/[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 107 by ph10, Wed Mar 7 11:02:28 2007 UTC revision 231 by ph10, Tue Sep 11 11:15:33 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  #ifdef SUPPORT_UTF8
1343      if (utf8) switch(c)      if (utf8) switch(c)
1344        {        {
# Line 1267  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  #endif
1364      }      }
1365    }    }
1366  }  }
# Line 1303  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  #ifdef SUPPORT_UTF8
1436      if (utf8) switch(c)      if (utf8) switch(c)
1437        {        {
# Line 1332  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  #endif
1457      }      }
1458    }    }
1459  }  }
# Line 1370  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 1386  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 1399  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 1465  for (code = first_significant_code(code Line 1597  for (code = first_significant_code(code
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    
1622      case OP_KET:      case OP_KET:
# Line 1627  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 1925  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 1959  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 1973  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 2091  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 2127  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 2239  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 2301  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 2451  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 2471  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 2519  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 2570  for (;; ptr++) Line 2976  for (;; ptr++)
2976    
2977        oldptr = ptr;        oldptr = ptr;
2978    
2979          /* Remember \r or \n */
2980    
2981          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
2982    
2983          /* Check for range */
2984    
2985        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
2986          {          {
2987          int d;          int d;
# Line 2637  for (;; ptr++) Line 3049  for (;; ptr++)
3049    
3050          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3051    
3052            /* Remember \r or \n */
3053    
3054            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3055    
3056          /* 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
3057          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3058          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2659  for (;; ptr++) Line 3075  for (;; ptr++)
3075              unsigned int origd = d;              unsigned int origd = d;
3076              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3077                {                {
3078                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3079                      ocd <= (unsigned int)d)
3080                    continue;                          /* Skip embedded ranges */
3081    
3082                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3083                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3084                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3085                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3086                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3087                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3088                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3089                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3090                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3091                  d = ocd;                  d = ocd;
3092                  continue;                  continue;
# Line 2786  for (;; ptr++) Line 3206  for (;; ptr++)
3206        goto FAILED;        goto FAILED;
3207        }        }
3208    
3209    
3210    /* This code has been disabled because it would mean that \s counts as
3211    an explicit \r or \n reference, and that's not really what is wanted. Now
3212    we set the flag only if there is a literal "\r" or "\n" in the class. */
3213    
3214    #if 0
3215        /* Remember whether \r or \n are in this class */
3216    
3217        if (negate_class)
3218          {
3219          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3220          }
3221        else
3222          {
3223          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3224          }
3225    #endif
3226    
3227    
3228      /* 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
3229      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
3230      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
3231      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3232      single-bytes only. This is an historical hangover. Maybe one day we can  
3233      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3234        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3235        operate on single-bytes only. This is an historical hangover. Maybe one day
3236        we can tidy these opcodes to handle multi-byte characters.
3237    
3238      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
3239      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 2801  for (;; ptr++) Line 3243  for (;; ptr++)
3243      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3244    
3245  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3246      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3247            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3248  #else  #else
3249      if (class_charcount == 1)      if (class_charcount == 1)
3250  #endif  #endif
# Line 3076  for (;; ptr++) Line 3516  for (;; ptr++)
3516        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3517        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3518    
3519        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3520    
3521        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3522    
# Line 3226  for (;; ptr++) Line 3666  for (;; ptr++)
3666        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3667        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3668    
3669        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3670    
3671        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3672          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3262  for (;; ptr++) Line 3702  for (;; ptr++)
3702          goto FAILED;          goto FAILED;
3703          }          }
3704    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3705        /* 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
