/[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 643 by ph10, Fri Jul 29 15:56:39 2011 UTC revision 746 by ph10, Tue Nov 15 15:07:02 2011 UTC
# Line 409  static const char error_texts[] = Line 409  static const char error_texts[] =
409    "(*MARK) must have an argument\0"    "(*MARK) must have an argument\0"
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    ;    ;
414    
415  /* 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 676  else
676    
677      case CHAR_l:      case CHAR_l:
678      case CHAR_L:      case CHAR_L:
679        *errorcodeptr = ERR37;
680        break;
681    
682      case CHAR_u:      case CHAR_u:
683        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
684          {
685          /* In JavaScript, \u must be followed by four hexadecimal numbers.
686          Otherwise it is a lowercase u letter. */
687          if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0
688               && (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0)
689            {
690            c = 0;
691            for (i = 0; i < 4; ++i)
692              {
693              register int cc = *(++ptr);
694    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
695              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
696              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
697    #else           /* EBCDIC coding */
698              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
699              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
700    #endif
701              }
702            }
703          }
704        else
705          *errorcodeptr = ERR37;
706        break;
707    
708      case CHAR_U:      case CHAR_U:
709      *errorcodeptr = ERR37;      /* In JavaScript, \U is an uppercase U letter. */
710        if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37;
711      break;      break;
712    
713      /* 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
714      class, \g must be followed by one of a number of specific things:      class, \g must be followed by one of a number of specific things:
715    
716      (1) A number, either plain or braced. If positive, it is an absolute      (1) A number, either plain or braced. If positive, it is an absolute
# Line 828  else Line 857  else
857      treated as a data character. */      treated as a data character. */
858    
859      case CHAR_x:      case CHAR_x:
860        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
861          {
862          /* In JavaScript, \x must be followed by two hexadecimal numbers.
863          Otherwise it is a lowercase x letter. */
864          if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0)
865            {
866            c = 0;
867            for (i = 0; i < 2; ++i)
868              {
869              register int cc = *(++ptr);
870    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
871              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
872              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
873    #else           /* EBCDIC coding */
874              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
875              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
876    #endif
877              }
878            }
879          break;
880          }
881    
882      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
883        {        {
884        const uschar *pt = ptr + 2;        const uschar *pt = ptr + 2;
# Line 921  else Line 972  else
972    }    }
973    
974  /* Perl supports \N{name} for character names, as well as plain \N for "not  /* Perl supports \N{name} for character names, as well as plain \N for "not
975  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
976  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
977    
978  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
# Line 1506  for (;;) Line 1557  for (;;)
1557      case OP_CBRA:      case OP_CBRA:
1558      case OP_BRA:      case OP_BRA:
1559      case OP_ONCE:      case OP_ONCE:
1560        case OP_ONCE_NC:
1561      case OP_COND:      case OP_COND:
1562      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd);      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd);
1563      if (d < 0) return d;      if (d < 0) return d;
# Line 1594  for (;;) Line 1646  for (;;)
1646      need to skip over a multibyte character in UTF8 mode.  */      need to skip over a multibyte character in UTF8 mode.  */
1647    
1648      case OP_EXACT:      case OP_EXACT:
1649        case OP_EXACTI:
1650      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1651      cc += 4;      cc += 4;
1652  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1761  for (;;) Line 1814  for (;;)
1814        break;        break;
1815    
1816        case OP_THEN_ARG:        case OP_THEN_ARG:
1817        code += code[1+LINK_SIZE];        code += code[1];
1818        break;        break;
1819        }        }
1820    
# Line 1880  for (;;) Line 1933  for (;;)
1933        break;        break;
