/[pcre]/code/branches/pcre16/pcre_compile.c
ViewVC logotype

Diff of /code/branches/pcre16/pcre_compile.c

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 406 by ph10, Mon Mar 23 12:05:43 2009 UTC revision 528 by ph10, Sat May 29 16:40:22 2010 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) 1997-2009 University of Cambridge             Copyright (c) 1997-2010 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 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When DEBUG is defined, we need the pcre_printint() function, which is also  /* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is
57  used by pcretest. DEBUG is not defined when building a production library. */  also used by pcretest. PCRE_DEBUG is not defined when building a production
58    library. */
59    
60  #ifdef DEBUG  #ifdef PCRE_DEBUG
61  #include "pcre_printint.src"  #include "pcre_printint.src"
62  #endif  #endif
63    
# Line 91  is 4 there is plenty of room. */ Line 92  is 4 there is plenty of room. */
92    
93  #define COMPILE_WORK_SIZE (4096)  #define COMPILE_WORK_SIZE (4096)
94    
95    /* The overrun tests check for a slightly smaller size so that they detect the
96    overrun before it actually does run off the end of the data block. */
97    
98    #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100)
99    
100    
101  /* Table for handling escaped characters in the range '0'-'z'. Positive returns  /* Table for handling escaped characters in the range '0'-'z'. Positive returns
102  are simple data values; negative values are for special things like \d and so  are simple data values; negative values are for special things like \d and so
# Line 118  static const short int escapes[] = { Line 124  static const short int escapes[] = {
124       -ESC_H,                  0,       -ESC_H,                  0,
125       0,                       -ESC_K,       0,                       -ESC_K,
126       0,                       0,       0,                       0,
127       0,                       0,       -ESC_N,                  0,
128       -ESC_P,                  -ESC_Q,       -ESC_P,                  -ESC_Q,
129       -ESC_R,                  -ESC_S,       -ESC_R,                  -ESC_S,
130       0,                       0,       0,                       0,
# Line 165  static const short int escapes[] = { Line 171  static const short int escapes[] = {
171  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',
172  /*  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,
173  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,
174  /*  D0 */   '}',     0, -ESC_K,       0,      0,     0,      0, -ESC_P,  /*  D0 */   '}',     0, -ESC_K,       0,      0,-ESC_N,      0, -ESC_P,
175  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,
176  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,
177  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,
# Line 182  string is built from string macros so th Line 188  string is built from string macros so th
188  platforms. */  platforms. */
189    
190  typedef struct verbitem {  typedef struct verbitem {
191    int   len;    int   len;                 /* Length of verb name */
192    int   op;    int   op;                  /* Op when no arg, or -1 if arg mandatory */
193      int   op_arg;              /* Op when arg present, or -1 if not allowed */
194  } verbitem;  } verbitem;
195    
196  static const char verbnames[] =  static const char verbnames[] =
197      "\0"                       /* Empty name is a shorthand for MARK */
198      STRING_MARK0
199    STRING_ACCEPT0    STRING_ACCEPT0
200    STRING_COMMIT0    STRING_COMMIT0
201    STRING_F0    STRING_F0
# Line 196  static const char verbnames[] = Line 205  static const char verbnames[] =
205    STRING_THEN;    STRING_THEN;
206    
207  static const verbitem verbs[] = {  static const verbitem verbs[] = {
208    { 6, OP_ACCEPT },    { 0, -1,        OP_MARK },
209    { 6, OP_COMMIT },    { 4, -1,        OP_MARK },
210    { 1, OP_FAIL },    { 6, OP_ACCEPT, -1 },
211    { 4, OP_FAIL },    { 6, OP_COMMIT, -1 },
212    { 5, OP_PRUNE },    { 1, OP_FAIL,   -1 },
213    { 4, OP_SKIP  },    { 4, OP_FAIL,   -1 },
214    { 4, OP_THEN  }    { 5, OP_PRUNE,  OP_PRUNE_ARG },
215      { 4, OP_SKIP,   OP_SKIP_ARG  },
216      { 4, OP_THEN,   OP_THEN_ARG  }
217  };  };
218    
219  static const int verbcount = sizeof(verbs)/sizeof(verbitem);  static const int verbcount = sizeof(verbs)/sizeof(verbitem);
# Line 250  static const int posix_class_maps[] = { Line 261  static const int posix_class_maps[] = {
261    cbit_xdigit,-1,          0              /* xdigit */    cbit_xdigit,-1,          0              /* xdigit */
262  };  };
263    
264    /* Table of substitutes for \d etc when PCRE_UCP is set. The POSIX class
265    substitutes must be in the order of the names, defined above, and there are
266    both positive and negative cases. NULL means no substitute. */
267    
268    #ifdef SUPPORT_UCP
269    static const uschar *substitutes[] = {
270      (uschar *)"\\P{Nd}",    /* \D */
271      (uschar *)"\\p{Nd}",    /* \d */
272      (uschar *)"\\P{Xsp}",   /* \S */       /* NOTE: Xsp is Perl space */
273      (uschar *)"\\p{Xsp}",   /* \s */
274      (uschar *)"\\P{Xwd}",   /* \W */
275      (uschar *)"\\p{Xwd}"    /* \w */
276    };
277    
278    static const uschar *posix_substitutes[] = {
279      (uschar *)"\\p{L}",     /* alpha */
280      (uschar *)"\\p{Ll}",    /* lower */
281      (uschar *)"\\p{Lu}",    /* upper */
282      (uschar *)"\\p{Xan}",   /* alnum */
283      NULL,                   /* ascii */
284      (uschar *)"\\h",        /* blank */
285      NULL,                   /* cntrl */
286      (uschar *)"\\p{Nd}",    /* digit */
287      NULL,                   /* graph */
288      NULL,                   /* print */
289      NULL,                   /* punct */
290      (uschar *)"\\p{Xps}",   /* space */    /* NOTE: Xps is POSIX space */
291      (uschar *)"\\p{Xwd}",   /* word */
292      NULL,                   /* xdigit */
293      /* Negated cases */
294      (uschar *)"\\P{L}",     /* ^alpha */
295      (uschar *)"\\P{Ll}",    /* ^lower */
296      (uschar *)"\\P{Lu}",    /* ^upper */
297      (uschar *)"\\P{Xan}",   /* ^alnum */
298      NULL,                   /* ^ascii */
299      (uschar *)"\\H",        /* ^blank */
300      NULL,                   /* ^cntrl */
301      (uschar *)"\\P{Nd}",    /* ^digit */
302      NULL,                   /* ^graph */
303      NULL,                   /* ^print */
304      NULL,                   /* ^punct */
305      (uschar *)"\\P{Xps}",   /* ^space */   /* NOTE: Xps is POSIX space */
306      (uschar *)"\\P{Xwd}",   /* ^word */
307      NULL                    /* ^xdigit */
308    };
309    #define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))
310    #endif
311    
312  #define STRING(a)  # a  #define STRING(a)  # a
313  #define XSTRING(s) STRING(s)  #define XSTRING(s) STRING(s)
# Line 262  the number of relocations needed when a Line 320  the number of relocations needed when a
320  it is now one long string. We cannot use a table of offsets, because the  it is now one long string. We cannot use a table of offsets, because the
321  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we
322  simply count through to the one we want - this isn't a performance issue  simply count through to the one we want - this isn't a performance issue
323  because these strings are used only when there is a compilation error. */  because these strings are used only when there is a compilation error.
324    
325    Each substring ends with \0 to insert a null character. This includes the final
326    substring, so that the whole string ends with \0\0, which can be detected when
327    counting through. */
328    
329  static const char error_texts[] =  static const char error_texts[] =
330    "no error\0"    "no error\0"
# Line 309  static const char error_texts[] = Line 371  static const char error_texts[] =
371    /* 35 */    /* 35 */
372    "invalid condition (?(0)\0"    "invalid condition (?(0)\0"
373    "\\C not allowed in lookbehind assertion\0"    "\\C not allowed in lookbehind assertion\0"
374    "PCRE does not support \\L, \\l, \\N, \\U, or \\u\0"    "PCRE does not support \\L, \\l, \\N{name}, \\U, or \\u\0"
375    "number after (?C is > 255\0"    "number after (?C is > 255\0"
376    "closing ) for (?C expected\0"    "closing ) for (?C expected\0"
377    /* 40 */    /* 40 */
# Line 335  static const char error_texts[] = Line 397  static const char error_texts[] =
397    "inconsistent NEWLINE options\0"    "inconsistent NEWLINE options\0"
398    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"
399    "a numbered reference must not be zero\0"    "a numbered reference must not be zero\0"
400    "(*VERB) with an argument is not supported\0"    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"
401    /* 60 */    /* 60 */
402    "(*VERB) not recognized\0"    "(*VERB) not recognized\0"
403    "number is too big\0"    "number is too big\0"
404    "subpattern name expected\0"    "subpattern name expected\0"
405    "digit expected after (?+\0"    "digit expected after (?+\0"
406    "] is an invalid data character in JavaScript compatibility mode";    "] is an invalid data character in JavaScript compatibility mode\0"
407      /* 65 */
408      "different names for subpatterns of the same number are not allowed\0"
409      "(*MARK) must have an argument\0"
410      "this version of PCRE is not compiled with PCRE_UCP support\0"
411      ;
412    
413  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
414  patterns. Note that the tables in chartables are dependent on the locale, and  patterns. Note that the tables in chartables are dependent on the locale, and
# Line 500  static const char * Line 566  static const char *
566  find_error_text(int n)  find_error_text(int n)
567  {  {
568  const char *s = error_texts;  const char *s = error_texts;
569  for (; n > 0; n--) while (*s++ != 0) {};  for (; n > 0; n--)
570      {
571      while (*s++ != 0) {};
572      if (*s == 0) return "Error text not found (please report)";
573      }
574  return s;  return s;
575  }  }
576    
# Line 571  else Line 641  else
641    
642      case CHAR_l:      case CHAR_l:
643      case CHAR_L:      case CHAR_L:
     case CHAR_N:  
644      case CHAR_u:      case CHAR_u:
645      case CHAR_U:      case CHAR_U:
646      *errorcodeptr = ERR37;      *errorcodeptr = ERR37;
# Line 809  else Line 878  else
878      }      }
879    }    }
880    
881    /* Perl supports \N{name} for character names, as well as plain \N for "not
882    newline". PCRE does not support \N{name}. */
883    
884    if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET)
885      *errorcodeptr = ERR37;
886    
887    /* If PCRE_UCP is set, we change the values for \d etc. */
888    
889    if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)
890      c -= (ESC_DU - ESC_D);
891    
892    /* Set the pointer to the final character before returning. */
893    
894  *ptrptr = ptr;  *ptrptr = ptr;
895  return c;  return c;
896  }  }
# Line 1009  return p; Line 1091  return p;
1091    
1092    
1093  /*************************************************  /*************************************************
1094  *       Find forward referenced subpattern       *  *  Subroutine for finding forward reference      *
1095  *************************************************/  *************************************************/
1096    
1097  /* This function scans along a pattern's text looking for capturing  /* This recursive function is called only from find_parens() below. The
1098    top-level call starts at the beginning of the pattern. All other calls must
1099    start at a parenthesis. It scans along a pattern's text looking for capturing
1100  subpatterns, and counting them. If it finds a named pattern that matches the  subpatterns, and counting them. If it finds a named pattern that matches the
