| 1831 | c = eptr[-1]; | |
| 1832 | if (c == '_') prev_is_word = TRUE; else | |
| 1833 | { | |
| 1834 | int cat = UCD_CATEGORY(c); | |
| 1835 | prev_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1836 | } | |
| 1837 | } | |
| 1838 | else | |
| 1839 | #endif | |
| 1840 | prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
| 1841 | } | |
| 1842 | ||
| 1843 | /* Get status of next character */ | |
| 1844 | ||
| 1845 | if (eptr >= md->end_subject) | |
| 1846 | { | |
| 1847 | SCHECK_PARTIAL(); | |
| 1848 | cur_is_word = FALSE; | |
| 1849 | } | |
| 1850 | else | |
| 1851 | #ifdef SUPPORT_UCP | |
| 1852 | if (md->use_ucp) | |
| 1853 | { | |
| 1854 | c = *eptr; | |
| 1855 | if (c == '_') cur_is_word = TRUE; else | |
| 1856 | { | |
| 1857 | int cat = UCD_CATEGORY(c); | |
| 1858 | cur_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1859 | } | |
| 1860 | } | |
| 1861 | else | |
| 1862 | #endif | |
| 1863 | cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0); | |
| 1864 | } | } |
| 1865 | ||
| 1866 | /* Now see if the situation is what we want */ | /* Now see if the situation is what we want */ |
| 1867 | ||
| 1868 | if ((*ecode++ == OP_WORD_BOUNDARY)? | if ((*ecode++ == OP_WORD_BOUNDARY)? |
| 1869 | cur_is_word == prev_is_word : cur_is_word != prev_is_word) | cur_is_word == prev_is_word : cur_is_word != prev_is_word) |
| 1870 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1871 | } | } |
| 1872 | break; | break; |
| 1873 | ||
| 1874 | /* Match a single character type; inline for speed */ | /* Match a single character type; inline for speed */ |
| 1875 | ||
| 1876 | case OP_ANY: | case OP_ANY: |
| 1877 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); |
| 1878 | /* Fall through */ | /* Fall through */ |
| 1879 | ||
| 1880 | case OP_ALLANY: | case OP_ALLANY: |
| 1881 | if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr++ >= md->end_subject) |
| 1882 | { | |
| 1883 | SCHECK_PARTIAL(); | |
| 1884 | MRRETURN(MATCH_NOMATCH); | |
| 1885 | } | |
| 1886 | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 1887 | ecode++; | ecode++; |
| 1888 | break; | break; |
| # | Line 1508 for (;;) | Line 1891 for (;;) |
| 1891 | any byte, even newline, independent of the setting of PCRE_DOTALL. */ | any byte, even newline, independent of the setting of PCRE_DOTALL. */ |
| 1892 | ||
| 1893 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 1894 | if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr++ >= md->end_subject) |
| 1895 | { | |
| 1896 | SCHECK_PARTIAL(); | |
| 1897 | MRRETURN(MATCH_NOMATCH); | |
| 1898 | } | |
| 1899 | ecode++; | ecode++; |
| 1900 | break; | break; |
| 1901 | ||
| 1902 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 1903 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1904 | { | |
| 1905 | SCHECK_PARTIAL(); | |
| 1906 | MRRETURN(MATCH_NOMATCH); | |
| 1907 | } | |
| 1908 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1909 | if ( | if ( |
| 1910 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1521 for (;;) | Line 1912 for (;;) |
| 1912 | #endif | #endif |
| 1913 | (md->ctypes[c] & ctype_digit) != 0 | (md->ctypes[c] & ctype_digit) != 0 |
| 1914 | ) | ) |
| 1915 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1916 | ecode++; | ecode++; |
| 1917 | break; | break; |
| 1918 | ||
| 1919 | case OP_DIGIT: | case OP_DIGIT: |
| 1920 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1921 | { | |
| 1922 | SCHECK_PARTIAL(); | |
| 1923 | MRRETURN(MATCH_NOMATCH); | |
| 1924 | } | |
| 1925 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1926 | if ( | if ( |
| 1927 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1534 for (;;) | Line 1929 for (;;) |
| 1929 | #endif | #endif |
| 1930 | (md->ctypes[c] & ctype_digit) == 0 | (md->ctypes[c] & ctype_digit) == 0 |
| 1931 | ) | ) |
| 1932 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1933 | ecode++; | ecode++; |
| 1934 | break; | break; |
| 1935 | ||
| 1936 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 1937 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1938 | { | |
| 1939 | SCHECK_PARTIAL(); | |
| 1940 | MRRETURN(MATCH_NOMATCH); | |
| 1941 | } | |
| 1942 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1943 | if ( | if ( |
| 1944 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1547 for (;;) | Line 1946 for (;;) |
| 1946 | #endif | #endif |
| 1947 | (md->ctypes[c] & ctype_space) != 0 | (md->ctypes[c] & ctype_space) != 0 |
| 1948 | ) | ) |
| 1949 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1950 | ecode++; | ecode++; |
| 1951 | break; | break; |
| 1952 | ||
| 1953 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 1954 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1955 | { | |
| 1956 | SCHECK_PARTIAL(); | |
| 1957 | MRRETURN(MATCH_NOMATCH); | |
| 1958 | } | |
| 1959 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1960 | if ( | if ( |
| 1961 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1560 for (;;) | Line 1963 for (;;) |
| 1963 | #endif | #endif |
| 1964 | (md->ctypes[c] & ctype_space) == 0 | (md->ctypes[c] & ctype_space) == 0 |
| 1965 | ) | ) |
| 1966 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1967 | ecode++; | ecode++; |
| 1968 | break; | break; |
| 1969 | ||
| 1970 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 1971 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1972 | { | |
| 1973 | SCHECK_PARTIAL(); | |
| 1974 | MRRETURN(MATCH_NOMATCH); | |
| 1975 | } | |
| 1976 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1977 | if ( | if ( |
| 1978 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1573 for (;;) | Line 1980 for (;;) |
| 1980 | #endif | #endif |
| 1981 | (md->ctypes[c] & ctype_word) != 0 | (md->ctypes[c] & ctype_word) != 0 |
| 1982 | ) | ) |
| 1983 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1984 | ecode++; | ecode++; |
| 1985 | break; | break; |
| 1986 | ||
| 1987 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 1988 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1989 | { | |
| 1990 | SCHECK_PARTIAL(); | |
| 1991 | MRRETURN(MATCH_NOMATCH); | |
| 1992 | } | |
| 1993 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1994 | if ( | if ( |
| 1995 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1586 for (;;) | Line 1997 for (;;) |
| 1997 | #endif | #endif |
| 1998 | (md->ctypes[c] & ctype_word) == 0 | (md->ctypes[c] & ctype_word) == 0 |
| 1999 | ) | ) |
| 2000 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2001 | ecode++; | ecode++; |
| 2002 | break; | break; |
| 2003 | ||
| 2004 | case OP_ANYNL: | case OP_ANYNL: |
| 2005 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2006 | { | |
| 2007 | SCHECK_PARTIAL(); | |
| 2008 | MRRETURN(MATCH_NOMATCH); | |
| 2009 | } | |
| 2010 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2011 | switch(c) | switch(c) |
| 2012 | { | { |
| 2013 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2014 | case 0x000d: | case 0x000d: |
| 2015 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 2016 | break; | break; |
| # | Line 1608 for (;;) | Line 2023 for (;;) |
| 2023 | case 0x0085: | case 0x0085: |
| 2024 | case 0x2028: | case 0x2028: |
| 2025 | case 0x2029: | case 0x2029: |
| 2026 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 2027 | break; | break; |
| 2028 | } | } |
| 2029 | ecode++; | ecode++; |
| 2030 | break; | break; |
| 2031 | ||
| 2032 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 2033 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2034 | { | |
| 2035 | SCHECK_PARTIAL(); | |
| 2036 | MRRETURN(MATCH_NOMATCH); | |
| 2037 | } | |
| 2038 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2039 | switch(c) | switch(c) |
| 2040 | { | { |
| # | Line 1639 for (;;) | Line 2058 for (;;) |
| 2058 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 2059 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 2060 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 2061 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2062 | } | } |
| 2063 | ecode++; | ecode++; |
| 2064 | break; | break; |
| 2065 | ||
| 2066 | case OP_HSPACE: | case OP_HSPACE: |
| 2067 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2068 | { | |
| 2069 | SCHECK_PARTIAL(); | |
| 2070 | MRRETURN(MATCH_NOMATCH); | |
| 2071 | } | |
| 2072 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2073 | switch(c) | switch(c) |
| 2074 | { | { |
| 2075 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2076 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 2077 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 2078 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 1675 for (;;) | Line 2098 for (;;) |
| 2098 | break; | break; |
| 2099 | ||
| 2100 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 2101 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2102 | { | |
| 2103 | SCHECK_PARTIAL(); | |
| 2104 | MRRETURN(MATCH_NOMATCH); | |
| 2105 | } | |
| 2106 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2107 | switch(c) | switch(c) |
| 2108 | { | { |
| # | Line 1687 for (;;) | Line 2114 for (;;) |
| 2114 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 2115 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 2116 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 2117 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2118 | } | } |
| 2119 | ecode++; | ecode++; |
| 2120 | break; | break; |
| 2121 | ||
| 2122 | case OP_VSPACE: | case OP_VSPACE: |
| 2123 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2124 | { | |
| 2125 | SCHECK_PARTIAL(); | |
| 2126 | MRRETURN(MATCH_NOMATCH); | |
| 2127 | } | |
| 2128 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2129 | switch(c) | switch(c) |
| 2130 | { | { |
| 2131 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2132 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 2133 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 2134 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 1716 for (;;) | Line 2147 for (;;) |
| 2147 | ||
| 2148 | case OP_PROP: | case OP_PROP: |
| 2149 | case OP_NOTPROP: | case OP_NOTPROP: |
| 2150 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2151 | { | |
| 2152 | SCHECK_PARTIAL(); | |
| 2153 | MRRETURN(MATCH_NOMATCH); | |
| 2154 | } | |
| 2155 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2156 | { | { |
| 2157 | const ucd_record *prop = GET_UCD(c); | const ucd_record *prop = GET_UCD(c); |
| # | Line 1724 for (;;) | Line 2159 for (;;) |
| 2159 | switch(ecode[1]) | switch(ecode[1]) |
| 2160 | { | { |
| 2161 | case PT_ANY: | case PT_ANY: |
| 2162 | if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); | if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH); |
| 2163 | break; | break; |
| 2164 | ||
| 2165 | case PT_LAMP: | case PT_LAMP: |
| 2166 | if ((prop->chartype == ucp_Lu || | if ((prop->chartype == ucp_Lu || |
