/[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 97 by ph10, Mon Mar 5 12:36:47 2007 UTC revision 224 by ph10, Mon Aug 20 12:58:29 2007 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) 1997-2006 University of Cambridge             Copyright (c) 1997-2007 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 42  POSSIBILITY OF SUCH DAMAGE. Line 42  POSSIBILITY OF SUCH DAMAGE.
42  supporting internal functions that are not used by other modules. */  supporting internal functions that are not used by other modules. */
43    
44    
45    #ifdef HAVE_CONFIG_H
46    #include <config.h>
47    #endif
48    
49  #define NLBLOCK cd             /* Block containing newline information */  #define NLBLOCK cd             /* Block containing newline information */
50  #define PSSTART start_pattern  /* Field containing processed string start */  #define PSSTART start_pattern  /* Field containing processed string start */
51  #define PSEND   end_pattern    /* Field containing processed string end */  #define PSEND   end_pattern    /* Field containing processed string end */
52    
   
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
# Line 58  used by pcretest. DEBUG is not defined w Line 61  used by pcretest. DEBUG is not defined w
61  #endif  #endif
62    
63    
64    /* Macro for setting individual bits in class bitmaps. */
65    
66    #define SETBIT(a,b) a[b/8] |= (1 << (b%8))
67    
68    /* Maximum length value to check against when making sure that the integer that
69    holds the compiled pattern length does not overflow. We make it a bit less than
70    INT_MAX to allow for adding in group terminating bytes, so that we don't have
71    to check them every time. */
72    
73    #define OFLOW_MAX (INT_MAX - 20)
74    
75    
76  /*************************************************  /*************************************************
77  *      Code parameters and static tables         *  *      Code parameters and static tables         *
78  *************************************************/  *************************************************/
# Line 87  static const short int escapes[] = { Line 102  static const short int escapes[] = {
102       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */
103       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */
104     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */
105       0,      0,      0,      0,      0,      0,      0,      0,   /* H - O */  -ESC_H,      0,      0, -ESC_K,      0,      0,      0,      0,   /* H - O */
106  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0,      0, -ESC_W,   /* P - W */  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0, -ESC_V, -ESC_W,   /* P - W */
107  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */
108     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */
109       0,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */  -ESC_h,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */
110  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0,      0, -ESC_w,   /* p - w */  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0, -ESC_v, -ESC_w,   /* p - w */
111       0,      0, -ESC_z                                            /* x - z */       0,      0, -ESC_z                                            /* x - z */
112  };  };
113    
# Line 106  static const short int escapes[] = { Line 121  static const short int escapes[] = {
121  /*  70 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  70 */     0,     0,      0,       0,      0,     0,      0,      0,
122  /*  78 */     0,   '`',    ':',     '#',    '@',  '\'',    '=',    '"',  /*  78 */     0,   '`',    ':',     '#',    '@',  '\'',    '=',    '"',
123  /*  80 */     0,     7, -ESC_b,       0, -ESC_d, ESC_e,  ESC_f,      0,  /*  80 */     0,     7, -ESC_b,       0, -ESC_d, ESC_e,  ESC_f,      0,
124  /*  88 */     0,     0,      0,     '{',      0,     0,      0,      0,  /*  88 */-ESC_h,     0,      0,     '{',      0,     0,      0,      0,
125  /*  90 */     0,     0, -ESC_k,     'l',      0, ESC_n,      0, -ESC_p,  /*  90 */     0,     0, -ESC_k,     'l',      0, ESC_n,      0, -ESC_p,
126  /*  98 */     0, ESC_r,      0,     '}',      0,     0,      0,      0,  /*  98 */     0, ESC_r,      0,     '}',      0,     0,      0,      0,
127  /*  A0 */     0,   '~', -ESC_s, ESC_tee,      0,     0, -ESC_w,      0,  /*  A0 */     0,   '~', -ESC_s, ESC_tee,      0,-ESC_v, -ESC_w,      0,
128  /*  A8 */     0,-ESC_z,      0,       0,      0,   '[',      0,      0,  /*  A8 */     0,-ESC_z,      0,       0,      0,   '[',      0,      0,
129  /*  B0 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  B0 */     0,     0,      0,       0,      0,     0,      0,      0,
130  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',
131  /*  C0 */   '{',-ESC_A, -ESC_B,  -ESC_C, -ESC_D,-ESC_E,      0, -ESC_G,  /*  C0 */   '{',-ESC_A, -ESC_B,  -ESC_C, -ESC_D,-ESC_E,      0, -ESC_G,
132  /*  C8 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,
133  /*  D0 */   '}',     0,      0,       0,      0,     0,      0, -ESC_P,  /*  D0 */   '}',     0, -ESC_K,       0,      0,     0,      0, -ESC_P,
134  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,
135  /*  E0 */  '\\',     0, -ESC_S,       0,      0,     0, -ESC_W, -ESC_X,  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,
136  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,
137  /*  F0 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  F0 */     0,     0,      0,       0,      0,     0,      0,      0,
138  /*  F8 */     0,     0,      0,       0,      0,     0,      0,      0  /*  F8 */     0,     0,      0,       0,      0,     0,      0,      0
# Line 125  static const short int escapes[] = { Line 140  static const short int escapes[] = {
140  #endif  #endif
141    
142    
143    /* Table of special "verbs" like (*PRUNE) */
144    
145    typedef struct verbitem {
146      const char *name;
147      int   len;
148      int   op;
149    } verbitem;
150    
151    static verbitem verbs[] = {
152      { "ACCEPT", 6, OP_ACCEPT },
153      { "COMMIT", 6, OP_COMMIT },
154      { "F",      1, OP_FAIL },
155      { "FAIL",   4, OP_FAIL },
156      { "PRUNE",  5, OP_PRUNE },
157      { "SKIP",   4, OP_SKIP  },
158      { "THEN",   4, OP_THEN  }
159    };
160    
161    static int verbcount = sizeof(verbs)/sizeof(verbitem);
162    
163    
164  /* Tables of names of POSIX character classes and their lengths. The list is  /* Tables of names of POSIX character classes and their lengths. The list is
165  terminated by a zero length entry. The first three must be alpha, lower, upper,  terminated by a zero length entry. The first three must be alpha, lower, upper,
166  as this is assumed for handling case independence. */  as this is assumed for handling case independence. */
# Line 198  static const char *error_texts[] = { Line 234  static const char *error_texts[] = {
234    "missing ) after comment",    "missing ) after comment",
235    "parentheses nested too deeply",  /** DEAD **/    "parentheses nested too deeply",  /** DEAD **/
236    /* 20 */    /* 20 */
237    "regular expression too large",    "regular expression is too large",
238    "failed to get memory",    "failed to get memory",
239    "unmatched parentheses",    "unmatched parentheses",
240    "internal error: code overflow",    "internal error: code overflow",
# Line 208  static const char *error_texts[] = { Line 244  static const char *error_texts[] = {
244    "malformed number or name after (?(",    "malformed number or name after (?(",
245    "conditional group contains more than two branches",    "conditional group contains more than two branches",
246    "assertion expected after (?(",    "assertion expected after (?(",
247    "(?R or (?digits must be followed by )",    "(?R or (?[+-]digits must be followed by )",
248    /* 30 */    /* 30 */
249    "unknown POSIX class name",    "unknown POSIX class name",
250    "POSIX collating elements are not supported",    "POSIX collating elements are not supported",
# Line 234  static const char *error_texts[] = { Line 270  static const char *error_texts[] = {
270    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",
271    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",
272    /* 50 */    /* 50 */
273    "repeated subpattern is too long",    "repeated subpattern is too long",    /** DEAD **/
274    "octal value is greater than \\377 (not in UTF-8 mode)",    "octal value is greater than \\377 (not in UTF-8 mode)",
275    "internal error: overran compiling workspace",    "internal error: overran compiling workspace",
276    "internal error: previously-checked referenced subpattern not found",    "internal error: previously-checked referenced subpattern not found",
# Line 242  static const char *error_texts[] = { Line 278  static const char *error_texts[] = {
278    /* 55 */    /* 55 */
