/[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 286 by ph10, Mon Dec 17 14:46:11 2007 UTC revision 391 by ph10, Tue Mar 17 21:16:01 2009 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2007 University of Cambridge             Copyright (c) 1997-2009 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 97  are simple data values; negative values Line 97  are simple data values; negative values
97  on. Zero means further processing is needed (for things like \x), or the escape  on. Zero means further processing is needed (for things like \x), or the escape
98  is invalid. */  is invalid. */
99    
100  #ifndef EBCDIC  /* This is the "normal" table for ASCII systems */  #ifndef EBCDIC
101  static const short int escapes[] = {  
102       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */  /* This is the "normal" table for ASCII systems or for EBCDIC systems running
103       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */  in UTF-8 mode. */
104     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */  
105  -ESC_H,      0,      0, -ESC_K,      0,      0,      0,      0,   /* H - O */  static const short int escapes[] = {
106  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0, -ESC_V, -ESC_W,   /* P - W */       0,                       0,
107  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */       0,                       0,
108     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */       0,                       0,
109  -ESC_h,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */       0,                       0,
110  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0, -ESC_v, -ESC_w,   /* p - w */       0,                       0,
111       0,      0, -ESC_z                                            /* x - z */       CHAR_COLON,              CHAR_SEMICOLON,
112         CHAR_LESS_THAN_SIGN,     CHAR_EQUALS_SIGN,
113         CHAR_GREATER_THAN_SIGN,  CHAR_QUESTION_MARK,
114         CHAR_COMMERCIAL_AT,      -ESC_A,
115         -ESC_B,                  -ESC_C,
116         -ESC_D,                  -ESC_E,
117         0,                       -ESC_G,
118         -ESC_H,                  0,
119         0,                       -ESC_K,
120         0,                       0,
121         0,                       0,
122         -ESC_P,                  -ESC_Q,
123         -ESC_R,                  -ESC_S,
124         0,                       0,
125         -ESC_V,                  -ESC_W,
126         -ESC_X,                  0,
127         -ESC_Z,                  CHAR_LEFT_SQUARE_BRACKET,
128         CHAR_BACKSLASH,          CHAR_RIGHT_SQUARE_BRACKET,
129         CHAR_CIRCUMFLEX_ACCENT,  CHAR_UNDERSCORE,
130         CHAR_GRAVE_ACCENT,       7,
131         -ESC_b,                  0,
132         -ESC_d,                  ESC_e,
133         ESC_f,                   0,
134         -ESC_h,                  0,
135         0,                       -ESC_k,
136         0,                       0,
137         ESC_n,                   0,
138         -ESC_p,                  0,
139         ESC_r,                   -ESC_s,
140         ESC_tee,                 0,
141         -ESC_v,                  -ESC_w,
142         0,                       0,
143         -ESC_z
144  };  };
145    
146  #else           /* This is the "abnormal" table for EBCDIC systems */  #else
147    
148    /* This is the "abnormal" table for EBCDIC systems without UTF-8 support. */
149    
150  static const short int escapes[] = {  static const short int escapes[] = {
151  /*  48 */     0,     0,      0,     '.',    '<',   '(',    '+',    '|',  /*  48 */     0,     0,      0,     '.',    '<',   '(',    '+',    '|',
152  /*  50 */   '&',     0,      0,       0,      0,     0,      0,      0,  /*  50 */   '&',     0,      0,       0,      0,     0,      0,      0,
# Line 142  static const short int escapes[] = { Line 177  static const short int escapes[] = {
177    
178  /* Table of special "verbs" like (*PRUNE). This is a short table, so it is  /* Table of special "verbs" like (*PRUNE). This is a short table, so it is
179  searched linearly. Put all the names into a single string, in order to reduce  searched linearly. Put all the names into a single string, in order to reduce
180  the number of relocations when a shared library is dynamically linked. */  the number of relocations when a shared library is dynamically linked. The
181    string is built from string macros so that it works in UTF-8 mode on EBCDIC
182    platforms. */
183    
184  typedef struct verbitem {  typedef struct verbitem {
185    int   len;    int   len;
# Line 150  typedef struct verbitem { Line 187  typedef struct verbitem {
187  } verbitem;  } verbitem;
188    
189  static const char verbnames[] =  static const char verbnames[] =
190    "ACCEPT\0"    STRING_ACCEPT0
191    "COMMIT\0"    STRING_COMMIT0
192    "F\0"    STRING_F0
193    "FAIL\0"    STRING_FAIL0
194    "PRUNE\0"    STRING_PRUNE0
195    "SKIP\0"    STRING_SKIP0
196    "THEN";    STRING_THEN;
197    
198  static verbitem verbs[] = {  static const verbitem verbs[] = {
199    { 6, OP_ACCEPT },    { 6, OP_ACCEPT },
200    { 6, OP_COMMIT },    { 6, OP_COMMIT },
201    { 1, OP_FAIL },    { 1, OP_FAIL },
# Line 168  static verbitem verbs[] = { Line 205  static verbitem verbs[] = {
205    { 4, OP_THEN  }    { 4, OP_THEN  }
206  };  };
207    
208  static int verbcount = sizeof(verbs)/sizeof(verbitem);  static const int verbcount = sizeof(verbs)/sizeof(verbitem);
209    
210    
211  /* Tables of names of POSIX character classes and their lengths. The names are  /* Tables of names of POSIX character classes and their lengths. The names are
# Line 178  length entry. The first three must be al Line 215  length entry. The first three must be al
215  for handling case independence. */  for handling case independence. */
216    
217  static const char posix_names[] =  static const char posix_names[] =
218    "alpha\0"  "lower\0"  "upper\0"  "alnum\0"  "ascii\0"  "blank\0"    STRING_alpha0 STRING_lower0 STRING_upper0 STRING_alnum0
219    "cntrl\0"  "digit\0"  "graph\0"  "print\0"  "punct\0"  "space\0"    STRING_ascii0 STRING_blank0 STRING_cntrl0 STRING_digit0
220    "word\0"   "xdigit";    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0
221      STRING_word0  STRING_xdigit;
222    
223  static const uschar posix_name_lengths[] = {  static const uschar posix_name_lengths[] = {
224    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };
# Line 295  static const char error_texts[] = Line 333  static const char error_texts[] =
333    /* 55 */    /* 55 */
334    "repeating a DEFINE group is not allowed\0"    "repeating a DEFINE group is not allowed\0"
335    "inconsistent NEWLINE options\0"    "inconsistent NEWLINE options\0"
336    "\\g is not followed by a braced name or an optionally braced non-zero number\0"    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"
337    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number\0"    "a numbered reference must not be zero\0"
338    "(*VERB) with an argument is not supported\0"    "(*VERB) with an argument is not supported\0"
339    /* 60 */    /* 60 */
340    "(*VERB) not recognized\0"    "(*VERB) not recognized\0"
341    "number is too big\0"    "number is too big\0"
342    "subpattern name expected\0"    "subpattern name expected\0"
343    "digit expected after (?+";    "digit expected after (?+\0"
344      "] is an invalid data character in JavaScript compatibility mode";
345    
346    
347  /* 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 321  For convenience, we use the same bit def Line 360  For convenience, we use the same bit def
360    
361  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
362    
363  #ifndef EBCDIC  /* This is the "normal" case, for ASCII systems */  #ifndef EBCDIC
364    
365    /* This is the "normal" case, for ASCII systems, and EBCDIC systems running in
366    UTF-8 mode. */
367    
368  static const unsigned char digitab[] =  static const unsigned char digitab[] =
369    {    {
370    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
# Line 357  static const unsigned char digitab[] = Line 400  static const unsigned char digitab[] =
400    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */
401    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */
402    
403  #else           /* This is the "abnormal" case, for EBCDIC systems */  #else
404    
405    /* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */
406    
407  static const unsigned char digitab[] =  static const unsigned char digitab[] =
408    {    {
409    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
# Line 454  static const char * Line 500  static const char *
500  find_error_text(int n)  find_error_text(int n)
501  {  {
502  const char *s = error_texts;  const char *s = error_texts;
503  for (; n > 0; n--) while (*s++ != 0);  for (; n > 0; n--) while (*s++ != 0) {};
504  return s;  return s;
505  }  }
506    
# Line 502  if (c == 0) *errorcodeptr = ERR1; Line 548  if (c == 0) *errorcodeptr = ERR1;
548  in a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
549  Otherwise further processing may be required. */  Otherwise further processing may be required. */
550    
551  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
552  else if (c < '0' || c > 'z') {}                           /* Not alphanumeric */  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */
553  else if ((i = escapes[c - '0']) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0) c = i;
554    
555  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
556  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */
# Line 523  else Line 569  else
569      /* A number of Perl escapes are not handled by PCRE. We give an explicit      /* A number of Perl escapes are not handled by PCRE. We give an explicit
570      error. */      error. */
571    
572      case 'l':      case CHAR_l:
573      case 'L':      case CHAR_L:
574      case 'N':      case CHAR_N:
575      case 'u':      case CHAR_u:
576      case 'U':      case CHAR_U:
577      *errorcodeptr = ERR37;      *errorcodeptr = ERR37;
578      break;      break;
579    
580      /* \g must be followed by a number, either plain or braced. If positive, it      /* \g must be followed by one of a number of specific things:
581      is an absolute backreference. If negative, it is a relative backreference.  
582      This is a Perl 5.10 feature. Perl 5.10 also supports \g{name} as a      (1) A number, either plain or braced. If positive, it is an absolute
583      reference to a named group. This is part of Perl's movement towards a      backreference. If negative, it is a relative backreference. This is a Perl
584      unified syntax for back references. As this is synonymous with \k{name}, we      5.10 feature.
585      fudge it up by pretending it really was \k. */  
586        (2) Perl 5.10 also supports \g{name} as a reference to a named group. This
587        is part of Perl's movement towards a unified syntax for back references. As
588        this is synonymous with \k{name}, we fudge it up by pretending it really
589        was \k.
