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

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

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

revision 496 by ph10, Tue Mar 2 19:11:17 2010 UTC revision 574 by ph10, Sat Nov 20 17:47:27 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
96    overrun before it actually does run off the end of the data block. */
97    
98    #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100)
99    
100    
101  /* Table for handling escaped characters in the range '0'-'z'. Positive returns  /* Table for handling escaped characters in the range '0'-'z'. Positive returns
102  are simple data values; negative values are for special things like \d and so  are simple data values; negative values are for special things like \d and so
# Line 119  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 166  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 183  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 197  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 251  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 263  the number of relocations needed when a Line 320  the number of relocations needed when a
320  it is now one long string. We cannot use a table of offsets, because the  it is now one long string. We cannot use a table of offsets, because the
321  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we
322  simply count through to the one we want - this isn't a performance issue  simply count through to the one we want - this isn't a performance issue
323  because these strings are used only when there is a compilation error. */  because these strings are used only when there is a compilation error.
324    
325    Each substring ends with \0 to insert a null character. This includes the final
326    substring, so that the whole string ends with \0\0, which can be detected when
327    counting through. */
328    
329  static const char error_texts[] =  static const char error_texts[] =
330    "no error\0"    "no error\0"
# Line 310  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 336  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 344  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";    "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 503  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--) while (*s++ != 0) {};  for (; n > 0; n--)
571      {
572      while (*s++ != 0) {};
573      if (*s == 0) return "Error text not found (please report)";
574      }
575  return s;  return s;
576  }  }
577    
# Line 574  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 775  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 785  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 812  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 1031  Arguments: Line 1116  Arguments:
1116    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1117    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1118    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1119      utf8         TRUE if we are in UTF-8 mode
1120    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1121    
1122  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 1038  Returns:       the number of the named s Line 1124  Returns:       the number of the named s
1124    
1125  static int  static int
1126  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,
1127    BOOL xmode, int *count)    BOOL xmode, BOOL utf8, int *count)
1128  {  {
1129  uschar *ptr = *ptrptr;  uschar *ptr = *ptrptr;
1130  int start_count = *count;  int start_count = *count;
# Line 1050  dealing with. The very first call may no Line 1136  dealing with. The very first call may no
1136    
1137  if (ptr[0] == CHAR_LEFT_PARENTHESIS)  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1138    {    {
1139    if (ptr[1] == CHAR_QUESTION_MARK &&    /* Handle specials such as (*SKIP) or (*UTF8) etc. */
1140        ptr[2] == CHAR_VERTICAL_LINE)  
1141      if (ptr[1] == CHAR_ASTERISK) ptr += 2;
1142    
1143      /* Handle a normal, unnamed capturing parenthesis. */
1144    
1145      else if (ptr[1] != CHAR_QUESTION_MARK)
1146        {
1147        *count += 1;
1148        if (name == NULL && *count == lorn) return *count;
1149        ptr++;
1150        }
1151    
1152      /* All cases now have (? at the start. Remember when we are in a group
1153      where the parenthesis numbers are duplicated. */
1154    
1155      else if (ptr[2] == CHAR_VERTICAL_LINE)
1156      {      {
1157      ptr += 3;      ptr += 3;
1158      dup_parens = TRUE;      dup_parens = TRUE;
1159      }      }
1160    
1161    /* Handle a normal, unnamed capturing parenthesis */    /* Handle comments; all characters are allowed until a ket is reached. */
1162    
1163    else if (ptr[1] != CHAR_QUESTION_MARK && ptr[1] != CHAR_ASTERISK)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1164      {      {
1165      *count += 1;      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1166      if (name == NULL && *count == lorn) return *count;      goto FAIL_EXIT;
     ptr++;  
1167      }      }
1168    
1169    /* 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
1170    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
1171    condition (there can't be any nested parens. */    condition (there can't be any nested parens). */
1172    
1173    else if (ptr[2] == CHAR_LEFT_PARENTHESIS)    else if (ptr[2] == CHAR_LEFT_PARENTHESIS)
1174      {      {
# Line 1080  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1180  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1180        }        }
1181      }      }
1182    
1183    /* We have either (? or (* and not a condition */    /* Start with (? but not a condition. */
1184    
1185    else    else
1186      {      {
# Line 1185  for (; *ptr != 0; ptr++) Line 1285  for (; *ptr != 0; ptr++)
1285    
1286    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1287      {      {
1288      while (*(++ptr) != 0 && *ptr != CHAR_NL) {};      ptr++;
1289        while (*ptr != 0)
1290          {
1291          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1292          ptr++;
1293    #ifdef SUPPORT_UTF8
1294          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
1295    #endif
1296          }
1297      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1298      continue;      continue;
1299      }      }
# Line 1194  for (; *ptr != 0; ptr++) Line 1302  for (; *ptr != 0; ptr++)
1302    
1303    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1304      {      {
1305      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);
1306      if (rc > 0) return rc;      if (rc > 0) return rc;
1307      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1308      }      }
# Line 1202  for (; *ptr != 0; ptr++) Line 1310  for (; *ptr != 0; ptr++)
1310    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
1311      {      {
1312      if (dup_parens && *count < hwm_count) *count = hwm_count;      if (dup_parens && *count < hwm_count) *count = hwm_count;
1313      *ptrptr = ptr;      goto FAIL_EXIT;
     return -1;  
1314      }      }
1315    
1316    else if (*ptr == CHAR_VERTICAL_LINE && dup_parens)    else if (*ptr == CHAR_VERTICAL_LINE && dup_parens)
# Line 1241  Arguments: Line 1348  Arguments:
1348    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1349    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1350    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1351      utf8         TRUE if we are in UTF-8 mode
1352    
1353  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1354  */  */
1355    
1356  static int  static int
1357  find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode)  find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode,
1358      BOOL utf8)
1359  {  {
1360  uschar *ptr = (uschar *)cd->start_pattern;  uschar *ptr = (uschar *)cd->start_pattern;
1361  int count = 0;  int count = 0;
# Line 1259  matching closing parens. That is why we Line 1368  matching closing parens. That is why we
1368    
1369  for (;;)  for (;;)
1370    {    {
1371    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);
1372    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == 0) break;
1373    }    }
1374    
# Line 1443  for (;;) Line 1552  for (;;)
1552      case OP_CALLOUT:      case OP_CALLOUT:
1553      case OP_SOD:      case OP_SOD:
1554      case OP_SOM:      case OP_SOM:
1555        case OP_SET_SOM:
1556      case OP_EOD:      case OP_EOD:
1557      case OP_EODN:      case OP_EODN:
1558      case OP_CIRC:      case OP_CIRC:
# Line 1602  for (;;) Line 1712  for (;;)
1712    
1713    /* 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
1714    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
1715    two bytes of parameters. */    two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we
1716      must add in its length. */
1717    
1718    else    else
1719      {      {
# Line 1626  for (;;) Line 1737  for (;;)
1737        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
1738        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1739        break;        break;
1740    
1741          case OP_MARK:
1742          case OP_PRUNE_ARG:
1743          case OP_SKIP_ARG:
1744          code += code[1];
1745          break;
1746    
1747          case OP_THEN_ARG:
1748          code += code[1+LINK_SIZE];
1749          break;
1750        }        }
1751    
1752      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
# Line 1697  for (;;) Line 1818  for (;;)
1818    
1819    /* 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
1820    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
1821    two bytes of parameters. */    two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we
1822      must add in its length. */
1823    
1824    else    else
1825      {      {
# Line 1721  for (;;) Line 1843  for (;;)
1843        case OP_TYPEEXACT:        case OP_TYPEEXACT:
1844        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1845        break;        break;
1846    
1847          case OP_MARK:
1848          case OP_PRUNE_ARG:
1849          case OP_SKIP_ARG:
1850          code += code[1];
1851          break;
1852    
1853          case OP_THEN_ARG:
1854          code += code[1+LINK_SIZE];
1855          break;
1856        }        }
1857    
1858      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
# Line 1777  Arguments: Line 1909  Arguments:
1909    code        points to start of search    code        points to start of search
1910    endcode     points to where to stop    endcode     points to where to stop
1911    utf8        TRUE if in UTF8 mode    utf8        TRUE if in UTF8 mode
1912      cd          contains pointers to tables etc.
