/[pcre]/code/trunk/pcre_study.c
ViewVC logotype

Diff of /code/trunk/pcre_study.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 539 by ph10, Sun Jun 13 21:35:04 2010 UTC revision 603 by ph10, Fri May 27 10:14:09 2011 UTC
# Line 73  Arguments: Line 73  Arguments:
73  Returns:   the minimum length  Returns:   the minimum length
74             -1 if \C was encountered             -1 if \C was encountered
75             -2 internal error (missing capturing bracket)             -2 internal error (missing capturing bracket)
76               -3 internal error (opcode not listed)
77  */  */
78    
79  static int  static int
# Line 160  for (;;) Line 161  for (;;)
161      case OP_RREF:      case OP_RREF:
162      case OP_NRREF:      case OP_NRREF:
163      case OP_DEF:      case OP_DEF:
     case OP_OPT:  
164      case OP_CALLOUT:      case OP_CALLOUT:
165      case OP_SOD:      case OP_SOD:
166      case OP_SOM:      case OP_SOM:
167      case OP_EOD:      case OP_EOD:
168      case OP_EODN:      case OP_EODN:
169      case OP_CIRC:      case OP_CIRC:
170        case OP_CIRCM:
171      case OP_DOLL:      case OP_DOLL:
172        case OP_DOLLM:
173      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
174      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
175      cc += _pcre_OP_lengths[*cc];      cc += _pcre_OP_lengths[*cc];
# Line 186  for (;;) Line 188  for (;;)
188      /* Handle literal characters and + repetitions */      /* Handle literal characters and + repetitions */
189    
190      case OP_CHAR:      case OP_CHAR:
191      case OP_CHARNC:      case OP_CHARI:
192      case OP_NOT:      case OP_NOT:
193        case OP_NOTI:
194      case OP_PLUS:      case OP_PLUS:
195        case OP_PLUSI:
196      case OP_MINPLUS:      case OP_MINPLUS:
197        case OP_MINPLUSI:
198      case OP_POSPLUS:      case OP_POSPLUS:
199        case OP_POSPLUSI:
200      case OP_NOTPLUS:      case OP_NOTPLUS:
201        case OP_NOTPLUSI:
202      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
203        case OP_NOTMINPLUSI:
204      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
205        case OP_NOTPOSPLUSI:
206      branchlength++;      branchlength++;
207      cc += 2;      cc += 2;
208  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 212  for (;;) Line 221  for (;;)
221      need to skip over a multibyte character in UTF8 mode.  */      need to skip over a multibyte character in UTF8 mode.  */
222    
223      case OP_EXACT:      case OP_EXACT:
224        case OP_EXACTI:
225      case OP_NOTEXACT:      case OP_NOTEXACT:
226        case OP_NOTEXACTI:
227      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
228      cc += 4;      cc += 4;
229  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 337  for (;;) Line 348  for (;;)
348      that case we must set the minimum length to zero. */      that case we must set the minimum length to zero. */
349    
350      case OP_REF:      case OP_REF:
351        case OP_REFI:
352      if ((options & PCRE_JAVASCRIPT_COMPAT) == 0)      if ((options & PCRE_JAVASCRIPT_COMPAT) == 0)
353        {        {
354        ce = cs = (uschar *)_pcre_find_bracket(startcode, utf8, GET2(cc, 1));        ce = cs = (uschar *)_pcre_find_bracket(startcode, utf8, GET2(cc, 1));
# Line 391  for (;;) Line 403  for (;;)
403    
404      /* Anything else does not or need not match a character. We can get the      /* Anything else does not or need not match a character. We can get the
405      item's length from the table, but for those that can match zero occurrences      item's length from the table, but for those that can match zero occurrences
406      of a character, we must take special action for UTF-8 characters. */      of a character, we must take special action for UTF-8 characters. As it
407        happens, the "NOT" versions of these opcodes are used at present only for
408        ASCII characters, so they could be omitted from this list. However, in
409        future that may change, so we include them here so as not to leave a
410        gotcha for a future maintainer. */
411    
412      case OP_UPTO:      case OP_UPTO:
413        case OP_UPTOI:
414      case OP_NOTUPTO:      case OP_NOTUPTO:
415        case OP_NOTUPTOI:
416      case OP_MINUPTO:      case OP_MINUPTO:
417        case OP_MINUPTOI:
