/[pcre]/code/branches/pcre16/pcre_compile.c
ViewVC logotype

Diff of /code/branches/pcre16/pcre_compile.c

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

revision 654 by ph10, Tue Aug 2 11:00:40 2011 UTC revision 750 by ph10, Fri Nov 18 11:07:14 2011 UTC
# Line 410  static const char error_texts[] = Line 410  static const char error_texts[] =
410    "this version of PCRE is not compiled with PCRE_UCP support\0"    "this version of PCRE is not compiled with PCRE_UCP support\0"
411    "\\c must be followed by an ASCII character\0"    "\\c must be followed by an ASCII character\0"
412    "\\k is not followed by a braced, angle-bracketed, or quoted name\0"    "\\k is not followed by a braced, angle-bracketed, or quoted name\0"
413      /* 70 */
414      "internal error: unknown opcode in find_fixedlength()\0"
415    ;    ;
416    
417  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 676  else Line 678  else
678    
679      case CHAR_l:      case CHAR_l:
680      case CHAR_L:      case CHAR_L:
681        *errorcodeptr = ERR37;
682        break;
683    
684      case CHAR_u:      case CHAR_u:
685        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
686          {
687          /* In JavaScript, \u must be followed by four hexadecimal numbers.
688          Otherwise it is a lowercase u letter. */
689          if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0
690               && (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0)
691            {
692            c = 0;
693            for (i = 0; i < 4; ++i)
694              {
695              register int cc = *(++ptr);
696    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
697              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
698              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
699    #else           /* EBCDIC coding */
700              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
701              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
702    #endif
703              }
704            }
705          }
706        else
707          *errorcodeptr = ERR37;
708        break;
709    
710      case CHAR_U:      case CHAR_U:
711      *errorcodeptr = ERR37;      /* In JavaScript, \U is an uppercase U letter. */
712        if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37;
713      break;      break;
714    
715      /* In a character class, \g is just a literal "g". Outside a character      /* In a character class, \g is just a literal "g". Outside a character
# Line 828  else Line 859  else
859      treated as a data character. */      treated as a data character. */
860    
861      case CHAR_x:      case CHAR_x:
862        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
863          {
864          /* In JavaScript, \x must be followed by two hexadecimal numbers.
865          Otherwise it is a lowercase x letter. */
866          if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0)
867            {
868            c = 0;
869            for (i = 0; i < 2; ++i)
870              {
871              register int cc = *(++ptr);
872    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
873              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
874              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
875    #else           /* EBCDIC coding */
876              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
877              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
878    #endif
879              }
880            }
881          break;
882          }
883    
884      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
885        {        {
886        const uschar *pt = ptr + 2;        const uschar *pt = ptr + 2;
# Line 1477  Returns:   the fixed length, Line 1530  Returns:   the fixed length,
1530               or -1 if there is no fixed length,               or -1 if there is no fixed length,
1531               or -2 if \C was encountered               or -2 if \C was encountered
1532               or -3 if an OP_RECURSE item was encountered and atend is FALSE               or -3 if an OP_RECURSE item was encountered and atend is FALSE
1533                 or -4 if an unknown opcode was encountered (internal error)
1534  */  */
1535    
1536  static int  static int
# Line 1500  for (;;) Line 1554  for (;;)
1554      /* We only need to continue for OP_CBRA (normal capturing bracket) and      /* We only need to continue for OP_CBRA (normal capturing bracket) and
1555      OP_BRA (normal non-capturing bracket) because the other variants of these      OP_BRA (normal non-capturing bracket) because the other variants of these
1556      opcodes are all concerned with unlimited repeated groups, which of course      opcodes are all concerned with unlimited repeated groups, which of course
1557      are not of fixed length. They will cause a -1 response from the default      are not of fixed length. */
     case of this switch. */  
1558    
1559      case OP_CBRA:      case OP_CBRA:
1560      case OP_BRA:      case OP_BRA:
1561      case OP_ONCE:      case OP_ONCE:
1562        case OP_ONCE_NC:
1563      case OP_COND:      case OP_COND:
1564      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd);      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd);
1565      if (d < 0) return d;      if (d < 0) return d;
# Line 1514  for (;;) Line 1568  for (;;)
1568      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
1569      break;      break;
1570    
1571      /* Reached end of a branch; if it's a ket it is the end of a nested      /* Reached end of a branch; if it's a ket it is the end of a nested call.
