/[pcre]/code/trunk/pcre_compile.c
ViewVC logotype

Diff of /code/trunk/pcre_compile.c

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

revision 505 by ph10, Tue Mar 9 16:50:47 2010 UTC revision 578 by ph10, Tue Nov 23 15:34:55 2010 UTC
# Line 92  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  /* 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. */  overrun before it actually does run off the end of the data block. */
97    
98  #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100)  #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100)
# Line 124  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 171  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 188  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 202  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 256  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 268  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  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  substring, so that the whole string ends with \0\0, which can be detected when
327  counting through. */  counting through. */
328    
329  static const char error_texts[] =  static const char error_texts[] =
# Line 319  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 345  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"
# Line 353  static const char error_texts[] = Line 405  static const char error_texts[] =
405    "digit expected after (?+\0"    "digit expected after (?+\0"
406    "] is an invalid data character in JavaScript compatibility mode\0"    "] is an invalid data character in JavaScript compatibility mode\0"
407    /* 65 */    /* 65 */
408    "different names for subpatterns of the same number are not allowed\0";    "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      "\\c must be followed by an ASCII character\0"
412      ;
413    
414  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
415  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 511  static const char * Line 567  static const char *
567  find_error_text(int n)  find_error_text(int n)
568  {  {
569  const char *s = error_texts;  const char *s = error_texts;
570  for (; n > 0; n--)  for (; n > 0; n--)
571    {    {
572    while (*s++ != 0) {};    while (*s++ != 0) {};
573    if (*s == 0) return "Error text not found (please report)";    if (*s == 0) return "Error text not found (please report)";
574    }    }
575  return s;  return s;
576  }  }
577    
# Line 586  else Line 642  else
642    
643      case CHAR_l:      case CHAR_l:
644      case CHAR_L:      case CHAR_L:
     case CHAR_N:  
645      case CHAR_u:      case CHAR_u:
646      case CHAR_U:      case CHAR_U:
647      *errorcodeptr = ERR37;      *errorcodeptr = ERR37;
# Line 787  else Line 842  else
842      break;      break;
843    
844      /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped.      /* For \c, a following letter is upper-cased; then the 0x40 bit is flipped.
845      This coding is ASCII-specific, but then the whole concept of \cx is      An error is given if the byte following \c is not an ASCII character. This
846        coding is ASCII-specific, but then the whole concept of \cx is
847      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */
848    
849      case CHAR_c:      case CHAR_c:
# Line 797  else Line 853  else
853        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
854        break;        break;
855        }        }
856    #ifndef EBCDIC    /* ASCII/UTF-8 coding */
857  #ifndef EBCDIC  /* ASCII/UTF-8 coding */      if (c > 127)  /* Excludes all non-ASCII in either mode */
858          {
859          *errorcodeptr = ERR68;
860          break;
861          }
862      if (c >= CHAR_a && c <= CHAR_z) c -= 32;      if (c >= CHAR_a && c <= CHAR_z) c -= 32;
863      c ^= 0x40;      c ^= 0x40;
864  #else           /* EBCDIC coding */  #else             /* EBCDIC coding */
865      if (c >= CHAR_a && c <= CHAR_z) c += 64;      if (c >= CHAR_a && c <= CHAR_z) c += 64;
866      c ^= 0xC0;      c ^= 0xC0;
867  #endif  #endif
# Line 824  else Line 884  else
884      }      }
885    }    }
886    
887    /* Perl supports \N{name} for character names, as well as plain \N for "not
888    newline". PCRE does not support \N{name}. */
889    
890    if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET)
891      *errorcodeptr = ERR37;
892    
893    /* If PCRE_UCP is set, we change the values for \d etc. */
894    
895    if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)
896      c -= (ESC_DU - ESC_D);
897    
898    /* Set the pointer to the final character before returning. */
899    
900  *ptrptr = ptr;  *ptrptr = ptr;
901  return c;  return c;
902  }  }
# Line 1032  top-level call starts at the beginning o Line 1105  top-level call starts at the beginning o
1105  start at a parenthesis. It scans along a pattern's text looking for capturing  start at a parenthesis. It scans along a pattern's text looking for capturing
1106  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
1107  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
1108  returns when it reaches a given numbered subpattern. We know that if (?P< is  returns when it reaches a given numbered subpattern. Recursion is used to keep
1109  encountered, the name will be terminated by '>' because that is checked in the  track of subpatterns that reset the capturing group numbers - the (?| feature.
1110  first pass. Recursion is used to keep track of subpatterns that reset the  
1111  capturing group numbers - the (?| feature.  This function was originally called only from the second pass, in which we know
1112    that if (?< or (?' or (?P< is encountered, the name will be correctly
1113    terminated because that is checked in the first pass. There is now one call to
1114    this function in the first pass, to check for a recursive back reference by
1115    name (so that we can make the whole group atomic). In this case, we need check
1116    only up to the current position in the pattern, and that is still OK because
1117    and previous occurrences will have been checked. To make this work, the test
1118    for "end of pattern" is a check against cd->end_pattern in the main loop,
1119    instead of looking for a binary zero. This means that the special first-pass
1120    call can adjust cd->end_pattern temporarily. (Checks for binary zero while
1121    processing items within the loop are OK, because afterwards the main loop will
1122    terminate.)
1123    
1124  Arguments:  Arguments:
1125    ptrptr       address of the current character pointer (updated)    ptrptr       address of the current character pointer (updated)
# Line 1043  Arguments: Line 1127  Arguments:
1127    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1128    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1129    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1130      utf8         TRUE if we are in UTF-8 mode
1131    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1132    
1133  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
# Line 1050  Returns:       the number of the named s Line 1135  Returns:       the number of the named s
1135    
1136  static int  static int
1137  find_parens_sub(uschar **ptrptr, compile_data *cd, const uschar *name, int lorn,  find_parens_sub(uschar **ptrptr, compile_data *cd, const uschar *name, int lorn,
1138    BOOL xmode, int *count)    BOOL xmode, BOOL utf8, int *count)
1139  {  {
1140  uschar *ptr = *ptrptr;  uschar *ptr = *ptrptr;
1141  int start_count = *count;  int start_count = *count;
# Line 1062  dealing with. The very first call may no Line 1147  dealing with. The very first call may no
1147    
1148  if (ptr[0] == CHAR_LEFT_PARENTHESIS)  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1149    {    {
1150    if (ptr[1] == CHAR_QUESTION_MARK &&    /* Handle specials such as (*SKIP) or (*UTF8) etc. */
1151        ptr[2] == CHAR_VERTICAL_LINE)  
1152      if (ptr[1] == CHAR_ASTERISK) ptr += 2;
1153    
1154      /* Handle a normal, unnamed capturing parenthesis. */
1155    
1156      else if (ptr[1] != CHAR_QUESTION_MARK)
1157        {
1158        *count += 1;
1159        if (name == NULL && *count == lorn) return *count;
1160        ptr++;
1161        }
1162    
1163      /* All cases now have (? at the start. Remember when we are in a group
1164      where the parenthesis numbers are duplicated. */
1165    
1166      else if (ptr[2] == CHAR_VERTICAL_LINE)
1167      {      {
1168      ptr += 3;      ptr += 3;
1169      dup_parens = TRUE;      dup_parens = TRUE;
1170      }      }
1171    
1172    /* Handle a normal, unnamed capturing parenthesis */    /* Handle comments; all characters are allowed until a ket is reached. */
1173    
1174    else if (ptr[1] != CHAR_QUESTION_MARK && ptr[1] != CHAR_ASTERISK)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1175      {      {
1176      *count += 1;      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1177      if (name == NULL && *count == lorn) return *count;      goto FAIL_EXIT;
     ptr++;  
1178      }      }
1179    
1180    /* Handle a condition. If it is an assertion, just carry on so that it    /* Handle a condition. If it is an assertion, just carry on so that it
1181    is processed as normal. If not, skip to the closing parenthesis of the    is processed as normal. If not, skip to the closing parenthesis of the
1182    condition (there can't be any nested parens. */    condition (there can't be any nested parens). */
1183    
1184    else if (ptr[2] == CHAR_LEFT_PARENTHESIS)    else if (ptr[2] == CHAR_LEFT_PARENTHESIS)
1185      {      {
# Line 1092  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1191  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1191        }        }
1192      }      }
1193    