1913    
1914  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
1915  */  */
1916    
1917  static BOOL  static BOOL
1918  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8)  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,
1919      compile_data *cd)
1920  {  {
1921  register int c;  register int c;
1922  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);
# Line 1813  for (code = first_significant_code(code Line 1947  for (code = first_significant_code(code
1947      continue;      continue;
1948      }      }
1949    
1950      /* For a recursion/subroutine call, if its end has been reached, which
1951      implies a subroutine call, we can scan it. */
1952    
1953      if (c == OP_RECURSE)
1954        {
1955        BOOL empty_branch = FALSE;
1956        const uschar *scode = cd->start_code + GET(code, 1);
1957        if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
1958        do
1959          {
1960          if (could_be_empty_branch(scode, endcode, utf8, cd))
1961            {
1962            empty_branch = TRUE;
1963            break;
1964            }
1965          scode += GET(scode, 1);
1966          }
1967        while (*scode == OP_ALT);
1968        if (!empty_branch) return FALSE;  /* All branches are non-empty */
1969        continue;
1970        }
1971    
1972    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1973    
1974    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
# Line 1831  for (code = first_significant_code(code Line 1987  for (code = first_significant_code(code
1987        empty_branch = FALSE;        empty_branch = FALSE;
1988        do        do
1989          {          {
1990          if (!empty_branch && could_be_empty_branch(code, endcode, utf8))          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))
1991            empty_branch = TRUE;            empty_branch = TRUE;
1992          code += GET(code, 1);          code += GET(code, 1);
1993          }          }
# Line 1965  for (code = first_significant_code(code Line 2121  for (code = first_significant_code(code
2121      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];
2122      break;      break;
2123  #endif  #endif
2124    
2125        /* MARK, and PRUNE/SKIP/THEN with an argument must skip over the argument
2126        string. */
2127    
2128        case OP_MARK:
2129        case OP_PRUNE_ARG:
2130        case OP_SKIP_ARG:
2131        code += code[1];
2132        break;
2133    
2134        case OP_THEN_ARG:
2135        code += code[1+LINK_SIZE];
2136        break;
2137    
2138        /* None of the remaining opcodes are required to match a character. */
2139    
2140        default:
2141        break;
2142      }      }
2143    }    }
2144    
# Line 1987  Arguments: Line 2161  Arguments:
2161    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2162    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2163    utf8        TRUE if in UTF-8 mode    utf8        TRUE if in UTF-8 mode
2164      cd          pointers to tables etc
2165    
2166  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2167  */  */
2168    
2169  static BOOL  static BOOL
2170  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,
2171    BOOL utf8)    BOOL utf8, compile_data *cd)
2172  {  {
2173  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2174    {    {
2175    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))
2176      return FALSE;      return FALSE;
2177    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2178    }    }
# Line 2180  auto_callout(uschar *code, const uschar Line 2355  auto_callout(uschar *code, const uschar
2355  {  {
2356  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2357  *code++ = 255;  *code++ = 255;
2358  PUT(code, 0, ptr - cd->start_pattern);  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2359  PUT(code, LINK_SIZE, 0);                /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2360  return code + 2*LINK_SIZE;  return code + 2*LINK_SIZE;
2361  }  }
2362    
# Line 2206  Returns:             nothing Line 2381  Returns:             nothing
2381  static void  static void
2382  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)
2383  {  {
2384  int length = ptr - cd->start_pattern - GET(previous_callout, 2);  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2385  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
2386  }  }
2387    
# Line 2256  for (++c; c <= d; c++) Line 2431  for (++c; c <= d; c++)
2431    
2432  return TRUE;  return TRUE;
2433  }  }
2434    
2435    
2436    
2437    /*************************************************
2438    *        Check a character and a property        *
2439    *************************************************/
2440    
2441    /* This function is called by check_auto_possessive() when a property item
2442    is adjacent to a fixed character.
2443    
2444    Arguments:
2445      c            the character
2446      ptype        the property type
2447      pdata        the data for the type
2448      negated      TRUE if it's a negated property (\P or \p{^)
2449    
2450    Returns:       TRUE if auto-possessifying is OK
2451    */
2452    
2453    static BOOL
2454    check_char_prop(int c, int ptype, int pdata, BOOL negated)
2455    {
2456    const ucd_record *prop = GET_UCD(c);
2457    switch(ptype)
2458      {
2459      case PT_LAMP:
2460      return (prop->chartype == ucp_Lu ||
2461              prop->chartype == ucp_Ll ||
2462              prop->chartype == ucp_Lt) == negated;
2463    
2464      case PT_GC:
2465      return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;
2466    
2467      case PT_PC:
2468      return (pdata == prop->chartype) == negated;
2469    
2470      case PT_SC:
2471      return (pdata == prop->script) == negated;
2472    
2473      /* These are specials */
2474    
2475      case PT_ALNUM:
2476      return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2477              _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;
2478    
2479      case PT_SPACE:    /* Perl space */
2480      return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2481              c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2482              == negated;
2483    
2484      case PT_PXSPACE:  /* POSIX space */
2485      return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2486              c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2487              c == CHAR_FF || c == CHAR_CR)
2488              == negated;
2489    
2490      case PT_WORD:
2491      return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2492              _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2493              c == CHAR_UNDERSCORE) == negated;
2494      }
2495    return FALSE;
2496    }
2497  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2498    
2499    
# Line 2269  whether the next thing could possibly ma Line 2507  whether the next thing could possibly ma
2507  sense to automatically possessify the repeated item.  sense to automatically possessify the repeated item.
