/[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 472 by ph10, Fri Dec 11 16:42:50 2009 UTC revision 545 by ph10, Wed Jun 16 10:51:15 2010 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2009 University of Cambridge             Copyright (c) 1997-2010 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When DEBUG is defined, we need the pcre_printint() function, which is also  /* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is
57  used by pcretest. DEBUG is not defined when building a production library. */  also used by pcretest. PCRE_DEBUG is not defined when building a production
58    library. */
59    
60  #ifdef DEBUG  #ifdef PCRE_DEBUG
61  #include "pcre_printint.src"  #include "pcre_printint.src"
62  #endif  #endif
63    
# Line 91  is 4 there is plenty of room. */ Line 92  is 4 there is plenty of room. */
92    
93  #define COMPILE_WORK_SIZE (4096)  #define COMPILE_WORK_SIZE (4096)
94    
95    /* The overrun tests check for a slightly smaller size so that they detect the
96    overrun before it actually does run off the end of the data block. */
97    
98    #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100)
99    
100    
101  /* Table for handling escaped characters in the range '0'-'z'. Positive returns  /* Table for handling escaped characters in the range '0'-'z'. Positive returns
102  are simple data values; negative values are for special things like \d and so  are simple data values; negative values are for special things like \d and so
# Line 118  static const short int escapes[] = { Line 124  static const short int escapes[] = {
124       -ESC_H,                  0,       -ESC_H,                  0,
125       0,                       -ESC_K,       0,                       -ESC_K,
126       0,                       0,       0,                       0,
127       0,                       0,       -ESC_N,                  0,
128       -ESC_P,                  -ESC_Q,       -ESC_P,                  -ESC_Q,
129       -ESC_R,                  -ESC_S,       -ESC_R,                  -ESC_S,
130       0,                       0,       0,                       0,
# Line 165  static const short int escapes[] = { Line 171  static const short int escapes[] = {
171  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',
172  /*  C0 */   '{',-ESC_A, -ESC_B,  -ESC_C, -ESC_D,-ESC_E,      0, -ESC_G,  /*  C0 */   '{',-ESC_A, -ESC_B,  -ESC_C, -ESC_D,-ESC_E,      0, -ESC_G,
173  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,
174  /*  D0 */   '}',     0, -ESC_K,       0,      0,     0,      0, -ESC_P,  /*  D0 */   '}',     0, -ESC_K,       0,      0,-ESC_N,      0, -ESC_P,
175  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,
176  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,
177  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,
# Line 182  string is built from string macros so th Line 188  string is built from string macros so th
188  platforms. */  platforms. */
189    
190  typedef struct verbitem {  typedef struct verbitem {
191    int   len;    int   len;                 /* Length of verb name */
192    int   op;    int   op;                  /* Op when no arg, or -1 if arg mandatory */
193      int   op_arg;              /* Op when arg present, or -1 if not allowed */
194  } verbitem;  } verbitem;
195    
196  static const char verbnames[] =  static const char verbnames[] =
197      "\0"                       /* Empty name is a shorthand for MARK */
198      STRING_MARK0
199    STRING_ACCEPT0    STRING_ACCEPT0
200    STRING_COMMIT0    STRING_COMMIT0
201    STRING_F0    STRING_F0
# Line 196  static const char verbnames[] = Line 205  static const char verbnames[] =
205    STRING_THEN;    STRING_THEN;
206    
207  static const verbitem verbs[] = {  static const verbitem verbs[] = {
208    { 6, OP_ACCEPT },    { 0, -1,        OP_MARK },
209    { 6, OP_COMMIT },    { 4, -1,        OP_MARK },
210    { 1, OP_FAIL },    { 6, OP_ACCEPT, -1 },
211    { 4, OP_FAIL },    { 6, OP_COMMIT, -1 },
212    { 5, OP_PRUNE },    { 1, OP_FAIL,   -1 },
213    { 4, OP_SKIP  },    { 4, OP_FAIL,   -1 },
214    { 4, OP_THEN  }    { 5, OP_PRUNE,  OP_PRUNE_ARG },
215      { 4, OP_SKIP,   OP_SKIP_ARG  },
216      { 4, OP_THEN,   OP_THEN_ARG  }
217  };  };
218    
219  static const int verbcount = sizeof(verbs)/sizeof(verbitem);  static const int verbcount = sizeof(verbs)/sizeof(verbitem);
# Line 250  static const int posix_class_maps[] = { Line 261  static const int posix_class_maps[] = {
261    cbit_xdigit,-1,          0              /* xdigit */    cbit_xdigit,-1,          0              /* xdigit */
262  };  };
263    
264    /* Table of substitutes for \d etc when PCRE_UCP is set. The POSIX class
265    substitutes must be in the order of the names, defined above, and there are
266    both positive and negative cases. NULL means no substitute. */
267    
268    #ifdef SUPPORT_UCP
269    static const uschar *substitutes[] = {
270      (uschar *)"\\P{Nd}",    /* \D */
271      (uschar *)"\\p{Nd}",    /* \d */
272      (uschar *)"\\P{Xsp}",   /* \S */       /* NOTE: Xsp is Perl space */
273      (uschar *)"\\p{Xsp}",   /* \s */
274      (uschar *)"\\P{Xwd}",   /* \W */
275      (uschar *)"\\p{Xwd}"    /* \w */
276    };
277    
278    static const uschar *posix_substitutes[] = {
279      (uschar *)"\\p{L}",     /* alpha */
280      (uschar *)"\\p{Ll}",    /* lower */
281      (uschar *)"\\p{Lu}",    /* upper */
282      (uschar *)"\\p{Xan}",   /* alnum */
283      NULL,                   /* ascii */
284      (uschar *)"\\h",        /* blank */
285      NULL,                   /* cntrl */
286      (uschar *)"\\p{Nd}",    /* digit */
287      NULL,                   /* graph */
288      NULL,                   /* print */
289      NULL,                   /* punct */
290      (uschar *)"\\p{Xps}",   /* space */    /* NOTE: Xps is POSIX space */
291      (uschar *)"\\p{Xwd}",   /* word */
292      NULL,                   /* xdigit */
293      /* Negated cases */
294      (uschar *)"\\P{L}",     /* ^alpha */
295      (uschar *)"\\P{Ll}",    /* ^lower */
296      (uschar *)"\\P{Lu}",    /* ^upper */
297      (uschar *)"\\P{Xan}",   /* ^alnum */
298      NULL,                   /* ^ascii */
299      (uschar *)"\\H",        /* ^blank */
300      NULL,                   /* ^cntrl */
301      (uschar *)"\\P{Nd}",    /* ^digit */
302      NULL,                   /* ^graph */
303      NULL,                   /* ^print */
304      NULL,                   /* ^punct */
305      (uschar *)"\\P{Xps}",   /* ^space */   /* NOTE: Xps is POSIX space */
306      (uschar *)"\\P{Xwd}",   /* ^word */
307      NULL                    /* ^xdigit */
308    };
309    #define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))
310    #endif
311    
312  #define STRING(a)  # a  #define STRING(a)  # a
313  #define XSTRING(s) STRING(s)  #define XSTRING(s) STRING(s)
# Line 262  the number of relocations needed when a Line 320  the number of relocations needed when a
320  it is now one long string. We cannot use a table of offsets, because the  it is now one long string. We cannot use a table of offsets, because the
321  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we
322  simply count through to the one we want - this isn't a performance issue  simply count through to the one we want - this isn't a performance issue
323  because these strings are used only when there is a compilation error. */  because these strings are used only when there is a compilation error.