1194    /* We have either (? or (* and not a condition */    /* Start with (? but not a condition. */
1195    
1196    else    else
1197      {      {
# Line 1121  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1220  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1220    }    }
1221    
1222  /* Past any initial parenthesis handling, scan for parentheses or vertical  /* Past any initial parenthesis handling, scan for parentheses or vertical
1223  bars. */  bars. Stop if we get to cd->end_pattern. Note that this is important for the
1224    first-pass call when this value is temporarily adjusted to stop at the current
1225    position. So DO NOT change this to a test for binary zero. */
1226    
1227  for (; *ptr != 0; ptr++)  for (; ptr < cd->end_pattern; ptr++)
1228    {    {
1229    /* Skip over backslashed characters and also entire \Q...\E */    /* Skip over backslashed characters and also entire \Q...\E */
1230    
# Line 1197  for (; *ptr != 0; ptr++) Line 1298  for (; *ptr != 0; ptr++)
1298    
1299    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1300      {      {
1301      while (*(++ptr) != 0 && *ptr != CHAR_NL) {};      ptr++;
1302        while (*ptr != 0)
1303          {
1304          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1305          ptr++;
1306    #ifdef SUPPORT_UTF8
1307          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
1308    #endif
1309          }
1310      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1311      continue;      continue;
1312      }      }
# Line 1206  for (; *ptr != 0; ptr++) Line 1315  for (; *ptr != 0; ptr++)
1315    
1316    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1317      {      {
1318      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);
1319      if (rc > 0) return rc;      if (rc > 0) return rc;
1320      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1321      }      }
# Line 1214  for (; *ptr != 0; ptr++) Line 1323  for (; *ptr != 0; ptr++)
1323    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
1324      {      {
1325      if (dup_parens && *count < hwm_count) *count = hwm_count;      if (dup_parens && *count < hwm_count) *count = hwm_count;
1326      *ptrptr = ptr;      goto FAIL_EXIT;
     return -1;  
1327      }      }
1328    
1329    else if (*ptr == CHAR_VERTICAL_LINE && dup_parens)    else if (*ptr == CHAR_VERTICAL_LINE && dup_parens)
# Line 1253  Arguments: Line 1361  Arguments:
1361    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1362    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1363    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1364      utf8         TRUE if we are in UTF-8 mode
1365    
1366  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1367  */  */
1368    
1369  static int  static int
1370  find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode)  find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode,
1371      BOOL utf8)
1372  {  {
1373  uschar *ptr = (uschar *)cd->start_pattern;  uschar *ptr = (uschar *)cd->start_pattern;
1374  int count = 0;  int count = 0;
# Line 1271  matching closing parens. That is why we Line 1381  matching closing parens. That is why we
1381    
1382  for (;;)  for (;;)
1383    {    {
1384    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);
1385    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == 0) break;
1386    }    }
1387    
# Line 1615  for (;;) Line 1725  for (;;)
1725    
1726    /* 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
1727    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
1728    two bytes of parameters. */    two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we
1729      must add in its length. */
1730    
1731    else    else
1732      {      {
# Line 1639  for (;;) Line 1750  for (;;)
1750        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
1751        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1752        break;        break;
1753    
1754          case OP_MARK:
1755          case OP_PRUNE_ARG:
1756          case OP_SKIP_ARG:
1757          code += code[1];
1758          break;
1759    
1760          case OP_THEN_ARG:
1761          code += code[1+LINK_SIZE];
1762          break;
1763        }        }
1764    
1765      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
# Line 1710  for (;;) Line 1831  for (;;)
1831    
1832    /* 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
1833    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
1834    two bytes of parameters. */    two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we
1835      must add in its length. */
1836    
1837    else    else
1838      {      {
# Line 1734  for (;;) Line 1856  for (;;)
1856        case OP_TYPEEXACT:        case OP_TYPEEXACT:
1857        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1858        break;        break;
1859    
1860          case OP_MARK:
1861          case OP_PRUNE_ARG:
1862          case OP_SKIP_ARG:
1863          code += code[1];
1864          break;
1865    
1866          case OP_THEN_ARG:
1867          code += code[1+LINK_SIZE];
1868          break;
1869        }        }
1870    
1871      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
# Line 1807  for (code = first_significant_code(code Line 1939  for (code = first_significant_code(code
1939    const uschar *ccode;    const uschar *ccode;
1940    
1941    c = *code;    c = *code;
1942    
1943    /* Skip over forward assertions; the other assertions are skipped by    /* Skip over forward assertions; the other assertions are skipped by
1944    first_significant_code() with a TRUE final argument. */    first_significant_code() with a TRUE final argument. */
1945    
# Line 1827  for (code = first_significant_code(code Line 1959  for (code = first_significant_code(code
1959      c = *code;      c = *code;
1960      continue;      continue;
1961      }      }
1962    
1963    /* For a recursion/subroutine call, if its end has been reached, which    /* For a recursion/subroutine call, if its end has been reached, which
1964    implies a subroutine call, we can scan it. */    implies a subroutine call, we can scan it. */
1965    
1966    if (c == OP_RECURSE)    if (c == OP_RECURSE)
1967      {      {
1968      BOOL empty_branch = FALSE;      BOOL empty_branch = FALSE;
1969      const uschar *scode = cd->start_code + GET(code, 1);      const uschar *scode = cd->start_code + GET(code, 1);
1970      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
1971      do      do
# Line 1841  for (code = first_significant_code(code Line 1973  for (code = first_significant_code(code
1973        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf8, cd))
1974          {          {
1975          empty_branch = TRUE;          empty_branch = TRUE;
1976          break;          break;
1977          }          }
1978        scode += GET(scode, 1);        scode += GET(scode, 1);
1979        }        }
1980      while (*scode == OP_ALT);      while (*scode == OP_ALT);
1981      if (!empty_branch) return FALSE;  /* All branches are non-empty */      if (!empty_branch) return FALSE;  /* All branches are non-empty */
1982      continue;      continue;
1983      }      }
1984    
1985    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1986    
# Line 2003  for (code = first_significant_code(code Line 2135  for (code = first_significant_code(code
2135      break;      break;
2136  #endif  #endif
2137    
2138        /* MARK, and PRUNE/SKIP/THEN with an argument must skip over the argument
2139        string. */
2140    
2141        case OP_MARK:
2142        case OP_PRUNE_ARG:
2143        case OP_SKIP_ARG:
2144        code += code[1];
2145        break;
2146    
2147        case OP_THEN_ARG:
2148        code += code[1+LINK_SIZE];
2149        break;
2150    
2151      /* None of the remaining opcodes are required to match a character. */      /* None of the remaining opcodes are required to match a character. */
2152    
2153      default:      default:
2154      break;      break;
2155      }      }
2156    }    }
2157    
# Line 2029  Arguments: Line 2174  Arguments:
2174    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2175    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2176    utf8        TRUE if in UTF-8 mode    utf8        TRUE if in UTF-8 mode
2177    cd          pointers to tables etc    cd          pointers to tables etc
2178    
2179  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2180  */  */
# Line 2223  auto_callout(uschar *code, const uschar Line 2368  auto_callout(uschar *code, const uschar
2368  {  {
2369  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2370  *code++ = 255;  *code++ = 255;
2371  PUT(code, 0, ptr - cd->start_pattern);  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2372  PUT(code, LINK_SIZE, 0);                /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2373  return code + 2*LINK_SIZE;  return code + 2*LINK_SIZE;
2374  }  }
2375    
# Line 2249  Returns:             nothing Line 2394  Returns:             nothing
2394  static void  static void
2395  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)
2396  {  {
2397  int length = ptr - cd->start_pattern - GET(previous_callout, 2);  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2398  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
2399  }  }
2400    
# Line 2299  for (++c; c <= d; c++) Line 2444  for (++c; c <= d; c++)
2444    
2445  return TRUE;  return TRUE;
2446  }  }
2447    
2448    
2449    
2450    /*************************************************
2451    *        Check a character and a property        *
2452    *************************************************/
2453    
2454    /* This function is called by check_auto_possessive() when a property item
2455    is adjacent to a fixed character.