| 2167 | prop->chartype == ucp_Ll || | prop->chartype == ucp_Ll || |
| 2168 | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) |
| 2169 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2170 | break; | break; |
| 2171 | ||
| 2172 | case PT_GC: | case PT_GC: |
| 2173 | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) |
| 2174 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2175 | break; | break; |
| 2176 | ||
| 2177 | case PT_PC: | case PT_PC: |
| 2178 | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) |
| 2179 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2180 | break; | break; |
| 2181 | ||
| 2182 | case PT_SC: | case PT_SC: |
| 2183 | if ((ecode[2] != prop->script) == (op == OP_PROP)) | if ((ecode[2] != prop->script) == (op == OP_PROP)) |
| 2184 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2185 | break; | |
| 2186 | ||
| 2187 | /* These are specials */ | |
| 2188 | ||
| 2189 | case PT_ALNUM: | |
| 2190 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | |
| 2191 | _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) | |
| 2192 | MRRETURN(MATCH_NOMATCH); | |
| 2193 | break; | |
| 2194 | ||
| 2195 | case PT_SPACE: /* Perl space */ | |
| 2196 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | |
| 2197 | c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) | |
| 2198 | == (op == OP_NOTPROP)) | |
| 2199 | MRRETURN(MATCH_NOMATCH); | |
| 2200 | break; | |
| 2201 | ||
| 2202 | case PT_PXSPACE: /* POSIX space */ | |
| 2203 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | |
| 2204 | c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || | |
| 2205 | c == CHAR_FF || c == CHAR_CR) | |
| 2206 | == (op == OP_NOTPROP)) | |
| 2207 | MRRETURN(MATCH_NOMATCH); | |
| 2208 | break; | |
| 2209 | ||
| 2210 | case PT_WORD: | |
| 2211 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | |
| 2212 | _pcre_ucp_gentype[prop->chartype] == ucp_N || | |
| 2213 | c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) | |
| 2214 | MRRETURN(MATCH_NOMATCH); | |
| 2215 | break; | break; |
| 2216 | ||
| 2217 | /* This should never occur */ | |
| 2218 | ||
| 2219 | default: | default: |
| 2220 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 2221 | } | } |
| # | Line 1761 for (;;) | Line 2228 for (;;) |
| 2228 | is in the binary; otherwise a compile-time error occurs. */ | is in the binary; otherwise a compile-time error occurs. */ |
| 2229 | ||
| 2230 | case OP_EXTUNI: | case OP_EXTUNI: |
| 2231 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2232 | { | |
| 2233 | SCHECK_PARTIAL(); | |
| 2234 | MRRETURN(MATCH_NOMATCH); | |
| 2235 | } | |
| 2236 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2237 | { | { |
| 2238 | int category = UCD_CATEGORY(c); | int category = UCD_CATEGORY(c); |
| 2239 | if (category == ucp_M) RRETURN(MATCH_NOMATCH); | if (category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 2240 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 2241 | { | { |
| 2242 | int len = 1; | int len = 1; |
| # | Line 1792 for (;;) | Line 2263 for (;;) |
| 2263 | loops). */ | loops). */ |
| 2264 | ||
| 2265 | case OP_REF: | case OP_REF: |
| 2266 | { | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ |
| 2267 | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ | ecode += 3; |
| ecode += 3; | ||
| 2268 | ||
| 2269 | /* If the reference is unset, there are two possibilities: | /* If the reference is unset, there are two possibilities: |
| 2270 | ||
| 2271 | (a) In the default, Perl-compatible state, set the length to be longer | (a) In the default, Perl-compatible state, set the length negative; |
| 2272 | than the amount of subject left; this ensures that every attempt at a | this ensures that every attempt at a match fails. We can't just fail |
| 2273 | match fails. We can't just fail here, because of the possibility of | here, because of the possibility of quantifiers with zero minima. |
| quantifiers with zero minima. | ||
| 2274 | ||
| 2275 | (b) If the JavaScript compatibility flag is set, set the length to zero | (b) If the JavaScript compatibility flag is set, set the length to zero |
| 2276 | so that the back reference matches an empty string. | so that the back reference matches an empty string. |
| 2277 | ||
| 2278 | Otherwise, set the length to the length of what was matched by the | Otherwise, set the length to the length of what was matched by the |
| 2279 | referenced subpattern. */ | referenced subpattern. */ |
| 2280 | ||
| 2281 | if (offset >= offset_top || md->offset_vector[offset] < 0) | if (offset >= offset_top || md->offset_vector[offset] < 0) |
| 2282 | length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1; | length = (md->jscript_compat)? 0 : -1; |
| 2283 | else | else |
| 2284 | length = md->offset_vector[offset+1] - md->offset_vector[offset]; | length = md->offset_vector[offset+1] - md->offset_vector[offset]; |
| 2285 | ||
| 2286 | /* Set up for repetition, or handle the non-repeated case */ | /* Set up for repetition, or handle the non-repeated case */ |
| 2287 | ||
| 2288 | switch (*ecode) | switch (*ecode) |
| 2289 | { | { |
| 2290 | case OP_CRSTAR: | case OP_CRSTAR: |
| 2291 | case OP_CRMINSTAR: | case OP_CRMINSTAR: |
| 2292 | case OP_CRPLUS: | case OP_CRPLUS: |
| 2293 | case OP_CRMINPLUS: | case OP_CRMINPLUS: |
| 2294 | case OP_CRQUERY: | case OP_CRQUERY: |
| 2295 | case OP_CRMINQUERY: | case OP_CRMINQUERY: |
| 2296 | c = *ecode++ - OP_CRSTAR; | c = *ecode++ - OP_CRSTAR; |
| 2297 | minimize = (c & 1) != 0; | minimize = (c & 1) != 0; |
| 2298 | min = rep_min[c]; /* Pick up values from tables; */ | min = rep_min[c]; /* Pick up values from tables; */ |
| 2299 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 2300 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2301 | break; | break; |
| 2302 | ||
| 2303 | case OP_CRRANGE: | case OP_CRRANGE: |
| 2304 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
| 2305 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
| 2306 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
| 2307 | max = GET2(ecode, 3); | max = GET2(ecode, 3); |
| 2308 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2309 | ecode += 5; | ecode += 5; |
| 2310 | break; | break; |
| 2311 | ||
| 2312 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| 2313 | if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); | if ((length = match_ref(offset, eptr, length, md, ims)) < 0) |
| 2314 | eptr += length; | { |
| 2315 | continue; /* With the main loop */ | CHECK_PARTIAL(); |
| 2316 | MRRETURN(MATCH_NOMATCH); | |
| 2317 | } | } |
| 2318 | eptr += length; | |
| 2319 | continue; /* With the main loop */ | |
| 2320 | } | |
| 2321 | ||
| 2322 | /* If the length of the reference is zero, just continue with the | /* Handle repeated back references. If the length of the reference is |
| 2323 | main loop. */ | zero, just continue with the main loop. */ |
| 2324 | ||
| 2325 | if (length == 0) continue; | if (length == 0) continue; |
| 2326 | ||
| 2327 | /* First, ensure the minimum number of matches are present. We get back | /* First, ensure the minimum number of matches are present. We get back |
| 2328 | the length of the reference string explicitly rather than passing the | the length of the reference string explicitly rather than passing the |
| 2329 | address of eptr, so that eptr can be a register variable. */ | address of eptr, so that eptr can be a register variable. */ |
| 2330 | ||
| 2331 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2332 | { | |
| 2333 | int slength; | |
| 2334 | if ((slength = match_ref(offset, eptr, length, md, ims)) < 0) | |
| 2335 | { | { |
| 2336 | if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2337 | eptr += length; | MRRETURN(MATCH_NOMATCH); |
| 2338 | } | } |
| 2339 | eptr += slength; | |
| 2340 | } | |
| 2341 | ||
| 2342 | /* If min = max, continue at the same level without recursion. | /* If min = max, continue at the same level without recursion. |
| 2343 | They are not both allowed to be zero. */ | They are not both allowed to be zero. */ |
| 2344 | ||
| 2345 | if (min == max) continue; | if (min == max) continue; |
| 2346 | ||
| 2347 | /* If minimizing, keep trying and advancing the pointer */ | /* If minimizing, keep trying and advancing the pointer */ |
| 2348 | ||
| 2349 | if (minimize) | if (minimize) |
| 2350 | { | |
| 2351 | for (fi = min;; fi++) | |
| 2352 | { | { |
| 2353 | for (fi = min;; fi++) | int slength; |
| 2354 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14); | |
| 2355 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 2356 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 2357 | if ((slength = match_ref(offset, eptr, length, md, ims)) < 0) | |
| 2358 | { | { |
| 2359 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14); | CHECK_PARTIAL(); |
| 2360 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | MRRETURN(MATCH_NOMATCH); |
| if (fi >= max || !match_ref(offset, eptr, length, md, ims)) | ||
| RRETURN(MATCH_NOMATCH); | ||
| eptr += length; | ||
| 2361 | } | } |
| 2362 | /* Control never gets here */ | eptr += slength; |
| 2363 | } | } |
| 2364 | /* Control never gets here */ | |
| 2365 | } | |
| 2366 | ||
| 2367 | /* If maximizing, find the longest string and work backwards */ | /* If maximizing, find the longest string and work backwards */ |
| 2368 | ||
| 2369 | else | else |
| 2370 | { | |
| 2371 | pp = eptr; | |
| 2372 | for (i = min; i < max; i++) | |
| 2373 | { | { |
| 2374 | pp = eptr; | int slength; |
| 2375 | for (i = min; i < max; i++) | if ((slength = match_ref(offset, eptr, length, md, ims)) < 0) |
| 2376 | { | { |
| 2377 | if (!match_ref(offset, eptr, length, md, ims)) break; | CHECK_PARTIAL(); |
| 2378 | eptr += length; | break; |
| } | ||
| while (eptr >= pp) | ||
| { | ||
| RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15); | ||
| if (rrc != MATCH_NOMATCH) RRETURN(rrc); | ||
| eptr -= length; | ||
| 2379 | } | } |
| 2380 | RRETURN(MATCH_NOMATCH); | eptr += slength; |
| 2381 | } | } |
| 2382 | while (eptr >= pp) | |
| 2383 | { | |
| 2384 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15); | |
| 2385 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 2386 | eptr -= length; | |
| 2387 | } | |
| 2388 | MRRETURN(MATCH_NOMATCH); | |
| 2389 | } | } |
| 2390 | /* Control never gets here */ | /* Control never gets here */ |
| 2391 | ||
| 2392 | /* Match a bit-mapped character class, possibly repeatedly. This op code is | /* Match a bit-mapped character class, possibly repeatedly. This op code is |
| 2393 | used when all the characters in the class have values in the range 0-255, | used when all the characters in the class have values in the range 0-255, |
| 2394 | and either the matching is caseful, or the characters are in the range | and either the matching is caseful, or the characters are in the range |