590    
591        (3) For Oniguruma compatibility we also support \g followed by a name or a
592        number either in angle brackets or in single quotes. However, these are
593        (possibly recursive) subroutine calls, _not_ backreferences. Just return
594        the -ESC_g code (cf \k). */
595    
596        case CHAR_g:
597        if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
598          {
599          c = -ESC_g;
600          break;
601          }
602    
603        /* Handle the Perl-compatible cases */
604    
605      case 'g':      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
     if (ptr[1] == '{')  
606        {        {
607        const uschar *p;        const uschar *p;
608        for (p = ptr+2; *p != 0 && *p != '}'; p++)        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
609          if (*p != '-' && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;
610        if (*p != 0 && *p != '}')        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
611          {          {
612          c = -ESC_k;          c = -ESC_k;
613          break;          break;
# Line 554  else Line 617  else
617        }        }
618      else braced = FALSE;      else braced = FALSE;
619    
620      if (ptr[1] == '-')      if (ptr[1] == CHAR_MINUS)
621        {        {
622        negated = TRUE;        negated = TRUE;
623        ptr++;        ptr++;
# Line 563  else Line 626  else
626    
627      c = 0;      c = 0;
628      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
629        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - CHAR_0;
630    
631      if (c < 0)      if (c < 0)   /* Integer overflow */
632        {        {
633        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
634        break;        break;
635        }        }
636    
637      if (c == 0 || (braced && *(++ptr) != '}'))      if (braced && *(++ptr) != CHAR_RIGHT_CURLY_BRACKET)
638        {        {
639        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
640        break;        break;
641        }        }
642    
643        if (c == 0)
644          {
645          *errorcodeptr = ERR58;
646          break;
647          }
648    
649      if (negated)      if (negated)
650        {        {
651        if (c > bracount)        if (c > bracount)
# Line 602  else Line 671  else
671      value is greater than 377, the least significant 8 bits are taken. Inside a      value is greater than 377, the least significant 8 bits are taken. Inside a
672      character class, \ followed by a digit is always an octal number. */      character class, \ followed by a digit is always an octal number. */
673    
674      case '1': case '2': case '3': case '4': case '5':      case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: case CHAR_5:
675      case '6': case '7': case '8': case '9':      case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:
676    
677      if (!isclass)      if (!isclass)
678        {        {
679        oldptr = ptr;        oldptr = ptr;
680        c -= '0';        c -= CHAR_0;
681        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
682          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - CHAR_0;
683        if (c < 0)        if (c < 0)    /* Integer overflow */
684          {          {
685          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
686          break;          break;
# Line 628  else Line 697  else
697      generates a binary zero byte and treats the digit as a following literal.      generates a binary zero byte and treats the digit as a following literal.
698      Thus we have to pull back the pointer by one. */      Thus we have to pull back the pointer by one. */
699    
700      if ((c = *ptr) >= '8')      if ((c = *ptr) >= CHAR_8)
701        {        {
702        ptr--;        ptr--;
703        c = 0;        c = 0;
# Line 641  else Line 710  else
710      to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more      to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more
711      than 3 octal digits. */      than 3 octal digits. */
712    
713      case '0':      case CHAR_0:
714      c -= '0';      c -= CHAR_0;
715      while(i++ < 2 && ptr[1] >= '0' && ptr[1] <= '7')      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
716          c = c * 8 + *(++ptr) - '0';          c = c * 8 + *(++ptr) - CHAR_0;
717      if (!utf8 && c > 255) *errorcodeptr = ERR51;      if (!utf8 && c > 255) *errorcodeptr = ERR51;
718      break;      break;
719    
# Line 652  else Line 721  else
721      than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is      than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is
722      treated as a data character. */      treated as a data character. */
723    
724      case 'x':      case CHAR_x:
725      if (ptr[1] == '{')      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
726        {        {
727        const uschar *pt = ptr + 2;        const uschar *pt = ptr + 2;
728        int count = 0;        int count = 0;
# Line 662  else Line 731  else
731        while ((digitab[*pt] & ctype_xdigit) != 0)        while ((digitab[*pt] & ctype_xdigit) != 0)
732          {          {
733          register int cc = *pt++;          register int cc = *pt++;
734          if (c == 0 && cc == '0') continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
735          count++;          count++;
736    
737  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
738          if (cc >= 'a') cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
739          c = (c << 4) + cc - ((cc < 'A')? '0' : ('A' - 10));          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
740  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
741          if (cc >= 'a' && cc <= 'z') cc += 64;  /* Convert to upper case */          if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
742          c = (c << 4) + cc - ((cc >= '0')? '0' : ('A' - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
743  #endif  #endif
744          }          }
745    
746        if (*pt == '}')        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
747          {          {
748          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;
749          ptr = pt;          ptr = pt;
# Line 690  else Line 759  else
759      c = 0;      c = 0;
760      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)
761        {        {
762        int cc;                               /* Some compilers don't like ++ */        int cc;                                  /* Some compilers don't like */
763        cc = *(++ptr);                        /* in initializers */        cc = *(++ptr);                           /* ++ in initializers */
764  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
765        if (cc >= 'a') cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
766        c = c * 16 + cc - ((cc < 'A')? '0' : ('A' - 10));        c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
767  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
768        if (cc <= 'z') cc += 64;              /* Convert to upper case */        if (cc <= CHAR_z) cc += 64;              /* Convert to upper case */
769        c = c * 16 + cc - ((cc >= '0')? '0' : ('A' - 10));        c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
770  #endif  #endif
771        }        }
772      break;      break;
# Line 706  else Line 775  else
775      This coding is ASCII-specific, but then the whole concept of \cx is      This coding is ASCII-specific, but then the whole concept of \cx is
776      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */
777    
778      case 'c':      case CHAR_c:
779      c = *(++ptr);      c = *(++ptr);
780      if (c == 0)      if (c == 0)
781        {        {
# Line 714  else Line 783  else
783        break;        break;
784        }        }
785    
786  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
787      if (c >= 'a' && c <= 'z') c -= 32;      if (c >= CHAR_a && c <= CHAR_z) c -= 32;
788      c ^= 0x40;      c ^= 0x40;
789  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
790      if (c >= 'a' && c <= 'z') c += 64;      if (c >= CHAR_a && c <= CHAR_z) c += 64;
791      c ^= 0xC0;      c ^= 0xC0;
792  #endif  #endif
793      break;      break;
# Line 780  if (c == 0) goto ERROR_RETURN; Line 849  if (c == 0) goto ERROR_RETURN;
849  /* \P or \p can be followed by a name in {}, optionally preceded by ^ for  /* \P or \p can be followed by a name in {}, optionally preceded by ^ for
850  negation. */  negation. */
851    
852  if (c == '{')  if (c == CHAR_LEFT_CURLY_BRACKET)
853    {    {
854    if (ptr[1] == '^')    if (ptr[1] == CHAR_CIRCUMFLEX_ACCENT)
855      {      {
856      *negptr = TRUE;      *negptr = TRUE;
857      ptr++;      ptr++;
# Line 791  if (c == '{') Line 860  if (c == '{')
860      {      {
861      c = *(++ptr);      c = *(++ptr);
862      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
863      if (c == '}') break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
864      name[i] = c;      name[i] = c;
865      }      }
866    if (c !='}') goto ERROR_RETURN;    if (c != CHAR_RIGHT_CURLY_BRACKET) goto ERROR_RETURN;
867    name[i] = 0;    name[i] = 0;
868    }    }
869    
# Line 859  is_counted_repeat(const uschar *p) Line 928  is_counted_repeat(const uschar *p)
928  {  {
929  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;
930  while ((digitab[*p] & ctype_digit) != 0) p++;  while ((digitab[*p] & ctype_digit) != 0) p++;
931  if (*p == '}') return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
932    
933  if (*p++ != ',') return FALSE;  if (*p++ != CHAR_COMMA) return FALSE;
934  if (*p == '}') return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
935    
936  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;
937  while ((digitab[*p] & ctype_digit) != 0) p++;  while ((digitab[*p] & ctype_digit) != 0) p++;
938    
939  return (*p == '}');  return (*p == CHAR_RIGHT_CURLY_BRACKET);
940  }  }
941    
942    
# Line 900  int max = -1; Line 969  int max = -1;
969  /* Read the minimum value and do a paranoid check: a negative value indicates  /* Read the minimum value and do a paranoid check: a negative value indicates
970  an integer overflow. */  an integer overflow. */
971    
972  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - '0';  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;
973  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
974    {    {
975    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 910  if (min < 0 || min > 65535) Line 979  if (min < 0 || min > 65535)
979  /* Read the maximum value if there is one, and again do a paranoid on its size.  /* Read the maximum value if there is one, and again do a paranoid on its size.
980  Also, max must not be less than min. */  Also, max must not be less than min. */
981    
982  if (*p == '}') max = min; else  if (*p == CHAR_RIGHT_CURLY_BRACKET) max = min; else
983    {    {
984    if (*(++p) != '}')    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
985      {      {
986      max = 0;      max = 0;
987      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - '0';      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;
988      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
989        {        {
990        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 952  be terminated by '>' because that is che Line 1021  be terminated by '>' because that is che
1021    
1022  Arguments:  Arguments:
1023    ptr          current position in the pattern    ptr          current position in the pattern
1024    count        current count of capturing parens so far encountered    cd           compile background data
1025    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1026    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1027    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
# Line 961  Returns:       the number of the named s Line 1030  Returns:       the number of the named s
1030  */  */
1031    
1032  static int  static int
1033  find_parens(const uschar *ptr, int count, const uschar *name, int lorn,  find_parens(const uschar *ptr, compile_data *cd, const uschar *name, int lorn,
1034    BOOL xmode)    BOOL xmode)
1035  {  {
1036  const uschar *thisname;  const uschar *thisname;
1037    int count = cd->bracount;
1038    
1039  for (; *ptr != 0; ptr++)  for (; *ptr != 0; ptr++)
1040    {    {
# Line 972  for (; *ptr != 0; ptr++) Line 1042  for (; *ptr != 0; ptr++)
1042    
1043    /* Skip over backslashed characters and also entire \Q...\E */    /* Skip over backslashed characters and also entire \Q...\E */
1044    
1045    if (*ptr == '\\')    if (*ptr == CHAR_BACKSLASH)
1046      {      {
1047      if (*(++ptr) == 0) return -1;      if (*(++ptr) == 0) return -1;
1048      if (*ptr == 'Q') for (;;)      if (*ptr == CHAR_Q) for (;;)
1049        {        {
1050        while (*(++ptr) != 0 && *ptr != '\\');        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};
1051        if (*ptr == 0) return -1;        if (*ptr == 0) return -1;
1052        if (*(++ptr) == 'E') break;        if (*(++ptr) == CHAR_E) break;
1053        }        }
1054      continue;      continue;
1055      }      }
1056    
1057    /* Skip over character classes */    /* Skip over character classes; this logic must be similar to the way they
1058      are handled for real. If the first character is '^', skip it. Also, if the
1059      first few characters (either before or after ^) are \Q\E or \E we skip them
1060      too. This makes for compatibility with Perl. Note the use of STR macros to
1061      encode "Q\\E" so that it works in UTF-8 on EBCDIC platforms. */
