Parent Directory
|
Revision Log
|
Patch
| revision 403 by ph10, Sat Mar 21 17:33:11 2009 UTC | revision 597 by ph10, Mon May 2 17:08:52 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 | if (eptr >= md->end_subject) return -1; | |
| 197 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 198 | GETCHARINC(d, p); | GETCHARINC(d, p); |
| 199 | if (c != d && c != UCD_OTHERCASE(d)) return FALSE; | if (c != d && c != UCD_OTHERCASE(d)) return -1; |
| 200 | } | } |
| 201 | } | } |
| 202 | else | else |
| # | Line 183 if ((ims & PCRE_CASELESS) != 0) | Line 205 if ((ims & PCRE_CASELESS) != 0) |
| 205 | ||
| 206 | /* The same code works when not in UTF-8 mode and in UTF-8 mode when there | /* The same code works when not in UTF-8 mode and in UTF-8 mode when there |
| 207 | is no UCP support. */ | is no UCP support. */ |
| 208 | { | |
| 209 | while (length-- > 0) | if (eptr + length > md->end_subject) return -1; |
| 210 | { if (md->lcc[*p++] != md->lcc[*eptr++]) return FALSE; } | while (length-- > 0) |
| 211 | { if (md->lcc[*p++] != md->lcc[*eptr++]) return -1; } | |
| 212 | } | |
| 213 | } | } |
| 214 | ||
| 215 | /* 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 |
| 216 | are in UTF-8 mode. */ | are in UTF-8 mode. */ |
| 217 | ||
| 218 | else | else |
| 219 | { while (length-- > 0) if (*p++ != *eptr++) return FALSE; } | { |
| 220 | if (eptr + length > md->end_subject) return -1; | |
| 221 | while (length-- > 0) if (*p++ != *eptr++) return -1; | |
| 222 | } | |
| 223 | ||
| 224 | return TRUE; | return eptr - eptr_start; |
| 225 | } | } |
| 226 | ||
| 227 | ||
| # | Line 245 enum { RM1=1, RM2, RM3, RM4, RM5, RM | Line 272 enum { RM1=1, RM2, RM3, RM4, RM5, RM |
| 272 | RM21, RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30, | RM21, RM22, RM23, RM24, RM25, RM26, RM27, RM28, RM29, RM30, |
| 273 | RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, | RM31, RM32, RM33, RM34, RM35, RM36, RM37, RM38, RM39, RM40, |
| 274 | RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, | RM41, RM42, RM43, RM44, RM45, RM46, RM47, RM48, RM49, RM50, |
| 275 | RM51, RM52, RM53, RM54 }; | RM51, RM52, RM53, RM54, RM55, RM56, RM57, RM58, RM59, RM60, |
| 276 | RM61, RM62 }; | |
| 277 | ||
| 278 | /* 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 |
| 279 | 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 |
| 280 | actuall used in this definition. */ | actually used in this definition. */ |
| 281 | ||
| 282 | #ifndef NO_RECURSE | #ifndef NO_RECURSE |
| 283 | #define REGISTER register | #define REGISTER register |
| 284 | ||
| 285 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 286 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ |
| 287 | { \ | { \ |
| 288 | printf("match() called in line %d\n", __LINE__); \ | printf("match() called in line %d\n", __LINE__); \ |
| 289 | rrc = match(ra,rb,mstart,rc,rd,re,rf,rg,rdepth+1); \ | rrc = match(ra,rb,mstart,markptr,rc,rd,re,rf,rg,rdepth+1); \ |
| 290 | printf("to line %d\n", __LINE__); \ | printf("to line %d\n", __LINE__); \ |
| 291 | } | } |
| 292 | #define RRETURN(ra) \ | #define RRETURN(ra) \ |
| # | Line 268 actuall used in this definition. */ | Line 296 actuall used in this definition. */ |
| 296 | } | } |
| 297 | #else | #else |
| 298 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw) \ |
| 299 | 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) |
| 300 | #define RRETURN(ra) return ra | #define RRETURN(ra) return ra |
| 301 | #endif | #endif |
| 302 | ||
| # | Line 283 argument of match(), which never changes | Line 311 argument of match(), which never changes |
| 311 | ||
| 312 | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\ | #define RMATCH(ra,rb,rc,rd,re,rf,rg,rw)\ |
| 313 | {\ | {\ |
| 314 | heapframe *newframe = (pcre_stack_malloc)(sizeof(heapframe));\ | heapframe *newframe = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe));\ |
| 315 | if (newframe == NULL) RRETURN(PCRE_ERROR_NOMEMORY);\ | |
| 316 | frame->Xwhere = rw; \ | frame->Xwhere = rw; \ |
| 317 | newframe->Xeptr = ra;\ | newframe->Xeptr = ra;\ |
| 318 | newframe->Xecode = rb;\ | newframe->Xecode = rb;\ |
| 319 | newframe->Xmstart = mstart;\ | newframe->Xmstart = mstart;\ |
| 320 | newframe->Xmarkptr = markptr;\ | |
| 321 | newframe->Xoffset_top = rc;\ | newframe->Xoffset_top = rc;\ |
| 322 | newframe->Xims = re;\ | newframe->Xims = re;\ |
| 323 | newframe->Xeptrb = rf;\ | newframe->Xeptrb = rf;\ |
| # | Line 303 argument of match(), which never changes | Line 333 argument of match(), which never changes |
| 333 | ||
| 334 | #define RRETURN(ra)\ | #define RRETURN(ra)\ |
| 335 | {\ | {\ |
| 336 | heapframe *newframe = frame;\ | heapframe *oldframe = frame;\ |
| 337 | frame = newframe->Xprevframe;\ | frame = oldframe->Xprevframe;\ |
| 338 | (pcre_stack_free)(newframe);\ | (pcre_stack_free)(oldframe);\ |
| 339 | if (frame != NULL)\ | if (frame != NULL)\ |
| 340 | {\ | {\ |
| 341 | rrc = ra;\ | rrc = ra;\ |
| # | Line 322 typedef struct heapframe { | Line 352 typedef struct heapframe { |
| 352 | ||
| 353 | /* Function arguments that may change */ | /* Function arguments that may change */ |
| 354 | ||
| 355 | const uschar *Xeptr; | USPTR Xeptr; |
| 356 | const uschar *Xecode; | const uschar *Xecode; |
| 357 | const uschar *Xmstart; | USPTR Xmstart; |
| 358 | USPTR Xmarkptr; | |
| 359 | int Xoffset_top; | int Xoffset_top; |
| 360 | long int Xims; | long int Xims; |
| 361 | eptrblock *Xeptrb; | eptrblock *Xeptrb; |
| # | Line 333 typedef struct heapframe { | Line 364 typedef struct heapframe { |
| 364 | ||
| 365 | /* Function local variables */ | /* Function local variables */ |
| 366 | ||
| 367 | const uschar *Xcallpat; | USPTR Xcallpat; |
| 368 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 369 | const uschar *Xcharptr; | USPTR Xcharptr; |
| 370 | #endif | #endif |
| 371 | const uschar *Xdata; | USPTR Xdata; |
| 372 | const uschar *Xnext; | USPTR Xnext; |
| 373 | const uschar *Xpp; | USPTR Xpp; |
| 374 | const uschar *Xprev; | USPTR Xprev; |
| 375 | const uschar *Xsaved_eptr; | USPTR Xsaved_eptr; |
| 376 | ||
| 377 | recursion_info Xnew_recursive; | recursion_info Xnew_recursive; |
| 378 | ||
| # | Line 398 typedef struct heapframe { | Line 429 typedef struct heapframe { |
| 429 | ||
| 430 | /* This function is called recursively in many circumstances. Whenever it | /* This function is called recursively in many circumstances. Whenever it |
| 431 | returns a negative (error) response, the outer incarnation must also return the | returns a negative (error) response, the outer incarnation must also return the |
| 432 | same response. | same response. */ |
| 433 | ||
| 434 | /* These macros pack up tests that are used for partial matching, and which | |
| 435 | appears several times in the code. We set the "hit end" flag if the pointer is | |
| 436 | at the end of the subject and also past the start of the subject (i.e. | |
| 437 | something has been matched). For hard partial matching, we then return | |
| 438 | immediately. The second one is used when we already know we are past the end of | |
| 439 | the subject. */ | |
| 440 | ||
| 441 | #define CHECK_PARTIAL()\ | |
| 442 | if (md->partial != 0 && eptr >= md->end_subject && \ | |
| 443 | eptr > md->start_used_ptr) \ | |
| 444 | { \ | |
| 445 | md->hitend = TRUE; \ | |
| 446 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \ | |
| 447 | } | |
| 448 | ||
| 449 | Performance note: It might be tempting to extract commonly used fields from the | #define SCHECK_PARTIAL()\ |
| 450 | md structure (e.g. utf8, end_subject) into individual variables to improve | if (md->partial != 0 && eptr > md->start_used_ptr) \ |
| 451 | { \ | |
| 452 | md->hitend = TRUE; \ | |
| 453 | if (md->partial > 1) MRRETURN(PCRE_ERROR_PARTIAL); \ | |
| 454 | } | |
| 455 | ||
| 456 | ||
| 457 | /* Performance note: It might be tempting to extract commonly used fields from | |
| 458 | the md structure (e.g. utf8, end_subject) into individual variables to improve | |
| 459 | 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 |
| 460 | made performance worse. | made performance worse. |
| 461 | ||
| # | Line 410 Arguments: | Line 464 Arguments: |
| 464 | ecode pointer to current position in compiled code | ecode pointer to current position in compiled code |
| 465 | mstart pointer to the current match start position (can be modified | mstart pointer to the current match start position (can be modified |
| 466 | by encountering \K) | by encountering \K) |
| 467 | markptr pointer to the most recent MARK name, or NULL | |
| 468 | offset_top current top pointer | offset_top current top pointer |
| 469 | md pointer to "static" info for the match | md pointer to "static" info for the match |
| 470 | ims current /i, /m, and /s options | ims current /i, /m, and /s options |
| # | Line 423 Arguments: | Line 478 Arguments: |
| 478 | ||
| 479 | Returns: MATCH_MATCH if matched ) these values are >= 0 | Returns: MATCH_MATCH if matched ) these values are >= 0 |
| 480 | MATCH_NOMATCH if failed to match ) | MATCH_NOMATCH if failed to match ) |
| 481 | a negative MATCH_xxx value for PRUNE, SKIP, etc | |
| 482 | a negative PCRE_ERROR_xxx value if aborted by an error condition | a negative PCRE_ERROR_xxx value if aborted by an error condition |
| 483 | (e.g. stopped by repeated call or recursion limit) | (e.g. stopped by repeated call or recursion limit) |
| 484 | */ | */ |
| 485 | ||
| 486 | static int | static int |
| 487 | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, const uschar *mstart, | match(REGISTER USPTR eptr, REGISTER const uschar *ecode, USPTR mstart, |
| 488 | 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, |
| 489 | int flags, unsigned int rdepth) | eptrblock *eptrb, int flags, unsigned int rdepth) |
| 490 | { | { |
| 491 | /* 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, |
| 492 | 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 506 heap storage. Set up the top-level frame |
| 506 | heap whenever RMATCH() does a "recursion". See the macro definitions above. */ | heap whenever RMATCH() does a "recursion". See the macro definitions above. */ |
| 507 | ||
| 508 | #ifdef NO_RECURSE | #ifdef NO_RECURSE |
| 509 | heapframe *frame = (pcre_stack_malloc)(sizeof(heapframe)); | heapframe *frame = (heapframe *)(pcre_stack_malloc)(sizeof(heapframe)); |
| 510 | if (frame == NULL) RRETURN(PCRE_ERROR_NOMEMORY); | |
| 511 | frame->Xprevframe = NULL; /* Marks the top level */ | frame->Xprevframe = NULL; /* Marks the top level */ |
| 512 | ||
| 513 | /* Copy in the original argument variables */ | /* Copy in the original argument variables */ |
| # | Line 458 frame->Xprevframe = NULL; /* | Line 515 frame->Xprevframe = NULL; /* |
| 515 | frame->Xeptr = eptr; | frame->Xeptr = eptr; |
| 516 | frame->Xecode = ecode; | frame->Xecode = ecode; |
| 517 | frame->Xmstart = mstart; | frame->Xmstart = mstart; |
| 518 | frame->Xmarkptr = markptr; | |
| 519 | frame->Xoffset_top = offset_top; | frame->Xoffset_top = offset_top; |
| 520 | frame->Xims = ims; | frame->Xims = ims; |
| 521 | frame->Xeptrb = eptrb; | frame->Xeptrb = eptrb; |
| # | Line 473 HEAP_RECURSE: | Line 531 HEAP_RECURSE: |
| 531 | #define eptr frame->Xeptr | #define eptr frame->Xeptr |
| 532 | #define ecode frame->Xecode | #define ecode frame->Xecode |
| 533 | #define mstart frame->Xmstart | #define mstart frame->Xmstart |
| 534 | #define markptr frame->Xmarkptr | |
| 535 | #define offset_top frame->Xoffset_top | #define offset_top frame->Xoffset_top |
| 536 | #define ims frame->Xims | #define ims frame->Xims |
| 537 | #define eptrb frame->Xeptrb | #define eptrb frame->Xeptrb |
| # | Line 600 TAIL_RECURSE: | Line 659 TAIL_RECURSE: |
| 659 | /* 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 |
| 660 | 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 |
| 661 | 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() |
| 662 | and a "return", respectively (possibly with some debugging if DEBUG is | and a "return", respectively (possibly with some debugging if PCRE_DEBUG is |
| 663 | 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 |
| 664 | 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, |
| 665 | however, impact performance when true recursion is being used. */ | however, impact performance when true recursion is being used. */ |
| # | Line 641 for (;;) | Line 700 for (;;) |
| 700 | { | { |
| 701 | minimize = possessive = FALSE; | minimize = possessive = FALSE; |
| 702 | op = *ecode; | op = *ecode; |
| /* For partial matching, remember if we ever hit the end of the subject after | ||
| matching at least one subject character. */ | ||
| if (md->partial && | ||
| eptr >= md->end_subject && | ||
| eptr > mstart) | ||
| md->hitend = TRUE; | ||
| 703 | ||
| 704 | switch(op) | switch(op) |
| 705 | { | { |
| 706 | case OP_MARK: | |
| 707 | markptr = ecode + 2; | |
| 708 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 709 | ims, eptrb, flags, RM55); | |
| 710 | ||
| 711 | /* A return of MATCH_SKIP_ARG means that matching failed at SKIP with an | |
| 712 | argument, and we must check whether that argument matches this MARK's | |
| 713 | argument. It is passed back in md->start_match_ptr (an overloading of that | |
| 714 | variable). If it does match, we reset that variable to the current subject | |
| 715 | position and return MATCH_SKIP. Otherwise, pass back the return code | |
| 716 | unaltered. */ | |
| 717 | ||
| 718 | if (rrc == MATCH_SKIP_ARG && | |
| 719 | strcmp((char *)markptr, (char *)(md->start_match_ptr)) == 0) | |
| 720 | { | |
| 721 | md->start_match_ptr = eptr; | |
| 722 | RRETURN(MATCH_SKIP); | |
| 723 | } | |
| 724 | ||
| 725 | if (md->mark == NULL) md->mark = markptr; | |
| 726 | RRETURN(rrc); | |
| 727 | ||
| 728 | case OP_FAIL: | case OP_FAIL: |
| 729 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 730 | ||
| 731 | /* COMMIT overrides PRUNE, SKIP, and THEN */ | |
| 732 | ||
| 733 | case OP_COMMIT: | |
| 734 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | |
| 735 | ims, eptrb, flags, RM52); | |
| 736 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && | |
| 737 | rrc != MATCH_SKIP && rrc != MATCH_SKIP_ARG && | |
| 738 | rrc != MATCH_THEN) | |
| 739 | RRETURN(rrc); | |
| 740 | MRRETURN(MATCH_COMMIT); | |
| 741 | ||
| 742 | /* PRUNE overrides THEN */ | |
| 743 | ||
| 744 | case OP_PRUNE: | case OP_PRUNE: |
| 745 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 746 | ims, eptrb, flags, RM51); | ims, eptrb, flags, RM51); |
| 747 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); |
| 748 | MRRETURN(MATCH_PRUNE); | |
| 749 | ||
| 750 | case OP_PRUNE_ARG: | |
| 751 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 752 | ims, eptrb, flags, RM56); | |
| 753 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | |
| 754 | md->mark = ecode + 2; | |
| 755 | RRETURN(MATCH_PRUNE); | RRETURN(MATCH_PRUNE); |
| 756 | ||
| 757 | 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); | ||
| 758 | ||
| 759 | case OP_SKIP: | case OP_SKIP: |
| 760 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 761 | ims, eptrb, flags, RM53); | ims, eptrb, flags, RM53); |
| 762 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) |
| 763 | RRETURN(rrc); | |
| 764 | md->start_match_ptr = eptr; /* Pass back current position */ | md->start_match_ptr = eptr; /* Pass back current position */ |
| 765 | RRETURN(MATCH_SKIP); | MRRETURN(MATCH_SKIP); |
| 766 | ||
| 767 | case OP_SKIP_ARG: | |
| 768 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1], offset_top, md, | |
| 769 | ims, eptrb, flags, RM57); | |
| 770 | if (rrc != MATCH_NOMATCH && rrc != MATCH_PRUNE && rrc != MATCH_THEN) | |
| 771 | RRETURN(rrc); | |
| 772 | ||
| 773 | /* Pass back the current skip name by overloading md->start_match_ptr and | |
| 774 | returning the special MATCH_SKIP_ARG return code. This will either be | |
| 775 | caught by a matching MARK, or get to the top, where it is treated the same | |
| 776 | as PRUNE. */ | |
| 777 | ||
| 778 | md->start_match_ptr = ecode + 2; | |
| 779 | RRETURN(MATCH_SKIP_ARG); | |
| 780 | ||
| 781 | /* For THEN (and THEN_ARG) we pass back the address of the bracket or | |
| 782 | the alt that is at the start of the current branch. This makes it possible | |
| 783 | to skip back past alternatives that precede the THEN within the current | |
| 784 | branch. */ | |
| 785 | ||
| 786 | case OP_THEN: | case OP_THEN: |
| 787 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 788 | ims, eptrb, flags, RM54); | ims, eptrb, flags, RM54); |
| 789 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 790 | md->start_match_ptr = ecode - GET(ecode, 1); | |
| 791 | MRRETURN(MATCH_THEN); | |
| 792 | ||
| 793 | case OP_THEN_ARG: | |
| 794 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode] + ecode[1+LINK_SIZE], | |
| 795 | offset_top, md, ims, eptrb, flags, RM58); | |
| 796 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 797 | md->start_match_ptr = ecode - GET(ecode, 1); | |
| 798 | md->mark = ecode + LINK_SIZE + 2; | |
| 799 | RRETURN(MATCH_THEN); | RRETURN(MATCH_THEN); |
| 800 | ||
| 801 | /* 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 817 for (;;) |
| 817 | number = GET2(ecode, 1+LINK_SIZE); | number = GET2(ecode, 1+LINK_SIZE); |
| 818 | offset = number << 1; | offset = number << 1; |
| 819 | ||
| 820 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 821 | printf("start bracket %d\n", number); | printf("start bracket %d\n", number); |
| 822 | printf("subject="); | printf("subject="); |
| 823 | pchars(eptr, 16, TRUE, md); | pchars(eptr, 16, TRUE, md); |
| # | Line 714 for (;;) | Line 832 for (;;) |
| 832 | save_capture_last = md->capture_last; | save_capture_last = md->capture_last; |
| 833 | ||
| 834 | DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); | DPRINTF(("saving %d %d %d\n", save_offset1, save_offset2, save_offset3)); |
| 835 | md->offset_vector[md->offset_end - number] = eptr - md->start_subject; | md->offset_vector[md->offset_end - number] = |
| 836 | (int)(eptr - md->start_subject); | |
| 837 | ||
| 838 | flags = (op == OP_SCBRA)? match_cbegroup : 0; | flags = (op == OP_SCBRA)? match_cbegroup : 0; |
| 839 | do | do |
| 840 | { | { |