418      case OP_NOTMINUPTO:      case OP_NOTMINUPTO:
419        case OP_NOTMINUPTOI:
420      case OP_POSUPTO:      case OP_POSUPTO:
421        case OP_POSUPTOI:
422        case OP_NOTPOSUPTO:
423        case OP_NOTPOSUPTOI:
424    
425      case OP_STAR:      case OP_STAR:
426        case OP_STARI:
427        case OP_NOTSTAR:
428        case OP_NOTSTARI:
429      case OP_MINSTAR:      case OP_MINSTAR:
430        case OP_MINSTARI:
431      case OP_NOTMINSTAR:      case OP_NOTMINSTAR:
432        case OP_NOTMINSTARI:
433      case OP_POSSTAR:      case OP_POSSTAR:
434        case OP_POSSTARI:
435      case OP_NOTPOSSTAR:      case OP_NOTPOSSTAR:
436        case OP_NOTPOSSTARI:
437    
438      case OP_QUERY:      case OP_QUERY:
439        case OP_QUERYI:
440        case OP_NOTQUERY:
441        case OP_NOTQUERYI:
442      case OP_MINQUERY:      case OP_MINQUERY:
443        case OP_MINQUERYI:
444      case OP_NOTMINQUERY:      case OP_NOTMINQUERY:
445        case OP_NOTMINQUERYI:
446      case OP_POSQUERY:      case OP_POSQUERY:
447        case OP_POSQUERYI:
448      case OP_NOTPOSQUERY:      case OP_NOTPOSQUERY:
449        case OP_NOTPOSQUERYI:
450    
451      cc += _pcre_OP_lengths[op];      cc += _pcre_OP_lengths[op];
452  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
453      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];
# Line 419  for (;;) Line 459  for (;;)
459      case OP_MARK:      case OP_MARK:
460      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
461      case OP_SKIP_ARG:      case OP_SKIP_ARG:
     case OP_THEN_ARG:  
462      cc += _pcre_OP_lengths[op] + cc[1];      cc += _pcre_OP_lengths[op] + cc[1];
463      break;      break;
464    
465      /* For the record, these are the opcodes that are matched by "default":      case OP_THEN_ARG:
466      OP_ACCEPT, OP_CLOSE, OP_COMMIT, OP_FAIL, OP_PRUNE, OP_SET_SOM, OP_SKIP,      cc += _pcre_OP_lengths[op] + cc[1+LINK_SIZE];
467      OP_THEN. */      break;
468    
469        /* The remaining opcodes are just skipped over. */
470    
471      default:      case OP_ACCEPT:
472        case OP_CLOSE:
473        case OP_COMMIT:
474        case OP_FAIL:
475        case OP_PRUNE:
476        case OP_SET_SOM:
477        case OP_SKIP:
478        case OP_THEN:
479      cc += _pcre_OP_lengths[op];      cc += _pcre_OP_lengths[op];
480      break;      break;
481    
482        /* This should not occur: we list all opcodes explicitly so that when
483        new ones get added they are properly considered. */
484    
485        default:
486        return -3;
487      }      }
488    }    }
489  /* Control never gets here */  /* Control never gets here */
# Line 504  Arguments: Line 558  Arguments:
558    start_bits     the starting bitmap    start_bits     the starting bitmap
559    cbit type      the type of character wanted    cbit type      the type of character wanted
560    table_limit    32 for non-UTF-8; 16 for UTF-8    table_limit    32 for non-UTF-8; 16 for UTF-8
561    cd             the block with char table pointers    cd             the block with char table pointers
562    
563  Returns:         nothing  Returns:         nothing
564  */  */
# Line 513  static void Line 567  static void
567  set_type_bits(uschar *start_bits, int cbit_type, int table_limit,  set_type_bits(uschar *start_bits, int cbit_type, int table_limit,
568    compile_data *cd)    compile_data *cd)
569  {  {
570  register int c;  register int c;
571  for (c = 0; c < table_limit; c++) start_bits[c] |= cd->cbits[c+cbit_type];  for (c = 0; c < table_limit; c++) start_bits[c] |= cd->cbits[c+cbit_type];
572  if (table_limit == 32) return;  if (table_limit == 32) return;
573  for (c = 128; c < 256; c++)  for (c = 128; c < 256; c++)
# Line 522  for (c = 128; c < 256; c++) Line 576  for (c = 128; c < 256; c++)
576      {      {
577      uschar buff[8];      uschar buff[8];
578      (void)_pcre_ord2utf8(c, buff);      (void)_pcre_ord2utf8(c, buff);
579      SET_BIT(buff[0]);      SET_BIT(buff[0]);
580      }      }
581    }    }
582  }  }
583    
584    
# Line 535  for (c = 128; c < 256; c++) Line 589  for (c = 128; c < 256; c++)
589  /* This function sets starting bits for a negative character type such as \D.  /* This function sets starting bits for a negative character type such as \D.