1572      call. If it's ALT it is an alternation in a nested call. If it is      If it's ALT it is an alternation in a nested call. An ACCEPT is effectively
1573      END it's the end of the outer call. All can be handled by the same code.      an ALT. If it is END it's the end of the outer call. All can be handled by
1574      Note that we must not include the OP_KETRxxx opcodes here, because they      the same code. Note that we must not include the OP_KETRxxx opcodes here,
1575      all imply an unlimited repeat. */      because they all imply an unlimited repeat. */
1576    
1577      case OP_ALT:      case OP_ALT:
1578      case OP_KET:      case OP_KET:
1579      case OP_END:      case OP_END:
1580        case OP_ACCEPT:
1581        case OP_ASSERT_ACCEPT:
1582      if (length < 0) length = branchlength;      if (length < 0) length = branchlength;
1583        else if (length != branchlength) return -1;        else if (length != branchlength) return -1;
1584      if (*cc != OP_ALT) return length;      if (*cc != OP_ALT) return length;
# Line 1556  for (;;) Line 1612  for (;;)
1612    
1613      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1614    
1615      case OP_REVERSE:      case OP_MARK:
1616      case OP_CREF:      case OP_PRUNE_ARG:
1617      case OP_NCREF:      case OP_SKIP_ARG:
1618      case OP_RREF:      case OP_THEN_ARG:
1619      case OP_NRREF:      cc += cc[1] + _pcre_OP_lengths[*cc];
1620      case OP_DEF:      break;
1621    
1622      case OP_CALLOUT:      case OP_CALLOUT:
     case OP_SOD:  
     case OP_SOM:  
     case OP_SET_SOM:  
     case OP_EOD:  
     case OP_EODN:  
1623      case OP_CIRC:      case OP_CIRC:
1624      case OP_CIRCM:      case OP_CIRCM:
1625        case OP_CLOSE:
1626        case OP_COMMIT:
1627        case OP_CREF:
1628        case OP_DEF:
1629      case OP_DOLL:      case OP_DOLL:
1630      case OP_DOLLM:      case OP_DOLLM:
1631        case OP_EOD:
1632        case OP_EODN:
1633        case OP_FAIL:
1634        case OP_NCREF:
1635        case OP_NRREF:
1636      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1637        case OP_PRUNE:
1638        case OP_REVERSE:
1639        case OP_RREF:
1640        case OP_SET_SOM:
1641        case OP_SKIP:
1642        case OP_SOD:
1643        case OP_SOM:
1644        case OP_THEN:
1645      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1646      cc += _pcre_OP_lengths[*cc];      cc += _pcre_OP_lengths[*cc];
1647      break;      break;
# Line 1594  for (;;) Line 1663  for (;;)
1663      need to skip over a multibyte character in UTF8 mode.  */      need to skip over a multibyte character in UTF8 mode.  */
1664    
1665      case OP_EXACT:      case OP_EXACT:
1666        case OP_EXACTI:
1667        case OP_NOTEXACT:
1668        case OP_NOTEXACTI:
1669      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1670      cc += 4;      cc += 4;
1671  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1614  for (;;) Line 1686  for (;;)
1686      cc += 2;      cc += 2;
1687      /* Fall through */      /* Fall through */
1688    
1689        case OP_HSPACE:
1690        case OP_VSPACE:
1691        case OP_NOT_HSPACE:
1692        case OP_NOT_VSPACE:
1693      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
1694      case OP_DIGIT:      case OP_DIGIT:
1695      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
# Line 1645  for (;;) Line 1721  for (;;)
1721    
1722      switch (*cc)      switch (*cc)
1723        {        {
1724          case OP_CRPLUS:
1725          case OP_CRMINPLUS:
1726        case OP_CRSTAR:        case OP_CRSTAR:
1727        case OP_CRMINSTAR:        case OP_CRMINSTAR:
1728        case OP_CRQUERY:        case OP_CRQUERY:
# Line 1665  for (;;) Line 1743  for (;;)
1743    
1744      /* Anything else is variable length */      /* Anything else is variable length */
1745    
1746      default:      case OP_ANYNL:
1747        case OP_BRAMINZERO:
1748        case OP_BRAPOS:
1749        case OP_BRAPOSZERO:
1750        case OP_BRAZERO:
1751        case OP_CBRAPOS:
1752        case OP_EXTUNI:
1753        case OP_KETRMAX:
1754        case OP_KETRMIN:
1755        case OP_KETRPOS:
1756        case OP_MINPLUS:
1757        case OP_MINPLUSI:
1758        case OP_MINQUERY:
1759        case OP_MINQUERYI:
1760        case OP_MINSTAR:
1761        case OP_MINSTARI:
1762        case OP_MINUPTO:
1763        case OP_MINUPTOI:
1764        case OP_NOTMINPLUS:
1765        case OP_NOTMINPLUSI:
1766        case OP_NOTMINQUERY:
1767        case OP_NOTMINQUERYI:
1768        case OP_NOTMINSTAR:
1769        case OP_NOTMINSTARI:
1770        case OP_NOTMINUPTO:
1771        case OP_NOTMINUPTOI:
1772        case OP_NOTPLUS:
1773        case OP_NOTPLUSI:
1774        case OP_NOTPOSPLUS:
1775        case OP_NOTPOSPLUSI:
1776        case OP_NOTPOSQUERY:
1777        case OP_NOTPOSQUERYI:
1778        case OP_NOTPOSSTAR:
1779        case OP_NOTPOSSTARI:
1780        case OP_NOTPOSUPTO:
1781        case OP_NOTPOSUPTOI:
1782        case OP_NOTQUERY:
1783        case OP_NOTQUERYI:
1784        case OP_NOTSTAR:
1785        case OP_NOTSTARI:
1786        case OP_NOTUPTO:
1787        case OP_NOTUPTOI:
1788        case OP_PLUS:
1789        case OP_PLUSI:
1790        case OP_POSPLUS:
1791        case OP_POSPLUSI:
1792        case OP_POSQUERY:
1793        case OP_POSQUERYI:
1794        case OP_POSSTAR:
1795        case OP_POSSTARI:
1796        case OP_POSUPTO:
1797        case OP_POSUPTOI:
1798        case OP_QUERY:
1799        case OP_QUERYI:
1800        case OP_REF:
1801        case OP_REFI:
1802        case OP_SBRA:
1803        case OP_SBRAPOS:
1804        case OP_SCBRA:
1805        case OP_SCBRAPOS:
1806        case OP_SCOND:
1807        case OP_SKIPZERO:
1808        case OP_STAR:
1809        case OP_STARI:
1810        case OP_TYPEMINPLUS:
1811        case OP_TYPEMINQUERY:
1812        case OP_TYPEMINSTAR:
1813        case OP_TYPEMINUPTO:
1814        case OP_TYPEPLUS:
1815        case OP_TYPEPOSPLUS:
1816        case OP_TYPEPOSQUERY:
1817        case OP_TYPEPOSSTAR:
1818        case OP_TYPEPOSUPTO:
1819        case OP_TYPEQUERY:
1820        case OP_TYPESTAR:
1821        case OP_TYPEUPTO:
1822        case OP_UPTO:
1823        case OP_UPTOI:
1824      return -1;      return -1;
1825    
1826        /* Catch unrecognized opcodes so that when new ones are added they
1827        are not forgotten, as has happened in the past. */
1828    
1829        default:
1830        return -4;
1831      }      }
1832    }    }
1833  /* Control never gets here */  /* Control never gets here */
# Line 1761  for (;;) Line 1922  for (;;)
1922        break;        break;
1923    
1924        case OP_THEN_ARG:        case OP_THEN_ARG:
1925        code += code[1+LINK_SIZE];        code += code[1];
1926        break;        break;
1927        }        }
1928    
# Line 1880  for (;;) Line 2041  for (;;)
2041        break;        break;
2042    
2043        case OP_THEN_ARG:        case OP_THEN_ARG:
2044        code += code[1+LINK_SIZE];        code += code[1];
2045        break;        break;
2046        }        }
2047    
# Line 2045  for (code = first_significant_code(code Line 2206  for (code = first_significant_code(code
2206    
2207    if (c == OP_BRA  || c == OP_BRAPOS ||    if (c == OP_BRA  || c == OP_BRAPOS ||
2208        c == OP_CBRA || c == OP_CBRAPOS ||        c == OP_CBRA || c == OP_CBRAPOS ||
2209        c == OP_ONCE || c == OP_COND)        c == OP_ONCE || c == OP_ONCE_NC ||
2210          c == OP_COND)
2211      {      {
2212      BOOL empty_branch;      BOOL empty_branch;
2213      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 2217  for (code = first_significant_code(code Line 2379  for (code = first_significant_code(code
2379      break;      break;
2380    
2381      case OP_THEN_ARG:      case OP_THEN_ARG:
2382      code += code[1+LINK_SIZE];      code += code[1];
2383      break;      break;
2384    
2385      /* None of the remaining opcodes are required to match a character. */      /* None of the remaining opcodes are required to match a character. */