1062    
1063    if (*ptr == '[')    if (*ptr == CHAR_LEFT_SQUARE_BRACKET)
1064      {      {
1065      while (*(++ptr) != ']')      BOOL negate_class = FALSE;
1066        for (;;)
1067          {
1068          int c = *(++ptr);
1069          if (c == CHAR_BACKSLASH)
1070            {
1071            if (ptr[1] == CHAR_E)
1072              ptr++;
1073            else if (strncmp((const char *)ptr+1,
1074                     STR_Q STR_BACKSLASH STR_E, 3) == 0)
1075              ptr += 3;
1076            else
1077              break;
1078            }
1079          else if (!negate_class && c == CHAR_CIRCUMFLEX_ACCENT)
1080            negate_class = TRUE;
1081          else break;
1082          }
1083    
1084        /* If the next character is ']', it is a data character that must be
1085        skipped, except in JavaScript compatibility mode. */
1086    
1087        if (ptr[1] == CHAR_RIGHT_SQUARE_BRACKET &&
1088            (cd->external_options & PCRE_JAVASCRIPT_COMPAT) == 0)
1089          ptr++;
1090    
1091        while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1092        {        {
1093        if (*ptr == 0) return -1;        if (*ptr == 0) return -1;
1094        if (*ptr == '\\')        if (*ptr == CHAR_BACKSLASH)
1095          {          {
1096          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
1097          if (*ptr == 'Q') for (;;)          if (*ptr == CHAR_Q) for (;;)
1098            {            {
1099            while (*(++ptr) != 0 && *ptr != '\\');            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};
1100            if (*ptr == 0) return -1;            if (*ptr == 0) return -1;
1101            if (*(++ptr) == 'E') break;            if (*(++ptr) == CHAR_E) break;
1102            }            }
1103          continue;          continue;
1104          }          }
# Line 1008  for (; *ptr != 0; ptr++) Line 1108  for (; *ptr != 0; ptr++)
1108    
1109    /* Skip comments in /x mode */    /* Skip comments in /x mode */
1110    
1111    if (xmode && *ptr == '#')    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1112      {      {
1113      while (*(++ptr) != 0 && *ptr != '\n');      while (*(++ptr) != 0 && *ptr != CHAR_NL) {};
1114      if (*ptr == 0) return -1;      if (*ptr == 0) return -1;
1115      continue;      continue;
1116      }      }
1117    
1118    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
1119    
1120    if (*ptr != '(') continue;    if (*ptr != CHAR_LEFT_PARENTHESIS) continue;
1121    if (ptr[1] != '?' && ptr[1] != '*')    if (ptr[1] != CHAR_QUESTION_MARK && ptr[1] != CHAR_ASTERISK)
1122      {      {
1123      count++;      count++;
1124      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 1026  for (; *ptr != 0; ptr++) Line 1126  for (; *ptr != 0; ptr++)
1126      }      }
1127    
1128    ptr += 2;    ptr += 2;
1129    if (*ptr == 'P') ptr++;                      /* Allow optional P */    if (*ptr == CHAR_P) ptr++;                      /* Allow optional P */
1130    
1131    /* We have to disambiguate (?<! and (?<= from (?<name> */    /* We have to disambiguate (?<! and (?<= from (?<name> */
1132    
1133    if ((*ptr != '<' || ptr[1] == '!' || ptr[1] == '=') &&    if ((*ptr != CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_EXCLAMATION_MARK ||
1134         *ptr != '\'')        ptr[1] == CHAR_EQUALS_SIGN) && *ptr != CHAR_APOSTROPHE)
1135      continue;      continue;
1136    
1137    count++;    count++;
1138    
1139    if (name == NULL && count == lorn) return count;    if (name == NULL && count == lorn) return count;
1140    term = *ptr++;    term = *ptr++;
1141    if (term == '<') term = '>';    if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1142    thisname = ptr;    thisname = ptr;
1143    while (*ptr != term) ptr++;    while (*ptr != term) ptr++;
1144    if (name != NULL && lorn == ptr - thisname &&    if (name != NULL && lorn == ptr - thisname &&
# Line 1252  for (;;) Line 1352  for (;;)
1352      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
1353      case OP_WORDCHAR:      case OP_WORDCHAR:
1354      case OP_ANY:      case OP_ANY:
1355        case OP_ALLANY:
1356      branchlength++;      branchlength++;
1357      cc++;      cc++;
1358      break;      break;
# Line 1400  for (;;) Line 1501  for (;;)
1501        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1502        break;        break;
1503        }        }
1504    #else
1505        (void)(utf8);  /* Keep compiler happy by referencing function argument */
1506  #endif  #endif
1507      }      }
1508    }    }
# Line 1493  for (;;) Line 1596  for (;;)
1596        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1597        break;        break;
1598        }        }
1599    #else
1600        (void)(utf8);  /* Keep compiler happy by referencing function argument */
1601  #endif  #endif
1602      }      }
1603    }    }
# Line 1544  for (code = first_significant_code(code Line 1649  for (code = first_significant_code(code
1649    
1650    /* Groups with zero repeats can of course be empty; skip them. */    /* Groups with zero repeats can of course be empty; skip them. */
1651    
1652    if (c == OP_BRAZERO || c == OP_BRAMINZERO)    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO)
1653      {      {
1654      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1655      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
# Line 1630  for (code = first_significant_code(code Line 1735  for (code = first_significant_code(code
1735      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
1736      case OP_WORDCHAR:      case OP_WORDCHAR:
1737      case OP_ANY:      case OP_ANY:
1738        case OP_ALLANY:
1739      case OP_ANYBYTE:      case OP_ANYBYTE:
1740      case OP_CHAR:      case OP_CHAR:
1741      case OP_CHARNC:      case OP_CHARNC:
# Line 1737  return TRUE; Line 1843  return TRUE;
1843  *************************************************/  *************************************************/
1844    
1845  /* This function is called when the sequence "[:" or "[." or "[=" is  /* This function is called when the sequence "[:" or "[." or "[=" is
1846  encountered in a character class. It checks whether this is followed by an  encountered in a character class. It checks whether this is followed by a
1847  optional ^ and then a sequence of letters, terminated by a matching ":]" or  sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
1848  ".]" or "=]".  reach an unescaped ']' without the special preceding character, return FALSE.
1849    
1850    Originally, this function only recognized a sequence of letters between the
1851    terminators, but it seems that Perl recognizes any sequence of characters,
1852    though of course unknown POSIX names are subsequently rejected. Perl gives an
1853    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
1854    didn't consider this to be a POSIX class. Likewise for [:1234:].
1855    
1856    The problem in trying to be exactly like Perl is in the handling of escapes. We
1857    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
1858    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
1859    below handles the special case of \], but does not try to do any other escape
1860    processing. This makes it different from Perl for cases such as [:l\ower:]
1861    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
1862    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
1863    I think.
1864    
1865  Argument:  Arguments:
1866    ptr      pointer to the initial [    ptr      pointer to the initial [
1867    endptr   where to return the end pointer    endptr   where to return the end pointer
   cd       pointer to compile data  
1868    
1869  Returns:   TRUE or FALSE  Returns:   TRUE or FALSE
1870  */  */
1871    
1872  static BOOL  static BOOL
1873  check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd)  check_posix_syntax(const uschar *ptr, const uschar **endptr)
1874  {  {
1875  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
1876  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
1877  if (*(++ptr) == '^') ptr++;  for (++ptr; *ptr != 0; ptr++)
 while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;  
 if (*ptr == terminator && ptr[1] == ']')  
1878    {    {
1879    *endptr = ptr;    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else
1880    return TRUE;      {
1881        if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
1882        if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
1883          {
1884          *endptr = ptr;
1885          return TRUE;
1886          }
1887        }
1888    }    }
1889  return FALSE;  return FALSE;
1890  }  }
# Line 1805  return -1; Line 1930  return -1;
1930  that is referenced. This means that groups can be replicated for fixed  that is referenced. This means that groups can be replicated for fixed
1931  repetition simply by copying (because the recursion is allowed to refer to  repetition simply by copying (because the recursion is allowed to refer to
1932  earlier groups that are outside the current group). However, when a group is  earlier groups that are outside the current group). However, when a group is
1933  optional (i.e. the minimum quantifier is zero), OP_BRAZERO is inserted before  optional (i.e. the minimum quantifier is zero), OP_BRAZERO or OP_SKIPZERO is
1934  it, after it has been compiled. This means that any OP_RECURSE items within it  inserted before it, after it has been compiled. This means that any OP_RECURSE
1935  that refer to the group itself or any contained groups have to have their  items within it that refer to the group itself or any contained groups have to
1936  offsets adjusted. That one of the jobs of this function. Before it is called,  have their offsets adjusted. That one of the jobs of this function. Before it
1937  the partially compiled regex must be temporarily terminated with OP_END.  is called, the partially compiled regex must be temporarily terminated with
1938    OP_END.
1939    
1940  This function has been extended with the possibility of forward references for  This function has been extended with the possibility of forward references for
1941  recursions and subroutine calls. It must also check the list of such references  recursions and subroutine calls. It must also check the list of such references
# Line 1944  get_othercase_range(unsigned int *cptr, Line 2070  get_othercase_range(unsigned int *cptr,
2070  unsigned int c, othercase, next;  unsigned int c, othercase, next;
2071    
2072  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2073    { if ((othercase = _pcre_ucp_othercase(c)) != NOTACHAR) break; }    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }
2074    
2075  if (c > d) return FALSE;  if (c > d) return FALSE;
2076    
# Line 1953  next = othercase + 1; Line 2079  next = othercase + 1;
2079    
2080  for (++c; c <= d; c++)  for (++c; c <= d; c++)
2081    {    {
2082    if (_pcre_ucp_othercase(c) != next) break;    if (UCD_OTHERCASE(c) != next) break;
2083    next++;    next++;
2084    }    }
2085    
# Line 1999  if ((options & PCRE_EXTENDED) != 0) Line 2125  if ((options & PCRE_EXTENDED) != 0)
2125    for (;;)    for (;;)
2126      {      {
2127      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2128      if (*ptr == '#')      if (*ptr == CHAR_NUMBER_SIGN)
2129        {        {
2130        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2131          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
# Line 2011  if ((options & PCRE_EXTENDED) != 0) Line 2137  if ((options & PCRE_EXTENDED) != 0)
2137  /* If the next item is one that we can handle, get its value. A non-negative  /* If the next item is one that we can handle, get its value. A non-negative
2138  value is a character, a negative value is an escape value. */  value is a character, a negative value is an escape value. */
2139    
2140  if (*ptr == '\\')  if (*ptr == CHAR_BACKSLASH)
2141    {    {
2142    int temperrorcode = 0;    int temperrorcode = 0;
2143    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);
# Line 2036  if ((options & PCRE_EXTENDED) != 0) Line 2162  if ((options & PCRE_EXTENDED) != 0)
2162    for (;;)    for (;;)
2163      {      {
2164      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2165      if (*ptr == '#')      if (*ptr == CHAR_NUMBER_SIGN)
2166        {        {
2167        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2168          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
# Line 2047  if ((options & PCRE_EXTENDED) != 0) Line 2173  if ((options & PCRE_EXTENDED) != 0)
2173    
2174  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
2175    
2176  if (*ptr == '*' || *ptr == '?' || strncmp((char *)ptr, "{0,", 3) == 0)  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
2177    return FALSE;    strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
2178        return FALSE;
2179    
2180  /* Now compare the next item with the previous opcode. If the previous is a  /* Now compare the next item with the previous opcode. If the previous is a
2181  positive single character match, "item" either contains the character or, if  positive single character match, "item" either contains the character or, if
# Line 2063  if (next >= 0) switch(op_code) Line 2190  if (next >= 0) switch(op_code)
2190    case OP_CHAR:    case OP_CHAR:
2191  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2192    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }
2193    #else
2194      (void)(utf8_char);  /* Keep compiler happy by referencing function argument */
2195  #endif  #endif
2196    return item != next;    return item != next;