| # | Line 1958 for (;;) | Line 2443 for (;;) |
| 2443 | { | { |
| 2444 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2445 | { | { |
| 2446 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2447 | { | |
| 2448 | SCHECK_PARTIAL(); | |
| 2449 | MRRETURN(MATCH_NOMATCH); | |
| 2450 | } | |
| 2451 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2452 | if (c > 255) | if (c > 255) |
| 2453 | { | { |
| 2454 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); |
| 2455 | } | } |
| 2456 | else | else |
| 2457 | { | { |
| 2458 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2459 | } | } |
| 2460 | } | } |
| 2461 | } | } |
| # | Line 1976 for (;;) | Line 2465 for (;;) |
| 2465 | { | { |
| 2466 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2467 | { | { |
| 2468 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2469 | { | |
| 2470 | SCHECK_PARTIAL(); | |
| 2471 | MRRETURN(MATCH_NOMATCH); | |
| 2472 | } | |
| 2473 | c = *eptr++; | c = *eptr++; |
| 2474 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2475 | } | } |
| 2476 | } | } |
| 2477 | ||
| # | Line 2000 for (;;) | Line 2493 for (;;) |
| 2493 | { | { |
| 2494 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16); |
| 2495 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2496 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2497 | if (eptr >= md->end_subject) | |
| 2498 | { | |
| 2499 | SCHECK_PARTIAL(); | |
| 2500 | MRRETURN(MATCH_NOMATCH); | |
| 2501 | } | |
| 2502 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2503 | if (c > 255) | if (c > 255) |
| 2504 | { | { |
| 2505 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); |
| 2506 | } | } |
| 2507 | else | else |
| 2508 | { | { |
| 2509 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2510 | } | } |
| 2511 | } | } |
| 2512 | } | } |
| # | Line 2020 for (;;) | Line 2518 for (;;) |
| 2518 | { | { |
| 2519 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17); |
| 2520 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2521 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2522 | if (eptr >= md->end_subject) | |
| 2523 | { | |
| 2524 | SCHECK_PARTIAL(); | |
| 2525 | MRRETURN(MATCH_NOMATCH); | |
| 2526 | } | |
| 2527 | c = *eptr++; | c = *eptr++; |
| 2528 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2529 | } | } |
| 2530 | } | } |
| 2531 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2041 for (;;) | Line 2544 for (;;) |
| 2544 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2545 | { | { |
| 2546 | int len = 1; | int len = 1; |
| 2547 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2548 | { | |
| 2549 | SCHECK_PARTIAL(); | |
| 2550 | break; | |
| 2551 | } | |
| 2552 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 2553 | if (c > 255) | if (c > 255) |
| 2554 | { | { |
| # | Line 2067 for (;;) | Line 2574 for (;;) |
| 2574 | { | { |
| 2575 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2576 | { | { |
| 2577 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2578 | { | |
| 2579 | SCHECK_PARTIAL(); | |
| 2580 | break; | |
| 2581 | } | |
| 2582 | c = *eptr; | c = *eptr; |
| 2583 | if ((data[c/8] & (1 << (c&7))) == 0) break; | if ((data[c/8] & (1 << (c&7))) == 0) break; |
| 2584 | eptr++; | eptr++; |
| # | Line 2080 for (;;) | Line 2591 for (;;) |
| 2591 | } | } |
| 2592 | } | } |
| 2593 | ||
| 2594 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2595 | } | } |
| 2596 | } | } |
| 2597 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2129 for (;;) | Line 2640 for (;;) |
| 2640 | ||
| 2641 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2642 | { | { |
| 2643 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2644 | { | |
| 2645 | SCHECK_PARTIAL(); | |
| 2646 | MRRETURN(MATCH_NOMATCH); | |
| 2647 | } | |
| 2648 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2649 | if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); |
| 2650 | } | } |
| 2651 | ||
| 2652 | /* If max == min we can continue with the main loop without the | /* If max == min we can continue with the main loop without the |
| # | Line 2148 for (;;) | Line 2663 for (;;) |
| 2663 | { | { |
| 2664 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20); |
| 2665 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2666 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2667 | if (eptr >= md->end_subject) | |
| 2668 | { | |
| 2669 | SCHECK_PARTIAL(); | |
| 2670 | MRRETURN(MATCH_NOMATCH); | |
| 2671 | } | |
| 2672 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2673 | if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); |
| 2674 | } | } |
| 2675 | /* Control never gets here */ | /* Control never gets here */ |
| 2676 | } | } |
| # | Line 2163 for (;;) | Line 2683 for (;;) |
| 2683 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2684 | { | { |
| 2685 | int len = 1; | int len = 1; |
| 2686 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2687 | { | |
| 2688 | SCHECK_PARTIAL(); | |
| 2689 | break; | |
| 2690 | } | |
| 2691 | GETCHARLENTEST(c, eptr, len); | GETCHARLENTEST(c, eptr, len); |
| 2692 | if (!_pcre_xclass(c, data)) break; | if (!_pcre_xclass(c, data)) break; |
| 2693 | eptr += len; | eptr += len; |
| # | Line 2175 for (;;) | Line 2699 for (;;) |
| 2699 | if (eptr-- == pp) break; /* Stop if tried at original pos */ | if (eptr-- == pp) break; /* Stop if tried at original pos */ |
| 2700 | if (utf8) BACKCHAR(eptr); | if (utf8) BACKCHAR(eptr); |
| 2701 | } | } |
| 2702 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2703 | } | } |
| 2704 | ||
| 2705 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2191 for (;;) | Line 2715 for (;;) |
| 2715 | length = 1; | length = 1; |
| 2716 | ecode++; | ecode++; |
| 2717 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 2718 | if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | if (length > md->end_subject - eptr) |
| 2719 | while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 2720 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | |
| 2721 | MRRETURN(MATCH_NOMATCH); | |
| 2722 | } | |
| 2723 | while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2724 | } | } |
| 2725 | else | else |
| 2726 | #endif | #endif |
| 2727 | ||
| 2728 | /* Non-UTF-8 mode */ | /* Non-UTF-8 mode */ |
| 2729 | { | { |
| 2730 | if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); | if (md->end_subject - eptr < 1) |
| 2731 | if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 2732 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | |
| 2733 | MRRETURN(MATCH_NOMATCH); | |
| 2734 | } | |
| 2735 | if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2736 | ecode += 2; | ecode += 2; |
| 2737 | } | } |
| 2738 | break; | break; |
| # | Line 2215 for (;;) | Line 2747 for (;;) |
| 2747 | ecode++; | ecode++; |
| 2748 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 2749 | ||
| 2750 | if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | if (length > md->end_subject - eptr) |
| 2751 | { | |
| 2752 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | |
| 2753 | MRRETURN(MATCH_NOMATCH); | |
| 2754 | } | |
| 2755 | ||
| 2756 | /* 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 |
| 2757 | can use the fast lookup table. */ | can use the fast lookup table. */ |
| 2758 | ||
| 2759 | if (fc < 128) | if (fc < 128) |
| 2760 | { | { |
| 2761 | if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); |
| 2762 | } | } |
| 2763 | ||
| 2764 | /* Otherwise we must pick up the subject character */ | /* Otherwise we must pick up the subject character */ |
| # | Line 2241 for (;;) | Line 2777 for (;;) |
| 2777 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2778 | if (dc != UCD_OTHERCASE(fc)) | if (dc != UCD_OTHERCASE(fc)) |
| 2779 | #endif | #endif |
| 2780 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2781 | } | } |
| 2782 | } | } |
| 2783 | } | } |
| # | Line 2250 for (;;) | Line 2786 for (;;) |
| 2786 | ||
| 2787 | /* Non-UTF-8 mode */ | /* Non-UTF-8 mode */ |
| 2788 | { | { |
| 2789 | if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); | if (md->end_subject - eptr < 1) |
| 2790 | if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 2791 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | |
| 2792 | MRRETURN(MATCH_NOMATCH); | |
| 2793 | } | |
| 2794 | if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2795 | ecode += 2; | ecode += 2; |
| 2796 | } | } |
| 2797 | break; | break; |
| # | Line 2304 for (;;) | Line 2844 for (;;) |
| 2844 | case OP_MINQUERY: | case OP_MINQUERY: |
| 2845 | c = *ecode++ - OP_STAR; | c = *ecode++ - OP_STAR; |
| 2846 | minimize = (c & 1) != 0; | minimize = (c & 1) != 0; |
| 2847 | ||
| 2848 | min = rep_min[c]; /* Pick up values from tables; */ | min = rep_min[c]; /* Pick up values from tables; */ |
| 2849 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 2850 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2851 | ||
| 2852 | /* Common code for all repeated single-character matches. We can give | /* Common code for all repeated single-character matches. */ |
| up quickly if there are fewer than the minimum number of characters left in | ||
| the subject. */ | ||
| 2853 | ||
| 2854 | REPEATCHAR: | REPEATCHAR: |
| 2855 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 2319 for (;;) | Line 2858 for (;;) |
| 2858 | length = 1; | length = 1; |
| 2859 | charptr = ecode; | charptr = ecode; |
| 2860 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 2861 | ecode += length; | ecode += length; |
| 2862 | ||
| 2863 | /* Handle multibyte character matching specially here. There is | /* Handle multibyte character matching specially here. There is |
| # | Line 2337 for (;;) | Line 2875 for (;;) |
| 2875 | ||
| 2876 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2877 | { | { |
| 2878 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | if (eptr <= md->end_subject - length && |
| 2879 | memcmp(eptr, charptr, length) == 0) eptr += length; | |
| 2880 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2881 | /* Need braces because of following else */ | else if (oclength > 0 && |
| 2882 | else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } | eptr <= md->end_subject - oclength && |
| 2883 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2884 | #endif /* SUPPORT_UCP */ | |
| 2885 | else | else |
| 2886 | { | { |
| 2887 | if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2888 | eptr += oclength; | MRRETURN(MATCH_NOMATCH); |
| 2889 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else { RRETURN(MATCH_NOMATCH); } | ||
| #endif /* SUPPORT_UCP */ | ||
| 2890 | } | } |
| 2891 | ||
| 2892 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2359 for (;;) | Line 2897 for (;;) |
| 2897 | { | { |