2456    
2457    Arguments:
2458      c            the character
2459      ptype        the property type
2460      pdata        the data for the type
2461      negated      TRUE if it's a negated property (\P or \p{^)
2462    
2463    Returns:       TRUE if auto-possessifying is OK
2464    */
2465    
2466    static BOOL
2467    check_char_prop(int c, int ptype, int pdata, BOOL negated)
2468    {
2469    const ucd_record *prop = GET_UCD(c);
2470    switch(ptype)
2471      {
2472      case PT_LAMP:
2473      return (prop->chartype == ucp_Lu ||
2474              prop->chartype == ucp_Ll ||
2475              prop->chartype == ucp_Lt) == negated;
2476    
2477      case PT_GC:
2478      return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;
2479    
2480      case PT_PC:
2481      return (pdata == prop->chartype) == negated;
2482    
2483      case PT_SC:
2484      return (pdata == prop->script) == negated;
2485    
2486      /* These are specials */
2487    
2488      case PT_ALNUM:
2489      return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2490              _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;
2491    
2492      case PT_SPACE:    /* Perl space */
2493      return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2494              c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2495              == negated;
2496    
2497      case PT_PXSPACE:  /* POSIX space */
2498      return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2499              c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2500              c == CHAR_FF || c == CHAR_CR)
2501              == negated;
2502    
2503      case PT_WORD:
2504      return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2505              _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2506              c == CHAR_UNDERSCORE) == negated;
2507      }
2508    return FALSE;
2509    }
2510  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2511    
2512    
# Line 2312  whether the next thing could possibly ma Line 2520  whether the next thing could possibly ma
2520  sense to automatically possessify the repeated item.  sense to automatically possessify the repeated item.
2521    
2522  Arguments:  Arguments:
2523    op_code       the repeated op code    previous      pointer to the repeated opcode
   this          data for this item, depends on the opcode  
2524    utf8          TRUE in UTF-8 mode    utf8          TRUE in UTF-8 mode
   utf8_char     used for utf8 character bytes, NULL if not relevant  
2525    ptr           next character in pattern    ptr           next character in pattern
2526    options       options bits    options       options bits
2527    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2324  Returns:        TRUE if possessifying is Line 2530  Returns:        TRUE if possessifying is
2530  */  */
2531    
2532  static BOOL  static BOOL
2533  check_auto_possessive(int op_code, int item, BOOL utf8, uschar *utf8_char,  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,
2534    const uschar *ptr, int options, compile_data *cd)    int options, compile_data *cd)
2535  {  {
2536  int next;  int c, next;
2537    int op_code = *previous++;
2538    
2539  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
2540    
# Line 2338  if ((options & PCRE_EXTENDED) != 0) Line 2545  if ((options & PCRE_EXTENDED) != 0)
2545      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2546      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
2547        {        {
2548        while (*(++ptr) != 0)        ptr++;
2549          while (*ptr != 0)
2550            {
2551          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
2552            ptr++;
2553    #ifdef SUPPORT_UTF8
2554            if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
2555    #endif
2556            }
2557        }        }
2558      else break;      else break;
2559      }      }
# Line 2375  if ((options & PCRE_EXTENDED) != 0) Line 2589  if ((options & PCRE_EXTENDED) != 0)
2589      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2590      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
2591        {        {
2592        while (*(++ptr) != 0)        ptr++;
2593          while (*ptr != 0)
2594            {
2595          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
2596            ptr++;
2597    #ifdef SUPPORT_UTF8
2598            if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
2599    #endif
2600            }
2601        }        }
2602      else break;      else break;
2603      }      }
# Line 2388  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 2609  if (*ptr == CHAR_ASTERISK || *ptr == CHA
2609    strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)    strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
2610      return FALSE;      return FALSE;
2611    
2612  /* Now compare the next item with the previous opcode. If the previous is a  /* Now compare the next item with the previous opcode. First, handle cases when
2613  positive single character match, "item" either contains the character or, if  the next item is a character. */
 "item" is greater than 127 in utf8 mode, the character's bytes are in  
 utf8_char. */  
   
   
 /* Handle cases when the next item is a character. */  
2614    
2615  if (next >= 0) switch(op_code)  if (next >= 0) switch(op_code)
2616    {    {
2617    case OP_CHAR:    case OP_CHAR:
2618  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2619    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2620  #else  #else
2621    (void)(utf8_char);  /* Keep compiler happy by referencing function argument */    c = *previous;
2622  #endif  #endif
2623    return item != next;    return c != next;
2624    
2625    /* For CHARNC (caseless character) we must check the other case. If we have    /* For CHARNC (caseless character) we must check the other case. If we have
2626    Unicode property support, we can use it to test the other case of    Unicode property support, we can use it to test the other case of
# Line 2412  if (next >= 0) switch(op_code) Line 2628  if (next >= 0) switch(op_code)
2628    
2629    case OP_CHARNC:    case OP_CHARNC:
2630  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2631    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2632    #else
2633      c = *previous;
2634  #endif  #endif
2635    if (item == next) return FALSE;    if (c == next) return FALSE;
2636  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2637    if (utf8)    if (utf8)
2638      {      {
# Line 2425  if (next >= 0) switch(op_code) Line 2643  if (next >= 0) switch(op_code)
2643  #else  #else
2644      othercase = NOTACHAR;      othercase = NOTACHAR;
2645  #endif  #endif
2646      return (unsigned int)item != othercase;      return (unsigned int)c != othercase;
2647      }      }
2648    else    else
2649  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2650    return (item != cd->fcc[next]);  /* Non-UTF-8 mode */    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */
2651    
2652    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, its data is always a single-byte character. */
2653    
2654    case OP_NOT:    case OP_NOT:
2655    if (item == next) return TRUE;    if ((c = *previous) == next) return TRUE;
2656    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2657  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2658    if (utf8)    if (utf8)
# Line 2446  if (next >= 0) switch(op_code) Line 2664  if (next >= 0) switch(op_code)
2664  #else  #else
2665      othercase = NOTACHAR;      othercase = NOTACHAR;
2666  #endif  #endif
2667      return (unsigned int)item == othercase;      return (unsigned int)c == othercase;
2668      }      }
2669    else    else
2670  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2671    return (item == cd->fcc[next]);  /* Non-UTF-8 mode */    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */
2672    
2673      /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.