1934    
1935        case OP_THEN_ARG:        case OP_THEN_ARG:
1936        code += code[1+LINK_SIZE];        code += code[1];
1937        break;        break;
1938        }        }
1939    
# Line 1982  for (code = first_significant_code(code Line 2035  for (code = first_significant_code(code
2035    /* For a recursion/subroutine call, if its end has been reached, which    /* For a recursion/subroutine call, if its end has been reached, which
2036    implies a backward reference subroutine call, we can scan it. If it's a    implies a backward reference subroutine call, we can scan it. If it's a
2037    forward reference subroutine call, we can't. To detect forward reference    forward reference subroutine call, we can't. To detect forward reference
2038    we have to scan up the list that is kept in the workspace. This function is    we have to scan up the list that is kept in the workspace. This function is
2039    called only when doing the real compile, not during the pre-compile that    called only when doing the real compile, not during the pre-compile that
2040    measures the size of the compiled pattern. */    measures the size of the compiled pattern. */
2041    
2042    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2043      {      {
2044      const uschar *scode;      const uschar *scode;
2045      BOOL empty_branch;      BOOL empty_branch;
2046    
2047      /* Test for forward reference */      /* Test for forward reference */
2048    
2049      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2050        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;
2051    
2052      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2053    
2054      empty_branch = FALSE;      empty_branch = FALSE;
2055      scode = cd->start_code + GET(code, 1);      scode = cd->start_code + GET(code, 1);
2056      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2057    
2058      /* Completed backwards reference */      /* Completed backwards reference */
2059    
2060      do      do
2061        {        {
2062        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf8, cd))
# Line 2014  for (code = first_significant_code(code Line 2067  for (code = first_significant_code(code
2067        scode += GET(scode, 1);        scode += GET(scode, 1);
2068        }        }
2069      while (*scode == OP_ALT);      while (*scode == OP_ALT);
2070    
2071      if (!empty_branch) return FALSE;  /* All branches are non-empty */      if (!empty_branch) return FALSE;  /* All branches are non-empty */
2072      continue;      continue;
2073      }      }
# Line 2045  for (code = first_significant_code(code Line 2098  for (code = first_significant_code(code
2098    
2099    if (c == OP_BRA  || c == OP_BRAPOS ||    if (c == OP_BRA  || c == OP_BRAPOS ||
2100        c == OP_CBRA || c == OP_CBRAPOS ||        c == OP_CBRA || c == OP_CBRAPOS ||
2101        c == OP_ONCE || c == OP_COND)        c == OP_ONCE || c == OP_ONCE_NC ||
2102          c == OP_COND)
2103      {      {
2104      BOOL empty_branch;      BOOL empty_branch;
2105      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 2271  for (code = first_significant_code(code
2271      break;      break;
2272    
2273      case OP_THEN_ARG:      case OP_THEN_ARG:
2274      code += code[1+LINK_SIZE];      code += code[1];
2275      break;      break;
2276    
2277      /* None of the remaining opcodes are required to match a character. */      /* None of the remaining opcodes are required to match a character. */
# Line 2240  return TRUE; Line 2294  return TRUE;
2294  the current branch of the current pattern to see if it could match the empty  the current branch of the current pattern to see if it could match the empty
2295  string. If it could, we must look outwards for branches at other levels,  string. If it could, we must look outwards for branches at other levels,
2296  stopping when we pass beyond the bracket which is the subject of the recursion.  stopping when we pass beyond the bracket which is the subject of the recursion.
2297  This function is called only during the real compile, not during the  This function is called only during the real compile, not during the
2298  pre-compile.  pre-compile.
2299    
2300  Arguments:  Arguments:
# Line 2295  I think. Line 2349  I think.
2349  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.
2350  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
2351  external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or  external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or
2352  a digit. Also, unescaped square brackets may also appear as part of class  a digit.
2353  names. For example, [:a[:abc]b:] gives unknown class "[:abc]b:]"in Perl.  
2354    In Perl, unescaped square brackets may also appear as part of class names. For
2355    example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for
2356    [:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not
2357    seem right at all. PCRE does not allow closing square brackets in POSIX class
2358    names.