# Line 2295  I think. Line 2457  I think.
2457  A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.  A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.
2458  It seems that the appearance of a nested POSIX class supersedes an apparent  It seems that the appearance of a nested POSIX class supersedes an apparent
2459  external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or  external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or
2460  a digit. Also, unescaped square brackets may also appear as part of class  a digit.
2461  names. For example, [:a[:abc]b:] gives unknown class "[:abc]b:]"in Perl.  
2462    In Perl, unescaped square brackets may also appear as part of class names. For
2463    example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for
2464    [:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not
2465    seem right at all. PCRE does not allow closing square brackets in POSIX class
2466    names.
2467    
2468  Arguments:  Arguments:
2469    ptr      pointer to the initial [    ptr      pointer to the initial [
# Line 2314  for (++ptr; *ptr != 0; ptr++) Line 2481  for (++ptr; *ptr != 0; ptr++)
2481    {    {
2482    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2483      ptr++;      ptr++;
2484      else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
2485    else    else
2486      {      {
2487      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
# Line 3086  uschar *class_utf8data_base; Line 3254  uschar *class_utf8data_base;
3254  uschar utf8_char[6];  uschar utf8_char[6];
3255  #else  #else
3256  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
 uschar *utf8_char = NULL;  
3257  #endif  #endif
3258    
3259  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3137  for (;; ptr++) Line 3304  for (;; ptr++)
3304    int subfirstbyte;    int subfirstbyte;
3305    int terminator;    int terminator;
3306    int mclength;    int mclength;
3307      int tempbracount;
3308    uschar mcbuffer[8];    uschar mcbuffer[8];
3309    
3310    /* Get next byte in the pattern */    /* Get next byte in the pattern */
# Line 4256  for (;; ptr++) Line 4424  for (;; ptr++)
4424      past, but it no longer happens for non-repeated recursions. In fact, the      past, but it no longer happens for non-repeated recursions. In fact, the
4425      repeated ones could be re-implemented independently so as not to need this,      repeated ones could be re-implemented independently so as not to need this,
4426      but for the moment we rely on the code for repeating groups. */      but for the moment we rely on the code for repeating groups. */
4427    
4428      if (*previous == OP_RECURSE)      if (*previous == OP_RECURSE)
4429        {        {
4430        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);
# Line 4814  for (;; ptr++) Line 4982  for (;; ptr++)
4982        ONCE brackets can be converted into non-capturing brackets, as the        ONCE brackets can be converted into non-capturing brackets, as the
4983        behaviour of (?:xx)++ is the same as (?>xx)++ and this saves having to        behaviour of (?:xx)++ is the same as (?>xx)++ and this saves having to
4984        deal with possessive ONCEs specially.        deal with possessive ONCEs specially.
4985    
4986        Otherwise, if the quantifier was possessive, we convert the BRA code to        Otherwise, when we are doing the actual compile phase, check to see
4987        the POS form, and the KET code to KETRPOS. (It turns out to be convenient        whether this group is one that could match an empty string. If so,
4988        at runtime to detect this kind of subpattern at both the start and at the        convert the initial operator to the S form (e.g. OP_BRA -> OP_SBRA) so
4989        end.) The use of special opcodes makes it possible to reduce greatly the        that runtime checking can be done. [This check is also applied to ONCE
4990        stack usage in pcre_exec(). If the group is preceded by OP_BRAZERO,        groups at runtime, but in a different way.]