2197    
# Line 2081  if (next >= 0) switch(op_code) Line 2210  if (next >= 0) switch(op_code)
2210      unsigned int othercase;      unsigned int othercase;
2211      if (next < 128) othercase = cd->fcc[next]; else      if (next < 128) othercase = cd->fcc[next]; else
2212  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2213      othercase = _pcre_ucp_othercase((unsigned int)next);      othercase = UCD_OTHERCASE((unsigned int)next);
2214  #else  #else
2215      othercase = NOTACHAR;      othercase = NOTACHAR;
2216  #endif  #endif
# Line 2094  if (next >= 0) switch(op_code) Line 2223  if (next >= 0) switch(op_code)
2223    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, "item" must be a single-byte character. */
2224    
2225    case OP_NOT:    case OP_NOT:
   if (next < 0) return FALSE;  /* Not a character */  
2226    if (item == next) return TRUE;    if (item == next) return TRUE;
2227    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2228  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 2103  if (next >= 0) switch(op_code) Line 2231  if (next >= 0) switch(op_code)
2231      unsigned int othercase;      unsigned int othercase;
2232      if (next < 128) othercase = cd->fcc[next]; else      if (next < 128) othercase = cd->fcc[next]; else
2233  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2234      othercase = _pcre_ucp_othercase(next);      othercase = UCD_OTHERCASE(next);
2235  #else  #else
2236      othercase = NOTACHAR;      othercase = NOTACHAR;
2237  #endif  #endif
# Line 2357  uschar classbits[32]; Line 2485  uschar classbits[32];
2485  BOOL class_utf8;  BOOL class_utf8;
2486  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8 = (options & PCRE_UTF8) != 0;
2487  uschar *class_utf8data;  uschar *class_utf8data;
2488    uschar *class_utf8data_base;
2489  uschar utf8_char[6];  uschar utf8_char[6];
2490  #else  #else
2491  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
# Line 2484  for (;; ptr++) Line 2613  for (;; ptr++)
2613    
2614    if (inescq && c != 0)    if (inescq && c != 0)
2615      {      {
2616      if (c == '\\' && ptr[1] == 'E')      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
2617        {        {
2618        inescq = FALSE;        inescq = FALSE;
2619        ptr++;        ptr++;
# Line 2510  for (;; ptr++) Line 2639  for (;; ptr++)
2639    /* Fill in length of a previous callout, except when the next thing is    /* Fill in length of a previous callout, except when the next thing is
2640    a quantifier. */    a quantifier. */
2641    
2642    is_quantifier = c == '*' || c == '+' || c == '?' ||    is_quantifier =
2643      (c == '{' && is_counted_repeat(ptr+1));      c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK ||
2644        (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1));
2645    
2646    if (!is_quantifier && previous_callout != NULL &&    if (!is_quantifier && previous_callout != NULL &&
2647         after_manual_callout-- <= 0)         after_manual_callout-- <= 0)
# Line 2526  for (;; ptr++) Line 2656  for (;; ptr++)
2656    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
2657      {      {
2658      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if ((cd->ctypes[c] & ctype_space) != 0) continue;
2659      if (c == '#')      if (c == CHAR_NUMBER_SIGN)
2660        {        {
2661        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2662          {          {
# Line 2551  for (;; ptr++) Line 2681  for (;; ptr++)
2681      {      {
2682      /* ===================================================================*/      /* ===================================================================*/
2683      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
2684      case '|':                      /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
2685      case ')':      case CHAR_RIGHT_PARENTHESIS:
2686      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
2687      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
2688      *codeptr = code;      *codeptr = code;
# Line 2574  for (;; ptr++) Line 2704  for (;; ptr++)
2704      /* Handle single-character metacharacters. In multiline mode, ^ disables      /* Handle single-character metacharacters. In multiline mode, ^ disables
2705      the setting of any following char as a first character. */      the setting of any following char as a first character. */
2706    
2707      case '^':      case CHAR_CIRCUMFLEX_ACCENT:
2708      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
2709        {        {
2710        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
# Line 2583  for (;; ptr++) Line 2713  for (;; ptr++)
2713      *code++ = OP_CIRC;      *code++ = OP_CIRC;
2714      break;      break;
2715    
2716      case '$':      case CHAR_DOLLAR_SIGN:
2717      previous = NULL;      previous = NULL;
2718      *code++ = OP_DOLL;      *code++ = OP_DOLL;
2719      break;      break;
# Line 2591  for (;; ptr++) Line 2721  for (;; ptr++)
2721      /* There can never be a first char if '.' is first, whatever happens about      /* There can never be a first char if '.' is first, whatever happens about
2722      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqbyte doesn't change either. */
2723    
2724      case '.':      case CHAR_DOT:
2725      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
2726      zerofirstbyte = firstbyte;      zerofirstbyte = firstbyte;
2727      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
2728      previous = code;      previous = code;
2729      *code++ = OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
2730      break;      break;
2731    
2732    
# Line 2611  for (;; ptr++) Line 2741  for (;; ptr++)
2741      opcode is compiled. It may optionally have a bit map for characters < 256,      opcode is compiled. It may optionally have a bit map for characters < 256,
2742      but those above are are explicitly listed afterwards. A flag byte tells      but those above are are explicitly listed afterwards. A flag byte tells
2743      whether the bitmap is present, and whether this is a negated class or not.      whether the bitmap is present, and whether this is a negated class or not.
     */  
2744    
2745      case '[':      In JavaScript compatibility mode, an isolated ']' causes an error. In
2746        default (Perl) mode, it is treated as a data character. */
2747    
2748        case CHAR_RIGHT_SQUARE_BRACKET:
2749        if ((cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
2750          {
2751          *errorcodeptr = ERR64;
2752          goto FAILED;
2753          }
2754        goto NORMAL_CHAR;
2755    
2756        case CHAR_LEFT_SQUARE_BRACKET:
2757      previous = code;      previous = code;
2758    
2759      /* PCRE supports POSIX class stuff inside a class. Perl gives an error if      /* PCRE supports POSIX class stuff inside a class. Perl gives an error if
2760      they are encountered at the top level, so we'll do that too. */      they are encountered at the top level, so we'll do that too. */
2761    
2762      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&      if ((ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2763          check_posix_syntax(ptr, &tempptr, cd))           ptr[1] == CHAR_EQUALS_SIGN) &&
2764            check_posix_syntax(ptr, &tempptr))
2765        {        {
2766        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;        *errorcodeptr = (ptr[1] == CHAR_COLON)? ERR13 : ERR31;
2767        goto FAILED;        goto FAILED;
2768        }        }
2769    
# Line 2634  for (;; ptr++) Line 2775  for (;; ptr++)
2775      for (;;)      for (;;)
2776        {        {
2777        c = *(++ptr);        c = *(++ptr);
2778        if (c == '\\')        if (c == CHAR_BACKSLASH)
2779          {          {
2780          if (ptr[1] == 'E') ptr++;          if (ptr[1] == CHAR_E)
2781            else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;            ptr++;
2782              else break;          else if (strncmp((const char *)ptr+1,
2783                              STR_Q STR_BACKSLASH STR_E, 3) == 0)
2784              ptr += 3;
2785            else
2786              break;
2787          }          }
2788        else if (!negate_class && c == '^')        else if (!negate_class && c == CHAR_CIRCUMFLEX_ACCENT)
2789          negate_class = TRUE;          negate_class = TRUE;
2790        else break;        else break;
2791        }        }
2792    
2793        /* Empty classes are allowed in JavaScript compatibility mode. Otherwise,
2794        an initial ']' is taken as a data character -- the code below handles
2795        that. In JS mode, [] must always fail, so generate OP_FAIL, whereas
2796        [^] must match any character, so generate OP_ALLANY. */
2797    
2798        if (c == CHAR_RIGHT_SQUARE_BRACKET &&
2799            (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
2800          {
2801          *code++ = negate_class? OP_ALLANY : OP_FAIL;
2802          if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
2803          zerofirstbyte = firstbyte;
2804          break;
2805          }
2806    
2807      /* If a class contains a negative special such as \S, we need to flip the      /* If a class contains a negative special such as \S, we need to flip the
2808      negation flag at the end, so that support for characters > 255 works      negation flag at the end, so that support for characters > 255 works
2809      correctly (they are all included in the class). */      correctly (they are all included in the class). */
# Line 2668  for (;; ptr++) Line 2827  for (;; ptr++)
2827  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2828      class_utf8 = FALSE;                       /* No chars >= 256 */      class_utf8 = FALSE;                       /* No chars >= 256 */
2829      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */
2830        class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */
2831  #endif  #endif
2832    
2833      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
# Line 2683  for (;; ptr++) Line 2843  for (;; ptr++)
2843          {                           /* Braces are required because the */          {                           /* Braces are required because the */
2844          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
2845          }          }
2846    
2847          /* In the pre-compile phase, accumulate the length of any UTF-8 extra
2848          data and reset the pointer. This is so that very large classes that
2849          contain a zillion UTF-8 characters no longer overwrite the work space
2850          (which is on the stack). */
2851    
2852          if (lengthptr != NULL)
2853            {
2854            *lengthptr += class_utf8data - class_utf8data_base;
2855            class_utf8data = class_utf8data_base;
2856            }
2857    
2858  #endif  #endif
2859    
2860        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
2861    
2862        if (inescq)        if (inescq)
2863          {          {
2864          if (c == '\\' && ptr[1] == 'E')     /* If we are at \E */          if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)  /* If we are at \E */
2865            {            {
2866            inescq = FALSE;                   /* Reset literal state */            inescq = FALSE;                   /* Reset literal state */
2867            ptr++;                            /* Skip the 'E' */            ptr++;                            /* Skip the 'E' */
# Line 2704  for (;; ptr++) Line 2876  for (;; ptr++)
2876        [.ch.] and [=ch=] ("collating elements") and fault them, as Perl        [.ch.] and [=ch=] ("collating elements") and fault them, as Perl
2877        5.6 and 5.8 do. */        5.6 and 5.8 do. */
2878    
2879        if (c == '[' &&        if (c == CHAR_LEFT_SQUARE_BRACKET &&
2880            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&            (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2881            check_posix_syntax(ptr, &tempptr, cd))             ptr[1] == CHAR_EQUALS_SIGN) && check_posix_syntax(ptr, &tempptr))
2882          {          {
2883          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
2884          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
2885          register const uschar *cbits = cd->cbits;          register const uschar *cbits = cd->cbits;
2886          uschar pbits[32];          uschar pbits[32];
2887    
2888          if (ptr[1] != ':')          if (ptr[1] != CHAR_COLON)
2889            {            {
2890            *errorcodeptr = ERR31;            *errorcodeptr = ERR31;
2891            goto FAILED;            goto FAILED;
2892            }            }
2893    
2894          ptr += 2;          ptr += 2;
2895          if (*ptr == '^')          if (*ptr == CHAR_CIRCUMFLEX_ACCENT)
2896            {            {
2897            local_negate = TRUE;            local_negate = TRUE;
2898            should_flip_negation = TRUE;  /* Note negative special */            should_flip_negation = TRUE;  /* Note negative special */
# Line 2793  for (;; ptr++) Line 2965  for (;; ptr++)
2965        to 'or' into the one we are building. We assume they have more than one        to 'or' into the one we are building. We assume they have more than one
2966        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2967    
2968        if (c == '\\')        if (c == CHAR_BACKSLASH)
2969          {          {
2970          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2971          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2972    
2973          if (-c == ESC_b) c = '\b';       /* \b is backspace in a class */          if (-c == ESC_b) c = CHAR_BS;       /* \b is backspace in a class */