1101  name it is given, it returns its number. Alternatively, if the name is NULL, it  name it is given, it returns its number. Alternatively, if the name is NULL, it
1102  returns when it reaches a given numbered subpattern. This is used for forward  returns when it reaches a given numbered subpattern. We know that if (?P< is
1103  references to subpatterns. We know that if (?P< is encountered, the name will  encountered, the name will be terminated by '>' because that is checked in the
1104  be terminated by '>' because that is checked in the first pass.  first pass. Recursion is used to keep track of subpatterns that reset the
1105    capturing group numbers - the (?| feature.
1106    
1107  Arguments:  Arguments:
1108    ptr          current position in the pattern    ptrptr       address of the current character pointer (updated)
1109    cd           compile background data    cd           compile background data
1110    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1111    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1112    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1113      count        pointer to the current capturing subpattern number (updated)
1114    
1115  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1116  */  */
1117    
1118  static int  static int
1119  find_parens(const uschar *ptr, compile_data *cd, const uschar *name, int lorn,  find_parens_sub(uschar **ptrptr, compile_data *cd, const uschar *name, int lorn,
1120    BOOL xmode)    BOOL xmode, int *count)
1121  {  {
1122  const uschar *thisname;  uschar *ptr = *ptrptr;
1123  int count = cd->bracount;  int start_count = *count;
1124    int hwm_count = start_count;
1125    BOOL dup_parens = FALSE;
1126    
1127  for (; *ptr != 0; ptr++)  /* If the first character is a parenthesis, check on the type of group we are
1128    dealing with. The very first call may not start with a parenthesis. */
1129    
1130    if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1131    {    {
1132    int term;    if (ptr[1] == CHAR_QUESTION_MARK &&
1133          ptr[2] == CHAR_VERTICAL_LINE)
1134        {
1135        ptr += 3;
1136        dup_parens = TRUE;
1137        }
1138    
1139      /* Handle a normal, unnamed capturing parenthesis */
1140    
1141      else if (ptr[1] != CHAR_QUESTION_MARK && ptr[1] != CHAR_ASTERISK)
1142        {
1143        *count += 1;
1144        if (name == NULL && *count == lorn) return *count;
1145        ptr++;
1146        }
1147    
1148      /* Handle a condition. If it is an assertion, just carry on so that it
1149      is processed as normal. If not, skip to the closing parenthesis of the
1150      condition (there can't be any nested parens. */
1151    
1152      else if (ptr[2] == CHAR_LEFT_PARENTHESIS)
1153        {
1154        ptr += 2;
1155        if (ptr[1] != CHAR_QUESTION_MARK)
1156          {
1157          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1158          if (*ptr != 0) ptr++;
1159          }
1160        }
1161    
1162      /* We have either (? or (* and not a condition */
1163    
1164      else
1165        {
1166        ptr += 2;
1167        if (*ptr == CHAR_P) ptr++;                      /* Allow optional P */
1168    
1169        /* We have to disambiguate (?<! and (?<= from (?<name> for named groups */
1170    
1171        if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1172            ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1173          {
1174          int term;
1175          const uschar *thisname;
1176          *count += 1;
1177          if (name == NULL && *count == lorn) return *count;
1178          term = *ptr++;
1179          if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1180          thisname = ptr;
1181          while (*ptr != term) ptr++;
1182          if (name != NULL && lorn == ptr - thisname &&
1183              strncmp((const char *)name, (const char *)thisname, lorn) == 0)
1184            return *count;
1185          term++;
1186          }
1187        }
1188      }
1189    
1190    /* Past any initial parenthesis handling, scan for parentheses or vertical
1191    bars. */
1192    
1193    for (; *ptr != 0; ptr++)
1194      {
1195    /* Skip over backslashed characters and also entire \Q...\E */    /* Skip over backslashed characters and also entire \Q...\E */
1196    
1197    if (*ptr == CHAR_BACKSLASH)    if (*ptr == CHAR_BACKSLASH)
1198      {      {
1199      if (*(++ptr) == 0) return -1;      if (*(++ptr) == 0) goto FAIL_EXIT;
1200      if (*ptr == CHAR_Q) for (;;)      if (*ptr == CHAR_Q) for (;;)
1201        {        {
1202        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};
1203        if (*ptr == 0) return -1;        if (*ptr == 0) goto FAIL_EXIT;
1204        if (*(++ptr) == CHAR_E) break;        if (*(++ptr) == CHAR_E) break;
1205        }        }
1206      continue;      continue;
# Line 1065  for (; *ptr != 0; ptr++) Line 1217  for (; *ptr != 0; ptr++)
1217      BOOL negate_class = FALSE;      BOOL negate_class = FALSE;
1218      for (;;)      for (;;)
1219        {        {
1220        int c = *(++ptr);        if (ptr[1] == CHAR_BACKSLASH)
       if (c == CHAR_BACKSLASH)  
1221          {          {
1222          if (ptr[1] == CHAR_E)          if (ptr[2] == CHAR_E)
1223            ptr++;            ptr+= 2;
1224          else if (strncmp((const char *)ptr+1,          else if (strncmp((const char *)ptr+2,
1225                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1226            ptr += 3;            ptr += 4;
1227          else          else
1228            break;            break;
1229          }          }
1230        else if (!negate_class && c == CHAR_CIRCUMFLEX_ACCENT)        else if (!negate_class && ptr[1] == CHAR_CIRCUMFLEX_ACCENT)
1231            {
1232          negate_class = TRUE;          negate_class = TRUE;
1233            ptr++;
1234            }
1235        else break;        else break;
1236        }        }
1237    
# Line 1093  for (; *ptr != 0; ptr++) Line 1247  for (; *ptr != 0; ptr++)
1247        if (*ptr == 0) return -1;        if (*ptr == 0) return -1;
1248        if (*ptr == CHAR_BACKSLASH)        if (*ptr == CHAR_BACKSLASH)
1249          {          {
1250          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) goto FAIL_EXIT;
1251          if (*ptr == CHAR_Q) for (;;)          if (*ptr == CHAR_Q) for (;;)
1252            {            {
1253            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};
1254            if (*ptr == 0) return -1;            if (*ptr == 0) goto FAIL_EXIT;
1255            if (*(++ptr) == CHAR_E) break;            if (*(++ptr) == CHAR_E) break;
1256            }            }
1257          continue;          continue;
# Line 1111  for (; *ptr != 0; ptr++) Line 1265  for (; *ptr != 0; ptr++)
1265    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1266      {      {
1267      while (*(++ptr) != 0 && *ptr != CHAR_NL) {};      while (*(++ptr) != 0 && *ptr != CHAR_NL) {};
1268      if (*ptr == 0) return -1;      if (*ptr == 0) goto FAIL_EXIT;
1269      continue;      continue;
1270      }      }
1271    
1272    /* An opening parens must now be a real metacharacter */    /* Check for the special metacharacters */
1273    
1274    if (*ptr != CHAR_LEFT_PARENTHESIS) continue;    if (*ptr == CHAR_LEFT_PARENTHESIS)
   if (ptr[1] != CHAR_QUESTION_MARK && ptr[1] != CHAR_ASTERISK)  
1275      {      {
1276      count++;      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, count);
1277      if (name == NULL && count == lorn) return count;      if (rc > 0) return rc;
1278      continue;      if (*ptr == 0) goto FAIL_EXIT;
1279      }      }
1280    
1281    ptr += 2;    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
1282    if (*ptr == CHAR_P) ptr++;                      /* Allow optional P */      {
1283        if (dup_parens && *count < hwm_count) *count = hwm_count;
1284        *ptrptr = ptr;
1285        return -1;
1286        }
1287    
1288    /* We have to disambiguate (?<! and (?<= from (?<name> */    else if (*ptr == CHAR_VERTICAL_LINE && dup_parens)
1289        {
1290        if (*count > hwm_count) hwm_count = *count;
1291        *count = start_count;
1292        }
1293      }
1294    
1295    if ((*ptr != CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_EXCLAMATION_MARK ||  FAIL_EXIT:
1296        ptr[1] == CHAR_EQUALS_SIGN) && *ptr != CHAR_APOSTROPHE)  *ptrptr = ptr;
1297      continue;  return -1;
1298    }
1299    
1300    
1301    
1302    
1303    /*************************************************
1304    *       Find forward referenced subpattern       *
1305    *************************************************/
1306    
1307    count++;  /* This function scans along a pattern's text looking for capturing
1308    subpatterns, and counting them. If it finds a named pattern that matches the
1309    name it is given, it returns its number. Alternatively, if the name is NULL, it
1310    returns when it reaches a given numbered subpattern. This is used for forward
1311    references to subpatterns. We used to be able to start this scan from the
1312    current compiling point, using the current count value from cd->bracount, and
1313    do it all in a single loop, but the addition of the possibility of duplicate
1314    subpattern numbers means that we have to scan from the very start, in order to
1315    take account of such duplicates, and to use a recursive function to keep track
1316    of the different types of group.
1317    
1318    Arguments:
1319      cd           compile background data
1320      name         name to seek, or NULL if seeking a numbered subpattern
1321      lorn         name length, or subpattern number if name is NULL
1322      xmode        TRUE if we are in /x mode
1323    
1324    if (name == NULL && count == lorn) return count;  Returns:       the number of the found subpattern, or -1 if not found
1325    term = *ptr++;  */
1326    if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;  
1327    thisname = ptr;  static int
1328    while (*ptr != term) ptr++;  find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode)
1329    if (name != NULL && lorn == ptr - thisname &&  {
1330        strncmp((const char *)name, (const char *)thisname, lorn) == 0)  uschar *ptr = (uschar *)cd->start_pattern;
1331      return count;  int count = 0;
1332    int rc;
1333    
1334    /* If the pattern does not start with an opening parenthesis, the first call
1335    to find_parens_sub() will scan right to the end (if necessary). However, if it
1336    does start with a parenthesis, find_parens_sub() will return when it hits the
1337    matching closing parens. That is why we have to have a loop. */
1338    
1339    for (;;)
1340      {
1341      rc = find_parens_sub(&ptr, cd, name, lorn, xmode, &count);
1342      if (rc > 0 || *ptr++ == 0) break;
1343    }    }
1344    
1345  return -1;  return rc;
1346  }  }
1347    
1348    
1349    
1350    
1351  /*************************************************  /*************************************************
1352  *      Find first significant op code            *  *      Find first significant op code            *
1353  *************************************************/  *************************************************/
# Line 1200  for (;;) Line 1397  for (;;)
1397    
1398      case OP_CALLOUT:      case OP_CALLOUT:
1399      case OP_CREF:      case OP_CREF:
1400        case OP_NCREF:
1401      case OP_RREF:      case OP_RREF:
1402        case OP_NRREF:
1403      case OP_DEF:      case OP_DEF:
1404      code += _pcre_OP_lengths[*code];      code += _pcre_OP_lengths[*code];
1405      break;      break;
# Line 1216  for (;;) Line 1415  for (;;)
1415    
1416    
1417  /*************************************************  /*************************************************
1418  *        Find the fixed length of a pattern      *  *        Find the fixed length of a branch       *