| 2898 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22); |
| 2899 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2900 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2901 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | if (eptr <= md->end_subject - length && |
| 2902 | memcmp(eptr, charptr, length) == 0) eptr += length; | |
| 2903 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2904 | /* Need braces because of following else */ | else if (oclength > 0 && |
| 2905 | else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } | eptr <= md->end_subject - oclength && |
| 2906 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2907 | #endif /* SUPPORT_UCP */ | |
| 2908 | else | else |
| 2909 | { | { |
| 2910 | if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2911 | eptr += oclength; | MRRETURN(MATCH_NOMATCH); |
| 2912 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else { RRETURN (MATCH_NOMATCH); } | ||
| #endif /* SUPPORT_UCP */ | ||
| 2913 | } | } |
| 2914 | /* Control never gets here */ | /* Control never gets here */ |
| 2915 | } | } |
| # | Line 2381 for (;;) | Line 2919 for (;;) |
| 2919 | pp = eptr; | pp = eptr; |
| 2920 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2921 | { | { |
| 2922 | if (eptr > md->end_subject - length) break; | if (eptr <= md->end_subject - length && |
| 2923 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, length) == 0) eptr += length; |
| 2924 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2925 | else if (oclength == 0) break; | else if (oclength > 0 && |
| 2926 | eptr <= md->end_subject - oclength && | |
| 2927 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2928 | #endif /* SUPPORT_UCP */ | |
| 2929 | else | else |
| 2930 | { | { |
| 2931 | if (memcmp(eptr, occhars, oclength) != 0) break; | CHECK_PARTIAL(); |
| 2932 | eptr += oclength; | break; |
| 2933 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else break; | ||
| #endif /* SUPPORT_UCP */ | ||
| 2934 | } | } |
| 2935 | ||
| 2936 | if (possessive) continue; | if (possessive) continue; |
| 2937 | ||
| 2938 | for(;;) | for(;;) |
| 2939 | { | { |
| 2940 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23); |
| 2941 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2942 | if (eptr == pp) RRETURN(MATCH_NOMATCH); | if (eptr == pp) { MRRETURN(MATCH_NOMATCH); } |
| 2943 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2944 | eptr--; | eptr--; |
| 2945 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 2946 | #else /* without SUPPORT_UCP */ | #else /* without SUPPORT_UCP */ |
| 2947 | eptr -= length; | eptr -= length; |
| 2948 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 2949 | } | } |
| 2950 | } | } |
| 2951 | /* Control never gets here */ | /* Control never gets here */ |
| 2952 | } | } |
| # | Line 2420 for (;;) | Line 2959 for (;;) |
| 2959 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF8 */ |
| 2960 | ||
| 2961 | /* When not in UTF-8 mode, load a single-byte character. */ | /* When not in UTF-8 mode, load a single-byte character. */ |
| 2962 | { | |
| 2963 | if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | fc = *ecode++; |
| fc = *ecode++; | ||
| } | ||
| 2964 | ||
| 2965 | /* The value of fc at this point is always less than 256, though we may or | /* The value of fc at this point is always less than 256, though we may or |
| 2966 | may not be in UTF-8 mode. The code is duplicated for the caseless and | may not be in UTF-8 mode. The code is duplicated for the caseless and |
| # | Line 2441 for (;;) | Line 2978 for (;;) |
| 2978 | { | { |
| 2979 | fc = md->lcc[fc]; | fc = md->lcc[fc]; |
| 2980 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2981 | if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 2982 | if (eptr >= md->end_subject) | |
| 2983 | { | |
| 2984 | SCHECK_PARTIAL(); | |
| 2985 | MRRETURN(MATCH_NOMATCH); | |
| 2986 | } | |
| 2987 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2988 | } | |
| 2989 | if (min == max) continue; | if (min == max) continue; |
| 2990 | if (minimize) | if (minimize) |
| 2991 | { | { |
| # | Line 2449 for (;;) | Line 2993 for (;;) |
| 2993 | { | { |
| 2994 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24); |
| 2995 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2996 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2997 | fc != md->lcc[*eptr++]) | if (eptr >= md->end_subject) |
| 2998 | RRETURN(MATCH_NOMATCH); | { |
| 2999 | SCHECK_PARTIAL(); | |
| 3000 | MRRETURN(MATCH_NOMATCH); | |
| 3001 | } | |
| 3002 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3003 | } | } |
| 3004 | /* Control never gets here */ | /* Control never gets here */ |
| 3005 | } | } |
| # | Line 2460 for (;;) | Line 3008 for (;;) |
| 3008 | pp = eptr; | pp = eptr; |
| 3009 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3010 | { | { |
| 3011 | if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break; | if (eptr >= md->end_subject) |
| 3012 | { | |
| 3013 | SCHECK_PARTIAL(); | |
| 3014 | break; | |
| 3015 | } | |
| 3016 | if (fc != md->lcc[*eptr]) break; | |
| 3017 | eptr++; | eptr++; |
| 3018 | } | } |
| 3019 | ||
| 3020 | if (possessive) continue; | if (possessive) continue; |
| 3021 | ||
| 3022 | while (eptr >= pp) | while (eptr >= pp) |
| 3023 | { | { |
| 3024 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25); |
| 3025 | eptr--; | eptr--; |
| 3026 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3027 | } | } |
| 3028 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3029 | } | } |
| 3030 | /* Control never gets here */ | /* Control never gets here */ |
| 3031 | } | } |
| # | Line 2479 for (;;) | Line 3034 for (;;) |
| 3034 | ||
| 3035 | else | else |
| 3036 | { | { |
| 3037 | for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH); | for (i = 1; i <= min; i++) |
| 3038 | { | |
| 3039 | if (eptr >= md->end_subject) | |
| 3040 | { | |
| 3041 | SCHECK_PARTIAL(); | |
| 3042 | MRRETURN(MATCH_NOMATCH); | |
| 3043 | } | |
| 3044 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3045 | } | |
| 3046 | ||
| 3047 | if (min == max) continue; | if (min == max) continue; |
| 3048 | ||
| 3049 | if (minimize) | if (minimize) |
| 3050 | { | { |
| 3051 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3052 | { | { |
| 3053 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26); |
| 3054 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3055 | if (fi >= max || eptr >= md->end_subject || fc != *eptr++) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3056 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3057 | { | |
| 3058 | SCHECK_PARTIAL(); | |
| 3059 | MRRETURN(MATCH_NOMATCH); | |
| 3060 | } | |
| 3061 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3062 | } | } |
| 3063 | /* Control never gets here */ | /* Control never gets here */ |
| 3064 | } | } |
| # | Line 2497 for (;;) | Line 3067 for (;;) |
| 3067 | pp = eptr; | pp = eptr; |
| 3068 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3069 | { | { |
| 3070 | if (eptr >= md->end_subject || fc != *eptr) break; | if (eptr >= md->end_subject) |
| 3071 | { | |
| 3072 | SCHECK_PARTIAL(); | |
| 3073 | break; | |
| 3074 | } | |
| 3075 | if (fc != *eptr) break; | |
| 3076 | eptr++; | eptr++; |
| 3077 | } | } |
| 3078 | if (possessive) continue; | if (possessive) continue; |
| 3079 | ||
| 3080 | while (eptr >= pp) | while (eptr >= pp) |
| 3081 | { | { |
| 3082 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27); |
| 3083 | eptr--; | eptr--; |
| 3084 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3085 | } | } |
| 3086 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3087 | } | } |
| 3088 | } | } |
| 3089 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2516 for (;;) | Line 3092 for (;;) |
| 3092 | checking can be multibyte. */ | checking can be multibyte. */ |
| 3093 | ||
| 3094 | case OP_NOT: | case OP_NOT: |
| 3095 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3096 | { | |
| 3097 | SCHECK_PARTIAL(); | |
| 3098 | MRRETURN(MATCH_NOMATCH); | |
| 3099 | } | |
| 3100 | ecode++; | ecode++; |
| 3101 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3102 | if ((ims & PCRE_CASELESS) != 0) | if ((ims & PCRE_CASELESS) != 0) |
| # | Line 2525 for (;;) | Line 3105 for (;;) |
| 3105 | if (c < 256) | if (c < 256) |
| 3106 | #endif | #endif |
| 3107 | c = md->lcc[c]; | c = md->lcc[c]; |
| 3108 | if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH); | if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH); |
| 3109 | } | } |
| 3110 | else | else |
| 3111 | { | { |
| 3112 | if (*ecode++ == c) RRETURN(MATCH_NOMATCH); | if (*ecode++ == c) MRRETURN(MATCH_NOMATCH); |
| 3113 | } | } |
| 3114 | break; | break; |
| 3115 | ||
| # | Line 2593 for (;;) | Line 3173 for (;;) |
| 3173 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 3174 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 3175 | ||
| 3176 | /* Common code for all repeated single-byte matches. We can give up quickly | /* Common code for all repeated single-byte matches. */ |
| if there are fewer than the minimum number of bytes left in the | ||
| subject. */ | ||
| 3177 | ||
| 3178 | REPEATNOTCHAR: | REPEATNOTCHAR: |
| if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 3179 | fc = *ecode++; | fc = *ecode++; |
| 3180 | ||
| 3181 | /* The code is duplicated for the caseless and caseful cases, for speed, | /* The code is duplicated for the caseless and caseful cases, for speed, |
| # | Line 2623 for (;;) | Line 3200 for (;;) |
| 3200 | register unsigned int d; | register unsigned int d; |
| 3201 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3202 | { | { |
| 3203 | if (eptr >= md->end_subject) | |
| 3204 | { | |
| 3205 | SCHECK_PARTIAL(); | |
| 3206 | MRRETURN(MATCH_NOMATCH); | |
| 3207 | } | |
| 3208 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3209 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3210 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3211 | } | } |
| 3212 | } | } |
| 3213 | else | else |
| # | Line 2634 for (;;) | Line 3216 for (;;) |
| 3216 | /* Not UTF-8 mode */ | /* Not UTF-8 mode */ |
| 3217 | { | { |
| 3218 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3219 | if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 3220 | if (eptr >= md->end_subject) | |
| 3221 | { | |
| 3222 | SCHECK_PARTIAL(); | |
| 3223 | MRRETURN(MATCH_NOMATCH); | |
| 3224 | } | |
| 3225 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3226 | } | |
| 3227 | } | } |
| 3228 | ||
| 3229 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2650 for (;;) | Line 3239 for (;;) |
| 3239 | { | { |
| 3240 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28); |
| 3241 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3242 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3243 | if (eptr >= md->end_subject) | |
