/[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 642 by ph10, Thu Jul 28 18:59:40 2011 UTC revision 701 by ph10, Tue Sep 20 11:30:56 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 681  else Line 681  else
681      *errorcodeptr = ERR37;      *errorcodeptr = ERR37;
682      break;      break;
683    
684      /* 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
685      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:
686    
687      (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 921  else Line 921  else
921    }    }
922    
923  /* 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
924  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
925  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
926    
927  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
# Line 1982  for (code = first_significant_code(code Line 1982  for (code = first_significant_code(code
1982    /* For a recursion/subroutine call, if its end has been reached, which    /* For a recursion/subroutine call, if its end has been reached, which
1983    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
1984    forward reference subroutine call, we can't. To detect forward reference    forward reference subroutine call, we can't. To detect forward reference
1985    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
1986    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
1987    measures the size of the compiled pattern. */    measures the size of the compiled pattern. */
1988    
1989    if (c == OP_RECURSE)    if (c == OP_RECURSE)
1990      {      {
1991      const uschar *scode;      const uschar *scode;
1992      BOOL empty_branch;      BOOL empty_branch;
1993    
1994      /* Test for forward reference */      /* Test for forward reference */
1995    
1996      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
1997        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;
1998    
1999      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2000    
2001      empty_branch = FALSE;      empty_branch = FALSE;
2002      scode = cd->start_code + GET(code, 1);      scode = cd->start_code + GET(code, 1);
2003      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2004    
2005      /* Completed backwards reference */      /* Completed backwards reference */
2006    
2007      do      do
2008        {        {
2009        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 2014  for (code = first_significant_code(code
2014        scode += GET(scode, 1);        scode += GET(scode, 1);
2015        }        }
2016      while (*scode == OP_ALT);      while (*scode == OP_ALT);
2017    
2018      if (!empty_branch) return FALSE;  /* All branches are non-empty */      if (!empty_branch) return FALSE;  /* All branches are non-empty */
2019      continue;      continue;
2020      }      }
# Line 2240  return TRUE; Line 2240  return TRUE;
2240  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
2241  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,
2242  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.
2243  This function is called only during the real compile, not during the  This function is called only during the real compile, not during the
2244  pre-compile.  pre-compile.
2245    
2246  Arguments:  Arguments:
# Line 2295  I think. Line 2295  I think.
2295  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.
2296  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
2297  external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or  external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or
2298  a digit. Also, unescaped square brackets may also appear as part of class  a digit.
2299  names. For example, [:a[:abc]b:] gives unknown class "[:abc]b:]"in Perl.  
2300    In Perl, unescaped square brackets may also appear as part of class names. For
2301    example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for
2302    [:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not
2303    seem right at all. PCRE does not allow closing square brackets in POSIX class
2304    names.
2305    
2306  Arguments:  Arguments:
2307    ptr      pointer to the initial [    ptr      pointer to the initial [
# Line 2312  int terminator;          /* Don't combin Line 2317  int terminator;          /* Don't combin
2317  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2318  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != 0; ptr++)
2319    {    {
2320    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2321      ptr++;      ptr++;
2322      else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
2323    else    else
2324      {      {
2325      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 2331  for (++ptr; *ptr != 0; ptr++)