| 841 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, |
| 842 | ims, eptrb, flags, RM1); | ims, eptrb, flags, RM1); |
| 843 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && |
| 844 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 845 | RRETURN(rrc); | |
| 846 | md->capture_last = save_capture_last; | md->capture_last = save_capture_last; |
| 847 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 848 | } | } |
| # | Line 733 for (;;) | Line 854 for (;;) |
| 854 | md->offset_vector[offset+1] = save_offset2; | md->offset_vector[offset+1] = save_offset2; |
| 855 | md->offset_vector[md->offset_end - number] = save_offset3; | md->offset_vector[md->offset_end - number] = save_offset3; |
| 856 | ||
| 857 | if (rrc != MATCH_THEN) md->mark = markptr; | |
| 858 | RRETURN(MATCH_NOMATCH); | RRETURN(MATCH_NOMATCH); |
| 859 | } | } |
| 860 | ||
| # | Line 772 for (;;) | Line 894 for (;;) |
| 894 | ||
| 895 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, |
| 896 | eptrb, flags, RM48); | eptrb, flags, RM48); |
| 897 | if (rrc == MATCH_NOMATCH) md->mark = markptr; | |
| 898 | RRETURN(rrc); | RRETURN(rrc); |
| 899 | } | } |
| 900 | ||
| # | Line 780 for (;;) | Line 903 for (;;) |
| 903 | ||
| 904 | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, | RMATCH(eptr, ecode + _pcre_OP_lengths[*ecode], offset_top, md, ims, |
| 905 | eptrb, flags, RM2); | eptrb, flags, RM2); |
| 906 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc != MATCH_NOMATCH && |
| 907 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 908 | RRETURN(rrc); | |
| 909 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 910 | } | } |
| 911 | /* Control never reaches here. */ | /* Control never reaches here. */ |
| # | Line 794 for (;;) | Line 919 for (;;) |
| 919 | case OP_COND: | case OP_COND: |
| 920 | case OP_SCOND: | case OP_SCOND: |
| 921 | codelink= GET(ecode, 1); | codelink= GET(ecode, 1); |
| 922 | ||
| 923 | /* Because of the way auto-callout works during compile, a callout item is | /* Because of the way auto-callout works during compile, a callout item is |
| 924 | inserted between OP_COND and an assertion condition. */ | inserted between OP_COND and an assertion condition. */ |
| 925 | ||
| # | Line 807 for (;;) | Line 932 for (;;) |
| 932 | cb.callout_number = ecode[LINK_SIZE+2]; | cb.callout_number = ecode[LINK_SIZE+2]; |
| 933 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
| 934 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
| 935 | cb.subject_length = md->end_subject - md->start_subject; | cb.subject_length = (int)(md->end_subject - md->start_subject); |
| 936 | cb.start_match = mstart - md->start_subject; | cb.start_match = (int)(mstart - md->start_subject); |
| 937 | cb.current_position = eptr - md->start_subject; | cb.current_position = (int)(eptr - md->start_subject); |
| 938 | cb.pattern_position = GET(ecode, LINK_SIZE + 3); | cb.pattern_position = GET(ecode, LINK_SIZE + 3); |
| 939 | cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE); | cb.next_item_length = GET(ecode, 3 + 2*LINK_SIZE); |
| 940 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 941 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 942 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 943 | if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH); | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
| 944 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 945 | } | } |
| 946 | ecode += _pcre_OP_lengths[OP_CALLOUT]; | ecode += _pcre_OP_lengths[OP_CALLOUT]; |
| 947 | } | } |
| 948 | ||
| 949 | condcode = ecode[LINK_SIZE+1]; | condcode = ecode[LINK_SIZE+1]; |
| 950 | ||
| 951 | /* Now see what the actual condition is */ | /* Now see what the actual condition is */ |
| 952 | ||
| 953 | if (condcode == OP_RREF) /* Recursion test */ | if (condcode == OP_RREF || condcode == OP_NRREF) /* Recursion test */ |
| 954 | { | { |
| 955 | offset = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | if (md->recursive == NULL) /* Not recursing => FALSE */ |
| 956 | condition = md->recursive != NULL && | { |
| 957 | (offset == RREF_ANY || offset == md->recursive->group_num); | condition = FALSE; |
| 958 | ecode += condition? 3 : GET(ecode, 1); | ecode += GET(ecode, 1); |
| 959 | } | |
| 960 | else | |
| 961 | { | |
| 962 | int recno = GET2(ecode, LINK_SIZE + 2); /* Recursion group number*/ | |
| 963 | condition = (recno == RREF_ANY || recno == md->recursive->group_num); | |
| 964 | ||
| 965 | /* If the test is for recursion into a specific subpattern, and it is | |
| 966 | false, but the test was set up by name, scan the table to see if the | |
| 967 | name refers to any other numbers, and test them. The condition is true | |
| 968 | if any one is set. */ | |
| 969 | ||
| 970 | if (!condition && condcode == OP_NRREF && recno != RREF_ANY) | |
| 971 | { | |
| 972 | uschar *slotA = md->name_table; | |
| 973 | for (i = 0; i < md->name_count; i++) | |
| 974 | { | |
| 975 | if (GET2(slotA, 0) == recno) break; | |
| 976 | slotA += md->name_entry_size; | |
| 977 | } | |
| 978 | ||
| 979 | /* Found a name for the number - there can be only one; duplicate | |
| 980 | names for different numbers are allowed, but not vice versa. First | |
| 981 | scan down for duplicates. */ | |
| 982 | ||
| 983 | if (i < md->name_count) | |
| 984 | { | |
| 985 | uschar *slotB = slotA; | |
| 986 | while (slotB > md->name_table) | |
| 987 | { | |
| 988 | slotB -= md->name_entry_size; | |
| 989 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 990 | { | |
| 991 | condition = GET2(slotB, 0) == md->recursive->group_num; | |
| 992 | if (condition) break; | |
| 993 | } | |
| 994 | else break; | |
| 995 | } | |
| 996 | ||
| 997 | /* Scan up for duplicates */ | |
| 998 | ||
| 999 | if (!condition) | |
| 1000 | { | |
| 1001 | slotB = slotA; | |
| 1002 | for (i++; i < md->name_count; i++) | |
| 1003 | { | |
| 1004 | slotB += md->name_entry_size; | |
| 1005 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1006 | { | |
| 1007 | condition = GET2(slotB, 0) == md->recursive->group_num; | |
| 1008 | if (condition) break; | |
| 1009 | } | |
| 1010 | else break; | |
| 1011 | } | |
| 1012 | } | |
| 1013 | } | |
| 1014 | } | |
| 1015 | ||
| 1016 | /* Chose branch according to the condition */ | |
| 1017 | ||
| 1018 | ecode += condition? 3 : GET(ecode, 1); | |
| 1019 | } | |
| 1020 | } | } |
| 1021 | ||
| 1022 | else if (condcode == OP_CREF) /* Group used test */ | else if (condcode == OP_CREF || condcode == OP_NCREF) /* Group used test */ |
| 1023 | { | { |
| 1024 | offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ | offset = GET2(ecode, LINK_SIZE+2) << 1; /* Doubled ref number */ |
| 1025 | condition = offset < offset_top && md->offset_vector[offset] >= 0; | condition = offset < offset_top && md->offset_vector[offset] >= 0; |
| 1026 | ||
| 1027 | /* If the numbered capture is unset, but the reference was by name, | |
| 1028 | scan the table to see if the name refers to any other numbers, and test | |
| 1029 | them. The condition is true if any one is set. This is tediously similar | |
| 1030 | to the code above, but not close enough to try to amalgamate. */ | |
| 1031 | ||
| 1032 | if (!condition && condcode == OP_NCREF) | |
| 1033 | { | |
| 1034 | int refno = offset >> 1; | |
| 1035 | uschar *slotA = md->name_table; | |
| 1036 | ||
| 1037 | for (i = 0; i < md->name_count; i++) | |
| 1038 | { | |
| 1039 | if (GET2(slotA, 0) == refno) break; | |
| 1040 | slotA += md->name_entry_size; | |
| 1041 | } | |
| 1042 | ||
| 1043 | /* Found a name for the number - there can be only one; duplicate names | |
| 1044 | for different numbers are allowed, but not vice versa. First scan down | |
| 1045 | for duplicates. */ | |
| 1046 | ||
| 1047 | if (i < md->name_count) | |
| 1048 | { | |
| 1049 | uschar *slotB = slotA; | |
| 1050 | while (slotB > md->name_table) | |
| 1051 | { | |
| 1052 | slotB -= md->name_entry_size; | |
| 1053 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1054 | { | |
| 1055 | offset = GET2(slotB, 0) << 1; | |
| 1056 | condition = offset < offset_top && | |
| 1057 | md->offset_vector[offset] >= 0; | |
| 1058 | if (condition) break; | |
| 1059 | } | |
| 1060 | else break; | |
| 1061 | } | |
| 1062 | ||
| 1063 | /* Scan up for duplicates */ | |
| 1064 | ||
| 1065 | if (!condition) | |
| 1066 | { | |
| 1067 | slotB = slotA; | |
| 1068 | for (i++; i < md->name_count; i++) | |
| 1069 | { | |
| 1070 | slotB += md->name_entry_size; | |
| 1071 | if (strcmp((char *)slotA + 2, (char *)slotB + 2) == 0) | |
| 1072 | { | |
| 1073 | offset = GET2(slotB, 0) << 1; | |
| 1074 | condition = offset < offset_top && | |
| 1075 | md->offset_vector[offset] >= 0; | |
| 1076 | if (condition) break; | |
| 1077 | } | |
| 1078 | else break; | |
| 1079 | } | |
| 1080 | } | |
| 1081 | } | |
| 1082 | } | |
| 1083 | ||
| 1084 | /* Chose branch according to the condition */ | |
| 1085 | ||
| 1086 | ecode += condition? 3 : GET(ecode, 1); | ecode += condition? 3 : GET(ecode, 1); |
| 1087 | } | } |
| 1088 | ||
| # | Line 860 for (;;) | Line 1106 for (;;) |
| 1106 | ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); | ecode += 1 + LINK_SIZE + GET(ecode, LINK_SIZE + 2); |
| 1107 | while (*ecode == OP_ALT) ecode += GET(ecode, 1); | while (*ecode == OP_ALT) ecode += GET(ecode, 1); |
| 1108 | } | } |
| 1109 | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) | else if (rrc != MATCH_NOMATCH && |
| 1110 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1111 | { | { |
| 1112 | RRETURN(rrc); /* Need braces because of following else */ | RRETURN(rrc); /* Need braces because of following else */ |
| 1113 | } | } |
| # | Line 897 for (;;) | Line 1144 for (;;) |
| 1144 | break; | break; |
| 1145 | ||
| 1146 | ||
| 1147 | /* Before OP_ACCEPT there may be any number of OP_CLOSE opcodes, | |
| 1148 | to close any currently open capturing brackets. */ | |
| 1149 | ||
| 1150 | case OP_CLOSE: | |
| 1151 | number = GET2(ecode, 1); | |
| 1152 | offset = number << 1; | |
| 1153 | ||
| 1154 | #ifdef PCRE_DEBUG | |
| 1155 | printf("end bracket %d at *ACCEPT", number); | |
| 1156 | printf("\n"); | |
| 1157 | #endif | |
| 1158 | ||
| 1159 | md->capture_last = number; | |
| 1160 | if (offset >= md->offset_max) md->offset_overflow = TRUE; else | |
| 1161 | { | |
| 1162 | md->offset_vector[offset] = | |
| 1163 | md->offset_vector[md->offset_end - number]; | |
| 1164 | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); | |
| 1165 | if (offset_top <= offset) offset_top = offset + 2; | |
| 1166 | } | |
| 1167 | ecode += 3; | |
| 1168 | break; | |
| 1169 | ||
| 1170 | ||
| 1171 | /* 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 |
| 1172 | recursion, we should restore the offsets appropriately and continue from | recursion, we should restore the offsets appropriately and continue from |
| 1173 | after the call. */ | after the call. */ |
| # | Line 910 for (;;) | Line 1181 for (;;) |
| 1181 | md->recursive = rec->prevrec; | md->recursive = rec->prevrec; |
| 1182 | memmove(md->offset_vector, rec->offset_save, | memmove(md->offset_vector, rec->offset_save, |
| 1183 | rec->saved_max * sizeof(int)); | rec->saved_max * sizeof(int)); |
| 1184 | mstart = rec->save_start; | offset_top = rec->save_offset_top; |
| 1185 | ims = original_ims; | ims = original_ims; |
| 1186 | ecode = rec->after_call; | ecode = rec->after_call; |
| 1187 | break; | break; |
| 1188 | } | } |
| 1189 | ||
| 1190 | /* 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 |
| 1191 | 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 |
| 1192 | the subject. In both cases, backtracking will then try other alternatives, | |
| 1193 | if any. */ | |
| 1194 | ||
| 1195 | if (eptr == mstart && | |
| 1196 | (md->notempty || | |
| 1197 | (md->notempty_atstart && | |
| 1198 | mstart == md->start_subject + md->start_offset))) | |
| 1199 | MRRETURN(MATCH_NOMATCH); | |
| 1200 | ||
| 1201 | /* Otherwise, we have a match. */ | |
| 1202 | ||
| if (md->notempty && eptr == mstart) RRETURN(MATCH_NOMATCH); | ||
| 1203 | md->end_match_ptr = eptr; /* Record where we ended */ | md->end_match_ptr = eptr; /* Record where we ended */ |
| 1204 | md->end_offset_top = offset_top; /* and how many extracts were taken */ | md->end_offset_top = offset_top; /* and how many extracts were taken */ |
| 1205 | md->start_match_ptr = mstart; /* and the start (\K can modify) */ | md->start_match_ptr = mstart; /* and the start (\K can modify) */ |
| 1206 | RRETURN(MATCH_MATCH); | |
| 1207 | /* For some reason, the macros don't work properly if an expression is | |
| 1208 | given as the argument to MRRETURN when the heap is in use. */ | |
| 1209 | ||
| 1210 | rrc = (op == OP_END)? MATCH_MATCH : MATCH_ACCEPT; | |
| 1211 | MRRETURN(rrc); | |
| 1212 | ||
| 1213 | /* Change option settings */ | /* Change option settings */ |
| 1214 | ||
| # | Line 945 for (;;) | Line 1230 for (;;) |
| 1230 | { | { |
| 1231 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, |
| 1232 | RM4); | RM4); |
| 1233 | if (rrc == MATCH_MATCH) break; | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
| 1234 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | { |
| 1235 | mstart = md->start_match_ptr; /* In case \K reset it */ | |
| 1236 | break; | |
| 1237 | } | |
| 1238 | if (rrc != MATCH_NOMATCH && | |
| 1239 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1240 | RRETURN(rrc); | |
| 1241 | ecode += GET(ecode, 1); | ecode += GET(ecode, 1); |
| 1242 | } | } |
| 1243 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| 1244 | if (*ecode == OP_KET) RRETURN(MATCH_NOMATCH); | if (*ecode == OP_KET) MRRETURN(MATCH_NOMATCH); |
| 1245 | ||
| 1246 | /* If checking an assertion for a condition, return MATCH_MATCH. */ | /* If checking an assertion for a condition, return MATCH_MATCH. */ |
| 1247 | ||
| # | Line 964 for (;;) | Line 1255 for (;;) |
| 1255 | offset_top = md->end_offset_top; | offset_top = md->end_offset_top; |
| 1256 | continue; | continue; |
| 1257 | ||
| 1258 | /* Negative assertion: all branches must fail to match */ | /* Negative assertion: all branches must fail to match. Encountering SKIP, |
| 1259 | PRUNE, or COMMIT means we must assume failure without checking subsequent | |
| 1260 | branches. */ | |
| 1261 | ||
| 1262 | case OP_ASSERT_NOT: | case OP_ASSERT_NOT: |
| 1263 | case OP_ASSERTBACK_NOT: | case OP_ASSERTBACK_NOT: |
| # | Line 972 for (;;) | Line 1265 for (;;) |
| 1265 | { | { |
| 1266 | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, | RMATCH(eptr, ecode + 1 + LINK_SIZE, offset_top, md, ims, NULL, 0, |
| 1267 | RM5); | RM5); |
| 1268 | if (rrc == MATCH_MATCH) RRETURN(MATCH_NOMATCH); | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) MRRETURN(MATCH_NOMATCH); |
| 1269 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | if (rrc == MATCH_SKIP || rrc == MATCH_PRUNE || rrc == MATCH_COMMIT) |
| 1270 | { | |
| 1271 | do ecode += GET(ecode,1); while (*ecode == OP_ALT); | |
| 1272 | break; | |
| 1273 | } | |
| 1274 | if (rrc != MATCH_NOMATCH && | |
| 1275 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1276 | RRETURN(rrc); | |
| 1277 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
| 1278 | } | } |
| 1279 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| # | Line 996 for (;;) | Line 1296 for (;;) |
| 1296 | while (i-- > 0) | while (i-- > 0) |
| 1297 | { | { |
| 1298 | eptr--; | eptr--; |
| 1299 | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1300 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 1301 | } | } |
| 1302 | } | } |
| # | Line 1007 for (;;) | Line 1307 for (;;) |
| 1307 | ||
| 1308 | { | { |
| 1309 | eptr -= GET(ecode, 1); | eptr -= GET(ecode, 1); |
| 1310 | if (eptr < md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1311 | } | } |
| 1312 | ||
| 1313 | /* Skip to next op code */ | /* Save the earliest consulted character, then skip to next op code */ |
| 1314 | ||
| 1315 | if (eptr < md->start_used_ptr) md->start_used_ptr = eptr; | |
| 1316 | ecode += 1 + LINK_SIZE; | ecode += 1 + LINK_SIZE; |
| 1317 | break; | break; |
| 1318 | ||
| # | Line 1027 for (;;) | Line 1328 for (;;) |
| 1328 | cb.callout_number = ecode[1]; | cb.callout_number = ecode[1]; |
| 1329 | cb.offset_vector = md->offset_vector; | cb.offset_vector = md->offset_vector; |
| 1330 | cb.subject = (PCRE_SPTR)md->start_subject; | cb.subject = (PCRE_SPTR)md->start_subject; |
| 1331 | cb.subject_length = md->end_subject - md->start_subject; | cb.subject_length = (int)(md->end_subject - md->start_subject); |
| 1332 | cb.start_match = mstart - md->start_subject; | cb.start_match = (int)(mstart - md->start_subject); |
| 1333 | cb.current_position = eptr - md->start_subject; | cb.current_position = (int)(eptr - md->start_subject); |
| 1334 | cb.pattern_position = GET(ecode, 2); | cb.pattern_position = GET(ecode, 2); |
| 1335 | cb.next_item_length = GET(ecode, 2 + LINK_SIZE); | cb.next_item_length = GET(ecode, 2 + LINK_SIZE); |
| 1336 | cb.capture_top = offset_top/2; | cb.capture_top = offset_top/2; |
| 1337 | cb.capture_last = md->capture_last; | cb.capture_last = md->capture_last; |
| 1338 | cb.callout_data = md->callout_data; | cb.callout_data = md->callout_data; |
| 1339 | if ((rrc = (*pcre_callout)(&cb)) > 0) RRETURN(MATCH_NOMATCH); | if ((rrc = (*pcre_callout)(&cb)) > 0) MRRETURN(MATCH_NOMATCH); |
| 1340 | if (rrc < 0) RRETURN(rrc); | if (rrc < 0) RRETURN(rrc); |
| 1341 | } | } |
| 1342 | ecode += 2 + 2*LINK_SIZE; | ecode += 2 + 2*LINK_SIZE; |
| # | Line 1090 for (;;) | Line 1391 for (;;) |
| 1391 | ||
| 1392 | memcpy(new_recursive.offset_save, md->offset_vector, | memcpy(new_recursive.offset_save, md->offset_vector, |
| 1393 | new_recursive.saved_max * sizeof(int)); | new_recursive.saved_max * sizeof(int)); |
| 1394 | new_recursive.save_start = mstart; | new_recursive.save_offset_top = offset_top; |
| mstart = eptr; | ||
| 1395 | ||
| 1396 | /* 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 |
| 1397 | restore the offset and recursion data. */ | restore the offset and recursion data. */ |
| # | Line 1102 for (;;) | Line 1402 for (;;) |
| 1402 | { | { |
| 1403 | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, | RMATCH(eptr, callpat + _pcre_OP_lengths[*callpat], offset_top, |