1419  *************************************************/  *************************************************/
1420    
1421  /* Scan a pattern and compute the fixed length of subject that will match it,  /* Scan a branch and compute the fixed length of subject that will match it,
1422  if the length is fixed. This is needed for dealing with backward assertions.  if the length is fixed. This is needed for dealing with backward assertions.
1423  In UTF8 mode, the result is in characters rather than bytes.  In UTF8 mode, the result is in characters rather than bytes. The branch is
1424    temporarily terminated with OP_END when this function is called.
1425    
1426    This function is called when a backward assertion is encountered, so that if it
1427    fails, the error message can point to the correct place in the pattern.
1428    However, we cannot do this when the assertion contains subroutine calls,
1429    because they can be forward references. We solve this by remembering this case
1430    and doing the check at the end; a flag specifies which mode we are running in.
1431    
1432  Arguments:  Arguments:
1433    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1434    options  the compiling options    options  the compiling options
1435      atend    TRUE if called when the pattern is complete
1436      cd       the "compile data" structure
1437    
1438  Returns:   the fixed length, or -1 if there is no fixed length,  Returns:   the fixed length,
1439                 or -1 if there is no fixed length,
1440               or -2 if \C was encountered               or -2 if \C was encountered
1441                 or -3 if an OP_RECURSE item was encountered and atend is FALSE
1442  */  */
1443    
1444  static int  static int
1445  find_fixedlength(uschar *code, int options)  find_fixedlength(uschar *code, int options, BOOL atend, compile_data *cd)
1446  {  {
1447  int length = -1;  int length = -1;
1448    
# Line 1245  branch, check the length against that of Line 1455  branch, check the length against that of
1455  for (;;)  for (;;)
1456    {    {
1457    int d;    int d;
1458      uschar *ce, *cs;
1459    register int op = *cc;    register int op = *cc;
1460    switch (op)    switch (op)
1461      {      {
# Line 1252  for (;;) Line 1463  for (;;)
1463      case OP_BRA:      case OP_BRA:
1464      case OP_ONCE:      case OP_ONCE:
1465      case OP_COND:      case OP_COND:
1466      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), options);      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), options, atend, cd);
1467      if (d < 0) return d;      if (d < 0) return d;
1468      branchlength += d;      branchlength += d;
1469      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
# Line 1275  for (;;) Line 1486  for (;;)
1486      branchlength = 0;      branchlength = 0;
1487      break;      break;
1488    
1489        /* A true recursion implies not fixed length, but a subroutine call may
1490        be OK. If the subroutine is a forward reference, we can't deal with
1491        it until the end of the pattern, so return -3. */
1492    
1493        case OP_RECURSE:
1494        if (!atend) return -3;
1495        cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1496        do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */
1497        if (cc > cs && cc < ce) return -1;                /* Recursion */
1498        d = find_fixedlength(cs + 2, options, atend, cd);
1499        if (d < 0) return d;
1500        branchlength += d;
1501        cc += 1 + LINK_SIZE;
1502        break;
1503    
1504      /* Skip over assertive subpatterns */      /* Skip over assertive subpatterns */
1505    
1506      case OP_ASSERT:      case OP_ASSERT:
# Line 1288  for (;;) Line 1514  for (;;)
1514    
1515      case OP_REVERSE:      case OP_REVERSE:
1516      case OP_CREF:      case OP_CREF:
1517        case OP_NCREF:
1518      case OP_RREF:      case OP_RREF:
1519        case OP_NRREF:
1520      case OP_DEF:      case OP_DEF:
1521      case OP_OPT:      case OP_OPT:
1522      case OP_CALLOUT:      case OP_CALLOUT:
1523      case OP_SOD:      case OP_SOD:
1524      case OP_SOM:      case OP_SOM:
1525        case OP_SET_SOM:
1526      case OP_EOD:      case OP_EOD:
1527      case OP_EODN:      case OP_EODN:
1528      case OP_CIRC:      case OP_CIRC:
# Line 1311  for (;;) Line 1540  for (;;)
1540      branchlength++;      branchlength++;
1541      cc += 2;      cc += 2;
1542  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1543      if ((options & PCRE_UTF8) != 0)      if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0)
1544        {        cc += _pcre_utf8_table4[cc[-1] & 0x3f];
       while ((*cc & 0xc0) == 0x80) cc++;  
       }  
1545  #endif  #endif
1546      break;      break;
1547    
# Line 1325  for (;;) Line 1552  for (;;)
1552      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1553      cc += 4;      cc += 4;
1554  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1555      if ((options & PCRE_UTF8) != 0)      if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0)
1556        {        cc += _pcre_utf8_table4[cc[-1] & 0x3f];
       while((*cc & 0x80) == 0x80) cc++;  
       }  
1557  #endif  #endif
1558      break;      break;
1559    
# Line 1407  for (;;) Line 1632  for (;;)
1632    
1633    
1634  /*************************************************  /*************************************************
1635  *    Scan compiled regex for numbered bracket    *  *    Scan compiled regex for specific bracket    *
1636  *************************************************/  *************************************************/
1637    
1638  /* This little function scans through a compiled pattern until it finds a  /* This little function scans through a compiled pattern until it finds a
1639  capturing bracket with the given number.  capturing bracket with the given number, or, if the number is negative, an
1640    instance of OP_REVERSE for a lookbehind. The function is global in the C sense
1641    so that it can be called from pcre_study() when finding the minimum matching
1642    length.
1643    
1644  Arguments:  Arguments:
1645    code        points to start of expression    code        points to start of expression
1646    utf8        TRUE in UTF-8 mode    utf8        TRUE in UTF-8 mode
1647    number      the required bracket number    number      the required bracket number or negative to find a lookbehind
1648    
1649  Returns:      pointer to the opcode for the bracket, or NULL if not found  Returns:      pointer to the opcode for the bracket, or NULL if not found
1650  */  */
1651    
1652  static const uschar *  const uschar *
1653  find_bracket(const uschar *code, BOOL utf8, int number)  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)
1654  {  {
1655  for (;;)  for (;;)
1656    {    {
# Line 1435  for (;;) Line 1663  for (;;)
1663    
1664    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1665    
1666      /* Handle recursion */
1667    
1668      else if (c == OP_REVERSE)
1669        {
1670        if (number < 0) return (uschar *)code;
1671        code += _pcre_OP_lengths[c];
1672        }
1673    
1674    /* Handle capturing bracket */    /* Handle capturing bracket */
1675    
1676    else if (c == OP_CBRA)    else if (c == OP_CBRA)
# Line 1446  for (;;) Line 1682  for (;;)
1682    
1683    /* Otherwise, we can get the item's length from the table, except that for    /* Otherwise, we can get the item's length from the table, except that for
1684    repeated character types, we have to test for \p and \P, which have an extra    repeated character types, we have to test for \p and \P, which have an extra
1685    two bytes of parameters. */    two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we
1686      must add in its length. */
1687    
1688    else    else
1689      {      {
# Line 1470  for (;;) Line 1707  for (;;)
1707        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
1708        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1709        break;        break;
1710    
1711          case OP_MARK:
1712          case OP_PRUNE_ARG:
1713          case OP_SKIP_ARG:
1714          case OP_THEN_ARG:
1715          code += code[1];
1716          break;
1717        }        }
1718    
1719      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
# Line 1541  for (;;) Line 1785  for (;;)
1785    
1786    /* Otherwise, we can get the item's length from the table, except that for    /* Otherwise, we can get the item's length from the table, except that for
1787    repeated character types, we have to test for \p and \P, which have an extra    repeated character types, we have to test for \p and \P, which have an extra
1788    two bytes of parameters. */    two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we
1789      must add in its length. */
1790    
1791    else    else
1792      {      {
# Line 1565  for (;;) Line 1810  for (;;)
1810        case OP_TYPEEXACT:        case OP_TYPEEXACT:
1811        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1812        break;        break;
1813    
1814          case OP_MARK:
1815          case OP_PRUNE_ARG:
1816          case OP_SKIP_ARG:
1817          case OP_THEN_ARG:
1818          code += code[1];
1819          break;
1820        }        }
1821    
1822      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
# Line 1621  Arguments: Line 1873  Arguments:
1873    code        points to start of search    code        points to start of search
1874    endcode     points to where to stop    endcode     points to where to stop
1875    utf8        TRUE if in UTF8 mode    utf8        TRUE if in UTF8 mode
1876      cd          contains pointers to tables etc.
1877    
1878  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
1879  */  */
1880    
1881  static BOOL  static BOOL
1882  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8)  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,
1883      compile_data *cd)
1884  {  {
1885  register int c;  register int c;
1886  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);
# Line 1657  for (code = first_significant_code(code Line 1911  for (code = first_significant_code(code
1911      continue;      continue;
1912      }      }
1913    
1914      /* For a recursion/subroutine call, if its end has been reached, which
1915      implies a subroutine call, we can scan it. */
1916    
1917      if (c == OP_RECURSE)
1918        {
1919        BOOL empty_branch = FALSE;
1920        const uschar *scode = cd->start_code + GET(code, 1);
1921        if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
1922        do
1923          {
1924          if (could_be_empty_branch(scode, endcode, utf8, cd))
1925            {
1926            empty_branch = TRUE;
1927            break;
1928            }
1929          scode += GET(scode, 1);
1930          }
1931        while (*scode == OP_ALT);
1932        if (!empty_branch) return FALSE;  /* All branches are non-empty */
1933        continue;
1934        }
1935    
1936    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1937    
1938    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
# Line 1675  for (code = first_significant_code(code Line 1951  for (code = first_significant_code(code
1951        empty_branch = FALSE;        empty_branch = FALSE;
1952        do        do
1953          {          {
1954          if (!empty_branch && could_be_empty_branch(code, endcode, utf8))          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))
1955            empty_branch = TRUE;            empty_branch = TRUE;
1956          code += GET(code, 1);          code += GET(code, 1);
1957          }          }
# Line 1800  for (code = first_significant_code(code Line 2076  for (code = first_significant_code(code
2076      case OP_QUERY:      case OP_QUERY:
2077      case OP_MINQUERY:      case OP_MINQUERY:
2078      case OP_POSQUERY:      case OP_POSQUERY:
2079        if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];