2331           (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||           (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2332            ptr[1] == CHAR_EQUALS_SIGN) &&            ptr[1] == CHAR_EQUALS_SIGN) &&
2333          check_posix_syntax(ptr, endptr))          check_posix_syntax(ptr, endptr))
2334        return FALSE;        return FALSE;
2335      }      }
2336    }    }
2337  return FALSE;  return FALSE;
# Line 3035  Arguments: Line 3041  Arguments:
3041    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)
3042    reqbyteptr     set to the last literal character required, else < 0    reqbyteptr     set to the last literal character required, else < 0
3043    bcptr          points to current branch chain    bcptr          points to current branch chain
3044    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3045    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
3046    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
3047                   points to length accumulator during pre-compile phase                   points to length accumulator during pre-compile phase
# Line 3075  uschar *save_hwm = NULL; Line 3081  uschar *save_hwm = NULL;
3081  uschar classbits[32];  uschar classbits[32];
3082    
3083  /* 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
3084  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
3085  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3086    
3087  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 3086  uschar *class_utf8data_base; Line 3092  uschar *class_utf8data_base;
3092  uschar utf8_char[6];  uschar utf8_char[6];
3093  #else  #else
3094  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
 uschar *utf8_char = NULL;  
3095  #endif  #endif
3096    
3097  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 4250  for (;; ptr++) Line 4255  for (;; ptr++)
4255        ptr++;        ptr++;
4256        }        }
4257      else repeat_type = greedy_default;      else repeat_type = greedy_default;
4258    
4259      /* 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
4260      previous becomes the atomic group. All recursions were so wrapped in the      previous becomes the atomic group. All recursions were so wrapped in the
4261      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
4262      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,
4263      but for the moment we rely on the code for repeating groups. */      but for the moment we rely on the code for repeating groups. */
4264    
4265      if (*previous == OP_RECURSE)      if (*previous == OP_RECURSE)
4266        {        {
4267        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);
# Line 4266  for (;; ptr++) Line 4271  for (;; ptr++)
4271        PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE);        PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE);
4272        code += 2 + 2 * LINK_SIZE;        code += 2 + 2 * LINK_SIZE;
4273        length_prevgroup = 3 + 3*LINK_SIZE;        length_prevgroup = 3 + 3*LINK_SIZE;
4274    
4275        /* When actually compiling, we need to check whether this was a forward        /* When actually compiling, we need to check whether this was a forward
4276        reference, and if so, adjust the offset. */        reference, and if so, adjust the offset. */
4277    
4278        if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE)        if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE)
4279          {          {
4280          int offset = GET(cd->hwm, -LINK_SIZE);          int offset = GET(cd->hwm, -LINK_SIZE);
4281          if (offset == previous + 1 - cd->start_code)          if (offset == previous + 1 - cd->start_code)
4282            PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE);            PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE);
4283          }          }
4284        }        }
4285    
4286      /* Now handle repetition for the different types of item. */      /* Now handle repetition for the different types of item. */
4287    
4288      /* 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 4580  for (;; ptr++)
4580      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
4581      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 >=
4582      OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK,      OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK,
4583      ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow      ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow
4584      repetition of assertions, but now it does, for Perl compatibility. */      repetition of assertions, but now it does, for Perl compatibility. */