279    "repeating a DEFINE group is not allowed",    "repeating a DEFINE group is not allowed",
280    "inconsistent NEWLINE options",    "inconsistent NEWLINE options",
281    "\\g is not followed by an (optionally braced) non-zero number"    "\\g is not followed by a braced name or an optionally braced non-zero number",
282      "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number",
283      "(*VERB) with an argument is not supported",
284      /* 60 */
285      "(*VERB) not recognized",
286      "number is too big"
287  };  };
288    
289    
# Line 373  static const unsigned char ebcdic_charta Line 414  static const unsigned char ebcdic_charta
414  /* Definition to allow mutual recursion */  /* Definition to allow mutual recursion */
415    
416  static BOOL  static BOOL
417    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, int, int *,    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,
418      int *, branch_chain *, compile_data *, int *);      int *, int *, branch_chain *, compile_data *, int *);
419    
420    
421    
# Line 399  Arguments: Line 440  Arguments:
440    
441  Returns:         zero or positive => a data character  Returns:         zero or positive => a data character
442                   negative => a special escape sequence                   negative => a special escape sequence
443                   on error, errorptr is set                   on error, errorcodeptr is set
444  */  */
445    
446  static int  static int
# Line 452  else Line 493  else
493    
494      /* \g must be followed by a number, either plain or braced. If positive, it      /* \g must be followed by a number, either plain or braced. If positive, it
495      is an absolute backreference. If negative, it is a relative backreference.      is an absolute backreference. If negative, it is a relative backreference.
496      This is a Perl 5.10 feature. */      This is a Perl 5.10 feature. Perl 5.10 also supports \g{name} as a
497        reference to a named group. This is part of Perl's movement towards a
498        unified syntax for back references. As this is synonymous with \k{name}, we
499        fudge it up by pretending it really was \k. */
500    
501      case 'g':      case 'g':
502      if (ptr[1] == '{')      if (ptr[1] == '{')
503        {        {
504          const uschar *p;
505          for (p = ptr+2; *p != 0 && *p != '}'; p++)
506            if (*p != '-' && (digitab[*p] & ctype_digit) == 0) break;
507          if (*p != 0 && *p != '}')
508            {
509            c = -ESC_k;
510            break;
511            }
512        braced = TRUE;        braced = TRUE;
513        ptr++;        ptr++;
514        }        }
# Line 473  else Line 525  else
525      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
526        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - '0';
527    
528        if (c < 0)
529          {
530          *errorcodeptr = ERR61;
531          break;
532          }
533    
534      if (c == 0 || (braced && *(++ptr) != '}'))      if (c == 0 || (braced && *(++ptr) != '}'))
535        {        {
536        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
537        return 0;        break;
538        }        }
539    
540      if (negated)      if (negated)
# Line 484  else Line 542  else
542        if (c > bracount)        if (c > bracount)
543          {          {
544          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
545          return 0;          break;
546          }          }
547        c = bracount - (c - 1);        c = bracount - (c - 1);
548        }        }
# Line 513  else Line 571  else
571        c -= '0';        c -= '0';
572        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
573          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - '0';
574          if (c < 0)
575            {
576            *errorcodeptr = ERR61;
577            break;
578            }
579        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
580          {          {
581          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 608  else Line 671  else
671      if (c == 0)      if (c == 0)
672        {        {
673        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
674        return 0;        break;
675        }        }
676    
677  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
# Line 684  if (c == '{') Line 747  if (c == '{')
747      *negptr = TRUE;      *negptr = TRUE;
748      ptr++;      ptr++;
749      }      }
750    for (i = 0; i < sizeof(name) - 1; i++)    for (i = 0; i < (int)sizeof(name) - 1; i++)
751      {      {
752      c = *(++ptr);      c = *(++ptr);
753      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 887  for (; *ptr != 0; ptr++) Line 950  for (; *ptr != 0; ptr++)
950      {      {
951      while (*(++ptr) != ']')      while (*(++ptr) != ']')
952        {        {
953          if (*ptr == 0) return -1;
954        if (*ptr == '\\')        if (*ptr == '\\')
955          {          {
956          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 914  for (; *ptr != 0; ptr++) Line 978  for (; *ptr != 0; ptr++)
978    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
979    
980    if (*ptr != '(') continue;    if (*ptr != '(') continue;
981    if (ptr[1] != '?')    if (ptr[1] != '?' && ptr[1] != '*')
982      {      {
983      count++;      count++;
984      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 1042  for (;;) Line 1106  for (;;)
1106    {    {
1107    int d;    int d;
1108    register int op = *cc;    register int op = *cc;
   
1109    switch (op)    switch (op)
1110      {      {
1111      case OP_CBRA:      case OP_CBRA:
# Line 1131  for (;;) Line 1194  for (;;)
1194    
1195      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1196      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1197        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1198      cc += 4;      cc += 4;
1199      break;      break;
1200    
# Line 1239  for (;;) Line 1303  for (;;)
1303      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1304      }      }
1305    
1306    /* In UTF-8 mode, opcodes that are followed by a character may be followed by    /* Otherwise, we can get the item's length from the table, except that for
1307    a multi-byte character. The length in the table is a minimum, so we have to    repeated character types, we have to test for \p and \P, which have an extra
1308    arrange to skip the extra bytes. */    two bytes of parameters. */
1309    
1310    else    else
1311      {      {
1312        switch(c)
1313          {
1314          case OP_TYPESTAR:
1315          case OP_TYPEMINSTAR:
1316          case OP_TYPEPLUS:
1317          case OP_TYPEMINPLUS:
1318          case OP_TYPEQUERY:
1319          case OP_TYPEMINQUERY:
1320          case OP_TYPEPOSSTAR:
1321          case OP_TYPEPOSPLUS:
1322          case OP_TYPEPOSQUERY:
1323          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1324          break;
1325    
1326          case OP_TYPEUPTO:
1327          case OP_TYPEMINUPTO:
1328          case OP_TYPEEXACT:
1329          case OP_TYPEPOSUPTO:
1330          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1331          break;
1332          }
1333    
1334        /* Add in the fixed length from the table */
1335    
1336      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1337    
1338      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1339      a multi-byte character. The length in the table is a minimum, so we have to
1340      arrange to skip the extra bytes. */
1341    
1342    #ifdef SUPPORT_UTF8
1343      if (utf8) switch(c)      if (utf8) switch(c)
1344        {        {
1345        case OP_CHAR:        case OP_CHAR:
# Line 1266  for (;;) Line 1360  for (;;)
1360        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1361        break;        break;
1362        }        }
1363    #endif
1364      }      }
1365    }    }
1366  }  }
# Line 1301  for (;;) Line 1396  for (;;)
1396    
1397    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1398    
1399    /* Otherwise, we get the item's length from the table. In UTF-8 mode, opcodes    /* Otherwise, we can get the item's length from the table, except that for
1400    that are followed by a character may be followed by a multi-byte character.    repeated character types, we have to test for \p and \P, which have an extra
1401    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1402    
1403    else    else
1404      {      {
1405        switch(c)
1406          {
1407          case OP_TYPESTAR:
1408          case OP_TYPEMINSTAR:
1409          case OP_TYPEPLUS:
1410          case OP_TYPEMINPLUS:
1411          case OP_TYPEQUERY:
1412          case OP_TYPEMINQUERY:
1413          case OP_TYPEPOSSTAR:
1414          case OP_TYPEPOSPLUS:
1415          case OP_TYPEPOSQUERY:
1416          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1417          break;
1418    
1419          case OP_TYPEPOSUPTO:
1420          case OP_TYPEUPTO:
1421          case OP_TYPEMINUPTO:
1422          case OP_TYPEEXACT:
1423          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1424          break;
1425          }
1426    
1427        /* Add in the fixed length from the table */
1428    
1429      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1430    
1431        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1432        by a multi-byte character. The length in the table is a minimum, so we have
1433        to arrange to skip the extra bytes. */
1434    
1435    #ifdef SUPPORT_UTF8
1436      if (utf8) switch(c)      if (utf8) switch(c)
1437        {        {
1438        case OP_CHAR:        case OP_CHAR:
# Line 1329  for (;;) Line 1453  for (;;)
1453        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1454        break;        break;
1455        }        }
1456    #endif
1457      }      }
1458    }    }
1459  }  }
# Line 1366  for (code = first_significant_code(code Line 1491  for (code = first_significant_code(code
1491    
1492    c = *code;    c = *code;
1493    
1494    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    /* Groups with zero repeats can of course be empty; skip them. */
1495    
1496      if (c == OP_BRAZERO || c == OP_BRAMINZERO)
1497        {
1498        code += _pcre_OP_lengths[c];
1499        do code += GET(code, 1); while (*code == OP_ALT);
1500        c = *code;
1501        continue;
1502        }
1503    
1504      /* For other groups, scan the branches. */
1505    
1506      if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1507      {      {
1508      BOOL empty_branch;      BOOL empty_branch;
1509      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 1382  for (code = first_significant_code(code Line 1519  for (code = first_significant_code(code
1519        }        }
1520      while (*code == OP_ALT);      while (*code == OP_ALT);
1521      if (!empty_branch) return FALSE;   /* All branches are non-empty */      if (!empty_branch) return FALSE;   /* All branches are non-empty */
1522        c = *code;
     /* Move past the KET and fudge things so that the increment in the "for"  
     above has no effect. */  
   
     c = OP_END;  
     code += 1 + LINK_SIZE - _pcre_OP_lengths[c];  
1523      continue;      continue;
1524      }      }
1525    
# Line 1395  for (code = first_significant_code(code Line 1527  for (code = first_significant_code(code
1527    
1528    switch (c)    switch (c)
1529      {      {
1530      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1531        cannot be represented just by a bit map. This includes negated single
1532        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1533        actual length is stored in the compiled code, so we must update "code"
1534        here. */
1535    
1536  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1537      case OP_XCLASS:      case OP_XCLASS:
1538      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1539      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1540  #endif  #endif
1541    
# Line 1460  for (code = first_significant_code(code Line 1596  for (code = first_significant_code(code
1596      case OP_TYPEPOSPLUS:      case OP_TYPEPOSPLUS:
1597      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1598      return FALSE;      return FALSE;
1599    
1600        /* These are going to continue, as they may be empty, but we have to
1601        fudge the length for the \p and \P cases. */
1602    
1603        case OP_TYPESTAR:
1604        case OP_TYPEMINSTAR:
1605        case OP_TYPEPOSSTAR:
1606        case OP_TYPEQUERY:
1607        case OP_TYPEMINQUERY:
1608        case OP_TYPEPOSQUERY:
1609        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1610        break;
1611    
1612        /* Same for these */
1613    
1614        case OP_TYPEUPTO:
1615        case OP_TYPEMINUPTO:
1616        case OP_TYPEPOSUPTO:
1617        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1618        break;
1619    
1620      /* End of branch */      /* End of branch */
1621    
# Line 1623  adjust_recurse(uschar *group, int adjust Line 1779  adjust_recurse(uschar *group, int adjust
1779    uschar *save_hwm)    uschar *save_hwm)
1780  {  {
1781  uschar *ptr = group;  uschar *ptr = group;
1782    
1783  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1784    {    {
1785    int offset;    int offset;
# Line 1921  if (next >= 0) switch(op_code) Line 2078  if (next >= 0) switch(op_code)
2078    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2079    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2080    
2081      case OP_HSPACE:
2082      case OP_NOT_HSPACE:
2083      switch(next)
2084        {
2085        case 0x09:
2086        case 0x20:
2087        case 0xa0:
2088        case 0x1680:
2089        case 0x180e:
2090        case 0x2000:
2091        case 0x2001:
2092        case 0x2002:
2093        case 0x2003:
2094        case 0x2004:
2095        case 0x2005:
2096        case 0x2006:
2097        case 0x2007:
2098        case 0x2008:
2099        case 0x2009:
2100        case 0x200A:
2101        case 0x202f:
2102        case 0x205f:
2103        case 0x3000:
2104        return op_code != OP_HSPACE;
2105        default:
2106        return op_code == OP_HSPACE;
2107        }
2108    
2109      case OP_VSPACE:
2110      case OP_NOT_VSPACE:
2111      switch(next)
2112        {
2113        case 0x0a:
2114        case 0x0b:
2115        case 0x0c:
2116        case 0x0d:
2117        case 0x85:
2118        case 0x2028:
2119        case 0x2029:
2120        return op_code != OP_VSPACE;
2121        default:
2122        return op_code == OP_VSPACE;
2123        }
2124    
2125    default:    default:
2126    return FALSE;    return FALSE;
2127    }    }
# Line 1955  switch(op_code) Line 2156  switch(op_code)
2156      case ESC_W:      case ESC_W:
2157      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2158    
2159        case ESC_h:
2160        case ESC_H:
2161        switch(item)
2162          {
2163          case 0x09:
2164          case 0x20:
2165          case 0xa0:
2166          case 0x1680:
2167          case 0x180e:
2168          case 0x2000:
2169          case 0x2001:
2170          case 0x2002:
2171          case 0x2003:
2172          case 0x2004:
2173          case 0x2005:
2174          case 0x2006:
2175          case 0x2007:
2176          case 0x2008:
2177          case 0x2009:
2178          case 0x200A:
2179          case 0x202f:
2180          case 0x205f:
2181          case 0x3000:
2182          return -next != ESC_h;
2183          default:
2184          return -next == ESC_h;
2185          }
2186    
2187        case ESC_v:
2188        case ESC_V:
2189        switch(item)
2190          {
2191          case 0x0a:
2192          case 0x0b:
2193          case 0x0c:
2194          case 0x0d:
2195          case 0x85:
2196          case 0x2028:
2197          case 0x2029:
2198          return -next != ESC_v;
2199          default:
2200          return -next == ESC_v;
2201          }
2202    
2203      default:      default:
2204      return FALSE;      return FALSE;
2205      }      }
2206    
2207    case OP_DIGIT:    case OP_DIGIT:
2208    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2209             next == -ESC_h || next == -ESC_v;
2210    
2211    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2212    return next == -ESC_d;    return next == -ESC_d;
# Line 1969  switch(op_code) Line 2215  switch(op_code)
2215    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2216    
2217    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2218    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2219    
2220      case OP_HSPACE:
2221      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2222    
2223      case OP_NOT_HSPACE:
2224      return next == -ESC_h;
2225    
2226      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2227      case OP_VSPACE:
2228      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2229    
2230      case OP_NOT_VSPACE:
2231      return next == -ESC_v;
2232    
2233    case OP_WORDCHAR:    case OP_WORDCHAR:
2234    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2235    
2236    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2237    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2087  for (;; ptr++) Line 2346  for (;; ptr++)
2346    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2347    BOOL is_quantifier;    BOOL is_quantifier;
2348    BOOL is_recurse;    BOOL is_recurse;
2349      BOOL reset_bracount;
2350    int class_charcount;    int class_charcount;
2351    int class_lastchar;    int class_lastchar;
2352    int newoptions;    int newoptions;