2080        break;
2081    
2082      case OP_UPTO:      case OP_UPTO:
2083      case OP_MINUPTO:      case OP_MINUPTO:
2084      case OP_POSUPTO:      case OP_POSUPTO:
2085      if (utf8) while ((code[2] & 0xc0) == 0x80) code++;      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];
2086      break;      break;
2087  #endif  #endif
2088    
2089        /* MARK, and PRUNE/SKIP/THEN with an argument must skip over the argument
2090        string. */
2091    
2092        case OP_MARK:
2093        case OP_PRUNE_ARG:
2094        case OP_SKIP_ARG:
2095        case OP_THEN_ARG:
2096        code += code[1];
2097        break;
2098    
2099        /* None of the remaining opcodes are required to match a character. */
2100    
2101        default:
2102        break;
2103      }      }
2104    }    }
2105    
# Line 1828  Arguments: Line 2122  Arguments:
2122    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2123    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2124    utf8        TRUE if in UTF-8 mode    utf8        TRUE if in UTF-8 mode
2125      cd          pointers to tables etc
2126    
2127  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2128  */  */
2129    
2130  static BOOL  static BOOL
2131  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,
2132    BOOL utf8)    BOOL utf8, compile_data *cd)
2133  {  {
2134  while (bcptr != NULL && bcptr->current >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2135    {    {
2136    if (!could_be_empty_branch(bcptr->current, endcode, utf8)) return FALSE;    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))
2137        return FALSE;
2138    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2139    }    }
2140  return TRUE;  return TRUE;
# Line 2248  if (next >= 0) switch(op_code) Line 2544  if (next >= 0) switch(op_code)
2544    else    else
2545  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2546    return (item == cd->fcc[next]);  /* Non-UTF-8 mode */    return (item == cd->fcc[next]);  /* Non-UTF-8 mode */
2547    
2548      /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.
2549      When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
2550    
2551    case OP_DIGIT:    case OP_DIGIT:
2552    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;
# Line 2290  if (next >= 0) switch(op_code) Line 2589  if (next >= 0) switch(op_code)
2589      case 0x202f:      case 0x202f:
2590      case 0x205f:      case 0x205f:
2591      case 0x3000:      case 0x3000:
2592      return op_code != OP_HSPACE;      return op_code == OP_NOT_HSPACE;
2593      default:      default:
2594      return op_code == OP_HSPACE;      return op_code != OP_NOT_HSPACE;
2595      }      }
2596    
2597      case OP_ANYNL:
2598    case OP_VSPACE:    case OP_VSPACE:
2599    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
2600    switch(next)    switch(next)
# Line 2306  if (next >= 0) switch(op_code) Line 2606  if (next >= 0) switch(op_code)
2606      case 0x85:      case 0x85:
2607      case 0x2028:      case 0x2028:
2608      case 0x2029:      case 0x2029:
2609      return op_code != OP_VSPACE;      return op_code == OP_NOT_VSPACE;
2610      default:      default:
2611      return op_code == OP_VSPACE;      return op_code != OP_NOT_VSPACE;
2612      }      }
2613    
2614    default:    default:
# Line 2316  if (next >= 0) switch(op_code) Line 2616  if (next >= 0) switch(op_code)
2616    }    }
2617    
2618    
2619  /* Handle the case when the next item is \d, \s, etc. */  /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP
2620    is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are
2621    generated only when PCRE_UCP is *not* set, that is, when only ASCII
2622    characteristics are recognized. */
2623    
2624  switch(op_code)  switch(op_code)
2625    {    {
# Line 2395  switch(op_code) Line 2698  switch(op_code)
2698    
2699    case OP_DIGIT:    case OP_DIGIT:
2700    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2701           next == -ESC_h || next == -ESC_v;           next == -ESC_h || next == -ESC_v || next == -ESC_R;
2702    
2703    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2704    return next == -ESC_d;    return next == -ESC_d;
2705    
2706    case OP_WHITESPACE:    case OP_WHITESPACE:
2707    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;
2708    
2709    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2710    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2711    
2712    case OP_HSPACE:    case OP_HSPACE:
2713    return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||
2714             next == -ESC_w || next == -ESC_v || next == -ESC_R;
2715    
2716    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
2717    return next == -ESC_h;    return next == -ESC_h;
2718    
2719    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
2720      case OP_ANYNL:
2721    case OP_VSPACE:    case OP_VSPACE:
2722    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2723    
2724    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
2725    return next == -ESC_v;    return next == -ESC_v || next == -ESC_R;
2726    
2727    case OP_WORDCHAR:    case OP_WORDCHAR:
2728    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||
2729             next == -ESC_v || next == -ESC_R;
2730    
2731    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2732    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2484  BOOL inescq = FALSE; Line 2790  BOOL inescq = FALSE;
2790  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstbyte = FALSE;
2791  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
2792  const uschar *tempptr;  const uschar *tempptr;
2793    const uschar *nestptr = NULL;
2794  uschar *previous = NULL;  uschar *previous = NULL;
2795  uschar *previous_callout = NULL;  uschar *previous_callout = NULL;
2796  uschar *save_hwm = NULL;  uschar *save_hwm = NULL;
# Line 2500  BOOL utf8 = FALSE; Line 2807  BOOL utf8 = FALSE;
2807  uschar *utf8_char = NULL;  uschar *utf8_char = NULL;
2808  #endif  #endif
2809    
2810  #ifdef DEBUG  #ifdef PCRE_DEBUG
2811  if (lengthptr != NULL) DPRINTF((">> start branch\n"));  if (lengthptr != NULL) DPRINTF((">> start branch\n"));
2812  #endif  #endif
2813    
# Line 2554  for (;; ptr++) Line 2861  for (;; ptr++)
2861    
2862    c = *ptr;    c = *ptr;
2863    
2864      /* If we are at the end of a nested substitution, revert to the outer level
2865      string. Nesting only happens one level deep. */
2866    
2867      if (c == 0 && nestptr != NULL)
2868        {
2869        ptr = nestptr;
2870        nestptr = NULL;
2871        c = *ptr;
2872        }
2873    
2874    /* If we are in the pre-compile phase, accumulate the length used for the    /* If we are in the pre-compile phase, accumulate the length used for the
2875    previous cycle of this loop. */    previous cycle of this loop. */
2876    
2877    if (lengthptr != NULL)    if (lengthptr != NULL)
2878      {      {
2879  #ifdef DEBUG  #ifdef PCRE_DEBUG
2880      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
2881  #endif  #endif
2882      if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */
2883        {        {
2884        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
2885        goto FAILED;        goto FAILED;
# Line 2611  for (;; ptr++) Line 2928  for (;; ptr++)
2928    /* In the real compile phase, just check the workspace used by the forward    /* In the real compile phase, just check the workspace used by the forward
2929    reference list. */    reference list. */
2930    
2931    else if (cd->hwm > cd->start_workspace + COMPILE_WORK_SIZE)    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)
2932      {      {
2933      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
2934      goto FAILED;      goto FAILED;
# Line 2851  for (;; ptr++) Line 3168  for (;; ptr++)
3168          {                           /* Braces are required because the */          {                           /* Braces are required because the */
3169          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
3170          }          }
3171    
3172        /* In the pre-compile phase, accumulate the length of any UTF-8 extra        /* In the pre-compile phase, accumulate the length of any UTF-8 extra
3173        data and reset the pointer. This is so that very large classes that        data and reset the pointer. This is so that very large classes that
3174        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion UTF-8 characters no longer overwrite the work space
# Line 2920  for (;; ptr++) Line 3237  for (;; ptr++)
3237    
3238          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
3239            posix_class = 0;            posix_class = 0;
3240    
3241          /* We build the bit map for the POSIX class in a chunk of local store          /* When PCRE_UCP is set, some of the POSIX classes are converted to
3242          because we may be adding and subtracting from it, and we don't want to          different escape sequences that use Unicode properties. */
3243          subtract bits that may be in the main map already. At the end we or the  
3244          result into the bit map that is being built. */  #ifdef SUPPORT_UCP
3245            if ((options & PCRE_UCP) != 0)
3246              {
3247              int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);
3248              if (posix_substitutes[pc] != NULL)
3249                {
3250                nestptr = tempptr + 1;
3251                ptr = posix_substitutes[pc] - 1;
3252                continue;
3253                }
3254              }
3255    #endif
3256            /* In the non-UCP case, we build the bit map for the POSIX class in a
3257            chunk of local store because we may be adding and subtracting from it,
3258            and we don't want to subtract bits that may be in the main map already.
3259            At the end we or the result into the bit map that is being built. */
3260    
3261          posix_class *= 3;          posix_class *= 3;
3262    
# Line 2968  for (;; ptr++) Line 3300  for (;; ptr++)
3300    
3301        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
3302        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
3303        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace. We
3304        Elsewhere it marks a word boundary. Other escapes have preset maps ready        assume that other escapes have more than one character in them, so set
3305        to 'or' into the one we are building. We assume they have more than one        class_charcount bigger than one. Unrecognized escapes fall through and
3306        character in them, so set class_charcount bigger than one. */        are either treated as literal characters (by default), or are faulted if
3307          PCRE_EXTRA is set. */
3308    
3309        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
3310          {          {
3311          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3312          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
3313    
3314          if (-c == ESC_b) c = CHAR_BS;       /* \b is backspace in a class */          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */
         else if (-c == ESC_X) c = CHAR_X;   /* \X is literal X in a class */  
         else if (-c == ESC_R) c = CHAR_R;   /* \R is literal R in a class */  
3315          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
3316            {            {
3317            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 2997  for (;; ptr++) Line 3328  for (;; ptr++)
3328            register const uschar *cbits = cd->cbits;            register const uschar *cbits = cd->cbits;
3329            class_charcount += 2;     /* Greater than 1 is what matters */            class_charcount += 2;     /* Greater than 1 is what matters */
3330    
3331            /* Save time by not doing this in the pre-compile phase. */            switch (-c)
   