4991        convert this to OP_BRAPOSZERO. Then cancel the possessive flag so that  
4992        the default action below, of wrapping everything inside atomic brackets,        Then, if the quantifier was possessive and the bracket is not a
4993        does not happen.        conditional, we convert the BRA code to the POS form, and the KET code to
4994          KETRPOS. (It turns out to be convenient at runtime to detect this kind of
4995        Then, when we are doing the actual compile phase, check to see whether        subpattern at both the start and at the end.) The use of special opcodes
4996        this group is one that could match an empty string. If so, convert the        makes it possible to reduce greatly the stack usage in pcre_exec(). If
4997        initial operator to the S form (e.g. OP_BRA -> OP_SBRA) so that runtime        the group is preceded by OP_BRAZERO, convert this to OP_BRAPOSZERO.
4998        checking can be done. [This check is also applied to ONCE groups at  
4999        runtime, but in a different way.] */        Then, if the minimum number of matches is 1 or 0, cancel the possessive
5000          flag so that the default action below, of wrapping everything inside
5001          atomic brackets, does not happen. When the minimum is greater than 1,
5002          there will be earlier copies of the group, and so we still have to wrap
5003          the whole thing. */
5004    
5005        else        else
5006          {          {
5007          uschar *ketcode = code - 1 - LINK_SIZE;          uschar *ketcode = code - 1 - LINK_SIZE;
5008          uschar *bracode = ketcode - GET(ketcode, 1);          uschar *bracode = ketcode - GET(ketcode, 1);
5009    
5010          if (*bracode == OP_ONCE && possessive_quantifier) *bracode = OP_BRA;          /* Convert possessive ONCE brackets to non-capturing */
5011          if (*bracode == OP_ONCE)  
5012            if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) &&
5013                possessive_quantifier) *bracode = OP_BRA;
5014    
5015            /* For non-possessive ONCE brackets, all we need to do is to
5016            set the KET. */
5017    
5018            if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC)
5019            *ketcode = OP_KETRMAX + repeat_type;            *ketcode = OP_KETRMAX + repeat_type;
5020    
5021            /* Handle non-ONCE brackets and possessive ONCEs (which have been
5022            converted to non-capturing above). */
5023    
5024          else          else
5025            {            {
5026            if (possessive_quantifier)            /* In the compile phase, check for empty string matching. */
5027              {  
             *bracode += 1;                   /* Switch to xxxPOS opcodes */  
             *ketcode = OP_KETRPOS;  
             if (brazeroptr != NULL) *brazeroptr = OP_BRAPOSZERO;  
             possessive_quantifier = FALSE;  
             }  
           else *ketcode = OP_KETRMAX + repeat_type;  
   
5028            if (lengthptr == NULL)            if (lengthptr == NULL)
5029              {              {
5030              uschar *scode = bracode;              uschar *scode = bracode;
# Line 4863  for (;; ptr++) Line 5039  for (;; ptr++)
5039                }                }
5040              while (*scode == OP_ALT);              while (*scode == OP_ALT);
5041              }              }
5042    
5043              /* Handle possessive quantifiers. */
5044    
5045              if (possessive_quantifier)
5046                {
5047                /* For COND brackets, we wrap the whole thing in a possessively
5048                repeated non-capturing bracket, because we have not invented POS
5049                versions of the COND opcodes. Because we are moving code along, we
5050                must ensure that any pending recursive references are updated. */
5051    
5052                if (*bracode == OP_COND || *bracode == OP_SCOND)
5053                  {
5054                  int nlen = (int)(code - bracode);
5055                  *code = OP_END;
5056                  adjust_recurse(bracode, 1 + LINK_SIZE, utf8, cd, save_hwm);
5057                  memmove(bracode + 1+LINK_SIZE, bracode, nlen);
5058                  code += 1 + LINK_SIZE;
5059                  nlen += 1 + LINK_SIZE;
5060                  *bracode = OP_BRAPOS;
5061                  *code++ = OP_KETRPOS;
5062                  PUTINC(code, 0, nlen);
5063                  PUT(bracode, 1, nlen);
5064                  }
5065    
5066                /* For non-COND brackets, we modify the BRA code and use KETRPOS. */
5067    
5068                else
5069                  {
5070                  *bracode += 1;              /* Switch to xxxPOS opcodes */
5071                  *ketcode = OP_KETRPOS;
5072                  }
5073    