3706        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
3707        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 3358  for (;; ptr++) Line 3790  for (;; ptr++)
3790          if (repeat_min > 1)          if (repeat_min > 1)
3791            {            {
3792            /* 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
3793            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3794              potential integer overflow. */
3795    
3796            if (lengthptr != NULL)            if (lengthptr != NULL)
3797              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3798                int delta = (repeat_min - 1)*length_prevgroup;
3799                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3800                                                                (double)INT_MAX ||
3801                    OFLOW_MAX - *lengthptr < delta)
3802                  {
3803                  *errorcodeptr = ERR20;
3804                  goto FAILED;
3805                  }
3806                *lengthptr += delta;
3807                }
3808    
3809            /* This is compiling for real */            /* This is compiling for real */
3810    
# Line 3399  for (;; ptr++) Line 3842  for (;; ptr++)
3842          /* 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
3843          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
3844          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
3845          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3846            paranoid checks to avoid integer overflow. */
3847    
3848          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3849            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3850              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3851                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3852              if ((double)repeat_max *
3853                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3854                      > (double)INT_MAX ||
3855                  OFLOW_MAX - *lengthptr < delta)
3856                {
3857                *errorcodeptr = ERR20;
3858                goto FAILED;
3859                }
3860              *lengthptr += delta;
3861              }
3862    
3863          /* This is compiling for real */          /* This is compiling for real */
3864    
# Line 3555  for (;; ptr++) Line 4010  for (;; ptr++)
4010      /* ===================================================================*/      /* ===================================================================*/
4011      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4012      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4013      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. */  
4014    
4015      case '(':      case '(':
4016      newoptions = options;      newoptions = options;
4017      skipbytes = 0;      skipbytes = 0;
4018      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4019      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4020        reset_bracount = FALSE;
4021    
4022        /* First deal with various "verbs" that can be introduced by '*'. */
4023    
4024      if (*(++ptr) == '?')      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4025          {
4026          int i, namelen;
4027          const uschar *name = ++ptr;
4028          previous = NULL;
4029          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
4030          if (*ptr == ':')
4031            {
4032            *errorcodeptr = ERR59;   /* Not supported */
4033            goto FAILED;
4034            }
4035          if (*ptr != ')')
4036            {
4037            *errorcodeptr = ERR60;
4038            goto FAILED;
4039            }
4040          namelen = ptr - name;
4041          for (i = 0; i < verbcount; i++)
4042            {
4043            if (namelen == verbs[i].len &&
4044                strncmp((char *)name, verbs[i].name, namelen) == 0)
4045              {
4046              *code = verbs[i].op;
4047              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4048              break;
4049              }
4050            }
4051          if (i < verbcount) continue;
4052          *errorcodeptr = ERR60;
4053          goto FAILED;
4054          }
4055    
4056        /* Deal with the extended parentheses; all are introduced by '?', and the
4057        appearance of any of them means that this is not a capturing group. */
4058    
4059        else if (*ptr == '?')
4060        {        {
4061        int i, set, unset, namelen;        int i, set, unset, namelen;
4062        int *optset;        int *optset;
# Line 3586  for (;; ptr++) Line 4077  for (;; ptr++)
4077    
4078    
4079          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4080            case '|':                 /* Reset capture count for each branch */
4081            reset_bracount = TRUE;
4082            /* Fall through */
4083    
4084            /* ------------------------------------------------------------ */
4085          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
4086          bravalue = OP_BRA;          bravalue = OP_BRA;
4087          ptr++;          ptr++;
# Line 3621  for (;; ptr++) Line 4117  for (;; ptr++)
4117    
4118          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
4119          skipbytes = 3;          skipbytes = 3;
4120            refsign = -1;
4121    
4122          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4123    
# Line 3644  for (;; ptr++) Line 4141  for (;; ptr++)
4141            terminator = '\'';            terminator = '\'';
4142            ptr++;            ptr++;
4143            }            }
4144          else terminator = 0;          else
4145              {
4146              terminator = 0;
4147              if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);
4148              }
4149    
4150          /* 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 */
4151    
# Line 3680  for (;; ptr++) Line 4181  for (;; ptr++)
4181          if (lengthptr != NULL) break;          if (lengthptr != NULL) break;
4182    
4183          /* 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
4184          reference. */          reference. If the string started with "+" or "-" we require the rest to
4185            be digits, in which case recno will be set. */
4186    
4187            if (refsign > 0)
4188              {
4189              if (recno <= 0)
4190                {
4191                *errorcodeptr = ERR58;
4192                goto FAILED;
4193                }
4194              if (refsign == '-')
4195                {
4196                recno = cd->bracount - recno + 1;
4197                if (recno <= 0)
4198                  {
4199                  *errorcodeptr = ERR15;