2353    int recno;    int recno;
2354      int refsign;
2355    int skipbytes;    int skipbytes;
2356    int subreqbyte;    int subreqbyte;
2357    int subfirstbyte;    int subfirstbyte;
# Line 2123  for (;; ptr++) Line 2384  for (;; ptr++)
2384      */      */
2385    
2386      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2387    
2388        /* Paranoid check for integer overflow */
2389    
2390        if (OFLOW_MAX - *lengthptr < code - last_code)
2391          {
2392          *errorcodeptr = ERR20;
2393          goto FAILED;
2394          }
2395    
2396      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2397      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));
2398    
# Line 2235  for (;; ptr++) Line 2505  for (;; ptr++)
2505      *ptrptr = ptr;      *ptrptr = ptr;
2506      if (lengthptr != NULL)      if (lengthptr != NULL)
2507        {        {
2508          if (OFLOW_MAX - *lengthptr < code - last_code)
2509            {
2510            *errorcodeptr = ERR20;
2511            goto FAILED;
2512            }
2513        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2514        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2515        }        }
# Line 2297  for (;; ptr++) Line 2572  for (;; ptr++)
2572        goto FAILED;        goto FAILED;
2573        }        }
2574    
2575      /* If the first character is '^', set the negation flag and skip it. */      /* If the first character is '^', set the negation flag and skip it. Also,
2576        if the first few characters (either before or after ^) are \Q\E or \E we
2577        skip them too. This makes for compatibility with Perl. */
2578    
2579      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2580        for (;;)
2581        {        {
       negate_class = TRUE;  
2582        c = *(++ptr);        c = *(++ptr);
2583        }        if (c == '\\')
2584      else          {
2585        {          if (ptr[1] == 'E') ptr++;
2586        negate_class = FALSE;            else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2587                else break;
2588            }
2589          else if (!negate_class && c == '^')
2590            negate_class = TRUE;
2591          else break;
2592        }        }
2593    
2594      /* Keep a count of chars with values < 256 so that we can optimize the case      /* Keep a count of chars with values < 256 so that we can optimize the case
# Line 2447  for (;; ptr++) Line 2729  for (;; ptr++)
2729        of the specials, which just set a flag. The sequence \b is a special        of the specials, which just set a flag. The sequence \b is a special
2730        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2731        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2732        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
2733        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2734    
2735        if (c == '\\')        if (c == '\\')
# Line 2467  for (;; ptr++) Line 2749  for (;; ptr++)
2749            else inescq = TRUE;            else inescq = TRUE;
2750            continue;            continue;
2751            }            }
2752            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2753    
2754          if (c < 0)          if (c < 0)
2755            {            {
# Line 2515  for (;; ptr++) Line 2798  for (;; ptr++)
2798            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
2799                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
2800    
2801              /* We need to deal with \H, \h, \V, and \v in both phases because
2802              they use extra memory. */
2803    
2804              if (-c == ESC_h)
2805                {
2806                SETBIT(classbits, 0x09); /* VT */
2807                SETBIT(classbits, 0x20); /* SPACE */
2808                SETBIT(classbits, 0xa0); /* NSBP */
2809    #ifdef SUPPORT_UTF8
2810                if (utf8)
2811                  {
2812                  class_utf8 = TRUE;
2813                  *class_utf8data++ = XCL_SINGLE;
2814                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
2815                  *class_utf8data++ = XCL_SINGLE;
2816                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
2817                  *class_utf8data++ = XCL_RANGE;
2818                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
2819                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
2820                  *class_utf8data++ = XCL_SINGLE;
2821                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
2822                  *class_utf8data++ = XCL_SINGLE;
2823                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
2824                  *class_utf8data++ = XCL_SINGLE;
2825                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
2826                  }
2827    #endif
2828                continue;
2829                }
2830    
2831              if (-c == ESC_H)
2832                {
2833                for (c = 0; c < 32; c++)
2834                  {
2835                  int x = 0xff;
2836                  switch (c)
2837                    {
2838                    case 0x09/8: x ^= 1 << (0x09%8); break;
2839                    case 0x20/8: x ^= 1 << (0x20%8); break;
2840                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
2841                    default: break;
2842                    }
2843                  classbits[c] |= x;
2844                  }
2845    
2846    #ifdef SUPPORT_UTF8
2847                if (utf8)
2848                  {
2849                  class_utf8 = TRUE;
2850                  *class_utf8data++ = XCL_RANGE;
2851                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2852                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
2853                  *class_utf8data++ = XCL_RANGE;
2854                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
2855                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
2856                  *class_utf8data++ = XCL_RANGE;
2857                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
2858                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
2859                  *class_utf8data++ = XCL_RANGE;
2860                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
2861                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
2862                  *class_utf8data++ = XCL_RANGE;
2863                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
2864                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
2865                  *class_utf8data++ = XCL_RANGE;
2866                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
2867                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
2868                  *class_utf8data++ = XCL_RANGE;
2869                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
2870                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2871                  }
2872    #endif
2873                continue;
2874                }
2875    
2876              if (-c == ESC_v)
2877                {
2878                SETBIT(classbits, 0x0a); /* LF */
2879                SETBIT(classbits, 0x0b); /* VT */
2880                SETBIT(classbits, 0x0c); /* FF */
2881                SETBIT(classbits, 0x0d); /* CR */
2882                SETBIT(classbits, 0x85); /* NEL */
2883    #ifdef SUPPORT_UTF8
2884                if (utf8)
2885                  {
2886                  class_utf8 = TRUE;
2887                  *class_utf8data++ = XCL_RANGE;
2888                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
2889                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2890                  }
2891    #endif
2892                continue;
2893                }
2894    
2895              if (-c == ESC_V)
2896                {
2897                for (c = 0; c < 32; c++)
2898                  {
2899                  int x = 0xff;
2900                  switch (c)
2901                    {
2902                    case 0x0a/8: x ^= 1 << (0x0a%8);
2903                                 x ^= 1 << (0x0b%8);
2904                                 x ^= 1 << (0x0c%8);
2905                                 x ^= 1 << (0x0d%8);
2906                                 break;
2907                    case 0x85/8: x ^= 1 << (0x85%8); break;
2908                    default: break;
2909                    }
2910                  classbits[c] |= x;
2911                  }
2912    
2913    #ifdef SUPPORT_UTF8
2914                if (utf8)
2915                  {
2916                  class_utf8 = TRUE;
2917                  *class_utf8data++ = XCL_RANGE;
2918                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2919                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