324    
325    Each substring ends with \0 to insert a null character. This includes the final
326    substring, so that the whole string ends with \0\0, which can be detected when
327    counting through. */
328    
329  static const char error_texts[] =  static const char error_texts[] =
330    "no error\0"    "no error\0"
# Line 309  static const char error_texts[] = Line 371  static const char error_texts[] =
371    /* 35 */    /* 35 */
372    "invalid condition (?(0)\0"    "invalid condition (?(0)\0"
373    "\\C not allowed in lookbehind assertion\0"    "\\C not allowed in lookbehind assertion\0"
374    "PCRE does not support \\L, \\l, \\N, \\U, or \\u\0"    "PCRE does not support \\L, \\l, \\N{name}, \\U, or \\u\0"
375    "number after (?C is > 255\0"    "number after (?C is > 255\0"
376    "closing ) for (?C expected\0"    "closing ) for (?C expected\0"
377    /* 40 */    /* 40 */
# Line 335  static const char error_texts[] = Line 397  static const char error_texts[] =
397    "inconsistent NEWLINE options\0"    "inconsistent NEWLINE options\0"
398    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"
399    "a numbered reference must not be zero\0"    "a numbered reference must not be zero\0"
400    "(*VERB) with an argument is not supported\0"    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"
401    /* 60 */    /* 60 */
402    "(*VERB) not recognized\0"    "(*VERB) not recognized\0"
403    "number is too big\0"    "number is too big\0"
# Line 343  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      ;
412    
413  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
414  patterns. Note that the tables in chartables are dependent on the locale, and  patterns. Note that the tables in chartables are dependent on the locale, and
# Line 502  static const char * Line 566  static const char *
566  find_error_text(int n)  find_error_text(int n)
567  {  {
568  const char *s = error_texts;  const char *s = error_texts;
569  for (; n > 0; n--) while (*s++ != 0) {};  for (; n > 0; n--)
570      {
571      while (*s++ != 0) {};
572      if (*s == 0) return "Error text not found (please report)";
573      }
574  return s;  return s;
575  }  }
576    
# Line 573  else Line 641  else
641    
642      case CHAR_l:      case CHAR_l:
643      case CHAR_L:      case CHAR_L:
     case CHAR_N:  
644      case CHAR_u:      case CHAR_u:
645      case CHAR_U:      case CHAR_U:
646      *errorcodeptr = ERR37;      *errorcodeptr = ERR37;
# Line 811  else Line 878  else
878      }      }
879    }    }
880    
881    /* Perl supports \N{name} for character names, as well as plain \N for "not
882    newline". PCRE does not support \N{name}. */
883    
884    if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET)
885      *errorcodeptr = ERR37;
886    
887    /* If PCRE_UCP is set, we change the values for \d etc. */
888    
889    if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)
890      c -= (ESC_DU - ESC_D);
891    
892    /* Set the pointer to the final character before returning. */
893    
894  *ptrptr = ptr;  *ptrptr = ptr;
895  return c;  return c;
896  }  }
# Line 1049  dealing with. The very first call may no Line 1129  dealing with. The very first call may no
1129    
1130  if (ptr[0] == CHAR_LEFT_PARENTHESIS)  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1131    {    {
1132    if (ptr[1] == CHAR_QUESTION_MARK &&    /* Handle specials such as (*SKIP) or (*UTF8) etc. */
1133        ptr[2] == CHAR_VERTICAL_LINE)  
1134      if (ptr[1] == CHAR_ASTERISK) ptr += 2;
1135    
1136      /* Handle a normal, unnamed capturing parenthesis. */
1137    
1138      else if (ptr[1] != CHAR_QUESTION_MARK)
1139        {
1140        *count += 1;
1141        if (name == NULL && *count == lorn) return *count;
1142        ptr++;
1143        }
1144    
1145      /* All cases now have (? at the start. Remember when we are in a group
1146      where the parenthesis numbers are duplicated. */
1147    
1148      else if (ptr[2] == CHAR_VERTICAL_LINE)
1149      {      {
1150      ptr += 3;      ptr += 3;
1151      dup_parens = TRUE;      dup_parens = TRUE;
1152      }      }
1153    
1154    /* Handle a normal, unnamed capturing parenthesis */    /* Handle comments; all characters are allowed until a ket is reached. */
1155    
1156    else if (ptr[1] != CHAR_QUESTION_MARK && ptr[1] != CHAR_ASTERISK)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1157      {      {
1158      *count += 1;      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1159      if (name == NULL && *count == lorn) return *count;      goto FAIL_EXIT;
     ptr++;  
1160      }      }
1161    
1162    /* 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
1163    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
1164    condition (there can't be any nested parens. */    condition (there can't be any nested parens). */
1165    
1166    else if (ptr[2] == CHAR_LEFT_PARENTHESIS)    else if (ptr[2] == CHAR_LEFT_PARENTHESIS)
1167      {      {
# Line 1079  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1173  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1173        }        }
1174      }      }
1175    
1176    /* We have either (? or (* and not a condition */    /* Start with (? but not a condition. */
1177    
1178    else    else
1179      {      {
# Line 1201  for (; *ptr != 0; ptr++) Line 1295  for (; *ptr != 0; ptr++)
1295    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
1296      {      {
1297      if (dup_parens && *count < hwm_count) *count = hwm_count;      if (dup_parens && *count < hwm_count) *count = hwm_count;
1298      *ptrptr = ptr;      goto FAIL_EXIT;
     return -1;  
1299      }      }
1300    
1301    else if (*ptr == CHAR_VERTICAL_LINE && dup_parens)    else if (*ptr == CHAR_VERTICAL_LINE && dup_parens)
# Line 1442  for (;;) Line 1535  for (;;)
1535      case OP_CALLOUT:      case OP_CALLOUT:
1536      case OP_SOD:      case OP_SOD:
1537      case OP_SOM:      case OP_SOM:
1538        case OP_SET_SOM:
1539      case OP_EOD:      case OP_EOD:
1540      case OP_EODN:      case OP_EODN:
1541      case OP_CIRC:      case OP_CIRC:
# Line 1601  for (;;) Line 1695  for (;;)
1695    
1696    /* 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
1697    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
1698    two bytes of parameters. */    two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we
1699      must add in its length. */
1700    
1701    else    else
1702      {      {
# Line 1625  for (;;) Line 1720  for (;;)
1720        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
1721        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1722        break;        break;
1723    
1724          case OP_MARK:
1725          case OP_PRUNE_ARG:
1726          case OP_SKIP_ARG:
1727          case OP_THEN_ARG:
1728          code += code[1];
1729          break;
1730        }        }
1731    
1732      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
# Line 1696  for (;;) Line 1798  for (;;)
1798    
1799    /* 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
1800    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
1801    two bytes of parameters. */    two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we
1802      must add in its length. */
1803    
1804    else    else
1805      {      {
# Line 1720  for (;;) Line 1823  for (;;)
1823        case OP_TYPEEXACT:        case OP_TYPEEXACT:
1824        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1825        break;        break;
1826    
1827          case OP_MARK:
1828          case OP_PRUNE_ARG:
1829          case OP_SKIP_ARG:
1830          case OP_THEN_ARG:
1831          code += code[1];
1832          break;
1833        }        }
1834    
1835      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
# Line 1776  Arguments: Line 1886  Arguments:
1886    code        points to start of search    code        points to start of search
1887    endcode     points to where to stop    endcode     points to where to stop
1888    utf8        TRUE if in UTF8 mode    utf8        TRUE if in UTF8 mode
1889      cd          contains pointers to tables etc.