2359    
2360  Arguments:  Arguments:
2361    ptr      pointer to the initial [    ptr      pointer to the initial [
# Line 2312  int terminator;          /* Don't combin Line 2371  int terminator;          /* Don't combin
2371  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2372  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != 0; ptr++)
2373    {    {
2374    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2375      ptr++;      ptr++;
2376      else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
2377    else    else
2378      {      {
2379      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
# Line 2325  for (++ptr; *ptr != 0; ptr++) Line 2385  for (++ptr; *ptr != 0; ptr++)
2385           (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||           (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2386            ptr[1] == CHAR_EQUALS_SIGN) &&            ptr[1] == CHAR_EQUALS_SIGN) &&
2387          check_posix_syntax(ptr, endptr))          check_posix_syntax(ptr, endptr))
2388        return FALSE;        return FALSE;
2389      }      }
2390    }    }
2391  return FALSE;  return FALSE;
# Line 3035  Arguments: Line 3095  Arguments:
3095    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)
3096    reqbyteptr     set to the last literal character required, else < 0    reqbyteptr     set to the last literal character required, else < 0
3097    bcptr          points to current branch chain    bcptr          points to current branch chain
3098    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3099    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
3100    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
3101                   points to length accumulator during pre-compile phase                   points to length accumulator during pre-compile phase
# Line 3075  uschar *save_hwm = NULL; Line 3135  uschar *save_hwm = NULL;
3135  uschar classbits[32];  uschar classbits[32];
3136    
3137  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we
3138  must not do this for other options (e.g. PCRE_EXTENDED) because they may change  must not do this for other options (e.g. PCRE_EXTENDED) because they may change
3139  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3140    
3141  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 3086  uschar *class_utf8data_base; Line 3146  uschar *class_utf8data_base;
3146  uschar utf8_char[6];  uschar utf8_char[6];
3147  #else  #else
3148  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
 uschar *utf8_char = NULL;  
3149  #endif  #endif
3150    
3151  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3137  for (;; ptr++) Line 3196  for (;; ptr++)
3196    int subfirstbyte;    int subfirstbyte;
3197    int terminator;    int terminator;
3198    int mclength;    int mclength;
3199      int tempbracount;
3200    uschar mcbuffer[8];    uschar mcbuffer[8];
3201    
3202    /* Get next byte in the pattern */    /* Get next byte in the pattern */
# Line 4250  for (;; ptr++) Line 4310  for (;; ptr++)
4310        ptr++;        ptr++;
4311        }        }
4312      else repeat_type = greedy_default;      else repeat_type = greedy_default;
4313    
4314      /* If previous was a recursion call, wrap it in atomic brackets so that      /* If previous was a recursion call, wrap it in atomic brackets so that
4315      previous becomes the atomic group. All recursions were so wrapped in the      previous becomes the atomic group. All recursions were so wrapped in the
4316      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
4317      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,
4318      but for the moment we rely on the code for repeating groups. */      but for the moment we rely on the code for repeating groups. */
4319    
4320      if (*previous == OP_RECURSE)      if (*previous == OP_RECURSE)
4321        {        {
4322        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);
# Line 4266  for (;; ptr++) Line 4326  for (;; ptr++)
4326        PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE);        PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE);
4327        code += 2 + 2 * LINK_SIZE;        code += 2 + 2 * LINK_SIZE;
4328        length_prevgroup = 3 + 3*LINK_SIZE;        length_prevgroup = 3 + 3*LINK_SIZE;
4329    
4330        /* When actually compiling, we need to check whether this was a forward        /* When actually compiling, we need to check whether this was a forward
4331        reference, and if so, adjust the offset. */        reference, and if so, adjust the offset. */
4332    
4333        if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE)        if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE)
4334          {          {
4335          int offset = GET(cd->hwm, -LINK_SIZE);          int offset = GET(cd->hwm, -LINK_SIZE);
4336          if (offset == previous + 1 - cd->start_code)          if (offset == previous + 1 - cd->start_code)
4337            PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE);            PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE);
4338          }          }
4339        }        }