2674      When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
2675    
2676    case OP_DIGIT:    case OP_DIGIT:
2677    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;
# Line 2493  if (next >= 0) switch(op_code) Line 2714  if (next >= 0) switch(op_code)
2714      case 0x202f:      case 0x202f:
2715      case 0x205f:      case 0x205f:
2716      case 0x3000:      case 0x3000:
2717      return op_code != OP_HSPACE;      return op_code == OP_NOT_HSPACE;
2718      default:      default:
2719      return op_code == OP_HSPACE;      return op_code != OP_NOT_HSPACE;
2720      }      }
2721    
2722      case OP_ANYNL:
2723    case OP_VSPACE:    case OP_VSPACE:
2724    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
2725    switch(next)    switch(next)
# Line 2509  if (next >= 0) switch(op_code) Line 2731  if (next >= 0) switch(op_code)
2731      case 0x85:      case 0x85:
2732      case 0x2028:      case 0x2028:
2733      case 0x2029:      case 0x2029:
2734      return op_code != OP_VSPACE;      return op_code == OP_NOT_VSPACE;
2735      default:      default:
2736      return op_code == OP_VSPACE;      return op_code != OP_NOT_VSPACE;
2737      }      }
2738    
2739    #ifdef SUPPORT_UCP
2740      case OP_PROP:
2741      return check_char_prop(next, previous[0], previous[1], FALSE);
2742    
2743      case OP_NOTPROP:
2744      return check_char_prop(next, previous[0], previous[1], TRUE);
2745    #endif
2746    
2747    default:    default:
2748    return FALSE;    return FALSE;
2749    }    }
2750    
2751    
2752  /* 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
2753    is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are
2754    generated only when PCRE_UCP is *not* set, that is, when only ASCII
2755    characteristics are recognized. Similarly, the opcodes OP_DIGIT etc. are
2756    replaced by OP_PROP codes when PCRE_UCP is set. */
2757    
2758  switch(op_code)  switch(op_code)
2759    {    {
2760    case OP_CHAR:    case OP_CHAR:
2761    case OP_CHARNC:    case OP_CHARNC:
2762  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2763    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2764    #else
2765      c = *previous;
2766  #endif  #endif
2767    switch(-next)    switch(-next)
2768      {      {
2769      case ESC_d:      case ESC_d:
2770      return item > 127 || (cd->ctypes[item] & ctype_digit) == 0;      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;
2771    
2772      case ESC_D:      case ESC_D:
2773      return item <= 127 && (cd->ctypes[item] & ctype_digit) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;
2774    
2775      case ESC_s:      case ESC_s:
2776      return item > 127 || (cd->ctypes[item] & ctype_space) == 0;      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;
2777    
2778      case ESC_S:      case ESC_S:
2779      return item <= 127 && (cd->ctypes[item] & ctype_space) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;
2780    
2781      case ESC_w:      case ESC_w:
2782      return item > 127 || (cd->ctypes[item] & ctype_word) == 0;      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;
2783    
2784      case ESC_W:      case ESC_W:
2785      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;
2786    
2787      case ESC_h:      case ESC_h:
2788      case ESC_H:      case ESC_H:
2789      switch(item)      switch(c)
2790        {        {
2791        case 0x09:        case 0x09:
2792        case 0x20:        case 0x20:
# Line 2578  switch(op_code) Line 2814  switch(op_code)
2814    
2815      case ESC_v:      case ESC_v:
2816      case ESC_V:      case ESC_V:
2817      switch(item)      switch(c)
2818        {        {
2819        case 0x0a:        case 0x0a:
2820        case 0x0b:        case 0x0b:
# Line 2592  switch(op_code) Line 2828  switch(op_code)
2828        return -next == ESC_v;        return -next == ESC_v;
2829        }        }
2830    
2831        /* When PCRE_UCP is set, these values get generated for \d etc. Find
2832        their substitutions and process them. The result will always be either
2833        -ESC_p or -ESC_P. Then fall through to process those values. */
2834    
2835    #ifdef SUPPORT_UCP
2836        case ESC_du:
2837        case ESC_DU:
2838        case ESC_wu:
2839        case ESC_WU:
2840        case ESC_su:
2841        case ESC_SU:
2842          {
2843          int temperrorcode = 0;
2844          ptr = substitutes[-next - ESC_DU];
2845          next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);
2846          if (temperrorcode != 0) return FALSE;
2847          ptr++;    /* For compatibility */
2848          }
2849        /* Fall through */
2850    
2851        case ESC_p:
2852        case ESC_P:
2853          {
2854          int ptype, pdata, errorcodeptr;
2855          BOOL negated;
2856    
2857          ptr--;      /* Make ptr point at the p or P */
2858          ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);
2859          if (ptype < 0) return FALSE;
2860          ptr++;      /* Point past the final curly ket */
2861    
2862          /* If the property item is optional, we have to give up. (When generated
2863          from \d etc by PCRE_UCP, this test will have been applied much earlier,
2864          to the original \d etc. At this point, ptr will point to a zero byte. */
2865    
2866          if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
2867            strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
2868              return FALSE;
2869    
2870          /* Do the property check. */
2871    
2872          return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);
2873          }
2874    #endif
2875    
2876      default:      default:
2877      return FALSE;      return FALSE;
2878      }      }
2879    
2880      /* In principle, support for Unicode properties should be integrated here as
2881      well. It means re-organizing the above code so as to get hold of the property
2882      values before switching on the op-code. However, I wonder how many patterns
2883      combine ASCII \d etc with Unicode properties? (Note that if PCRE_UCP is set,
2884      these op-codes are never generated.) */
2885    
2886    case OP_DIGIT:    case OP_DIGIT:
2887    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2888           next == -ESC_h || next == -ESC_v;           next == -ESC_h || next == -ESC_v || next == -ESC_R;
2889    
2890    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2891    return next == -ESC_d;    return next == -ESC_d;
2892    
2893    case OP_WHITESPACE:    case OP_WHITESPACE:
2894    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;
2895    
2896    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2897    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2898    
2899    case OP_HSPACE:    case OP_HSPACE:
2900    return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||
2901             next == -ESC_w || next == -ESC_v || next == -ESC_R;
2902    
2903    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
2904    return next == -ESC_h;    return next == -ESC_h;
2905    
2906    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
2907      case OP_ANYNL:
2908    case OP_VSPACE:    case OP_VSPACE:
2909    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2910    
2911    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
2912    return next == -ESC_v;    return next == -ESC_v || next == -ESC_R;
2913    
2914    case OP_WORDCHAR:    case OP_WORDCHAR:
2915    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||
2916             next == -ESC_v || next == -ESC_R;
2917    
2918    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2919    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2687  BOOL inescq = FALSE; Line 2977  BOOL inescq = FALSE;
2977  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstbyte = FALSE;
2978  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
2979  const uschar *tempptr;  const uschar *tempptr;
2980    const uschar *nestptr = NULL;
2981  uschar *previous = NULL;  uschar *previous = NULL;
2982  uschar *previous_callout = NULL;  uschar *previous_callout = NULL;
2983  uschar *save_hwm = NULL;  uschar *save_hwm = NULL;
# Line 2757  for (;; ptr++) Line 3048  for (;; ptr++)
3048    
3049    c = *ptr;    c = *ptr;
3050    
3051      /* If we are at the end of a nested substitution, revert to the outer level
3052      string. Nesting only happens one level deep. */
3053    
3054      if (c == 0 && nestptr != NULL)
3055        {
3056        ptr = nestptr;
3057        nestptr = NULL;
3058        c = *ptr;
3059        }
3060    
3061    /* 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
3062    previous cycle of this loop. */    previous cycle of this loop. */
3063    
# Line 2787  for (;; ptr++) Line 3088  for (;; ptr++)
3088        goto FAILED;        goto FAILED;
3089        }        }
3090    
3091      *lengthptr += code - last_code;      *lengthptr += (int)(code - last_code);
3092      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));
3093    
3094      /* If "previous" is set and it is not at the start of the work space, move      /* If "previous" is set and it is not at the start of the work space, move
# Line 2869  for (;; ptr++) Line 3170  for (;; ptr++)
3170      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if ((cd->ctypes[c] & ctype_space) != 0) continue;
3171      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3172        {        {
3173        while (*(++ptr) != 0)        ptr++;
3174          while (*ptr != 0)
3175          {          {
3176          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3177            ptr++;
3178    #ifdef SUPPORT_UTF8
3179            if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
3180    #endif
3181          }          }
3182        if (*ptr != 0) continue;        if (*ptr != 0) continue;
3183    
# Line 2905  for (;; ptr++) Line 3211  for (;; ptr++)
3211          *errorcodeptr = ERR20;          *errorcodeptr = ERR20;
3212          goto FAILED;          goto FAILED;
3213          }          }
3214        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += (int)(code - last_code);   /* To include callout length */
3215        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
3216        }        }
3217      return TRUE;      return TRUE;
# Line 3110  for (;; ptr++) Line 3416  for (;; ptr++)
3416            ptr++;            ptr++;
3417            }            }
3418    
3419          posix_class = check_posix_name(ptr, tempptr - ptr);          posix_class = check_posix_name(ptr, (int)(tempptr - ptr));
3420          if (posix_class < 0)          if (posix_class < 0)
3421            {            {
3422            *errorcodeptr = ERR30;            *errorcodeptr = ERR30;
# Line 3124  for (;; ptr++) Line 3430  for (;; ptr++)
3430          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
3431            posix_class = 0;            posix_class = 0;
3432    
3433          /* 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
3434          because we may be adding and subtracting from it, and we don't want to          different escape sequences that use Unicode properties. */
3435          subtract bits that may be in the main map already. At the end we or the  
3436          result into the bit map that is being built. */  #ifdef SUPPORT_UCP
3437            if ((options & PCRE_UCP) != 0)
3438              {
3439              int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);
3440              if (posix_substitutes[pc] != NULL)
3441                {
3442                nestptr = tempptr + 1;
3443                ptr = posix_substitutes[pc] - 1;
3444                continue;
3445                }
3446              }
3447    #endif
3448            /* In the non-UCP case, we build the bit map for the POSIX class in a
3449            chunk of local store because we may be adding and subtracting from it,
3450            and we don't want to subtract bits that may be in the main map already.