5074                /* If the minimum is zero, mark it as possessive, then unset the
5075                possessive flag when the minimum is 0 or 1. */
5076    
5077                if (brazeroptr != NULL) *brazeroptr = OP_BRAPOSZERO;
5078                if (repeat_min < 2) possessive_quantifier = FALSE;
5079                }
5080    
5081              /* Non-possessive quantifier */
5082    
5083              else *ketcode = OP_KETRMAX + repeat_type;
5084            }            }
5085          }          }
5086        }        }
# Line 4889  for (;; ptr++) Line 5107  for (;; ptr++)
5107      notation is just syntactic sugar, taken from Sun's Java package, but the      notation is just syntactic sugar, taken from Sun's Java package, but the
5108      special opcodes can optimize it.      special opcodes can optimize it.
5109    
5110      Possessively repeated subpatterns have already been handled in the code      Some (but not all) possessively repeated subpatterns have already been
5111      just above, so possessive_quantifier is always FALSE for them at this      completely handled in the code just above. For them, possessive_quantifier
5112      stage.      is always FALSE at this stage.
5113    
5114      Note that the repeated item starts at tempcode, not at previous, which      Note that the repeated item starts at tempcode, not at previous, which
5115      might be the first part of a string whose (former) last char we repeated.      might be the first part of a string whose (former) last char we repeated.
# Line 5040  for (;; ptr++) Line 5258  for (;; ptr++)
5258                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5259                }                }
5260              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5261    
5262                /* Do not set firstbyte after *ACCEPT */
5263                if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
5264              }              }
5265    
5266            /* Handle other cases with/without an argument */            /* Handle other cases with/without an argument */
# Line 5052  for (;; ptr++) Line 5273  for (;; ptr++)
5273                goto FAILED;                goto FAILED;
5274                }                }
5275              *code = verbs[i].op;              *code = verbs[i].op;
5276              if (*code++ == OP_THEN)              if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;
               {  
               PUT(code, 0, code - bcptr->current_branch - 1);  
               code += LINK_SIZE;  
               }  
5277              }              }
5278    
5279            else            else
# Line 5067  for (;; ptr++) Line 5284  for (;; ptr++)
5284                goto FAILED;                goto FAILED;
5285                }                }
5286              *code = verbs[i].op_arg;              *code = verbs[i].op_arg;
5287              if (*code++ == OP_THEN_ARG)              if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;
               {  
               PUT(code, 0, code - bcptr->current_branch - 1);  
               code += LINK_SIZE;  
               }  
5288              *code++ = arglen;              *code++ = arglen;
5289              memcpy(code, arg, arglen);              memcpy(code, arg, arglen);
5290              code += arglen;              code += arglen;
# Line 5906  for (;; ptr++) Line 6119  for (;; ptr++)
6119      *code = bravalue;      *code = bravalue;
6120      tempcode = code;      tempcode = code;
6121      tempreqvary = cd->req_varyopt;        /* Save value before bracket */      tempreqvary = cd->req_varyopt;        /* Save value before bracket */
6122        tempbracount = cd->bracount;          /* Save value before bracket */
6123      length_prevgroup = 0;                 /* Initialize for pre-compile phase */      length_prevgroup = 0;                 /* Initialize for pre-compile phase */
6124    
6125      if (!compile_regex(      if (!compile_regex(
# Line 5928  for (;; ptr++) Line 6142  for (;; ptr++)
6142           ))           ))
6143        goto FAILED;        goto FAILED;
6144    
6145        /* If this was an atomic group and there are no capturing groups within it,
6146        generate OP_ONCE_NC instead of OP_ONCE. */
6147    
6148        if (bravalue == OP_ONCE && cd->bracount <= tempbracount)
6149          *code = OP_ONCE_NC;
6150    
6151      if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT)      if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT)
6152        cd->assert_depth -= 1;        cd->assert_depth -= 1;
6153    
6154      /* At the end of compiling, code is still pointing to the start of the      /* At the end of compiling, code is still pointing to the start of the
6155      group, while tempcode has been updated to point past the end of the group      group, while tempcode has been updated to point past the end of the group.