1890    
1891  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
1892  */  */
1893    
1894  static BOOL  static BOOL
1895  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8)  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,
1896      compile_data *cd)
1897  {  {
1898  register int c;  register int c;
1899  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 1812  for (code = first_significant_code(code Line 1924  for (code = first_significant_code(code
1924      continue;      continue;
1925      }      }
1926    
1927      /* For a recursion/subroutine call, if its end has been reached, which
1928      implies a subroutine call, we can scan it. */
1929    
1930      if (c == OP_RECURSE)
1931        {
1932        BOOL empty_branch = FALSE;
1933        const uschar *scode = cd->start_code + GET(code, 1);
1934        if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
1935        do
1936          {
1937          if (could_be_empty_branch(scode, endcode, utf8, cd))
1938            {
1939            empty_branch = TRUE;
1940            break;
1941            }
1942          scode += GET(scode, 1);
1943          }
1944        while (*scode == OP_ALT);
1945        if (!empty_branch) return FALSE;  /* All branches are non-empty */
1946        continue;
1947        }
1948    
1949    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1950    
1951    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 1830  for (code = first_significant_code(code Line 1964  for (code = first_significant_code(code
1964        empty_branch = FALSE;        empty_branch = FALSE;
1965        do        do
1966          {          {
1967          if (!empty_branch && could_be_empty_branch(code, endcode, utf8))          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))
1968            empty_branch = TRUE;            empty_branch = TRUE;
1969          code += GET(code, 1);          code += GET(code, 1);
1970          }          }
# Line 1964  for (code = first_significant_code(code Line 2098  for (code = first_significant_code(code
2098      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];
2099      break;      break;
2100  #endif  #endif
2101    
2102        /* MARK, and PRUNE/SKIP/THEN with an argument must skip over the argument
2103        string. */
2104    
2105        case OP_MARK:
2106        case OP_PRUNE_ARG:
2107        case OP_SKIP_ARG:
2108        case OP_THEN_ARG:
2109        code += code[1];
2110        break;
2111    
2112        /* None of the remaining opcodes are required to match a character. */
2113    
2114        default:
2115        break;
2116      }      }
2117    }    }
2118    
# Line 1986  Arguments: Line 2135  Arguments:
2135    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2136    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2137    utf8        TRUE if in UTF-8 mode    utf8        TRUE if in UTF-8 mode
2138      cd          pointers to tables etc
2139    
2140  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2141  */  */
2142    
2143  static BOOL  static BOOL
2144  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,
2145    BOOL utf8)    BOOL utf8, compile_data *cd)
2146  {  {
2147  while (bcptr != NULL && bcptr->current >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2148    {    {
2149    if (!could_be_empty_branch(bcptr->current, endcode, utf8)) return FALSE;    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))
2150        return FALSE;
2151    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2152    }    }
2153  return TRUE;  return TRUE;
# Line 2178  auto_callout(uschar *code, const uschar Line 2329  auto_callout(uschar *code, const uschar
2329  {  {
2330  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2331  *code++ = 255;  *code++ = 255;
2332  PUT(code, 0, ptr - cd->start_pattern);  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2333  PUT(code, LINK_SIZE, 0);                /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2334  return code + 2*LINK_SIZE;  return code + 2*LINK_SIZE;
2335  }  }
2336    
# Line 2204  Returns:             nothing Line 2355  Returns:             nothing
2355  static void  static void
2356  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)
2357  {  {
2358  int length = ptr - cd->start_pattern - GET(previous_callout, 2);  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2359  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
2360  }  }
2361    
# Line 2254  for (++c; c <= d; c++) Line 2405  for (++c; c <= d; c++)
2405    
2406  return TRUE;  return TRUE;
2407  }  }
2408    
2409    
2410    
2411    /*************************************************
2412    *        Check a character and a property        *
2413    *************************************************/
2414    
2415    /* This function is called by check_auto_possessive() when a property item
2416    is adjacent to a fixed character.
2417    
2418    Arguments:
2419      c            the character
2420      ptype        the property type
2421      pdata        the data for the type
2422      negated      TRUE if it's a negated property (\P or \p{^)
2423    
2424    Returns:       TRUE if auto-possessifying is OK
2425    */
2426    
2427    static BOOL
2428    check_char_prop(int c, int ptype, int pdata, BOOL negated)
2429    {
2430    const ucd_record *prop = GET_UCD(c);
2431    switch(ptype)
2432      {
2433      case PT_LAMP:
2434      return (prop->chartype == ucp_Lu ||
2435              prop->chartype == ucp_Ll ||
2436              prop->chartype == ucp_Lt) == negated;
2437    
2438      case PT_GC:
2439      return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;
2440    
2441      case PT_PC:
2442      return (pdata == prop->chartype) == negated;
2443    
2444      case PT_SC:
2445      return (pdata == prop->script) == negated;
2446    
2447      /* These are specials */
2448    
2449      case PT_ALNUM:
2450      return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2451              _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;
2452    
2453      case PT_SPACE:    /* Perl space */
2454      return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2455              c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2456              == negated;
2457    
2458      case PT_PXSPACE:  /* POSIX space */
2459      return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2460              c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2461              c == CHAR_FF || c == CHAR_CR)
2462              == negated;
2463    
2464      case PT_WORD:
2465      return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2466              _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2467              c == CHAR_UNDERSCORE) == negated;
2468      }
2469    return FALSE;
2470    }
2471  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2472    
2473    
# Line 2267  whether the next thing could possibly ma Line 2481  whether the next thing could possibly ma
2481  sense to automatically possessify the repeated item.  sense to automatically possessify the repeated item.
2482    
2483  Arguments:  Arguments:
2484    op_code       the repeated op code    previous      pointer to the repeated opcode
   this          data for this item, depends on the opcode  
2485    utf8          TRUE in UTF-8 mode    utf8          TRUE in UTF-8 mode
   utf8_char     used for utf8 character bytes, NULL if not relevant  
2486    ptr           next character in pattern    ptr           next character in pattern
2487    options       options bits    options       options bits
2488    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2279  Returns:        TRUE if possessifying is Line 2491  Returns:        TRUE if possessifying is
2491  */  */
2492    
2493  static BOOL  static BOOL
2494  check_auto_possessive(int op_code, int item, BOOL utf8, uschar *utf8_char,  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,
2495    const uschar *ptr, int options, compile_data *cd)    int options, compile_data *cd)
2496  {  {
2497  int next;  int c, next;
2498    int op_code = *previous++;
2499    
2500  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
2501    
# Line 2343  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 2556  if (*ptr == CHAR_ASTERISK || *ptr == CHA
2556    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)
2557      return FALSE;      return FALSE;
2558    
2559  /* 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
2560  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. */  
2561    
2562  if (next >= 0) switch(op_code)  if (next >= 0) switch(op_code)
2563    {    {
2564    case OP_CHAR:    case OP_CHAR:
2565  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2566    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2567  #else  #else
2568    (void)(utf8_char);  /* Keep compiler happy by referencing function argument */    c = *previous;
2569  #endif  #endif
2570    return item != next;    return c != next;
2571    
2572    /* 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
2573    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 2367  if (next >= 0) switch(op_code) Line 2575  if (next >= 0) switch(op_code)
2575    
2576    case OP_CHARNC:    case OP_CHARNC:
2577  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2578    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2579    #else
2580      c = *previous;
2581  #endif  #endif
2582    if (item == next) return FALSE;    if (c == next) return FALSE;
2583  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2584    if (utf8)    if (utf8)
2585      {      {
# Line 2380  if (next >= 0) switch(op_code) Line 2590  if (next >= 0) switch(op_code)
2590  #else  #else
2591      othercase = NOTACHAR;      othercase = NOTACHAR;
2592  #endif  #endif
2593      return (unsigned int)item != othercase;      return (unsigned int)c != othercase;
2594      }      }
2595    else    else
2596  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2597    return (item != cd->fcc[next]);  /* Non-UTF-8 mode */    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */
2598    
2599    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, its data is always a single-byte character. */
2600    
2601    case OP_NOT:    case OP_NOT:
2602    if (item == next) return TRUE;    if ((c = *previous) == next) return TRUE;
2603    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2604  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2605    if (utf8)    if (utf8)
# Line 2401  if (next >= 0) switch(op_code) Line 2611  if (next >= 0) switch(op_code)
2611  #else  #else
2612      othercase = NOTACHAR;      othercase = NOTACHAR;
2613  #endif  #endif
2614      return (unsigned int)item == othercase;      return (unsigned int)c == othercase;
2615      }      }
2616    else    else
2617  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2618    return (item == cd->fcc[next]);  /* Non-UTF-8 mode */    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */
2619    
2620      /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.