| 1404 | md, ims, eptrb, flags, RM6); | md, ims, eptrb, flags, RM6); |
| 1405 | if (rrc == MATCH_MATCH) | if (rrc == MATCH_MATCH || rrc == MATCH_ACCEPT) |
| 1406 | { | { |
| 1407 | DPRINTF(("Recursion matched\n")); | DPRINTF(("Recursion matched\n")); |
| 1408 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1409 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1410 | (pcre_free)(new_recursive.offset_save); | (pcre_free)(new_recursive.offset_save); |
| 1411 | RRETURN(MATCH_MATCH); | MRRETURN(MATCH_MATCH); |
| 1412 | } | } |
| 1413 | else if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) | else if (rrc != MATCH_NOMATCH && |
| 1414 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1415 | { | { |
| 1416 | DPRINTF(("Recursion gave error %d\n", rrc)); | DPRINTF(("Recursion gave error %d\n", rrc)); |
| 1417 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| # | Line 1129 for (;;) | Line 1430 for (;;) |
| 1430 | md->recursive = new_recursive.prevrec; | md->recursive = new_recursive.prevrec; |
| 1431 | if (new_recursive.offset_save != stacksave) | if (new_recursive.offset_save != stacksave) |
| 1432 | (pcre_free)(new_recursive.offset_save); | (pcre_free)(new_recursive.offset_save); |
| 1433 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1434 | } | } |
| 1435 | /* Control never reaches here */ | /* Control never reaches here */ |
| 1436 | ||
| # | Line 1138 for (;;) | Line 1439 for (;;) |
| 1439 | a move back into the brackets. Friedl calls these "atomic" subpatterns. | a move back into the brackets. Friedl calls these "atomic" subpatterns. |
| 1440 | 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 |
| 1441 | 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 |
| 1442 | the end of a normal bracket, leaving the subject pointer. */ | the end of a normal bracket, leaving the subject pointer, but resetting |
| 1443 | the start-of-match value in case it was changed by \K. */ | |
| 1444 | ||
| 1445 | case OP_ONCE: | case OP_ONCE: |
| 1446 | prev = ecode; | prev = ecode; |
| # | Line 1147 for (;;) | Line 1449 for (;;) |
| 1449 | do | do |
| 1450 | { | { |
| 1451 | 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); |
| 1452 | if (rrc == MATCH_MATCH) break; | if (rrc == MATCH_MATCH) /* Note: _not_ MATCH_ACCEPT */ |
| 1453 | if (rrc != MATCH_NOMATCH && rrc != MATCH_THEN) RRETURN(rrc); | { |
| 1454 | mstart = md->start_match_ptr; | |
| 1455 | break; | |
| 1456 | } | |
| 1457 | if (rrc != MATCH_NOMATCH && | |
| 1458 | (rrc != MATCH_THEN || md->start_match_ptr != ecode)) | |
| 1459 | RRETURN(rrc); | |
| 1460 | ecode += GET(ecode,1); | ecode += GET(ecode,1); |
| 1461 | } | } |
| 1462 | while (*ecode == OP_ALT); | while (*ecode == OP_ALT); |
| # | Line 1266 for (;;) | Line 1574 for (;;) |
| 1574 | } | } |
| 1575 | else saved_eptr = NULL; | else saved_eptr = NULL; |
| 1576 | ||
| 1577 | /* 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 |
| 1578 | 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 |
| 1579 | assertions. Do this also for the "once" (atomic) groups. */ | use by positive assertions. We also need to record the match start in case |
| 1580 | it was changed by \K. */ | |
| 1581 | ||
| 1582 | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || | if (*prev == OP_ASSERT || *prev == OP_ASSERT_NOT || |
| 1583 | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || | *prev == OP_ASSERTBACK || *prev == OP_ASSERTBACK_NOT || |
| # | Line 1276 for (;;) | Line 1585 for (;;) |
| 1585 | { | { |
| 1586 | md->end_match_ptr = eptr; /* For ONCE */ | md->end_match_ptr = eptr; /* For ONCE */ |
| 1587 | md->end_offset_top = offset_top; | md->end_offset_top = offset_top; |
| 1588 | RRETURN(MATCH_MATCH); | md->start_match_ptr = mstart; |
| 1589 | MRRETURN(MATCH_MATCH); | |
| 1590 | } | } |
| 1591 | ||
| 1592 | /* 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 1600 for (;;) |
| 1600 | number = GET2(prev, 1+LINK_SIZE); | number = GET2(prev, 1+LINK_SIZE); |
| 1601 | offset = number << 1; | offset = number << 1; |
| 1602 | ||
| 1603 | #ifdef DEBUG | #ifdef PCRE_DEBUG |
| 1604 | printf("end bracket %d", number); | printf("end bracket %d", number); |
| 1605 | printf("\n"); | printf("\n"); |
| 1606 | #endif | #endif |
| # | Line 1300 for (;;) | Line 1610 for (;;) |
| 1610 | { | { |
| 1611 | md->offset_vector[offset] = | md->offset_vector[offset] = |
| 1612 | md->offset_vector[md->offset_end - number]; | md->offset_vector[md->offset_end - number]; |
| 1613 | md->offset_vector[offset+1] = eptr - md->start_subject; | md->offset_vector[offset+1] = (int)(eptr - md->start_subject); |
| 1614 | if (offset_top <= offset) offset_top = offset + 2; | if (offset_top <= offset) offset_top = offset + 2; |
| 1615 | } | } |
| 1616 | ||
| # | Line 1312 for (;;) | Line 1622 for (;;) |
| 1622 | recursion_info *rec = md->recursive; | recursion_info *rec = md->recursive; |
| 1623 | DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); | DPRINTF(("Recursion (%d) succeeded - continuing\n", number)); |
| 1624 | md->recursive = rec->prevrec; | md->recursive = rec->prevrec; |
| mstart = rec->save_start; | ||
| 1625 | memcpy(md->offset_vector, rec->offset_save, | memcpy(md->offset_vector, rec->offset_save, |
| 1626 | rec->saved_max * sizeof(int)); | rec->saved_max * sizeof(int)); |
| 1627 | offset_top = rec->save_offset_top; | |
| 1628 | ecode = rec->after_call; | ecode = rec->after_call; |
| 1629 | ims = original_ims; | ims = original_ims; |
| 1630 | break; | break; |
| # | Line 1371 for (;;) | Line 1681 for (;;) |
| 1681 | /* Start of subject unless notbol, or after internal newline if multiline */ | /* Start of subject unless notbol, or after internal newline if multiline */ |
| 1682 | ||
| 1683 | case OP_CIRC: | case OP_CIRC: |
| 1684 | if (md->notbol && eptr == md->start_subject) RRETURN(MATCH_NOMATCH); | if (md->notbol && eptr == md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1685 | if ((ims & PCRE_MULTILINE) != 0) | if ((ims & PCRE_MULTILINE) != 0) |
| 1686 | { | { |
| 1687 | if (eptr != md->start_subject && | if (eptr != md->start_subject && |
| 1688 | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) | (eptr == md->end_subject || !WAS_NEWLINE(eptr))) |
| 1689 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1690 | ecode++; | ecode++; |
| 1691 | break; | break; |
| 1692 | } | } |
| # | Line 1385 for (;;) | Line 1695 for (;;) |
| 1695 | /* Start of subject assertion */ | /* Start of subject assertion */ |
| 1696 | ||
| 1697 | case OP_SOD: | case OP_SOD: |
| 1698 | if (eptr != md->start_subject) RRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject) MRRETURN(MATCH_NOMATCH); |
| 1699 | ecode++; | ecode++; |
| 1700 | break; | break; |
| 1701 | ||
| 1702 | /* Start of match assertion */ | /* Start of match assertion */ |
| 1703 | ||
| 1704 | case OP_SOM: | case OP_SOM: |
| 1705 | if (eptr != md->start_subject + md->start_offset) RRETURN(MATCH_NOMATCH); | if (eptr != md->start_subject + md->start_offset) MRRETURN(MATCH_NOMATCH); |
| 1706 | ecode++; | ecode++; |
| 1707 | break; | break; |
| 1708 | ||
| # | Line 1410 for (;;) | Line 1720 for (;;) |
| 1720 | if ((ims & PCRE_MULTILINE) != 0) | if ((ims & PCRE_MULTILINE) != 0) |
| 1721 | { | { |
| 1722 | if (eptr < md->end_subject) | if (eptr < md->end_subject) |
| 1723 | { if (!IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); } | { if (!IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); } |
| 1724 | else | else |
| 1725 | { if (md->noteol) RRETURN(MATCH_NOMATCH); } | { |
| 1726 | if (md->noteol) MRRETURN(MATCH_NOMATCH); | |
| 1727 | SCHECK_PARTIAL(); | |
| 1728 | } | |
| 1729 | ecode++; | ecode++; |
| 1730 | break; | break; |
| 1731 | } | } |
| 1732 | else | else /* Not multiline */ |
| 1733 | { | { |
| 1734 | if (md->noteol) RRETURN(MATCH_NOMATCH); | if (md->noteol) MRRETURN(MATCH_NOMATCH); |
| 1735 | 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; | ||
| } | ||
| 1736 | } | } |
| 1737 | ||
| 1738 | /* ... else fall through for endonly */ | /* ... else fall through for endonly */ |
| 1739 | ||
| 1740 | /* End of subject assertion (\z) */ | /* End of subject assertion (\z) */ |
| 1741 | ||
| 1742 | case OP_EOD: | case OP_EOD: |
| 1743 | if (eptr < md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr < md->end_subject) MRRETURN(MATCH_NOMATCH); |
| 1744 | SCHECK_PARTIAL(); | |
| 1745 | ecode++; | ecode++; |
| 1746 | break; | break; |
| 1747 | ||
| 1748 | /* End of subject or ending \n assertion (\Z) */ | /* End of subject or ending \n assertion (\Z) */ |
| 1749 | ||
| 1750 | case OP_EODN: | case OP_EODN: |
| 1751 | if (eptr != md->end_subject && | ASSERT_NL_OR_EOS: |
| 1752 | if (eptr < md->end_subject && | |
| 1753 | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) | (!IS_NEWLINE(eptr) || eptr != md->end_subject - md->nllen)) |
| 1754 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1755 | ||
| 1756 | /* Either at end of string or \n before end. */ | |
| 1757 | ||
| 1758 | SCHECK_PARTIAL(); | |
| 1759 | ecode++; | ecode++; |
| 1760 | break; | break; |
| 1761 | ||
| # | Line 1454 for (;;) | Line 1767 for (;;) |
| 1767 | ||
| 1768 | /* Find out if the previous and current characters are "word" characters. | /* Find out if the previous and current characters are "word" characters. |
| 1769 | 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 |
| 1770 | be "non-word" characters. */ | be "non-word" characters. Remember the earliest consulted character for |
| 1771 | partial matching. */ | |
| 1772 | ||
| 1773 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| 1774 | if (utf8) | if (utf8) |
| 1775 | { | { |
| 1776 | /* Get status of previous character */ | |
| 1777 | ||
| 1778 | if (eptr == md->start_subject) prev_is_word = FALSE; else | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 1779 | { | { |
| 1780 | const uschar *lastptr = eptr - 1; | USPTR lastptr = eptr - 1; |
| 1781 | while((*lastptr & 0xc0) == 0x80) lastptr--; | while((*lastptr & 0xc0) == 0x80) lastptr--; |
| 1782 | if (lastptr < md->start_used_ptr) md->start_used_ptr = lastptr; | |
| 1783 | GETCHAR(c, lastptr); | GETCHAR(c, lastptr); |
| 1784 | #ifdef SUPPORT_UCP | |
| 1785 | if (md->use_ucp) | |
| 1786 | { | |
| 1787 | if (c == '_') prev_is_word = TRUE; else | |
| 1788 | { | |
| 1789 | int cat = UCD_CATEGORY(c); | |
| 1790 | prev_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1791 | } | |
| 1792 | } | |
| 1793 | else | |
| 1794 | #endif | |
| 1795 | prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; | prev_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; |
| 1796 | } | } |
| 1797 | if (eptr >= md->end_subject) cur_is_word = FALSE; else | |
| 1798 | /* Get status of next character */ | |
| 1799 | ||
| 1800 | if (eptr >= md->end_subject) | |
| 1801 | { | |
| 1802 | SCHECK_PARTIAL(); | |
| 1803 | cur_is_word = FALSE; | |
| 1804 | } | |
| 1805 | else | |
| 1806 | { | { |
| 1807 | GETCHAR(c, eptr); | GETCHAR(c, eptr); |
| 1808 | #ifdef SUPPORT_UCP | |
| 1809 | if (md->use_ucp) | |
| 1810 | { | |
| 1811 | if (c == '_') cur_is_word = TRUE; else | |
| 1812 | { | |
| 1813 | int cat = UCD_CATEGORY(c); | |
| 1814 | cur_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1815 | } | |
| 1816 | } | |
| 1817 | else | |
| 1818 | #endif | |
| 1819 | cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; | cur_is_word = c < 256 && (md->ctypes[c] & ctype_word) != 0; |
| 1820 | } | } |
| 1821 | } | } |
| 1822 | else | else |
| 1823 | #endif | #endif |
| 1824 | ||
| 1825 | /* More streamlined when not in UTF-8 mode */ | /* Not in UTF-8 mode, but we may still have PCRE_UCP set, and for |
| 1826 | consistency with the behaviour of \w we do use it in this case. */ | |
| 1827 | ||
| 1828 | { | { |
| 1829 | prev_is_word = (eptr != md->start_subject) && | /* Get status of previous character */ |
| 1830 | ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
| 1831 | cur_is_word = (eptr < md->end_subject) && | if (eptr == md->start_subject) prev_is_word = FALSE; else |
| 1832 | ((md->ctypes[*eptr] & ctype_word) != 0); | { |
| 1833 | if (eptr <= md->start_used_ptr) md->start_used_ptr = eptr - 1; | |
| 1834 | #ifdef SUPPORT_UCP | |
| 1835 | if (md->use_ucp) | |
| 1836 | { | |
| 1837 | c = eptr[-1]; | |
| 1838 | if (c == '_') prev_is_word = TRUE; else | |
| 1839 | { | |
| 1840 | int cat = UCD_CATEGORY(c); | |
| 1841 | prev_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1842 | } | |
| 1843 | } | |
| 1844 | else | |
| 1845 | #endif | |
| 1846 | prev_is_word = ((md->ctypes[eptr[-1]] & ctype_word) != 0); | |
| 1847 | } | |
| 1848 | ||
| 1849 | /* Get status of next character */ | |
| 1850 | ||
| 1851 | if (eptr >= md->end_subject) | |
| 1852 | { | |
| 1853 | SCHECK_PARTIAL(); | |
| 1854 | cur_is_word = FALSE; | |
| 1855 | } | |
| 1856 | else | |
| 1857 | #ifdef SUPPORT_UCP | |
| 1858 | if (md->use_ucp) | |
| 1859 | { | |
| 1860 | c = *eptr; | |
| 1861 | if (c == '_') cur_is_word = TRUE; else | |
| 1862 | { | |
| 1863 | int cat = UCD_CATEGORY(c); | |
| 1864 | cur_is_word = (cat == ucp_L || cat == ucp_N); | |
| 1865 | } | |
| 1866 | } | |
| 1867 | else | |
| 1868 | #endif | |
| 1869 | cur_is_word = ((md->ctypes[*eptr] & ctype_word) != 0); | |
| 1870 | } | } |
| 1871 | ||
| 1872 | /* Now see if the situation is what we want */ | /* Now see if the situation is what we want */ |
| 1873 | ||
| 1874 | if ((*ecode++ == OP_WORD_BOUNDARY)? | if ((*ecode++ == OP_WORD_BOUNDARY)? |
| 1875 | cur_is_word == prev_is_word : cur_is_word != prev_is_word) | cur_is_word == prev_is_word : cur_is_word != prev_is_word) |
| 1876 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1877 | } | } |
| 1878 | break; | break; |
| 1879 | ||
| 1880 | /* Match a single character type; inline for speed */ | /* Match a single character type; inline for speed */ |
| 1881 | ||
| 1882 | case OP_ANY: | case OP_ANY: |
| 1883 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); |
| 1884 | /* Fall through */ | /* Fall through */ |
| 1885 | ||
| 1886 | case OP_ALLANY: | case OP_ALLANY: |
| 1887 | if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr++ >= md->end_subject) |
| 1888 | { | |
| 1889 | SCHECK_PARTIAL(); | |
| 1890 | MRRETURN(MATCH_NOMATCH); | |
| 1891 | } | |
| 1892 | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | if (utf8) while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 1893 | ecode++; | ecode++; |
| 1894 | break; | break; |
| # | Line 1508 for (;;) | Line 1897 for (;;) |
| 1897 | any byte, even newline, independent of the setting of PCRE_DOTALL. */ | any byte, even newline, independent of the setting of PCRE_DOTALL. */ |
| 1898 | ||
| 1899 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 1900 | if (eptr++ >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr++ >= md->end_subject) |
| 1901 | { | |
| 1902 | SCHECK_PARTIAL(); | |
| 1903 | MRRETURN(MATCH_NOMATCH); | |
| 1904 | } | |
| 1905 | ecode++; | ecode++; |
| 1906 | break; | break; |
| 1907 | ||
| 1908 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 1909 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1910 | { | |
| 1911 | SCHECK_PARTIAL(); | |
| 1912 | MRRETURN(MATCH_NOMATCH); | |
| 1913 | } | |
| 1914 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1915 | if ( | if ( |
| 1916 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1521 for (;;) | Line 1918 for (;;) |
| 1918 | #endif | #endif |
| 1919 | (md->ctypes[c] & ctype_digit) != 0 | (md->ctypes[c] & ctype_digit) != 0 |
| 1920 | ) | ) |
| 1921 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1922 | ecode++; | ecode++; |
| 1923 | break; | break; |
| 1924 | ||
| 1925 | case OP_DIGIT: | case OP_DIGIT: |
| 1926 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1927 | { | |
| 1928 | SCHECK_PARTIAL(); | |
| 1929 | MRRETURN(MATCH_NOMATCH); | |
| 1930 | } | |
| 1931 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1932 | if ( | if ( |
| 1933 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1534 for (;;) | Line 1935 for (;;) |
| 1935 | #endif | #endif |
| 1936 | (md->ctypes[c] & ctype_digit) == 0 | (md->ctypes[c] & ctype_digit) == 0 |
| 1937 | ) | ) |
| 1938 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1939 | ecode++; | ecode++; |
| 1940 | break; | break; |
| 1941 | ||
| 1942 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 1943 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1944 | { | |
| 1945 | SCHECK_PARTIAL(); | |
| 1946 | MRRETURN(MATCH_NOMATCH); | |
| 1947 | } | |
| 1948 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1949 | if ( | if ( |
| 1950 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1547 for (;;) | Line 1952 for (;;) |
| 1952 | #endif | #endif |
| 1953 | (md->ctypes[c] & ctype_space) != 0 | (md->ctypes[c] & ctype_space) != 0 |
| 1954 | ) | ) |
| 1955 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1956 | ecode++; | ecode++; |
| 1957 | break; | break; |
| 1958 | ||
| 1959 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 1960 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1961 | { | |
| 1962 | SCHECK_PARTIAL(); | |
| 1963 | MRRETURN(MATCH_NOMATCH); | |
| 1964 | } | |
| 1965 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1966 | if ( | if ( |
| 1967 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1560 for (;;) | Line 1969 for (;;) |
| 1969 | #endif | #endif |
| 1970 | (md->ctypes[c] & ctype_space) == 0 | (md->ctypes[c] & ctype_space) == 0 |
| 1971 | ) | ) |
| 1972 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1973 | ecode++; | ecode++; |
| 1974 | break; | break; |
| 1975 | ||
| 1976 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 1977 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1978 | { | |
| 1979 | SCHECK_PARTIAL(); | |
| 1980 | MRRETURN(MATCH_NOMATCH); | |
| 1981 | } | |
| 1982 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 1983 | if ( | if ( |
| 1984 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1573 for (;;) | Line 1986 for (;;) |
| 1986 | #endif | #endif |
| 1987 | (md->ctypes[c] & ctype_word) != 0 | (md->ctypes[c] & ctype_word) != 0 |
| 1988 | ) | ) |
| 1989 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 1990 | ecode++; | ecode++; |
| 1991 | break; | break; |
| 1992 | ||
| 1993 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 1994 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 1995 | { | |