3451            At the end we or the result into the bit map that is being built. */
3452    
3453          posix_class *= 3;          posix_class *= 3;
3454    
# Line 3171  for (;; ptr++) Line 3492  for (;; ptr++)
3492    
3493        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
3494        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
3495        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
3496        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
3497        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
3498        character in them, so set class_charcount bigger than one. */        are either treated as literal characters (by default), or are faulted if
3499          PCRE_EXTRA is set. */
3500    
3501        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
3502          {          {
3503          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3504          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
3505    
3506          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 */  
3507          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
3508            {            {
3509            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3200  for (;; ptr++) Line 3520  for (;; ptr++)
3520            register const uschar *cbits = cd->cbits;            register const uschar *cbits = cd->cbits;
3521            class_charcount += 2;     /* Greater than 1 is what matters */            class_charcount += 2;     /* Greater than 1 is what matters */
3522    
3523            /* Save time by not doing this in the pre-compile phase. */            switch (-c)
   
           if (lengthptr == NULL) switch (-c)  
3524              {              {
3525    #ifdef SUPPORT_UCP
3526                case ESC_du:     /* These are the values given for \d etc */
3527                case ESC_DU:     /* when PCRE_UCP is set. We replace the */
3528                case ESC_wu:     /* escape sequence with an appropriate \p */
3529                case ESC_WU:     /* or \P to test Unicode properties instead */
3530                case ESC_su:     /* of the default ASCII testing. */
3531                case ESC_SU:
3532                nestptr = ptr;
3533                ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
3534                class_charcount -= 2;                /* Undo! */
3535                continue;
3536    #endif
3537              case ESC_d:              case ESC_d:
3538              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];
3539              continue;              continue;
# Line 3222  for (;; ptr++) Line 3552  for (;; ptr++)
3552              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
3553              continue;              continue;
3554    
3555                /* Perl 5.004 onwards omits VT from \s, but we must preserve it
3556                if it was previously set by something earlier in the character
3557                class. */
3558    
3559              case ESC_s:              case ESC_s:
3560              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_space];              classbits[0] |= cbits[cbit_space];
3561              classbits[1] &= ~0x08;   /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= cbits[cbit_space+1] & ~0x08;
3562                for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];
3563              continue;              continue;
3564    
3565              case ESC_S:              case ESC_S:
# Line 3233  for (;; ptr++) Line 3568  for (;; ptr++)
3568              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
3569              continue;              continue;
3570    
3571              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)  
             {  
3572              SETBIT(classbits, 0x09); /* VT */              SETBIT(classbits, 0x09); /* VT */
3573              SETBIT(classbits, 0x20); /* SPACE */              SETBIT(classbits, 0x20); /* SPACE */
3574              SETBIT(classbits, 0xa0); /* NSBP */              SETBIT(classbits, 0xa0); /* NSBP */
# Line 3270  for (;; ptr++) Line 3592  for (;; ptr++)
3592                }                }
3593  #endif  #endif
3594              continue;              continue;
             }  
3595    
3596            if (-c == ESC_H)              case ESC_H:
             {  
3597              for (c = 0; c < 32; c++)              for (c = 0; c < 32; c++)
3598                {                {
3599                int x = 0xff;                int x = 0xff;
# Line 3315  for (;; ptr++) Line 3635  for (;; ptr++)
3635                }                }
3636  #endif  #endif
3637              continue;              continue;
             }  
3638    
3639            if (-c == ESC_v)              case ESC_v:
             {  
3640              SETBIT(classbits, 0x0a); /* LF */              SETBIT(classbits, 0x0a); /* LF */
3641              SETBIT(classbits, 0x0b); /* VT */              SETBIT(classbits, 0x0b); /* VT */
3642              SETBIT(classbits, 0x0c); /* FF */              SETBIT(classbits, 0x0c); /* FF */
# Line 3334  for (;; ptr++) Line 3652  for (;; ptr++)
3652                }                }
3653  #endif  #endif
3654              continue;              continue;
             }  
3655    
3656            if (-c == ESC_V)              case ESC_V:
             {  
3657              for (c = 0; c < 32; c++)              for (c = 0; c < 32; c++)
3658                {                {
3659                int x = 0xff;                int x = 0xff;
# Line 3367  for (;; ptr++) Line 3683  for (;; ptr++)
3683                }                }
3684  #endif  #endif
3685              continue;              continue;
             }  
   
           /* We need to deal with \P and \p in both phases. */  
3686    
3687  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3688            if (-c == ESC_p || -c == ESC_P)              case ESC_p:
3689              {              case ESC_P:
3690              BOOL negated;                {
3691              int pdata;                BOOL negated;
3692              int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int pdata;
3693              if (ptype < 0) goto FAILED;                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
3694              class_utf8 = TRUE;                if (ptype < 0) goto FAILED;
3695              *class_utf8data++ = ((-c == ESC_p) != negated)?                class_utf8 = TRUE;
3696                XCL_PROP : XCL_NOTPROP;                *class_utf8data++ = ((-c == ESC_p) != negated)?
3697              *class_utf8data++ = ptype;                  XCL_PROP : XCL_NOTPROP;
3698              *class_utf8data++ = pdata;                *class_utf8data++ = ptype;
3699              class_charcount -= 2;   /* Not a < 256 character */                *class_utf8data++ = pdata;
3700              continue;                class_charcount -= 2;   /* Not a < 256 character */
3701              }                continue;
3702                  }
3703  #endif  #endif
3704            /* Unrecognized escapes are faulted if PCRE is running in its              /* Unrecognized escapes are faulted if PCRE is running in its
3705            strict mode. By default, for compatibility with Perl, they are              strict mode. By default, for compatibility with Perl, they are
3706            treated as literals. */              treated as literals. */
3707    
3708            if ((options & PCRE_EXTRA) != 0)              default:
3709              {              if ((options & PCRE_EXTRA) != 0)
3710              *errorcodeptr = ERR7;                {
3711              goto FAILED;                *errorcodeptr = ERR7;
3712                  goto FAILED;
3713                  }
3714                class_charcount -= 2;  /* Undo the default count from above */
3715                c = *ptr;              /* Get the final character and fall through */
3716                break;
3717              }              }
   
           class_charcount -= 2;  /* Undo the default count from above */  
           c = *ptr;              /* Get the final character and fall through */  
3718            }            }
3719    
3720          /* 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 3468  for (;; ptr++) Line 3784  for (;; ptr++)
3784            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3785            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3786    
3787            /* \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 */  
3788    
3789            if (d < 0)            if (d < 0)
3790              {              {
3791              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  
3792                {                {
3793                ptr = oldptr;                ptr = oldptr;
3794                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto LONE_SINGLE_CHARACTER;  /* A few lines below */
# Line 3641  for (;; ptr++) Line 3954  for (;; ptr++)
3954          }          }
3955        }        }
3956    
3957      /* 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.