2621      When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
2622    
2623    case OP_DIGIT:    case OP_DIGIT:
2624    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;
# Line 2448  if (next >= 0) switch(op_code) Line 2661  if (next >= 0) switch(op_code)
2661      case 0x202f:      case 0x202f:
2662      case 0x205f:      case 0x205f:
2663      case 0x3000:      case 0x3000:
2664      return op_code != OP_HSPACE;      return op_code == OP_NOT_HSPACE;
2665      default:      default:
2666      return op_code == OP_HSPACE;      return op_code != OP_NOT_HSPACE;
2667      }      }
2668    
2669      case OP_ANYNL:
2670    case OP_VSPACE:    case OP_VSPACE:
2671    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
2672    switch(next)    switch(next)
# Line 2464  if (next >= 0) switch(op_code) Line 2678  if (next >= 0) switch(op_code)
2678      case 0x85:      case 0x85:
2679      case 0x2028:      case 0x2028:
2680      case 0x2029:      case 0x2029:
2681      return op_code != OP_VSPACE;      return op_code == OP_NOT_VSPACE;
2682      default:      default:
2683      return op_code == OP_VSPACE;      return op_code != OP_NOT_VSPACE;
2684      }      }
2685    
2686    #ifdef SUPPORT_UCP
2687      case OP_PROP:
2688      return check_char_prop(next, previous[0], previous[1], FALSE);
2689    
2690      case OP_NOTPROP:
2691      return check_char_prop(next, previous[0], previous[1], TRUE);
2692    #endif
2693    
2694    default:    default:
2695    return FALSE;    return FALSE;
2696    }    }
2697    
2698    
2699  /* 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
2700    is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are
2701    generated only when PCRE_UCP is *not* set, that is, when only ASCII
2702    characteristics are recognized. Similarly, the opcodes OP_DIGIT etc. are
2703    replaced by OP_PROP codes when PCRE_UCP is set. */
2704    
2705  switch(op_code)  switch(op_code)
2706    {    {
2707    case OP_CHAR:    case OP_CHAR:
2708    case OP_CHARNC:    case OP_CHARNC:
2709  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2710    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2711    #else
2712      c = *previous;
2713  #endif  #endif
2714    switch(-next)    switch(-next)
2715      {      {
2716      case ESC_d:      case ESC_d:
2717      return item > 127 || (cd->ctypes[item] & ctype_digit) == 0;      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;
2718    
2719      case ESC_D:      case ESC_D:
2720      return item <= 127 && (cd->ctypes[item] & ctype_digit) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;
2721    
2722      case ESC_s:      case ESC_s:
2723      return item > 127 || (cd->ctypes[item] & ctype_space) == 0;      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;
2724    
2725      case ESC_S:      case ESC_S:
2726      return item <= 127 && (cd->ctypes[item] & ctype_space) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;
2727    
2728      case ESC_w:      case ESC_w:
2729      return item > 127 || (cd->ctypes[item] & ctype_word) == 0;      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;
2730    
2731      case ESC_W:      case ESC_W:
2732      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;
2733    
2734      case ESC_h:      case ESC_h:
2735      case ESC_H:      case ESC_H:
2736      switch(item)      switch(c)
2737        {        {
2738        case 0x09:        case 0x09:
2739        case 0x20:        case 0x20:
# Line 2533  switch(op_code) Line 2761  switch(op_code)
2761    
2762      case ESC_v:      case ESC_v:
2763      case ESC_V:      case ESC_V:
2764      switch(item)      switch(c)
2765        {        {
2766        case 0x0a:        case 0x0a:
2767        case 0x0b:        case 0x0b:
# Line 2547  switch(op_code) Line 2775  switch(op_code)
2775        return -next == ESC_v;        return -next == ESC_v;
2776        }        }
2777    
2778        /* When PCRE_UCP is set, these values get generated for \d etc. Find
2779        their substitutions and process them. The result will always be either
2780        -ESC_p or -ESC_P. Then fall through to process those values. */
2781    
2782    #ifdef SUPPORT_UCP
2783        case ESC_du:
2784        case ESC_DU:
2785        case ESC_wu:
2786        case ESC_WU:
2787        case ESC_su:
2788        case ESC_SU:
2789          {
2790          int temperrorcode = 0;
2791          ptr = substitutes[-next - ESC_DU];
2792          next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);
2793          if (temperrorcode != 0) return FALSE;
2794          ptr++;    /* For compatibility */
2795          }
2796        /* Fall through */
2797    
2798        case ESC_p:
2799        case ESC_P:
2800          {
2801          int ptype, pdata, errorcodeptr;
2802          BOOL negated;
2803    
2804          ptr--;      /* Make ptr point at the p or P */
2805          ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);
2806          if (ptype < 0) return FALSE;
2807          ptr++;      /* Point past the final curly ket */
2808    
2809          /* If the property item is optional, we have to give up. (When generated
2810          from \d etc by PCRE_UCP, this test will have been applied much earlier,
2811          to the original \d etc. At this point, ptr will point to a zero byte. */
2812    
2813          if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
2814            strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
2815              return FALSE;
2816    
2817          /* Do the property check. */
2818    
2819          return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);
2820          }
2821    #endif
2822    
2823      default:      default:
2824      return FALSE;      return FALSE;
2825      }      }
2826    
2827      /* In principle, support for Unicode properties should be integrated here as
2828      well. It means re-organizing the above code so as to get hold of the property
2829      values before switching on the op-code. However, I wonder how many patterns
2830      combine ASCII \d etc with Unicode properties? (Note that if PCRE_UCP is set,
2831      these op-codes are never generated.) */
2832    
2833    case OP_DIGIT:    case OP_DIGIT:
2834    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2835           next == -ESC_h || next == -ESC_v;           next == -ESC_h || next == -ESC_v || next == -ESC_R;
2836    
2837    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2838    return next == -ESC_d;    return next == -ESC_d;
2839    
2840    case OP_WHITESPACE:    case OP_WHITESPACE:
2841    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;
2842    
2843    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2844    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2845    
2846    case OP_HSPACE:    case OP_HSPACE:
2847    return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||
2848             next == -ESC_w || next == -ESC_v || next == -ESC_R;
2849    
2850    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
2851    return next == -ESC_h;    return next == -ESC_h;
2852    
2853    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
2854      case OP_ANYNL:
2855    case OP_VSPACE:    case OP_VSPACE:
2856    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2857    
2858    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
2859    return next == -ESC_v;    return next == -ESC_v || next == -ESC_R;
2860    
2861    case OP_WORDCHAR:    case OP_WORDCHAR:
2862    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||
2863             next == -ESC_v || next == -ESC_R;
2864    
2865    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2866    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2642  BOOL inescq = FALSE; Line 2924  BOOL inescq = FALSE;
2924  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstbyte = FALSE;
2925  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
2926  const uschar *tempptr;  const uschar *tempptr;
2927    const uschar *nestptr = NULL;
2928  uschar *previous = NULL;  uschar *previous = NULL;
2929  uschar *previous_callout = NULL;  uschar *previous_callout = NULL;
2930  uschar *save_hwm = NULL;  uschar *save_hwm = NULL;
# Line 2658  BOOL utf8 = FALSE; Line 2941  BOOL utf8 = FALSE;
2941  uschar *utf8_char = NULL;  uschar *utf8_char = NULL;
2942  #endif  #endif
2943    
2944  #ifdef DEBUG  #ifdef PCRE_DEBUG
2945  if (lengthptr != NULL) DPRINTF((">> start branch\n"));  if (lengthptr != NULL) DPRINTF((">> start branch\n"));
2946  #endif  #endif
2947    
# Line 2712  for (;; ptr++) Line 2995  for (;; ptr++)
2995    
2996    c = *ptr;    c = *ptr;
2997    
2998      /* If we are at the end of a nested substitution, revert to the outer level
2999      string. Nesting only happens one level deep. */
3000    
3001      if (c == 0 && nestptr != NULL)
3002        {
3003        ptr = nestptr;
3004        nestptr = NULL;
3005        c = *ptr;
3006        }
3007    