4340    
4341      /* Now handle repetition for the different types of item. */      /* Now handle repetition for the different types of item. */
4342    
4343      /* If previous was a character match, abolish the item and generate a      /* If previous was a character match, abolish the item and generate a
# Line 4575  for (;; ptr++) Line 4635  for (;; ptr++)
4635      opcodes such as BRA and CBRA, as this is the place where they get converted      opcodes such as BRA and CBRA, as this is the place where they get converted
4636      into the more special varieties such as BRAPOS and SBRA. A test for >=      into the more special varieties such as BRAPOS and SBRA. A test for >=
4637      OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK,      OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK,
4638      ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow      ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow
4639      repetition of assertions, but now it does, for Perl compatibility. */      repetition of assertions, but now it does, for Perl compatibility. */
4640    
4641      else if (*previous >= OP_ASSERT && *previous <= OP_COND)      else if (*previous >= OP_ASSERT && *previous <= OP_COND)
4642        {        {
4643        register int i;        register int i;
4644        int len = (int)(code - previous);        int len = (int)(code - previous);
4645        uschar *bralink = NULL;        uschar *bralink = NULL;
4646        uschar *brazeroptr = NULL;        uschar *brazeroptr = NULL;
4647    
4648        /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so        /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so
4649        we just ignore the repeat. */        we just ignore the repeat. */
4650    
4651        if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF)        if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF)
4652          goto END_REPEAT;          goto END_REPEAT;
4653    
4654        /* There is no sense in actually repeating assertions. The only potential        /* There is no sense in actually repeating assertions. The only potential
4655        use of repetition is in cases when the assertion is optional. Therefore,        use of repetition is in cases when the assertion is optional. Therefore,
4656        if the minimum is greater than zero, just ignore the repeat. If the        if the minimum is greater than zero, just ignore the repeat. If the
4657        maximum is not not zero or one, set it to 1. */        maximum is not not zero or one, set it to 1. */
4658    
4659        if (*previous < OP_ONCE)    /* Assertion */        if (*previous < OP_ONCE)    /* Assertion */
4660          {          {
4661          if (repeat_min > 0) goto END_REPEAT;          if (repeat_min > 0) goto END_REPEAT;
4662          if (repeat_max < 0 || repeat_max > 1) repeat_max = 1;          if (repeat_max < 0 || repeat_max > 1) repeat_max = 1;
4663          }          }
4664    
4665        /* The case of a zero minimum is special because of the need to stick        /* The case of a zero minimum is special because of the need to stick
4666        OP_BRAZERO in front of it, and because the group appears once in the        OP_BRAZERO in front of it, and because the group appears once in the
# Line 4834  for (;; ptr++) Line 4894  for (;; ptr++)
4894          {          {
4895          uschar *ketcode = code - 1 - LINK_SIZE;          uschar *ketcode = code - 1 - LINK_SIZE;
4896          uschar *bracode = ketcode - GET(ketcode, 1);          uschar *bracode = ketcode - GET(ketcode, 1);
4897    
4898          if (*bracode == OP_ONCE && possessive_quantifier) *bracode = OP_BRA;          if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) &&
4899          if (*bracode == OP_ONCE)              possessive_quantifier) *bracode = OP_BRA;
4900    
4901            if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC)
4902            *ketcode = OP_KETRMAX + repeat_type;            *ketcode = OP_KETRMAX + repeat_type;
4903          else          else
4904            {            {
# Line 4887  for (;; ptr++) Line 4949  for (;; ptr++)
4949      there are special alternative opcodes for this case. For anything else, we      there are special alternative opcodes for this case. For anything else, we
4950      wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+'      wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+'
4951      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
4952      special opcodes can optimize it.      special opcodes can optimize it.
4953    
4954      Possessively repeated subpatterns have already been handled in the code      Possessively repeated subpatterns have already been handled in the code
4955      just above, so possessive_quantifier is always FALSE for them at this      just above, so possessive_quantifier is always FALSE for them at this
4956      stage.      stage.
4957    
4958      Note that the repeated item starts at tempcode, not at previous, which      Note that the repeated item starts at tempcode, not at previous, which
4959      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.