4585    
4586      else if (*previous >= OP_ASSERT && *previous <= OP_COND)      else if (*previous >= OP_ASSERT && *previous <= OP_COND)
4587        {        {
4588        register int i;        register int i;
4589        int len = (int)(code - previous);        int len = (int)(code - previous);
4590        uschar *bralink = NULL;        uschar *bralink = NULL;
4591        uschar *brazeroptr = NULL;        uschar *brazeroptr = NULL;
4592    
4593        /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so        /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so
4594        we just ignore the repeat. */        we just ignore the repeat. */
4595    
4596        if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF)        if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF)
4597          goto END_REPEAT;          goto END_REPEAT;
4598    
4599        /* There is no sense in actually repeating assertions. The only potential        /* There is no sense in actually repeating assertions. The only potential
4600        use of repetition is in cases when the assertion is optional. Therefore,        use of repetition is in cases when the assertion is optional. Therefore,
4601        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
4602        maximum is not not zero or one, set it to 1. */        maximum is not not zero or one, set it to 1. */
4603    
4604        if (*previous < OP_ONCE)    /* Assertion */        if (*previous < OP_ONCE)    /* Assertion */
4605          {          {
4606          if (repeat_min > 0) goto END_REPEAT;          if (repeat_min > 0) goto END_REPEAT;
4607          if (repeat_max < 0 || repeat_max > 1) repeat_max = 1;          if (repeat_max < 0 || repeat_max > 1) repeat_max = 1;
4608          }          }
4609    
4610        /* 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
4611        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 4839  for (;; ptr++)
4839          {          {
4840          uschar *ketcode = code - 1 - LINK_SIZE;          uschar *ketcode = code - 1 - LINK_SIZE;
4841          uschar *bracode = ketcode - GET(ketcode, 1);          uschar *bracode = ketcode - GET(ketcode, 1);
4842    
4843          if (*bracode == OP_ONCE && possessive_quantifier) *bracode = OP_BRA;          if (*bracode == OP_ONCE && possessive_quantifier) *bracode = OP_BRA;
4844          if (*bracode == OP_ONCE)          if (*bracode == OP_ONCE)
4845            *ketcode = OP_KETRMAX + repeat_type;            *ketcode = OP_KETRMAX + repeat_type;
4846          else          else
4847            {            {
# Line 4887  for (;; ptr++) Line 4892  for (;; ptr++)
4892      there are special alternative opcodes for this case. For anything else, we      there are special alternative opcodes for this case. For anything else, we
4893      wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+'      wrap the entire repeated item inside OP_ONCE brackets. Logically, the '+'
4894      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
4895      special opcodes can optimize it.      special opcodes can optimize it.
4896    
4897      Possessively repeated subpatterns have already been handled in the code      Possessively repeated subpatterns have already been handled in the code
4898      just above, so possessive_quantifier is always FALSE for them at this      just above, so possessive_quantifier is always FALSE for them at this
4899      stage.      stage.
4900    
4901      Note that the repeated item starts at tempcode, not at previous, which      Note that the repeated item starts at tempcode, not at previous, which
4902      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.
4903    
# Line 4998  for (;; ptr++) Line 5003  for (;; ptr++)
5003        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
5004        namelen = (int)(ptr - name);        namelen = (int)(ptr - name);
5005    
5006          /* It appears that Perl allows any characters whatsoever, other than
5007          a closing parenthesis, to appear in arguments, so we no longer insist on
5008          letters, digits, and underscores. */
5009    
5010        if (*ptr == CHAR_COLON)        if (*ptr == CHAR_COLON)
5011          {          {
5012          arg = ++ptr;          arg = ++ptr;
5013          while ((cd->ctypes[*ptr] & (ctype_letter|ctype_digit)) != 0          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
           || *ptr == '_') ptr++;  
5014          arglen = (int)(ptr - arg);          arglen = (int)(ptr - arg);
5015          }          }
5016    
# Line 5019  for (;; ptr++) Line 5027  for (;; ptr++)
5027          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5028              strncmp((char *)name, vn, namelen) == 0)              strncmp((char *)name, vn, namelen) == 0)