3008    /* 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
3009    previous cycle of this loop. */    previous cycle of this loop. */
3010    
3011    if (lengthptr != NULL)    if (lengthptr != NULL)
3012      {      {
3013  #ifdef DEBUG  #ifdef PCRE_DEBUG
3014      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3015  #endif  #endif
3016      if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */
3017        {        {
3018        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3019        goto FAILED;        goto FAILED;
# Line 2742  for (;; ptr++) Line 3035  for (;; ptr++)
3035        goto FAILED;        goto FAILED;
3036        }        }
3037    
3038      *lengthptr += code - last_code;      *lengthptr += (int)(code - last_code);
3039      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));
3040    
3041      /* 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 2769  for (;; ptr++) Line 3062  for (;; ptr++)
3062    /* 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
3063    reference list. */    reference list. */
3064    
3065    else if (cd->hwm > cd->start_workspace + COMPILE_WORK_SIZE)    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)
3066      {      {
3067      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3068      goto FAILED;      goto FAILED;
# Line 2860  for (;; ptr++) Line 3153  for (;; ptr++)
3153          *errorcodeptr = ERR20;          *errorcodeptr = ERR20;
3154          goto FAILED;          goto FAILED;
3155          }          }
3156        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += (int)(code - last_code);   /* To include callout length */
3157        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
3158        }        }
3159      return TRUE;      return TRUE;
# Line 3065  for (;; ptr++) Line 3358  for (;; ptr++)
3358            ptr++;            ptr++;
3359            }            }
3360    
3361          posix_class = check_posix_name(ptr, tempptr - ptr);          posix_class = check_posix_name(ptr, (int)(tempptr - ptr));
3362          if (posix_class < 0)          if (posix_class < 0)
3363            {            {
3364            *errorcodeptr = ERR30;            *errorcodeptr = ERR30;
# Line 3079  for (;; ptr++) Line 3372  for (;; ptr++)
3372          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
3373            posix_class = 0;            posix_class = 0;
3374    
3375          /* 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
3376          because we may be adding and subtracting from it, and we don't want to          different escape sequences that use Unicode properties. */
3377          subtract bits that may be in the main map already. At the end we or the  
3378          result into the bit map that is being built. */  #ifdef SUPPORT_UCP
3379            if ((options & PCRE_UCP) != 0)
3380              {
3381              int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);
3382              if (posix_substitutes[pc] != NULL)
3383                {
3384                nestptr = tempptr + 1;
3385                ptr = posix_substitutes[pc] - 1;
3386                continue;
3387                }
3388              }
3389    #endif
3390            /* In the non-UCP case, we build the bit map for the POSIX class in a
3391            chunk of local store because we may be adding and subtracting from it,
3392            and we don't want to subtract bits that may be in the main map already.
3393            At the end we or the result into the bit map that is being built. */
3394    
3395          posix_class *= 3;          posix_class *= 3;
3396    
# Line 3126  for (;; ptr++) Line 3434  for (;; ptr++)
3434    
3435        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
3436        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
3437        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
3438        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
3439        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
3440        character in them, so set class_charcount bigger than one. */        are either treated as literal characters (by default), or are faulted if
3441          PCRE_EXTRA is set. */
3442    
3443        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
3444          {          {
3445          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3446          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
3447    
3448          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 */  
3449          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
3450            {            {
3451            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3155  for (;; ptr++) Line 3462  for (;; ptr++)
3462            register const uschar *cbits = cd->cbits;            register const uschar *cbits = cd->cbits;
3463            class_charcount += 2;     /* Greater than 1 is what matters */            class_charcount += 2;     /* Greater than 1 is what matters */
3464    
3465            /* Save time by not doing this in the pre-compile phase. */            switch (-c)
   
           if (lengthptr == NULL) switch (-c)  
3466              {              {
3467    #ifdef SUPPORT_UCP
3468                case ESC_du:     /* These are the values given for \d etc */
3469                case ESC_DU:     /* when PCRE_UCP is set. We replace the */
3470                case ESC_wu:     /* escape sequence with an appropriate \p */
3471                case ESC_WU:     /* or \P to test Unicode properties instead */
3472                case ESC_su:     /* of the default ASCII testing. */
3473                case ESC_SU:
3474                nestptr = ptr;
3475                ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
3476                class_charcount -= 2;                /* Undo! */
3477                continue;
3478    #endif
3479              case ESC_d:              case ESC_d:
3480              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];
3481              continue;              continue;
# Line 3188  for (;; ptr++) Line 3505  for (;; ptr++)
3505              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
3506              continue;              continue;
3507    
3508              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)  
             {  
3509              SETBIT(classbits, 0x09); /* VT */              SETBIT(classbits, 0x09); /* VT */
3510              SETBIT(classbits, 0x20); /* SPACE */              SETBIT(classbits, 0x20); /* SPACE */
3511              SETBIT(classbits, 0xa0); /* NSBP */              SETBIT(classbits, 0xa0); /* NSBP */
# Line 3225  for (;; ptr++) Line 3529  for (;; ptr++)
3529                }                }
3530  #endif  #endif
3531              continue;              continue;
             }  
3532    
3533            if (-c == ESC_H)              case ESC_H:
             {  
3534              for (c = 0; c < 32; c++)              for (c = 0; c < 32; c++)
3535                {                {
3536                int x = 0xff;                int x = 0xff;
# Line 3270  for (;; ptr++) Line 3572  for (;; ptr++)
3572                }                }
3573  #endif  #endif
3574              continue;              continue;
             }  
3575    
3576            if (-c == ESC_v)              case ESC_v:
             {  
3577              SETBIT(classbits, 0x0a); /* LF */              SETBIT(classbits, 0x0a); /* LF */
3578              SETBIT(classbits, 0x0b); /* VT */              SETBIT(classbits, 0x0b); /* VT */
3579              SETBIT(classbits, 0x0c); /* FF */              SETBIT(classbits, 0x0c); /* FF */
# Line 3289  for (;; ptr++) Line 3589  for (;; ptr++)
3589                }                }
3590  #endif  #endif
3591              continue;              continue;
             }  
3592    
3593            if (-c == ESC_V)              case ESC_V:
             {  
3594              for (c = 0; c < 32; c++)              for (c = 0; c < 32; c++)
3595                {                {
3596                int x = 0xff;                int x = 0xff;
# Line 3322  for (;; ptr++) Line 3620  for (;; ptr++)
3620                }                }
3621  #endif  #endif
3622              continue;              continue;
             }  
   
           /* We need to deal with \P and \p in both phases. */  
3623    
3624  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3625            if (-c == ESC_p || -c == ESC_P)              case ESC_p:
3626              {              case ESC_P:
3627              BOOL negated;                {
3628              int pdata;                BOOL negated;
3629              int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int pdata;
3630              if (ptype < 0) goto FAILED;                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
3631              class_utf8 = TRUE;                if (ptype < 0) goto FAILED;
3632              *class_utf8data++ = ((-c == ESC_p) != negated)?                class_utf8 = TRUE;