| 1996 | SCHECK_PARTIAL(); | |
| 1997 | MRRETURN(MATCH_NOMATCH); | |
| 1998 | } | |
| 1999 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2000 | if ( | if ( |
| 2001 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 1586 for (;;) | Line 2003 for (;;) |
| 2003 | #endif | #endif |
| 2004 | (md->ctypes[c] & ctype_word) == 0 | (md->ctypes[c] & ctype_word) == 0 |
| 2005 | ) | ) |
| 2006 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2007 | ecode++; | ecode++; |
| 2008 | break; | break; |
| 2009 | ||
| 2010 | case OP_ANYNL: | case OP_ANYNL: |
| 2011 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2012 | { | |
| 2013 | SCHECK_PARTIAL(); | |
| 2014 | MRRETURN(MATCH_NOMATCH); | |
| 2015 | } | |
| 2016 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2017 | switch(c) | switch(c) |
| 2018 | { | { |
| 2019 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2020 | case 0x000d: | case 0x000d: |
| 2021 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 2022 | break; | break; |
| # | Line 1608 for (;;) | Line 2029 for (;;) |
| 2029 | case 0x0085: | case 0x0085: |
| 2030 | case 0x2028: | case 0x2028: |
| 2031 | case 0x2029: | case 0x2029: |
| 2032 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 2033 | break; | break; |
| 2034 | } | } |
| 2035 | ecode++; | ecode++; |
| 2036 | break; | break; |
| 2037 | ||
| 2038 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 2039 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2040 | { | |
| 2041 | SCHECK_PARTIAL(); | |
| 2042 | MRRETURN(MATCH_NOMATCH); | |
| 2043 | } | |
| 2044 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2045 | switch(c) | switch(c) |
| 2046 | { | { |
| # | Line 1639 for (;;) | Line 2064 for (;;) |
| 2064 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 2065 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 2066 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 2067 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2068 | } | } |
| 2069 | ecode++; | ecode++; |
| 2070 | break; | break; |
| 2071 | ||
| 2072 | case OP_HSPACE: | case OP_HSPACE: |
| 2073 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2074 | { | |
| 2075 | SCHECK_PARTIAL(); | |
| 2076 | MRRETURN(MATCH_NOMATCH); | |
| 2077 | } | |
| 2078 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2079 | switch(c) | switch(c) |
| 2080 | { | { |
| 2081 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2082 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 2083 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 2084 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 1675 for (;;) | Line 2104 for (;;) |
| 2104 | break; | break; |
| 2105 | ||
| 2106 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 2107 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2108 | { | |
| 2109 | SCHECK_PARTIAL(); | |
| 2110 | MRRETURN(MATCH_NOMATCH); | |
| 2111 | } | |
| 2112 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2113 | switch(c) | switch(c) |
| 2114 | { | { |
| # | Line 1687 for (;;) | Line 2120 for (;;) |
| 2120 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 2121 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 2122 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 2123 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2124 | } | } |
| 2125 | ecode++; | ecode++; |
| 2126 | break; | break; |
| 2127 | ||
| 2128 | case OP_VSPACE: | case OP_VSPACE: |
| 2129 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2130 | { | |
| 2131 | SCHECK_PARTIAL(); | |
| 2132 | MRRETURN(MATCH_NOMATCH); | |
| 2133 | } | |
| 2134 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2135 | switch(c) | switch(c) |
| 2136 | { | { |
| 2137 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 2138 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 2139 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 2140 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 1716 for (;;) | Line 2153 for (;;) |
| 2153 | ||
| 2154 | case OP_PROP: | case OP_PROP: |
| 2155 | case OP_NOTPROP: | case OP_NOTPROP: |
| 2156 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2157 | { | |
| 2158 | SCHECK_PARTIAL(); | |
| 2159 | MRRETURN(MATCH_NOMATCH); | |
| 2160 | } | |
| 2161 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2162 | { | { |
| 2163 | const ucd_record *prop = GET_UCD(c); | const ucd_record *prop = GET_UCD(c); |
| # | Line 1724 for (;;) | Line 2165 for (;;) |
| 2165 | switch(ecode[1]) | switch(ecode[1]) |
| 2166 | { | { |
| 2167 | case PT_ANY: | case PT_ANY: |
| 2168 | if (op == OP_NOTPROP) RRETURN(MATCH_NOMATCH); | if (op == OP_NOTPROP) MRRETURN(MATCH_NOMATCH); |
| 2169 | break; | break; |
| 2170 | ||
| 2171 | case PT_LAMP: | case PT_LAMP: |
| 2172 | if ((prop->chartype == ucp_Lu || | if ((prop->chartype == ucp_Lu || |
| 2173 | prop->chartype == ucp_Ll || | prop->chartype == ucp_Ll || |
| 2174 | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) | prop->chartype == ucp_Lt) == (op == OP_NOTPROP)) |
| 2175 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2176 | break; | break; |
| 2177 | ||
| 2178 | case PT_GC: | case PT_GC: |
| 2179 | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) | if ((ecode[2] != _pcre_ucp_gentype[prop->chartype]) == (op == OP_PROP)) |
| 2180 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2181 | break; | break; |
| 2182 | ||
| 2183 | case PT_PC: | case PT_PC: |
| 2184 | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) | if ((ecode[2] != prop->chartype) == (op == OP_PROP)) |
| 2185 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2186 | break; | break; |
| 2187 | ||
| 2188 | case PT_SC: | case PT_SC: |
| 2189 | if ((ecode[2] != prop->script) == (op == OP_PROP)) | if ((ecode[2] != prop->script) == (op == OP_PROP)) |
| 2190 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2191 | break; | |
| 2192 | ||
| 2193 | /* These are specials */ | |
| 2194 | ||
| 2195 | case PT_ALNUM: | |
| 2196 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | |
| 2197 | _pcre_ucp_gentype[prop->chartype] == ucp_N) == (op == OP_NOTPROP)) | |
| 2198 | MRRETURN(MATCH_NOMATCH); | |
| 2199 | break; | |
| 2200 | ||
| 2201 | case PT_SPACE: /* Perl space */ | |
| 2202 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | |
| 2203 | c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR) | |
| 2204 | == (op == OP_NOTPROP)) | |
| 2205 | MRRETURN(MATCH_NOMATCH); | |
| 2206 | break; | |
| 2207 | ||
| 2208 | case PT_PXSPACE: /* POSIX space */ | |
| 2209 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_Z || | |
| 2210 | c == CHAR_HT || c == CHAR_NL || c == CHAR_VT || | |
| 2211 | c == CHAR_FF || c == CHAR_CR) | |
| 2212 | == (op == OP_NOTPROP)) | |
| 2213 | MRRETURN(MATCH_NOMATCH); | |
| 2214 | break; | |
| 2215 | ||
| 2216 | case PT_WORD: | |
| 2217 | if ((_pcre_ucp_gentype[prop->chartype] == ucp_L || | |
| 2218 | _pcre_ucp_gentype[prop->chartype] == ucp_N || | |
| 2219 | c == CHAR_UNDERSCORE) == (op == OP_NOTPROP)) | |
| 2220 | MRRETURN(MATCH_NOMATCH); | |
| 2221 | break; | break; |
| 2222 | ||
| 2223 | /* This should never occur */ | |
| 2224 | ||
| 2225 | default: | default: |
| 2226 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 2227 | } | } |
| # | Line 1761 for (;;) | Line 2234 for (;;) |
| 2234 | is in the binary; otherwise a compile-time error occurs. */ | is in the binary; otherwise a compile-time error occurs. */ |
| 2235 | ||
| 2236 | case OP_EXTUNI: | case OP_EXTUNI: |
| 2237 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2238 | { | |
| 2239 | SCHECK_PARTIAL(); | |
| 2240 | MRRETURN(MATCH_NOMATCH); | |
| 2241 | } | |
| 2242 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2243 | { | { |
| 2244 | int category = UCD_CATEGORY(c); | int category = UCD_CATEGORY(c); |
| 2245 | if (category == ucp_M) RRETURN(MATCH_NOMATCH); | if (category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 2246 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 2247 | { | { |
| 2248 | int len = 1; | int len = 1; |
| # | Line 1792 for (;;) | Line 2269 for (;;) |
| 2269 | loops). */ | loops). */ |
| 2270 | ||
| 2271 | case OP_REF: | case OP_REF: |
| 2272 | { | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ |
| 2273 | offset = GET2(ecode, 1) << 1; /* Doubled ref number */ | ecode += 3; |
| ecode += 3; | ||
| 2274 | ||
| 2275 | /* If the reference is unset, there are two possibilities: | /* If the reference is unset, there are two possibilities: |
| 2276 | ||
| 2277 | (a) In the default, Perl-compatible state, set the length to be longer | (a) In the default, Perl-compatible state, set the length negative; |
| 2278 | 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 |
| 2279 | 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. | ||
| 2280 | ||
| 2281 | (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 |
| 2282 | so that the back reference matches an empty string. | so that the back reference matches an empty string. |
| 2283 | ||
| 2284 | 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 |
| 2285 | referenced subpattern. */ | referenced subpattern. */ |
| 2286 | ||
| 2287 | if (offset >= offset_top || md->offset_vector[offset] < 0) | if (offset >= offset_top || md->offset_vector[offset] < 0) |
| 2288 | length = (md->jscript_compat)? 0 : md->end_subject - eptr + 1; | length = (md->jscript_compat)? 0 : -1; |
| 2289 | else | else |
| 2290 | length = md->offset_vector[offset+1] - md->offset_vector[offset]; | length = md->offset_vector[offset+1] - md->offset_vector[offset]; |
| 2291 | ||
| 2292 | /* Set up for repetition, or handle the non-repeated case */ | /* Set up for repetition, or handle the non-repeated case */ |
| 2293 | ||
| 2294 | switch (*ecode) | switch (*ecode) |
| 2295 | { | { |
| 2296 | case OP_CRSTAR: | case OP_CRSTAR: |
| 2297 | case OP_CRMINSTAR: | case OP_CRMINSTAR: |
| 2298 | case OP_CRPLUS: | case OP_CRPLUS: |
| 2299 | case OP_CRMINPLUS: | case OP_CRMINPLUS: |
| 2300 | case OP_CRQUERY: | case OP_CRQUERY: |
| 2301 | case OP_CRMINQUERY: | case OP_CRMINQUERY: |
| 2302 | c = *ecode++ - OP_CRSTAR; | c = *ecode++ - OP_CRSTAR; |
| 2303 | minimize = (c & 1) != 0; | minimize = (c & 1) != 0; |
| 2304 | min = rep_min[c]; /* Pick up values from tables; */ | min = rep_min[c]; /* Pick up values from tables; */ |
| 2305 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 2306 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2307 | break; | break; |
| 2308 | ||
| 2309 | case OP_CRRANGE: | case OP_CRRANGE: |
| 2310 | case OP_CRMINRANGE: | case OP_CRMINRANGE: |
| 2311 | minimize = (*ecode == OP_CRMINRANGE); | minimize = (*ecode == OP_CRMINRANGE); |
| 2312 | min = GET2(ecode, 1); | min = GET2(ecode, 1); |
| 2313 | max = GET2(ecode, 3); | max = GET2(ecode, 3); |
| 2314 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2315 | ecode += 5; | ecode += 5; |
| 2316 | break; | break; |
| 2317 | ||
| 2318 | default: /* No repeat follows */ | default: /* No repeat follows */ |
| 2319 | if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); | if ((length = match_ref(offset, eptr, length, md, ims)) < 0) |
| 2320 | eptr += length; | { |
| 2321 | continue; /* With the main loop */ | CHECK_PARTIAL(); |
| 2322 | MRRETURN(MATCH_NOMATCH); | |
| 2323 | } | } |
| 2324 | eptr += length; | |
| 2325 | continue; /* With the main loop */ | |
| 2326 | } | |
| 2327 | ||
| 2328 | /* If the length of the reference is zero, just continue with the | /* Handle repeated back references. If the length of the reference is |
| 2329 | main loop. */ | zero, just continue with the main loop. */ |
| 2330 | ||
| 2331 | if (length == 0) continue; | if (length == 0) continue; |
| 2332 | ||
| 2333 | /* First, ensure the minimum number of matches are present. We get back | /* First, ensure the minimum number of matches are present. We get back |
| 2334 | the length of the reference string explicitly rather than passing the | the length of the reference string explicitly rather than passing the |
| 2335 | address of eptr, so that eptr can be a register variable. */ | address of eptr, so that eptr can be a register variable. */ |
| 2336 | ||
| 2337 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2338 | { | |
| 2339 | int slength; | |
| 2340 | if ((slength = match_ref(offset, eptr, length, md, ims)) < 0) | |
| 2341 | { | { |
| 2342 | if (!match_ref(offset, eptr, length, md, ims)) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2343 | eptr += length; | MRRETURN(MATCH_NOMATCH); |
| 2344 | } | } |
| 2345 | eptr += slength; | |
| 2346 | } | |
| 2347 | ||
| 2348 | /* If min = max, continue at the same level without recursion. | /* If min = max, continue at the same level without recursion. |
| 2349 | They are not both allowed to be zero. */ | They are not both allowed to be zero. */ |
| 2350 | ||
| 2351 | if (min == max) continue; | if (min == max) continue; |
| 2352 | ||
| 2353 | /* If minimizing, keep trying and advancing the pointer */ | /* If minimizing, keep trying and advancing the pointer */ |
| 2354 | ||
| 2355 | if (minimize) | if (minimize) |
| 2356 | { | |
| 2357 | for (fi = min;; fi++) | |
| 2358 | { | { |
| 2359 | for (fi = min;; fi++) | int slength; |
| 2360 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14); | |
| 2361 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 2362 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 2363 | if ((slength = match_ref(offset, eptr, length, md, ims)) < 0) | |
| 2364 | { | { |
| 2365 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM14); | CHECK_PARTIAL(); |
| 2366 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | MRRETURN(MATCH_NOMATCH); |
| if (fi >= max || !match_ref(offset, eptr, length, md, ims)) | ||
| RRETURN(MATCH_NOMATCH); | ||
| eptr += length; | ||
| 2367 | } | } |
| 2368 | /* Control never gets here */ | eptr += slength; |
| 2369 | } | } |
| 2370 | /* Control never gets here */ | |
| 2371 | } | |
| 2372 | ||
| 2373 | /* If maximizing, find the longest string and work backwards */ | /* If maximizing, find the longest string and work backwards */ |
| 2374 | ||
| 2375 | else | else |
| 2376 | { | |
| 2377 | pp = eptr; | |
| 2378 | for (i = min; i < max; i++) | |
| 2379 | { | { |
| 2380 | pp = eptr; | int slength; |
| 2381 | for (i = min; i < max; i++) | if ((slength = match_ref(offset, eptr, length, md, ims)) < 0) |
| 2382 | { | { |
| 2383 | if (!match_ref(offset, eptr, length, md, ims)) break; | CHECK_PARTIAL(); |
| 2384 | eptr += length; | break; |
| } | ||
| while (eptr >= pp) | ||
| { | ||
| RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15); | ||
| if (rrc != MATCH_NOMATCH) RRETURN(rrc); | ||
| eptr -= length; | ||
| 2385 | } | } |
| 2386 | RRETURN(MATCH_NOMATCH); | eptr += slength; |
| 2387 | } | } |
| 2388 | while (eptr >= pp) | |
| 2389 | { | |
| 2390 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM15); | |
| 2391 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 2392 | eptr -= length; | |
| 2393 | } | |
| 2394 | MRRETURN(MATCH_NOMATCH); | |
| 2395 | } | } |
| 2396 | /* Control never gets here */ | /* Control never gets here */ |
| 2397 | ||
| 2398 | /* Match a bit-mapped character class, possibly repeatedly. This op code is | /* Match a bit-mapped character class, possibly repeatedly. This op code is |
| 2399 | 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, |
| 2400 | 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 2449 for (;;) |
| 2449 | { | { |
| 2450 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2451 | { | { |
| 2452 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2453 | { | |
| 2454 | SCHECK_PARTIAL(); | |
| 2455 | MRRETURN(MATCH_NOMATCH); | |
| 2456 | } | |
| 2457 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2458 | if (c > 255) | if (c > 255) |
| 2459 | { | { |
| 2460 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); |
| 2461 | } | } |
| 2462 | else | else |
| 2463 | { | { |
| 2464 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2465 | } | } |
| 2466 | } | } |
| 2467 | } | } |
| # | Line 1976 for (;;) | Line 2471 for (;;) |
| 2471 | { | { |
| 2472 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2473 | { | { |
| 2474 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2475 | { | |
| 2476 | SCHECK_PARTIAL(); | |
| 2477 | MRRETURN(MATCH_NOMATCH); | |
| 2478 | } | |
| 2479 | c = *eptr++; | c = *eptr++; |
| 2480 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2481 | } | } |
| 2482 | } | } |
| 2483 | ||
| # | Line 2000 for (;;) | Line 2499 for (;;) |
| 2499 | { | { |
| 2500 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM16); |
| 2501 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2502 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2503 | if (eptr >= md->end_subject) | |
| 2504 | { | |
| 2505 | SCHECK_PARTIAL(); | |
| 2506 | MRRETURN(MATCH_NOMATCH); | |
| 2507 | } | |
| 2508 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 2509 | if (c > 255) | if (c > 255) |
| 2510 | { | { |
| 2511 | if (op == OP_CLASS) RRETURN(MATCH_NOMATCH); | if (op == OP_CLASS) MRRETURN(MATCH_NOMATCH); |
| 2512 | } | } |
| 2513 | else | else |
| 2514 | { | { |
| 2515 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2516 | } | } |
| 2517 | } | } |
| 2518 | } | } |
| # | Line 2020 for (;;) | Line 2524 for (;;) |
| 2524 | { | { |
| 2525 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM17); |
| 2526 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2527 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2528 | if (eptr >= md->end_subject) | |
| 2529 | { | |
| 2530 | SCHECK_PARTIAL(); | |
| 2531 | MRRETURN(MATCH_NOMATCH); | |
| 2532 | } | |
| 2533 | c = *eptr++; | c = *eptr++; |
| 2534 | if ((data[c/8] & (1 << (c&7))) == 0) RRETURN(MATCH_NOMATCH); | if ((data[c/8] & (1 << (c&7))) == 0) MRRETURN(MATCH_NOMATCH); |
| 2535 | } | } |
| 2536 | } | } |
| 2537 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2041 for (;;) | Line 2550 for (;;) |
| 2550 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2551 | { | { |
| 2552 | int len = 1; | int len = 1; |
| 2553 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2554 | { | |
| 2555 | SCHECK_PARTIAL(); | |
| 2556 | break; | |
| 2557 | } | |
| 2558 | GETCHARLEN(c, eptr, len); | GETCHARLEN(c, eptr, len); |