2920                  *class_utf8data++ = XCL_RANGE;
2921                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2922                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2923                  }
2924    #endif
2925                continue;
2926                }
2927    
2928            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
2929    
2930  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2655  for (;; ptr++) Line 3065  for (;; ptr++)
3065              unsigned int origd = d;              unsigned int origd = d;
3066              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3067                {                {
3068                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3069                      ocd <= (unsigned int)d)
3070                    continue;                          /* Skip embedded ranges */
3071    
3072                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3073                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3074                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3075                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3076                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3077                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3078                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3079                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3080                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3081                  d = ocd;                  d = ocd;
3082                  continue;                  continue;
# Line 2783  for (;; ptr++) Line 3197  for (;; ptr++)
3197        }        }
3198    
3199      /* If class_charcount is 1, we saw precisely one character whose value is      /* If class_charcount is 1, we saw precisely one character whose value is
3200      less than 256. In non-UTF-8 mode we can always optimize. In UTF-8 mode, we      less than 256. As long as there were no characters >= 128 and there was no
3201      can optimize the negative case only if there were no characters >= 128      use of \p or \P, in other words, no use of any XCLASS features, we can
3202      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3203      single-bytes only. This is an historical hangover. Maybe one day we can  
3204      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3205        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3206        operate on single-bytes only. This is an historical hangover. Maybe one day
3207        we can tidy these opcodes to handle multi-byte characters.
3208    
3209      The optimization throws away the bit map. We turn the item into a      The optimization throws away the bit map. We turn the item into a
3210      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note
# Line 2797  for (;; ptr++) Line 3214  for (;; ptr++)
3214      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3215    
3216  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3217      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3218            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3219  #else  #else
3220      if (class_charcount == 1)      if (class_charcount == 1)
3221  #endif  #endif
# Line 3258  for (;; ptr++) Line 3673  for (;; ptr++)
3673          goto FAILED;          goto FAILED;
3674          }          }
3675    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3676        /* If the maximum repeat count is unlimited, find the end of the bracket        /* If the maximum repeat count is unlimited, find the end of the bracket
3677        by scanning through from the start, and compute the offset back to it        by scanning through from the start, and compute the offset back to it
3678        from the current code pointer. There may be an OP_OPT setting following        from the current code pointer. There may be an OP_OPT setting following
# Line 3354  for (;; ptr++) Line 3761  for (;; ptr++)
3761          if (repeat_min > 1)          if (repeat_min > 1)
3762            {            {
3763            /* In the pre-compile phase, we don't actually do the replication. We            /* In the pre-compile phase, we don't actually do the replication. We
3764            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3765              potential integer overflow. */
3766    
3767            if (lengthptr != NULL)            if (lengthptr != NULL)
3768              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3769                int delta = (repeat_min - 1)*length_prevgroup;
3770                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3771                                                                (double)INT_MAX ||
3772                    OFLOW_MAX - *lengthptr < delta)
3773                  {
3774                  *errorcodeptr = ERR20;
3775                  goto FAILED;
3776                  }
3777                *lengthptr += delta;
3778                }
3779    
3780            /* This is compiling for real */            /* This is compiling for real */
3781    
# Line 3395  for (;; ptr++) Line 3813  for (;; ptr++)
3813          /* In the pre-compile phase, we don't actually do the replication. We          /* In the pre-compile phase, we don't actually do the replication. We
3814          just adjust the length as if we had. For each repetition we must add 1          just adjust the length as if we had. For each repetition we must add 1
3815          to the length for BRAZERO and for all but the last repetition we must          to the length for BRAZERO and for all but the last repetition we must
3816          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3817            paranoid checks to avoid integer overflow. */
3818    
3819          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3820            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3821              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3822                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3823              if ((double)repeat_max *
3824                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3825                      > (double)INT_MAX ||
3826                  OFLOW_MAX - *lengthptr < delta)
3827                {
3828                *errorcodeptr = ERR20;
3829                goto FAILED;
3830                }
3831              *lengthptr += delta;
3832              }
3833    
3834          /* This is compiling for real */          /* This is compiling for real */
3835    
# Line 3551  for (;; ptr++) Line 3981  for (;; ptr++)
3981      /* ===================================================================*/      /* ===================================================================*/
3982      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
3983      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
3984      parenthesis forms. First deal with the specials; all are introduced by ?,      parenthesis forms.  */
     and the appearance of any of them means that this is not a capturing  
     group. */  
3985    
3986      case '(':      case '(':
3987      newoptions = options;      newoptions = options;
3988      skipbytes = 0;      skipbytes = 0;
3989      bravalue = OP_CBRA;      bravalue = OP_CBRA;
3990      save_hwm = cd->hwm;      save_hwm = cd->hwm;
3991        reset_bracount = FALSE;
3992    
3993      if (*(++ptr) == '?')      /* First deal with various "verbs" that can be introduced by '*'. */
3994    
3995        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
3996          {
3997          int i, namelen;
3998          const uschar *name = ++ptr;
3999          previous = NULL;
4000          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
4001          if (*ptr == ':')
4002            {
4003            *errorcodeptr = ERR59;   /* Not supported */
4004            goto FAILED;
4005            }
4006          if (*ptr != ')')
4007            {
4008            *errorcodeptr = ERR60;
4009            goto FAILED;
4010            }
4011          namelen = ptr - name;
4012          for (i = 0; i < verbcount; i++)
4013            {
4014            if (namelen == verbs[i].len &&
4015                strncmp((char *)name, verbs[i].name, namelen) == 0)
4016              {
4017              *code = verbs[i].op;
4018              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4019              break;
4020              }
4021            }
4022          if (i < verbcount) continue;
4023          *errorcodeptr = ERR60;
4024          goto FAILED;
4025          }
4026    
4027        /* Deal with the extended parentheses; all are introduced by '?', and the
4028        appearance of any of them means that this is not a capturing group. */
4029    
4030        else if (*ptr == '?')