4200                  goto FAILED;
4201                  }
4202                }
4203              else recno += cd->bracount;
4204              PUT2(code, 2+LINK_SIZE, recno);
4205              break;
4206              }
4207    
4208            /* Otherwise (did not start with "+" or "-"), start by looking for the
4209            name. */
4210    
4211          slot = cd->name_table;          slot = cd->name_table;
4212          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
# Line 3772  for (;; ptr++) Line 4298  for (;; ptr++)
4298    
4299          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4300          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4301          ptr++;          ptr++;
4302            if (*ptr == ')')          /* Optimize (?!) */
4303              {
4304              *code++ = OP_FAIL;
4305              previous = NULL;
4306              continue;
4307              }
4308            bravalue = OP_ASSERT_NOT;
4309          break;          break;
4310    
4311    
# Line 3999  for (;; ptr++) Line 4531  for (;; ptr++)
4531    
4532    
4533          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4534            case '-': case '+':
4535          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */
4536          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case '5': case '6': case '7': case '8': case '9':   /* subroutine */
4537            {            {
4538            const uschar *called;            const uschar *called;
4539    
4540              if ((refsign = *ptr) == '+') ptr++;
4541              else if (refsign == '-')
4542                {
4543                if ((digitab[ptr[1]] & ctype_digit) == 0)
4544                  goto OTHER_CHAR_AFTER_QUERY;
4545                ptr++;
4546                }
4547    
4548            recno = 0;            recno = 0;
4549            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4550              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - '0';
4551    
4552            if (*ptr != ')')            if (*ptr != ')')
4553              {              {
4554              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4555              goto FAILED;              goto FAILED;
4556              }              }
4557    
4558              if (refsign == '-')
4559                {
4560                if (recno == 0)
4561                  {
4562                  *errorcodeptr = ERR58;
4563                  goto FAILED;
4564                  }
4565                recno = cd->bracount - recno + 1;
4566                if (recno <= 0)
4567                  {
4568                  *errorcodeptr = ERR15;
4569                  goto FAILED;
4570                  }
4571                }
4572              else if (refsign == '+')
4573                {
4574                if (recno == 0)
4575                  {
4576                  *errorcodeptr = ERR58;
4577                  goto FAILED;
4578                  }
4579                recno += cd->bracount;
4580                }
4581    
4582            /* Come here from code above that handles a named recursion */            /* Come here from code above that handles a named recursion */
4583    
4584            HANDLE_RECURSION:            HANDLE_RECURSION:
# Line 4084  for (;; ptr++) Line 4651  for (;; ptr++)
4651    
4652          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4653          default:              /* Other characters: check option setting */          default:              /* Other characters: check option setting */
4654            OTHER_CHAR_AFTER_QUERY:
4655          set = unset = 0;          set = unset = 0;
4656          optset = &set;          optset = &set;
4657    
# Line 4095  for (;; ptr++) Line 4663  for (;; ptr++)
4663    
4664              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4665              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4666              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4667              break;              break;
4668    
4669              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4218  for (;; ptr++) Line 4786  for (;; ptr++)
4786           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
4787           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
4788            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
4789             reset_bracount,               /* True if (?| group */
4790           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
4791           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
4792           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4234  for (;; ptr++) Line 4803  for (;; ptr++)
4803      is on the bracket. */      is on the bracket. */
4804    
4805      /* 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
4806      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
4807        in the real compile phase, not in the pre-pass, where the whole group may
4808        not be available. */
4809    
4810      if (bravalue == OP_COND)      if (bravalue == OP_COND && lengthptr == NULL)
4811        {        {
4812        uschar *tc = code;        uschar *tc = code;
4813        int condcount = 0;        int condcount = 0;
# Line 4283  for (;; ptr++) Line 4854  for (;; ptr++)
4854        goto FAILED;        goto FAILED;
4855        }        }
4856    
4857      /* 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,
4858      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
4859      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
4860      a quantifier. */      duplicated by a quantifier.*/
4861    
4862      if (lengthptr != NULL)      if (lengthptr != NULL)
4863        {        {
4864          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
4865            {
4866            *errorcodeptr = ERR20;
4867            goto FAILED;
4868            }
4869        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4870        code++;        *code++ = OP_BRA;
4871        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4872        *code++ = OP_KET;        *code++ = OP_KET;
4873        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4874          break;    /* No need to waste time with special character handling */
4875        }        }
4876    
4877      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4878    
4879      else code = tempcode;      code = tempcode;
4880    
4881      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4882      relevant. */      relevant. */
# Line 4396  for (;; ptr++) Line 4973  for (;; ptr++)
4973        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
4974        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
4975    
4976        /* \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).