| 2559 | if (c > 255) | if (c > 255) |
| 2560 | { | { |
| # | Line 2067 for (;;) | Line 2580 for (;;) |
| 2580 | { | { |
| 2581 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2582 | { | { |
| 2583 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2584 | { | |
| 2585 | SCHECK_PARTIAL(); | |
| 2586 | break; | |
| 2587 | } | |
| 2588 | c = *eptr; | c = *eptr; |
| 2589 | if ((data[c/8] & (1 << (c&7))) == 0) break; | if ((data[c/8] & (1 << (c&7))) == 0) break; |
| 2590 | eptr++; | eptr++; |
| # | Line 2080 for (;;) | Line 2597 for (;;) |
| 2597 | } | } |
| 2598 | } | } |
| 2599 | ||
| 2600 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2601 | } | } |
| 2602 | } | } |
| 2603 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2129 for (;;) | Line 2646 for (;;) |
| 2646 | ||
| 2647 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2648 | { | { |
| 2649 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 2650 | { | |
| 2651 | SCHECK_PARTIAL(); | |
| 2652 | MRRETURN(MATCH_NOMATCH); | |
| 2653 | } | |
| 2654 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2655 | if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); |
| 2656 | } | } |
| 2657 | ||
| 2658 | /* 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 2669 for (;;) |
| 2669 | { | { |
| 2670 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM20); |
| 2671 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2672 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2673 | if (eptr >= md->end_subject) | |
| 2674 | { | |
| 2675 | SCHECK_PARTIAL(); | |
| 2676 | MRRETURN(MATCH_NOMATCH); | |
| 2677 | } | |
| 2678 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 2679 | if (!_pcre_xclass(c, data)) RRETURN(MATCH_NOMATCH); | if (!_pcre_xclass(c, data)) MRRETURN(MATCH_NOMATCH); |
| 2680 | } | } |
| 2681 | /* Control never gets here */ | /* Control never gets here */ |
| 2682 | } | } |
| # | Line 2163 for (;;) | Line 2689 for (;;) |
| 2689 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2690 | { | { |
| 2691 | int len = 1; | int len = 1; |
| 2692 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 2693 | { | |
| 2694 | SCHECK_PARTIAL(); | |
| 2695 | break; | |
| 2696 | } | |
| 2697 | GETCHARLENTEST(c, eptr, len); | GETCHARLENTEST(c, eptr, len); |
| 2698 | if (!_pcre_xclass(c, data)) break; | if (!_pcre_xclass(c, data)) break; |
| 2699 | eptr += len; | eptr += len; |
| # | Line 2175 for (;;) | Line 2705 for (;;) |
| 2705 | if (eptr-- == pp) break; /* Stop if tried at original pos */ | if (eptr-- == pp) break; /* Stop if tried at original pos */ |
| 2706 | if (utf8) BACKCHAR(eptr); | if (utf8) BACKCHAR(eptr); |
| 2707 | } | } |
| 2708 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2709 | } | } |
| 2710 | ||
| 2711 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2191 for (;;) | Line 2721 for (;;) |
| 2721 | length = 1; | length = 1; |
| 2722 | ecode++; | ecode++; |
| 2723 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 2724 | if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | if (length > md->end_subject - eptr) |
| 2725 | while (length-- > 0) if (*ecode++ != *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 2726 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | |
| 2727 | MRRETURN(MATCH_NOMATCH); | |
| 2728 | } | |
| 2729 | while (length-- > 0) if (*ecode++ != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2730 | } | } |
| 2731 | else | else |
| 2732 | #endif | #endif |
| 2733 | ||
| 2734 | /* Non-UTF-8 mode */ | /* Non-UTF-8 mode */ |
| 2735 | { | { |
| 2736 | if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); | if (md->end_subject - eptr < 1) |
| 2737 | if (ecode[1] != *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 2738 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | |
| 2739 | MRRETURN(MATCH_NOMATCH); | |
| 2740 | } | |
| 2741 | if (ecode[1] != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 2742 | ecode += 2; | ecode += 2; |
| 2743 | } | } |
| 2744 | break; | break; |
| # | Line 2215 for (;;) | Line 2753 for (;;) |
| 2753 | ecode++; | ecode++; |
| 2754 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| 2755 | ||
| 2756 | if (length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | if (length > md->end_subject - eptr) |
| 2757 | { | |
| 2758 | CHECK_PARTIAL(); /* Not SCHECK_PARTIAL() */ | |
| 2759 | MRRETURN(MATCH_NOMATCH); | |
| 2760 | } | |
| 2761 | ||
| 2762 | /* 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 |
| 2763 | can use the fast lookup table. */ | can use the fast lookup table. */ |
| 2764 | ||
| 2765 | if (fc < 128) | if (fc < 128) |
| 2766 | { | { |
| 2767 | if (md->lcc[*ecode++] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | if (md->lcc[*ecode++] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); |
| 2768 | } | } |
| 2769 | ||
| 2770 | /* Otherwise we must pick up the subject character */ | /* Otherwise we must pick up the subject character */ |
| # | Line 2241 for (;;) | Line 2783 for (;;) |
| 2783 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2784 | if (dc != UCD_OTHERCASE(fc)) | if (dc != UCD_OTHERCASE(fc)) |
| 2785 | #endif | #endif |
| 2786 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 2787 | } | } |
| 2788 | } | } |
| 2789 | } | } |
| # | Line 2250 for (;;) | Line 2792 for (;;) |
| 2792 | ||
| 2793 | /* Non-UTF-8 mode */ | /* Non-UTF-8 mode */ |
| 2794 | { | { |
| 2795 | if (md->end_subject - eptr < 1) RRETURN(MATCH_NOMATCH); | if (md->end_subject - eptr < 1) |
| 2796 | if (md->lcc[ecode[1]] != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 2797 | SCHECK_PARTIAL(); /* This one can use SCHECK_PARTIAL() */ | |
| 2798 | MRRETURN(MATCH_NOMATCH); | |
| 2799 | } | |
| 2800 | if (md->lcc[ecode[1]] != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2801 | ecode += 2; | ecode += 2; |
| 2802 | } | } |
| 2803 | break; | break; |
| # | Line 2304 for (;;) | Line 2850 for (;;) |
| 2850 | case OP_MINQUERY: | case OP_MINQUERY: |
| 2851 | c = *ecode++ - OP_STAR; | c = *ecode++ - OP_STAR; |
| 2852 | minimize = (c & 1) != 0; | minimize = (c & 1) != 0; |
| 2853 | ||
| 2854 | min = rep_min[c]; /* Pick up values from tables; */ | min = rep_min[c]; /* Pick up values from tables; */ |
| 2855 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 2856 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 2857 | ||
| 2858 | /* 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. */ | ||
| 2859 | ||
| 2860 | REPEATCHAR: | REPEATCHAR: |
| 2861 | #ifdef SUPPORT_UTF8 | #ifdef SUPPORT_UTF8 |
| # | Line 2319 for (;;) | Line 2864 for (;;) |
| 2864 | length = 1; | length = 1; |
| 2865 | charptr = ecode; | charptr = ecode; |
| 2866 | GETCHARLEN(fc, ecode, length); | GETCHARLEN(fc, ecode, length); |
| if (min * length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 2867 | ecode += length; | ecode += length; |
| 2868 | ||
| 2869 | /* Handle multibyte character matching specially here. There is | /* Handle multibyte character matching specially here. There is |
| # | Line 2337 for (;;) | Line 2881 for (;;) |
| 2881 | ||
| 2882 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2883 | { | { |
| 2884 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | if (eptr <= md->end_subject - length && |
| 2885 | memcmp(eptr, charptr, length) == 0) eptr += length; | |
| 2886 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2887 | /* Need braces because of following else */ | else if (oclength > 0 && |
| 2888 | else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } | eptr <= md->end_subject - oclength && |
| 2889 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2890 | #endif /* SUPPORT_UCP */ | |
| 2891 | else | else |
| 2892 | { | { |
| 2893 | if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2894 | eptr += oclength; | MRRETURN(MATCH_NOMATCH); |
| 2895 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else { RRETURN(MATCH_NOMATCH); } | ||
| #endif /* SUPPORT_UCP */ | ||
| 2896 | } | } |
| 2897 | ||
| 2898 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2359 for (;;) | Line 2903 for (;;) |
| 2903 | { | { |
| 2904 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM22); |
| 2905 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2906 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 2907 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | if (eptr <= md->end_subject - length && |
| 2908 | memcmp(eptr, charptr, length) == 0) eptr += length; | |
| 2909 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2910 | /* Need braces because of following else */ | else if (oclength > 0 && |
| 2911 | else if (oclength == 0) { RRETURN(MATCH_NOMATCH); } | eptr <= md->end_subject - oclength && |
| 2912 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2913 | #endif /* SUPPORT_UCP */ | |
| 2914 | else | else |
| 2915 | { | { |
| 2916 | if (memcmp(eptr, occhars, oclength) != 0) RRETURN(MATCH_NOMATCH); | CHECK_PARTIAL(); |
| 2917 | eptr += oclength; | MRRETURN(MATCH_NOMATCH); |
| 2918 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else { RRETURN (MATCH_NOMATCH); } | ||
| #endif /* SUPPORT_UCP */ | ||
| 2919 | } | } |
| 2920 | /* Control never gets here */ | /* Control never gets here */ |
| 2921 | } | } |
| # | Line 2381 for (;;) | Line 2925 for (;;) |
| 2925 | pp = eptr; | pp = eptr; |
| 2926 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 2927 | { | { |
| 2928 | if (eptr > md->end_subject - length) break; | if (eptr <= md->end_subject - length && |
| 2929 | if (memcmp(eptr, charptr, length) == 0) eptr += length; | memcmp(eptr, charptr, length) == 0) eptr += length; |
| 2930 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2931 | else if (oclength == 0) break; | else if (oclength > 0 && |
| 2932 | eptr <= md->end_subject - oclength && | |
| 2933 | memcmp(eptr, occhars, oclength) == 0) eptr += oclength; | |
| 2934 | #endif /* SUPPORT_UCP */ | |
| 2935 | else | else |
| 2936 | { | { |
| 2937 | if (memcmp(eptr, occhars, oclength) != 0) break; | CHECK_PARTIAL(); |
| 2938 | eptr += oclength; | break; |
| 2939 | } | } |
| #else /* without SUPPORT_UCP */ | ||
| else break; | ||
| #endif /* SUPPORT_UCP */ | ||
| 2940 | } | } |
| 2941 | ||
| 2942 | if (possessive) continue; | if (possessive) continue; |
| 2943 | ||
| 2944 | for(;;) | for(;;) |
| 2945 | { | { |
| 2946 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM23); |
| 2947 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 2948 | if (eptr == pp) RRETURN(MATCH_NOMATCH); | if (eptr == pp) { MRRETURN(MATCH_NOMATCH); } |
| 2949 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| 2950 | eptr--; | eptr--; |
| 2951 | BACKCHAR(eptr); | BACKCHAR(eptr); |
| 2952 | #else /* without SUPPORT_UCP */ | #else /* without SUPPORT_UCP */ |
| 2953 | eptr -= length; | eptr -= length; |
| 2954 | #endif /* SUPPORT_UCP */ | #endif /* SUPPORT_UCP */ |
| 2955 | } | } |
| 2956 | } | } |
| 2957 | /* Control never gets here */ | /* Control never gets here */ |
| 2958 | } | } |
| # | Line 2420 for (;;) | Line 2965 for (;;) |
| 2965 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF8 */ |
| 2966 | ||
| 2967 | /* When not in UTF-8 mode, load a single-byte character. */ | /* When not in UTF-8 mode, load a single-byte character. */ |
| 2968 | { | |
| 2969 | if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | fc = *ecode++; |
| fc = *ecode++; | ||
| } | ||
| 2970 | ||
| 2971 | /* 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 |
| 2972 | 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 2984 for (;;) |
| 2984 | { | { |
| 2985 | fc = md->lcc[fc]; | fc = md->lcc[fc]; |
| 2986 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 2987 | if (fc != md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 2988 | if (eptr >= md->end_subject) | |
| 2989 | { | |
| 2990 | SCHECK_PARTIAL(); | |
| 2991 | MRRETURN(MATCH_NOMATCH); | |
| 2992 | } | |
| 2993 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 2994 | } | |
| 2995 | if (min == max) continue; | if (min == max) continue; |
| 2996 | if (minimize) | if (minimize) |
| 2997 | { | { |
| # | Line 2449 for (;;) | Line 2999 for (;;) |
| 2999 | { | { |
| 3000 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM24); |
| 3001 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3002 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3003 | fc != md->lcc[*eptr++]) | if (eptr >= md->end_subject) |
| 3004 | RRETURN(MATCH_NOMATCH); | { |
| 3005 | SCHECK_PARTIAL(); | |
| 3006 | MRRETURN(MATCH_NOMATCH); | |
| 3007 | } | |
| 3008 | if (fc != md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3009 | } | } |
| 3010 | /* Control never gets here */ | /* Control never gets here */ |
| 3011 | } | } |
| # | Line 2460 for (;;) | Line 3014 for (;;) |
| 3014 | pp = eptr; | pp = eptr; |
| 3015 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3016 | { | { |
| 3017 | if (eptr >= md->end_subject || fc != md->lcc[*eptr]) break; | if (eptr >= md->end_subject) |
| 3018 | { | |
| 3019 | SCHECK_PARTIAL(); | |
| 3020 | break; | |
| 3021 | } | |
| 3022 | if (fc != md->lcc[*eptr]) break; | |
| 3023 | eptr++; | eptr++; |
| 3024 | } | } |
| 3025 | ||
| 3026 | if (possessive) continue; | if (possessive) continue; |
| 3027 | ||
| 3028 | while (eptr >= pp) | while (eptr >= pp) |
| 3029 | { | { |
| 3030 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM25); |
| 3031 | eptr--; | eptr--; |
| 3032 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3033 | } | } |
| 3034 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3035 | } | } |
| 3036 | /* Control never gets here */ | /* Control never gets here */ |
| 3037 | } | } |
| # | Line 2479 for (;;) | Line 3040 for (;;) |
| 3040 | ||
| 3041 | else | else |
| 3042 | { | { |
| 3043 | for (i = 1; i <= min; i++) if (fc != *eptr++) RRETURN(MATCH_NOMATCH); | for (i = 1; i <= min; i++) |
| 3044 | { | |
| 3045 | if (eptr >= md->end_subject) | |
| 3046 | { | |
| 3047 | SCHECK_PARTIAL(); | |
| 3048 | MRRETURN(MATCH_NOMATCH); | |
| 3049 | } | |
| 3050 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3051 | } | |
| 3052 | ||
| 3053 | if (min == max) continue; | if (min == max) continue; |
| 3054 | ||
| 3055 | if (minimize) | if (minimize) |
| 3056 | { | { |
| 3057 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 3058 | { | { |
| 3059 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM26); |
| 3060 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3061 | if (fi >= max || eptr >= md->end_subject || fc != *eptr++) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3062 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3063 | { | |
| 3064 | SCHECK_PARTIAL(); | |
| 3065 | MRRETURN(MATCH_NOMATCH); | |
| 3066 | } | |
| 3067 | if (fc != *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3068 | } | } |
| 3069 | /* Control never gets here */ | /* Control never gets here */ |
| 3070 | } | } |
| # | Line 2497 for (;;) | Line 3073 for (;;) |
| 3073 | pp = eptr; | pp = eptr; |
| 3074 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3075 | { | { |
| 3076 | if (eptr >= md->end_subject || fc != *eptr) break; | if (eptr >= md->end_subject) |
| 3077 | { | |
| 3078 | SCHECK_PARTIAL(); | |
| 3079 | break; | |
| 3080 | } | |
| 3081 | if (fc != *eptr) break; | |
| 3082 | eptr++; | eptr++; |
| 3083 | } | } |
| 3084 | if (possessive) continue; | if (possessive) continue; |
| 3085 | ||
| 3086 | while (eptr >= pp) | while (eptr >= pp) |
| 3087 | { | { |
| 3088 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM27); |
| 3089 | eptr--; | eptr--; |
| 3090 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3091 | } | } |
| 3092 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3093 | } | } |
| 3094 | } | } |
| 3095 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2516 for (;;) | Line 3098 for (;;) |
| 3098 | checking can be multibyte. */ | checking can be multibyte. */ |
| 3099 | ||
| 3100 | case OP_NOT: | case OP_NOT: |
| 3101 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3102 | { | |
| 3103 | SCHECK_PARTIAL(); | |
| 3104 | MRRETURN(MATCH_NOMATCH); | |
| 3105 | } | |
| 3106 | ecode++; | ecode++; |
| 3107 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3108 | if ((ims & PCRE_CASELESS) != 0) | if ((ims & PCRE_CASELESS) != 0) |
| # | Line 2525 for (;;) | Line 3111 for (;;) |
| 3111 | if (c < 256) | if (c < 256) |
| 3112 | #endif | #endif |
| 3113 | c = md->lcc[c]; | c = md->lcc[c]; |
| 3114 | if (md->lcc[*ecode++] == c) RRETURN(MATCH_NOMATCH); | if (md->lcc[*ecode++] == c) MRRETURN(MATCH_NOMATCH); |
| 3115 | } | } |
| 3116 | else | else |
| 3117 | { | { |
| 3118 | if (*ecode++ == c) RRETURN(MATCH_NOMATCH); | if (*ecode++ == c) MRRETURN(MATCH_NOMATCH); |
| 3119 | } | } |
| 3120 | break; | break; |
| 3121 | ||
| # | Line 2593 for (;;) | Line 3179 for (;;) |
| 3179 | max = rep_max[c]; /* zero for max => infinity */ | max = rep_max[c]; /* zero for max => infinity */ |
| 3180 | if (max == 0) max = INT_MAX; | if (max == 0) max = INT_MAX; |
| 3181 | ||
| 3182 | /* 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. */ | ||
| 3183 | ||
| 3184 | REPEATNOTCHAR: | REPEATNOTCHAR: |
| if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 3185 | fc = *ecode++; | fc = *ecode++; |
| 3186 | ||
| 3187 | /* 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 3206 for (;;) |
| 3206 | register unsigned int d; | register unsigned int d; |
| 3207 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3208 | { | { |
| 3209 | if (eptr >= md->end_subject) | |
| 3210 | { | |
| 3211 | SCHECK_PARTIAL(); | |
| 3212 | MRRETURN(MATCH_NOMATCH); | |
| 3213 | } | |
| 3214 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3215 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3216 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3217 | } | } |
| 3218 | } | } |
| 3219 | else | else |
| # | Line 2634 for (;;) | Line 3222 for (;;) |
| 3222 | /* Not UTF-8 mode */ | /* Not UTF-8 mode */ |
| 3223 | { | { |
| 3224 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3225 | if (fc == md->lcc[*eptr++]) RRETURN(MATCH_NOMATCH); | { |
| 3226 | if (eptr >= md->end_subject) | |
| 3227 | { | |
| 3228 | SCHECK_PARTIAL(); | |
| 3229 | MRRETURN(MATCH_NOMATCH); | |
| 3230 | } | |
| 3231 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3232 | } | |
| 3233 | } | } |