4031        {        {
4032        int i, set, unset, namelen;        int i, set, unset, namelen;
4033        int *optset;        int *optset;
# Line 3582  for (;; ptr++) Line 4048  for (;; ptr++)
4048    
4049    
4050          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4051            case '|':                 /* Reset capture count for each branch */
4052            reset_bracount = TRUE;
4053            /* Fall through */
4054    
4055            /* ------------------------------------------------------------ */
4056          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
4057          bravalue = OP_BRA;          bravalue = OP_BRA;
4058          ptr++;          ptr++;
# Line 3617  for (;; ptr++) Line 4088  for (;; ptr++)
4088    
4089          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
4090          skipbytes = 3;          skipbytes = 3;
4091            refsign = -1;
4092    
4093          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4094    
# Line 3640  for (;; ptr++) Line 4112  for (;; ptr++)
4112            terminator = '\'';            terminator = '\'';
4113            ptr++;            ptr++;
4114            }            }
4115          else terminator = 0;          else
4116              {
4117              terminator = 0;
4118              if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);
4119              }
4120    
4121          /* 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 */
4122    
# Line 3676  for (;; ptr++) Line 4152  for (;; ptr++)
4152          if (lengthptr != NULL) break;          if (lengthptr != NULL) break;
4153    
4154          /* In the real compile we do the work of looking for the actual          /* In the real compile we do the work of looking for the actual
4155          reference. */          reference. If the string started with "+" or "-" we require the rest to
4156            be digits, in which case recno will be set. */
4157    
4158            if (refsign > 0)
4159              {
4160              if (recno <= 0)
4161                {
4162                *errorcodeptr = ERR58;
4163                goto FAILED;
4164                }
4165              if (refsign == '-')
4166                {
4167                recno = cd->bracount - recno + 1;
4168                if (recno <= 0)
4169                  {
4170                  *errorcodeptr = ERR15;
4171                  goto FAILED;
4172                  }
4173                }
4174              else recno += cd->bracount;
4175              PUT2(code, 2+LINK_SIZE, recno);
4176              break;
4177              }
4178    
4179            /* Otherwise (did not start with "+" or "-"), start by looking for the
4180            name. */
4181    
4182          slot = cd->name_table;          slot = cd->name_table;
4183          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
# Line 3768  for (;; ptr++) Line 4269  for (;; ptr++)
4269    
4270          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4271          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4272          ptr++;          ptr++;
4273            if (*ptr == ')')          /* Optimize (?!) */
4274              {
4275              *code++ = OP_FAIL;
4276              previous = NULL;
4277              continue;
4278              }
4279            bravalue = OP_ASSERT_NOT;
4280          break;          break;
4281    
4282    
# Line 3995  for (;; ptr++) Line 4502  for (;; ptr++)
4502    
4503    
4504          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4505            case '-': case '+':
4506          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */
4507          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case '5': case '6': case '7': case '8': case '9':   /* subroutine */
4508            {            {
4509            const uschar *called;            const uschar *called;
4510    
4511              if ((refsign = *ptr) == '+') ptr++;
4512              else if (refsign == '-')
4513                {
4514                if ((digitab[ptr[1]] & ctype_digit) == 0)
4515                  goto OTHER_CHAR_AFTER_QUERY;
4516                ptr++;
4517                }
4518    
4519            recno = 0;            recno = 0;
4520            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4521              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - '0';
4522    
4523            if (*ptr != ')')            if (*ptr != ')')
4524              {              {
4525              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4526              goto FAILED;              goto FAILED;
4527              }              }
4528    
4529              if (refsign == '-')
4530                {
4531                if (recno == 0)
4532                  {
4533                  *errorcodeptr = ERR58;
4534                  goto FAILED;
4535                  }
4536                recno = cd->bracount - recno + 1;
4537                if (recno <= 0)
4538                  {
4539                  *errorcodeptr = ERR15;
4540                  goto FAILED;
4541                  }
4542                }
4543              else if (refsign == '+')
4544                {
4545                if (recno == 0)
4546                  {
4547                  *errorcodeptr = ERR58;
4548                  goto FAILED;
4549                  }
4550                recno += cd->bracount;
4551                }
4552    
4553            /* Come here from code above that handles a named recursion */            /* Come here from code above that handles a named recursion */
4554    
4555            HANDLE_RECURSION:            HANDLE_RECURSION:
# Line 4080  for (;; ptr++) Line 4622  for (;; ptr++)
4622    
4623          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4624          default:              /* Other characters: check option setting */          default:              /* Other characters: check option setting */
4625            OTHER_CHAR_AFTER_QUERY:
4626          set = unset = 0;          set = unset = 0;
4627          optset = &set;          optset = &set;
4628    
# Line 4214  for (;; ptr++) Line 4757  for (;; ptr++)
4757           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
4758           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
4759            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
4760             reset_bracount,               /* True if (?| group */
4761           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
4762           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
4763           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4230  for (;; ptr++) Line 4774  for (;; ptr++)
4774      is on the bracket. */      is on the bracket. */
4775    
4776      /* If this is a conditional bracket, check that there are no more than      /* If this is a conditional bracket, check that there are no more than
4777      two branches in the group, or just one if it's a DEFINE group. */      two branches in the group, or just one if it's a DEFINE group. We do this
4778        in the real compile phase, not in the pre-pass, where the whole group may
4779        not be available. */
4780    
4781      if (bravalue == OP_COND)      if (bravalue == OP_COND && lengthptr == NULL)
4782        {        {
4783        uschar *tc = code;        uschar *tc = code;
4784        int condcount = 0;        int condcount = 0;
# Line 4279  for (;; ptr++) Line 4825  for (;; ptr++)
4825        goto FAILED;        goto FAILED;
4826        }        }
4827    
4828      /* In the pre-compile phase, update the length by the length of the nested      /* In the pre-compile phase, update the length by the length of the group,
4829      group, less the brackets at either end. Then reduce the compiled code to      less the brackets at either end. Then reduce the compiled code to just a
4830      just the brackets so that it doesn't use much memory if it is duplicated by      set of non-capturing brackets so that it doesn't use much memory if it is
4831      a quantifier. */      duplicated by a quantifier.*/
4832    
4833      if (lengthptr != NULL)      if (lengthptr != NULL)
4834        {        {
4835          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
4836            {
4837            *errorcodeptr = ERR20;
4838            goto FAILED;
4839            }
4840        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4841        code++;        *code++ = OP_BRA;
4842        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4843        *code++ = OP_KET;        *code++ = OP_KET;
4844        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4845          break;    /* No need to waste time with special character handling */
4846        }        }
4847    
4848      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4849    
4850      else code = tempcode;      code = tempcode;
4851    
4852      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4853      relevant. */      relevant. */
# Line 4392  for (;; ptr++) Line 4944  for (;; ptr++)
4944        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
4945        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
4946    
4947        /* \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).