4960    
# Line 4997  for (;; ptr++) Line 5059  for (;; ptr++)
5059        previous = NULL;        previous = NULL;
5060        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
5061        namelen = (int)(ptr - name);        namelen = (int)(ptr - name);
5062    
5063        /* It appears that Perl allows any characters whatsoever, other than        /* It appears that Perl allows any characters whatsoever, other than
5064        a closing parenthesis, to appear in arguments, so we no longer insist on        a closing parenthesis, to appear in arguments, so we no longer insist on
5065        letters, digits, and underscores. */        letters, digits, and underscores. */
5066    
5067        if (*ptr == CHAR_COLON)        if (*ptr == CHAR_COLON)
5068          {          {
# Line 5008  for (;; ptr++) Line 5070  for (;; ptr++)
5070          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
5071          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5072          }          }
5073    
5074        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
5075          {          {
5076          *errorcodeptr = ERR60;          *errorcodeptr = ERR60;
# Line 5022  for (;; ptr++) Line 5084  for (;; ptr++)
5084          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5085              strncmp((char *)name, vn, namelen) == 0)              strncmp((char *)name, vn, namelen) == 0)
5086            {            {
5087            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5088            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5089    
5090            if (verbs[i].op == OP_ACCEPT)            if (verbs[i].op == OP_ACCEPT)
# Line 5032  for (;; ptr++) Line 5094  for (;; ptr++)
5094                {                {
5095                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5096                goto FAILED;                goto FAILED;
5097                }                }
5098              cd->had_accept = TRUE;              cd->had_accept = TRUE;
5099              for (oc = cd->open_caps; oc != NULL; oc = oc->next)              for (oc = cd->open_caps; oc != NULL; oc = oc->next)
5100                {                {
# Line 5040  for (;; ptr++) Line 5102  for (;; ptr++)
5102                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5103                }                }
5104              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5105    
5106                /* Do not set firstbyte after *ACCEPT */
5107                if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
5108              }              }
5109    
5110            /* Handle other cases with/without an argument */            /* Handle other cases with/without an argument */
# Line 5052  for (;; ptr++) Line 5117  for (;; ptr++)
5117                goto FAILED;                goto FAILED;
5118                }                }
5119              *code = verbs[i].op;              *code = verbs[i].op;
5120              if (*code++ == OP_THEN)              if (*code++ == OP_THEN) cd->external_flags |= PCRE_HASTHEN;
               {  
               PUT(code, 0, code - bcptr->current_branch - 1);  
               code += LINK_SIZE;  
               }  
5121              }              }
5122    
5123            else            else
# Line 5067  for (;; ptr++) Line 5128  for (;; ptr++)
5128                goto FAILED;                goto FAILED;
5129                }                }
5130              *code = verbs[i].op_arg;              *code = verbs[i].op_arg;
5131              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;  
               }  
5132              *code++ = arglen;              *code++ = arglen;
5133              memcpy(code, arg, arglen);              memcpy(code, arg, arglen);
5134              code += arglen;              code += arglen;
# Line 5333  for (;; ptr++) Line 5390  for (;; ptr++)
5390          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
5391          case CHAR_EQUALS_SIGN:                 /* Positive lookahead */          case CHAR_EQUALS_SIGN:                 /* Positive lookahead */
5392          bravalue = OP_ASSERT;          bravalue = OP_ASSERT;
5393          cd->assert_depth += 1;          cd->assert_depth += 1;
5394          ptr++;          ptr++;
5395          break;          break;
5396    
# Line 5348  for (;; ptr++) Line 5405  for (;; ptr++)
5405            continue;            continue;
5406            }            }
5407          bravalue = OP_ASSERT_NOT;          bravalue = OP_ASSERT_NOT;
5408          cd->assert_depth += 1;          cd->assert_depth += 1;
5409          break;          break;
5410    
5411    
# Line 5358  for (;; ptr++) Line 5415  for (;; ptr++)
5415            {            {
5416            case CHAR_EQUALS_SIGN:               /* Positive lookbehind */            case CHAR_EQUALS_SIGN:               /* Positive lookbehind */
5417            bravalue = OP_ASSERTBACK;            bravalue = OP_ASSERTBACK;
5418            cd->assert_depth += 1;            cd->assert_depth += 1;
5419            ptr += 2;            ptr += 2;
5420            break;            break;
5421    
5422            case CHAR_EXCLAMATION_MARK:          /* Negative lookbehind */            case CHAR_EXCLAMATION_MARK:          /* Negative lookbehind */
5423            bravalue = OP_ASSERTBACK_NOT;            bravalue = OP_ASSERTBACK_NOT;
5424            cd->assert_depth += 1;            cd->assert_depth += 1;
5425            ptr += 2;            ptr += 2;
5426            break;            break;
5427    
# Line 5608  for (;; ptr++) Line 5665  for (;; ptr++)
5665    