| 3244 | { | |
| 3245 | SCHECK_PARTIAL(); | |
| 3246 | MRRETURN(MATCH_NOMATCH); | |
| 3247 | } | |
| 3248 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3249 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3250 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3251 | } | } |
| 3252 | } | } |
| 3253 | else | else |
| # | Line 2665 for (;;) | Line 3258 for (;;) |
| 3258 | { | { |
| 3259 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29); |
| 3260 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3261 | if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++]) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3262 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3263 | { | |
| 3264 | SCHECK_PARTIAL(); | |
| 3265 | MRRETURN(MATCH_NOMATCH); | |
| 3266 | } | |
| 3267 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3268 | } | } |
| 3269 | } | } |
| 3270 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2686 for (;;) | Line 3284 for (;;) |
| 3284 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3285 | { | { |
| 3286 | int len = 1; | int len = 1; |
| 3287 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 3288 | { | |
| 3289 | SCHECK_PARTIAL(); | |
| 3290 | break; | |
| 3291 | } | |
| 3292 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3293 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3294 | if (fc == d) break; | if (fc == d) break; |
| # | Line 2707 for (;;) | Line 3309 for (;;) |
| 3309 | { | { |
| 3310 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3311 | { | { |
| 3312 | if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break; | if (eptr >= md->end_subject) |
| 3313 | { | |
| 3314 | SCHECK_PARTIAL(); | |
| 3315 | break; | |
| 3316 | } | |
| 3317 | if (fc == md->lcc[*eptr]) break; | |
| 3318 | eptr++; | eptr++; |
| 3319 | } | } |
| 3320 | if (possessive) continue; | if (possessive) continue; |
| # | Line 2719 for (;;) | Line 3326 for (;;) |
| 3326 | } | } |
| 3327 | } | } |
| 3328 | ||
| 3329 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3330 | } | } |
| 3331 | /* Control never gets here */ | /* Control never gets here */ |
| 3332 | } | } |
| # | Line 2735 for (;;) | Line 3342 for (;;) |
| 3342 | register unsigned int d; | register unsigned int d; |
| 3343 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3344 | { | { |
| 3345 | if (eptr >= md->end_subject) | |
| 3346 | { | |
| 3347 | SCHECK_PARTIAL(); | |
| 3348 | MRRETURN(MATCH_NOMATCH); | |
| 3349 | } | |
| 3350 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3351 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3352 | } | } |
| 3353 | } | } |
| 3354 | else | else |
| # | Line 2744 for (;;) | Line 3356 for (;;) |
| 3356 | /* Not UTF-8 mode */ | /* Not UTF-8 mode */ |
| 3357 | { | { |
| 3358 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3359 | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 3360 | if (eptr >= md->end_subject) | |
| 3361 | { | |
| 3362 | SCHECK_PARTIAL(); | |
| 3363 | MRRETURN(MATCH_NOMATCH); | |
| 3364 | } | |
| 3365 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3366 | } | |
| 3367 | } | } |
| 3368 | ||
| 3369 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2760 for (;;) | Line 3379 for (;;) |
| 3379 | { | { |
| 3380 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32); |
| 3381 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3382 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3383 | if (eptr >= md->end_subject) | |
| 3384 | { | |
| 3385 | SCHECK_PARTIAL(); | |
| 3386 | MRRETURN(MATCH_NOMATCH); | |
| 3387 | } | |
| 3388 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3389 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3390 | } | } |
| 3391 | } | } |
| 3392 | else | else |
| # | Line 2773 for (;;) | Line 3397 for (;;) |
| 3397 | { | { |
| 3398 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33); |
| 3399 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3400 | if (fi >= max || eptr >= md->end_subject || fc == *eptr++) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3401 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3402 | { | |
| 3403 | SCHECK_PARTIAL(); | |
| 3404 | MRRETURN(MATCH_NOMATCH); | |
| 3405 | } | |
| 3406 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3407 | } | } |
| 3408 | } | } |
| 3409 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2794 for (;;) | Line 3423 for (;;) |
| 3423 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3424 | { | { |
| 3425 | int len = 1; | int len = 1; |
| 3426 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 3427 | { | |
| 3428 | SCHECK_PARTIAL(); | |
| 3429 | break; | |
| 3430 | } | |
| 3431 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3432 | if (fc == d) break; | if (fc == d) break; |
| 3433 | eptr += len; | eptr += len; |
| # | Line 2814 for (;;) | Line 3447 for (;;) |
| 3447 | { | { |
| 3448 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3449 | { | { |
| 3450 | if (eptr >= md->end_subject || fc == *eptr) break; | if (eptr >= md->end_subject) |
| 3451 | { | |
| 3452 | SCHECK_PARTIAL(); | |
| 3453 | break; | |
| 3454 | } | |
| 3455 | if (fc == *eptr) break; | |
| 3456 | eptr++; | eptr++; |
| 3457 | } | } |
| 3458 | if (possessive) continue; | if (possessive) continue; |
| # | Line 2826 for (;;) | Line 3464 for (;;) |
| 3464 | } | } |
| 3465 | } | } |
| 3466 | ||
| 3467 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3468 | } | } |
| 3469 | } | } |
| 3470 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2908 for (;;) | Line 3546 for (;;) |
| 3546 | ||
| 3547 | /* First, ensure the minimum number of matches are present. Use inline | /* First, ensure the minimum number of matches are present. Use inline |
| 3548 | code for maximizing the speed, and do the type test once at the start | code for maximizing the speed, and do the type test once at the start |
| 3549 | (i.e. keep it out of the loop). Also we can test that there are at least | (i.e. keep it out of the loop). Separate the UTF-8 code completely as that |
| the minimum number of bytes before we start. This isn't as effective in | ||
| UTF-8 mode, but it does no harm. Separate the UTF-8 code completely as that | ||
| 3550 | is tidier. Also separate the UCP code, which can be the same for both UTF-8 | is tidier. Also separate the UCP code, which can be the same for both UTF-8 |
| 3551 | and single-bytes. */ | and single-bytes. */ |
| 3552 | ||
| if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 3553 | if (min > 0) | if (min > 0) |
| 3554 | { | { |
| 3555 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| # | Line 2923 for (;;) | Line 3558 for (;;) |
| 3558 | switch(prop_type) | switch(prop_type) |
| 3559 | { | { |
| 3560 | case PT_ANY: | case PT_ANY: |
| 3561 | if (prop_fail_result) RRETURN(MATCH_NOMATCH); | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); |
| 3562 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3563 | { | { |
| 3564 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3565 | { | |
| 3566 | SCHECK_PARTIAL(); | |
| 3567 | MRRETURN(MATCH_NOMATCH); | |
| 3568 | } | |
| 3569 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3570 | } | } |
| 3571 | break; | break; |
| # | Line 2934 for (;;) | Line 3573 for (;;) |
| 3573 | case PT_LAMP: | case PT_LAMP: |
| 3574 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3575 | { | { |
| 3576 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3577 | { | |
| 3578 | SCHECK_PARTIAL(); | |
| 3579 | MRRETURN(MATCH_NOMATCH); | |
| 3580 | } | |
| 3581 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3582 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 3583 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 3584 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| 3585 | prop_chartype == ucp_Lt) == prop_fail_result) | prop_chartype == ucp_Lt) == prop_fail_result) |
| 3586 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3587 | } | } |
| 3588 | break; | break; |
| 3589 | ||
| 3590 | case PT_GC: | case PT_GC: |
| 3591 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3592 | { | { |
| 3593 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3594 | { | |
| 3595 | SCHECK_PARTIAL(); | |
| 3596 | MRRETURN(MATCH_NOMATCH); | |
| 3597 | } | |
| 3598 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3599 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3600 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| 3601 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3602 | } | } |
| 3603 | break; | break; |
| 3604 | ||
| 3605 | case PT_PC: | case PT_PC: |
| 3606 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3607 | { | { |
| 3608 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3609 | { | |
| 3610 | SCHECK_PARTIAL(); | |
| 3611 | MRRETURN(MATCH_NOMATCH); | |
| 3612 | } | |
| 3613 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3614 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 3615 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| 3616 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3617 | } | } |
| 3618 | break; | break; |
| 3619 | ||
| 3620 | case PT_SC: | case PT_SC: |
| 3621 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3622 | { | { |
| 3623 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3624 | { | |
| 3625 | SCHECK_PARTIAL(); | |
| 3626 | MRRETURN(MATCH_NOMATCH); | |
| 3627 | } | |
| 3628 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3629 | prop_script = UCD_SCRIPT(c); | prop_script = UCD_SCRIPT(c); |
| 3630 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| 3631 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3632 | } | |
| 3633 | break; | |
| 3634 | ||
| 3635 | case PT_ALNUM: | |
| 3636 | for (i = 1; i <= min; i++) | |
| 3637 | { | |
| 3638 | if (eptr >= md->end_subject) | |
| 3639 | { | |
| 3640 | SCHECK_PARTIAL(); | |
| 3641 | MRRETURN(MATCH_NOMATCH); | |
| 3642 | } | |
| 3643 | GETCHARINCTEST(c, eptr); | |
| 3644 | prop_category = UCD_CATEGORY(c); | |
| 3645 | if ((prop_category == ucp_L || prop_category == ucp_N) | |
| 3646 | == prop_fail_result) | |
| 3647 | MRRETURN(MATCH_NOMATCH); | |
| 3648 | } | } |
| 3649 | break; | break; |
| 3650 | ||
| 3651 | case PT_SPACE: /* Perl space */ | |
| 3652 | for (i = 1; i <= min; i++) | |
| 3653 | { | |
| 3654 | if (eptr >= md->end_subject) | |
| 3655 | { | |
| 3656 | SCHECK_PARTIAL(); | |
| 3657 | MRRETURN(MATCH_NOMATCH); | |
| 3658 | } | |
| 3659 | GETCHARINCTEST(c, eptr); | |
| 3660 | prop_category = UCD_CATEGORY(c); | |
| 3661 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 3662 | c == CHAR_FF || c == CHAR_CR) | |
| 3663 | == prop_fail_result) | |
| 3664 | MRRETURN(MATCH_NOMATCH); | |
| 3665 | } | |
| 3666 | break; | |
| 3667 | ||
| 3668 | case PT_PXSPACE: /* POSIX space */ | |