4948          We also support \k{name} (.NET syntax) */
4949    
4950        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\''))        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\'' || ptr[1] == '{'))
4951          {          {
4952          is_recurse = FALSE;          is_recurse = FALSE;
4953          terminator = (*(++ptr) == '<')? '>' : '\'';          terminator = (*(++ptr) == '<')? '>' : (*ptr == '\'')? '\'' : '}';
4954          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
4955          }          }
4956    
# Line 4563  This function is used during the pre-com Line 5116  This function is used during the pre-com
5116  out the amount of memory needed, as well as during the real compile phase. The  out the amount of memory needed, as well as during the real compile phase. The
5117  value of lengthptr distinguishes the two phases.  value of lengthptr distinguishes the two phases.
5118    
5119  Argument:  Arguments:
5120    options        option bits, including any changes for this subpattern    options        option bits, including any changes for this subpattern
5121    oldims         previous settings of ims option bits    oldims         previous settings of ims option bits
5122    codeptr        -> the address of the current code pointer    codeptr        -> the address of the current code pointer
5123    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5124    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5125    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5126      reset_bracount TRUE to reset the count for each branch
5127    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5128    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5129    reqbyteptr     place to put the last required character, or a negative number    reqbyteptr     place to put the last required character, or a negative number
# Line 4583  Returns: TRUE on success Line 5137  Returns: TRUE on success
5137    
5138  static BOOL  static BOOL
5139  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5140    int *errorcodeptr, BOOL lookbehind, int skipbytes, int *firstbyteptr,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5141    int *reqbyteptr, branch_chain *bcptr, compile_data *cd, int *lengthptr)    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5142      int *lengthptr)
5143  {  {
5144  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
5145  uschar *code = *codeptr;  uschar *code = *codeptr;
# Line 4594  uschar *reverse_count = NULL; Line 5149  uschar *reverse_count = NULL;
5149  int firstbyte, reqbyte;  int firstbyte, reqbyte;
5150  int branchfirstbyte, branchreqbyte;  int branchfirstbyte, branchreqbyte;
5151  int length;  int length;
5152    int orig_bracount;
5153    int max_bracount;
5154  branch_chain bc;  branch_chain bc;
5155    
5156  bc.outer = bcptr;  bc.outer = bcptr;
# Line 4622  code += 1 + LINK_SIZE + skipbytes; Line 5179  code += 1 + LINK_SIZE + skipbytes;
5179    
5180  /* Loop for each alternative branch */  /* Loop for each alternative branch */
5181    
5182    orig_bracount = max_bracount = cd->bracount;
5183  for (;;)  for (;;)
5184    {    {
5185      /* For a (?| group, reset the capturing bracket count so that each branch
5186      uses the same numbers. */
5187    
5188      if (reset_bracount) cd->bracount = orig_bracount;
5189    
5190    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5191    
5192    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4653  for (;;) Line 5216  for (;;)
5216      return FALSE;      return FALSE;
5217      }      }
5218    
5219      /* Keep the highest bracket count in case (?| was used and some branch
5220      has fewer than the rest. */
5221    
5222      if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5223    
5224    /* In the real compile phase, there is some post-processing to be done. */    /* In the real compile phase, there is some post-processing to be done. */
5225    
5226    if (lengthptr == NULL)    if (lengthptr == NULL)
# Line 4716  for (;;) Line 5284  for (;;)
5284        }        }
5285      }      }
5286    
5287    /* Reached end of expression, either ')' or end of pattern. Go back through    /* Reached end of expression, either ')' or end of pattern. In the real
5288    the alternative branches and reverse the chain of offsets, with the field in    compile phase, go back through the alternative branches and reverse the chain
5289    the BRA item now becoming an offset to the first alternative. If there are    of offsets, with the field in the BRA item now becoming an offset to the
5290    no alternatives, it points to the end of the group. The length in the    first alternative. If there are no alternatives, it points to the end of the
5291    terminating ket is always the length of the whole bracketed item. If any of    group. The length in the terminating ket is always the length of the whole
5292    the ims options were changed inside the group, compile a resetting op-code    bracketed item. If any of the ims options were changed inside the group,
5293    following, except at the very end of the pattern. Return leaving the pointer    compile a resetting op-code following, except at the very end of the pattern.
5294    at the terminating char. */    Return leaving the pointer at the terminating char. */
5295    
5296    if (*ptr != '|')    if (*ptr != '|')
5297      {      {
5298      int branch_length = code - last_branch;      if (lengthptr == NULL)
     do  
5299        {        {
5300        int prev_length = GET(last_branch, 1);        int branch_length = code - last_branch;
5301        PUT(last_branch, 1, branch_length);        do
5302        branch_length = prev_length;          {
5303        last_branch -= branch_length;          int prev_length = GET(last_branch, 1);
5304            PUT(last_branch, 1, branch_length);
5305            branch_length = prev_length;
5306            last_branch -= branch_length;
5307            }
5308          while (branch_length > 0);
5309        }        }
     while (branch_length > 0);  
5310    
5311      /* Fill in the ket */      /* Fill in the ket */
5312    
# Line 4752  for (;;) Line 5323  for (;;)
5323        length += 2;        length += 2;
5324        }        }
5325    
5326        /* Retain the highest bracket number, in case resetting was used. */
5327    
5328        cd->bracount = max_bracount;
5329    
5330      /* Set values to pass back */      /* Set values to pass back */
5331    
5332      *codeptr = code;      *codeptr = code;
5333      *ptrptr = ptr;      *ptrptr = ptr;
5334      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5335      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5336      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5337          {
5338          if (OFLOW_MAX - *lengthptr < length)
5339            {
5340            *errorcodeptr = ERR20;
5341            return FALSE;
5342            }
5343          *lengthptr += length;
5344          }
5345      return TRUE;      return TRUE;
5346      }      }
5347    
5348    /* Another branch follows; insert an "or" node. Its length field points back    /* Another branch follows. In the pre-compile phase, we can move the code
5349      pointer back to where it was for the start of the first branch. (That is,
5350      pretend that each branch is the only one.)