3958        If we are at the end of an internal nested string, revert to the outer
3959        string. */
3960    
3961        while (((c = *(++ptr)) != 0 ||
3962               (nestptr != NULL &&
3963                 (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&
3964               (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
3965    
3966      while ((c = *(++ptr)) != 0 && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));      /* Check for missing terminating ']' */
3967    
3968      if (c == 0)                          /* Missing terminating ']' */      if (c == 0)
3969        {        {
3970        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
3971        goto FAILED;        goto FAILED;
3972        }        }
3973    
   
 /* 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  
   
   
3974      /* 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
3975      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
3976      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 3733  we set the flag only if there is a liter Line 4034  we set the flag only if there is a liter
4034    
4035      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4036      extended class, with its own opcode, unless there was a negated special      extended class, with its own opcode, unless there was a negated special
4037      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
4038      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
4039      (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
4040      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
4041        actual compiled code. */
4042    
4043  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
4044      if (class_utf8 && !should_flip_negation)      if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))
4045        {        {
4046        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
4047        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3765  we set the flag only if there is a liter Line 4067  we set the flag only if there is a liter
4067        }        }
4068  #endif  #endif
4069    
4070      /* 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
4071      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
4072      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
4073      map into the code vector, negating it if necessary. */      (non-UCP) in the class. Then copy the 32-byte map into the code vector,
4074        negating it if necessary. */
4075    
4076      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
4077      if (negate_class)      if (negate_class)
# Line 3892  we set the flag only if there is a liter Line 4195  we set the flag only if there is a liter
4195    
4196        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4197            repeat_max < 0 &&            repeat_max < 0 &&
4198            check_auto_possessive(*previous, c, utf8, utf8_char, ptr + 1,            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
             options, cd))  
4199          {          {
4200          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4201          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 3914  we set the flag only if there is a liter Line 4216  we set the flag only if there is a liter
4216        c = previous[1];        c = previous[1];
4217        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4218            repeat_max < 0 &&            repeat_max < 0 &&
4219            check_auto_possessive(OP_NOT, c, utf8, NULL, ptr + 1, options, cd))            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
4220          {          {
4221          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4222          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 3938  we set the flag only if there is a liter Line 4240  we set the flag only if there is a liter
4240    
4241        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4242            repeat_max < 0 &&            repeat_max < 0 &&
4243            check_auto_possessive(c, 0, utf8, NULL, ptr + 1, options, cd))            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
4244          {          {
4245          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4246          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 4148  we set the flag only if there is a liter Line 4450  we set the flag only if there is a liter
4450        {        {
4451        register int i;        register int i;
4452        int ketoffset = 0;        int ketoffset = 0;
4453        int len = code - previous;        int len = (int)(code - previous);
4454        uschar *bralink = NULL;        uschar *bralink = NULL;
4455    
4456        /* Repeating a DEFINE group is pointless */        /* Repeating a DEFINE group is pointless */
# Line 4169  we set the flag only if there is a liter Line 4471  we set the flag only if there is a liter
4471          {          {
4472          register uschar *ket = previous;          register uschar *ket = previous;
4473          do ket += GET(ket, 1); while (*ket != OP_KET);          do ket += GET(ket, 1); while (*ket != OP_KET);
4474          ketoffset = code - ket;          ketoffset = (int)(code - ket);
4475          }          }
4476    
4477        /* The case of a zero minimum is special because of the need to stick        /* The case of a zero minimum is special because of the need to stick
# Line 4237  we set the flag only if there is a liter Line 4539  we set the flag only if there is a liter
4539            /* We chain together the bracket offset fields that have to be            /* We chain together the bracket offset fields that have to be
4540            filled in later when the ends of the brackets are reached. */            filled in later when the ends of the brackets are reached. */
4541    
4542            offset = (bralink == NULL)? 0 : previous - bralink;            offset = (bralink == NULL)? 0 : (int)(previous - bralink);
4543            bralink = previous;            bralink = previous;
4544            PUTINC(previous, 0, offset);            PUTINC(previous, 0, offset);
4545            }            }
# Line 4346  we set the flag only if there is a liter Line 4648  we set the flag only if there is a liter
4648              {              {
4649              int offset;              int offset;
4650              *code++ = OP_BRA;              *code++ = OP_BRA;
4651              offset = (bralink == NULL)? 0 : code - bralink;              offset = (bralink == NULL)? 0 : (int)(code - bralink);
4652              bralink = code;              bralink = code;
4653              PUTINC(code, 0, offset);              PUTINC(code, 0, offset);
4654              }              }
# Line 4367  we set the flag only if there is a liter Line 4669  we set the flag only if there is a liter
4669          while (bralink != NULL)          while (bralink != NULL)
4670            {            {
4671            int oldlinkoffset;            int oldlinkoffset;
4672            int offset = code - bralink + 1;            int offset = (int)(code - bralink + 1);
4673            uschar *bra = code - offset;            uschar *bra = code - offset;
4674            oldlinkoffset = GET(bra, 1);            oldlinkoffset = GET(bra, 1);
4675            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;
# Line 4455  we set the flag only if there is a liter Line 4757  we set the flag only if there is a liter
4757  #endif  #endif
4758          }          }
4759    
4760        len = code - tempcode;        len = (int)(code - tempcode);
4761        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4762          {          {
4763          case OP_STAR:  *tempcode = OP_POSSTAR; break;          case OP_STAR:  *tempcode = OP_POSSTAR; break;
# Line 4475  we set the flag only if there is a liter Line 4777  we set the flag only if there is a liter
4777    
4778          /* Because we are moving code along, we must ensure that any          /* Because we are moving code along, we must ensure that any
4779          pending recursive references are updated. */          pending recursive references are updated. */
4780    
4781          default:          default:
4782          *code = OP_END;          *code = OP_END;
4783          adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);          adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);
# Line 4514  we set the flag only if there is a liter Line 4816  we set the flag only if there is a liter
4816    
4817      /* First deal with various "verbs" that can be introduced by '*'. */      /* First deal with various "verbs" that can be introduced by '*'. */
4818    
4819      if (*(++ptr) == CHAR_ASTERISK && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == CHAR_ASTERISK &&
4820             ((cd->ctypes[ptr[1]] & ctype_letter) != 0 || ptr[1] == ':'))
4821        {        {
4822        int i, namelen;        int i, namelen;
4823          int arglen = 0;
4824        const char *vn = verbnames;        const char *vn = verbnames;
4825        const uschar *name = ++ptr;        const uschar *name = ptr + 1;
4826          const uschar *arg = NULL;
4827        previous = NULL;        previous = NULL;
4828        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
4829          namelen = (int)(ptr - name);
4830    
4831        if (*ptr == CHAR_COLON)        if (*ptr == CHAR_COLON)
4832          {          {
4833          *errorcodeptr = ERR59;   /* Not supported */          arg = ++ptr;
4834          goto FAILED;          while ((cd->ctypes[*ptr] & (ctype_letter|ctype_digit)) != 0
4835              || *ptr == '_') ptr++;
4836            arglen = (int)(ptr - arg);
4837          }          }
4838    
4839        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
4840          {          {
4841          *errorcodeptr = ERR60;          *errorcodeptr = ERR60;
4842          goto FAILED;          goto FAILED;
4843          }          }
4844        namelen = ptr - name;  
4845          /* Scan the table of verb names */
4846    
4847        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4848          {          {
4849          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
# Line 4549  we set the flag only if there is a liter Line 4861  we set the flag only if there is a liter
4861                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
4862                }                }