2974          else if (-c == ESC_X) c = 'X';   /* \X is literal X in a class */          else if (-c == ESC_X) c = CHAR_X;   /* \X is literal X in a class */
2975          else if (-c == ESC_R) c = 'R';   /* \R is literal R in a class */          else if (-c == ESC_R) c = CHAR_R;   /* \R is literal R in a class */
2976          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
2977            {            {
2978            if (ptr[1] == '\\' && ptr[2] == 'E')            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
2979              {              {
2980              ptr += 2; /* avoid empty string */              ptr += 2; /* avoid empty string */
2981              }              }
# Line 3029  for (;; ptr++) Line 3201  for (;; ptr++)
3201        entirely. The code for handling \Q and \E is messy. */        entirely. The code for handling \Q and \E is messy. */
3202    
3203        CHECK_RANGE:        CHECK_RANGE:
3204        while (ptr[1] == '\\' && ptr[2] == 'E')        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
3205          {          {
3206          inescq = FALSE;          inescq = FALSE;
3207          ptr += 2;          ptr += 2;
# Line 3039  for (;; ptr++) Line 3211  for (;; ptr++)
3211    
3212        /* Remember \r or \n */        /* Remember \r or \n */
3213    
3214        if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
3215    
3216        /* Check for range */        /* Check for range */
3217    
3218        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == CHAR_MINUS)
3219          {          {
3220          int d;          int d;
3221          ptr += 2;          ptr += 2;
3222          while (*ptr == '\\' && ptr[1] == 'E') ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
3223    
3224          /* If we hit \Q (not followed by \E) at this point, go into escaped          /* If we hit \Q (not followed by \E) at this point, go into escaped
3225          mode. */          mode. */
3226    
3227          while (*ptr == '\\' && ptr[1] == 'Q')          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_Q)
3228            {            {
3229            ptr += 2;            ptr += 2;
3230            if (*ptr == '\\' && ptr[1] == 'E') { ptr += 2; continue; }            if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3231                { ptr += 2; continue; }
3232            inescq = TRUE;            inescq = TRUE;
3233            break;            break;
3234            }            }
3235    
3236          if (*ptr == 0 || (!inescq && *ptr == ']'))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
3237            {            {
3238            ptr = oldptr;            ptr = oldptr;
3239            goto LONE_SINGLE_CHARACTER;            goto LONE_SINGLE_CHARACTER;
# Line 3079  for (;; ptr++) Line 3252  for (;; ptr++)
3252          not any of the other escapes. Perl 5.6 treats a hyphen as a literal          not any of the other escapes. Perl 5.6 treats a hyphen as a literal
3253          in such circumstances. */          in such circumstances. */
3254    
3255          if (!inescq && d == '\\')          if (!inescq && d == CHAR_BACKSLASH)
3256            {            {
3257            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3258            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
# Line 3089  for (;; ptr++) Line 3262  for (;; ptr++)
3262    
3263            if (d < 0)            if (d < 0)
3264              {              {
3265              if (d == -ESC_b) d = '\b';              if (d == -ESC_b) d = CHAR_BS;
3266              else if (d == -ESC_X) d = 'X';              else if (d == -ESC_X) d = CHAR_X;
3267              else if (d == -ESC_R) d = 'R'; else              else if (d == -ESC_R) d = CHAR_R; else
3268                {                {
3269                ptr = oldptr;                ptr = oldptr;
3270                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto LONE_SINGLE_CHARACTER;  /* A few lines below */
# Line 3112  for (;; ptr++) Line 3285  for (;; ptr++)
3285    
3286          /* Remember \r or \n */          /* Remember \r or \n */
3287    
3288          if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
3289    
3290          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless
3291          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
# Line 3232  for (;; ptr++) Line 3405  for (;; ptr++)
3405          if ((options & PCRE_CASELESS) != 0)          if ((options & PCRE_CASELESS) != 0)
3406            {            {
3407            unsigned int othercase;            unsigned int othercase;
3408            if ((othercase = _pcre_ucp_othercase(c)) != NOTACHAR)            if ((othercase = UCD_OTHERCASE(c)) != c)
3409              {              {
3410              *class_utf8data++ = XCL_SINGLE;              *class_utf8data++ = XCL_SINGLE;
3411              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);
# Line 3259  for (;; ptr++) Line 3432  for (;; ptr++)
3432    
3433      /* Loop until ']' reached. This "while" is the end of the "do" above. */      /* Loop until ']' reached. This "while" is the end of the "do" above. */
3434    
3435      while ((c = *(++ptr)) != 0 && (c != ']' || inescq));      while ((c = *(++ptr)) != 0 && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
3436    
3437      if (c == 0)                          /* Missing terminating ']' */      if (c == 0)                          /* Missing terminating ']' */
3438        {        {
# Line 3404  we set the flag only if there is a liter Line 3577  we set the flag only if there is a liter
3577      /* Various kinds of repeat; '{' is not necessarily a quantifier, but this      /* Various kinds of repeat; '{' is not necessarily a quantifier, but this
3578      has been tested above. */      has been tested above. */
3579    
3580      case '{':      case CHAR_LEFT_CURLY_BRACKET:
3581      if (!is_quantifier) goto NORMAL_CHAR;      if (!is_quantifier) goto NORMAL_CHAR;
3582      ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr);      ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr);
3583      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
3584      goto REPEAT;      goto REPEAT;
3585    
3586      case '*':      case CHAR_ASTERISK:
3587      repeat_min = 0;      repeat_min = 0;
3588      repeat_max = -1;      repeat_max = -1;
3589      goto REPEAT;      goto REPEAT;
3590    
3591      case '+':      case CHAR_PLUS:
3592      repeat_min = 1;      repeat_min = 1;
3593      repeat_max = -1;      repeat_max = -1;
3594      goto REPEAT;      goto REPEAT;
3595    
3596      case '?':      case CHAR_QUESTION_MARK:
3597      repeat_min = 0;      repeat_min = 0;
3598      repeat_max = 1;      repeat_max = 1;
3599    
# Line 3455  we set the flag only if there is a liter Line 3628  we set the flag only if there is a liter
3628      but if PCRE_UNGREEDY is set, it works the other way round. We change the      but if PCRE_UNGREEDY is set, it works the other way round. We change the
3629      repeat type to the non-default. */      repeat type to the non-default. */
3630    
3631      if (ptr[1] == '+')      if (ptr[1] == CHAR_PLUS)
3632        {        {
3633        repeat_type = 0;                  /* Force greedy */        repeat_type = 0;                  /* Force greedy */
3634        possessive_quantifier = TRUE;        possessive_quantifier = TRUE;
3635        ptr++;        ptr++;
3636        }        }
3637      else if (ptr[1] == '?')      else if (ptr[1] == CHAR_QUESTION_MARK)
3638        {        {
3639        repeat_type = greedy_non_default;        repeat_type = greedy_non_default;
3640        ptr++;        ptr++;
# Line 3787  we set the flag only if there is a liter Line 3960  we set the flag only if there is a liter
3960    
3961        if (repeat_min == 0)        if (repeat_min == 0)
3962          {          {
3963          /* If the maximum is also zero, we just omit the group from the output          /* If the maximum is also zero, we used to just omit the group from the
3964          altogether. */          output altogether, like this:
3965    
3966          if (repeat_max == 0)          ** if (repeat_max == 0)
3967            {          **   {
3968            code = previous;          **   code = previous;
3969            goto END_REPEAT;          **   goto END_REPEAT;
3970            }          **   }
3971    
3972            However, that fails when a group is referenced as a subroutine from
3973            elsewhere in the pattern, so now we stick in OP_SKIPZERO in front of it
3974            so that it is skipped on execution. As we don't have a list of which
3975            groups are referenced, we cannot do this selectively.
3976    
3977            If the maximum is 1 or unlimited, we just have to stick in the BRAZERO
3978            and do no more at this point. However, we do need to adjust any
3979            OP_RECURSE calls inside the group that refer to the group itself or any
3980            internal or forward referenced group, because the offset is from the
3981            start of the whole regex. Temporarily terminate the pattern while doing
3982            this. */
3983    
3984          /* If the maximum is 1 or unlimited, we just have to stick in the          if (repeat_max <= 1)    /* Covers 0, 1, and unlimited */
         BRAZERO and do no more at this point. However, we do need to adjust  
         any OP_RECURSE calls inside the group that refer to the group itself or  
         any internal or forward referenced group, because the offset is from  
         the start of the whole regex. Temporarily terminate the pattern while  
         doing this. */  
   
         if (repeat_max <= 1)  
3985            {            {
3986            *code = OP_END;            *code = OP_END;
3987            adjust_recurse(previous, 1, utf8, cd, save_hwm);            adjust_recurse(previous, 1, utf8, cd, save_hwm);
3988            memmove(previous+1, previous, len);            memmove(previous+1, previous, len);
3989            code++;            code++;
3990              if (repeat_max == 0)
3991                {
3992                *previous++ = OP_SKIPZERO;
3993                goto END_REPEAT;
3994                }
3995            *previous++ = OP_BRAZERO + repeat_type;            *previous++ = OP_BRAZERO + repeat_type;
3996            }            }
3997    
# Line 4003  we set the flag only if there is a liter Line 4186  we set the flag only if there is a liter
4186          }          }
4187        }        }
4188    
4189        /* If previous is OP_FAIL, it was generated by an empty class [] in
4190        JavaScript mode. The other ways in which OP_FAIL can be generated, that is
4191        by (*FAIL) or (?!) set previous to NULL, which gives a "nothing to repeat"
4192        error above. We can just ignore the repeat in JS case. */
4193    
4194        else if (*previous == OP_FAIL) goto END_REPEAT;
4195    
4196      /* Else there's some kind of shambles */      /* Else there's some kind of shambles */
4197    
4198      else      else
# Line 4077  we set the flag only if there is a liter Line 4267  we set the flag only if there is a liter
4267      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4268      parenthesis forms.  */      parenthesis forms.  */
4269    
4270      case '(':      case CHAR_LEFT_PARENTHESIS:
4271      newoptions = options;      newoptions = options;
4272      skipbytes = 0;      skipbytes = 0;
4273      bravalue = OP_CBRA;      bravalue = OP_CBRA;
# Line 4086  we set the flag only if there is a liter Line 4276  we set the flag only if there is a liter
4276    
4277      /* First deal with various "verbs" that can be introduced by '*'. */      /* First deal with various "verbs" that can be introduced by '*'. */
4278    
4279      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == CHAR_ASTERISK && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4280        {        {
4281        int i, namelen;        int i, namelen;
4282        const char *vn = verbnames;        const char *vn = verbnames;
4283        const uschar *name = ++ptr;        const uschar *name = ++ptr;
4284        previous = NULL;        previous = NULL;
4285        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
4286        if (*ptr == ':')        if (*ptr == CHAR_COLON)
4287          {          {
4288          *errorcodeptr = ERR59;   /* Not supported */          *errorcodeptr = ERR59;   /* Not supported */
4289          goto FAILED;          goto FAILED;
4290          }          }
4291        if (*ptr != ')')        if (*ptr != CHAR_RIGHT_PARENTHESIS)
4292          {          {
4293          *errorcodeptr = ERR60;          *errorcodeptr = ERR60;
4294          goto FAILED;          goto FAILED;
# Line 4123  we set the flag only if there is a liter Line 4313  we set the flag only if there is a liter
4313      /* Deal with the extended parentheses; all are introduced by '?', and the      /* Deal with the extended parentheses; all are introduced by '?', and the
4314      appearance of any of them means that this is not a capturing group. */      appearance of any of them means that this is not a capturing group. */
4315    
4316      else if (*ptr == '?')      else if (*ptr == CHAR_QUESTION_MARK)
4317        {        {
4318        int i, set, unset, namelen;        int i, set, unset, namelen;
4319        int *optset;        int *optset;
# Line 4132  we set the flag only if there is a liter Line 4322  we set the flag only if there is a liter
4322    
4323        switch (*(++ptr))        switch (*(++ptr))
4324          {          {