2508    
2509  Arguments:  Arguments:
2510    op_code       the repeated op code    previous      pointer to the repeated opcode
   this          data for this item, depends on the opcode  
2511    utf8          TRUE in UTF-8 mode    utf8          TRUE in UTF-8 mode
   utf8_char     used for utf8 character bytes, NULL if not relevant  
2512    ptr           next character in pattern    ptr           next character in pattern
2513    options       options bits    options       options bits
2514    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2281  Returns:        TRUE if possessifying is Line 2517  Returns:        TRUE if possessifying is
2517  */  */
2518    
2519  static BOOL  static BOOL
2520  check_auto_possessive(int op_code, int item, BOOL utf8, uschar *utf8_char,  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,
2521    const uschar *ptr, int options, compile_data *cd)    int options, compile_data *cd)
2522  {  {
2523  int next;  int c, next;
2524    int op_code = *previous++;
2525    
2526  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
2527    
# Line 2295  if ((options & PCRE_EXTENDED) != 0) Line 2532  if ((options & PCRE_EXTENDED) != 0)
2532      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2533      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
2534        {        {
2535        while (*(++ptr) != 0)        ptr++;
2536          while (*ptr != 0)
2537            {
2538          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
2539            ptr++;
2540    #ifdef SUPPORT_UTF8
2541            if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
2542    #endif
2543            }
2544        }        }
2545      else break;      else break;
2546      }      }
# Line 2332  if ((options & PCRE_EXTENDED) != 0) Line 2576  if ((options & PCRE_EXTENDED) != 0)
2576      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2577      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
2578        {        {
2579        while (*(++ptr) != 0)        ptr++;
2580          while (*ptr != 0)
2581            {
2582          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
2583            ptr++;
2584    #ifdef SUPPORT_UTF8
2585            if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
2586    #endif
2587            }
2588        }        }
2589      else break;      else break;
2590      }      }
# Line 2345  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 2596  if (*ptr == CHAR_ASTERISK || *ptr == CHA
2596    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)
2597      return FALSE;      return FALSE;
2598    
2599  /* 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
2600  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. */  
2601    
2602  if (next >= 0) switch(op_code)  if (next >= 0) switch(op_code)
2603    {    {
2604    case OP_CHAR:    case OP_CHAR:
2605  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2606    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2607  #else  #else
2608    (void)(utf8_char);  /* Keep compiler happy by referencing function argument */    c = *previous;
2609  #endif  #endif
2610    return item != next;    return c != next;
2611    
2612    /* 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
2613    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 2369  if (next >= 0) switch(op_code) Line 2615  if (next >= 0) switch(op_code)
2615    
2616    case OP_CHARNC:    case OP_CHARNC:
2617  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2618    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2619    #else
2620      c = *previous;
2621  #endif  #endif
2622    if (item == next) return FALSE;    if (c == next) return FALSE;
2623  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2624    if (utf8)    if (utf8)
2625      {      {
# Line 2382  if (next >= 0) switch(op_code) Line 2630  if (next >= 0) switch(op_code)
2630  #else  #else
2631      othercase = NOTACHAR;      othercase = NOTACHAR;
2632  #endif  #endif
2633      return (unsigned int)item != othercase;      return (unsigned int)c != othercase;
2634      }      }
2635    else    else
2636  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2637    return (item != cd->fcc[next]);  /* Non-UTF-8 mode */    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */
2638    
2639    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, its data is always a single-byte character. */
2640    
2641    case OP_NOT:    case OP_NOT:
2642    if (item == next) return TRUE;    if ((c = *previous) == next) return TRUE;
2643    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2644  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2645    if (utf8)    if (utf8)
# Line 2403  if (next >= 0) switch(op_code) Line 2651  if (next >= 0) switch(op_code)
2651  #else  #else
2652      othercase = NOTACHAR;      othercase = NOTACHAR;
2653  #endif  #endif
2654      return (unsigned int)item == othercase;      return (unsigned int)c == othercase;
2655      }      }
2656    else    else
2657  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2658    return (item == cd->fcc[next]);  /* Non-UTF-8 mode */    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */
2659    
2660      /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.
2661      When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
2662    
2663    case OP_DIGIT:    case OP_DIGIT:
2664    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;
# Line 2450  if (next >= 0) switch(op_code) Line 2701  if (next >= 0) switch(op_code)
2701      case 0x202f:      case 0x202f:
2702      case 0x205f:      case 0x205f:
2703      case 0x3000:      case 0x3000:
2704      return op_code != OP_HSPACE;      return op_code == OP_NOT_HSPACE;
2705      default:      default:
2706      return op_code == OP_HSPACE;      return op_code != OP_NOT_HSPACE;
2707      }      }
2708    
2709      case OP_ANYNL:
2710    case OP_VSPACE:    case OP_VSPACE:
2711    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
2712    switch(next)    switch(next)
# Line 2466  if (next >= 0) switch(op_code) Line 2718  if (next >= 0) switch(op_code)
2718      case 0x85:      case 0x85:
2719      case 0x2028:      case 0x2028:
2720      case 0x2029:      case 0x2029:
2721      return op_code != OP_VSPACE;      return op_code == OP_NOT_VSPACE;
2722      default:      default:
2723      return op_code == OP_VSPACE;      return op_code != OP_NOT_VSPACE;
2724      }      }
2725    
2726    #ifdef SUPPORT_UCP
2727      case OP_PROP:
2728      return check_char_prop(next, previous[0], previous[1], FALSE);
2729    
2730      case OP_NOTPROP:
2731      return check_char_prop(next, previous[0], previous[1], TRUE);
2732    #endif
2733    
2734    default:    default:
2735    return FALSE;    return FALSE;
2736    }    }
2737    
2738    
2739  /* 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
2740    is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are
2741    generated only when PCRE_UCP is *not* set, that is, when only ASCII
2742    characteristics are recognized. Similarly, the opcodes OP_DIGIT etc. are
2743    replaced by OP_PROP codes when PCRE_UCP is set. */
2744    
2745  switch(op_code)  switch(op_code)
2746    {    {
2747    case OP_CHAR:    case OP_CHAR:
2748    case OP_CHARNC:    case OP_CHARNC:
2749  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2750    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2751    #else
2752      c = *previous;
2753  #endif  #endif
2754    switch(-next)    switch(-next)
2755      {      {
2756      case ESC_d:      case ESC_d:
2757      return item > 127 || (cd->ctypes[item] & ctype_digit) == 0;      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;
2758    
2759      case ESC_D:      case ESC_D:
2760      return item <= 127 && (cd->ctypes[item] & ctype_digit) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;
2761    
2762      case ESC_s:      case ESC_s:
2763      return item > 127 || (cd->ctypes[item] & ctype_space) == 0;      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;
2764    
2765      case ESC_S:      case ESC_S:
2766      return item <= 127 && (cd->ctypes[item] & ctype_space) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;
2767    
2768      case ESC_w:      case ESC_w:
2769      return item > 127 || (cd->ctypes[item] & ctype_word) == 0;      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;
2770    
2771      case ESC_W:      case ESC_W:
2772      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;
2773    
2774      case ESC_h:      case ESC_h:
2775      case ESC_H:      case ESC_H:
2776      switch(item)      switch(c)
2777        {        {
2778        case 0x09:        case 0x09:
2779        case 0x20:        case 0x20:
# Line 2535  switch(op_code) Line 2801  switch(op_code)
2801    
2802      case ESC_v:      case ESC_v:
2803      case ESC_V:      case ESC_V:
2804      switch(item)      switch(c)
2805        {        {
2806        case 0x0a:        case 0x0a:
2807        case 0x0b:        case 0x0b:
# Line 2549  switch(op_code) Line 2815  switch(op_code)
2815        return -next == ESC_v;        return -next == ESC_v;
2816        }        }
2817    
2818        /* When PCRE_UCP is set, these values get generated for \d etc. Find
2819        their substitutions and process them. The result will always be either
2820        -ESC_p or -ESC_P. Then fall through to process those values. */
2821    
2822    #ifdef SUPPORT_UCP
2823        case ESC_du:
2824        case ESC_DU:
2825        case ESC_wu:
2826        case ESC_WU:
2827        case ESC_su:
2828        case ESC_SU:
2829          {
2830          int temperrorcode = 0;
2831          ptr = substitutes[-next - ESC_DU];
2832          next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);
2833          if (temperrorcode != 0) return FALSE;
2834          ptr++;    /* For compatibility */
2835          }
2836        /* Fall through */
2837    
2838        case ESC_p:
2839        case ESC_P:
2840          {
2841          int ptype, pdata, errorcodeptr;
2842          BOOL negated;
2843    
2844          ptr--;      /* Make ptr point at the p or P */
2845          ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);
2846          if (ptype < 0) return FALSE;
2847          ptr++;      /* Point past the final curly ket */
2848    
2849          /* If the property item is optional, we have to give up. (When generated
2850          from \d etc by PCRE_UCP, this test will have been applied much earlier,
2851          to the original \d etc. At this point, ptr will point to a zero byte. */
2852    
2853          if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
2854            strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
2855              return FALSE;
2856    
2857          /* Do the property check. */
2858    
2859          return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);
2860          }
2861    #endif
2862    
2863      default:      default:
2864      return FALSE;      return FALSE;
2865      }      }
2866    
2867      /* In principle, support for Unicode properties should be integrated here as
2868      well. It means re-organizing the above code so as to get hold of the property
2869      values before switching on the op-code. However, I wonder how many patterns
2870      combine ASCII \d etc with Unicode properties? (Note that if PCRE_UCP is set,
2871      these op-codes are never generated.) */
2872    
2873    case OP_DIGIT:    case OP_DIGIT:
2874    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2875           next == -ESC_h || next == -ESC_v;           next == -ESC_h || next == -ESC_v || next == -ESC_R;
2876    
2877    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2878    return next == -ESC_d;    return next == -ESC_d;
2879    
2880    case OP_WHITESPACE:    case OP_WHITESPACE:
2881    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;
2882    
2883    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2884    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2885    
2886    case OP_HSPACE:    case OP_HSPACE:
2887    return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||
2888             next == -ESC_w || next == -ESC_v || next == -ESC_R;
2889    
2890    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
2891    return next == -ESC_h;    return next == -ESC_h;
2892    
2893    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
2894      case OP_ANYNL:
2895    case OP_VSPACE:    case OP_VSPACE:
2896    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2897    
2898    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
2899    return next == -ESC_v;    return next == -ESC_v || next == -ESC_R;
2900    
2901    case OP_WORDCHAR:    case OP_WORDCHAR:
2902    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||
2903             next == -ESC_v || next == -ESC_R;
2904    
2905    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2906    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2644  BOOL inescq = FALSE; Line 2964  BOOL inescq = FALSE;
2964  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstbyte = FALSE;
2965  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
2966  const uschar *tempptr;  const uschar *tempptr;
2967    const uschar *nestptr = NULL;
2968  uschar *previous = NULL;  uschar *previous = NULL;
2969  uschar *previous_callout = NULL;  uschar *previous_callout = NULL;
2970  uschar *save_hwm = NULL;  uschar *save_hwm = NULL;
# Line 2714  for (;; ptr++) Line 3035  for (;; ptr++)
3035    
3036    c = *ptr;    c = *ptr;
3037    
3038      /* If we are at the end of a nested substitution, revert to the outer level
3039      string. Nesting only happens one level deep. */
3040    
3041      if (c == 0 && nestptr != NULL)
3042        {
3043        ptr = nestptr;
3044        nestptr = NULL;
3045        c = *ptr;
3046        }
3047    
3048    /* 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
3049    previous cycle of this loop. */    previous cycle of this loop. */
3050    
# Line 2722  for (;; ptr++) Line 3053  for (;; ptr++)
3053  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3054      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3055  #endif  #endif
3056      if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */
3057        {        {
3058        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3059        goto FAILED;        goto FAILED;
# Line 2744  for (;; ptr++) Line 3075  for (;; ptr++)
3075        goto FAILED;        goto FAILED;
3076        }        }
3077    
3078      *lengthptr += code - last_code;      *lengthptr += (int)(code - last_code);
3079      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));
3080    
3081      /* 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 2771  for (;; ptr++) Line 3102  for (;; ptr++)
3102    /* In the real compile phase, just check the workspace used by the forward    /* In the real compile phase, just check the workspace used by the forward
3103    reference list. */    reference list. */
3104    
3105    else if (cd->hwm > cd->start_workspace + COMPILE_WORK_SIZE)    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)
3106      {      {
3107      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3108      goto FAILED;      goto FAILED;
# Line 2826  for (;; ptr++) Line 3157  for (;; ptr++)
3157      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if ((cd->ctypes[c] & ctype_space) != 0) continue;
3158      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3159        {        {
3160        while (*(++ptr) != 0)        ptr++;
3161          while (*ptr != 0)
3162          {          {
3163          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3164            ptr++;
3165    #ifdef SUPPORT_UTF8
3166            if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;
3167    #endif
3168          }          }
3169        if (*ptr != 0) continue;        if (*ptr != 0) continue;
3170    
# Line 2862  for (;; ptr++) Line 3198  for (;; ptr++)
3198          *errorcodeptr = ERR20;          *errorcodeptr = ERR20;
3199          goto FAILED;          goto FAILED;
3200          }          }
3201        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += (int)(code - last_code);   /* To include callout length */
3202        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
3203        }        }
3204      return TRUE;      return TRUE;
# Line 3067  for (;; ptr++) Line 3403  for (;; ptr++)
3403            ptr++;            ptr++;
3404            }            }
3405    
3406          posix_class = check_posix_name(ptr, tempptr - ptr);          posix_class = check_posix_name(ptr, (int)(tempptr - ptr));
3407          if (posix_class < 0)          if (posix_class < 0)
3408            {            {
3409            *errorcodeptr = ERR30;            *errorcodeptr = ERR30;
# Line 3081  for (;; ptr++) Line 3417  for (;; ptr++)
3417          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
3418            posix_class = 0;            posix_class = 0;
3419    
3420          /* 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
3421          because we may be adding and subtracting from it, and we don't want to          different escape sequences that use Unicode properties. */
3422          subtract bits that may be in the main map already. At the end we or the  
3423          result into the bit map that is being built. */  #ifdef SUPPORT_UCP
3424            if ((options & PCRE_UCP) != 0)
3425              {
3426              int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);
3427              if (posix_substitutes[pc] != NULL)
3428                {
3429                nestptr = tempptr + 1;
3430                ptr = posix_substitutes[pc] - 1;
3431                continue;
3432                }
3433              }
3434    #endif
3435            /* In the non-UCP case, we build the bit map for the POSIX class in a
3436            chunk of local store because we may be adding and subtracting from it,
3437            and we don't want to subtract bits that may be in the main map already.