5666            temp = cd->end_pattern;            temp = cd->end_pattern;
5667            cd->end_pattern = ptr;            cd->end_pattern = ptr;
5668            recno = find_parens(cd, name, namelen,            recno = find_parens(cd, name, namelen,
5669              (options & PCRE_EXTENDED) != 0, utf8);              (options & PCRE_EXTENDED) != 0, utf8);
5670            cd->end_pattern = temp;            cd->end_pattern = temp;
5671            if (recno < 0) recno = 0;    /* Forward ref; set dummy number */            if (recno < 0) recno = 0;    /* Forward ref; set dummy number */
# Line 5764  for (;; ptr++) Line 5821  for (;; ptr++)
5821              /* If not a forward reference, and the subpattern is still open,              /* If not a forward reference, and the subpattern is still open,
5822              this is a recursive call. We check to see if this is a left              this is a recursive call. We check to see if this is a left
5823              recursion that could loop for ever, and diagnose that case. We              recursion that could loop for ever, and diagnose that case. We
5824              must not, however, do this check if we are in a conditional              must not, however, do this check if we are in a conditional
5825              subpattern because the condition might be testing for recursion in              subpattern because the condition might be testing for recursion in
5826              a pattern such as /(?(R)a+|(?R)b)/, which is perfectly valid.              a pattern such as /(?(R)a+|(?R)b)/, which is perfectly valid.
5827              Forever loops are also detected at runtime, so those that occur in              Forever loops are also detected at runtime, so those that occur in
5828              conditional subpatterns will be picked up then. */              conditional subpatterns will be picked up then. */
5829    
5830              else if (GET(called, 1) == 0 && cond_depth <= 0 &&              else if (GET(called, 1) == 0 && cond_depth <= 0 &&
# Line 5779  for (;; ptr++) Line 5836  for (;; ptr++)
5836              }              }
5837    
5838            /* Insert the recursion/subroutine item. */            /* Insert the recursion/subroutine item. */
5839    
5840            *code = OP_RECURSE;            *code = OP_RECURSE;
5841            PUT(code, 1, (int)(called - cd->start_code));            PUT(code, 1, (int)(called - cd->start_code));
5842            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
# Line 5903  for (;; ptr++) Line 5960  for (;; ptr++)
5960      be able to pass its address because some compilers complain otherwise. */      be able to pass its address because some compilers complain otherwise. */
5961    
5962      previous = code;                      /* For handling repetition */      previous = code;                      /* For handling repetition */
5963      *code = bravalue;      *code = bravalue;
5964      tempcode = code;      tempcode = code;
5965      tempreqvary = cd->req_varyopt;        /* Save value before bracket */      tempreqvary = cd->req_varyopt;        /* Save value before bracket */
5966        tempbracount = cd->bracount;          /* Save value before bracket */
5967      length_prevgroup = 0;                 /* Initialize for pre-compile phase */      length_prevgroup = 0;                 /* Initialize for pre-compile phase */
5968    
5969      if (!compile_regex(      if (!compile_regex(
5970           newoptions,                      /* The complete new option state */           newoptions,                      /* The complete new option state */
5971           &tempcode,                       /* Where to put code (updated) */           &tempcode,                       /* Where to put code (updated) */
5972           &ptr,                            /* Input pointer (updated) */           &ptr,                            /* Input pointer (updated) */
# Line 5917  for (;; ptr++) Line 5975  for (;; ptr++)
5975            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
5976           reset_bracount,                  /* True if (?| group */           reset_bracount,                  /* True if (?| group */
5977           skipbytes,                       /* Skip over bracket number */           skipbytes,                       /* Skip over bracket number */
5978           cond_depth +           cond_depth +
5979             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */
5980           &subfirstbyte,                   /* For possible first char */           &subfirstbyte,                   /* For possible first char */
5981           &subreqbyte,                     /* For possible last char */           &subreqbyte,                     /* For possible last char */
# Line 5927  for (;; ptr++) Line 5985  for (;; ptr++)
5985             &length_prevgroup              /* Pre-compile phase */             &length_prevgroup              /* Pre-compile phase */
5986           ))           ))
5987        goto FAILED;        goto FAILED;
5988    
5989        /* If this was an atomic group and there are no capturing groups within it,
5990        generate OP_ONCE_NC instead of OP_ONCE. */
5991    
5992        if (bravalue == OP_ONCE && cd->bracount <= tempbracount)
5993          *code = OP_ONCE_NC;
5994    
5995      if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT)      if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT)
5996        cd->assert_depth -= 1;        cd->assert_depth -= 1;
5997    
5998      /* 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
5999      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.