6156      and any option resetting that may follow it. The pattern pointer (ptr)      The pattern pointer (ptr) is on the bracket.
     is on the bracket. */  
6157    
6158      /* If this is a conditional bracket, check that there are no more than      If this is a conditional bracket, check that there are no more than
6159      two branches in the group, or just one if it's a DEFINE group. We do this      two branches in the group, or just one if it's a DEFINE group. We do this
6160      in the real compile phase, not in the pre-pass, where the whole group may      in the real compile phase, not in the pre-pass, where the whole group may
6161      not be available. */      not be available. */
# Line 6335  for (;; ptr++) Line 6554  for (;; ptr++)
6554        else firstbyte = reqbyte = REQ_NONE;        else firstbyte = reqbyte = REQ_NONE;
6555        }        }
6556    
6557      /* firstbyte was previously set; we can set reqbyte only the length is      /* firstbyte was previously set; we can set reqbyte only if the length is
6558      1 or the matching is caseful. */      1 or the matching is caseful. */
6559    
6560      else      else
# Line 6552  for (;;) Line 6771  for (;;)
6771          }          }
6772        else if (fixed_length < 0)        else if (fixed_length < 0)
6773          {          {
6774          *errorcodeptr = (fixed_length == -2)? ERR36 : ERR25;          *errorcodeptr = (fixed_length == -2)? ERR36 :
6775                            (fixed_length == -4)? ERR70: ERR25;
6776          *ptrptr = ptr;          *ptrptr = ptr;
6777          return FALSE;          return FALSE;
6778          }          }
# Line 6727  do { Line 6947  do {
6947    
6948     /* Other brackets */     /* Other brackets */
6949    
6950     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_COND)     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||
6951                op == OP_COND)
6952       {       {
6953       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;
6954       }       }
# Line 6831  do { Line 7052  do {
7052    
7053     /* Other brackets */     /* Other brackets */
7054    
7055     else if (op == OP_ASSERT || op == OP_ONCE)     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC)
7056       {       {
7057       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;
7058       }       }
# Line 6901  do { Line 7122  do {
7122       case OP_SCBRAPOS:       case OP_SCBRAPOS:
7123       case OP_ASSERT:       case OP_ASSERT:
7124       case OP_ONCE:       case OP_ONCE:
7125         case OP_ONCE_NC:
7126       case OP_COND:       case OP_COND:
7127       if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0)       if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0)
7128         return -1;         return -1;
# Line 7282  re->top_bracket = cd->bracount; Line 7504  re->top_bracket = cd->bracount;
7504  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
7505  re->flags = cd->external_flags;  re->flags = cd->external_flags;
7506    
7507  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */  if (cd->had_accept) reqbyte = REQ_NONE;   /* Must disable after (*ACCEPT) */
7508    
7509  /* If not reached end of pattern on success, there's an excess bracket. */  /* If not reached end of pattern on success, there's an excess bracket. */
7510    
# Line 7349  if (cd->check_lookbehind) Line 7571  if (cd->check_lookbehind)
7571        DPRINTF(("fixed length = %d\n", fixed_length));        DPRINTF(("fixed length = %d\n", fixed_length));
7572        if (fixed_length < 0)        if (fixed_length < 0)
7573          {          {
7574          errorcode = (fixed_length == -2)? ERR36 : ERR25;          errorcode = (fixed_length == -2)? ERR36 :
7575                        (fixed_length == -4)? ERR70 : ERR25;
7576          break;          break;
7577          }          }
7578        PUT(cc, 1, fixed_length);        PUT(cc, 1, fixed_length);

Legend:
Removed from v.654  
changed lines
  Added in v.750

  ViewVC Help
Powered by ViewVC 1.1.5