3438            At the end we or the result into the bit map that is being built. */
3439    
3440          posix_class *= 3;          posix_class *= 3;
3441    
# Line 3128  for (;; ptr++) Line 3479  for (;; ptr++)
3479    
3480        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
3481        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
3482        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
3483        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
3484        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
3485        character in them, so set class_charcount bigger than one. */        are either treated as literal characters (by default), or are faulted if
3486          PCRE_EXTRA is set. */
3487    
3488        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
3489          {          {
3490          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3491          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
3492    
3493          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 */  
3494          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
3495            {            {
3496            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3157  for (;; ptr++) Line 3507  for (;; ptr++)
3507            register const uschar *cbits = cd->cbits;            register const uschar *cbits = cd->cbits;
3508            class_charcount += 2;     /* Greater than 1 is what matters */            class_charcount += 2;     /* Greater than 1 is what matters */
3509    
3510            /* Save time by not doing this in the pre-compile phase. */            switch (-c)
   
           if (lengthptr == NULL) switch (-c)  
3511              {              {
3512    #ifdef SUPPORT_UCP
3513                case ESC_du:     /* These are the values given for \d etc */
3514                case ESC_DU:     /* when PCRE_UCP is set. We replace the */
3515                case ESC_wu:     /* escape sequence with an appropriate \p */
3516                case ESC_WU:     /* or \P to test Unicode properties instead */
3517                case ESC_su:     /* of the default ASCII testing. */
3518                case ESC_SU:
3519                nestptr = ptr;
3520                ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
3521                class_charcount -= 2;                /* Undo! */
3522                continue;
3523    #endif
3524              case ESC_d:              case ESC_d:
3525              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];
3526              continue;              continue;
# Line 3179  for (;; ptr++) Line 3539  for (;; ptr++)
3539              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
3540              continue;              continue;
3541    
3542                /* Perl 5.004 onwards omits VT from \s, but we must preserve it
3543                if it was previously set by something earlier in the character
3544                class. */
3545    
3546              case ESC_s:              case ESC_s:
3547              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_space];              classbits[0] |= cbits[cbit_space];
3548              classbits[1] &= ~0x08;   /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= cbits[cbit_space+1] & ~0x08;
3549                for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];
3550              continue;              continue;
3551    
3552              case ESC_S:              case ESC_S:
# Line 3190  for (;; ptr++) Line 3555  for (;; ptr++)
3555              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
3556              continue;              continue;
3557    
3558              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)  
             {  
3559              SETBIT(classbits, 0x09); /* VT */              SETBIT(classbits, 0x09); /* VT */
3560              SETBIT(classbits, 0x20); /* SPACE */              SETBIT(classbits, 0x20); /* SPACE */
3561              SETBIT(classbits, 0xa0); /* NSBP */              SETBIT(classbits, 0xa0); /* NSBP */
# Line 3227  for (;; ptr++) Line 3579  for (;; ptr++)
3579                }                }
3580  #endif  #endif
3581              continue;              continue;
             }  
3582    
3583            if (-c == ESC_H)              case ESC_H:
             {  
3584              for (c = 0; c < 32; c++)              for (c = 0; c < 32; c++)
3585                {                {
3586                int x = 0xff;                int x = 0xff;
# Line 3272  for (;; ptr++) Line 3622  for (;; ptr++)
3622                }                }
3623  #endif  #endif
3624              continue;              continue;
             }  
3625    
3626            if (-c == ESC_v)              case ESC_v:
             {  
3627              SETBIT(classbits, 0x0a); /* LF */              SETBIT(classbits, 0x0a); /* LF */
3628              SETBIT(classbits, 0x0b); /* VT */              SETBIT(classbits, 0x0b); /* VT */
3629              SETBIT(classbits, 0x0c); /* FF */              SETBIT(classbits, 0x0c); /* FF */
# Line 3291  for (;; ptr++) Line 3639  for (;; ptr++)
3639                }                }
3640  #endif  #endif
3641              continue;              continue;
             }  
3642    
3643            if (-c == ESC_V)              case ESC_V:
             {  
3644              for (c = 0; c < 32; c++)              for (c = 0; c < 32; c++)
3645                {                {
3646                int x = 0xff;                int x = 0xff;
# Line 3324  for (;; ptr++) Line 3670  for (;; ptr++)
3670                }                }
3671  #endif  #endif
3672              continue;              continue;
             }  
   
           /* We need to deal with \P and \p in both phases. */  
3673    
3674  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3675            if (-c == ESC_p || -c == ESC_P)              case ESC_p:
3676              {              case ESC_P:
3677              BOOL negated;                {
3678              int pdata;                BOOL negated;
3679              int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int pdata;
3680              if (ptype < 0) goto FAILED;                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
3681              class_utf8 = TRUE;                if (ptype < 0) goto FAILED;
3682              *class_utf8data++ = ((-c == ESC_p) != negated)?                class_utf8 = TRUE;
3683                XCL_PROP : XCL_NOTPROP;                *class_utf8data++ = ((-c == ESC_p) != negated)?
3684              *class_utf8data++ = ptype;                  XCL_PROP : XCL_NOTPROP;
3685              *class_utf8data++ = pdata;                *class_utf8data++ = ptype;
3686              class_charcount -= 2;   /* Not a < 256 character */                *class_utf8data++ = pdata;
3687              continue;                class_charcount -= 2;   /* Not a < 256 character */
3688              }                continue;
3689                  }
3690  #endif  #endif
3691            /* Unrecognized escapes are faulted if PCRE is running in its              /* Unrecognized escapes are faulted if PCRE is running in its
3692            strict mode. By default, for compatibility with Perl, they are              strict mode. By default, for compatibility with Perl, they are
3693            treated as literals. */              treated as literals. */
3694    
3695            if ((options & PCRE_EXTRA) != 0)              default:
3696              {              if ((options & PCRE_EXTRA) != 0)
3697              *errorcodeptr = ERR7;                {
3698              goto FAILED;                *errorcodeptr = ERR7;
3699                  goto FAILED;
3700                  }
3701                class_charcount -= 2;  /* Undo the default count from above */
3702                c = *ptr;              /* Get the final character and fall through */
3703                break;
3704              }              }
   
           class_charcount -= 2;  /* Undo the default count from above */  
           c = *ptr;              /* Get the final character and fall through */  
3705            }            }
3706    
3707          /* 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 3425  for (;; ptr++) Line 3771  for (;; ptr++)
3771            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3772            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3773    
3774            /* \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 */  
3775    
3776            if (d < 0)            if (d < 0)
3777              {              {
3778              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  
3779                {                {
3780                ptr = oldptr;                ptr = oldptr;
3781                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto LONE_SINGLE_CHARACTER;  /* A few lines below */
# Line 3598  for (;; ptr++) Line 3941  for (;; ptr++)
3941          }          }
3942        }        }
3943    
3944      /* 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.