5029            {            {
5030            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5031            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
5032    
5033            if (verbs[i].op == OP_ACCEPT)            if (verbs[i].op == OP_ACCEPT)
# Line 5029  for (;; ptr++) Line 5037  for (;; ptr++)
5037                {                {
5038                *errorcodeptr = ERR59;                *errorcodeptr = ERR59;
5039                goto FAILED;                goto FAILED;
5040                }                }
5041              cd->had_accept = TRUE;              cd->had_accept = TRUE;
5042              for (oc = cd->open_caps; oc != NULL; oc = oc->next)              for (oc = cd->open_caps; oc != NULL; oc = oc->next)
5043                {                {
# Line 5037  for (;; ptr++) Line 5045  for (;; ptr++)
5045                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
5046                }                }
5047              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5048    
5049                /* Do not set firstbyte after *ACCEPT */
5050                if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
5051              }              }
5052    
5053            /* Handle other cases with/without an argument */            /* Handle other cases with/without an argument */
# Line 5330  for (;; ptr++) Line 5341  for (;; ptr++)
5341          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
5342          case CHAR_EQUALS_SIGN:                 /* Positive lookahead */          case CHAR_EQUALS_SIGN:                 /* Positive lookahead */
5343          bravalue = OP_ASSERT;          bravalue = OP_ASSERT;
5344          cd->assert_depth += 1;          cd->assert_depth += 1;
5345          ptr++;          ptr++;
5346          break;          break;
5347    
# Line 5345  for (;; ptr++) Line 5356  for (;; ptr++)
5356            continue;            continue;
5357            }            }
5358          bravalue = OP_ASSERT_NOT;          bravalue = OP_ASSERT_NOT;
5359          cd->assert_depth += 1;          cd->assert_depth += 1;
5360          break;          break;
5361    
5362    
# Line 5355  for (;; ptr++) Line 5366  for (;; ptr++)
5366            {            {
5367            case CHAR_EQUALS_SIGN:               /* Positive lookbehind */            case CHAR_EQUALS_SIGN:               /* Positive lookbehind */
5368            bravalue = OP_ASSERTBACK;            bravalue = OP_ASSERTBACK;
5369            cd->assert_depth += 1;            cd->assert_depth += 1;
5370            ptr += 2;            ptr += 2;
5371            break;            break;
5372    
5373            case CHAR_EXCLAMATION_MARK:          /* Negative lookbehind */            case CHAR_EXCLAMATION_MARK:          /* Negative lookbehind */
5374            bravalue = OP_ASSERTBACK_NOT;            bravalue = OP_ASSERTBACK_NOT;
5375            cd->assert_depth += 1;            cd->assert_depth += 1;
5376            ptr += 2;            ptr += 2;
5377            break;            break;
5378    
# Line 5605  for (;; ptr++) Line 5616  for (;; ptr++)
5616    
5617            temp = cd->end_pattern;            temp = cd->end_pattern;
5618            cd->end_pattern = ptr;            cd->end_pattern = ptr;
5619            recno = find_parens(cd, name, namelen,            recno = find_parens(cd, name, namelen,
5620              (options & PCRE_EXTENDED) != 0, utf8);              (options & PCRE_EXTENDED) != 0, utf8);
5621            cd->end_pattern = temp;            cd->end_pattern = temp;
5622            if (recno < 0) recno = 0;    /* Forward ref; set dummy number */            if (recno < 0) recno = 0;    /* Forward ref; set dummy number */
# Line 5761  for (;; ptr++) Line 5772  for (;; ptr++)
5772              /* If not a forward reference, and the subpattern is still open,              /* If not a forward reference, and the subpattern is still open,
5773              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
5774              recursion that could loop for ever, and diagnose that case. We              recursion that could loop for ever, and diagnose that case. We
5775              must not, however, do this check if we are in a conditional              must not, however, do this check if we are in a conditional
5776              subpattern because the condition might be testing for recursion in              subpattern because the condition might be testing for recursion in
5777              a pattern such as /(?(R)a+|(?R)b)/, which is perfectly valid.              a pattern such as /(?(R)a+|(?R)b)/, which is perfectly valid.
