/[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 475 by ph10, Sat Jan 2 18:21:30 2010 UTC revision 524 by ph10, Mon May 24 17:06:28 2010 UTC
# Line 48  supporting functions. */ Line 48  supporting functions. */
48    
49  #include "pcre_internal.h"  #include "pcre_internal.h"
50    
51    #define SET_BIT(c) start_bits[c/8] |= (1 << (c&7))
52    
53  /* Returns from set_start_bits() */  /* Returns from set_start_bits() */
54    
# Line 413  for (;;) Line 414  for (;;)
414  #endif  #endif
415      break;      break;
416    
417        /* Skip these, but we need to add in the name length. */
418    
419        case OP_MARK:
420        case OP_PRUNE_ARG:
421        case OP_SKIP_ARG:
422        case OP_THEN_ARG:
423        cc += _pcre_OP_lengths[op] + cc[1];
424        break;
425    
426      /* For the record, these are the opcodes that are matched by "default":      /* For the record, these are the opcodes that are matched by "default":
427      OP_ACCEPT, OP_CLOSE, OP_COMMIT, OP_FAIL, OP_PRUNE, OP_SET_SOM, OP_SKIP,      OP_ACCEPT, OP_CLOSE, OP_COMMIT, OP_FAIL, OP_PRUNE, OP_SET_SOM, OP_SKIP,
428      OP_THEN. */      OP_THEN. */
# Line 431  for (;;) Line 441  for (;;)
441  *      Set a bit and maybe its alternate case    *  *      Set a bit and maybe its alternate case    *
442  *************************************************/  *************************************************/
443    
444  /* Given a character, set its bit in the table, and also the bit for the other  /* Given a character, set its first byte's bit in the table, and also the
445  version of a letter if we are caseless.  corresponding bit for the other version of a letter if we are caseless. In
446    UTF-8 mode, for characters greater than 127, we can only do the caseless thing
447    when Unicode property support is available.
448    
449  Arguments:  Arguments:
450    start_bits    points to the bit map    start_bits    points to the bit map
451    c             is the character    p             points to the character
452    caseless      the caseless flag    caseless      the caseless flag
453    cd            the block with char table pointers    cd            the block with char table pointers
454      utf8          TRUE for UTF-8 mode
455    
456  Returns:        nothing  Returns:        pointer after the character
457  */  */
458    
459  static void  static const uschar *
460  set_table_bit(uschar *start_bits, unsigned int c, BOOL caseless,  set_table_bit(uschar *start_bits, const uschar *p, BOOL caseless,
461    compile_data *cd)    compile_data *cd, BOOL utf8)
462  {  {
463  start_bits[c/8] |= (1 << (c&7));  unsigned int c = *p;
464  if (caseless && (cd->ctypes[c] & ctype_letter) != 0)  
465    start_bits[cd->fcc[c]/8] |= (1 << (cd->fcc[c]&7));  SET_BIT(c);
466    
467    #ifdef SUPPORT_UTF8
468    if (utf8 && c > 127)
469      {
470      GETCHARINC(c, p);
471    #ifdef SUPPORT_UCP
472      if (caseless)
473        {
474        uschar buff[8];
475        c = UCD_OTHERCASE(c);
476        (void)_pcre_ord2utf8(c, buff);
477        SET_BIT(buff[0]);
478        }
479    #endif
480      return p;
481      }
482    #endif
483    
484    /* Not UTF-8 mode, or character is less than 127. */
485    
486    if (caseless && (cd->ctypes[c] & ctype_letter) != 0) SET_BIT(cd->fcc[c]);
487    return p + 1;
488  }  }
489    
490    
# Line 607  do Line 642  do
642        case OP_QUERY:        case OP_QUERY:
643        case OP_MINQUERY:        case OP_MINQUERY:
644        case OP_POSQUERY:        case OP_POSQUERY:
645        set_table_bit(start_bits, tcode[1], caseless, cd);        tcode = set_table_bit(start_bits, tcode + 1, caseless, cd, utf8);
       tcode += 2;  
 #ifdef SUPPORT_UTF8  
       if (utf8 && tcode[-1] >= 0xc0)  
         tcode += _pcre_utf8_table4[tcode[-1] & 0x3f];  
 #endif  
646        break;        break;
647    
648        /* Single-char upto sets the bit and tries the next */        /* Single-char upto sets the bit and tries the next */
# Line 620  do Line 650  do
650        case OP_UPTO:        case OP_UPTO:
651        case OP_MINUPTO:        case OP_MINUPTO:
652        case OP_POSUPTO:        case OP_POSUPTO:
653        set_table_bit(start_bits, tcode[3], caseless, cd);        tcode = set_table_bit(start_bits, tcode + 3, caseless, cd, utf8);
       tcode += 4;  
 #ifdef SUPPORT_UTF8  
       if (utf8 && tcode[-1] >= 0xc0)  
         tcode += _pcre_utf8_table4[tcode[-1] & 0x3f];  
 #endif  
654        break;        break;
655    
656        /* At least one single char sets the bit and stops */        /* At least one single char sets the bit and stops */
# Line 638  do Line 663  do
663        case OP_PLUS:        case OP_PLUS:
664        case OP_MINPLUS:        case OP_MINPLUS:
665        case OP_POSPLUS:        case OP_POSPLUS:
666        set_table_bit(start_bits, tcode[1], caseless, cd);        (void)set_table_bit(start_bits, tcode + 1, caseless, cd, utf8);
667        try_next = FALSE;        try_next = FALSE;
668        break;        break;
669    
670          /* Special spacing and line-terminating items. These recognize specific
671          lists of characters. The difference between VSPACE and ANYNL is that the
672          latter can match the two-character CRLF sequence, but that is not
673          relevant for finding the first character, so their code here is
674          identical. */
675    
676          case OP_HSPACE:
677          SET_BIT(0x09);
678          SET_BIT(0x20);
679          SET_BIT(0xA0);
680          if (utf8)
681            {
682            SET_BIT(0xE1);  /* For U+1680, U+180E */
683            SET_BIT(0xE2);  /* For U+2000 - U+200A, U+202F, U+205F */
684            SET_BIT(0xE3);  /* For U+3000 */
685            }
686          try_next = FALSE;
687          break;
688    
689          case OP_ANYNL:
690          case OP_VSPACE:
691          SET_BIT(0x0A);
692          SET_BIT(0x0B);
693          SET_BIT(0x0C);
694          SET_BIT(0x0D);
695          SET_BIT(0x85);
696          if (utf8) SET_BIT(0xE2);    /* For U+2028, U+2029 */
697          try_next = FALSE;
698          break;
699    
700        /* Single character type sets the bits and stops */        /* Single character types set the bits and stop. Note that if PCRE_UCP
701          is set, we do not see these op codes because \d etc are converted to
702          properties. Therefore, these apply in the case when only ASCII characters
703          are recognized to match the types. */
704    
705        case OP_NOT_DIGIT:        case OP_NOT_DIGIT:
706        for (c = 0; c < 32; c++)        for (c = 0; c < 32; c++)
# Line 699  do Line 757  do
757    
758        case OP_TYPEPLUS:        case OP_TYPEPLUS:
759        case OP_TYPEMINPLUS:        case OP_TYPEMINPLUS:
760          case OP_TYPEPOSPLUS:
761        tcode++;        tcode++;
762        break;        break;
763    
# Line 722  do Line 781  do
781        case OP_TYPEPOSQUERY:        case OP_TYPEPOSQUERY:
782        switch(tcode[1])        switch(tcode[1])
783          {          {
784            default:
785          case OP_ANY:          case OP_ANY:
786          case OP_ALLANY:          case OP_ALLANY:
787          return SSB_FAIL;          return SSB_FAIL;
788    
789            case OP_HSPACE:
790            SET_BIT(0x09);
791            SET_BIT(0x20);
792            SET_BIT(0xA0);
793            if (utf8)
794              {
795              SET_BIT(0xE1);  /* For U+1680, U+180E */
796              SET_BIT(0xE2);  /* For U+2000 - U+200A, U+202F, U+205F */
797              SET_BIT(0xE3);  /* For U+3000 */
798              }
799            break;
800    
801            case OP_ANYNL:
802            case OP_VSPACE:
803            SET_BIT(0x0A);
804            SET_BIT(0x0B);
805            SET_BIT(0x0C);
806            SET_BIT(0x0D);
807            SET_BIT(0x85);
808            if (utf8) SET_BIT(0xE2);    /* For U+2028, U+2029 */
809            break;
810    
811          case OP_NOT_DIGIT:          case OP_NOT_DIGIT:
812          for (c = 0; c < 32; c++)          for (c = 0; c < 32; c++)
813            start_bits[c] |= ~cd->cbits[c+cbit_digit];            start_bits[c] |= ~cd->cbits[c+cbit_digit];
Legend:
Removed from v.475  
changed lines
  Added in v.524
webmaster@exim.org
 ViewVC Help
Powered by ViewVC 1.1.12