590  In UTF-8 mode, we can only do a direct setting for bytes less than 128, as  In UTF-8 mode, we can only do a direct setting for bytes less than 128, as
591  otherwise there can be confusion with bytes in the middle of UTF-8 characters.  otherwise there can be confusion with bytes in the middle of UTF-8 characters.
592  Unlike in the positive case, where we can set appropriate starting bits for  Unlike in the positive case, where we can set appropriate starting bits for
593  specific high-valued UTF-8 characters, in this case we have to set the bits for  specific high-valued UTF-8 characters, in this case we have to set the bits for
594  all high-valued characters. The lowest is 0xc2, but we overkill by starting at  all high-valued characters. The lowest is 0xc2, but we overkill by starting at
595  0xc0 (192) for simplicity.  0xc0 (192) for simplicity.
596    
597  Arguments:  Arguments:
598    start_bits     the starting bitmap    start_bits     the starting bitmap
599    cbit type      the type of character wanted    cbit type      the type of character wanted
600    table_limit    32 for non-UTF-8; 16 for UTF-8    table_limit    32 for non-UTF-8; 16 for UTF-8
601    cd             the block with char table pointers    cd             the block with char table pointers
602    
603  Returns:         nothing  Returns:         nothing
604  */  */
# Line 553  static void Line 607  static void
607  set_nottype_bits(uschar *start_bits, int cbit_type, int table_limit,  set_nottype_bits(uschar *start_bits, int cbit_type, int table_limit,
608    compile_data *cd)    compile_data *cd)
609  {  {
610  register int c;  register int c;
611  for (c = 0; c < table_limit; c++) start_bits[c] |= ~cd->cbits[c+cbit_type];  for (c = 0; c < table_limit; c++) start_bits[c] |= ~cd->cbits[c+cbit_type];
612  if (table_limit != 32) for (c = 24; c < 32; c++) start_bits[c] = 0xff;  if (table_limit != 32) for (c = 24; c < 32; c++) start_bits[c] = 0xff;
613  }  }
# Line 575  function fails unless the result is SSB_ Line 629  function fails unless the result is SSB_
629  Arguments:  Arguments:
630    code         points to an expression    code         points to an expression
631    start_bits   points to a 32-byte table, initialized to 0    start_bits   points to a 32-byte table, initialized to 0
   caseless     the current state of the caseless flag  
632    utf8         TRUE if in UTF-8 mode    utf8         TRUE if in UTF-8 mode
633    cd           the block with char table pointers    cd           the block with char table pointers
634    
# Line 585  Returns: SSB_FAIL => Failed to Line 638  Returns: SSB_FAIL => Failed to
638  */  */
639    
640  static int  static int
641  set_start_bits(const uschar *code, uschar *start_bits, BOOL caseless,  set_start_bits(const uschar *code, uschar *start_bits, BOOL utf8,
642    BOOL utf8, compile_data *cd)    compile_data *cd)
643  {  {
644  register int c;  register int c;
645  int yield = SSB_DONE;  int yield = SSB_DONE;
# Line 635  do Line 688  do
688        case OP_SCBRA:        case OP_SCBRA:
689        case OP_ONCE:        case OP_ONCE:
690        case OP_ASSERT:        case OP_ASSERT:
691        rc = set_start_bits(tcode, start_bits, caseless, utf8, cd);        rc = set_start_bits(tcode, start_bits, utf8, cd);
692        if (rc == SSB_FAIL) return SSB_FAIL;        if (rc == SSB_FAIL) return SSB_FAIL;
693        if (rc == SSB_DONE) try_next = FALSE; else        if (rc == SSB_DONE) try_next = FALSE; else
694          {          {
# Line 676  do Line 729  do
729        tcode += 1 + LINK_SIZE;        tcode += 1 + LINK_SIZE;
730        break;        break;
731    
       /* Skip over an option setting, changing the caseless flag */  
   
       case OP_OPT:  
       caseless = (tcode[1] & PCRE_CASELESS) != 0;  
       tcode += 2;  
       break;  
   
732        /* BRAZERO does the bracket, but carries on. */        /* BRAZERO does the bracket, but carries on. */
733    
734        case OP_BRAZERO:        case OP_BRAZERO:
735        case OP_BRAMINZERO:        case OP_BRAMINZERO:
736        if (set_start_bits(++tcode, start_bits, caseless, utf8, cd) == SSB_FAIL)        if (set_start_bits(++tcode, start_bits, utf8, cd) == SSB_FAIL)
737          return SSB_FAIL;          return SSB_FAIL;
738  /* =========================================================================  /* =========================================================================
739        See the comment at the head of this function concerning the next line,        See the comment at the head of this function concerning the next line,
# Line 714  do Line 760  do
760        case OP_QUERY:        case OP_QUERY:
761        case OP_MINQUERY:        case OP_MINQUERY:
762        case OP_POSQUERY:        case OP_POSQUERY:
763        tcode = set_table_bit(start_bits, tcode + 1, caseless, cd, utf8);        tcode = set_table_bit(start_bits, tcode + 1, FALSE, cd, utf8);
764          break;
765    
766          case OP_STARI:
767          case OP_MINSTARI:
768          case OP_POSSTARI:
769          case OP_QUERYI:
770          case OP_MINQUERYI:
771          case OP_POSQUERYI:
772          tcode = set_table_bit(start_bits, tcode + 1, TRUE, cd, utf8);
773        break;        break;
774    
775        /* Single-char upto sets the bit and tries the next */        /* Single-char upto sets the bit and tries the next */
# Line 722  do Line 777  do
777        case OP_UPTO:        case OP_UPTO:
778        case OP_MINUPTO:        case OP_MINUPTO:
779        case OP_POSUPTO:        case OP_POSUPTO:
780        tcode = set_table_bit(start_bits, tcode + 3, caseless, cd, utf8);        tcode = set_table_bit(start_bits, tcode + 3, FALSE, cd, utf8);
781          break;
782    
783          case OP_UPTOI:
784          case OP_MINUPTOI:
785          case OP_POSUPTOI:
786          tcode = set_table_bit(start_bits, tcode + 3, TRUE, cd, utf8);
787        break;        break;
788    
789        /* At least one single char sets the bit and stops */        /* At least one single char sets the bit and stops */
790    
791        case OP_EXACT:       /* Fall through */        case OP_EXACT:
792        tcode += 2;        tcode += 2;
793          /* Fall through */
794        case OP_CHAR:        case OP_CHAR:
       case OP_CHARNC:  
795        case OP_PLUS:        case OP_PLUS:
796        case OP_MINPLUS:        case OP_MINPLUS:
797        case OP_POSPLUS:        case OP_POSPLUS:
798        (void)set_table_bit(start_bits, tcode + 1, caseless, cd, utf8);        (void)set_table_bit(start_bits, tcode + 1, FALSE, cd, utf8);
799          try_next = FALSE;
800          break;
801    
802          case OP_EXACTI:
803          tcode += 2;
804          /* Fall through */
805          case OP_CHARI:
806          case OP_PLUSI:
807          case OP_MINPLUSI:
808          case OP_POSPLUSI:
809          (void)set_table_bit(start_bits, tcode + 1, TRUE, cd, utf8);
810        try_next = FALSE;        try_next = FALSE;
811        break;        break;
812    
# Line 750  do Line 821  do
821        SET_BIT(0x20);        SET_BIT(0x20);
822        if (utf8)        if (utf8)
823          {          {
824          SET_BIT(0xC2);  /* For U+00A0 */          SET_BIT(0xC2);  /* For U+00A0 */
825          SET_BIT(0xE1);  /* For U+1680, U+180E */          SET_BIT(0xE1);  /* For U+1680, U+180E */
826          SET_BIT(0xE2);  /* For U+2000 - U+200A, U+202F, U+205F */          SET_BIT(0xE2);  /* For U+2000 - U+200A, U+202F, U+205F */
827          SET_BIT(0xE3);  /* For U+3000 */          SET_BIT(0xE3);  /* For U+3000 */
# Line 765  do Line 836  do
836        SET_BIT(0x0B);        SET_BIT(0x0B);
837        SET_BIT(0x0C);        SET_BIT(0x0C);
838        SET_BIT(0x0D);        SET_BIT(0x0D);
839        if (utf8)        if (utf8)
840          {          {
841          SET_BIT(0xC2);  /* For U+0085 */          SET_BIT(0xC2);  /* For U+0085 */
842          SET_BIT(0xE2);  /* For U+2028, U+2029 */          SET_BIT(0xE2);  /* For U+2028, U+2029 */
843          }          }
844        else SET_BIT(0x85);        else SET_BIT(0x85);
845        try_next = FALSE;        try_next = FALSE;
846        break;        break;
847    
848        /* Single character types set the bits and stop. Note that if PCRE_UCP        /* Single character types set the bits and stop. Note that if PCRE_UCP