| 3234 | ||
| 3235 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2650 for (;;) | Line 3245 for (;;) |
| 3245 | { | { |
| 3246 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM28); |
| 3247 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3248 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3249 | if (eptr >= md->end_subject) | |
| 3250 | { | |
| 3251 | SCHECK_PARTIAL(); | |
| 3252 | MRRETURN(MATCH_NOMATCH); | |
| 3253 | } | |
| 3254 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3255 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3256 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3257 | } | } |
| 3258 | } | } |
| 3259 | else | else |
| # | Line 2665 for (;;) | Line 3264 for (;;) |
| 3264 | { | { |
| 3265 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM29); |
| 3266 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3267 | if (fi >= max || eptr >= md->end_subject || fc == md->lcc[*eptr++]) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3268 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3269 | { | |
| 3270 | SCHECK_PARTIAL(); | |
| 3271 | MRRETURN(MATCH_NOMATCH); | |
| 3272 | } | |
| 3273 | if (fc == md->lcc[*eptr++]) MRRETURN(MATCH_NOMATCH); | |
| 3274 | } | } |
| 3275 | } | } |
| 3276 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2686 for (;;) | Line 3290 for (;;) |
| 3290 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3291 | { | { |
| 3292 | int len = 1; | int len = 1; |
| 3293 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 3294 | { | |
| 3295 | SCHECK_PARTIAL(); | |
| 3296 | break; | |
| 3297 | } | |
| 3298 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3299 | if (d < 256) d = md->lcc[d]; | if (d < 256) d = md->lcc[d]; |
| 3300 | if (fc == d) break; | if (fc == d) break; |
| # | Line 2707 for (;;) | Line 3315 for (;;) |
| 3315 | { | { |
| 3316 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3317 | { | { |
| 3318 | if (eptr >= md->end_subject || fc == md->lcc[*eptr]) break; | if (eptr >= md->end_subject) |
| 3319 | { | |
| 3320 | SCHECK_PARTIAL(); | |
| 3321 | break; | |
| 3322 | } | |
| 3323 | if (fc == md->lcc[*eptr]) break; | |
| 3324 | eptr++; | eptr++; |
| 3325 | } | } |
| 3326 | if (possessive) continue; | if (possessive) continue; |
| # | Line 2719 for (;;) | Line 3332 for (;;) |
| 3332 | } | } |
| 3333 | } | } |
| 3334 | ||
| 3335 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3336 | } | } |
| 3337 | /* Control never gets here */ | /* Control never gets here */ |
| 3338 | } | } |
| # | Line 2735 for (;;) | Line 3348 for (;;) |
| 3348 | register unsigned int d; | register unsigned int d; |
| 3349 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3350 | { | { |
| 3351 | if (eptr >= md->end_subject) | |
| 3352 | { | |
| 3353 | SCHECK_PARTIAL(); | |
| 3354 | MRRETURN(MATCH_NOMATCH); | |
| 3355 | } | |
| 3356 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3357 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3358 | } | } |
| 3359 | } | } |
| 3360 | else | else |
| # | Line 2744 for (;;) | Line 3362 for (;;) |
| 3362 | /* Not UTF-8 mode */ | /* Not UTF-8 mode */ |
| 3363 | { | { |
| 3364 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3365 | if (fc == *eptr++) RRETURN(MATCH_NOMATCH); | { |
| 3366 | if (eptr >= md->end_subject) | |
| 3367 | { | |
| 3368 | SCHECK_PARTIAL(); | |
| 3369 | MRRETURN(MATCH_NOMATCH); | |
| 3370 | } | |
| 3371 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3372 | } | |
| 3373 | } | } |
| 3374 | ||
| 3375 | if (min == max) continue; | if (min == max) continue; |
| # | Line 2760 for (;;) | Line 3385 for (;;) |
| 3385 | { | { |
| 3386 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM32); |
| 3387 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3388 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3389 | if (eptr >= md->end_subject) | |
| 3390 | { | |
| 3391 | SCHECK_PARTIAL(); | |
| 3392 | MRRETURN(MATCH_NOMATCH); | |
| 3393 | } | |
| 3394 | GETCHARINC(d, eptr); | GETCHARINC(d, eptr); |
| 3395 | if (fc == d) RRETURN(MATCH_NOMATCH); | if (fc == d) MRRETURN(MATCH_NOMATCH); |
| 3396 | } | } |
| 3397 | } | } |
| 3398 | else | else |
| # | Line 2773 for (;;) | Line 3403 for (;;) |
| 3403 | { | { |
| 3404 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM33); |
| 3405 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 3406 | if (fi >= max || eptr >= md->end_subject || fc == *eptr++) | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 3407 | RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3408 | { | |
| 3409 | SCHECK_PARTIAL(); | |
| 3410 | MRRETURN(MATCH_NOMATCH); | |
| 3411 | } | |
| 3412 | if (fc == *eptr++) MRRETURN(MATCH_NOMATCH); | |
| 3413 | } | } |
| 3414 | } | } |
| 3415 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2794 for (;;) | Line 3429 for (;;) |
| 3429 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3430 | { | { |
| 3431 | int len = 1; | int len = 1; |
| 3432 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 3433 | { | |
| 3434 | SCHECK_PARTIAL(); | |
| 3435 | break; | |
| 3436 | } | |
| 3437 | GETCHARLEN(d, eptr, len); | GETCHARLEN(d, eptr, len); |
| 3438 | if (fc == d) break; | if (fc == d) break; |
| 3439 | eptr += len; | eptr += len; |
| # | Line 2814 for (;;) | Line 3453 for (;;) |
| 3453 | { | { |
| 3454 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 3455 | { | { |
| 3456 | if (eptr >= md->end_subject || fc == *eptr) break; | if (eptr >= md->end_subject) |
| 3457 | { | |
| 3458 | SCHECK_PARTIAL(); | |
| 3459 | break; | |
| 3460 | } | |
| 3461 | if (fc == *eptr) break; | |
| 3462 | eptr++; | eptr++; |
| 3463 | } | } |
| 3464 | if (possessive) continue; | if (possessive) continue; |
| # | Line 2826 for (;;) | Line 3470 for (;;) |
| 3470 | } | } |
| 3471 | } | } |
| 3472 | ||
| 3473 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3474 | } | } |
| 3475 | } | } |
| 3476 | /* Control never gets here */ | /* Control never gets here */ |
| # | Line 2908 for (;;) | Line 3552 for (;;) |
| 3552 | ||
| 3553 | /* First, ensure the minimum number of matches are present. Use inline | /* First, ensure the minimum number of matches are present. Use inline |
| 3554 | 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 |
| 3555 | (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 | ||
| 3556 | 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 |
| 3557 | and single-bytes. */ | and single-bytes. */ |
| 3558 | ||
| if (min > md->end_subject - eptr) RRETURN(MATCH_NOMATCH); | ||
| 3559 | if (min > 0) | if (min > 0) |
| 3560 | { | { |
| 3561 | #ifdef SUPPORT_UCP | #ifdef SUPPORT_UCP |
| # | Line 2923 for (;;) | Line 3564 for (;;) |
| 3564 | switch(prop_type) | switch(prop_type) |
| 3565 | { | { |
| 3566 | case PT_ANY: | case PT_ANY: |
| 3567 | if (prop_fail_result) RRETURN(MATCH_NOMATCH); | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); |
| 3568 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3569 | { | { |
| 3570 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3571 | { | |
| 3572 | SCHECK_PARTIAL(); | |
| 3573 | MRRETURN(MATCH_NOMATCH); | |
| 3574 | } | |
| 3575 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3576 | } | } |
| 3577 | break; | break; |
| # | Line 2934 for (;;) | Line 3579 for (;;) |
| 3579 | case PT_LAMP: | case PT_LAMP: |
| 3580 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3581 | { | { |
| 3582 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3583 | { | |
| 3584 | SCHECK_PARTIAL(); | |
| 3585 | MRRETURN(MATCH_NOMATCH); | |
| 3586 | } | |
| 3587 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3588 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 3589 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 3590 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| 3591 | prop_chartype == ucp_Lt) == prop_fail_result) | prop_chartype == ucp_Lt) == prop_fail_result) |
| 3592 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3593 | } | } |
| 3594 | break; | break; |
| 3595 | ||
| 3596 | case PT_GC: | case PT_GC: |
| 3597 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3598 | { | { |
| 3599 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3600 | { | |
| 3601 | SCHECK_PARTIAL(); | |
| 3602 | MRRETURN(MATCH_NOMATCH); | |
| 3603 | } | |
| 3604 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3605 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3606 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| 3607 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3608 | } | } |
| 3609 | break; | break; |
| 3610 | ||
| 3611 | case PT_PC: | case PT_PC: |
| 3612 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3613 | { | { |
| 3614 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3615 | { | |
| 3616 | SCHECK_PARTIAL(); | |
| 3617 | MRRETURN(MATCH_NOMATCH); | |
| 3618 | } | |
| 3619 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3620 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 3621 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| 3622 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3623 | } | } |
| 3624 | break; | break; |
| 3625 | ||
| 3626 | case PT_SC: | case PT_SC: |
| 3627 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3628 | { | { |
| 3629 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3630 | { | |
| 3631 | SCHECK_PARTIAL(); | |
| 3632 | MRRETURN(MATCH_NOMATCH); | |
| 3633 | } | |
| 3634 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3635 | prop_script = UCD_SCRIPT(c); | prop_script = UCD_SCRIPT(c); |
| 3636 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| 3637 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3638 | } | |
| 3639 | break; | |
| 3640 | ||
| 3641 | case PT_ALNUM: | |
| 3642 | for (i = 1; i <= min; i++) | |
| 3643 | { | |
| 3644 | if (eptr >= md->end_subject) | |
| 3645 | { | |
| 3646 | SCHECK_PARTIAL(); | |
| 3647 | MRRETURN(MATCH_NOMATCH); | |
| 3648 | } | |
| 3649 | GETCHARINCTEST(c, eptr); | |
| 3650 | prop_category = UCD_CATEGORY(c); | |
| 3651 | if ((prop_category == ucp_L || prop_category == ucp_N) | |
| 3652 | == prop_fail_result) | |
| 3653 | MRRETURN(MATCH_NOMATCH); | |
| 3654 | } | |
| 3655 | break; | |
| 3656 | ||
| 3657 | case PT_SPACE: /* Perl space */ | |
| 3658 | for (i = 1; i <= min; i++) | |
| 3659 | { | |
| 3660 | if (eptr >= md->end_subject) | |
| 3661 | { | |
| 3662 | SCHECK_PARTIAL(); | |
| 3663 | MRRETURN(MATCH_NOMATCH); | |
| 3664 | } | |
| 3665 | GETCHARINCTEST(c, eptr); | |
| 3666 | prop_category = UCD_CATEGORY(c); | |
| 3667 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 3668 | c == CHAR_FF || c == CHAR_CR) | |
| 3669 | == prop_fail_result) | |
| 3670 | MRRETURN(MATCH_NOMATCH); | |
| 3671 | } | |
| 3672 | break; | |
| 3673 | ||
| 3674 | case PT_PXSPACE: /* POSIX space */ | |
| 3675 | for (i = 1; i <= min; i++) | |
| 3676 | { | |
| 3677 | if (eptr >= md->end_subject) | |
| 3678 | { | |
| 3679 | SCHECK_PARTIAL(); | |
| 3680 | MRRETURN(MATCH_NOMATCH); | |
| 3681 | } | |
| 3682 | GETCHARINCTEST(c, eptr); | |
| 3683 | prop_category = UCD_CATEGORY(c); | |
| 3684 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 3685 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | |
| 3686 | == prop_fail_result) | |
| 3687 | MRRETURN(MATCH_NOMATCH); | |
| 3688 | } | } |
| 3689 | break; | break; |
| 3690 | ||
| 3691 | case PT_WORD: | |
| 3692 | for (i = 1; i <= min; i++) | |
| 3693 | { | |
| 3694 | if (eptr >= md->end_subject) | |
| 3695 | { | |
| 3696 | SCHECK_PARTIAL(); | |
| 3697 | MRRETURN(MATCH_NOMATCH); | |
| 3698 | } | |
| 3699 | GETCHARINCTEST(c, eptr); | |
| 3700 | prop_category = UCD_CATEGORY(c); | |
| 3701 | if ((prop_category == ucp_L || prop_category == ucp_N || | |
| 3702 | c == CHAR_UNDERSCORE) | |
| 3703 | == prop_fail_result) | |
| 3704 | MRRETURN(MATCH_NOMATCH); | |
| 3705 | } | |
| 3706 | break; | |
| 3707 | ||
| 3708 | /* This should not occur */ | |
| 3709 | ||
| 3710 | default: | default: |
| 3711 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 3712 | } | } |
| # | Line 2989 for (;;) | Line 3719 for (;;) |
| 3719 | { | { |
| 3720 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3721 | { | { |
| 3722 | if (eptr >= md->end_subject) | |
| 3723 | { | |
| 3724 | SCHECK_PARTIAL(); | |
| 3725 | MRRETURN(MATCH_NOMATCH); | |
| 3726 | } | |
| 3727 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 3728 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3729 | if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); | if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 3730 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 3731 | { | { |
| 3732 | int len = 1; | int len = 1; |
| 3733 | if (!utf8) c = *eptr; else | if (!utf8) c = *eptr; |
| 3734 | { | else { GETCHARLEN(c, eptr, len); } |
| GETCHARLEN(c, eptr, len); | ||
| } | ||
| 3735 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 3736 | if (prop_category != ucp_M) break; | if (prop_category != ucp_M) break; |
| 3737 | eptr += len; | eptr += len; |
| # | Line 3017 for (;;) | Line 3750 for (;;) |
| 3750 | case OP_ANY: | case OP_ANY: |
| 3751 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3752 | { | { |
| 3753 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) | if (eptr >= md->end_subject) |
| 3754 | RRETURN(MATCH_NOMATCH); | { |
| 3755 | SCHECK_PARTIAL(); | |
| 3756 | MRRETURN(MATCH_NOMATCH); | |
| 3757 | } | |
| 3758 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | |
| 3759 | eptr++; | eptr++; |
| 3760 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 3761 | } | } |
| # | Line 3027 for (;;) | Line 3764 for (;;) |
| 3764 | case OP_ALLANY: | case OP_ALLANY: |
| 3765 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3766 | { | { |
| 3767 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3768 | { | |
| 3769 | SCHECK_PARTIAL(); | |
| 3770 | MRRETURN(MATCH_NOMATCH); | |
| 3771 | } | |
| 3772 | eptr++; | eptr++; |
| 3773 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 3774 | } | } |
| 3775 | break; | break; |
| 3776 | ||
| 3777 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 3778 | if (eptr > md->end_subject - min) MRRETURN(MATCH_NOMATCH); | |
| 3779 | eptr += min; | eptr += min; |
| 3780 | break; | break; |
| 3781 | ||
| 3782 | case OP_ANYNL: | case OP_ANYNL: |
| 3783 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3784 | { | { |
| 3785 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3786 | { | |
| 3787 | SCHECK_PARTIAL(); | |
| 3788 | MRRETURN(MATCH_NOMATCH); | |
| 3789 | } | |
| 3790 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3791 | switch(c) | switch(c) |
| 3792 | { | { |
| 3793 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3794 | case 0x000d: | case 0x000d: |
| 3795 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 3796 | break; | break; |
| # | Line 3057 for (;;) | Line 3803 for (;;) |
| 3803 | case 0x0085: | case 0x0085: |
| 3804 | case 0x2028: | case 0x2028: |
| 3805 | case 0x2029: | case 0x2029: |
| 3806 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 3807 | break; | break; |
| 3808 | } | } |
| 3809 | } | } |
| # | Line 3066 for (;;) | Line 3812 for (;;) |
| 3812 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 3813 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3814 | { | { |
| 3815 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3816 | { | |
| 3817 | SCHECK_PARTIAL(); | |
| 3818 | MRRETURN(MATCH_NOMATCH); | |
| 3819 | } | |
| 3820 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3821 | switch(c) | switch(c) |
| 3822 | { | { |
| # | Line 3090 for (;;) | Line 3840 for (;;) |
| 3840 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 3841 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 3842 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 3843 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3844 | } | } |
| 3845 | } | } |
| 3846 | break; | break; |
| # | Line 3098 for (;;) | Line 3848 for (;;) |
| 3848 | case OP_HSPACE: | case OP_HSPACE: |
| 3849 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3850 | { | { |
| 3851 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3852 | { | |
| 3853 | SCHECK_PARTIAL(); | |
| 3854 | MRRETURN(MATCH_NOMATCH); | |
| 3855 | } | |
| 3856 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3857 | switch(c) | switch(c) |
| 3858 | { | { |
| 3859 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3860 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 3861 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 3862 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3130 for (;;) | Line 3884 for (;;) |
| 3884 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 3885 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3886 | { | { |
| 3887 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3888 | { | |
| 3889 | SCHECK_PARTIAL(); | |
| 3890 | MRRETURN(MATCH_NOMATCH); | |
| 3891 | } | |
| 3892 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3893 | switch(c) | switch(c) |
| 3894 | { | { |
| # | Line 3142 for (;;) | Line 3900 for (;;) |
| 3900 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 3901 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 3902 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 3903 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3904 | } | } |
| 3905 | } | } |
| 3906 | break; | break; |
| # | Line 3150 for (;;) | Line 3908 for (;;) |
| 3908 | case OP_VSPACE: | case OP_VSPACE: |
| 3909 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3910 | { | { |
| 3911 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3912 | { | |
| 3913 | SCHECK_PARTIAL(); | |
| 3914 | MRRETURN(MATCH_NOMATCH); | |
| 3915 | } | |
| 3916 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3917 | switch(c) | switch(c) |
| 3918 | { | { |
| 3919 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 3920 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 3921 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 3922 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3170 for (;;) | Line 3932 for (;;) |
| 3932 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 3933 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3934 | { | { |
| 3935 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 3936 | { | |