           if (lengthptr == NULL) switch (-c)  
3332              {              {
3333    #ifdef SUPPORT_UCP
3334                case ESC_du:     /* These are the values given for \d etc */
3335                case ESC_DU:     /* when PCRE_UCP is set. We replace the */
3336                case ESC_wu:     /* escape sequence with an appropriate \p */
3337                case ESC_WU:     /* or \P to test Unicode properties instead */
3338                case ESC_su:     /* of the default ASCII testing. */
3339                case ESC_SU:
3340                nestptr = ptr;
3341                ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
3342                class_charcount -= 2;                /* Undo! */
3343                continue;
3344    #endif
3345              case ESC_d:              case ESC_d:
3346              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];
3347              continue;              continue;
# Line 3030  for (;; ptr++) Line 3371  for (;; ptr++)
3371              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
3372              continue;              continue;
3373    
3374              default:    /* Not recognized; fall through */              case ESC_h:
             break;      /* Need "default" setting to stop compiler warning. */  
             }  
   
           /* In the pre-compile phase, just do the recognition. */  
   
           else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||  
                    c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;  
   
           /* We need to deal with \H, \h, \V, and \v in both phases because  
           they use extra memory. */  
   
           if (-c == ESC_h)  
             {  
3375              SETBIT(classbits, 0x09); /* VT */              SETBIT(classbits, 0x09); /* VT */
3376              SETBIT(classbits, 0x20); /* SPACE */              SETBIT(classbits, 0x20); /* SPACE */
3377              SETBIT(classbits, 0xa0); /* NSBP */              SETBIT(classbits, 0xa0); /* NSBP */
# Line 3067  for (;; ptr++) Line 3395  for (;; ptr++)
3395                }                }
3396  #endif  #endif
3397              continue;              continue;
             }  
3398    
3399            if (-c == ESC_H)              case ESC_H:
             {  
3400              for (c = 0; c < 32; c++)              for (c = 0; c < 32; c++)
3401                {                {
3402                int x = 0xff;                int x = 0xff;
# Line 3112  for (;; ptr++) Line 3438  for (;; ptr++)
3438                }                }
3439  #endif  #endif
3440              continue;              continue;
             }  
3441    
3442            if (-c == ESC_v)              case ESC_v:
             {  
3443              SETBIT(classbits, 0x0a); /* LF */              SETBIT(classbits, 0x0a); /* LF */
3444              SETBIT(classbits, 0x0b); /* VT */              SETBIT(classbits, 0x0b); /* VT */
3445              SETBIT(classbits, 0x0c); /* FF */              SETBIT(classbits, 0x0c); /* FF */
# Line 3131  for (;; ptr++) Line 3455  for (;; ptr++)
3455                }                }
3456  #endif  #endif
3457              continue;              continue;
             }  
3458    
3459            if (-c == ESC_V)              case ESC_V:
             {  
3460              for (c = 0; c < 32; c++)              for (c = 0; c < 32; c++)
3461                {                {
3462                int x = 0xff;                int x = 0xff;
# Line 3164  for (;; ptr++) Line 3486  for (;; ptr++)
3486                }                }
3487  #endif  #endif
3488              continue;              continue;
             }  
   
           /* We need to deal with \P and \p in both phases. */  
3489    
3490  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3491            if (-c == ESC_p || -c == ESC_P)              case ESC_p:
3492              {              case ESC_P:
3493              BOOL negated;                {
3494              int pdata;                BOOL negated;
3495              int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int pdata;
3496              if (ptype < 0) goto FAILED;                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
3497              class_utf8 = TRUE;                if (ptype < 0) goto FAILED;
3498              *class_utf8data++ = ((-c == ESC_p) != negated)?                class_utf8 = TRUE;
3499                XCL_PROP : XCL_NOTPROP;                *class_utf8data++ = ((-c == ESC_p) != negated)?
3500              *class_utf8data++ = ptype;                  XCL_PROP : XCL_NOTPROP;
3501              *class_utf8data++ = pdata;                *class_utf8data++ = ptype;
3502              class_charcount -= 2;   /* Not a < 256 character */                *class_utf8data++ = pdata;
3503              continue;                class_charcount -= 2;   /* Not a < 256 character */
3504              }                continue;
3505                  }
3506  #endif  #endif
3507            /* Unrecognized escapes are faulted if PCRE is running in its              /* Unrecognized escapes are faulted if PCRE is running in its
3508            strict mode. By default, for compatibility with Perl, they are              strict mode. By default, for compatibility with Perl, they are
3509            treated as literals. */              treated as literals. */
3510    
3511            if ((options & PCRE_EXTRA) != 0)              default:
3512              {              if ((options & PCRE_EXTRA) != 0)
3513              *errorcodeptr = ERR7;                {
3514              goto FAILED;                *errorcodeptr = ERR7;
3515                  goto FAILED;
3516                  }
3517                class_charcount -= 2;  /* Undo the default count from above */
3518                c = *ptr;              /* Get the final character and fall through */
3519                break;
3520              }              }
   
           class_charcount -= 2;  /* Undo the default count from above */  
           c = *ptr;              /* Get the final character and fall through */  
3521            }            }
3522    
3523          /* Fall through if we have a single character (c >= 0). This may be          /* Fall through if we have a single character (c >= 0). This may be
# Line 3265  for (;; ptr++) Line 3587  for (;; ptr++)
3587            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3588            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3589    
3590            /* \b is backspace; \X is literal X; \R is literal R; any other            /* \b is backspace; any other special means the '-' was literal */
           special means the '-' was literal */  
3591    
3592            if (d < 0)            if (d < 0)
3593              {              {
3594              if (d == -ESC_b) d = CHAR_BS;              if (d == -ESC_b) d = CHAR_BS; else
             else if (d == -ESC_X) d = CHAR_X;  
             else if (d == -ESC_R) d = CHAR_R; else  
3595                {                {
3596                ptr = oldptr;                ptr = oldptr;
3597                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto LONE_SINGLE_CHARACTER;  /* A few lines below */
# Line 3438  for (;; ptr++) Line 3757  for (;; ptr++)
3757          }          }
3758        }        }
3759    
3760      /* Loop until ']' reached. This "while" is the end of the "do" above. */      /* Loop until ']' reached. This "while" is the end of the "do" far above.
3761        If we are at the end of an internal nested string, revert to the outer
3762        string. */
3763    
3764        while (((c = *(++ptr)) != 0 ||
3765               (nestptr != NULL &&
3766                 (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&
3767               (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
3768    
3769      while ((c = *(++ptr)) != 0 && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));      /* Check for missing terminating ']' */
3770    
3771      if (c == 0)                          /* Missing terminating ']' */      if (c == 0)
3772        {        {
3773        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
3774        goto FAILED;        goto FAILED;
3775        }        }
3776    
   
 /* This code has been disabled because it would mean that \s counts as  
 an explicit \r or \n reference, and that's not really what is wanted. Now  
 we set the flag only if there is a literal "\r" or "\n" in the class. */  
   
 #if 0  
     /* Remember whether \r or \n are in this class */  
   
     if (negate_class)  
       {  
       if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;  
       }  
     else  
       {  
       if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;  
       }  
 #endif  
   