3633                XCL_PROP : XCL_NOTPROP;                *class_utf8data++ = ((-c == ESC_p) != negated)?
3634              *class_utf8data++ = ptype;                  XCL_PROP : XCL_NOTPROP;
3635              *class_utf8data++ = pdata;                *class_utf8data++ = ptype;
3636              class_charcount -= 2;   /* Not a < 256 character */                *class_utf8data++ = pdata;
3637              continue;                class_charcount -= 2;   /* Not a < 256 character */
3638              }                continue;
3639                  }
3640  #endif  #endif
3641            /* Unrecognized escapes are faulted if PCRE is running in its              /* Unrecognized escapes are faulted if PCRE is running in its
3642            strict mode. By default, for compatibility with Perl, they are              strict mode. By default, for compatibility with Perl, they are
3643            treated as literals. */              treated as literals. */
3644    
3645            if ((options & PCRE_EXTRA) != 0)              default:
3646              {              if ((options & PCRE_EXTRA) != 0)
3647              *errorcodeptr = ERR7;                {
3648              goto FAILED;                *errorcodeptr = ERR7;
3649                  goto FAILED;
3650                  }
3651                class_charcount -= 2;  /* Undo the default count from above */
3652                c = *ptr;              /* Get the final character and fall through */
3653                break;
3654              }              }
   
           class_charcount -= 2;  /* Undo the default count from above */  
           c = *ptr;              /* Get the final character and fall through */  
3655            }            }
3656    
3657          /* 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 3423  for (;; ptr++) Line 3721  for (;; ptr++)
3721            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3722            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3723    
3724            /* \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 */  
3725    
3726            if (d < 0)            if (d < 0)
3727              {              {
3728              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  
3729                {                {
3730                ptr = oldptr;                ptr = oldptr;
3731                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto LONE_SINGLE_CHARACTER;  /* A few lines below */
# Line 3596  for (;; ptr++) Line 3891  for (;; ptr++)
3891          }          }
3892        }        }
3893    
3894      /* 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.
3895        If we are at the end of an internal nested string, revert to the outer
3896        string. */
3897    
3898        while (((c = *(++ptr)) != 0 ||
3899               (nestptr != NULL &&
3900                 (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&
3901               (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
3902    
3903      while ((c = *(++ptr)) != 0 && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));      /* Check for missing terminating ']' */
3904    
3905      if (c == 0)                          /* Missing terminating ']' */      if (c == 0)
3906        {        {
3907        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
3908        goto FAILED;        goto FAILED;
3909        }        }
3910    
   
 /* 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  
   
   
3911      /* 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
3912      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
3913      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 3688  we set the flag only if there is a liter Line 3971  we set the flag only if there is a liter
3971    
3972      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3973      extended class, with its own opcode, unless there was a negated special      extended class, with its own opcode, unless there was a negated special
3974      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
3975      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
3976      (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
3977      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
3978        actual compiled code. */
3979    
3980  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3981      if (class_utf8 && !should_flip_negation)      if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))
3982        {        {
3983        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3984        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3720  we set the flag only if there is a liter Line 4004  we set the flag only if there is a liter
4004        }        }
4005  #endif  #endif
4006    
4007      /* 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
4008      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
4009      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
4010      map into the code vector, negating it if necessary. */      (non-UCP) in the class. Then copy the 32-byte map into the code vector,
4011        negating it if necessary. */
4012    
4013      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
4014      if (negate_class)      if (negate_class)
# Line 3847  we set the flag only if there is a liter Line 4132  we set the flag only if there is a liter
4132    
4133        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4134            repeat_max < 0 &&            repeat_max < 0 &&
4135            check_auto_possessive(*previous, c, utf8, utf8_char, ptr + 1,            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
             options, cd))  
4136          {          {
4137          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4138          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 3869  we set the flag only if there is a liter Line 4153  we set the flag only if there is a liter
4153        c = previous[1];        c = previous[1];
4154        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4155            repeat_max < 0 &&            repeat_max < 0 &&
4156            check_auto_possessive(OP_NOT, c, utf8, NULL, ptr + 1, options, cd))            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
4157          {          {
4158          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4159          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 3893  we set the flag only if there is a liter Line 4177  we set the flag only if there is a liter
4177    
4178        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4179            repeat_max < 0 &&            repeat_max < 0 &&
4180            check_auto_possessive(c, 0, utf8, NULL, ptr + 1, options, cd))            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
4181          {          {
4182          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4183          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 4103  we set the flag only if there is a liter Line 4387  we set the flag only if there is a liter
4387        {        {
4388        register int i;        register int i;
4389        int ketoffset = 0;        int ketoffset = 0;
4390        int len = code - previous;        int len = (int)(code - previous);
4391        uschar *bralink = NULL;        uschar *bralink = NULL;
4392    
4393        /* Repeating a DEFINE group is pointless */        /* Repeating a DEFINE group is pointless */
# Line 4124  we set the flag only if there is a liter Line 4408  we set the flag only if there is a liter
4408          {          {
4409          register uschar *ket = previous;          register uschar *ket = previous;
4410          do ket += GET(ket, 1); while (*ket != OP_KET);          do ket += GET(ket, 1); while (*ket != OP_KET);
4411          ketoffset = code - ket;          ketoffset = (int)(code - ket);
4412          }          }
4413    
4414        /* 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 4192  we set the flag only if there is a liter Line 4476  we set the flag only if there is a liter
4476            /* We chain together the bracket offset fields that have to be            /* We chain together the bracket offset fields that have to be
4477            filled in later when the ends of the brackets are reached. */            filled in later when the ends of the brackets are reached. */
4478    
4479            offset = (bralink == NULL)? 0 : previous - bralink;            offset = (bralink == NULL)? 0 : (int)(previous - bralink);
4480            bralink = previous;            bralink = previous;
4481            PUTINC(previous, 0, offset);            PUTINC(previous, 0, offset);
4482            }            }
# Line 4213  we set the flag only if there is a liter Line 4497  we set the flag only if there is a liter
4497            {            {
4498            /* In the pre-compile phase, we don't actually do the replication. We            /* In the pre-compile phase, we don't actually do the replication. We
4499            just adjust the length as if we had. Do some paranoid checks for            just adjust the length as if we had. Do some paranoid checks for
4500            potential integer overflow. */            potential integer overflow. The INT64_OR_DOUBLE type is a 64-bit
4501              integer type when available, otherwise double. */
4502    
4503            if (lengthptr != NULL)            if (lengthptr != NULL)
4504              {              {
4505              int delta = (repeat_min - 1)*length_prevgroup;              int delta = (repeat_min - 1)*length_prevgroup;
4506              if ((double)(repeat_min - 1)*(double)length_prevgroup >              if ((INT64_OR_DOUBLE)(repeat_min - 1)*
4507                                                              (double)INT_MAX ||                    (INT64_OR_DOUBLE)length_prevgroup >
4508                        (INT64_OR_DOUBLE)INT_MAX ||
4509                  OFLOW_MAX - *lengthptr < delta)                  OFLOW_MAX - *lengthptr < delta)
4510                {                {
4511                *errorcodeptr = ERR20;                *errorcodeptr = ERR20;
# Line 4265  we set the flag only if there is a liter Line 4551  we set the flag only if there is a liter
4551          just adjust the length as if we had. For each repetition we must add 1          just adjust the length as if we had. For each repetition we must add 1
4552          to the length for BRAZERO and for all but the last repetition we must          to the length for BRAZERO and for all but the last repetition we must
4553          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
4554          paranoid checks to avoid integer overflow. */          paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is
4555            a 64-bit integer type when available, otherwise double. */
4556    
4557          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
4558            {            {
4559            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
4560                        2 - 2*LINK_SIZE;   /* Last one doesn't nest */                        2 - 2*LINK_SIZE;   /* Last one doesn't nest */
4561            if ((double)repeat_max *            if ((INT64_OR_DOUBLE)repeat_max *
4562                  (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)                  (INT64_OR_DOUBLE)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
4563                    > (double)INT_MAX ||                    > (INT64_OR_DOUBLE)INT_MAX ||
4564                OFLOW_MAX - *lengthptr < delta)                OFLOW_MAX - *lengthptr < delta)
4565              {              {