4977          We also support \k{name} (.NET syntax) */
4978    
4979        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\''))        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\'' || ptr[1] == '{'))
4980          {          {
4981          is_recurse = FALSE;          is_recurse = FALSE;
4982          terminator = (*(++ptr) == '<')? '>' : '\'';          terminator = (*(++ptr) == '<')? '>' : (*ptr == '\'')? '\'' : '}';
4983          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
4984          }          }
4985    
# Line 4502  for (;; ptr++) Line 5080  for (;; ptr++)
5080      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5081      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5082    
5083        /* Remember if \r or \n were seen */
5084    
5085        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5086          cd->external_flags |= PCRE_HASCRORLF;
5087    
5088      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5089      byte, set it from this character, but revert to none on a zero repeat.      byte, set it from this character, but revert to none on a zero repeat.
5090      Otherwise, leave the firstbyte value alone, and don't change it on a zero      Otherwise, leave the firstbyte value alone, and don't change it on a zero
# Line 4567  This function is used during the pre-com Line 5150  This function is used during the pre-com
5150  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
5151  value of lengthptr distinguishes the two phases.  value of lengthptr distinguishes the two phases.
5152    
5153  Argument:  Arguments:
5154    options        option bits, including any changes for this subpattern    options        option bits, including any changes for this subpattern
5155    oldims         previous settings of ims option bits    oldims         previous settings of ims option bits
5156    codeptr        -> the address of the current code pointer    codeptr        -> the address of the current code pointer
5157    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5158    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5159    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5160      reset_bracount TRUE to reset the count for each branch
5161    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5162    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5163    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 4587  Returns: TRUE on success Line 5171  Returns: TRUE on success
5171    
5172  static BOOL  static BOOL
5173  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5174    int *errorcodeptr, BOOL lookbehind, int skipbytes, int *firstbyteptr,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5175    int *reqbyteptr, branch_chain *bcptr, compile_data *cd, int *lengthptr)    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5176      int *lengthptr)
5177  {  {
5178  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
5179  uschar *code = *codeptr;  uschar *code = *codeptr;
# Line 4598  uschar *reverse_count = NULL; Line 5183  uschar *reverse_count = NULL;
5183  int firstbyte, reqbyte;  int firstbyte, reqbyte;
5184  int branchfirstbyte, branchreqbyte;  int branchfirstbyte, branchreqbyte;
5185  int length;  int length;
5186    int orig_bracount;
5187    int max_bracount;
5188  branch_chain bc;  branch_chain bc;
5189    
5190  bc.outer = bcptr;  bc.outer = bcptr;
# Line 4626  code += 1 + LINK_SIZE + skipbytes; Line 5213  code += 1 + LINK_SIZE + skipbytes;
5213    
5214  /* Loop for each alternative branch */  /* Loop for each alternative branch */
5215    
5216    orig_bracount = max_bracount = cd->bracount;
5217  for (;;)  for (;;)
5218    {    {
5219      /* For a (?| group, reset the capturing bracket count so that each branch
5220      uses the same numbers. */
5221    
5222      if (reset_bracount) cd->bracount = orig_bracount;
5223    
5224    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5225    
5226    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4657  for (;;) Line 5250  for (;;)
5250      return FALSE;      return FALSE;
5251      }      }
5252    
5253      /* Keep the highest bracket count in case (?| was used and some branch
5254      has fewer than the rest. */
5255    
5256      if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5257    
5258    /* 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. */
5259    
5260    if (lengthptr == NULL)    if (lengthptr == NULL)
# Line 4720  for (;;) Line 5318  for (;;)
5318        }        }
5319      }      }
5320    
5321    /* Reached end of expression, either ')' or end of pattern. Go back through    /* Reached end of expression, either ')' or end of pattern. In the real
5322    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
5323    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
5324    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
5325    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
5326    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,
5327    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.