   
3777      /* 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
3778      less than 256. As long as there were no characters >= 128 and there was no      less than 256. As long as there were no characters >= 128 and there was no
3779      use of \p or \P, in other words, no use of any XCLASS features, we can      use of \p or \P, in other words, no use of any XCLASS features, we can
# Line 3483  we set the flag only if there is a liter Line 3790  we set the flag only if there is a liter
3790      can cause firstbyte to be set. Otherwise, there can be no first char if      can cause firstbyte to be set. Otherwise, there can be no first char if
3791      this item is first, whatever repeat count may follow. In the case of      this item is first, whatever repeat count may follow. In the case of
3792      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3793    
3794  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3795      if (class_charcount == 1 && !class_utf8 &&      if (class_charcount == 1 && !class_utf8 &&
3796        (!utf8 || !negate_class || class_lastchar < 128))        (!utf8 || !negate_class || class_lastchar < 128))
# Line 3530  we set the flag only if there is a liter Line 3837  we set the flag only if there is a liter
3837    
3838      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3839      extended class, with its own opcode, unless there was a negated special      extended class, with its own opcode, unless there was a negated special
3840      such as \S in the class, because in that case all characters > 255 are in      such as \S in the class, and PCRE_UCP is not set, because in that case all
3841      the class, so any that were explicitly given as well can be ignored. If      characters > 255 are in the class, so any that were explicitly given as
3842      (when there are explicit characters > 255 that must be listed) there are no      well can be ignored. If (when there are explicit characters > 255 that must
3843      characters < 256, we can omit the bitmap in the actual compiled code. */      be listed) there are no characters < 256, we can omit the bitmap in the
3844        actual compiled code. */
3845    
3846  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3847      if (class_utf8 && !should_flip_negation)      if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))
3848        {        {
3849        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3850        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3562  we set the flag only if there is a liter Line 3870  we set the flag only if there is a liter
3870        }        }
3871  #endif  #endif
3872    
3873      /* If there are no characters > 255, set the opcode to OP_CLASS or      /* If there are no characters > 255, or they are all to be included or
3874      OP_NCLASS, depending on whether the whole class was negated and whether      excluded, set the opcode to OP_CLASS or OP_NCLASS, depending on whether the
3875      there were negative specials such as \S in the class. Then copy the 32-byte      whole class was negated and whether there were negative specials such as \S
3876      map into the code vector, negating it if necessary. */      (non-UCP) in the class. Then copy the 32-byte map into the code vector,
3877        negating it if necessary. */
3878    
3879      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3880      if (negate_class)      if (negate_class)
# Line 3757  we set the flag only if there is a liter Line 4066  we set the flag only if there is a liter
4066    
4067        if (repeat_max == 0) goto END_REPEAT;        if (repeat_max == 0) goto END_REPEAT;
4068    
4069          /*--------------------------------------------------------------------*/
4070          /* This code is obsolete from release 8.00; the restriction was finally
4071          removed: */
4072    
4073        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
4074        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
4075    
4076        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;        /* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */
4077          /*--------------------------------------------------------------------*/
4078    
4079        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
4080    
# Line 3907  we set the flag only if there is a liter Line 4221  we set the flag only if there is a liter
4221          goto END_REPEAT;          goto END_REPEAT;
4222          }          }
4223    
4224          /*--------------------------------------------------------------------*/
4225          /* This code is obsolete from release 8.00; the restriction was finally
4226          removed: */
4227    
4228        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
4229        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
4230    
4231        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;        /* if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL; */
4232          /*--------------------------------------------------------------------*/
4233    
4234        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
4235          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 4045  we set the flag only if there is a liter Line 4364  we set the flag only if there is a liter
4364            {            {
4365            /* 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
4366            just adjust the length as if we had. Do some paranoid checks for            just adjust the length as if we had. Do some paranoid checks for
4367            potential integer overflow. */            potential integer overflow. The INT64_OR_DOUBLE type is a 64-bit
4368              integer type when available, otherwise double. */
4369    
4370            if (lengthptr != NULL)            if (lengthptr != NULL)
4371              {              {
4372              int delta = (repeat_min - 1)*length_prevgroup;              int delta = (repeat_min - 1)*length_prevgroup;
4373              if ((double)(repeat_min - 1)*(double)length_prevgroup >              if ((INT64_OR_DOUBLE)(repeat_min - 1)*
4374                                                              (double)INT_MAX ||                    (INT64_OR_DOUBLE)length_prevgroup >
4375                        (INT64_OR_DOUBLE)INT_MAX ||
4376                  OFLOW_MAX - *lengthptr < delta)                  OFLOW_MAX - *lengthptr < delta)
4377                {                {
4378                *errorcodeptr = ERR20;                *errorcodeptr = ERR20;
# Line 4097  we set the flag only if there is a liter Line 4418  we set the flag only if there is a liter
4418          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
4419          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
4420          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
4421          paranoid checks to avoid integer overflow. */          paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is
4422            a 64-bit integer type when available, otherwise double. */
4423    
4424          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
4425            {            {
4426            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
4427                        2 - 2*LINK_SIZE;   /* Last one doesn't nest */                        2 - 2*LINK_SIZE;   /* Last one doesn't nest */
4428            if ((double)repeat_max *            if ((INT64_OR_DOUBLE)repeat_max *
4429                  (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)                  (INT64_OR_DOUBLE)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
4430                    > (double)INT_MAX ||                    > (INT64_OR_DOUBLE)INT_MAX ||
4431                OFLOW_MAX - *lengthptr < delta)                OFLOW_MAX - *lengthptr < delta)
4432              {              {
4433              *errorcodeptr = ERR20;              *errorcodeptr = ERR20;
# Line 4182  we set the flag only if there is a liter Line 4504  we set the flag only if there is a liter
4504            uschar *scode = bracode;            uschar *scode = bracode;
4505            do            do
4506              {              {
4507              if (could_be_empty_branch(scode, ketcode, utf8))              if (could_be_empty_branch(scode, ketcode, utf8, cd))
4508                {                {
4509                *bracode += OP_SBRA - OP_BRA;                *bracode += OP_SBRA - OP_BRA;
4510                break;                break;
# Line 4225  we set the flag only if there is a liter Line 4547  we set the flag only if there is a liter
4547      if (possessive_quantifier)      if (possessive_quantifier)
4548        {        {
4549        int len;        int len;
4550        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||  
4551            *tempcode == OP_NOTEXACT)        if (*tempcode == OP_TYPEEXACT)
4552          tempcode += _pcre_OP_lengths[*tempcode] +          tempcode += _pcre_OP_lengths[*tempcode] +
4553            ((*tempcode == OP_TYPEEXACT &&            ((tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP)? 2 : 0);
4554               (tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0);  
4555          else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)
4556            {
4557            tempcode += _pcre_OP_lengths[*tempcode];
4558    #ifdef SUPPORT_UTF8
4559            if (utf8 && tempcode[-1] >= 0xc0)
4560              tempcode += _pcre_utf8_table4[tempcode[-1] & 0x3f];
4561    #endif
4562            }
4563    
4564        len = code - tempcode;        len = code - tempcode;
4565        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4566          {          {
# Line 4248  we set the flag only if there is a liter Line 4579  we set the flag only if there is a liter
4579          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;
4580          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;
4581    
4582            /* Because we are moving code along, we must ensure that any
4583            pending recursive references are updated. */
4584    
4585          default:          default:
4586            *code = OP_END;
4587            adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);
4588          memmove(tempcode + 1+LINK_SIZE, tempcode, len);          memmove(tempcode + 1+LINK_SIZE, tempcode, len);
4589          code += 1 + LINK_SIZE;          code += 1 + LINK_SIZE;
4590          len += 1 + LINK_SIZE;          len += 1 + LINK_SIZE;
# Line 4284  we set the flag only if there is a liter Line 4620  we set the flag only if there is a liter
4620    
4621      /* First deal with various "verbs" that can be introduced by '*'. */      /* First deal with various "verbs" that can be introduced by '*'. */
4622    
4623      if (*(++ptr) == CHAR_ASTERISK && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == CHAR_ASTERISK &&
4624             ((cd->ctypes[ptr[1]] & ctype_letter) != 0 || ptr[1] == ':'))
4625        {        {
4626        int i, namelen;        int i, namelen;
4627          int arglen = 0;
4628        const char *vn = verbnames;        const char *vn = verbnames;
4629        const uschar *name = ++ptr;        const uschar *name = ptr + 1;
4630          const uschar *arg = NULL;
4631        previous = NULL;        previous = NULL;
4632        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
4633          namelen = ptr - name;
4634    
4635        if (*ptr == CHAR_COLON)        if (*ptr == CHAR_COLON)
4636          {          {
4637          *errorcodeptr = ERR59;   /* Not supported */          arg = ++ptr;
4638          goto FAILED;          while ((cd->ctypes[*ptr] & (ctype_letter|ctype_digit)) != 0
4639              || *ptr == '_') ptr++;
4640            arglen = ptr - arg;
4641          }          }
4642    
4643        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
4644          {          {
4645          *errorcodeptr = ERR60;          *errorcodeptr = ERR60;
4646          goto FAILED;          goto FAILED;
4647          }          }
4648        namelen = ptr - name;  
4649          /* Scan the table of verb names */
4650    
4651        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4652          {          {
4653          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
4654              strncmp((char *)name, vn, namelen) == 0)              strncmp((char *)name, vn, namelen) == 0)
4655            {            {
4656            *code = verbs[i].op;            /* Check for open captures before ACCEPT */
4657            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;  
4658            break;            if (verbs[i].op == OP_ACCEPT)
4659                {
4660                open_capitem *oc;
4661                cd->had_accept = TRUE;
4662                for (oc = cd->open_caps; oc != NULL; oc = oc->next)
4663                  {
4664                  *code++ = OP_CLOSE;
4665                  PUT2INC(code, 0, oc->number);
4666                  }
4667                }
4668    
4669              /* Handle the cases with/without an argument */
4670    
4671              if (arglen == 0)
4672                {
4673                if (verbs[i].op < 0)   /* Argument is mandatory */
4674                  {
4675                  *errorcodeptr = ERR66;
4676                  goto FAILED;
4677                  }
4678                *code++ = verbs[i].op;
4679                }
4680    
4681              else
4682                {
4683                if (verbs[i].op_arg < 0)   /* Argument is forbidden */
4684                  {
4685                  *errorcodeptr = ERR59;
4686                  goto FAILED;
4687                  }
4688                *code++ = verbs[i].op_arg;
4689                *code++ = arglen;
4690                memcpy(code, arg, arglen);
4691                code += arglen;
4692                *code++ = 0;
4693                }
4694    
4695              break;  /* Found verb, exit loop */
4696            }            }
4697    
4698          vn += verbs[i].len + 1;          vn += verbs[i].len + 1;
4699          }          }
4700        if (i < verbcount) continue;  
4701        *errorcodeptr = ERR60;        if (i < verbcount) continue;    /* Successfully handled a verb */
4702          *errorcodeptr = ERR60;          /* Verb not recognized */
4703        goto FAILED;        goto FAILED;
4704        }        }
4705    
# Line 4470  we set the flag only if there is a liter Line 4855  we set the flag only if there is a liter
4855            }            }
4856    
4857          /* Otherwise (did not start with "+" or "-"), start by looking for the          /* Otherwise (did not start with "+" or "-"), start by looking for the
4858          name. */          name. If we find a name, add one to the opcode to change OP_CREF or
4859            OP_RREF into OP_NCREF or OP_NRREF. These behave exactly the same,
4860            except they record that the reference was originally to a name. The
4861            information is used to check duplicate names. */
4862    
4863          slot = cd->name_table;          slot = cd->name_table;
4864          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
# Line 4485  we set the flag only if there is a liter Line 4873  we set the flag only if there is a liter
4873            {            {
4874            recno = GET2(slot, 0);            recno = GET2(slot, 0);
4875            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4876              code[1+LINK_SIZE]++;
4877            }            }
4878    
4879          /* Search the pattern for a forward reference */          /* Search the pattern for a forward reference */
4880    
4881          else if ((i = find_parens(ptr, cd, name, namelen,          else if ((i = find_parens(cd, name, namelen,
4882                          (options & PCRE_EXTENDED) != 0)) > 0)                          (options & PCRE_EXTENDED) != 0)) > 0)
4883            {            {
4884            PUT2(code, 2+LINK_SIZE, i);            PUT2(code, 2+LINK_SIZE, i);
4885              code[1+LINK_SIZE]++;
4886            }            }
4887    
4888          /* If terminator == 0 it means that the name followed directly after          /* If terminator == 0 it means that the name followed directly after
# Line 4685  we set the flag only if there is a liter Line 5075  we set the flag only if there is a liter
5075                }                }
5076              }              }
5077    
5078            /* In the real compile, create the entry in the table */            /* In the real compile, create the entry in the table, maintaining
5079              alphabetical order. Duplicate names for different numbers are
5080              permitted only if PCRE_DUPNAMES is set. Duplicate names for the same
5081              number are always OK. (An existing number can be re-used if (?|
5082              appears in the pattern.) In either event, a duplicate name results in
5083              a duplicate entry in the table, even if the number is the same. This
5084              is because the number of names, and hence the table size, is computed
5085              in the pre-compile, and it affects various numbers and pointers which
5086              would all have to be modified, and the compiled code moved down, if
5087              duplicates with the same number were omitted from the table. This
5088              doesn't seem worth the hassle. However, *different* names for the
5089              same number are not permitted. */
5090    
5091            else            else
5092              {              {
5093                BOOL dupname = FALSE;
5094              slot = cd->name_table;              slot = cd->name_table;
5095    
5096              for (i = 0; i < cd->names_found; i++)              for (i = 0; i < cd->names_found; i++)
5097                {                {
5098                int crc = memcmp(name, slot+2, namelen);                int crc = memcmp(name, slot+2, namelen);
# Line 4697  we set the flag only if there is a liter Line 5100  we set the flag only if there is a liter
5100                  {                  {
5101                  if (slot[2+namelen] == 0)                  if (slot[2+namelen] == 0)
5102                    {                    {
5103                    if ((options & PCRE_DUPNAMES) == 0)                    if (GET2(slot, 0) != cd->bracount + 1 &&
5104                          (options & PCRE_DUPNAMES) == 0)
5105                      {                      {
5106                      *errorcodeptr = ERR43;                      *errorcodeptr = ERR43;