4566              *errorcodeptr = ERR20;              *errorcodeptr = ERR20;
# Line 4298  we set the flag only if there is a liter Line 4585  we set the flag only if there is a liter
4585              {              {
4586              int offset;              int offset;
4587              *code++ = OP_BRA;              *code++ = OP_BRA;
4588              offset = (bralink == NULL)? 0 : code - bralink;              offset = (bralink == NULL)? 0 : (int)(code - bralink);
4589              bralink = code;              bralink = code;
4590              PUTINC(code, 0, offset);              PUTINC(code, 0, offset);
4591              }              }
# Line 4319  we set the flag only if there is a liter Line 4606  we set the flag only if there is a liter
4606          while (bralink != NULL)          while (bralink != NULL)
4607            {            {
4608            int oldlinkoffset;            int oldlinkoffset;
4609            int offset = code - bralink + 1;            int offset = (int)(code - bralink + 1);
4610            uschar *bra = code - offset;            uschar *bra = code - offset;
4611            oldlinkoffset = GET(bra, 1);            oldlinkoffset = GET(bra, 1);
4612            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;
# Line 4350  we set the flag only if there is a liter Line 4637  we set the flag only if there is a liter
4637            uschar *scode = bracode;            uschar *scode = bracode;
4638            do            do
4639              {              {
4640              if (could_be_empty_branch(scode, ketcode, utf8))              if (could_be_empty_branch(scode, ketcode, utf8, cd))
4641                {                {
4642                *bracode += OP_SBRA - OP_BRA;                *bracode += OP_SBRA - OP_BRA;
4643                break;                break;
# Line 4407  we set the flag only if there is a liter Line 4694  we set the flag only if there is a liter
4694  #endif  #endif
4695          }          }
4696    
4697        len = code - tempcode;        len = (int)(code - tempcode);
4698        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4699          {          {
4700          case OP_STAR:  *tempcode = OP_POSSTAR; break;          case OP_STAR:  *tempcode = OP_POSSTAR; break;
# Line 4425  we set the flag only if there is a liter Line 4712  we set the flag only if there is a liter
4712          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;
4713          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;
4714    
4715            /* Because we are moving code along, we must ensure that any
4716            pending recursive references are updated. */
4717    
4718          default:          default:
4719            *code = OP_END;
4720            adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);
4721          memmove(tempcode + 1+LINK_SIZE, tempcode, len);          memmove(tempcode + 1+LINK_SIZE, tempcode, len);
4722          code += 1 + LINK_SIZE;          code += 1 + LINK_SIZE;
4723          len += 1 + LINK_SIZE;          len += 1 + LINK_SIZE;
# Line 4461  we set the flag only if there is a liter Line 4753  we set the flag only if there is a liter
4753    
4754      /* First deal with various "verbs" that can be introduced by '*'. */      /* First deal with various "verbs" that can be introduced by '*'. */
4755    
4756      if (*(++ptr) == CHAR_ASTERISK && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == CHAR_ASTERISK &&
4757             ((cd->ctypes[ptr[1]] & ctype_letter) != 0 || ptr[1] == ':'))
4758        {        {
4759        int i, namelen;        int i, namelen;
4760          int arglen = 0;
4761        const char *vn = verbnames;        const char *vn = verbnames;
4762        const uschar *name = ++ptr;        const uschar *name = ptr + 1;
4763          const uschar *arg = NULL;
4764        previous = NULL;        previous = NULL;
4765        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
4766          namelen = (int)(ptr - name);
4767    
4768        if (*ptr == CHAR_COLON)        if (*ptr == CHAR_COLON)
4769          {          {
4770          *errorcodeptr = ERR59;   /* Not supported */          arg = ++ptr;
4771          goto FAILED;          while ((cd->ctypes[*ptr] & (ctype_letter|ctype_digit)) != 0
4772              || *ptr == '_') ptr++;
4773            arglen = (int)(ptr - arg);
4774          }          }
4775    
4776        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
4777          {          {
4778          *errorcodeptr = ERR60;          *errorcodeptr = ERR60;
4779          goto FAILED;          goto FAILED;
4780          }          }
4781        namelen = ptr - name;  
4782          /* Scan the table of verb names */
4783    
4784        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4785          {          {
4786          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
# Line 4496  we set the flag only if there is a liter Line 4798  we set the flag only if there is a liter
4798                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
4799                }                }
4800              }              }
4801            *code++ = verbs[i].op;  
4802            break;            /* Handle the cases with/without an argument */
4803    
4804              if (arglen == 0)
4805                {
4806                if (verbs[i].op < 0)   /* Argument is mandatory */
4807                  {
4808                  *errorcodeptr = ERR66;
4809                  goto FAILED;
4810                  }
4811                *code++ = verbs[i].op;
4812                }
4813    
4814              else
4815                {
4816                if (verbs[i].op_arg < 0)   /* Argument is forbidden */
4817                  {
4818                  *errorcodeptr = ERR59;
4819                  goto FAILED;
4820                  }
4821                *code++ = verbs[i].op_arg;
4822                *code++ = arglen;
4823                memcpy(code, arg, arglen);
4824                code += arglen;
4825                *code++ = 0;
4826                }
4827    
4828              break;  /* Found verb, exit loop */
4829            }            }
4830    
4831          vn += verbs[i].len + 1;          vn += verbs[i].len + 1;
4832          }          }
4833        if (i < verbcount) continue;  
4834        *errorcodeptr = ERR60;        if (i < verbcount) continue;    /* Successfully handled a verb */
4835          *errorcodeptr = ERR60;          /* Verb not recognized */
4836        goto FAILED;        goto FAILED;
4837        }        }
4838    
# Line 4621  we set the flag only if there is a liter Line 4951  we set the flag only if there is a liter
4951                recno * 10 + *ptr - CHAR_0 : -1;                recno * 10 + *ptr - CHAR_0 : -1;
4952            ptr++;            ptr++;
4953            }            }
4954          namelen = ptr - name;          namelen = (int)(ptr - name);
4955    
4956          if ((terminator > 0 && *ptr++ != terminator) ||          if ((terminator > 0 && *ptr++ != terminator) ||
4957              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
# Line 4817  we set the flag only if there is a liter Line 5147  we set the flag only if there is a liter
5147              goto FAILED;              goto FAILED;
5148              }              }
5149            *code++ = n;            *code++ = n;
5150            PUT(code, 0, ptr - cd->start_pattern + 1);  /* Pattern offset */            PUT(code, 0, (int)(ptr - cd->start_pattern + 1)); /* Pattern offset */
5151            PUT(code, LINK_SIZE, 0);                    /* Default length */            PUT(code, LINK_SIZE, 0);                          /* Default length */
5152            code += 2 * LINK_SIZE;            code += 2 * LINK_SIZE;
5153            }            }
5154          previous = NULL;          previous = NULL;
# Line 4851  we set the flag only if there is a liter Line 5181  we set the flag only if there is a liter
5181            name = ++ptr;            name = ++ptr;
5182    
5183            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
5184            namelen = ptr - name;            namelen = (int)(ptr - name);
5185    
5186            /* In the pre-compile phase, just do a syntax check. */            /* In the pre-compile phase, just do a syntax check. */
5187    
# Line 4981  we set the flag only if there is a liter Line 5311  we set the flag only if there is a liter
5311          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
5312          name = ++ptr;          name = ++ptr;
5313          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
5314          namelen = ptr - name;          namelen = (int)(ptr - name);
5315    
5316          /* 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
5317          reference number. */          reference number. */
# Line 5144  we set the flag only if there is a liter Line 5474  we set the flag only if there is a liter
5474                  *errorcodeptr = ERR15;                  *errorcodeptr = ERR15;
5475                  goto FAILED;                  goto FAILED;
5476                  }                  }
5477    
5478                  /* Fudge the value of "called" so that when it is inserted as an
5479                  offset below, what it actually inserted is the reference number
5480                  of the group. */
5481    
5482                called = cd->start_code + recno;                called = cd->start_code + recno;
5483                PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code);                PUTINC(cd->hwm, 0, (int)(code + 2 + LINK_SIZE - cd->start_code));
5484                }                }
5485    
5486              /* If not a forward reference, and the subpattern is still open,              /* If not a forward reference, and the subpattern is still open,
# Line 5153  we set the flag only if there is a liter Line 5488  we set the flag only if there is a liter
5488              recursion that could loop for ever, and diagnose that case. */              recursion that could loop for ever, and diagnose that case. */
5489    
5490              else if (GET(called, 1) == 0 &&              else if (GET(called, 1) == 0 &&
5491                       could_be_empty(called, code, bcptr, utf8))                       could_be_empty(called, code, bcptr, utf8, cd))
5492                {                {
5493                *errorcodeptr = ERR40;                *errorcodeptr = ERR40;
5494                goto FAILED;                goto FAILED;
# Line 5169  we set the flag only if there is a liter Line 5504  we set the flag only if there is a liter
5504            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
5505    
5506            *code = OP_RECURSE;            *code = OP_RECURSE;
5507            PUT(code, 1, called - cd->start_code);            PUT(code, 1, (int)(called - cd->start_code));
5508            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
5509    
5510            *code = OP_KET;            *code = OP_KET;
# Line 5280  we set the flag only if there is a liter Line 5615  we set the flag only if there is a liter
5615          }     /* End of switch for character following (? */          }     /* End of switch for character following (? */
5616        }       /* End of (? handling */        }       /* End of (? handling */
5617    
5618      /* Opening parenthesis not followed by '?'. If PCRE_NO_AUTO_CAPTURE is set,      /* Opening parenthesis not followed by '*' or '?'. If PCRE_NO_AUTO_CAPTURE
5619      all unadorned brackets become non-capturing and behave like (?:...)      is set, all unadorned brackets become non-capturing and behave like (?:...)