4863              }              }
4864            *code++ = verbs[i].op;  
4865            break;            /* Handle the cases with/without an argument */
4866    
4867              if (arglen == 0)
4868                {
4869                if (verbs[i].op < 0)   /* Argument is mandatory */
4870                  {
4871                  *errorcodeptr = ERR66;
4872                  goto FAILED;
4873                  }
4874                *code = verbs[i].op;
4875                if (*code++ == OP_THEN)
4876                  {
4877                  PUT(code, 0, code - bcptr->current_branch - 1);
4878                  code += LINK_SIZE;
4879                  }
4880                }
4881    
4882              else
4883                {
4884                if (verbs[i].op_arg < 0)   /* Argument is forbidden */
4885                  {
4886                  *errorcodeptr = ERR59;
4887                  goto FAILED;
4888                  }
4889                *code = verbs[i].op_arg;
4890                if (*code++ == OP_THEN_ARG)
4891                  {
4892                  PUT(code, 0, code - bcptr->current_branch - 1);
4893                  code += LINK_SIZE;
4894                  }
4895                *code++ = arglen;
4896                memcpy(code, arg, arglen);
4897                code += arglen;
4898                *code++ = 0;
4899                }
4900    
4901              break;  /* Found verb, exit loop */
4902            }            }
4903    
4904          vn += verbs[i].len + 1;          vn += verbs[i].len + 1;
4905          }          }
4906        if (i < verbcount) continue;  
4907        *errorcodeptr = ERR60;        if (i < verbcount) continue;    /* Successfully handled a verb */
4908          *errorcodeptr = ERR60;          /* Verb not recognized */
4909        goto FAILED;        goto FAILED;
4910        }        }
4911    
# Line 4674  we set the flag only if there is a liter Line 5024  we set the flag only if there is a liter
5024                recno * 10 + *ptr - CHAR_0 : -1;                recno * 10 + *ptr - CHAR_0 : -1;
5025            ptr++;            ptr++;
5026            }            }
5027          namelen = ptr - name;          namelen = (int)(ptr - name);
5028    
5029          if ((terminator > 0 && *ptr++ != terminator) ||          if ((terminator > 0 && *ptr++ != terminator) ||
5030              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
# Line 4735  we set the flag only if there is a liter Line 5085  we set the flag only if there is a liter
5085          /* Search the pattern for a forward reference */          /* Search the pattern for a forward reference */
5086    
5087          else if ((i = find_parens(cd, name, namelen,          else if ((i = find_parens(cd, name, namelen,
5088                          (options & PCRE_EXTENDED) != 0)) > 0)                          (options & PCRE_EXTENDED) != 0, utf8)) > 0)
5089            {            {
5090            PUT2(code, 2+LINK_SIZE, i);            PUT2(code, 2+LINK_SIZE, i);
5091            code[1+LINK_SIZE]++;            code[1+LINK_SIZE]++;
# Line 4870  we set the flag only if there is a liter Line 5220  we set the flag only if there is a liter
5220              goto FAILED;              goto FAILED;
5221              }              }
5222            *code++ = n;            *code++ = n;
5223            PUT(code, 0, ptr - cd->start_pattern + 1);  /* Pattern offset */            PUT(code, 0, (int)(ptr - cd->start_pattern + 1)); /* Pattern offset */
5224            PUT(code, LINK_SIZE, 0);                    /* Default length */            PUT(code, LINK_SIZE, 0);                          /* Default length */
5225            code += 2 * LINK_SIZE;            code += 2 * LINK_SIZE;
5226            }            }
5227          previous = NULL;          previous = NULL;
# Line 4904  we set the flag only if there is a liter Line 5254  we set the flag only if there is a liter
5254            name = ++ptr;            name = ++ptr;
5255    
5256            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
5257            namelen = ptr - name;            namelen = (int)(ptr - name);
5258    
5259            /* In the pre-compile phase, just do a syntax check. */            /* In the pre-compile phase, just do a syntax check. */
5260    
# Line 5034  we set the flag only if there is a liter Line 5384  we set the flag only if there is a liter
5384          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
5385          name = ++ptr;          name = ++ptr;
5386          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
5387          namelen = ptr - name;          namelen = (int)(ptr - name);
5388    
5389          /* In the pre-compile phase, do a syntax check and set a dummy          /* In the pre-compile phase, do a syntax check. We used to just set
5390          reference number. */          a dummy reference number, because it was not used in the first pass.
5391            However, with the change of recursive back references to be atomic,
5392            we have to look for the number so that this state can be identified, as
5393            otherwise the incorrect length is computed. If it's not a backwards
5394            reference, the dummy number will do. */
5395    
5396          if (lengthptr != NULL)          if (lengthptr != NULL)
5397            {            {
5398              const uschar *temp;
5399    
5400            if (namelen == 0)            if (namelen == 0)
5401              {              {
5402              *errorcodeptr = ERR62;              *errorcodeptr = ERR62;
# Line 5056  we set the flag only if there is a liter Line 5412  we set the flag only if there is a liter
5412              *errorcodeptr = ERR48;              *errorcodeptr = ERR48;
5413              goto FAILED;              goto FAILED;
5414              }              }
5415            recno = 0;  
5416              /* The name table does not exist in the first pass, so we cannot
5417              do a simple search as in the code below. Instead, we have to scan the
5418              pattern to find the number. It is important that we scan it only as
5419              far as we have got because the syntax of named subpatterns has not
5420              been checked for the rest of the pattern, and find_parens() assumes
5421              correct syntax. In any case, it's a waste of resources to scan
5422              further. We stop the scan at the current point by temporarily
5423              adjusting the value of cd->endpattern. */
5424    
5425              temp = cd->end_pattern;
5426              cd->end_pattern = ptr;
5427              recno = find_parens(cd, name, namelen,
5428                (options & PCRE_EXTENDED) != 0, utf8);
5429              cd->end_pattern = temp;
5430              if (recno < 0) recno = 0;    /* Forward ref; set dummy number */
5431            }            }
5432    
5433          /* In the real compile, seek the name in the table. We check the name          /* In the real compile, seek the name in the table. We check the name
# Line 5081  we set the flag only if there is a liter Line 5452  we set the flag only if there is a liter
5452              }              }
5453            else if ((recno =                /* Forward back reference */            else if ((recno =                /* Forward back reference */
5454                      find_parens(cd, name, namelen,                      find_parens(cd, name, namelen,
5455                        (options & PCRE_EXTENDED) != 0)) <= 0)                        (options & PCRE_EXTENDED) != 0, utf8)) <= 0)
5456              {              {
5457              *errorcodeptr = ERR15;              *errorcodeptr = ERR15;
5458              goto FAILED;              goto FAILED;
# Line 5192  we set the flag only if there is a liter Line 5563  we set the flag only if there is a liter
5563              if (called == NULL)              if (called == NULL)
5564                {                {
5565                if (find_parens(cd, NULL, recno,                if (find_parens(cd, NULL, recno,
5566                      (options & PCRE_EXTENDED) != 0) < 0)                      (options & PCRE_EXTENDED) != 0, utf8) < 0)
5567                  {                  {
5568                  *errorcodeptr = ERR15;                  *errorcodeptr = ERR15;
5569                  goto FAILED;                  goto FAILED;
5570                  }                  }
5571    
5572                /* Fudge the value of "called" so that when it is inserted as an                /* Fudge the value of "called" so that when it is inserted as an
5573                offset below, what it actually inserted is the reference number                offset below, what it actually inserted is the reference number
5574                of the group. */                of the group. */
5575    
5576                called = cd->start_code + recno;                called = cd->start_code + recno;
5577                PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code);                PUTINC(cd->hwm, 0, (int)(code + 2 + LINK_SIZE - cd->start_code));
5578                }                }
5579    
5580              /* If not a forward reference, and the subpattern is still open,              /* If not a forward reference, and the subpattern is still open,
# Line 5227  we set the flag only if there is a liter Line 5598  we set the flag only if there is a liter
5598            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
5599    
5600            *code = OP_RECURSE;            *code = OP_RECURSE;
5601            PUT(code, 1, called - cd->start_code);            PUT(code, 1, (int)(called - cd->start_code));
5602            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
5603    
5604            *code = OP_KET;            *code = OP_KET;
# Line 5338  we set the flag only if there is a liter Line 5709  we set the flag only if there is a liter
5709          }     /* End of switch for character following (? */          }     /* End of switch for character following (? */
5710        }       /* End of (? handling */        }       /* End of (? handling */
5711    
5712      /* Opening parenthesis not followed by '?'. If PCRE_NO_AUTO_CAPTURE is set,      /* Opening parenthesis not followed by '*' or '?'. If PCRE_NO_AUTO_CAPTURE
5713      all unadorned brackets become non-capturing and behave like (?:...)      is set, all unadorned brackets become non-capturing and behave like (?:...)