849        is set, we do not see these op codes because \d etc are converted to        is set, we do not see these op codes because \d etc are converted to
850        properties. Therefore, these apply in the case when only characters less        properties. Therefore, these apply in the case when only characters less
851        than 256 are recognized to match the types. */        than 256 are recognized to match the types. */
852    
853        case OP_NOT_DIGIT:        case OP_NOT_DIGIT:
# Line 817  do Line 888  do
888        set_type_bits(start_bits, cbit_word, table_limit, cd);        set_type_bits(start_bits, cbit_word, table_limit, cd);
889        try_next = FALSE;        try_next = FALSE;
890        break;        break;
891    
892        /* One or more character type fudges the pointer and restarts, knowing        /* One or more character type fudges the pointer and restarts, knowing
893        it will hit a single character type and stop there. */        it will hit a single character type and stop there. */
894    
# Line 857  do Line 928  do
928          SET_BIT(0x20);          SET_BIT(0x20);
929          if (utf8)          if (utf8)
930            {            {
931            SET_BIT(0xC2);  /* For U+00A0 */            SET_BIT(0xC2);  /* For U+00A0 */
932            SET_BIT(0xE1);  /* For U+1680, U+180E */            SET_BIT(0xE1);  /* For U+1680, U+180E */
933            SET_BIT(0xE2);  /* For U+2000 - U+200A, U+202F, U+205F */            SET_BIT(0xE2);  /* For U+2000 - U+200A, U+202F, U+205F */
934            SET_BIT(0xE3);  /* For U+3000 */            SET_BIT(0xE3);  /* For U+3000 */
# Line 871  do Line 942  do
942          SET_BIT(0x0B);          SET_BIT(0x0B);
943          SET_BIT(0x0C);          SET_BIT(0x0C);
944          SET_BIT(0x0D);          SET_BIT(0x0D);
945          if (utf8)          if (utf8)
946            {            {
947            SET_BIT(0xC2);  /* For U+0085 */            SET_BIT(0xC2);  /* For U+0085 */
948            SET_BIT(0xE2);  /* For U+2028, U+2029 */            SET_BIT(0xE2);  /* For U+2028, U+2029 */
949            }            }
950          else SET_BIT(0x85);          else SET_BIT(0x85);
951          break;          break;
952    
# Line 892  do Line 963  do
963    
964          case OP_NOT_WHITESPACE:          case OP_NOT_WHITESPACE:
965          set_nottype_bits(start_bits, cbit_space, table_limit, cd);          set_nottype_bits(start_bits, cbit_space, table_limit, cd);
966          start_bits[1] |= 0x08;          start_bits[1] |= 0x08;
967          break;          break;
968    
969          /* The cbit_space table has vertical tab as whitespace; we have to          /* The cbit_space table has vertical tab as whitespace; we have to
# Line 1072  if ((re->options & PCRE_ANCHORED) == 0 & Line 1143  if ((re->options & PCRE_ANCHORED) == 0 &
1143    /* See if we can find a fixed set of initial characters for the pattern. */    /* See if we can find a fixed set of initial characters for the pattern. */
1144    
1145    memset(start_bits, 0, 32 * sizeof(uschar));    memset(start_bits, 0, 32 * sizeof(uschar));
1146    bits_set = set_start_bits(code, start_bits,    bits_set = set_start_bits(code, start_bits, (re->options & PCRE_UTF8) != 0,
     (re->options & PCRE_CASELESS) != 0, (re->options & PCRE_UTF8) != 0,  
1147      &compile_block) == SSB_DONE;      &compile_block) == SSB_DONE;
1148    }    }
1149    
1150  /* Find the minimum length of subject string. */  /* Find the minimum length of subject string. */
1151    
1152  min = find_minlength(code, code, re->options);  switch(min = find_minlength(code, code, re->options))
1153      {
1154      case -2: *errorptr = "internal error: missing capturing bracket"; break;
1155      case -3: *errorptr = "internal error: opcode not recognized"; break;
1156      default: break;
1157      }
1158    
1159  /* Return NULL if no optimization is possible. */  /* Return NULL if no optimization is possible. */
1160    
Legend:
Removed from v.539  
changed lines
  Added in v.603
webmaster@exim.org
 ViewVC Help
Powered by ViewVC 1.1.12