3945        If we are at the end of an internal nested string, revert to the outer
3946        string. */
3947    
3948        while (((c = *(++ptr)) != 0 ||
3949               (nestptr != NULL &&
3950                 (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&
3951               (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
3952    
3953      while ((c = *(++ptr)) != 0 && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));      /* Check for missing terminating ']' */
3954    
3955      if (c == 0)                          /* Missing terminating ']' */      if (c == 0)
3956        {        {
3957        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
3958        goto FAILED;        goto FAILED;
3959        }        }
3960    
   
 /* 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  
   
   
3961      /* 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
3962      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
3963      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 3690  we set the flag only if there is a liter Line 4021  we set the flag only if there is a liter
4021    
4022      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4023      extended class, with its own opcode, unless there was a negated special      extended class, with its own opcode, unless there was a negated special
4024      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
4025      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
4026      (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
4027      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
4028        actual compiled code. */
4029    
4030  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
4031      if (class_utf8 && !should_flip_negation)      if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))
4032        {        {
4033        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
4034        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3722  we set the flag only if there is a liter Line 4054  we set the flag only if there is a liter
4054        }        }
4055  #endif  #endif
4056    
4057      /* 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
4058      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
4059      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
4060      map into the code vector, negating it if necessary. */      (non-UCP) in the class. Then copy the 32-byte map into the code vector,
4061        negating it if necessary. */
4062    
4063      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
4064      if (negate_class)      if (negate_class)
# Line 3849  we set the flag only if there is a liter Line 4182  we set the flag only if there is a liter
4182    
4183        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4184            repeat_max < 0 &&            repeat_max < 0 &&
4185            check_auto_possessive(*previous, c, utf8, utf8_char, ptr + 1,            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
             options, cd))  
4186          {          {
4187          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4188          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 3871  we set the flag only if there is a liter Line 4203  we set the flag only if there is a liter
4203        c = previous[1];        c = previous[1];
4204        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4205            repeat_max < 0 &&            repeat_max < 0 &&
4206            check_auto_possessive(OP_NOT, c, utf8, NULL, ptr + 1, options, cd))            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
4207          {          {
4208          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4209          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 3895  we set the flag only if there is a liter Line 4227  we set the flag only if there is a liter
4227    
4228        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4229            repeat_max < 0 &&            repeat_max < 0 &&
4230            check_auto_possessive(c, 0, utf8, NULL, ptr + 1, options, cd))            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
4231          {          {
4232          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4233          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 4105  we set the flag only if there is a liter Line 4437  we set the flag only if there is a liter
4437        {        {
4438        register int i;        register int i;
4439        int ketoffset = 0;        int ketoffset = 0;
4440        int len = code - previous;        int len = (int)(code - previous);
4441        uschar *bralink = NULL;        uschar *bralink = NULL;
4442    
4443        /* Repeating a DEFINE group is pointless */        /* Repeating a DEFINE group is pointless */
# Line 4126  we set the flag only if there is a liter Line 4458  we set the flag only if there is a liter
4458          {          {
4459          register uschar *ket = previous;          register uschar *ket = previous;
4460          do ket += GET(ket, 1); while (*ket != OP_KET);          do ket += GET(ket, 1); while (*ket != OP_KET);
4461          ketoffset = code - ket;          ketoffset = (int)(code - ket);
4462          }          }
4463    
4464        /* 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 4194  we set the flag only if there is a liter Line 4526  we set the flag only if there is a liter
4526            /* We chain together the bracket offset fields that have to be            /* We chain together the bracket offset fields that have to be
4527            filled in later when the ends of the brackets are reached. */            filled in later when the ends of the brackets are reached. */
4528    
4529            offset = (bralink == NULL)? 0 : previous - bralink;            offset = (bralink == NULL)? 0 : (int)(previous - bralink);
4530            bralink = previous;            bralink = previous;
4531            PUTINC(previous, 0, offset);            PUTINC(previous, 0, offset);
4532            }            }
# Line 4303  we set the flag only if there is a liter Line 4635  we set the flag only if there is a liter
4635              {              {
4636              int offset;              int offset;
4637              *code++ = OP_BRA;              *code++ = OP_BRA;
4638              offset = (bralink == NULL)? 0 : code - bralink;              offset = (bralink == NULL)? 0 : (int)(code - bralink);
4639              bralink = code;              bralink = code;
4640              PUTINC(code, 0, offset);              PUTINC(code, 0, offset);
4641              }              }
# Line 4324  we set the flag only if there is a liter Line 4656  we set the flag only if there is a liter
4656          while (bralink != NULL)          while (bralink != NULL)
4657            {            {
4658            int oldlinkoffset;            int oldlinkoffset;
4659            int offset = code - bralink + 1;            int offset = (int)(code - bralink + 1);
4660            uschar *bra = code - offset;            uschar *bra = code - offset;
4661            oldlinkoffset = GET(bra, 1);            oldlinkoffset = GET(bra, 1);
4662            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;
# Line 4355  we set the flag only if there is a liter Line 4687  we set the flag only if there is a liter
4687            uschar *scode = bracode;            uschar *scode = bracode;
4688            do            do
4689              {              {
4690              if (could_be_empty_branch(scode, ketcode, utf8))              if (could_be_empty_branch(scode, ketcode, utf8, cd))
4691                {                {
4692                *bracode += OP_SBRA - OP_BRA;                *bracode += OP_SBRA - OP_BRA;
4693                break;                break;
# Line 4412  we set the flag only if there is a liter Line 4744  we set the flag only if there is a liter
4744  #endif  #endif
4745          }          }
4746    
4747        len = code - tempcode;        len = (int)(code - tempcode);
4748        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4749          {          {
4750          case OP_STAR:  *tempcode = OP_POSSTAR; break;          case OP_STAR:  *tempcode = OP_POSSTAR; break;
# Line 4432  we set the flag only if there is a liter Line 4764  we set the flag only if there is a liter
4764    
4765          /* Because we are moving code along, we must ensure that any          /* Because we are moving code along, we must ensure that any
4766          pending recursive references are updated. */          pending recursive references are updated. */
4767    
4768          default:          default:
4769          *code = OP_END;          *code = OP_END;
4770          adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);          adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);
# Line 4471  we set the flag only if there is a liter Line 4803  we set the flag only if there is a liter
4803    
4804      /* First deal with various "verbs" that can be introduced by '*'. */      /* First deal with various "verbs" that can be introduced by '*'. */
4805    
4806      if (*(++ptr) == CHAR_ASTERISK && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == CHAR_ASTERISK &&
4807             ((cd->ctypes[ptr[1]] & ctype_letter) != 0 || ptr[1] == ':'))
4808        {        {
4809        int i, namelen;        int i, namelen;
4810          int arglen = 0;
4811        const char *vn = verbnames;        const char *vn = verbnames;
4812        const uschar *name = ++ptr;        const uschar *name = ptr + 1;
4813          const uschar *arg = NULL;
4814        previous = NULL;        previous = NULL;
4815        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
4816          namelen = (int)(ptr - name);
4817    
4818        if (*ptr == CHAR_COLON)        if (*ptr == CHAR_COLON)
4819          {          {