| 3937 | SCHECK_PARTIAL(); | |
| 3938 | MRRETURN(MATCH_NOMATCH); | |
| 3939 | } | |
| 3940 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 3941 | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 128 && (md->ctypes[c] & ctype_digit) != 0) |
| 3942 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 3943 | } | } |
| 3944 | break; | break; |
| 3945 | ||
| 3946 | case OP_DIGIT: | case OP_DIGIT: |
| 3947 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3948 | { | { |
| 3949 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3950 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | { |
| 3951 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3952 | MRRETURN(MATCH_NOMATCH); | |
| 3953 | } | |
| 3954 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_digit) == 0) | |
| 3955 | MRRETURN(MATCH_NOMATCH); | |
| 3956 | /* 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 */ |
| 3957 | } | } |
| 3958 | break; | break; |
| # | Line 3190 for (;;) | Line 3960 for (;;) |
| 3960 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 3961 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3962 | { | { |
| 3963 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3964 | (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0)) | { |
| 3965 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3966 | MRRETURN(MATCH_NOMATCH); | |
| 3967 | } | |
| 3968 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_space) != 0) | |
| 3969 | MRRETURN(MATCH_NOMATCH); | |
| 3970 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); |
| 3971 | } | } |
| 3972 | break; | break; |
| # | Line 3200 for (;;) | Line 3974 for (;;) |
| 3974 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 3975 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3976 | { | { |
| 3977 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3978 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | { |
| 3979 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3980 | MRRETURN(MATCH_NOMATCH); | |
| 3981 | } | |
| 3982 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_space) == 0) | |
| 3983 | MRRETURN(MATCH_NOMATCH); | |
| 3984 | /* 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 */ |
| 3985 | } | } |
| 3986 | break; | break; |
| # | Line 3210 for (;;) | Line 3988 for (;;) |
| 3988 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 3989 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 3990 | { | { |
| 3991 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 3992 | (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0)) | { |
| 3993 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 3994 | MRRETURN(MATCH_NOMATCH); | |
| 3995 | } | |
| 3996 | if (*eptr < 128 && (md->ctypes[*eptr] & ctype_word) != 0) | |
| 3997 | MRRETURN(MATCH_NOMATCH); | |
| 3998 | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); | while (++eptr < md->end_subject && (*eptr & 0xc0) == 0x80); |
| 3999 | } | } |
| 4000 | break; | break; |
| # | Line 3220 for (;;) | Line 4002 for (;;) |
| 4002 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4003 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4004 | { | { |
| 4005 | if (eptr >= md->end_subject || | if (eptr >= md->end_subject) |
| 4006 | *eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | { |
| 4007 | RRETURN(MATCH_NOMATCH); | SCHECK_PARTIAL(); |
| 4008 | MRRETURN(MATCH_NOMATCH); | |
| 4009 | } | |
| 4010 | if (*eptr >= 128 || (md->ctypes[*eptr++] & ctype_word) == 0) | |
| 4011 | MRRETURN(MATCH_NOMATCH); | |
| 4012 | /* 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 */ |
| 4013 | } | } |
| 4014 | break; | break; |
| # | Line 3235 for (;;) | Line 4021 for (;;) |
| 4021 | #endif /* SUPPORT_UTF8 */ | #endif /* SUPPORT_UTF8 */ |
| 4022 | ||
| 4023 | /* 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 |
| 4024 | 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. */ | ||
| 4025 | ||
| 4026 | switch(ctype) | switch(ctype) |
| 4027 | { | { |
| 4028 | case OP_ANY: | case OP_ANY: |
| 4029 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4030 | { | { |
| 4031 | if (IS_NEWLINE(eptr)) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4032 | { | |
| 4033 | SCHECK_PARTIAL(); | |
| 4034 | MRRETURN(MATCH_NOMATCH); | |
| 4035 | } | |
| 4036 | if (IS_NEWLINE(eptr)) MRRETURN(MATCH_NOMATCH); | |
| 4037 | eptr++; | eptr++; |
| 4038 | } | } |
| 4039 | break; | break; |
| 4040 | ||
| 4041 | case OP_ALLANY: | case OP_ALLANY: |
| 4042 | if (eptr > md->end_subject - min) | |
| 4043 | { | |
| 4044 | SCHECK_PARTIAL(); | |
| 4045 | MRRETURN(MATCH_NOMATCH); | |
| 4046 | } | |
| 4047 | eptr += min; | eptr += min; |
| 4048 | break; | break; |
| 4049 | ||
| 4050 | case OP_ANYBYTE: | case OP_ANYBYTE: |
| 4051 | if (eptr > md->end_subject - min) | |
| 4052 | { | |
| 4053 | SCHECK_PARTIAL(); | |
| 4054 | MRRETURN(MATCH_NOMATCH); | |
| 4055 | } | |
| 4056 | eptr += min; | eptr += min; |
| 4057 | break; | break; |
| 4058 | ||
| /* Because of the CRLF case, we can't assume the minimum number of | ||
| bytes are present in this case. */ | ||
| 4059 | case OP_ANYNL: | case OP_ANYNL: |
| 4060 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4061 | { | { |
| 4062 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4063 | { | |
| 4064 | SCHECK_PARTIAL(); | |
| 4065 | MRRETURN(MATCH_NOMATCH); | |
| 4066 | } | |
| 4067 | switch(*eptr++) | switch(*eptr++) |
| 4068 | { | { |
| 4069 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4070 | case 0x000d: | case 0x000d: |
| 4071 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4072 | break; | break; |
| # | Line 3275 for (;;) | Line 4076 for (;;) |
| 4076 | case 0x000b: | case 0x000b: |
| 4077 | case 0x000c: | case 0x000c: |
| 4078 | case 0x0085: | case 0x0085: |
| 4079 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4080 | break; | break; |
| 4081 | } | } |
| 4082 | } | } |
| # | Line 3284 for (;;) | Line 4085 for (;;) |
| 4085 | case OP_NOT_HSPACE: | case OP_NOT_HSPACE: |
| 4086 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4087 | { | { |
| 4088 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4089 | { | |
| 4090 | SCHECK_PARTIAL(); | |
| 4091 | MRRETURN(MATCH_NOMATCH); | |
| 4092 | } | |
| 4093 | switch(*eptr++) | switch(*eptr++) |
| 4094 | { | { |
| 4095 | default: break; | default: break; |
| 4096 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4097 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4098 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4099 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4100 | } | } |
| 4101 | } | } |
| 4102 | break; | break; |
| # | Line 3299 for (;;) | Line 4104 for (;;) |
| 4104 | case OP_HSPACE: | case OP_HSPACE: |
| 4105 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4106 | { | { |
| 4107 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4108 | { | |
| 4109 | SCHECK_PARTIAL(); | |
| 4110 | MRRETURN(MATCH_NOMATCH); | |
| 4111 | } | |
| 4112 | switch(*eptr++) | switch(*eptr++) |
| 4113 | { | { |
| 4114 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4115 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4116 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4117 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3314 for (;;) | Line 4123 for (;;) |
| 4123 | case OP_NOT_VSPACE: | case OP_NOT_VSPACE: |
| 4124 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4125 | { | { |
| 4126 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4127 | { | |
| 4128 | SCHECK_PARTIAL(); | |
| 4129 | MRRETURN(MATCH_NOMATCH); | |
| 4130 | } | |
| 4131 | switch(*eptr++) | switch(*eptr++) |
| 4132 | { | { |
| 4133 | default: break; | default: break; |
| # | Line 3323 for (;;) | Line 4136 for (;;) |
| 4136 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4137 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4138 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4139 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4140 | } | } |
| 4141 | } | } |
| 4142 | break; | break; |
| # | Line 3331 for (;;) | Line 4144 for (;;) |
| 4144 | case OP_VSPACE: | case OP_VSPACE: |
| 4145 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4146 | { | { |
| 4147 | if (eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (eptr >= md->end_subject) |
| 4148 | { | |
| 4149 | SCHECK_PARTIAL(); | |
| 4150 | MRRETURN(MATCH_NOMATCH); | |
| 4151 | } | |
| 4152 | switch(*eptr++) | switch(*eptr++) |
| 4153 | { | { |
| 4154 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4155 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4156 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4157 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3347 for (;;) | Line 4164 for (;;) |
| 4164 | ||
| 4165 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4166 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4167 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); | { |
| 4168 | if (eptr >= md->end_subject) | |
| 4169 | { | |
| 4170 | SCHECK_PARTIAL(); | |
| 4171 | MRRETURN(MATCH_NOMATCH); | |
| 4172 | } | |
| 4173 | if ((md->ctypes[*eptr++] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); | |
| 4174 | } | |
| 4175 | break; | break; |
| 4176 | ||
| 4177 | case OP_DIGIT: | case OP_DIGIT: |
| 4178 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4179 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); | { |
| 4180 | if (eptr >= md->end_subject) | |
| 4181 | { | |
| 4182 | SCHECK_PARTIAL(); | |
| 4183 | MRRETURN(MATCH_NOMATCH); | |
| 4184 | } | |
| 4185 | if ((md->ctypes[*eptr++] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); | |
| 4186 | } | |
| 4187 | break; | break; |
| 4188 | ||
| 4189 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4190 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4191 | if ((md->ctypes[*eptr++] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); | { |
| 4192 | if (eptr >= md->end_subject) | |
| 4193 | { | |
| 4194 | SCHECK_PARTIAL(); | |
| 4195 | MRRETURN(MATCH_NOMATCH); | |
| 4196 | } | |
| 4197 | if ((md->ctypes[*eptr++] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); | |
| 4198 | } | |
| 4199 | break; | break; |
| 4200 | ||
| 4201 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4202 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4203 | if ((md->ctypes[*eptr++] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); | { |
| 4204 | if (eptr >= md->end_subject) | |
| 4205 | { | |
| 4206 | SCHECK_PARTIAL(); | |
| 4207 | MRRETURN(MATCH_NOMATCH); | |
| 4208 | } | |
| 4209 | if ((md->ctypes[*eptr++] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); | |
| 4210 | } | |
| 4211 | break; | break; |
| 4212 | ||
| 4213 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4214 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4215 | { | |
| 4216 | if (eptr >= md->end_subject) | |
| 4217 | { | |
| 4218 | SCHECK_PARTIAL(); | |
| 4219 | MRRETURN(MATCH_NOMATCH); | |
| 4220 | } | |
| 4221 | if ((md->ctypes[*eptr++] & ctype_word) != 0) | if ((md->ctypes[*eptr++] & ctype_word) != 0) |
| 4222 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4223 | } | |
| 4224 | break; | break; |
| 4225 | ||
| 4226 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4227 | for (i = 1; i <= min; i++) | for (i = 1; i <= min; i++) |
| 4228 | { | |
| 4229 | if (eptr >= md->end_subject) | |
| 4230 | { | |
| 4231 | SCHECK_PARTIAL(); | |
| 4232 | MRRETURN(MATCH_NOMATCH); | |
| 4233 | } | |
| 4234 | if ((md->ctypes[*eptr++] & ctype_word) == 0) | if ((md->ctypes[*eptr++] & ctype_word) == 0) |
| 4235 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4236 | } | |
| 4237 | break; | break; |
| 4238 | ||
| 4239 | default: | default: |
| # | Line 3402 for (;;) | Line 4261 for (;;) |
| 4261 | { | { |
| 4262 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM36); |
| 4263 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4264 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4265 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4266 | if (prop_fail_result) RRETURN(MATCH_NOMATCH); | { |
| 4267 | SCHECK_PARTIAL(); | |
| 4268 | MRRETURN(MATCH_NOMATCH); | |
| 4269 | } | |
| 4270 | GETCHARINCTEST(c, eptr); | |
| 4271 | if (prop_fail_result) MRRETURN(MATCH_NOMATCH); | |
| 4272 | } | } |
| 4273 | /* Control never gets here */ | /* Control never gets here */ |
| 4274 | ||
| # | Line 3413 for (;;) | Line 4277 for (;;) |
| 4277 | { | { |
| 4278 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM37); |
| 4279 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4280 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4281 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4282 | { | |
| 4283 | SCHECK_PARTIAL(); | |
| 4284 | MRRETURN(MATCH_NOMATCH); | |
| 4285 | } | |
| 4286 | GETCHARINCTEST(c, eptr); | |
| 4287 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4288 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 4289 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| 4290 | prop_chartype == ucp_Lt) == prop_fail_result) | prop_chartype == ucp_Lt) == prop_fail_result) |
| 4291 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4292 | } | |
| 4293 | /* Control never gets here */ | |
| 4294 | ||
| 4295 | case PT_GC: | |
| 4296 | for (fi = min;; fi++) | |
| 4297 | { | |
| 4298 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38); | |
| 4299 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4300 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4301 | if (eptr >= md->end_subject) | |
| 4302 | { | |
| 4303 | SCHECK_PARTIAL(); | |
| 4304 | MRRETURN(MATCH_NOMATCH); | |
| 4305 | } | |
| 4306 | GETCHARINCTEST(c, eptr); | |
| 4307 | prop_category = UCD_CATEGORY(c); | |
| 4308 | if ((prop_category == prop_value) == prop_fail_result) | |
| 4309 | MRRETURN(MATCH_NOMATCH); | |
| 4310 | } | |
| 4311 | /* Control never gets here */ | |
| 4312 | ||
| 4313 | case PT_PC: | |
| 4314 | for (fi = min;; fi++) | |
| 4315 | { | |
| 4316 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39); | |
| 4317 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4318 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4319 | if (eptr >= md->end_subject) | |
| 4320 | { | |
| 4321 | SCHECK_PARTIAL(); | |
| 4322 | MRRETURN(MATCH_NOMATCH); | |
| 4323 | } | |
| 4324 | GETCHARINCTEST(c, eptr); | |
| 4325 | prop_chartype = UCD_CHARTYPE(c); | |
| 4326 | if ((prop_chartype == prop_value) == prop_fail_result) | |
| 4327 | MRRETURN(MATCH_NOMATCH); | |
| 4328 | } | |
| 4329 | /* Control never gets here */ | |
| 4330 | ||
| 4331 | case PT_SC: | |
| 4332 | for (fi = min;; fi++) | |
| 4333 | { | |
| 4334 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40); | |
| 4335 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4336 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4337 | if (eptr >= md->end_subject) | |
| 4338 | { | |
| 4339 | SCHECK_PARTIAL(); | |
| 4340 | MRRETURN(MATCH_NOMATCH); | |
| 4341 | } | |
| 4342 | GETCHARINCTEST(c, eptr); | |
| 4343 | prop_script = UCD_SCRIPT(c); | |
| 4344 | if ((prop_script == prop_value) == prop_fail_result) | |
| 4345 | MRRETURN(MATCH_NOMATCH); | |
| 4346 | } | |
| 4347 | /* Control never gets here */ | |
| 4348 | ||
| 4349 | case PT_ALNUM: | |
| 4350 | for (fi = min;; fi++) | |
| 4351 | { | |
| 4352 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM59); | |
| 4353 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | |
| 4354 | if (fi >= max) MRRETURN(MATCH_NOMATCH); | |
| 4355 | if (eptr >= md->end_subject) | |
| 4356 | { | |
| 4357 | SCHECK_PARTIAL(); | |
| 4358 | MRRETURN(MATCH_NOMATCH); | |
| 4359 | } | |
| 4360 | GETCHARINCTEST(c, eptr); | |
| 4361 | prop_category = UCD_CATEGORY(c); | |
| 4362 | if ((prop_category == ucp_L || prop_category == ucp_N) | |
| 4363 | == prop_fail_result) | |
| 4364 | MRRETURN(MATCH_NOMATCH); | |
| 4365 | } | } |
| 4366 | /* Control never gets here */ | /* Control never gets here */ |
| 4367 | ||
| 4368 | case PT_GC: | case PT_SPACE: /* Perl space */ |
| 4369 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 4370 | { | { |
| 4371 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM38); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM60); |
| 4372 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4373 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4374 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4375 | { | |
| 4376 | SCHECK_PARTIAL(); | |
| 4377 | MRRETURN(MATCH_NOMATCH); | |
| 4378 | } | |
| 4379 | GETCHARINCTEST(c, eptr); | |
| 4380 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4381 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || |
| 4382 | RRETURN(MATCH_NOMATCH); | c == CHAR_FF || c == CHAR_CR) |
| 4383 | == prop_fail_result) | |
| 4384 | MRRETURN(MATCH_NOMATCH); | |
| 4385 | } | } |
| 4386 | /* Control never gets here */ | /* Control never gets here */ |
| 4387 | ||
| 4388 | case PT_PC: | case PT_PXSPACE: /* POSIX space */ |
| 4389 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 4390 | { | { |
| 4391 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM39); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM61); |
| 4392 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4393 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4394 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4395 | prop_chartype = UCD_CHARTYPE(c); | { |
| 4396 | if ((prop_chartype == prop_value) == prop_fail_result) | SCHECK_PARTIAL(); |
| 4397 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4398 | } | |
| 4399 | GETCHARINCTEST(c, eptr); | |
| 4400 | prop_category = UCD_CATEGORY(c); | |
| 4401 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 4402 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | |
| 4403 | == prop_fail_result) | |
| 4404 | MRRETURN(MATCH_NOMATCH); | |
| 4405 | } | } |
| 4406 | /* Control never gets here */ | /* Control never gets here */ |
| 4407 | ||
| 4408 | case PT_SC: | case PT_WORD: |
| 4409 | for (fi = min;; fi++) | for (fi = min;; fi++) |
| 4410 | { | { |
| 4411 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM40); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM62); |
| 4412 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4413 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4414 | GETCHARINC(c, eptr); | if (eptr >= md->end_subject) |