5107                      goto FAILED;                      goto FAILED;
5108                      }                      }
5109                      else dupname = TRUE;
5110                    }                    }
5111                  else crc = -1;      /* Current name is substring */                  else crc = -1;      /* Current name is a substring */
5112                  }                  }
5113    
5114                  /* Make space in the table and break the loop for an earlier
5115                  name. For a duplicate or later name, carry on. We do this for
5116                  duplicates so that in the simple case (when ?(| is not used) they
5117                  are in order of their numbers. */
5118    
5119                if (crc < 0)                if (crc < 0)
5120                  {                  {
5121                  memmove(slot + cd->name_entry_size, slot,                  memmove(slot + cd->name_entry_size, slot,
5122                    (cd->names_found - i) * cd->name_entry_size);                    (cd->names_found - i) * cd->name_entry_size);
5123                  break;                  break;
5124                  }                  }
5125    
5126                  /* Continue the loop for a later or duplicate name */
5127    
5128                slot += cd->name_entry_size;                slot += cd->name_entry_size;
5129                }                }
5130    
5131                /* For non-duplicate names, check for a duplicate number before
5132                adding the new name. */
5133    
5134                if (!dupname)
5135                  {
5136                  uschar *cslot = cd->name_table;
5137                  for (i = 0; i < cd->names_found; i++)
5138                    {
5139                    if (cslot != slot)
5140                      {
5141                      if (GET2(cslot, 0) == cd->bracount + 1)
5142                        {
5143                        *errorcodeptr = ERR65;
5144                        goto FAILED;
5145                        }
5146                      }
5147                    else i--;
5148                    cslot += cd->name_entry_size;
5149                    }
5150                  }
5151    
5152              PUT2(slot, 0, cd->bracount + 1);              PUT2(slot, 0, cd->bracount + 1);
5153              memcpy(slot + 2, name, namelen);              memcpy(slot + 2, name, namelen);
5154              slot[2+namelen] = 0;              slot[2+namelen] = 0;
5155              }              }
5156            }            }
5157    
5158          /* In both cases, count the number of names we've encountered. */          /* In both pre-compile and compile, count the number of names we've
5159            encountered. */
5160    
         ptr++;                    /* Move past > or ' */  
5161          cd->names_found++;          cd->names_found++;
5162            ptr++;                    /* Move past > or ' */
5163          goto NUMBERED_GROUP;          goto NUMBERED_GROUP;
5164    
5165    
# Line 4788  we set the flag only if there is a liter Line 5224  we set the flag only if there is a liter
5224              recno = GET2(slot, 0);              recno = GET2(slot, 0);
5225              }              }
5226            else if ((recno =                /* Forward back reference */            else if ((recno =                /* Forward back reference */
5227                      find_parens(ptr, cd, name, namelen,                      find_parens(cd, name, namelen,
5228                        (options & PCRE_EXTENDED) != 0)) <= 0)                        (options & PCRE_EXTENDED) != 0)) <= 0)
5229              {              {
5230              *errorcodeptr = ERR15;              *errorcodeptr = ERR15;
# Line 4892  we set the flag only if there is a liter Line 5328  we set the flag only if there is a liter
5328            if (lengthptr == NULL)            if (lengthptr == NULL)
5329              {              {
5330              *code = OP_END;              *code = OP_END;
5331              if (recno != 0) called = find_bracket(cd->start_code, utf8, recno);              if (recno != 0)
5332                  called = _pcre_find_bracket(cd->start_code, utf8, recno);
5333    
5334              /* Forward reference */              /* Forward reference */
5335    
5336              if (called == NULL)              if (called == NULL)
5337                {                {
5338                if (find_parens(ptr, cd, NULL, recno,                if (find_parens(cd, NULL, recno,
5339                      (options & PCRE_EXTENDED) != 0) < 0)                      (options & PCRE_EXTENDED) != 0) < 0)
5340                  {                  {
5341                  *errorcodeptr = ERR15;                  *errorcodeptr = ERR15;
5342                  goto FAILED;                  goto FAILED;
5343                  }                  }
5344    
5345                  /* Fudge the value of "called" so that when it is inserted as an
5346                  offset below, what it actually inserted is the reference number
5347                  of the group. */
5348    
5349                called = cd->start_code + recno;                called = cd->start_code + recno;
5350                PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code);                PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code);
5351                }                }
# Line 4913  we set the flag only if there is a liter Line 5355  we set the flag only if there is a liter
5355              recursion that could loop for ever, and diagnose that case. */              recursion that could loop for ever, and diagnose that case. */
5356    
5357              else if (GET(called, 1) == 0 &&              else if (GET(called, 1) == 0 &&
5358                       could_be_empty(called, code, bcptr, utf8))                       could_be_empty(called, code, bcptr, utf8, cd))
5359                {                {
5360                *errorcodeptr = ERR40;                *errorcodeptr = ERR40;
5361                goto FAILED;                goto FAILED;
# Line 5008  we set the flag only if there is a liter Line 5450  we set the flag only if there is a liter
5450              {              {
5451              cd->external_options = newoptions;              cd->external_options = newoptions;
5452              }              }
5453           else            else
5454              {              {
5455              if ((options & PCRE_IMS) != (newoptions & PCRE_IMS))              if ((options & PCRE_IMS) != (newoptions & PCRE_IMS))
5456                {                {
# Line 5040  we set the flag only if there is a liter Line 5482  we set the flag only if there is a liter
5482          }     /* End of switch for character following (? */          }     /* End of switch for character following (? */
5483        }       /* End of (? handling */        }       /* End of (? handling */
5484    
5485      /* Opening parenthesis not followed by '?'. If PCRE_NO_AUTO_CAPTURE is set,      /* Opening parenthesis not followed by '*' or '?'. If PCRE_NO_AUTO_CAPTURE
5486      all unadorned brackets become non-capturing and behave like (?:...)      is set, all unadorned brackets become non-capturing and behave like (?:...)
5487      brackets. */      brackets. */
5488    
5489      else if ((options & PCRE_NO_AUTO_CAPTURE) != 0)      else if ((options & PCRE_NO_AUTO_CAPTURE) != 0)
# Line 5233  we set the flag only if there is a liter Line 5675  we set the flag only if there is a liter
5675    
5676      /* ===================================================================*/      /* ===================================================================*/
5677      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
5678      are arranged to be the negation of the corresponding OP_values. For the      are arranged to be the negation of the corresponding OP_values in the
5679      back references, the values are ESC_REF plus the reference number. Only      default case when PCRE_UCP is not set. For the back references, the values
5680      back references and those types that consume a character may be repeated.      are ESC_REF plus the reference number. Only back references and those types
5681      We can test for values between ESC_b and ESC_Z for the latter; this may      that consume a character may be repeated. We can test for values between
5682      have to change if any new ones are ever created. */      ESC_b and ESC_Z for the latter; this may have to change if any new ones are
5683        ever created. */
5684    
5685      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
5686      tempptr = ptr;      tempptr = ptr;
# Line 5345  we set the flag only if there is a liter Line 5788  we set the flag only if there is a liter
5788    
5789        if (-c >= ESC_REF)        if (-c >= ESC_REF)
5790          {          {
5791            open_capitem *oc;
5792          recno = -c - ESC_REF;          recno = -c - ESC_REF;
5793    
5794          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
# Line 5354  we set the flag only if there is a liter Line 5798  we set the flag only if there is a liter
5798          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
5799          cd->backref_map |= (recno < 32)? (1 << recno) : 1;          cd->backref_map |= (recno < 32)? (1 << recno) : 1;
5800          if (recno > cd->top_backref) cd->top_backref = recno;          if (recno > cd->top_backref) cd->top_backref = recno;
5801    
5802            /* Check to see if this back reference is recursive, that it, it
5803            is inside the group that it references. A flag is set so that the
5804            group can be made atomic. */
5805    
5806            for (oc = cd->open_caps; oc != NULL; oc = oc->next)
5807              {
5808              if (oc->number == recno)
5809                {
5810                oc->flag = TRUE;
5811                break;
5812                }
5813              }
5814          }          }
5815    
5816        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
# Line 5383  we set the flag only if there is a liter Line 5840  we set the flag only if there is a liter
5840  #endif  #endif
5841    
5842        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
5843        can obtain the OP value by negating the escape value. */        can obtain the OP value by negating the escape value in the default
5844          situation when PCRE_UCP is not set. When it *is* set, we substitute
5845          Unicode property tests. */
5846    
5847        else        else
5848          {          {
5849          previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;  #ifdef SUPPORT_UCP
5850          *code++ = -c;          if (-c >= ESC_DU && -c <= ESC_wu)
5851              {
5852              nestptr = ptr + 1;                   /* Where to resume */
5853              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
5854              }
5855            else
5856    #endif
5857              {
5858              previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;
5859              *code++ = -c;
5860              }
5861          }          }
5862        continue;        continue;
5863        }        }
# Line 5536  uschar *code = *codeptr; Line 6005  uschar *code = *codeptr;
6005  uschar *last_branch = code;  uschar *last_branch = code;
6006  uschar *start_bracket = code;  uschar *start_bracket = code;
6007  uschar *reverse_count = NULL;  uschar *reverse_count = NULL;
6008    open_capitem capitem;
6009    int capnumber = 0;
6010  int firstbyte, reqbyte;  int firstbyte, reqbyte;
6011  int branchfirstbyte, branchreqbyte;  int branchfirstbyte, branchreqbyte;
6012  int length;  int length;
6013  int orig_bracount;  int orig_bracount;
6014  int max_bracount;  int max_bracount;
6015    int old_external_options = cd->external_options;
6016  branch_chain bc;  branch_chain bc;
6017    
6018  bc.outer = bcptr;  bc.outer = bcptr;
6019  bc.current = code;  bc.current_branch = code;
6020    
6021  firstbyte = reqbyte = REQ_UNSET;  firstbyte = reqbyte = REQ_UNSET;
6022    
# Line 5562  the code that abstracts option settings Line 6034  the code that abstracts option settings
6034  them global. It tests the value of length for (2 + 2*LINK_SIZE) in the  them global. It tests the value of length for (2 + 2*LINK_SIZE) in the
6035  pre-compile phase to find out whether anything has yet been compiled or not. */  pre-compile phase to find out whether anything has yet been compiled or not. */
6036    
6037    /* If this is a capturing subpattern, add to the chain of open capturing items
6038    so that we can detect them if (*ACCEPT) is encountered. This is also used to
6039    detect groups that contain recursive back references to themselves. */
6040    
6041    if (*code == OP_CBRA)
6042      {
6043      capnumber = GET2(code, 1 + LINK_SIZE);
6044      capitem.number = capnumber;
6045      capitem.next = cd->open_caps;
6046      capitem.flag = FALSE;
6047      cd->open_caps = &capitem;
6048      }
6049    
6050  /* Offset is set zero to mark that this bracket is still open */  /* Offset is set zero to mark that this bracket is still open */
6051    
6052  PUT(code, 1, 0);  PUT(code, 1, 0);
# Line 5606  for (;;) Line 6091  for (;;)
6091      return FALSE;      return FALSE;
6092      }      }
6093    
6094      /* If the external options have changed during this branch, it means that we
6095      are at the top level, and a leading option setting has been encountered. We
6096      need to re-set the original option values to take account of this so that,
6097      during the pre-compile phase, we know to allow for a re-set at the start of
6098      subsequent branches. */
6099    
6100      if (old_external_options != cd->external_options)
6101        oldims = cd->external_options & PCRE_IMS;
6102    
6103    /* Keep the highest bracket count in case (?| was used and some branch    /* Keep the highest bracket count in case (?| was used and some branch
6104    has fewer than the rest. */    has fewer than the rest. */
6105    
# Line 5656  for (;;) Line 6150  for (;;)
6150    
6151      /* If lookbehind, check that this branch matches a fixed-length string, and      /* If lookbehind, check that this branch matches a fixed-length string, and
6152      put the length into the OP_REVERSE item. Temporarily mark the end of the      put the length into the OP_REVERSE item. Temporarily mark the end of the
6153      branch with OP_END. */      branch with OP_END. If the branch contains OP_RECURSE, the result is -3
6154        because there may be forward references that we can't check here. Set a
6155        flag to cause another lookbehind check at the end. Why not do it all at the
6156        end? Because common, erroneous checks are picked up here and the offset of
6157        the problem can be shown. */
6158    
6159      if (lookbehind)      if (lookbehind)
6160        {        {
6161        int fixed_length;        int fixed_length;
6162        *code = OP_END;        *code = OP_END;
6163        fixed_length = find_fixedlength(last_branch, options);        fixed_length = find_fixedlength(last_branch, options, FALSE, cd);
6164        DPRINTF(("fixed length = %d\n", fixed_length));        DPRINTF(("fixed length = %d\n", fixed_length));
6165        if (fixed_length < 0)        if (fixed_length == -3)
6166            {
6167            cd->check_lookbehind = TRUE;
6168            }
6169          else if (fixed_length < 0)
6170          {          {
6171          *errorcodeptr = (fixed_length == -2)? ERR36 : ERR25;          *errorcodeptr = (fixed_length == -2)? ERR36 : ERR25;
6172          *ptrptr = ptr;          *ptrptr = ptr;
6173          return FALSE;          return FALSE;
6174          }          }
6175        PUT(reverse_count, 0, fixed_length);        else { PUT(reverse_count, 0, fixed_length); }
6176        }        }
6177      }      }
6178    
# Line 5704  for (;;) Line 6206  for (;;)
6206      PUT(code, 1, code - start_bracket);      PUT(code, 1, code - start_bracket);
6207      code += 1 + LINK_SIZE;      code += 1 + LINK_SIZE;
6208    
6209      /* Resetting option if needed */      /* If it was a capturing subpattern, check to see if it contained any
6210        recursive back references. If so, we must wrap it in atomic brackets.
6211        In any event, remove the block from the chain. */
6212    
6213        if (capnumber > 0)
6214          {
6215          if (cd->open_caps->flag)
6216            {
6217            memmove(start_bracket + 1 + LINK_SIZE, start_bracket,
6218              code - start_bracket);
6219            *start_bracket = OP_ONCE;
6220            code += 1 + LINK_SIZE;
6221            PUT(start_bracket, 1, code - start_bracket);
6222            *code = OP_KET;
6223            PUT(code, 1, code - start_bracket);
6224            code += 1 + LINK_SIZE;
6225            length += 2 + 2*LINK_SIZE;
6226            }
6227          cd->open_caps = cd->open_caps->next;
6228          }
6229    
6230        /* Reset options if needed. */
6231    
6232      if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS)      if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS)
6233        {        {
# Line 5753  for (;;) Line 6276  for (;;)
6276      {      {
6277      *code = OP_ALT;      *code = OP_ALT;
6278      PUT(code, 1, code - last_branch);      PUT(code, 1, code - last_branch);
6279      bc.current = last_branch = code;      bc.current_branch = last_branch = code;
6280      code += 1 + LINK_SIZE;      code += 1 + LINK_SIZE;
6281      }      }
6282    
# Line 5900  do { Line 6423  do {
6423       switch (*scode)       switch (*scode)
6424         {         {
6425         case OP_CREF:         case OP_CREF:
6426           case OP_NCREF:
6427         case OP_RREF:         case OP_RREF:
6428           case OP_NRREF:
6429         case OP_DEF:         case OP_DEF:
6430         return FALSE;         return FALSE;
6431    
# Line 6069  int length = 1;  /* For final END opcode Line 6594  int length = 1;  /* For final END opcode
6594  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
6595  int errorcode = 0;  int errorcode = 0;
6596  int skipatstart = 0;  int skipatstart = 0;
 #ifdef SUPPORT_UTF8  
6597  BOOL utf8;  BOOL utf8;
 #endif  
6598  size_t size;  size_t size;
6599  uschar *code;  uschar *code;
6600  const uschar *codestart;  const uschar *codestart;
# Line 6114  if (erroroffset == NULL) Line 6637  if (erroroffset == NULL)
6637    
6638  *erroroffset = 0;  *erroroffset = 0;
6639    
 /* Can't support UTF8 unless PCRE has been compiled to include the code. */  
   