4325          case '#':                 /* Comment; skip to ket */          case CHAR_NUMBER_SIGN:                 /* Comment; skip to ket */
4326          ptr++;          ptr++;
4327          while (*ptr != 0 && *ptr != ')') ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
4328          if (*ptr == 0)          if (*ptr == 0)
4329            {            {
4330            *errorcodeptr = ERR18;            *errorcodeptr = ERR18;
# Line 4144  we set the flag only if there is a liter Line 4334  we set the flag only if there is a liter
4334    
4335    
4336          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4337          case '|':                 /* Reset capture count for each branch */          case CHAR_VERTICAL_LINE:  /* Reset capture count for each branch */
4338          reset_bracount = TRUE;          reset_bracount = TRUE;
4339          /* Fall through */          /* Fall through */
4340    
4341          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4342          case ':':                 /* Non-capturing bracket */          case CHAR_COLON:          /* Non-capturing bracket */
4343          bravalue = OP_BRA;          bravalue = OP_BRA;
4344          ptr++;          ptr++;
4345          break;          break;
4346    
4347    
4348          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4349          case '(':          case CHAR_LEFT_PARENTHESIS:
4350          bravalue = OP_COND;       /* Conditional group */          bravalue = OP_COND;       /* Conditional group */
4351    
4352          /* A condition can be an assertion, a number (referring to a numbered          /* A condition can be an assertion, a number (referring to a numbered
# Line 4176  we set the flag only if there is a liter Line 4366  we set the flag only if there is a liter
4366          the switch. This will take control down to where bracketed groups,          the switch. This will take control down to where bracketed groups,
4367          including assertions, are processed. */          including assertions, are processed. */
4368    
4369          if (ptr[1] == '?' && (ptr[2] == '=' || ptr[2] == '!' || ptr[2] == '<'))          if (ptr[1] == CHAR_QUESTION_MARK && (ptr[2] == CHAR_EQUALS_SIGN ||
4370                ptr[2] == CHAR_EXCLAMATION_MARK || ptr[2] == CHAR_LESS_THAN_SIGN))
4371            break;            break;
4372    
4373          /* Most other conditions use OP_CREF (a couple change to OP_RREF          /* Most other conditions use OP_CREF (a couple change to OP_RREF
# Line 4188  we set the flag only if there is a liter Line 4379  we set the flag only if there is a liter
4379    
4380          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4381    
4382          if (ptr[1] == 'R' && ptr[2] == '&')          if (ptr[1] == CHAR_R && ptr[2] == CHAR_AMPERSAND)
4383            {            {
4384            terminator = -1;            terminator = -1;
4385            ptr += 2;            ptr += 2;
# Line 4198  we set the flag only if there is a liter Line 4389  we set the flag only if there is a liter
4389          /* Check for a test for a named group's having been set, using the Perl          /* Check for a test for a named group's having been set, using the Perl
4390          syntax (?(<name>) or (?('name') */          syntax (?(<name>) or (?('name') */
4391    
4392          else if (ptr[1] == '<')          else if (ptr[1] == CHAR_LESS_THAN_SIGN)
4393            {            {
4394            terminator = '>';            terminator = CHAR_GREATER_THAN_SIGN;
4395            ptr++;            ptr++;
4396            }            }
4397          else if (ptr[1] == '\'')          else if (ptr[1] == CHAR_APOSTROPHE)
4398            {            {
4399            terminator = '\'';            terminator = CHAR_APOSTROPHE;
4400            ptr++;            ptr++;
4401            }            }
4402          else          else
4403            {            {
4404            terminator = 0;            terminator = 0;
4405            if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);
4406            }            }
4407    
4408          /* We now expect to read a name; any thing else is an error */          /* We now expect to read a name; any thing else is an error */
# Line 4231  we set the flag only if there is a liter Line 4422  we set the flag only if there is a liter
4422            {            {
4423            if (recno >= 0)            if (recno >= 0)
4424              recno = ((digitab[*ptr] & ctype_digit) != 0)?              recno = ((digitab[*ptr] & ctype_digit) != 0)?
4425                recno * 10 + *ptr - '0' : -1;                recno * 10 + *ptr - CHAR_0 : -1;
4426            ptr++;            ptr++;
4427            }            }
4428          namelen = ptr - name;          namelen = ptr - name;
4429    
4430          if ((terminator > 0 && *ptr++ != terminator) || *ptr++ != ')')          if ((terminator > 0 && *ptr++ != terminator) ||
4431                *ptr++ != CHAR_RIGHT_PARENTHESIS)
4432            {            {
4433            ptr--;      /* Error offset */            ptr--;      /* Error offset */
4434            *errorcodeptr = ERR26;            *errorcodeptr = ERR26;
# Line 4258  we set the flag only if there is a liter Line 4450  we set the flag only if there is a liter
4450              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4451              goto FAILED;              goto FAILED;
4452              }              }
4453            recno = (refsign == '-')?            recno = (refsign == CHAR_MINUS)?
4454              cd->bracount - recno + 1 : recno +cd->bracount;              cd->bracount - recno + 1 : recno +cd->bracount;
4455            if (recno <= 0 || recno > cd->final_bracount)            if (recno <= 0 || recno > cd->final_bracount)
4456              {              {
# Line 4289  we set the flag only if there is a liter Line 4481  we set the flag only if there is a liter
4481    
4482          /* Search the pattern for a forward reference */          /* Search the pattern for a forward reference */
4483    
4484          else if ((i = find_parens(ptr, cd->bracount, name, namelen,          else if ((i = find_parens(ptr, cd, name, namelen,
4485                          (options & PCRE_EXTENDED) != 0)) > 0)                          (options & PCRE_EXTENDED) != 0)) > 0)
4486            {            {
4487            PUT2(code, 2+LINK_SIZE, i);            PUT2(code, 2+LINK_SIZE, i);
# Line 4310  we set the flag only if there is a liter Line 4502  we set the flag only if there is a liter
4502          /* Check for (?(R) for recursion. Allow digits after R to specify a          /* Check for (?(R) for recursion. Allow digits after R to specify a
4503          specific group number. */          specific group number. */
4504    
4505          else if (*name == 'R')          else if (*name == CHAR_R)
4506            {            {
4507            recno = 0;            recno = 0;
4508            for (i = 1; i < namelen; i++)            for (i = 1; i < namelen; i++)
# Line 4320  we set the flag only if there is a liter Line 4512  we set the flag only if there is a liter
4512                *errorcodeptr = ERR15;                *errorcodeptr = ERR15;
4513                goto FAILED;                goto FAILED;
4514                }                }
4515              recno = recno * 10 + name[i] - '0';              recno = recno * 10 + name[i] - CHAR_0;
4516              }              }
4517            if (recno == 0) recno = RREF_ANY;            if (recno == 0) recno = RREF_ANY;
4518            code[1+LINK_SIZE] = OP_RREF;      /* Change test type */            code[1+LINK_SIZE] = OP_RREF;      /* Change test type */
# Line 4330  we set the flag only if there is a liter Line 4522  we set the flag only if there is a liter
4522          /* Similarly, check for the (?(DEFINE) "condition", which is always          /* Similarly, check for the (?(DEFINE) "condition", which is always
4523          false. */          false. */
4524    
4525          else if (namelen == 6 && strncmp((char *)name, "DEFINE", 6) == 0)          else if (namelen == 6 && strncmp((char *)name, STRING_DEFINE, 6) == 0)
4526            {            {
4527            code[1+LINK_SIZE] = OP_DEF;            code[1+LINK_SIZE] = OP_DEF;
4528            skipbytes = 1;            skipbytes = 1;
# Line 4355  we set the flag only if there is a liter Line 4547  we set the flag only if there is a liter
4547    
4548    
4549          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4550          case '=':                 /* Positive lookahead */          case CHAR_EQUALS_SIGN:                 /* Positive lookahead */
4551          bravalue = OP_ASSERT;          bravalue = OP_ASSERT;
4552          ptr++;          ptr++;
4553          break;          break;
4554    
4555    
4556          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4557          case '!':                 /* Negative lookahead */          case CHAR_EXCLAMATION_MARK:            /* Negative lookahead */
4558          ptr++;          ptr++;
4559          if (*ptr == ')')          /* Optimize (?!) */          if (*ptr == CHAR_RIGHT_PARENTHESIS)    /* Optimize (?!) */
4560            {            {
4561            *code++ = OP_FAIL;            *code++ = OP_FAIL;
4562            previous = NULL;            previous = NULL;
# Line 4375  we set the flag only if there is a liter Line 4567  we set the flag only if there is a liter
4567    
4568    
4569          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4570          case '<':                 /* Lookbehind or named define */          case CHAR_LESS_THAN_SIGN:              /* Lookbehind or named define */
4571          switch (ptr[1])          switch (ptr[1])
4572            {            {
4573            case '=':               /* Positive lookbehind */            case CHAR_EQUALS_SIGN:               /* Positive lookbehind */
4574            bravalue = OP_ASSERTBACK;            bravalue = OP_ASSERTBACK;
4575            ptr += 2;            ptr += 2;
4576            break;            break;
4577    
4578            case '!':               /* Negative lookbehind */            case CHAR_EXCLAMATION_MARK:          /* Negative lookbehind */
4579            bravalue = OP_ASSERTBACK_NOT;            bravalue = OP_ASSERTBACK_NOT;
4580            ptr += 2;            ptr += 2;
4581            break;            break;
# Line 4398  we set the flag only if there is a liter Line 4590  we set the flag only if there is a liter
4590    
4591    
4592          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4593          case '>':                 /* One-time brackets */          case CHAR_GREATER_THAN_SIGN:           /* One-time brackets */
4594          bravalue = OP_ONCE;          bravalue = OP_ONCE;
4595          ptr++;          ptr++;
4596          break;          break;
4597    
4598    
4599          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4600          case 'C':                 /* Callout - may be followed by digits; */          case CHAR_C:                 /* Callout - may be followed by digits; */
4601          previous_callout = code;  /* Save for later completion */          previous_callout = code;  /* Save for later completion */
4602          after_manual_callout = 1; /* Skip one item before completing */          after_manual_callout = 1; /* Skip one item before completing */
4603          *code++ = OP_CALLOUT;          *code++ = OP_CALLOUT;
4604            {            {
4605            int n = 0;            int n = 0;
4606            while ((digitab[*(++ptr)] & ctype_digit) != 0)            while ((digitab[*(++ptr)] & ctype_digit) != 0)
4607              n = n * 10 + *ptr - '0';              n = n * 10 + *ptr - CHAR_0;
4608            if (*ptr != ')')            if (*ptr != CHAR_RIGHT_PARENTHESIS)
4609              {              {
4610              *errorcodeptr = ERR39;              *errorcodeptr = ERR39;
4611              goto FAILED;              goto FAILED;
# Line 4433  we set the flag only if there is a liter Line 4625  we set the flag only if there is a liter
4625    
4626    
4627          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4628          case 'P':                 /* Python-style named subpattern handling */          case CHAR_P:              /* Python-style named subpattern handling */
4629          if (*(++ptr) == '=' || *ptr == '>')  /* Reference or recursion */          if (*(++ptr) == CHAR_EQUALS_SIGN ||
4630                *ptr == CHAR_GREATER_THAN_SIGN)  /* Reference or recursion */
4631            {            {
4632            is_recurse = *ptr == '>';            is_recurse = *ptr == CHAR_GREATER_THAN_SIGN;
4633            terminator = ')';            terminator = CHAR_RIGHT_PARENTHESIS;
4634            goto NAMED_REF_OR_RECURSE;            goto NAMED_REF_OR_RECURSE;
4635            }            }
4636          else if (*ptr != '<')    /* Test for Python-style definition */          else if (*ptr != CHAR_LESS_THAN_SIGN)  /* Test for Python-style defn */
4637            {            {
4638            *errorcodeptr = ERR41;            *errorcodeptr = ERR41;
4639            goto FAILED;            goto FAILED;
# Line 4450  we set the flag only if there is a liter Line 4643  we set the flag only if there is a liter
4643    
4644          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4645          DEFINE_NAME:    /* Come here from (?< handling */          DEFINE_NAME:    /* Come here from (?< handling */
4646          case '\'':          case CHAR_APOSTROPHE:
4647            {            {
4648            terminator = (*ptr == '<')? '>' : '\'';            terminator = (*ptr == CHAR_LESS_THAN_SIGN)?