5714      brackets. */      brackets. */
5715    
5716      else if ((options & PCRE_NO_AUTO_CAPTURE) != 0)      else if ((options & PCRE_NO_AUTO_CAPTURE) != 0)
# Line 5531  we set the flag only if there is a liter Line 5902  we set the flag only if there is a liter
5902    
5903      /* ===================================================================*/      /* ===================================================================*/
5904      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
5905      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
5906      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
5907      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
5908      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
5909      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
5910        ever created. */
5911    
5912      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
5913      tempptr = ptr;      tempptr = ptr;
# Line 5695  we set the flag only if there is a liter Line 6067  we set the flag only if there is a liter
6067  #endif  #endif
6068    
6069        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6070        can obtain the OP value by negating the escape value. */        can obtain the OP value by negating the escape value in the default
6071          situation when PCRE_UCP is not set. When it *is* set, we substitute
6072          Unicode property tests. */
6073    
6074        else        else
6075          {          {
6076          previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;  #ifdef SUPPORT_UCP
6077          *code++ = -c;          if (-c >= ESC_DU && -c <= ESC_wu)
6078              {
6079              nestptr = ptr + 1;                   /* Where to resume */
6080              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
6081              }
6082            else
6083    #endif
6084              {
6085              previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;
6086              *code++ = -c;
6087              }
6088          }          }
6089        continue;        continue;
6090        }        }
# Line 6032  for (;;) Line 6416  for (;;)
6416      {      {
6417      if (lengthptr == NULL)      if (lengthptr == NULL)
6418        {        {
6419        int branch_length = code - last_branch;        int branch_length = (int)(code - last_branch);
6420        do        do
6421          {          {
6422          int prev_length = GET(last_branch, 1);          int prev_length = GET(last_branch, 1);
# Line 6046  for (;;) Line 6430  for (;;)
6430      /* Fill in the ket */      /* Fill in the ket */
6431    
6432      *code = OP_KET;      *code = OP_KET;
6433      PUT(code, 1, code - start_bracket);      PUT(code, 1, (int)(code - start_bracket));
6434      code += 1 + LINK_SIZE;      code += 1 + LINK_SIZE;
6435    
6436      /* If it was a capturing subpattern, check to see if it contained any      /* If it was a capturing subpattern, check to see if it contained any
# Line 6061  for (;;) Line 6445  for (;;)
6445            code - start_bracket);            code - start_bracket);
6446          *start_bracket = OP_ONCE;          *start_bracket = OP_ONCE;
6447          code += 1 + LINK_SIZE;          code += 1 + LINK_SIZE;
6448          PUT(start_bracket, 1, code - start_bracket);          PUT(start_bracket, 1, (int)(code - start_bracket));
6449          *code = OP_KET;          *code = OP_KET;
6450          PUT(code, 1, code - start_bracket);          PUT(code, 1, (int)(code - start_bracket));
6451          code += 1 + LINK_SIZE;          code += 1 + LINK_SIZE;
6452          length += 2 + 2*LINK_SIZE;          length += 2 + 2*LINK_SIZE;
6453          }          }
# Line 6118  for (;;) Line 6502  for (;;)
6502    else    else
6503      {      {
6504      *code = OP_ALT;      *code = OP_ALT;
6505      PUT(code, 1, code - last_branch);      PUT(code, 1, (int)(code - last_branch));
6506      bc.current_branch = last_branch = code;      bc.current_branch = last_branch = code;
6507      code += 1 + LINK_SIZE;      code += 1 + LINK_SIZE;
6508      }      }
# Line 6437  int length = 1;  /* For final END opcode Line 6821  int length = 1;  /* For final END opcode
6821  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
6822  int errorcode = 0;  int errorcode = 0;
6823  int skipatstart = 0;  int skipatstart = 0;
6824  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8;
6825  size_t size;  size_t size;
6826  uschar *code;  uschar *code;
6827  const uschar *codestart;  const uschar *codestart;
# Line 6507  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6891  while (ptr[skipatstart] == CHAR_LEFT_PAR
6891    
6892    if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)
6893      { skipatstart += 7; options |= PCRE_UTF8; continue; }      { skipatstart += 7; options |= PCRE_UTF8; continue; }
6894      else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0)
6895        { skipatstart += 6; options |= PCRE_UCP; continue; }
6896      else if (strncmp((char *)(ptr+skipatstart+2), STRING_NO_START_OPT_RIGHTPAR, 13) == 0)
6897        { skipatstart += 15; options |= PCRE_NO_START_OPTIMIZE; continue; }
6898    
6899    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)
6900      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
# Line 6531  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6919  while (ptr[skipatstart] == CHAR_LEFT_PAR
6919    else break;    else break;
6920    }    }
6921    
6922    utf8 = (options & PCRE_UTF8) != 0;
6923    
6924  /* Can't support UTF8 unless PCRE has been compiled to include the code. */  /* Can't support UTF8 unless PCRE has been compiled to include the code. */
6925    
6926  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 6548  if (utf8) Line 6938  if (utf8)
6938    }    }
6939  #endif  #endif
6940    
6941    /* Can't support UCP unless PCRE has been compiled to include the code. */
6942    
6943    #ifndef SUPPORT_UCP
6944    if ((options & PCRE_UCP) != 0)
6945      {
6946      errorcode = ERR67;
6947      goto PCRE_EARLY_ERROR_RETURN;
6948      }
6949    #endif
6950    
6951  /* Check validity of \R options. */  /* Check validity of \R options. */
6952    
6953  switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))  switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
# Line 6676  regex compiled on a system with 4-byte p Line 7076  regex compiled on a system with 4-byte p
7076  pointers. */  pointers. */
7077    
7078  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
7079  re->size = size;  re->size = (int)size;
7080  re->options = cd->external_options;  re->options = cd->external_options;
7081  re->flags = cd->external_flags;  re->flags = cd->external_flags;
7082  re->dummy1 = 0;  re->dummy1 = 0;
# Line 6747  while (errorcode == 0 && cd->hwm > cwork Line 7147  while (errorcode == 0 && cd->hwm > cwork
7147    recno = GET(codestart, offset);    recno = GET(codestart, offset);
7148    groupptr = _pcre_find_bracket(codestart, utf8, recno);    groupptr = _pcre_find_bracket(codestart, utf8, recno);
7149    if (groupptr == NULL) errorcode = ERR53;    if (groupptr == NULL) errorcode = ERR53;
7150      else PUT(((uschar *)codestart), offset, groupptr - codestart);      else PUT(((uschar *)codestart), offset, (int)(groupptr - codestart));
7151    }    }
7152    
7153  /* Give an error if there's back reference to a non-existent capturing  /* Give an error if there's back reference to a non-existent capturing
# Line 6802  if (errorcode != 0) Line 7202  if (errorcode != 0)
7202    {    {
7203    (pcre_free)(re);    (pcre_free)(re);
7204    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
7205    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = (int)(ptr - (const uschar *)pattern);
7206    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
7207    *errorptr = find_error_text(errorcode);    *errorptr = find_error_text(errorcode);
7208    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;

Legend:
Removed from v.505  
changed lines
  Added in v.578

  ViewVC Help
Powered by ViewVC 1.1.5