4820          *errorcodeptr = ERR59;   /* Not supported */          arg = ++ptr;
4821          goto FAILED;          while ((cd->ctypes[*ptr] & (ctype_letter|ctype_digit)) != 0
4822              || *ptr == '_') ptr++;
4823            arglen = (int)(ptr - arg);
4824          }          }
4825    
4826        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
4827          {          {
4828          *errorcodeptr = ERR60;          *errorcodeptr = ERR60;
4829          goto FAILED;          goto FAILED;
4830          }          }
4831        namelen = ptr - name;  
4832          /* Scan the table of verb names */
4833    
4834        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4835          {          {
4836          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
# Line 4506  we set the flag only if there is a liter Line 4848  we set the flag only if there is a liter
4848                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
4849                }                }
4850              }              }
4851            *code++ = verbs[i].op;  
4852            break;            /* Handle the cases with/without an argument */
4853    
4854              if (arglen == 0)
4855                {
4856                if (verbs[i].op < 0)   /* Argument is mandatory */
4857                  {
4858                  *errorcodeptr = ERR66;
4859                  goto FAILED;
4860                  }
4861                *code = verbs[i].op;
4862                if (*code++ == OP_THEN)
4863                  {
4864                  PUT(code, 0, code - bcptr->current_branch - 1);
4865                  code += LINK_SIZE;
4866                  }
4867                }
4868    
4869              else
4870                {
4871                if (verbs[i].op_arg < 0)   /* Argument is forbidden */
4872                  {
4873                  *errorcodeptr = ERR59;
4874                  goto FAILED;
4875                  }
4876                *code = verbs[i].op_arg;
4877                if (*code++ == OP_THEN_ARG)
4878                  {
4879                  PUT(code, 0, code - bcptr->current_branch - 1);
4880                  code += LINK_SIZE;
4881                  }
4882                *code++ = arglen;
4883                memcpy(code, arg, arglen);
4884                code += arglen;
4885                *code++ = 0;
4886                }
4887    
4888              break;  /* Found verb, exit loop */
4889            }            }
4890    
4891          vn += verbs[i].len + 1;          vn += verbs[i].len + 1;
4892          }          }
4893        if (i < verbcount) continue;  
4894        *errorcodeptr = ERR60;        if (i < verbcount) continue;    /* Successfully handled a verb */
4895          *errorcodeptr = ERR60;          /* Verb not recognized */
4896        goto FAILED;        goto FAILED;
4897        }        }
4898    
# Line 4631  we set the flag only if there is a liter Line 5011  we set the flag only if there is a liter
5011                recno * 10 + *ptr - CHAR_0 : -1;                recno * 10 + *ptr - CHAR_0 : -1;
5012            ptr++;            ptr++;
5013            }            }
5014          namelen = ptr - name;          namelen = (int)(ptr - name);
5015    
5016          if ((terminator > 0 && *ptr++ != terminator) ||          if ((terminator > 0 && *ptr++ != terminator) ||
5017              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
# Line 4692  we set the flag only if there is a liter Line 5072  we set the flag only if there is a liter
5072          /* Search the pattern for a forward reference */          /* Search the pattern for a forward reference */
5073    
5074          else if ((i = find_parens(cd, name, namelen,          else if ((i = find_parens(cd, name, namelen,
5075                          (options & PCRE_EXTENDED) != 0)) > 0)                          (options & PCRE_EXTENDED) != 0, utf8)) > 0)
5076            {            {
5077            PUT2(code, 2+LINK_SIZE, i);            PUT2(code, 2+LINK_SIZE, i);
5078            code[1+LINK_SIZE]++;            code[1+LINK_SIZE]++;
# Line 4827  we set the flag only if there is a liter Line 5207  we set the flag only if there is a liter
5207              goto FAILED;              goto FAILED;
5208              }              }
5209            *code++ = n;            *code++ = n;
5210            PUT(code, 0, ptr - cd->start_pattern + 1);  /* Pattern offset */            PUT(code, 0, (int)(ptr - cd->start_pattern + 1)); /* Pattern offset */
5211            PUT(code, LINK_SIZE, 0);                    /* Default length */            PUT(code, LINK_SIZE, 0);                          /* Default length */
5212            code += 2 * LINK_SIZE;            code += 2 * LINK_SIZE;
5213            }            }
5214          previous = NULL;          previous = NULL;
# Line 4861  we set the flag only if there is a liter Line 5241  we set the flag only if there is a liter
5241            name = ++ptr;            name = ++ptr;
5242    
5243            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
5244            namelen = ptr - name;            namelen = (int)(ptr - name);
5245    
5246            /* In the pre-compile phase, just do a syntax check. */            /* In the pre-compile phase, just do a syntax check. */
5247    
# Line 4991  we set the flag only if there is a liter Line 5371  we set the flag only if there is a liter
5371          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
5372          name = ++ptr;          name = ++ptr;
5373          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
5374          namelen = ptr - name;          namelen = (int)(ptr - name);
5375    
5376          /* In the pre-compile phase, do a syntax check and set a dummy          /* In the pre-compile phase, do a syntax check and set a dummy
5377          reference number. */          reference number. */
# Line 5038  we set the flag only if there is a liter Line 5418  we set the flag only if there is a liter
5418              }              }
5419            else if ((recno =                /* Forward back reference */            else if ((recno =                /* Forward back reference */
5420                      find_parens(cd, name, namelen,                      find_parens(cd, name, namelen,
5421                        (options & PCRE_EXTENDED) != 0)) <= 0)                        (options & PCRE_EXTENDED) != 0, utf8)) <= 0)
5422              {              {
5423              *errorcodeptr = ERR15;              *errorcodeptr = ERR15;
5424              goto FAILED;              goto FAILED;
# Line 5149  we set the flag only if there is a liter Line 5529  we set the flag only if there is a liter
5529              if (called == NULL)              if (called == NULL)
5530                {                {
5531                if (find_parens(cd, NULL, recno,                if (find_parens(cd, NULL, recno,
5532                      (options & PCRE_EXTENDED) != 0) < 0)                      (options & PCRE_EXTENDED) != 0, utf8) < 0)
5533                  {                  {
5534                  *errorcodeptr = ERR15;                  *errorcodeptr = ERR15;
5535                  goto FAILED;                  goto FAILED;
5536                  }                  }
5537    
5538                /* 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
5539                offset below, what it actually inserted is the reference number                offset below, what it actually inserted is the reference number
5540                of the group. */                of the group. */
5541    
5542                called = cd->start_code + recno;                called = cd->start_code + recno;
5543                PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code);                PUTINC(cd->hwm, 0, (int)(code + 2 + LINK_SIZE - cd->start_code));
5544                }                }
5545    
5546              /* If not a forward reference, and the subpattern is still open,              /* If not a forward reference, and the subpattern is still open,
# Line 5168  we set the flag only if there is a liter Line 5548  we set the flag only if there is a liter
5548              recursion that could loop for ever, and diagnose that case. */              recursion that could loop for ever, and diagnose that case. */
5549    
5550              else if (GET(called, 1) == 0 &&              else if (GET(called, 1) == 0 &&
5551                       could_be_empty(called, code, bcptr, utf8))                       could_be_empty(called, code, bcptr, utf8, cd))
5552                {                {
5553                *errorcodeptr = ERR40;                *errorcodeptr = ERR40;
5554                goto FAILED;                goto FAILED;
# Line 5184  we set the flag only if there is a liter Line 5564  we set the flag only if there is a liter
5564            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
5565    
5566            *code = OP_RECURSE;            *code = OP_RECURSE;
5567            PUT(code, 1, called - cd->start_code);            PUT(code, 1, (int)(called - cd->start_code));
5568            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
5569    
5570            *code = OP_KET;            *code = OP_KET;
# Line 5295  we set the flag only if there is a liter Line 5675  we set the flag only if there is a liter
5675          }     /* End of switch for character following (? */          }     /* End of switch for character following (? */
5676        }       /* End of (? handling */        }       /* End of (? handling */
5677    
5678      /* Opening parenthesis not followed by '?'. If PCRE_NO_AUTO_CAPTURE is set,      /* Opening parenthesis not followed by '*' or '?'. If PCRE_NO_AUTO_CAPTURE
5679      all unadorned brackets become non-capturing and behave like (?:...)      is set, all unadorned brackets become non-capturing and behave like (?:...)