4649                CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
4650            name = ++ptr;            name = ++ptr;
4651    
4652            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
# Line 4526  we set the flag only if there is a liter Line 4720  we set the flag only if there is a liter
4720    
4721    
4722          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4723          case '&':                 /* Perl recursion/subroutine syntax */          case CHAR_AMPERSAND:            /* Perl recursion/subroutine syntax */
4724          terminator = ')';          terminator = CHAR_RIGHT_PARENTHESIS;
4725          is_recurse = TRUE;          is_recurse = TRUE;
4726          /* Fall through */          /* Fall through */
4727    
# Line 4535  we set the flag only if there is a liter Line 4729  we set the flag only if there is a liter
4729          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4730          through from the Perl recursion syntax (?&name). We also come here from          through from the Perl recursion syntax (?&name). We also come here from
4731          the Perl \k<name> or \k'name' back reference syntax and the \k{name}          the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4732          .NET syntax. */          .NET syntax, and the Oniguruma \g<...> and \g'...' subroutine syntax. */
4733    
4734          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4735          name = ++ptr;          name = ++ptr;
# Line 4586  we set the flag only if there is a liter Line 4780  we set the flag only if there is a liter
4780              recno = GET2(slot, 0);              recno = GET2(slot, 0);
4781              }              }
4782            else if ((recno =                /* Forward back reference */            else if ((recno =                /* Forward back reference */
4783                      find_parens(ptr, cd->bracount, name, namelen,                      find_parens(ptr, cd, name, namelen,
4784                        (options & PCRE_EXTENDED) != 0)) <= 0)                        (options & PCRE_EXTENDED) != 0)) <= 0)
4785              {              {
4786              *errorcodeptr = ERR15;              *errorcodeptr = ERR15;
# Line 4602  we set the flag only if there is a liter Line 4796  we set the flag only if there is a liter
4796    
4797    
4798          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4799          case 'R':                 /* Recursion */          case CHAR_R:              /* Recursion */
4800          ptr++;                    /* Same as (?0)      */          ptr++;                    /* Same as (?0)      */
4801          /* Fall through */          /* Fall through */
4802    
4803    
4804          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4805          case '-': case '+':          case CHAR_MINUS: case CHAR_PLUS:  /* Recursion or subroutine */
4806          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */          case CHAR_0: case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4:
4807          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case CHAR_5: case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:
4808            {            {
4809            const uschar *called;            const uschar *called;
4810              terminator = CHAR_RIGHT_PARENTHESIS;
4811    
4812              /* Come here from the \g<...> and \g'...' code (Oniguruma
4813              compatibility). However, the syntax has been checked to ensure that
4814              the ... are a (signed) number, so that neither ERR63 nor ERR29 will
4815              be called on this path, nor with the jump to OTHER_CHAR_AFTER_QUERY
4816              ever be taken. */
4817    
4818              HANDLE_NUMERICAL_RECURSION:
4819    
4820            if ((refsign = *ptr) == '+')            if ((refsign = *ptr) == CHAR_PLUS)
4821              {              {
4822              ptr++;              ptr++;
4823              if ((digitab[*ptr] & ctype_digit) == 0)              if ((digitab[*ptr] & ctype_digit) == 0)
# Line 4623  we set the flag only if there is a liter Line 4826  we set the flag only if there is a liter
4826                goto FAILED;                goto FAILED;
4827                }                }
4828              }              }
4829            else if (refsign == '-')            else if (refsign == CHAR_MINUS)
4830              {              {
4831              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
4832                goto OTHER_CHAR_AFTER_QUERY;                goto OTHER_CHAR_AFTER_QUERY;
# Line 4632  we set the flag only if there is a liter Line 4835  we set the flag only if there is a liter
4835    
4836            recno = 0;            recno = 0;
4837            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4838              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - CHAR_0;
4839    
4840            if (*ptr != ')')            if (*ptr != terminator)
4841              {              {
4842              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4843              goto FAILED;              goto FAILED;
4844              }              }
4845    
4846            if (refsign == '-')            if (refsign == CHAR_MINUS)
4847              {              {
4848              if (recno == 0)              if (recno == 0)
4849                {                {
# Line 4654  we set the flag only if there is a liter Line 4857  we set the flag only if there is a liter
4857                goto FAILED;                goto FAILED;
4858                }                }
4859              }              }
4860            else if (refsign == '+')            else if (refsign == CHAR_PLUS)
4861              {              {
4862              if (recno == 0)              if (recno == 0)
4863                {                {
# Line 4687  we set the flag only if there is a liter Line 4890  we set the flag only if there is a liter
4890    
4891              if (called == NULL)              if (called == NULL)
4892                {                {
4893                if (find_parens(ptr, cd->bracount, NULL, recno,                if (find_parens(ptr, cd, NULL, recno,
4894                     (options & PCRE_EXTENDED) != 0) < 0)                      (options & PCRE_EXTENDED) != 0) < 0)
4895                  {                  {
4896                  *errorcodeptr = ERR15;                  *errorcodeptr = ERR15;
4897                  goto FAILED;                  goto FAILED;
# Line 4740  we set the flag only if there is a liter Line 4943  we set the flag only if there is a liter
4943          set = unset = 0;          set = unset = 0;
4944          optset = &set;          optset = &set;
4945    
4946          while (*ptr != ')' && *ptr != ':')          while (*ptr != CHAR_RIGHT_PARENTHESIS && *ptr != CHAR_COLON)
4947            {            {
4948            switch (*ptr++)            switch (*ptr++)
4949              {              {
4950              case '-': optset = &unset; break;              case CHAR_MINUS: optset = &unset; break;
4951    
4952              case 'J':    /* Record that it changed in the external options */              case CHAR_J:    /* Record that it changed in the external options */
4953              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4954              cd->external_flags |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4955              break;              break;
4956    
4957              case 'i': *optset |= PCRE_CASELESS; break;              case CHAR_i: *optset |= PCRE_CASELESS; break;
4958              case 'm': *optset |= PCRE_MULTILINE; break;              case CHAR_m: *optset |= PCRE_MULTILINE; break;
4959              case 's': *optset |= PCRE_DOTALL; break;              case CHAR_s: *optset |= PCRE_DOTALL; break;
4960              case 'x': *optset |= PCRE_EXTENDED; break;              case CHAR_x: *optset |= PCRE_EXTENDED; break;
4961              case 'U': *optset |= PCRE_UNGREEDY; break;              case CHAR_U: *optset |= PCRE_UNGREEDY; break;
4962              case 'X': *optset |= PCRE_EXTRA; break;              case CHAR_X: *optset |= PCRE_EXTRA; break;
4963    
4964              default:  *errorcodeptr = ERR12;              default:  *errorcodeptr = ERR12;
4965                        ptr--;    /* Correct the offset */                        ptr--;    /* Correct the offset */
# Line 4787  we set the flag only if there is a liter Line 4990  we set the flag only if there is a liter
4990          both phases.          both phases.
4991    
4992          If we are not at the pattern start, compile code to change the ims          If we are not at the pattern start, compile code to change the ims
4993          options if this setting actually changes any of them. We also pass the          options if this setting actually changes any of them, and reset the
4994          new setting back so that it can be put at the start of any following          greedy defaults and the case value for firstbyte and reqbyte. */
         branches, and when this group ends (if we are in a group), a resetting  
         item can be compiled. */  
4995    
4996          if (*ptr == ')')          if (*ptr == CHAR_RIGHT_PARENTHESIS)
4997            {            {
4998            if (code == cd->start_code + 1 + LINK_SIZE &&            if (code == cd->start_code + 1 + LINK_SIZE &&
4999                 (lengthptr == NULL || *lengthptr == 2 + 2*LINK_SIZE))                 (lengthptr == NULL || *lengthptr == 2 + 2*LINK_SIZE))
5000              {              {
5001              cd->external_options = newoptions;              cd->external_options = newoptions;
             options = newoptions;  
5002              }              }
5003           else           else
5004              {              {
# Line 4807  we set the flag only if there is a liter Line 5007  we set the flag only if there is a liter
5007                *code++ = OP_OPT;                *code++ = OP_OPT;
5008                *code++ = newoptions & PCRE_IMS;                *code++ = newoptions & PCRE_IMS;
5009                }                }
   
             /* Change options at this level, and pass them back for use  
             in subsequent branches. Reset the greedy defaults and the case  
             value for firstbyte and reqbyte. */  
   
             *optionsptr = options = newoptions;  
5010              greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);              greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);
5011              greedy_non_default = greedy_default ^ 1;              greedy_non_default = greedy_default ^ 1;
5012              req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;              req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;
5013              }              }
5014    
5015              /* Change options at this level, and pass them back for use
5016              in subsequent branches. When not at the start of the pattern, this
5017              information is also necessary so that a resetting item can be
5018              compiled at the end of a group (if we are in a group). */
5019    
5020              *optionsptr = options = newoptions;
5021            previous = NULL;       /* This item can't be repeated */            previous = NULL;       /* This item can't be repeated */
5022            continue;              /* It is complete */            continue;              /* It is complete */
5023            }            }
# Line 4933  we set the flag only if there is a liter Line 5133  we set the flag only if there is a liter
5133    
5134      /* Error if hit end of pattern */      /* Error if hit end of pattern */
5135    
5136      if (*ptr != ')')      if (*ptr != CHAR_RIGHT_PARENTHESIS)
5137        {        {
5138        *errorcodeptr = ERR14;        *errorcodeptr = ERR14;
5139        goto FAILED;        goto FAILED;
# Line 5031  we set the flag only if there is a liter Line 5231  we set the flag only if there is a liter
5231      We can test for values between ESC_b and ESC_Z for the latter; this may      We can test for values between ESC_b and ESC_Z for the latter; this may
5232      have to change if any new ones are ever created. */      have to change if any new ones are ever created. */
5233    
5234      case '\\':      case CHAR_BACKSLASH:
5235      tempptr = ptr;      tempptr = ptr;
5236      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);
5237      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
# Line 5040  we set the flag only if there is a liter Line 5240  we set the flag only if there is a liter
5240        {        {
5241        if (-c == ESC_Q)            /* Handle start of quoted string */        if (-c == ESC_Q)            /* Handle start of quoted string */
5242          {          {
5243          if (ptr[1] == '\\' && ptr[2] == 'E') ptr += 2; /* avoid empty string */          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
5244            else inescq = TRUE;            ptr += 2;               /* avoid empty string */
5245                else inescq = TRUE;
5246          continue;          continue;
5247          }          }
5248    
# Line 5058  we set the flag only if there is a liter Line 5259  we set the flag only if there is a liter
5259        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
5260        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
5261    
5262          /* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'
5263          is a subroutine call by number (Oniguruma syntax). In fact, the value
5264          -ESC_g is returned only for these cases. So we don't need to check for <
5265          or ' if the value is -ESC_g. For the Perl syntax \g{n} the value is
5266          -ESC_REF+n, and for the Perl syntax \g{name} the result is -ESC_k (as
5267          that is a synonym for a named back reference). */
5268    
5269          if (-c == ESC_g)
5270            {
5271            const uschar *p;
5272            save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
5273            terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?