| 3669 | for (i = 1; i <= min; i++) | |
| 3670 | { | |
| 3671 | if (eptr >= md->end_subject) | |
| 3672 | { | |
| 3673 | SCHECK_PARTIAL(); | |
| 3674 | MRRETURN(MATCH_NOMATCH); | |
| 3675 | } | |
| 3676 | GETCHARINCTEST(c, eptr); | |
| 3677 | prop_category = UCD_CATEGORY(c); | |
| 3678 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 3679 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | |
| 3680 | == prop_fail_result) | |
| 3681 | MRRETURN(MATCH_NOMATCH); | |
| 3682 | } | |
| 3683 | break; | |
| 3684 | ||
| 3685 | case PT_WORD: | |
| 3686 | for (i = 1; i <= min; i++) | |
| 3687 | { | |
| 3688 | if (eptr >= md->end_subject) | |
| 3689 | { | |
| 3690 | SCHECK_PARTIAL(); | |
| 3691 | MRRETURN(MATCH_NOMATCH); | |
| 3692 | } | |
| 3693 | GETCHARINCTEST(c, eptr); | |
| 3694 | prop_category = UCD_CATEGORY(c); | |
| 3695 | if ((prop_category == ucp_L || prop_category == ucp_N || | |
| 3696 | c == CHAR_UNDERSCORE) | |
| 3697 | == prop_fail_result) | |
| 3698 | MRRETURN(MATCH_NOMATCH); | |
| 3699 | } | |
| 3700 | break; | |
| 3701 | ||
| 3702 | /* This should not occur */ | |
| 3703 | ||
| 3704 | default: | default: |
| 3705 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 3706 | } | } |
| # | Line 2989 for (;;) | Line 3713 for (;;) |
| 3713 | { | { |
| 3714 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3715 | { | { |
| 3716 | if (eptr >= md->end_subject) | |
| 3717 | { | |
| 3718 | SCHECK_PARTIAL(); | |
| 3719 | MRRETURN(MATCH_NOMATCH); | |
| 3720 | } | |
| 3721 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3722 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3723 | if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); | if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 3724 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 3725 | { | { |
| 3726 | int len = 1; | int len = 1; |
| 3727 | if (!utf8) c = *eptr; else | if (!utf8) c = *eptr; |
| 3728 | { | else { GETCHARLEN(c, eptr, len); } |
| GETCHARLEN(c, eptr, len); | ||
| } | ||
| 3729 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3730 | if (prop_category != ucp_M) break; | if (prop_category != ucp_M) break; |
| 3731 | eptr += len; | eptr += len; |
| # | Line 3017 for (;;) | Line 3744 for (;;) |
| 3744 | case OP_ANY: | case OP_ANY: |
| 3745 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3746 | { | { |
| 3747 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) | if (eptr >= md->end_subject) |
| 3748 | RRETURN(MATCH_NOMATCH); | { |
| 3749 | SCHECK_PARTIAL(); | |
| 3750 | MRRETURN(MATCH_NOMATCH); | |
| 3751 | } | |
| 3752 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | |
| 3753 | eptr++; | eptr++; |
| 3754 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 3755 | } | } |
| # | Line 3027 for (;;) | Line 3758 for (;;) |
| 3758 | case OP_ALLANY: | case OP_ALLANY: |
| 3759 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3760 | { | { |
| 3761 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3762 | { | |
| 3763 | SCHECK_PARTIAL(); | |
| 3764 | MRRETURN(MATCH_NOMATCH); | |
| 3765 | } | |
| 3766 | eptr++; | eptr++; |
| 3767 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 3768 | } | } |
| 3769 | break; | break; |
| 3770 | ||
| 3771 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 3772 | if (eptr > md->end_subject - min) MRRETURN(MATCH_NOMATCH); | |
| 3773 | eptr += min; | eptr += min; |
| 3774 | break; | break; |
| 3775 | ||
| 3776 | case OP_ANYNL: | case OP_ANYNL: |
| 3777 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3778 | { | { |
| 3779 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3780 | { | |
| 3781 | SCHECK_PARTIAL(); | |
| 3782 | MRRETURN(MATCH_NOMATCH); | |
| 3783 | } | |
| 3784 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3785 | switch(c) | switch(c) |
| 3786 | { | { |
| 3787 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3788 | case 0x000d: | case 0x000d: |
| 3789 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 3790 | break; | break; |
| # | Line 3057 for (;;) | Line 3797 for (;;) |
| 3797 | case 0x0085: | case 0x0085: |
| 3798 | case 0x2028: | case 0x2028: |
| 3799 | case 0x2029: | case 0x2029: |
| 3800 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 3801 | break; | break; |
| 3802 | } | } |
| 3803 | } | } |
| # | Line 3066 for (;;) | Line 3806 for (;;) |
| 3806 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 3807 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3808 | { | { |
| 3809 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3810 | { | |
| 3811 | SCHECK_PARTIAL(); | |
| 3812 | MRRETURN(MATCH_NOMATCH); | |
| 3813 | } | |
| 3814 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3815 | switch(c) | switch(c) |
| 3816 | { | { |
| # | Line 3090 for (;;) | Line 3834 for (;;) |
| 3834 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 3835 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 3836 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 3837 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3838 | } | } |
| 3839 | } | } |
| 3840 | break; | break; |
| # | Line 3098 for (;;) | Line 3842 for (;;) |
| 3842 | case OP_HSPACE: | case OP_HSPACE: |
| 3843 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3844 | { | { |
| 3845 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3846 | { | |
| 3847 | SCHECK_PARTIAL(); | |
| 3848 | MRRETURN(MATCH_NOMATCH); | |
| 3849 | } | |
| 3850 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3851 | switch(c) | switch(c) |
| 3852 | { | { |
| 3853 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3854 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 3855 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 3856 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3130 for (;;) | Line 3878 for (;;) |
| 3878 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 3879 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3880 | { | { |
| 3881 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3882 | { | |
| 3883 | SCHECK_PARTIAL(); | |
| 3884 | MRRETURN(MATCH_NOMATCH); | |
| 3885 | } | |
| 3886 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3887 | switch(c) | switch(c) |
| 3888 | { | { |
| # | Line 3142 for (;;) | Line 3894 for (;;) |
| 3894 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 3895 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 3896 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 3897 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3898 | } | } |
| 3899 | } | } |
| 3900 | break; | break; |
| # | Line 3150 for (;;) | Line 3902 for (;;) |
| 3902 | case OP_VSPACE: | case OP_VSPACE: |
| 3903 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3904 | { | { |
| 3905 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3906 | { | |
| 3907 | SCHECK_PARTIAL(); | |
| 3908 | MRRETURN(MATCH_NOMATCH); | |
| 3909 | } | |
| 3910 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3911 | switch(c) | switch(c) |
| 3912 | { | { |
| 3913 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3914 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 3915 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 3916 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3170 for (;;) | Line 3926 for (;;) |
| 3926 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 3927 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3928 | { | { |
| 3929 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3930 | { | |
| 3931 | SCHECK_PARTIAL(); | |
| 3932 | MRRETURN(MATCH_NOMATCH); | |
| 3933 | } | |
| 3934 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3935 | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) |
| 3936 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3937 | } | } |
| 3938 | break; | break; |
| 3939 | ||
| 3940 | case OP_DIGIT: | case OP_DIGIT: |
| 3941 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3942 | { | { |
| 3943 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3944 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | { |
| 3945 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3946 | MRRETURN(MATCH_NOMATCH); | |
| 3947 | } | |
| 3948 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | |
| 3949 | MRRETURN(MATCH_NOMATCH); | |
| 3950 | /* 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 */ |
| 3951 | } | } |
| 3952 | break; | break; |
| # | Line 3190 for (;;) | Line 3954 for (;;) |
| 3954 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 3955 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3956 | { | { |
| 3957 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3958 | (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0)) | { |
| 3959 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3960 | MRRETURN(MATCH_NOMATCH); | |
| 3961 | } | |
| 3962 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) | |
| 3963 | MRRETURN(MATCH_NOMATCH); | |
| 3964 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); |
| 3965 | } | } |
| 3966 | break; | break; |
| # | Line 3200 for (;;) | Line 3968 for (;;) |
| 3968 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 3969 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3970 | { | { |
| 3971 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3972 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | { |
| 3973 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3974 | MRRETURN(MATCH_NOMATCH); | |
| 3975 | } | |
| 3976 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | |
| 3977 | MRRETURN(MATCH_NOMATCH); | |
| 3978 | /* 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 */ |
| 3979 | } | } |
| 3980 | break; | break; |
| # | Line 3210 for (;;) | Line 3982 for (;;) |
| 3982 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 3983 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3984 | { | { |
| 3985 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3986 | (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0)) | { |
| 3987 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3988 | MRRETURN(MATCH_NOMATCH); | |
| 3989 | } | |
| 3990 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) | |
| 3991 | MRRETURN(MATCH_NOMATCH); | |
| 3992 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); |
| 3993 | } | } |
| 3994 | break; | break; |
| # | Line 3220 for (;;) | Line 3996 for (;;) |
| 3996 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 3997 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3998 | { | { |
| 3999 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 4000 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | { |
| 4001 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 4002 | MRRETURN(MATCH_NOMATCH); | |
| 4003 | } | |
| 4004 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | |
| 4005 | MRRETURN(MATCH_NOMATCH); | |