5351    
5352      In the real compile phase, insert an ALT node. Its length field points back
5353    to the previous branch while the bracket remains open. At the end the chain    to the previous branch while the bracket remains open. At the end the chain
5354    is reversed. It's done like this so that the start of the bracket has a    is reversed. It's done like this so that the start of the bracket has a
5355    zero offset until it is closed, making it possible to detect recursion. */    zero offset until it is closed, making it possible to detect recursion. */
5356    
5357    *code = OP_ALT;    if (lengthptr != NULL)
5358    PUT(code, 1, code - last_branch);      {
5359    bc.current = last_branch = code;      code = *codeptr + 1 + LINK_SIZE + skipbytes;
5360    code += 1 + LINK_SIZE;      length += 1 + LINK_SIZE;
5361        }
5362      else
5363        {
5364        *code = OP_ALT;
5365        PUT(code, 1, code - last_branch);
5366        bc.current = last_branch = code;
5367        code += 1 + LINK_SIZE;
5368        }
5369    
5370    ptr++;    ptr++;
   length += 1 + LINK_SIZE;  
5371    }    }
5372  /* Control never reaches here */  /* Control never reaches here */
5373  }  }
# Line 5039  Returns: pointer to compiled data Line 5634  Returns: pointer to compiled data
5634                  with errorptr and erroroffset set                  with errorptr and erroroffset set
5635  */  */
5636    
5637  PCRE_DATA_SCOPE pcre *  PCRE_EXP_DEFN pcre *
5638  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
5639    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
5640  {  {
# Line 5047  return pcre_compile2(pattern, options, N Line 5642  return pcre_compile2(pattern, options, N
5642  }  }
5643    
5644    
5645  PCRE_DATA_SCOPE pcre *  PCRE_EXP_DEFN pcre *
5646  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
5647    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
5648  {  {
# Line 5096  if (errorcodeptr != NULL) *errorcodeptr Line 5691  if (errorcodeptr != NULL) *errorcodeptr
5691  if (erroroffset == NULL)  if (erroroffset == NULL)
5692    {    {
5693    errorcode = ERR16;    errorcode = ERR16;
5694    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN2;
5695    }    }
5696    
5697  *erroroffset = 0;  *erroroffset = 0;
# Line 5109  if (utf8 && (options & PCRE_NO_UTF8_CHEC Line 5704  if (utf8 && (options & PCRE_NO_UTF8_CHEC
5704       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)
5705    {    {
5706    errorcode = ERR44;    errorcode = ERR44;
5707    goto PCRE_UTF8_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN2;
5708    }    }
5709  #else  #else
5710  if ((options & PCRE_UTF8) != 0)  if ((options & PCRE_UTF8) != 0)
# Line 5134  cd->cbits = tables + cbits_offset; Line 5729  cd->cbits = tables + cbits_offset;
5729  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5730    
5731  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5732  current code allows for fixed one- or two-byte sequences, plus "any". */  current code allows for fixed one- or two-byte sequences, plus "any" and
5733    "anycrlf". */
5734    
5735  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))
5736    {    {
# Line 5144  switch (options & (PCRE_NEWLINE_CRLF | P Line 5740  switch (options & (PCRE_NEWLINE_CRLF | P
5740    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
5741         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;
5742    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
5743      case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
5744    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5745    }    }
5746    
5747  if (newline < 0)  if (newline == -2)
5748      {
5749      cd->nltype = NLTYPE_ANYCRLF;
5750      }
5751    else if (newline < 0)
5752    {    {
5753    cd->nltype = NLTYPE_ANY;    cd->nltype = NLTYPE_ANY;
5754    }    }
# Line 5208  outside can help speed up starting point Line 5809  outside can help speed up starting point
5809  code = cworkspace;  code = cworkspace;
5810  *code = OP_BRA;  *code = OP_BRA;
5811  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
5812    &code, &ptr, &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length);    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
5813      &length);
5814  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
5815    
5816  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 5267  cd->start_code = codestart; Line 5869  cd->start_code = codestart;
5869  cd->hwm = cworkspace;  cd->hwm = cworkspace;
5870  cd->req_varyopt = 0;  cd->req_varyopt = 0;
5871  cd->nopartial = FALSE;  cd->nopartial = FALSE;
5872    cd->had_accept = FALSE;
5873    
5874  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
5875  error, errorcode will be set non-zero, so we don't need to look at the result  error, errorcode will be set non-zero, so we don't need to look at the result
# Line 5276  ptr = (const uschar *)pattern; Line 5879  ptr = (const uschar *)pattern;
5879  code = (uschar *)codestart;  code = (uschar *)codestart;
5880  *code = OP_BRA;  *code = OP_BRA;
5881  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
5882    &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
5883  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
5884  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
5885    
5886  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;
5887    if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
5888    
5889  /* If not reached end of pattern on success, there's an excess bracket. */  /* If not reached end of pattern on success, there's an excess bracket. */
5890    
# Line 5321  if (errorcode != 0) Line 5925  if (errorcode != 0)
5925    (pcre_free)(re);    (pcre_free)(re);
5926    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
5927    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
5928  #ifdef SUPPORT_UTF8    PCRE_EARLY_ERROR_RETURN2:
   PCRE_UTF8_ERROR_RETURN:  
 #endif  
5929    *errorptr = error_texts[errorcode];    *errorptr = error_texts[errorcode];
5930    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
5931    return NULL;    return NULL;
# Line 5413  if ((re->options & PCRE_REQCHSET) != 0) Line 6015  if ((re->options & PCRE_REQCHSET) != 0)
6015      else printf("Req char = \\x%02x%s\n", ch, caseless);      else printf("Req char = \\x%02x%s\n", ch, caseless);
6016    }    }
6017    
6018  pcre_printint(re, stdout);  pcre_printint(re, stdout, TRUE);
6019    
6020  /* This check is done here in the debugging case so that the code that  /* This check is done here in the debugging case so that the code that
6021  was compiled can be seen. */  was compiled can be seen. */

Legend:
Removed from v.97  
changed lines
  Added in v.224

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12