5274              CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
5275    
5276            /* These two statements stop the compiler for warning about possibly
5277            unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In
5278            fact, because we actually check for a number below, the paths that
5279            would actually be in error are never taken. */
5280    
5281            skipbytes = 0;
5282            reset_bracount = FALSE;
5283    
5284            /* Test for a name */
5285    
5286            if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)
5287              {
5288              BOOL isnumber = TRUE;
5289              for (p = ptr + 1; *p != 0 && *p != terminator; p++)
5290                {
5291                if ((cd->ctypes[*p] & ctype_digit) == 0) isnumber = FALSE;
5292                if ((cd->ctypes[*p] & ctype_word) == 0) break;
5293                }
5294              if (*p != terminator)
5295                {
5296                *errorcodeptr = ERR57;
5297                break;
5298                }
5299              if (isnumber)
5300                {
5301                ptr++;
5302                goto HANDLE_NUMERICAL_RECURSION;
5303                }
5304              is_recurse = TRUE;
5305              goto NAMED_REF_OR_RECURSE;
5306              }
5307    
5308            /* Test a signed number in angle brackets or quotes. */
5309    
5310            p = ptr + 2;
5311            while ((digitab[*p] & ctype_digit) != 0) p++;
5312            if (*p != terminator)
5313              {
5314              *errorcodeptr = ERR57;
5315              break;
5316              }
5317            ptr++;
5318            goto HANDLE_NUMERICAL_RECURSION;
5319            }
5320    
5321        /* \k<name> or \k'name' is a back reference by name (Perl syntax).        /* \k<name> or \k'name' is a back reference by name (Perl syntax).
5322        We also support \k{name} (.NET syntax) */        We also support \k{name} (.NET syntax) */
5323    
5324        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\'' || ptr[1] == '{'))        if (-c == ESC_k && (ptr[1] == CHAR_LESS_THAN_SIGN ||
5325              ptr[1] == CHAR_APOSTROPHE || ptr[1] == CHAR_LEFT_CURLY_BRACKET))
5326          {          {
5327          is_recurse = FALSE;          is_recurse = FALSE;
5328          terminator = (*(++ptr) == '<')? '>' : (*ptr == '\'')? '\'' : '}';          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?
5329              CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)?
5330              CHAR_APOSTROPHE : CHAR_RIGHT_CURLY_BRACKET;
5331          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
5332          }          }
5333    
# Line 5167  we set the flag only if there is a liter Line 5430  we set the flag only if there is a liter
5430    
5431      /* Remember if \r or \n were seen */      /* Remember if \r or \n were seen */
5432    
5433      if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')      if (mcbuffer[0] == CHAR_CR || mcbuffer[0] == CHAR_NL)
5434        cd->external_flags |= PCRE_HASCRORLF;        cd->external_flags |= PCRE_HASCRORLF;
5435    
5436      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
# Line 5412  for (;;) Line 5675  for (;;)
5675    compile a resetting op-code following, except at the very end of the pattern.    compile a resetting op-code following, except at the very end of the pattern.
5676    Return leaving the pointer at the terminating char. */    Return leaving the pointer at the terminating char. */
5677    
5678    if (*ptr != '|')    if (*ptr != CHAR_VERTICAL_LINE)
5679      {      {
5680      if (lengthptr == NULL)      if (lengthptr == NULL)
5681        {        {
# Line 5435  for (;;) Line 5698  for (;;)
5698    
5699      /* Resetting option if needed */      /* Resetting option if needed */
5700    
5701      if ((options & PCRE_IMS) != oldims && *ptr == ')')      if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS)
5702        {        {
5703        *code++ = OP_OPT;        *code++ = OP_OPT;
5704        *code++ = oldims;        *code++ = oldims;
# Line 5564  do { Line 5827  do {
5827       if (!is_anchored(scode, options, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, options, bracket_map, backref_map)) return FALSE;
5828       }       }
5829    
5830     /* .* is not anchored unless DOTALL is set and it isn't in brackets that     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
5831     are or may be referenced. */     it isn't in brackets that are or may be referenced. */
5832    
5833     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
5834               op == OP_TYPEPOSSTAR) &&               op == OP_TYPEPOSSTAR))
             (*options & PCRE_DOTALL) != 0)  
5835       {       {
5836       if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE;       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)
5837           return FALSE;
5838       }       }
5839    
5840     /* Check for explicit anchoring */     /* Check for explicit anchoring */
# Line 5616  do { Line 5879  do {
5879     const uschar *scode = first_significant_code(code + _pcre_OP_lengths[*code],     const uschar *scode = first_significant_code(code + _pcre_OP_lengths[*code],
5880       NULL, 0, FALSE);       NULL, 0, FALSE);
5881     register int op = *scode;     register int op = *scode;
5882    
5883       /* If we are at the start of a conditional assertion group, *both* the
5884       conditional assertion *and* what follows the condition must satisfy the test
5885       for start of line. Other kinds of condition fail. Note that there may be an
5886       auto-callout at the start of a condition. */
5887    
5888       if (op == OP_COND)
5889         {
5890         scode += 1 + LINK_SIZE;
5891         if (*scode == OP_CALLOUT) scode += _pcre_OP_lengths[OP_CALLOUT];
5892         switch (*scode)
5893           {
5894           case OP_CREF:
5895           case OP_RREF:
5896           case OP_DEF:
5897           return FALSE;
5898    
5899           default:     /* Assertion */
5900           if (!is_startline(scode, bracket_map, backref_map)) return FALSE;
5901           do scode += GET(scode, 1); while (*scode == OP_ALT);
5902           scode += 1 + LINK_SIZE;
5903           break;
5904           }
5905         scode = first_significant_code(scode, NULL, 0, FALSE);
5906         op = *scode;
5907         }
5908    
5909     /* Non-capturing brackets */     /* Non-capturing brackets */
5910    
# Line 5635  do { Line 5924  do {
5924    
5925     /* Other brackets */     /* Other brackets */
5926    
5927     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_COND)     else if (op == OP_ASSERT || op == OP_ONCE)
5928       { if (!is_startline(scode, bracket_map, backref_map)) return FALSE; }       {
5929         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;
5930         }
5931    
5932     /* .* means "start at start or after \n" if it isn't in brackets that     /* .* means "start at start or after \n" if it isn't in brackets that
5933     may be referenced. */     may be referenced. */
# Line 5753  Returns:        pointer to compiled data Line 6044  Returns:        pointer to compiled data
6044                  with errorptr and erroroffset set                  with errorptr and erroroffset set
6045  */  */
6046    
6047  PCRE_EXP_DEFN pcre *  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
6048  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
6049    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
6050  {  {
# Line 5761  return pcre_compile2(pattern, options, N Line 6052  return pcre_compile2(pattern, options, N
6052  }  }
6053    
6054    
6055  PCRE_EXP_DEFN pcre *  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
6056  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
6057    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
6058  {  {
# Line 5788  to fill in forward references to subpatt Line 6079  to fill in forward references to subpatt
6079    
6080  uschar cworkspace[COMPILE_WORK_SIZE];  uschar cworkspace[COMPILE_WORK_SIZE];
6081    
   
6082  /* Set this early so that early errors get offset 0. */  /* Set this early so that early errors get offset 0. */
6083    
6084  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern;
# Line 5834  if ((options & PCRE_UTF8) != 0) Line 6124  if ((options & PCRE_UTF8) != 0)
6124    }    }
6125  #endif  #endif
6126    
6127  if ((options & ~PUBLIC_OPTIONS) != 0)  if ((options & ~PUBLIC_COMPILE_OPTIONS) != 0)
6128    {    {
6129    errorcode = ERR17;    errorcode = ERR17;
6130    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
# Line 5851  cd->ctypes = tables + ctypes_offset; Line 6141  cd->ctypes = tables + ctypes_offset;
6141  /* Check for global one-time settings at the start of the pattern, and remember  /* Check for global one-time settings at the start of the pattern, and remember
6142  the offset for later. */  the offset for later. */
6143    
6144  while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')  while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS &&
6145           ptr[skipatstart+1] == CHAR_ASTERISK)
6146    {    {
6147    int newnl = 0;    int newnl = 0;
6148    int newbsr = 0;    int newbsr = 0;
6149    
6150    if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)
6151      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
6152    else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)    else if (strncmp((char *)(ptr+skipatstart+2), STRING_LF_RIGHTPAR, 3)  == 0)
6153      { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }      { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
6154    else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)    else if (strncmp((char *)(ptr+skipatstart+2), STRING_CRLF_RIGHTPAR, 5)  == 0)
6155      { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }      { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
6156    else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)    else if (strncmp((char *)(ptr+skipatstart+2), STRING_ANY_RIGHTPAR, 4) == 0)
6157      { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }      { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
6158    else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)    else if (strncmp((char *)(ptr+skipatstart+2), STRING_ANYCRLF_RIGHTPAR, 8) == 0)
6159      { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }      { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
6160    
6161    else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)    else if (strncmp((char *)(ptr+skipatstart+2), STRING_BSR_ANYCRLF_RIGHTPAR, 12) == 0)
6162      { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }      { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
6163    else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)    else if (strncmp((char *)(ptr+skipatstart+2), STRING_BSR_UNICODE_RIGHTPAR, 12) == 0)
6164      { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }      { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
6165    
6166    if (newnl != 0)    if (newnl != 0)
# Line 5897  current code allows for fixed one- or tw Line 6188  current code allows for fixed one- or tw
6188  switch (options & PCRE_NEWLINE_BITS)  switch (options & PCRE_NEWLINE_BITS)
6189    {    {
6190    case 0: newline = NEWLINE; break;   /* Build-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
6191    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = CHAR_CR; break;
6192    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = CHAR_NL; break;
6193    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
6194         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break;
6195    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
6196    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
6197    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;

Legend:
Removed from v.286  
changed lines
  Added in v.391

  ViewVC Help
Powered by ViewVC 1.1.5