| 4415 | prop_script = UCD_SCRIPT(c); | { |
| 4416 | if ((prop_script == prop_value) == prop_fail_result) | SCHECK_PARTIAL(); |
| 4417 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4418 | } | |
| 4419 | GETCHARINCTEST(c, eptr); | |
| 4420 | prop_category = UCD_CATEGORY(c); | |
| 4421 | if ((prop_category == ucp_L || | |
| 4422 | prop_category == ucp_N || | |
| 4423 | c == CHAR_UNDERSCORE) | |
| 4424 | == prop_fail_result) | |
| 4425 | MRRETURN(MATCH_NOMATCH); | |
| 4426 | } | } |
| 4427 | /* Control never gets here */ | /* Control never gets here */ |
| 4428 | ||
| 4429 | /* This should never occur */ | |
| 4430 | ||
| 4431 | default: | default: |
| 4432 | RRETURN(PCRE_ERROR_INTERNAL); | RRETURN(PCRE_ERROR_INTERNAL); |
| 4433 | } | } |
| # | Line 3476 for (;;) | Line 4442 for (;;) |
| 4442 | { | { |
| 4443 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM41); |
| 4444 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4445 | if (fi >= max || eptr >= md->end_subject) RRETURN(MATCH_NOMATCH); | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4446 | if (eptr >= md->end_subject) | |
| 4447 | { | |
| 4448 | SCHECK_PARTIAL(); | |
| 4449 | MRRETURN(MATCH_NOMATCH); | |
| 4450 | } | |
| 4451 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4452 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4453 | if (prop_category == ucp_M) RRETURN(MATCH_NOMATCH); | if (prop_category == ucp_M) MRRETURN(MATCH_NOMATCH); |
| 4454 | while (eptr < md->end_subject) | while (eptr < md->end_subject) |
| 4455 | { | { |
| 4456 | int len = 1; | int len = 1; |
| 4457 | if (!utf8) c = *eptr; else | if (!utf8) c = *eptr; |
| 4458 | { | else { GETCHARLEN(c, eptr, len); } |
| GETCHARLEN(c, eptr, len); | ||
| } | ||
| 4459 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4460 | if (prop_category != ucp_M) break; | if (prop_category != ucp_M) break; |
| 4461 | eptr += len; | eptr += len; |
| # | Line 3505 for (;;) | Line 4474 for (;;) |
| 4474 | { | { |
| 4475 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM42); |
| 4476 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4477 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4478 | (ctype == OP_ANY && IS_NEWLINE(eptr))) | if (eptr >= md->end_subject) |
| 4479 | RRETURN(MATCH_NOMATCH); | { |
| 4480 | SCHECK_PARTIAL(); | |
| 4481 | MRRETURN(MATCH_NOMATCH); | |
| 4482 | } | |
| 4483 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | |
| 4484 | MRRETURN(MATCH_NOMATCH); | |
| 4485 | GETCHARINC(c, eptr); | GETCHARINC(c, eptr); |
| 4486 | switch(ctype) | switch(ctype) |
| 4487 | { | { |
| # | Line 3520 for (;;) | Line 4493 for (;;) |
| 4493 | case OP_ANYNL: | case OP_ANYNL: |
| 4494 | switch(c) | switch(c) |
| 4495 | { | { |
| 4496 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4497 | case 0x000d: | case 0x000d: |
| 4498 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4499 | break; | break; |
| # | Line 3532 for (;;) | Line 4505 for (;;) |
| 4505 | case 0x0085: | case 0x0085: |
| 4506 | case 0x2028: | case 0x2028: |
| 4507 | case 0x2029: | case 0x2029: |
| 4508 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4509 | break; | break; |
| 4510 | } | } |
| 4511 | break; | break; |
| # | Line 3560 for (;;) | Line 4533 for (;;) |
| 4533 | case 0x202f: /* NARROW NO-BREAK SPACE */ | case 0x202f: /* NARROW NO-BREAK SPACE */ |
| 4534 | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ | case 0x205f: /* MEDIUM MATHEMATICAL SPACE */ |
| 4535 | case 0x3000: /* IDEOGRAPHIC SPACE */ | case 0x3000: /* IDEOGRAPHIC SPACE */ |
| 4536 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4537 | } | } |
| 4538 | break; | break; |
| 4539 | ||
| 4540 | case OP_HSPACE: | case OP_HSPACE: |
| 4541 | switch(c) | switch(c) |
| 4542 | { | { |
| 4543 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4544 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4545 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4546 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3602 for (;;) | Line 4575 for (;;) |
| 4575 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4576 | case 0x2028: /* LINE SEPARATOR */ | case 0x2028: /* LINE SEPARATOR */ |
| 4577 | case 0x2029: /* PARAGRAPH SEPARATOR */ | case 0x2029: /* PARAGRAPH SEPARATOR */ |
| 4578 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4579 | } | } |
| 4580 | break; | break; |
| 4581 | ||
| 4582 | case OP_VSPACE: | case OP_VSPACE: |
| 4583 | switch(c) | switch(c) |
| 4584 | { | { |
| 4585 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4586 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4587 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4588 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3623 for (;;) | Line 4596 for (;;) |
| 4596 | ||
| 4597 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4598 | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) | if (c < 256 && (md->ctypes[c] & ctype_digit) != 0) |
| 4599 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4600 | break; | break; |
| 4601 | ||
| 4602 | case OP_DIGIT: | case OP_DIGIT: |
| 4603 | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_digit) == 0) |
| 4604 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4605 | break; | break; |
| 4606 | ||
| 4607 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4608 | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) | if (c < 256 && (md->ctypes[c] & ctype_space) != 0) |
| 4609 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4610 | break; | break; |
| 4611 | ||
| 4612 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4613 | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_space) == 0) |
| 4614 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4615 | break; | break; |
| 4616 | ||
| 4617 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4618 | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) | if (c < 256 && (md->ctypes[c] & ctype_word) != 0) |
| 4619 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4620 | break; | break; |
| 4621 | ||
| 4622 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4623 | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) | if (c >= 256 || (md->ctypes[c] & ctype_word) == 0) |
| 4624 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4625 | break; | break; |
| 4626 | ||
| 4627 | default: | default: |
| # | Line 3664 for (;;) | Line 4637 for (;;) |
| 4637 | { | { |
| 4638 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM43); |
| 4639 | if (rrc != MATCH_NOMATCH) RRETURN(rrc); | if (rrc != MATCH_NOMATCH) RRETURN(rrc); |
| 4640 | if (fi >= max || eptr >= md->end_subject || | if (fi >= max) MRRETURN(MATCH_NOMATCH); |
| 4641 | (ctype == OP_ANY && IS_NEWLINE(eptr))) | if (eptr >= md->end_subject) |
| 4642 | RRETURN(MATCH_NOMATCH); | { |
| 4643 | SCHECK_PARTIAL(); | |
| 4644 | MRRETURN(MATCH_NOMATCH); | |
| 4645 | } | |
| 4646 | if (ctype == OP_ANY && IS_NEWLINE(eptr)) | |
| 4647 | MRRETURN(MATCH_NOMATCH); | |
| 4648 | c = *eptr++; | c = *eptr++; |
| 4649 | switch(ctype) | switch(ctype) |
| 4650 | { | { |
| # | Line 3679 for (;;) | Line 4656 for (;;) |
| 4656 | case OP_ANYNL: | case OP_ANYNL: |
| 4657 | switch(c) | switch(c) |
| 4658 | { | { |
| 4659 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4660 | case 0x000d: | case 0x000d: |
| 4661 | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; | if (eptr < md->end_subject && *eptr == 0x0a) eptr++; |
| 4662 | break; | break; |
| # | Line 3690 for (;;) | Line 4667 for (;;) |
| 4667 | case 0x000b: | case 0x000b: |
| 4668 | case 0x000c: | case 0x000c: |
| 4669 | case 0x0085: | case 0x0085: |
| 4670 | if (md->bsr_anycrlf) RRETURN(MATCH_NOMATCH); | if (md->bsr_anycrlf) MRRETURN(MATCH_NOMATCH); |
| 4671 | break; | break; |
| 4672 | } | } |
| 4673 | break; | break; |
| # | Line 3702 for (;;) | Line 4679 for (;;) |
| 4679 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4680 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4681 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| 4682 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4683 | } | } |
| 4684 | break; | break; |
| 4685 | ||
| 4686 | case OP_HSPACE: | case OP_HSPACE: |
| 4687 | switch(c) | switch(c) |
| 4688 | { | { |
| 4689 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4690 | case 0x09: /* HT */ | case 0x09: /* HT */ |
| 4691 | case 0x20: /* SPACE */ | case 0x20: /* SPACE */ |
| 4692 | case 0xa0: /* NBSP */ | case 0xa0: /* NBSP */ |
| # | Line 3726 for (;;) | Line 4703 for (;;) |
| 4703 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| 4704 | case 0x0d: /* CR */ | case 0x0d: /* CR */ |
| 4705 | case 0x85: /* NEL */ | case 0x85: /* NEL */ |
| 4706 | RRETURN(MATCH_NOMATCH); | MRRETURN(MATCH_NOMATCH); |
| 4707 | } | } |
| 4708 | break; | break; |
| 4709 | ||
| 4710 | case OP_VSPACE: | case OP_VSPACE: |
| 4711 | switch(c) | switch(c) |
| 4712 | { | { |
| 4713 | default: RRETURN(MATCH_NOMATCH); | default: MRRETURN(MATCH_NOMATCH); |
| 4714 | case 0x0a: /* LF */ | case 0x0a: /* LF */ |
| 4715 | case 0x0b: /* VT */ | case 0x0b: /* VT */ |
| 4716 | case 0x0c: /* FF */ | case 0x0c: /* FF */ |
| # | Line 3744 for (;;) | Line 4721 for (;;) |
| 4721 | break; | break; |
| 4722 | ||
| 4723 | case OP_NOT_DIGIT: | case OP_NOT_DIGIT: |
| 4724 | if ((md->ctypes[c] & ctype_digit) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_digit) != 0) MRRETURN(MATCH_NOMATCH); |
| 4725 | break; | break; |
| 4726 | ||
| 4727 | case OP_DIGIT: | case OP_DIGIT: |
| 4728 | if ((md->ctypes[c] & ctype_digit) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_digit) == 0) MRRETURN(MATCH_NOMATCH); |
| 4729 | break; | break; |
| 4730 | ||
| 4731 | case OP_NOT_WHITESPACE: | case OP_NOT_WHITESPACE: |
| 4732 | if ((md->ctypes[c] & ctype_space) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_space) != 0) MRRETURN(MATCH_NOMATCH); |
| 4733 | break; | break; |
| 4734 | ||
| 4735 | case OP_WHITESPACE: | case OP_WHITESPACE: |
| 4736 | if ((md->ctypes[c] & ctype_space) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_space) == 0) MRRETURN(MATCH_NOMATCH); |
| 4737 | break; | break; |
| 4738 | ||
| 4739 | case OP_NOT_WORDCHAR: | case OP_NOT_WORDCHAR: |
| 4740 | if ((md->ctypes[c] & ctype_word) != 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_word) != 0) MRRETURN(MATCH_NOMATCH); |
| 4741 | break; | break; |
| 4742 | ||
| 4743 | case OP_WORDCHAR: | case OP_WORDCHAR: |
| 4744 | if ((md->ctypes[c] & ctype_word) == 0) RRETURN(MATCH_NOMATCH); | if ((md->ctypes[c] & ctype_word) == 0) MRRETURN(MATCH_NOMATCH); |
| 4745 | break; | break; |
| 4746 | ||
| 4747 | default: | default: |
| # | Line 3792 for (;;) | Line 4769 for (;;) |
| 4769 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4770 | { | { |
| 4771 | int len = 1; | int len = 1; |
| 4772 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4773 | GETCHARLEN(c, eptr, len); | { |
| 4774 | SCHECK_PARTIAL(); | |
| 4775 | break; | |
| 4776 | } | |
| 4777 | GETCHARLENTEST(c, eptr, len); | |
| 4778 | if (prop_fail_result) break; | if (prop_fail_result) break; |
| 4779 | eptr+= len; | eptr+= len; |
| 4780 | } | } |
| # | Line 3803 for (;;) | Line 4784 for (;;) |
| 4784 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4785 | { | { |
| 4786 | int len = 1; | int len = 1; |
| 4787 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4788 | GETCHARLEN(c, eptr, len); | { |
| 4789 | SCHECK_PARTIAL(); | |
| 4790 | break; | |
| 4791 | } | |
| 4792 | GETCHARLENTEST(c, eptr, len); | |
| 4793 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4794 | if ((prop_chartype == ucp_Lu || | if ((prop_chartype == ucp_Lu || |
| 4795 | prop_chartype == ucp_Ll || | prop_chartype == ucp_Ll || |
| # | Line 3818 for (;;) | Line 4803 for (;;) |
| 4803 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4804 | { | { |
| 4805 | int len = 1; | int len = 1; |
| 4806 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4807 | GETCHARLEN(c, eptr, len); | { |
| 4808 | SCHECK_PARTIAL(); | |
| 4809 | break; | |
| 4810 | } | |
| 4811 | GETCHARLENTEST(c, eptr, len); | |
| 4812 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4813 | if ((prop_category == prop_value) == prop_fail_result) | if ((prop_category == prop_value) == prop_fail_result) |
| 4814 | break; | break; |
| # | Line 3831 for (;;) | Line 4820 for (;;) |
| 4820 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4821 | { | { |
| 4822 | int len = 1; | int len = 1; |
| 4823 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4824 | GETCHARLEN(c, eptr, len); | { |
| 4825 | SCHECK_PARTIAL(); | |
| 4826 | break; | |
| 4827 | } | |
| 4828 | GETCHARLENTEST(c, eptr, len); | |
| 4829 | prop_chartype = UCD_CHARTYPE(c); | prop_chartype = UCD_CHARTYPE(c); |
| 4830 | if ((prop_chartype == prop_value) == prop_fail_result) | if ((prop_chartype == prop_value) == prop_fail_result) |
| 4831 | break; | break; |
| # | Line 3844 for (;;) | Line 4837 for (;;) |
| 4837 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4838 | { | { |
| 4839 | int len = 1; | int len = 1; |
| 4840 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4841 | GETCHARLEN(c, eptr, len); | { |
| 4842 | SCHECK_PARTIAL(); | |
| 4843 | break; | |
| 4844 | } | |
| 4845 | GETCHARLENTEST(c, eptr, len); | |
| 4846 | prop_script = UCD_SCRIPT(c); | prop_script = UCD_SCRIPT(c); |
| 4847 | if ((prop_script == prop_value) == prop_fail_result) | if ((prop_script == prop_value) == prop_fail_result) |
| 4848 | break; | break; |
| 4849 | eptr+= len; | eptr+= len; |
| 4850 | } | } |
| 4851 | break; | break; |
| 4852 | ||
| 4853 | case PT_ALNUM: | |
| 4854 | for (i = min; i < max; i++) | |
| 4855 | { | |
| 4856 | int len = 1; | |
| 4857 | if (eptr >= md->end_subject) | |
| 4858 | { | |
| 4859 | SCHECK_PARTIAL(); | |
| 4860 | break; | |
| 4861 | } | |
| 4862 | GETCHARLENTEST(c, eptr, len); | |
| 4863 | prop_category = UCD_CATEGORY(c); | |
| 4864 | if ((prop_category == ucp_L || prop_category == ucp_N) | |
| 4865 | == prop_fail_result) | |
| 4866 | break; | |
| 4867 | eptr+= len; | |
| 4868 | } | |
| 4869 | break; | |
| 4870 | ||
| 4871 | case PT_SPACE: /* Perl space */ | |
| 4872 | for (i = min; i < max; i++) | |
| 4873 | { | |
| 4874 | int len = 1; | |
| 4875 | if (eptr >= md->end_subject) | |
| 4876 | { | |
| 4877 | SCHECK_PARTIAL(); | |
| 4878 | break; | |
| 4879 | } | |
| 4880 | GETCHARLENTEST(c, eptr, len); | |
| 4881 | prop_category = UCD_CATEGORY(c); | |
| 4882 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 4883 | c == CHAR_FF || c == CHAR_CR) | |
| 4884 | == prop_fail_result) | |
| 4885 | break; | |
| 4886 | eptr+= len; | |
| 4887 | } | |
| 4888 | break; | |
| 4889 | ||
| 4890 | case PT_PXSPACE: /* POSIX space */ | |
| 4891 | for (i = min; i < max; i++) | |
| 4892 | { | |
| 4893 | int len = 1; | |
| 4894 | if (eptr >= md->end_subject) | |
| 4895 | { | |
| 4896 | SCHECK_PARTIAL(); | |
| 4897 | break; | |
| 4898 | } | |
| 4899 | GETCHARLENTEST(c, eptr, len); | |
| 4900 | prop_category = UCD_CATEGORY(c); | |
| 4901 | if ((prop_category == ucp_Z || c == CHAR_HT || c == CHAR_NL || | |
| 4902 | c == CHAR_VT || c == CHAR_FF || c == CHAR_CR) | |
| 4903 | == prop_fail_result) | |
| 4904 | break; | |
| 4905 | eptr+= len; | |
| 4906 | } | |
| 4907 | break; | |
| 4908 | ||
| 4909 | case PT_WORD: | |
| 4910 | for (i = min; i < max; i++) | |
| 4911 | { | |
| 4912 | int len = 1; | |
| 4913 | if (eptr >= md->end_subject) | |
| 4914 | { | |
| 4915 | SCHECK_PARTIAL(); | |
| 4916 | break; | |
| 4917 | } | |
| 4918 | GETCHARLENTEST(c, eptr, len); | |
| 4919 | prop_category = UCD_CATEGORY(c); | |
| 4920 | if ((prop_category == ucp_L || prop_category == ucp_N || | |
| 4921 | c == CHAR_UNDERSCORE) == prop_fail_result) | |
| 4922 | break; | |
| 4923 | eptr+= len; | |
| 4924 | } | |
| 4925 | break; | |
| 4926 | ||
| 4927 | default: | |
| 4928 | RRETURN(PCRE_ERROR_INTERNAL); | |
| 4929 | } | } |
| 4930 | ||
| 4931 | /* eptr is now past the end of the maximum run */ | /* eptr is now past the end of the maximum run */ |
| # | Line 3873 for (;;) | Line 4947 for (;;) |
| 4947 | { | { |
| 4948 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 4949 | { | { |
| 4950 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 4951 | { | |
| 4952 | SCHECK_PARTIAL(); | |
| 4953 | break; | |
| 4954 | } | |
| 4955 | GETCHARINCTEST(c, eptr); | GETCHARINCTEST(c, eptr); |
| 4956 | prop_category = UCD_CATEGORY(c); | prop_category = UCD_CATEGORY(c); |
| 4957 | if (prop_category == ucp_M) break; | if (prop_category == ucp_M) break; |
| # | Line 3893 for (;;) | Line 4971 for (;;) |
| 4971 | /* eptr is now past the end of the maximum run */ | /* eptr is now past the end of the maximum run */ |
| 4972 | ||
| 4973 | if (possessive) continue; | if (possessive) continue; |
| 4974 | ||
| 4975 | for(;;) | for(;;) |
| 4976 | { | { |
| 4977 | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45); | RMATCH(eptr, ecode, offset_top, md, ims, eptrb, 0, RM45); |
| # | Line 3928 for (;;) | Line 5007 for (;;) |
| 5007 | { | { |
| 5008 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5009 | { | { |
| 5010 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break; | if (eptr >= md->end_subject) |
| 5011 | { | |
| 5012 | SCHECK_PARTIAL(); | |
| 5013 | break; | |
| 5014 | } | |
| 5015 | if (IS_NEWLINE(eptr)) break; | |
| 5016 | eptr++; | eptr++; |
| 5017 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 5018 | } | } |
| # | Line 3940 for (;;) | Line 5024 for (;;) |
| 5024 | { | { |
| 5025 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5026 | { | { |
| 5027 | if (eptr >= md->end_subject || IS_NEWLINE(eptr)) break; | if (eptr >= md->end_subject) |
| 5028 | { | |
| 5029 | SCHECK_PARTIAL(); | |
| 5030 | break; | |
| 5031 | } | |
| 5032 | if (IS_NEWLINE(eptr)) break; | |
| 5033 | eptr++; | eptr++; |
| 5034 | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; | while (eptr < md->end_subject && (*eptr & 0xc0) == 0x80) eptr++; |
| 5035 | } | } |
| # | Line 3952 for (;;) | Line 5041 for (;;) |
| 5041 | { | { |
| 5042 | for (i = min; i < max; i++) | for (i = min; i < max; i++) |
| 5043 | { | { |
| 5044 | if (eptr >= md->end_subject) break; | if (eptr >= md->end_subject) |
| 5045 | { | |
| 5046 | SCHECK_PARTIAL(); | |
| 5047 | break; | |
| 5048 | } | |
| 5049 | eptr++; | eptr++; |