5778              Forever loops are also detected at runtime, so those that occur in              Forever loops are also detected at runtime, so those that occur in
5779              conditional subpatterns will be picked up then. */              conditional subpatterns will be picked up then. */
5780    
5781              else if (GET(called, 1) == 0 && cond_depth <= 0 &&              else if (GET(called, 1) == 0 && cond_depth <= 0 &&
# Line 5776  for (;; ptr++) Line 5787  for (;; ptr++)
5787              }              }
5788    
5789            /* Insert the recursion/subroutine item. */            /* Insert the recursion/subroutine item. */
5790    
5791            *code = OP_RECURSE;            *code = OP_RECURSE;
5792            PUT(code, 1, (int)(called - cd->start_code));            PUT(code, 1, (int)(called - cd->start_code));
5793            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
# Line 5900  for (;; ptr++) Line 5911  for (;; ptr++)
5911      be able to pass its address because some compilers complain otherwise. */      be able to pass its address because some compilers complain otherwise. */
5912    
5913      previous = code;                      /* For handling repetition */      previous = code;                      /* For handling repetition */
5914      *code = bravalue;      *code = bravalue;
5915      tempcode = code;      tempcode = code;
5916      tempreqvary = cd->req_varyopt;        /* Save value before bracket */      tempreqvary = cd->req_varyopt;        /* Save value before bracket */
5917      length_prevgroup = 0;                 /* Initialize for pre-compile phase */      length_prevgroup = 0;                 /* Initialize for pre-compile phase */
5918    
5919      if (!compile_regex(      if (!compile_regex(
5920           newoptions,                      /* The complete new option state */           newoptions,                      /* The complete new option state */
5921           &tempcode,                       /* Where to put code (updated) */           &tempcode,                       /* Where to put code (updated) */
5922           &ptr,                            /* Input pointer (updated) */           &ptr,                            /* Input pointer (updated) */
# Line 5914  for (;; ptr++) Line 5925  for (;; ptr++)
5925            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
5926           reset_bracount,                  /* True if (?| group */           reset_bracount,                  /* True if (?| group */
5927           skipbytes,                       /* Skip over bracket number */           skipbytes,                       /* Skip over bracket number */
5928           cond_depth +           cond_depth +
5929             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */
5930           &subfirstbyte,                   /* For possible first char */           &subfirstbyte,                   /* For possible first char */
5931           &subreqbyte,                     /* For possible last char */           &subreqbyte,                     /* For possible last char */
# Line 5924  for (;; ptr++) Line 5935  for (;; ptr++)
5935             &length_prevgroup              /* Pre-compile phase */             &length_prevgroup              /* Pre-compile phase */
5936           ))           ))
5937        goto FAILED;        goto FAILED;
5938    
5939      if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT)      if (bravalue >= OP_ASSERT && bravalue <= OP_ASSERTBACK_NOT)
5940        cd->assert_depth -= 1;        cd->assert_depth -= 1;
5941    
5942      /* 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
5943      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
# Line 6174  for (;; ptr++) Line 6185  for (;; ptr++)
6185            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))            ptr[1] != CHAR_APOSTROPHE && ptr[1] != CHAR_LEFT_CURLY_BRACKET))
6186            {            {
6187            *errorcodeptr = ERR69;            *errorcodeptr = ERR69;
6188            break;            break;
6189            }            }
6190          is_recurse = FALSE;          is_recurse = FALSE;
6191          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?
6192            CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)?            CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)?
6193            CHAR_APOSTROPHE : CHAR_RIGHT_CURLY_BRACKET;            CHAR_APOSTROPHE : CHAR_RIGHT_CURLY_BRACKET;
6194          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
6195          }          }
6196    
6197        /* Back references are handled specially; must disable firstbyte if        /* Back references are handled specially; must disable firstbyte if
6198        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 6315  for (;; ptr++) Line 6326  for (;; ptr++)
6326      byte, set it from this character, but revert to none on a zero repeat.      byte, set it from this character, but revert to none on a zero repeat.