6000      and any option resetting that may follow it. The pattern pointer (ptr)      The pattern pointer (ptr) is on the bracket.
     is on the bracket. */  
6001    
6002      /* 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
6003      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
6004      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
6005      not be available. */      not be available. */
# Line 6177  for (;; ptr++) Line 6240  for (;; ptr++)
6240            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))
6241            {            {
6242            *errorcodeptr = ERR69;            *errorcodeptr = ERR69;
6243            break;            break;
6244            }            }
6245          is_recurse = FALSE;          is_recurse = FALSE;
6246          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?
6247            CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)?            CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)?
6248            CHAR_APOSTROPHE : CHAR_RIGHT_CURLY_BRACKET;            CHAR_APOSTROPHE : CHAR_RIGHT_CURLY_BRACKET;
6249          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
6250          }          }
6251    
6252        /* Back references are handled specially; must disable firstbyte if        /* Back references are handled specially; must disable firstbyte if
6253        not set to cope with cases like (?=(\w+))\1: which would otherwise set        not set to cope with cases like (?=(\w+))\1: which would otherwise set
# Line 6335  for (;; ptr++) Line 6398  for (;; ptr++)
6398        else firstbyte = reqbyte = REQ_NONE;        else firstbyte = reqbyte = REQ_NONE;
6399        }        }
6400    
6401      /* 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
6402      1 or the matching is caseful. */      1 or the matching is caseful. */
6403    
6404      else      else
# Line 6382  Arguments: Line 6445  Arguments:
6445    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
6446    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
6447    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
6448    cond_depth     depth of nesting for conditional subpatterns    cond_depth     depth of nesting for conditional subpatterns
6449    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
6450    reqbyteptr     place to put the last required character, or a negative number    reqbyteptr     place to put the last required character, or a negative number
6451    bcptr          pointer to the chain of currently open branches    bcptr          pointer to the chain of currently open branches
# Line 6396  Returns:         TRUE on success Line 6459  Returns:         TRUE on success
6459  static BOOL  static BOOL
6460  compile_regex(int options, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, uschar **codeptr, const uschar **ptrptr,
6461    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
6462    int cond_depth, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    int cond_depth, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,
6463    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
6464  {  {
6465  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
# Line 6434  pre-compile phase to find out whether an Line 6497  pre-compile phase to find out whether an
6497    
6498  /* If this is a capturing subpattern, add to the chain of open capturing items  /* If this is a capturing subpattern, add to the chain of open capturing items
6499  so that we can detect them if (*ACCEPT) is encountered. This is also used to  so that we can detect them if (*ACCEPT) is encountered. This is also used to
6500  detect groups that contain recursive back references to themselves. Note that  detect groups that contain recursive back references to themselves. Note that
6501  only OP_CBRA need be tested here; changing this opcode to one of its variants,  only OP_CBRA need be tested here; changing this opcode to one of its variants,
6502  e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */  e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */
6503    
6504  if (*code == OP_CBRA)  if (*code == OP_CBRA)
# Line 6476  for (;;) Line 6539  for (;;)
6539    into the length. */    into the length. */
6540    
6541    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte,    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte,