5328    at the terminating char. */    Return leaving the pointer at the terminating char. */
5329    
5330    if (*ptr != '|')    if (*ptr != '|')
5331      {      {
5332      int branch_length = code - last_branch;      if (lengthptr == NULL)
     do  
5333        {        {
5334        int prev_length = GET(last_branch, 1);        int branch_length = code - last_branch;
5335        PUT(last_branch, 1, branch_length);        do
5336        branch_length = prev_length;          {
5337        last_branch -= branch_length;          int prev_length = GET(last_branch, 1);
5338            PUT(last_branch, 1, branch_length);
5339            branch_length = prev_length;
5340            last_branch -= branch_length;
5341            }
5342          while (branch_length > 0);
5343        }        }
     while (branch_length > 0);  
5344    
5345      /* Fill in the ket */      /* Fill in the ket */
5346    
# Line 4756  for (;;) Line 5357  for (;;)
5357        length += 2;        length += 2;
5358        }        }
5359    
5360        /* Retain the highest bracket number, in case resetting was used. */
5361    
5362        cd->bracount = max_bracount;
5363    
5364      /* Set values to pass back */      /* Set values to pass back */
5365    
5366      *codeptr = code;      *codeptr = code;
5367      *ptrptr = ptr;      *ptrptr = ptr;
5368      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5369      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5370      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5371          {
5372          if (OFLOW_MAX - *lengthptr < length)
5373            {
5374            *errorcodeptr = ERR20;
5375            return FALSE;
5376            }
5377          *lengthptr += length;
5378          }
5379      return TRUE;      return TRUE;
5380      }      }
5381    
5382    /* 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
5383      pointer back to where it was for the start of the first branch. (That is,
5384      pretend that each branch is the only one.)
5385    
5386      In the real compile phase, insert an ALT node. Its length field points back
5387    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
5388    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
5389    zero offset until it is closed, making it possible to detect recursion. */    zero offset until it is closed, making it possible to detect recursion. */
5390    
5391    *code = OP_ALT;    if (lengthptr != NULL)
5392    PUT(code, 1, code - last_branch);      {
5393    bc.current = last_branch = code;      code = *codeptr + 1 + LINK_SIZE + skipbytes;
5394    code += 1 + LINK_SIZE;      length += 1 + LINK_SIZE;
5395        }
5396      else
5397        {
5398        *code = OP_ALT;
5399        PUT(code, 1, code - last_branch);
5400        bc.current = last_branch = code;
5401        code += 1 + LINK_SIZE;
5402        }
5403    
5404    ptr++;    ptr++;
   length += 1 + LINK_SIZE;  
5405    }    }
5406  /* Control never reaches here */  /* Control never reaches here */
5407  }  }
# Line 5043  Returns: pointer to compiled data Line 5668  Returns: pointer to compiled data
5668                  with errorptr and erroroffset set                  with errorptr and erroroffset set
5669  */  */
5670    
5671  PCRE_DATA_SCOPE pcre *  PCRE_EXP_DEFN pcre *
5672  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
5673    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
5674  {  {
# Line 5051  return pcre_compile2(pattern, options, N Line 5676  return pcre_compile2(pattern, options, N
5676  }  }
5677    
5678    
5679  PCRE_DATA_SCOPE pcre *  PCRE_EXP_DEFN pcre *
5680  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
5681    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
5682  {  {
# Line 5059  real_pcre *re; Line 5684  real_pcre *re;
5684  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5685  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5686  int errorcode = 0;  int errorcode = 0;
5687    int skipatstart = 0;
5688  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5689  BOOL utf8;  BOOL utf8;
5690  #endif  #endif
# Line 5100  if (errorcodeptr != NULL) *errorcodeptr Line 5726  if (errorcodeptr != NULL) *errorcodeptr
5726  if (erroroffset == NULL)  if (erroroffset == NULL)
5727    {    {
5728    errorcode = ERR16;    errorcode = ERR16;
5729    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN2;
5730    }    }
5731    
5732  *erroroffset = 0;  *erroroffset = 0;
# Line 5113  if (utf8 && (options & PCRE_NO_UTF8_CHEC Line 5739  if (utf8 && (options & PCRE_NO_UTF8_CHEC
5739       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)
5740    {    {
5741    errorcode = ERR44;    errorcode = ERR44;
5742    goto PCRE_UTF8_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN2;
5743    }    }
5744  #else  #else
5745  if ((options & PCRE_UTF8) != 0)  if ((options & PCRE_UTF8) != 0)
# Line 5137  cd->fcc = tables + fcc_offset; Line 5763  cd->fcc = tables + fcc_offset;
5763  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5764  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5765    
