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

Diff of /code/trunk/pcre_compile.c

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

revision 745 by ph10, Mon Nov 14 11:41:03 2011 UTC revision 749 by ph10, Fri Nov 18 10:36:45 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 1528  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 1551  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:
# Line 1566  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 1608  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 1646  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 1666  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 1697  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 1717  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 4315  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 4873  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. Then
4998        checking can be done. [This check is also applied to ONCE groups at        cancel the possessive flag so that the default action below, of wrapping
4999        runtime, but in a different way.] */        everything inside atomic brackets, does not happen. */
5000    
5001        else        else
5002          {          {
5003          uschar *ketcode = code - 1 - LINK_SIZE;          uschar *ketcode = code - 1 - LINK_SIZE;
5004          uschar *bracode = ketcode - GET(ketcode, 1);          uschar *bracode = ketcode - GET(ketcode, 1);
5005    
5006            /* Convert possessive ONCE brackets to non-capturing */
5007    
5008          if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) &&          if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) &&
5009              possessive_quantifier) *bracode = OP_BRA;              possessive_quantifier) *bracode = OP_BRA;
5010    
5011            /* For non-possessive ONCE brackets, all we need to do is to
5012            set the KET. */
5013    
5014          if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC)          if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC)
5015            *ketcode = OP_KETRMAX + repeat_type;            *ketcode = OP_KETRMAX + repeat_type;
5016    
5017            /* Handle non-ONCE brackets and possessive ONCEs (which have been
5018            converted to non-capturing above). */
5019    
5020          else          else
5021            {            {
5022            if (possessive_quantifier)            /* In the compile phase, check for empty string matching. */
5023              {  
             *bracode += 1;                   /* Switch to xxxPOS opcodes */  
             *ketcode = OP_KETRPOS;  
             if (brazeroptr != NULL) *brazeroptr = OP_BRAPOSZERO;  
             possessive_quantifier = FALSE;  
             }  
           else *ketcode = OP_KETRMAX + repeat_type;  
   
5024            if (lengthptr == NULL)            if (lengthptr == NULL)
5025              {              {
5026              uschar *scode = bracode;              uschar *scode = bracode;
# Line 4924  for (;; ptr++) Line 5035  for (;; ptr++)
5035                }                }
5036              while (*scode == OP_ALT);              while (*scode == OP_ALT);
5037              }              }
5038    
5039              /* Handle possessive quantifiers. */
5040    
5041              if (possessive_quantifier)
5042                {
5043                /* For COND brackets, we wrap the whole thing in a possessively
5044                repeated non-capturing bracket, because we have not invented POS
5045                versions of the COND opcodes. Because we are moving code along, we
5046                must ensure that any pending recursive references are updated. */
5047    
5048                if (*bracode == OP_COND || *bracode == OP_SCOND)
5049                  {
5050                  int nlen = (int)(code - bracode);
5051                  *code = OP_END;
5052                  adjust_recurse(bracode, 1 + LINK_SIZE, utf8, cd, save_hwm);
5053                  memmove(bracode + 1+LINK_SIZE, bracode, nlen);
5054                  code += 1 + LINK_SIZE;
5055                  nlen += 1 + LINK_SIZE;
5056                  *bracode = OP_BRAPOS;
5057                  *code++ = OP_KETRPOS;
5058                  PUTINC(code, 0, nlen);
5059                  PUT(bracode, 1, nlen);
5060                  }
5061    
5062                /* For non-COND brackets, we modify the BRA code and use KETRPOS. */
5063    
5064                else
5065                  {
5066                  *bracode += 1;              /* Switch to xxxPOS opcodes */
5067                  *ketcode = OP_KETRPOS;
5068                  }
5069    
5070                /* If the minimum is zero, mark it as possessive, then unset the
5071                possessive flag. */
5072    
5073                if (brazeroptr != NULL) *brazeroptr = OP_BRAPOSZERO;
5074                possessive_quantifier = FALSE;
5075                }
5076    
5077              /* Non-possessive quantifier */
5078    
5079              else *ketcode = OP_KETRMAX + repeat_type;
5080            }            }
5081          }          }
5082        }        }
# Line 6614  for (;;) Line 6767  for (;;)
6767          }          }
6768        else if (fixed_length < 0)        else if (fixed_length < 0)
6769          {          {
6770          *errorcodeptr = (fixed_length == -2)? ERR36 : ERR25;          *errorcodeptr = (fixed_length == -2)? ERR36 :
6771                            (fixed_length == -4)? ERR70: ERR25;
6772          *ptrptr = ptr;          *ptrptr = ptr;
6773          return FALSE;          return FALSE;
6774          }          }
# Line 7413  if (cd->check_lookbehind) Line 7567  if (cd->check_lookbehind)
7567        DPRINTF(("fixed length = %d\n", fixed_length));        DPRINTF(("fixed length = %d\n", fixed_length));
7568        if (fixed_length < 0)        if (fixed_length < 0)
7569          {          {
7570          errorcode = (fixed_length == -2)? ERR36 : ERR25;          errorcode = (fixed_length == -2)? ERR36 :
7571                        (fixed_length == -4)? ERR70 : ERR25;
7572          break;          break;
7573          }          }
7574        PUT(cc, 1, fixed_length);        PUT(cc, 1, fixed_length);

Legend:
Removed from v.745  
changed lines
  Added in v.749

  ViewVC Help
Powered by ViewVC 1.1.5