5680      brackets. */      brackets. */
5681    
5682      else if ((options & PCRE_NO_AUTO_CAPTURE) != 0)      else if ((options & PCRE_NO_AUTO_CAPTURE) != 0)
# Line 5488  we set the flag only if there is a liter Line 5868  we set the flag only if there is a liter
5868    
5869      /* ===================================================================*/      /* ===================================================================*/
5870      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
5871      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
5872      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
5873      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
5874      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
5875      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
5876        ever created. */
5877    
5878      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
5879      tempptr = ptr;      tempptr = ptr;
# Line 5652  we set the flag only if there is a liter Line 6033  we set the flag only if there is a liter
6033  #endif  #endif
6034    
6035        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6036        can obtain the OP value by negating the escape value. */        can obtain the OP value by negating the escape value in the default
6037          situation when PCRE_UCP is not set. When it *is* set, we substitute
6038          Unicode property tests. */
6039    
6040        else        else
6041          {          {
6042          previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;  #ifdef SUPPORT_UCP
6043          *code++ = -c;          if (-c >= ESC_DU && -c <= ESC_wu)
6044              {
6045              nestptr = ptr + 1;                   /* Where to resume */
6046              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
6047              }
6048            else
6049    #endif
6050              {
6051              previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;
6052              *code++ = -c;
6053              }
6054          }          }
6055        continue;        continue;
6056        }        }
# Line 5989  for (;;) Line 6382  for (;;)
6382      {      {
6383      if (lengthptr == NULL)      if (lengthptr == NULL)
6384        {        {
6385        int branch_length = code - last_branch;        int branch_length = (int)(code - last_branch);
6386        do        do
6387          {          {
6388          int prev_length = GET(last_branch, 1);          int prev_length = GET(last_branch, 1);
# Line 6003  for (;;) Line 6396  for (;;)
6396      /* Fill in the ket */      /* Fill in the ket */
6397    
6398      *code = OP_KET;      *code = OP_KET;
6399      PUT(code, 1, code - start_bracket);      PUT(code, 1, (int)(code - start_bracket));
6400      code += 1 + LINK_SIZE;      code += 1 + LINK_SIZE;
6401    
6402      /* 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 6018  for (;;) Line 6411  for (;;)
6411            code - start_bracket);            code - start_bracket);
6412          *start_bracket = OP_ONCE;          *start_bracket = OP_ONCE;
6413          code += 1 + LINK_SIZE;          code += 1 + LINK_SIZE;
6414          PUT(start_bracket, 1, code - start_bracket);          PUT(start_bracket, 1, (int)(code - start_bracket));
6415          *code = OP_KET;          *code = OP_KET;
6416          PUT(code, 1, code - start_bracket);          PUT(code, 1, (int)(code - start_bracket));
6417          code += 1 + LINK_SIZE;          code += 1 + LINK_SIZE;
6418          length += 2 + 2*LINK_SIZE;          length += 2 + 2*LINK_SIZE;
6419          }          }
# Line 6075  for (;;) Line 6468  for (;;)
6468    else    else
6469      {      {
6470      *code = OP_ALT;      *code = OP_ALT;
6471      PUT(code, 1, code - last_branch);      PUT(code, 1, (int)(code - last_branch));
6472      bc.current_branch = last_branch = code;      bc.current_branch = last_branch = code;
6473      code += 1 + LINK_SIZE;      code += 1 + LINK_SIZE;
6474      }      }
# Line 6394  int length = 1;  /* For final END opcode Line 6787  int length = 1;  /* For final END opcode
6787  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
6788  int errorcode = 0;  int errorcode = 0;
6789  int skipatstart = 0;  int skipatstart = 0;
6790  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8;
6791  size_t size;  size_t size;
6792  uschar *code;  uschar *code;
6793  const uschar *codestart;  const uschar *codestart;
# Line 6464  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6857  while (ptr[skipatstart] == CHAR_LEFT_PAR
6857    
6858    if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)
6859      { skipatstart += 7; options |= PCRE_UTF8; continue; }      { skipatstart += 7; options |= PCRE_UTF8; continue; }
6860      else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0)
6861        { skipatstart += 6; options |= PCRE_UCP; continue; }
6862    
6863    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)
6864      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
# Line 6488  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6883  while (ptr[skipatstart] == CHAR_LEFT_PAR
6883    else break;    else break;
6884    }    }
6885    
6886    utf8 = (options & PCRE_UTF8) != 0;
6887    
6888  /* 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. */
6889    
6890  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 6505  if (utf8) Line 6902  if (utf8)
6902    }    }
6903  #endif  #endif
6904    
6905    /* Can't support UCP unless PCRE has been compiled to include the code. */
6906    
6907    #ifndef SUPPORT_UCP
6908    if ((options & PCRE_UCP) != 0)
6909      {
6910      errorcode = ERR67;
6911      goto PCRE_EARLY_ERROR_RETURN;
6912      }
6913    #endif
6914    
6915  /* Check validity of \R options. */  /* Check validity of \R options. */
6916    
6917  switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))  switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
# Line 6633  regex compiled on a system with 4-byte p Line 7040  regex compiled on a system with 4-byte p
7040  pointers. */  pointers. */
7041    
7042  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
7043  re->size = size;  re->size = (int)size;
7044  re->options = cd->external_options;  re->options = cd->external_options;
7045  re->flags = cd->external_flags;  re->flags = cd->external_flags;
7046  re->dummy1 = 0;  re->dummy1 = 0;
# Line 6704  while (errorcode == 0 && cd->hwm > cwork Line 7111  while (errorcode == 0 && cd->hwm > cwork
7111    recno = GET(codestart, offset);    recno = GET(codestart, offset);
7112    groupptr = _pcre_find_bracket(codestart, utf8, recno);    groupptr = _pcre_find_bracket(codestart, utf8, recno);
7113    if (groupptr == NULL) errorcode = ERR53;    if (groupptr == NULL) errorcode = ERR53;
7114      else PUT(((uschar *)codestart), offset, groupptr - codestart);      else PUT(((uschar *)codestart), offset, (int)(groupptr - codestart));
7115    }    }
7116    
7117  /* 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 6759  if (errorcode != 0) Line 7166  if (errorcode != 0)
7166    {    {
7167    (pcre_free)(re);    (pcre_free)(re);
7168    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
7169    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = (int)(ptr - (const uschar *)pattern);
7170    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
7171    *errorptr = find_error_text(errorcode);    *errorptr = find_error_text(errorcode);
7172    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;

Legend:
Removed from v.496  
changed lines
  Added in v.574

  ViewVC Help
Powered by ViewVC 1.1.5