 #ifdef SUPPORT_UTF8  
 utf8 = (options & PCRE_UTF8) != 0;  
 if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 &&  
      (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)  
   {  
   errorcode = ERR44;  
   goto PCRE_EARLY_ERROR_RETURN2;  
   }  
 #else  
 if ((options & PCRE_UTF8) != 0)  
   {  
   errorcode = ERR32;  
   goto PCRE_EARLY_ERROR_RETURN;  
   }  
 #endif  
   
 if ((options & ~PUBLIC_COMPILE_OPTIONS) != 0)  
   {  
   errorcode = ERR17;  
   goto PCRE_EARLY_ERROR_RETURN;  
   }  
   
6640  /* Set up pointers to the individual character tables */  /* Set up pointers to the individual character tables */
6641    
6642  if (tables == NULL) tables = _pcre_default_tables;  if (tables == NULL) tables = _pcre_default_tables;
# Line 6146  cd->fcc = tables + fcc_offset; Line 6645  cd->fcc = tables + fcc_offset;
6645  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
6646  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
6647    
6648    /* Check that all undefined public option bits are zero */
6649    
6650    if ((options & ~PUBLIC_COMPILE_OPTIONS) != 0)
6651      {
6652      errorcode = ERR17;
6653      goto PCRE_EARLY_ERROR_RETURN;
6654      }
6655    
6656  /* Check for global one-time settings at the start of the pattern, and remember  /* Check for global one-time settings at the start of the pattern, and remember
6657  the offset for later. */  the offset for later. */
6658    
# Line 6155  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6662  while (ptr[skipatstart] == CHAR_LEFT_PAR
6662    int newnl = 0;    int newnl = 0;
6663    int newbsr = 0;    int newbsr = 0;
6664    
6665      if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)
6666        { skipatstart += 7; options |= PCRE_UTF8; continue; }
6667      else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0)
6668        { skipatstart += 6; options |= PCRE_UCP; continue; }
6669    
6670    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)
6671      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
6672    else if (strncmp((char *)(ptr+skipatstart+2), STRING_LF_RIGHTPAR, 3)  == 0)    else if (strncmp((char *)(ptr+skipatstart+2), STRING_LF_RIGHTPAR, 3)  == 0)
# Line 6177  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6689  while (ptr[skipatstart] == CHAR_LEFT_PAR
6689      options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;      options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
6690    else break;    else break;
6691    }    }
6692    
6693    utf8 = (options & PCRE_UTF8) != 0;
6694    
6695    /* Can't support UTF8 unless PCRE has been compiled to include the code. */
6696    
6697    #ifdef SUPPORT_UTF8
6698    if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 &&
6699         (*erroroffset = _pcre_valid_utf8((USPTR)pattern, -1)) >= 0)
6700      {
6701      errorcode = ERR44;
6702      goto PCRE_EARLY_ERROR_RETURN2;
6703      }
6704    #else
6705    if (utf8)
6706      {
6707      errorcode = ERR32;
6708      goto PCRE_EARLY_ERROR_RETURN;
6709      }
6710    #endif
6711    
6712    /* Can't support UCP unless PCRE has been compiled to include the code. */
6713    
6714    #ifndef SUPPORT_UCP
6715    if ((options & PCRE_UCP) != 0)
6716      {
6717      errorcode = ERR67;
6718      goto PCRE_EARLY_ERROR_RETURN;
6719      }
6720    #endif
6721    
6722  /* Check validity of \R options. */  /* Check validity of \R options. */
6723    
# Line 6260  cd->end_pattern = (const uschar *)(patte Line 6801  cd->end_pattern = (const uschar *)(patte
6801  cd->req_varyopt = 0;  cd->req_varyopt = 0;
6802  cd->external_options = options;  cd->external_options = options;
6803  cd->external_flags = 0;  cd->external_flags = 0;
6804    cd->open_caps = NULL;
6805    
6806  /* 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
6807  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 6334  cd->start_code = codestart; Line 6876  cd->start_code = codestart;
6876  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6877  cd->req_varyopt = 0;  cd->req_varyopt = 0;
6878  cd->had_accept = FALSE;  cd->had_accept = FALSE;
6879    cd->check_lookbehind = FALSE;
6880    cd->open_caps = NULL;
6881    
6882  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6883  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 6359  if debugging, leave the test till after Line 6903  if debugging, leave the test till after
6903    
6904  *code++ = OP_END;  *code++ = OP_END;
6905    
6906  #ifndef DEBUG  #ifndef PCRE_DEBUG
6907  if (code - codestart > length) errorcode = ERR23;  if (code - codestart > length) errorcode = ERR23;
6908  #endif  #endif
6909    
# Line 6372  while (errorcode == 0 && cd->hwm > cwork Line 6916  while (errorcode == 0 && cd->hwm > cwork
6916    cd->hwm -= LINK_SIZE;    cd->hwm -= LINK_SIZE;
6917    offset = GET(cd->hwm, 0);    offset = GET(cd->hwm, 0);
6918    recno = GET(codestart, offset);    recno = GET(codestart, offset);
6919    groupptr = find_bracket(codestart, (re->options & PCRE_UTF8) != 0, recno);    groupptr = _pcre_find_bracket(codestart, utf8, recno);
6920    if (groupptr == NULL) errorcode = ERR53;    if (groupptr == NULL) errorcode = ERR53;
6921      else PUT(((uschar *)codestart), offset, groupptr - codestart);      else PUT(((uschar *)codestart), offset, groupptr - codestart);
6922    }    }
# Line 6382  subpattern. */ Line 6926  subpattern. */
6926    
6927  if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15;  if (errorcode == 0 && re->top_backref > re->top_bracket) errorcode = ERR15;
6928    
6929    /* If there were any lookbehind assertions that contained OP_RECURSE
6930    (recursions or subroutine calls), a flag is set for them to be checked here,
6931    because they may contain forward references. Actual recursions can't be fixed
6932    length, but subroutine calls can. It is done like this so that those without
6933    OP_RECURSE that are not fixed length get a diagnosic with a useful offset. The
6934    exceptional ones forgo this. We scan the pattern to check that they are fixed
6935    length, and set their lengths. */
6936    
6937    if (cd->check_lookbehind)
6938      {
6939      uschar *cc = (uschar *)codestart;
6940    
6941      /* Loop, searching for OP_REVERSE items, and process those that do not have
6942      their length set. (Actually, it will also re-process any that have a length
6943      of zero, but that is a pathological case, and it does no harm.) When we find
6944      one, we temporarily terminate the branch it is in while we scan it. */
6945    
6946      for (cc = (uschar *)_pcre_find_bracket(codestart, utf8, -1);
6947           cc != NULL;
6948           cc = (uschar *)_pcre_find_bracket(cc, utf8, -1))
6949        {
6950        if (GET(cc, 1) == 0)
6951          {
6952          int fixed_length;
6953          uschar *be = cc - 1 - LINK_SIZE + GET(cc, -LINK_SIZE);
6954          int end_op = *be;
6955          *be = OP_END;
6956          fixed_length = find_fixedlength(cc, re->options, TRUE, cd);
6957          *be = end_op;
6958          DPRINTF(("fixed length = %d\n", fixed_length));
6959          if (fixed_length < 0)
6960            {
6961            errorcode = (fixed_length == -2)? ERR36 : ERR25;
6962            break;
6963            }
6964          PUT(cc, 1, fixed_length);
6965          }
6966        cc += 1 + LINK_SIZE;
6967        }
6968      }
6969    
6970  /* Failed to compile, or error while post-processing */  /* Failed to compile, or error while post-processing */
6971    
6972  if (errorcode != 0)  if (errorcode != 0)
# Line 6442  if (reqbyte >= 0 && Line 7027  if (reqbyte >= 0 &&
7027  /* 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
7028  case when building a production library. */  case when building a production library. */
7029    
7030  #ifdef DEBUG  #ifdef PCRE_DEBUG
   
7031  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
7032    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
7033    
# Line 6480  if (code - codestart > length) Line 7064  if (code - codestart > length)
7064    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
7065    return NULL;    return NULL;
7066    }    }
7067  #endif   /* DEBUG */  #endif   /* PCRE_DEBUG */
7068    
7069  return (pcre *)re;  return (pcre *)re;
7070  }  }

Legend:
Removed from v.406  
changed lines
  Added in v.528

  ViewVC Help
Powered by ViewVC 1.1.5