| 4006 | /* 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 */ |
| 4007 | } | } |
| 4008 | break; | break; |
| # | Line 3235 for (;;) | Line 4015 for (;;) |
| 4015 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF8 */ |
| 4016 | ||
| 4017 | /* 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 |
| 4018 | than OP_PROP and OP_NOTPROP. We can assume that there are the minimum | than OP_PROP and OP_NOTPROP. */ |
| number of bytes present, as this was tested above. */ | ||
| 4019 | ||
| 4020 | switch(ctype) | switch(ctype) |
| 4021 | { | { |
| 4022 | case OP_ANY: | case OP_ANY: |
| 4023 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4024 | { | { |
| 4025 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4026 | { | |
| 4027 | SCHECK_PARTIAL(); | |
| 4028 | MRRETURN(MATCH_NOMATCH); | |
| 4029 | } | |
| 4030 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | |
| 4031 | eptr++; | eptr++; |
| 4032 | } | } |
| 4033 | break; | break; |
| 4034 | ||
| 4035 | case OP_ALLANY: | case OP_ALLANY: |
| 4036 | if (eptr > md->end_subject - min) | |
| 4037 | { | |
| 4038 | SCHECK_PARTIAL(); | |
| 4039 | MRRETURN(MATCH_NOMATCH); | |
| 4040 | } | |
| 4041 | eptr += min; | eptr += min; |
| 4042 | break; | break; |
| 4043 | ||
| 4044 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 4045 | if (eptr > md->end_subject - min) | |
| 4046 | { | |
| 4047 | SCHECK_PARTIAL(); | |
| 4048 | MRRETURN(MATCH_NOMATCH); | |
| 4049 | } | |
| 4050 | eptr += min; | eptr += min; |
| 4051 | break; | break; |
| 4052 | ||
| /* Because of the CRLF case, we can't assume the minimum number of | ||
| bytes are present in this case. */ | ||
| 4053 | case OP_ANYNL: | case OP_ANYNL: |
| 4054 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4055 | { | { |
| 4056 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4057 | { | |
| 4058 | SCHECK_PARTIAL(); | |
| 4059 | MRRETURN(MATCH_NOMATCH); | |
| 4060 | } | |
| 4061 | switch(*eptr++) | switch(*eptr++) |
| 4062 | { | { |
| 4063 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4064 | case 0x000d: | case 0x000d: |
| 4065 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4066 | break; | break; |
| # | Line 3275 for (;;) | Line 4070 for (;;) |
| 4070 | case 0x000b: | case 0x000b: |
| 4071 | case 0x000c: | case 0x000c: |
| 4072 | case 0x0085: | case 0x0085: |
| 4073 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4074 | break; | break; |
| 4075 | } | } |
| 4076 | } | } |
| # | Line 3284 for (;;) | Line 4079 for (;;) |
| 4079 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 4080 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4081 | { | { |
| 4082 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4083 | { | |
| 4084 | SCHECK_PARTIAL(); | |
| 4085 | MRRETURN(MATCH_NOMATCH); | |
| 4086 | } | |
| 4087 | switch(*eptr++) | switch(*eptr++) |
| 4088 | { | { |
| 4089 | default: break; | default: break; |
| 4090 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4091 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4092 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4093 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4094 | } | } |
| 4095 | } | } |
| 4096 | break; | break; |
| # | Line 3299 for (;;) | Line 4098 for (;;) |
| 4098 | case OP_HSPACE: | case OP_HSPACE: |
| 4099 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4100 | { | { |
| 4101 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4102 | { | |
| 4103 | SCHECK_PARTIAL(); | |
| 4104 | MRRETURN(MATCH_NOMATCH); | |
| 4105 | } | |
| 4106 | switch(*eptr++) | switch(*eptr++) |
| 4107 | { | { |
| 4108 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4109 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4110 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4111 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3314 for (;;) | Line 4117 for (;;) |
| 4117 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 4118 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4119 | { | { |
| 4120 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4121 | { | |
| 4122 | SCHECK_PARTIAL(); | |
| 4123 | MRRETURN(MATCH_NOMATCH); | |
| 4124 | } | |
| 4125 | switch(*eptr++) | switch(*eptr++) |
| 4126 | { | { |
| 4127 | default: break; | default: break; |
| # | Line 3323 for (;;) | Line 4130 for (;;) |
| 4130 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4131 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4132 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4133 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4134 | } | } |
| 4135 | } | } |
| 4136 | break; | break; |
| # | Line 3331 for (;;) | Line 4138 for (;;) |
| 4138 | case OP_VSPACE: | case OP_VSPACE: |
| 4139 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4140 | { | { |
| 4141 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4142 | { | |
| 4143 | SCHECK_PARTIAL(); | |
| 4144 | MRRETURN(MATCH_NOMATCH); | |
| 4145 | } | |
| 4146 | switch(*eptr++) | switch(*eptr++) |
| 4147 | { | { |
| 4148 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4149 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4150 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4151 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3347 for (;;) | Line 4158 for (;;) |
| 4158 | ||
| 4159 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4160 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4161 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); | { |
| 4162 | if (eptr >= md->end_subject) | |
| 4163 | { | |
| 4164 | SCHECK_PARTIAL(); | |
| 4165 | MRRETURN(MATCH_NOMATCH); | |
| 4166 | } | |
| 4167 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); | |
| 4168 | } | |
| 4169 | break; | break; |
| 4170 | ||
| 4171 | case OP_DIGIT: | case OP_DIGIT: |
| 4172 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4173 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); | { |
| 4174 | if (eptr >= md->end_subject) | |
| 4175 | { | |
| 4176 | SCHECK_PARTIAL(); | |
| 4177 | MRRETURN(MATCH_NOMATCH); | |
| 4178 | } | |
| 4179 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); | |
| 4180 | } | |
| 4181 | break; | break; |
| 4182 | ||
| 4183 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4184 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4185 | if ((md->ctypes[*eptr++] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); | { |
| 4186 | if (eptr >= md->end_subject) | |
| 4187 | { | |
| 4188 | SCHECK_PARTIAL(); | |
| 4189 | MRRETURN(MATCH_NOMATCH); | |
| 4190 | } | |
| 4191 | if ((md->ctypes[*eptr++] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); | |
| 4192 | } | |
| 4193 | break; | break; |
| 4194 | ||
| 4195 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4196 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4197 | if ((md->ctypes[*eptr++] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); | { |
| 4198 | if (eptr >= md->end_subject) | |
| 4199 | { | |
| 4200 | SCHECK_PARTIAL(); | |
| 4201 | MRRETURN(MATCH_NOMATCH); | |
| 4202 | } | |
| 4203 | if ((md->ctypes[*eptr++] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); | |
| 4204 | } | |
| 4205 | break; | break; |
| 4206 | ||
| 4207 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4208 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4209 | { | |
| 4210 | if (eptr >= md->end_subject) | |
| 4211 | { | |
| 4212 | SCHECK_PARTIAL(); | |
| 4213 | MRRETURN(MATCH_NOMATCH); | |
| 4214 | } | |
| 4215 | if ((md->ctypes[*eptr++] & ctype_word) != 0) | if ((md->ctypes[*eptr++] & ctype_word) != 0) |
| 4216 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4217 | } | |
| 4218 | break; | break; |
| 4219 | ||
| 4220 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4221 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4222 | { | |
| 4223 | if (eptr >= md->end_subject) | |
| 4224 | { | |
| 4225 | SCHECK_PARTIAL(); | |
| 4226 | MRRETURN(MATCH_NOMATCH); | |
| 4227 | } | |
| 4228 | if ((md->ctypes[*eptr++] & ctype_word) == 0) | if ((md->ctypes[*eptr++] & ctype_word) == 0) |
| 4229 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4230 | } | |
| 4231 | break; | break; |
| 4232 | ||
| 4233 | default: | default: |
| # | Line 3402 for (;;) | Line 4255 for (;;) |
| 4255 | { | { |
| 4256 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36); |
| 4257 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4258 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4259 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4260 | if (prop_fail_result) RRETURN(MATCH_NOMATCH); | { |
| 4261 | SCHECK_PARTIAL(); | |
| 4262 | MRRETURN(MATCH_NOMATCH); | |
| 4263 | } | |
| 4264 | GETCHARINCTEST(c, eptr); | |
| 4265 | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); | |
| 4266 | } | } |
| 4267 | /* Control never gets here */ | /* Control never gets here */ |
| 4268 | ||
| # | Line 3413 for (;;) | Line 4271 for (;;) |
| 4271 | { | { |
| 4272 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37); |
| 4273 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4274 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4275 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4276 | { | |
| 4277 | SCHECK_PARTIAL(); | |
| 4278 | MRRETURN(MATCH_NOMATCH); | |
| 4279 | } | |
| 4280 | GETCHARINCTEST(c, eptr); | |
| 4281 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4282 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 4283 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| 4284 | prop_chartype == ucp_Lt) == prop_fail_result) | prop_chartype == ucp_Lt) == prop_fail_result) |
| 4285 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4286 | } | |
| 4287 | /* Control never gets here */ | |
| 4288 | ||
| 4289 | case PT_GC: | |
| 4290 | for (fi = min;; fi++) | |
| 4291 | { | |
| 4292 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38); | |
| 4293 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4294 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4295 | if (eptr >= md->end_subject) | |
| 4296 | { | |
| 4297 | SCHECK_PARTIAL(); | |
| 4298 | MRRETURN(MATCH_NOMATCH); | |
| 4299 | } | |
| 4300 | GETCHARINCTEST(c, eptr); | |
| 4301 | prop_category = UCD_CATEGORY(c); | |
| 4302 | if ((prop_category == prop_value) == prop_fail_result) | |
| 4303 | MRRETURN(MATCH_NOMATCH); | |
| 4304 | } | |
| 4305 | /* Control never gets here */ | |
| 4306 | ||
| 4307 | case PT_PC: | |
| 4308 | for (fi = min;; fi++) | |
| 4309 | { | |
| 4310 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39); | |
| 4311 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4312 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4313 | if (eptr >= md->end_subject) | |
| 4314 | { | |
| 4315 | SCHECK_PARTIAL(); | |
| 4316 | MRRETURN(MATCH_NOMATCH); | |
| 4317 | } | |
| 4318 | GETCHARINCTEST(c, eptr); | |
| 4319 | prop_chartype = UCD_CHARTYPE(c); | |