5766    /* Check for global one-time settings at the start of the pattern, and remember
5767    the offset for later. */
5768    
5769    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5770      {
5771      int newnl = 0;
5772      int newbsr = 0;
5773    
5774      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5775        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5776      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5777        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5778      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5779        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5780      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5781        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5782      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5783        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5784    
5785      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5786        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5787      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5788        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5789    
5790      if (newnl != 0)
5791        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5792      else if (newbsr != 0)
5793        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5794      else break;
5795      }
5796    
5797    /* Check validity of \R options. */
5798    
5799    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5800      {
5801      case 0:
5802      case PCRE_BSR_ANYCRLF:
5803      case PCRE_BSR_UNICODE:
5804      break;
5805      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5806      }
5807    
5808  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5809  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
5810    "anycrlf". */
5811    
5812  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5813    {    {
5814    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5815    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5816    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5817    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
5818         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;
5819    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
5820      case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
5821    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5822    }    }
5823    
5824  if (newline < 0)  if (newline == -2)
5825      {
5826      cd->nltype = NLTYPE_ANYCRLF;
5827      }
5828    else if (newline < 0)
5829    {    {
5830    cd->nltype = NLTYPE_ANY;    cd->nltype = NLTYPE_ANY;
5831    }    }
# Line 5200  cd->hwm = cworkspace; Line 5874  cd->hwm = cworkspace;
5874  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5875  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5876  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5877  cd->external_options = options;  cd->external_options = options;
5878    cd->external_flags = 0;
5879    
5880  /* Now do the pre-compile. On error, errorcode will be set non-zero, so we  /* Now do the pre-compile. On error, errorcode will be set non-zero, so we
5881  don't need to look at the result of the function here. The initial options have  don't need to look at the result of the function here. The initial options have
# Line 5209  been put into the cd block so that they Line 5883  been put into the cd block so that they
5883  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5884  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5885    
5886    ptr += skipatstart;
5887  code = cworkspace;  code = cworkspace;
5888  *code = OP_BRA;  *code = OP_BRA;
5889  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
5890    &code, &ptr, &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length);    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
5891      &length);
5892  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
5893    
5894  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 5238  if (re == NULL) Line 5914  if (re == NULL)
5914    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
5915    }    }
5916    
5917  /* Put in the magic number, and save the sizes, initial options, and character  /* Put in the magic number, and save the sizes, initial options, internal
5918  table pointer. NULL is used for the default character tables. The nullpad field  flags, and character table pointer. NULL is used for the default character
5919  is at the end; it's there to help in the case when a regex compiled on a system  tables. The nullpad field is at the end; it's there to help in the case when a
5920  with 4-byte pointers is run on another with 8-byte pointers. */  regex compiled on a system with 4-byte pointers is run on another with 8-byte
5921    pointers. */
5922    
5923  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