6542          &branchreqbyte, &bc, cond_depth, cd,          &branchreqbyte, &bc, cond_depth, cd,
6543          (lengthptr == NULL)? NULL : &length))          (lengthptr == NULL)? NULL : &length))
6544      {      {
6545      *ptrptr = ptr;      *ptrptr = ptr;
# Line 6727  do { Line 6790  do {
6790    
6791     /* Other brackets */     /* Other brackets */
6792    
6793     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_COND)     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC ||
6794                op == OP_COND)
6795       {       {
6796       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, bracket_map, backref_map)) return FALSE;
6797       }       }
# Line 6831  do { Line 6895  do {
6895    
6896     /* Other brackets */     /* Other brackets */
6897    
6898     else if (op == OP_ASSERT || op == OP_ONCE)     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_ONCE_NC)
6899       {       {
6900       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;       if (!is_startline(scode, bracket_map, backref_map)) return FALSE;
6901       }       }
# Line 6901  do { Line 6965  do {
6965       case OP_SCBRAPOS:       case OP_SCBRAPOS:
6966       case OP_ASSERT:       case OP_ASSERT:
6967       case OP_ONCE:       case OP_ONCE:
6968         case OP_ONCE_NC:
6969       case OP_COND:       case OP_COND:
6970       if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0)       if ((d = find_firstassertedchar(scode, op == OP_ASSERT)) < 0)
6971         return -1;         return -1;
# Line 6910  do { Line 6975  do {
6975       case OP_EXACT:       case OP_EXACT:
6976       scode += 2;       scode += 2;
6977       /* Fall through */       /* Fall through */
6978    
6979       case OP_CHAR:       case OP_CHAR:
6980       case OP_PLUS:       case OP_PLUS:
6981       case OP_MINPLUS:       case OP_MINPLUS:
# Line 6923  do { Line 6988  do {
6988       case OP_EXACTI:       case OP_EXACTI:
6989       scode += 2;       scode += 2;
6990       /* Fall through */       /* Fall through */
6991    
6992       case OP_CHARI:       case OP_CHARI:
6993       case OP_PLUSI:       case OP_PLUSI:
6994       case OP_MINPLUSI:       case OP_MINPLUSI:
# Line 7083  utf8 = (options & PCRE_UTF8) != 0; Line 7148  utf8 = (options & PCRE_UTF8) != 0;
7148    
7149  /* Can't support UTF8 unless PCRE has been compiled to include the code. The  /* Can't support UTF8 unless PCRE has been compiled to include the code. The
7150  return of an error code from _pcre_valid_utf8() is a new feature, introduced in  return of an error code from _pcre_valid_utf8() is a new feature, introduced in
7151  release 8.13. It is passed back from pcre_[dfa_]exec(), but at the moment is  release 8.13. It is passed back from pcre_[dfa_]exec(), but at the moment is
7152  not used here. */  not used here. */
7153    
7154  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 7116  if ((options & PCRE_UCP) != 0) Line 7181  if ((options & PCRE_UCP) != 0)
7181  if ((options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) ==  if ((options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) ==
7182       (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))       (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
7183    {    {
7184    errorcode = ERR56;    errorcode = ERR56;
7185    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
7186    }    }
7187    
# Line 7202  outside can help speed up starting point Line 7267  outside can help speed up starting point
7267  ptr += skipatstart;  ptr += skipatstart;
7268  code = cworkspace;  code = cworkspace;
7269  *code = OP_BRA;  *code = OP_BRA;
7270  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,
7271    FALSE, 0, 0, &firstbyte, &reqbyte, NULL, cd, &length);    FALSE, 0, 0, &firstbyte, &reqbyte, NULL, cd, &length);
7272  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
7273    
# Line 7282  re->top_bracket = cd->bracount; Line 7347  re->top_bracket = cd->bracount;
7347  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
7348  re->flags = cd->external_flags;  re->flags = cd->external_flags;
7349    
7350  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */  if (cd->had_accept) reqbyte = REQ_NONE;   /* Must disable after (*ACCEPT) */
7351    
7352  /* 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. */
7353    

Legend:
Removed from v.643  
changed lines
  Added in v.746

  ViewVC Help
Powered by ViewVC 1.1.5