| 4320 | if ((prop_chartype == prop_value) == prop_fail_result) | |
| 4321 | MRRETURN(MATCH_NOMATCH); | |
| 4322 | } | |
| 4323 | /* Control never gets here */ | |
| 4324 | ||
| 4325 | case PT_SC: | |
| 4326 | for (fi = min;; fi++) | |
| 4327 | { | |
| 4328 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40); | |
| 4329 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4330 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4331 | if (eptr >= md->end_subject) | |
| 4332 | { | |
| 4333 | SCHECK_PARTIAL(); | |
| 4334 | MRRETURN(MATCH_NOMATCH); | |
| 4335 | } | |
| 4336 | GETCHARINCTEST(c, eptr); | |
| 4337 | prop_script = UCD_SCRIPT(c); | |
| 4338 | if ((prop_script == prop_value) == prop_fail_result) | |
| 4339 | MRRETURN(MATCH_NOMATCH); | |
| 4340 | } | |
| 4341 | /* Control never gets here */ | |
| 4342 | ||
| 4343 | case PT_ALNUM: | |
| 4344 | for (fi = min;; fi++) | |
| 4345 | { | |
| 4346 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM59); | |
| 4347 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4348 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4349 | if (eptr >= md->end_subject) | |
| 4350 | { | |
| 4351 | SCHECK_PARTIAL(); | |
| 4352 | MRRETURN(MATCH_NOMATCH); | |
| 4353 | } | |
| 4354 | GETCHARINCTEST(c, eptr); | |
| 4355 | prop_category = UCD_CATEGORY(c); | |
| 4356 | if ((prop_category == ucp_L || prop_category == ucp_N) | |
| 4357 | == prop_fail_result) | |
| 4358 | MRRETURN(MATCH_NOMATCH); | |
| 4359 | } | } |
| 4360 | /* Control never gets here */ | /* Control never gets here */ |
| 4361 | ||
| 4362 | case PT_GC: | case PT_SPACE: /* Perl space */ |
| 4363 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 4364 | { | { |
| 4365 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM60); |
| 4366 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4367 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4368 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4369 | { | |
| 4370 | SCHECK_PARTIAL(); | |
| 4371 | MRRETURN(MATCH_NOMATCH); | |
| 4372 | } | |
| 4373 | GETCHARINCTEST(c, eptr); | |
| 4374 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4375 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
| 4376 | RRETURN(MATCH_NOMATCH); | c == CHAR_FF || c == CHAR_CR) |
| 4377 | == prop_fail_result) | |
| 4378 | MRRETURN(MATCH_NOMATCH); | |
| 4379 | } | } |
| 4380 | /* Control never gets here */ | /* Control never gets here */ |
| 4381 | ||
| 4382 | case PT_PC: | case PT_PXSPACE: /* POSIX space */ |
| 4383 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 4384 | { | { |
| 4385 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM61); |
| 4386 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4387 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4388 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4389 | prop_chartype = UCD_CHARTYPE(c); | { |
| 4390 | if ((prop_chartype == prop_value) == prop_fail_result) | SCHECK_PARTIAL(); |
| 4391 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4392 | } | |
| 4393 | GETCHARINCTEST(c, eptr); | |
| 4394 | prop_category = UCD_CATEGORY(c); | |
| 4395 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 4396 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | |
| 4397 | == prop_fail_result) | |
| 4398 | MRRETURN(MATCH_NOMATCH); | |
| 4399 | } | } |
| 4400 | /* Control never gets here */ | /* Control never gets here */ |
| 4401 | ||
| 4402 | case PT_SC: | case PT_WORD: |
| 4403 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 4404 | { | { |
| 4405 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM62); |
| 4406 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4407 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4408 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4409 | prop_script = UCD_SCRIPT(c); | { |
| 4410 | if ((prop_script == prop_value) == prop_fail_result) | SCHECK_PARTIAL(); |
| 4411 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4412 | } | |
| 4413 | GETCHARINCTEST(c, eptr); | |
| 4414 | prop_category = UCD_CATEGORY(c); | |
| 4415 | if ((prop_category == ucp_L || | |
| 4416 | prop_category == ucp_N || | |
| 4417 | c == CHAR_UNDERSCORE) | |
| 4418 | == prop_fail_result) | |
| 4419 | MRRETURN(MATCH_NOMATCH); | |
| 4420 | } | } |
| 4421 | /* Control never gets here */ | /* Control never gets here */ |
| 4422 | ||
| 4423 | /* This should never occur */ | |
| 4424 | ||
| 4425 | default: | default: |
| 4426 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 4427 | } | } |
| # | Line 3476 for (;;) | Line 4436 for (;;) |
| 4436 | { | { |
| 4437 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41); |
| 4438 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4439 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4440 | if (eptr >= md->end_subject) | |
| 4441 | { | |
| 4442 | SCHECK_PARTIAL(); | |
| 4443 | MRRETURN(MATCH_NOMATCH); | |
| 4444 | } | |
| 4445 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4446 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4447 | if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); | if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 4448 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 4449 | { | { |
| 4450 | int len = 1; | int len = 1; |
| 4451 | if (!utf8) c = *eptr; else | if (!utf8) c = *eptr; |
| 4452 | { | else { GETCHARLEN(c, eptr, len); } |
| GETCHARLEN(c, eptr, len); | ||
| } | ||
| 4453 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4454 | if (prop_category != ucp_M) break; | if (prop_category != ucp_M) break; |
| 4455 | eptr += len; | eptr += len; |
| # | Line 3505 for (;;) | Line 4468 for (;;) |
| 4468 | { | { |
| 4469 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42); |
| 4470 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4471 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4472 | (ctype == OP_ANY && IS_NEWLINE(eptr))) | if (eptr >= md->end_subject) |
| 4473 | RRETURN(MATCH_NOMATCH); | { |
| 4474 | SCHECK_PARTIAL(); | |
| 4475 | MRRETURN(MATCH_NOMATCH); | |
| 4476 | } | |
| 4477 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | |
| 4478 | MRRETURN(MATCH_NOMATCH); | |
| 4479 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4480 | switch(ctype) | switch(ctype) |
| 4481 | { | { |
| # | Line 3520 for (;;) | Line 4487 for (;;) |
| 4487 | case OP_ANYNL: | case OP_ANYNL: |
| 4488 | switch(c) | switch(c) |
| 4489 | { | { |
| 4490 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4491 | case 0x000d: | case 0x000d: |
| 4492 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4493 | break; | break; |
| # | Line 3532 for (;;) | Line 4499 for (;;) |
| 4499 | case 0x0085: | case 0x0085: |
| 4500 | case 0x2028: | case 0x2028: |
| 4501 | case 0x2029: | case 0x2029: |
| 4502 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4503 | break; | break; |
| 4504 | } | } |
| 4505 | break; | break; |
| # | Line 3560 for (;;) | Line 4527 for (;;) |
| 4527 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 4528 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 4529 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 4530 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4531 | } | } |
| 4532 | break; | break; |
| 4533 | ||
| 4534 | case OP_HSPACE: | case OP_HSPACE: |
| 4535 | switch(c) | switch(c) |
| 4536 | { | { |
| 4537 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4538 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4539 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4540 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3602 for (;;) | Line 4569 for (;;) |
| 4569 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4570 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 4571 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 4572 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4573 | } | } |
| 4574 | break; | break; |
| 4575 | ||
| 4576 | case OP_VSPACE: | case OP_VSPACE: |
| 4577 | switch(c) | switch(c) |
| 4578 | { | { |
| 4579 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4580 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4581 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4582 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3623 for (;;) | Line 4590 for (;;) |
| 4590 | ||
| 4591 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4592 | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) |
| 4593 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4594 | break; | break; |
| 4595 | ||
| 4596 | case OP_DIGIT: | case OP_DIGIT: |
| 4597 | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) |
| 4598 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4599 | break; | break; |
| 4600 | ||
| 4601 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4602 | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) |
| 4603 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4604 | break; | break; |
| 4605 | ||
| 4606 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4607 | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) |
| 4608 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4609 | break; | break; |
| 4610 | ||
| 4611 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4612 | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) |
| 4613 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4614 | break; | break; |
| 4615 | ||
| 4616 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4617 | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) |
| 4618 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4619 | break; | break; |
| 4620 | ||
| 4621 | default: | default: |
| # | Line 3664 for (;;) | Line 4631 for (;;) |
| 4631 | { | { |
| 4632 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43); |
| 4633 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4634 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4635 | (ctype == OP_ANY && IS_NEWLINE(eptr))) | if (eptr >= md->end_subject) |
| 4636 | RRETURN(MATCH_NOMATCH); | { |
| 4637 | SCHECK_PARTIAL(); | |
| 4638 | MRRETURN(MATCH_NOMATCH); | |
| 4639 | } | |
| 4640 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | |
| 4641 | MRRETURN(MATCH_NOMATCH); | |
| 4642 | c = *eptr++; | c = *eptr++; |
| 4643 | switch(ctype) | switch(ctype) |
| 4644 | { | { |
| # | Line 3679 for (;;) | Line 4650 for (;;) |
| 4650 | case OP_ANYNL: | case OP_ANYNL: |
| 4651 | switch(c) | switch(c) |
| 4652 | { | { |
| 4653 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4654 | case 0x000d: | case 0x000d: |
| 4655 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4656 | break; | break; |
| # | Line 3690 for (;;) | Line 4661 for (;;) |
| 4661 | case 0x000b: | case 0x000b: |
| 4662 | case 0x000c: | case 0x000c: |
| 4663 | case 0x0085: | case 0x0085: |
| 4664 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4665 | break; | break; |
| 4666 |