5924  re->size = size;  re->size = size;
5925  re->options = cd->external_options;  re->options = cd->external_options;
5926    re->flags = cd->external_flags;
5927  re->dummy1 = 0;  re->dummy1 = 0;
5928  re->first_byte = 0;  re->first_byte = 0;
5929  re->req_byte = 0;  re->req_byte = 0;
# Line 5270  codestart = cd->name_table + re->name_en Line 5948  codestart = cd->name_table + re->name_en
5948  cd->start_code = codestart;  cd->start_code = codestart;
5949  cd->hwm = cworkspace;  cd->hwm = cworkspace;
5950  cd->req_varyopt = 0;  cd->req_varyopt = 0;
5951  cd->nopartial = FALSE;  cd->had_accept = FALSE;
5952    
5953  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
5954  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
5955  of the function here. */  of the function here. */
5956    
5957  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
5958  code = (uschar *)codestart;  code = (uschar *)codestart;
5959  *code = OP_BRA;  *code = OP_BRA;
5960  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
5961    &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
5962  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
5963  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
5964    re->flags = cd->external_flags;
5965    
5966  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
5967    
5968  /* 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. */
5969    
# Line 5325  if (errorcode != 0) Line 6004  if (errorcode != 0)
6004    (pcre_free)(re);    (pcre_free)(re);
6005    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6006    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6007  #ifdef SUPPORT_UTF8    PCRE_EARLY_ERROR_RETURN2:
   PCRE_UTF8_ERROR_RETURN:  
 #endif  
6008    *errorptr = error_texts[errorcode];    *errorptr = error_texts[errorcode];
6009    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6010    return NULL;    return NULL;
# Line 5357  if ((re->options & PCRE_ANCHORED) == 0) Line 6034  if ((re->options & PCRE_ANCHORED) == 0)
6034        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6035        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6036           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6037        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6038        }        }
6039      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6040        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6041      }      }
6042    }    }
6043    
# Line 5374  if (reqbyte >= 0 && Line 6051  if (reqbyte >= 0 &&
6051    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6052    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6053      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6054    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6055    }    }
6056    
6057  /* Print out the compiled data if debugging is enabled. This is never the  /* Print out the compiled data if debugging is enabled. This is never the
# Line 5385  case when building a production library. Line 6062  case when building a production library.
6062  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6063    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6064    
6065  if (re->options != 0)  printf("Options=%08x\n", re->options);
   {  
   printf("%s%s%s%s%s%s%s%s%s\n",  
     ((re->options & PCRE_NOPARTIAL) != 0)? "nopartial " : "",  
     ((re->options & PCRE_ANCHORED) != 0)? "anchored " : "",  
     ((re->options & PCRE_CASELESS) != 0)? "caseless " : "",  
     ((re->options & PCRE_EXTENDED) != 0)? "extended " : "",  
     ((re->options & PCRE_MULTILINE) != 0)? "multiline " : "",  
     ((re->options & PCRE_DOTALL) != 0)? "dotall " : "",  
     ((re->options & PCRE_DOLLAR_ENDONLY) != 0)? "endonly " : "",  
     ((re->options & PCRE_EXTRA) != 0)? "extra " : "",  
     ((re->options & PCRE_UNGREEDY) != 0)? "ungreedy " : "");  
   }  
6066    
6067  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6068    {    {
6069    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6070    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5408  if ((re->options & PCRE_FIRSTSET) != 0) Line 6073  if ((re->options & PCRE_FIRSTSET) != 0)
6073      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6074    }    }
6075    
6076  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6077    {    {
6078    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6079    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5417  if ((re->options & PCRE_REQCHSET) != 0) Line 6082  if ((re->options & PCRE_REQCHSET) != 0)
6082      else printf("Req char = \\x%02x%s\n", ch, caseless);      else printf("Req char = \\x%02x%s\n", ch, caseless);
6083    }    }
6084    
6085  pcre_printint(re, stdout);  pcre_printint(re, stdout, TRUE);
6086    
6087  /* 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
6088  was compiled can be seen. */  was compiled can be seen. */

Legend:
Removed from v.107  
changed lines
  Added in v.231

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12