6327      Otherwise, leave the firstbyte value alone, and don't change it on a zero      Otherwise, leave the firstbyte value alone, and don't change it on a zero
6328      repeat. */      repeat. */
6329    
6330      if (firstbyte == REQ_UNSET)      if (firstbyte == REQ_UNSET)
6331        {        {
6332        zerofirstbyte = REQ_NONE;        zerofirstbyte = REQ_NONE;
# Line 6332  for (;; ptr++) Line 6343  for (;; ptr++)
6343        else firstbyte = reqbyte = REQ_NONE;        else firstbyte = reqbyte = REQ_NONE;
6344        }        }
6345    
6346      /* 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
6347      1 or the matching is caseful. */      1 or the matching is caseful. */
6348    
6349      else      else
# Line 6379  Arguments: Line 6390  Arguments:
6390    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
6391    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
6392    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
6393    cond_depth     depth of nesting for conditional subpatterns    cond_depth     depth of nesting for conditional subpatterns
6394    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
6395    reqbyteptr     place to put the last required character, or a negative number    reqbyteptr     place to put the last required character, or a negative number
6396    bcptr          pointer to the chain of currently open branches    bcptr          pointer to the chain of currently open branches
# Line 6393  Returns:         TRUE on success Line 6404  Returns:         TRUE on success
6404  static BOOL  static BOOL
6405  compile_regex(int options, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, uschar **codeptr, const uschar **ptrptr,
6406    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
6407    int cond_depth, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    int cond_depth, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,
6408    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
6409  {  {
6410  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
# Line 6431  pre-compile phase to find out whether an Line 6442  pre-compile phase to find out whether an
6442    
6443  /* 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
6444  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
6445  detect groups that contain recursive back references to themselves. Note that  detect groups that contain recursive back references to themselves. Note that
6446  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,
6447  e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */  e.g. OP_SCBRAPOS, happens later, after the group has been compiled. */
6448    
6449  if (*code == OP_CBRA)  if (*code == OP_CBRA)
# Line 6473  for (;;) Line 6484  for (;;)
6484    into the length. */    into the length. */
6485    
6486    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte,    if (!compile_branch(&options, &code, &ptr, errorcodeptr, &branchfirstbyte,
6487          &branchreqbyte, &bc, cond_depth, cd,          &branchreqbyte, &bc, cond_depth, cd,
6488          (lengthptr == NULL)? NULL : &length))          (lengthptr == NULL)? NULL : &length))
6489      {      {
6490      *ptrptr = ptr;      *ptrptr = ptr;
# Line 6907  do { Line 6918  do {
6918       case OP_EXACT:       case OP_EXACT:
6919       scode += 2;       scode += 2;
6920       /* Fall through */       /* Fall through */
6921    
6922       case OP_CHAR:       case OP_CHAR:
6923       case OP_PLUS:       case OP_PLUS:
6924       case OP_MINPLUS:       case OP_MINPLUS:
# Line 6920  do { Line 6931  do {
6931       case OP_EXACTI:       case OP_EXACTI:
6932       scode += 2;       scode += 2;
6933       /* Fall through */       /* Fall through */
6934    
6935       case OP_CHARI:       case OP_CHARI:
6936       case OP_PLUSI:       case OP_PLUSI:
6937       case OP_MINPLUSI:       case OP_MINPLUSI:
# Line 7080  utf8 = (options & PCRE_UTF8) != 0; Line 7091  utf8 = (options & PCRE_UTF8) != 0;
7091    
7092  /* 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
7093  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
7094  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
7095  not used here. */  not used here. */
7096    
7097  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 7113  if ((options & PCRE_UCP) != 0) Line 7124  if ((options & PCRE_UCP) != 0)
7124  if ((options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) ==  if ((options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) ==
7125       (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))       (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
7126    {    {
7127    errorcode = ERR56;    errorcode = ERR56;
7128    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
7129    }    }
7130    
# Line 7199  outside can help speed up starting point Line 7210  outside can help speed up starting point
7210  ptr += skipatstart;  ptr += skipatstart;
7211  code = cworkspace;  code = cworkspace;
7212  *code = OP_BRA;  *code = OP_BRA;
7213  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,  (void)compile_regex(cd->external_options, &code, &ptr, &errorcode, FALSE,
7214    FALSE, 0, 0, &firstbyte, &reqbyte, NULL, cd, &length);    FALSE, 0, 0, &firstbyte, &reqbyte, NULL, cd, &length);
7215  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
7216    
# Line 7279  re->top_bracket = cd->bracount; Line 7290  re->top_bracket = cd->bracount;
7290  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
7291  re->flags = cd->external_flags;  re->flags = cd->external_flags;
7292    
7293  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */  if (cd->had_accept) reqbyte = REQ_NONE;   /* Must disable after (*ACCEPT) */
7294    
7295  /* 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. */
7296    

Legend:
Removed from v.642  
changed lines
  Added in v.701

  ViewVC Help
Powered by ViewVC 1.1.5