5620      brackets. */      brackets. */
5621    
5622      else if ((options & PCRE_NO_AUTO_CAPTURE) != 0)      else if ((options & PCRE_NO_AUTO_CAPTURE) != 0)
# Line 5473  we set the flag only if there is a liter Line 5808  we set the flag only if there is a liter
5808    
5809      /* ===================================================================*/      /* ===================================================================*/
5810      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
5811      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
5812      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
5813      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
5814      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
5815      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
5816        ever created. */
5817    
5818      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
5819      tempptr = ptr;      tempptr = ptr;
# Line 5585  we set the flag only if there is a liter Line 5921  we set the flag only if there is a liter
5921    
5922        if (-c >= ESC_REF)        if (-c >= ESC_REF)
5923          {          {
5924            open_capitem *oc;
5925          recno = -c - ESC_REF;          recno = -c - ESC_REF;
5926    
5927          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
# Line 5594  we set the flag only if there is a liter Line 5931  we set the flag only if there is a liter
5931          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
5932          cd->backref_map |= (recno < 32)? (1 << recno) : 1;          cd->backref_map |= (recno < 32)? (1 << recno) : 1;
5933          if (recno > cd->top_backref) cd->top_backref = recno;          if (recno > cd->top_backref) cd->top_backref = recno;
5934    
5935            /* Check to see if this back reference is recursive, that it, it
5936            is inside the group that it references. A flag is set so that the
5937            group can be made atomic. */
5938    
5939            for (oc = cd->open_caps; oc != NULL; oc = oc->next)
5940              {
5941              if (oc->number == recno)
5942                {
5943                oc->flag = TRUE;
5944                break;
5945                }
5946              }
5947          }          }
5948    
5949        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
# Line 5623  we set the flag only if there is a liter Line 5973  we set the flag only if there is a liter
5973  #endif  #endif
5974    
5975        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
5976        can obtain the OP value by negating the escape value. */        can obtain the OP value by negating the escape value in the default
5977          situation when PCRE_UCP is not set. When it *is* set, we substitute
5978          Unicode property tests. */
5979    
5980        else        else
5981          {          {
5982          previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;  #ifdef SUPPORT_UCP
5983          *code++ = -c;          if (-c >= ESC_DU && -c <= ESC_wu)
5984              {
5985              nestptr = ptr + 1;                   /* Where to resume */
5986              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
5987              }
5988            else
5989    #endif
5990              {
5991              previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;
5992              *code++ = -c;
5993              }
5994          }          }
5995        continue;        continue;
5996        }        }
# Line 5787  int old_external_options = cd->external_ Line 6149  int old_external_options = cd->external_
6149  branch_chain bc;  branch_chain bc;
6150    
6151  bc.outer = bcptr;  bc.outer = bcptr;
6152  bc.current = code;  bc.current_branch = code;
6153    
6154  firstbyte = reqbyte = REQ_UNSET;  firstbyte = reqbyte = REQ_UNSET;
6155    
# Line 5806  them global. It tests the value of lengt Line 6168  them global. It tests the value of lengt
6168  pre-compile phase to find out whether anything has yet been compiled or not. */  pre-compile phase to find out whether anything has yet been compiled or not. */
6169    
6170  /* If this is a capturing subpattern, add to the chain of open capturing items  /* If this is a capturing subpattern, add to the chain of open capturing items
6171  so that we can detect them if (*ACCEPT) is encountered. */  so that we can detect them if (*ACCEPT) is encountered. This is also used to
6172    detect groups that contain recursive back references to themselves. */
6173    
6174  if (*code == OP_CBRA)  if (*code == OP_CBRA)
6175    {    {
6176    capnumber = GET2(code, 1 + LINK_SIZE);    capnumber = GET2(code, 1 + LINK_SIZE);
6177    capitem.number = capnumber;    capitem.number = capnumber;
6178    capitem.next = cd->open_caps;    capitem.next = cd->open_caps;
6179      capitem.flag = FALSE;
6180    cd->open_caps = &capitem;    cd->open_caps = &capitem;
6181    }    }
6182    
# Line 5958  for (;;) Line 6322  for (;;)
6322      {      {
6323      if (lengthptr == NULL)      if (lengthptr == NULL)
6324        {        {
6325        int branch_length = code - last_branch;        int branch_length = (int)(code - last_branch);
6326        do        do
6327          {          {
6328          int prev_length = GET(last_branch, 1);          int prev_length = GET(last_branch, 1);
# Line 5969  for (;;) Line 6333  for (;;)
6333        while (branch_length > 0);        while (branch_length > 0);
6334        }        }
6335    
     /* If it was a capturing subpattern, remove it from the chain. */  
   
     if (capnumber > 0) cd->open_caps = cd->open_caps->next;  
   
6336      /* Fill in the ket */      /* Fill in the ket */
6337    
6338      *code = OP_KET;      *code = OP_KET;
6339      PUT(code, 1, code - start_bracket);      PUT(code, 1, (int)(code - start_bracket));
6340      code += 1 + LINK_SIZE;      code += 1 + LINK_SIZE;
6341    
6342        /* If it was a capturing subpattern, check to see if it contained any
6343        recursive back references. If so, we must wrap it in atomic brackets.
6344        In any event, remove the block from the chain. */
6345    
6346        if (capnumber > 0)
6347          {
6348          if (cd->open_caps->flag)
6349            {
6350            memmove(start_bracket + 1 + LINK_SIZE, start_bracket,
6351              code - start_bracket);
6352            *start_bracket = OP_ONCE;
6353            code += 1 + LINK_SIZE;
6354            PUT(start_bracket, 1, (int)(code - start_bracket));
6355            *code = OP_KET;
6356            PUT(code, 1, (int)(code - start_bracket));
6357            code += 1 + LINK_SIZE;
6358            length += 2 + 2*LINK_SIZE;
6359            }
6360          cd->open_caps = cd->open_caps->next;
6361          }
6362    
6363      /* Reset options if needed. */      /* Reset options if needed. */
6364    
6365      if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS)      if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS)
# Line 6027  for (;;) Line 6408  for (;;)
6408    else    else
6409      {      {
6410      *code = OP_ALT;      *code = OP_ALT;
6411      PUT(code, 1, code - last_branch);      PUT(code, 1, (int)(code - last_branch));
6412      bc.current = last_branch = code;      bc.current_branch = last_branch = code;
6413      code += 1 + LINK_SIZE;      code += 1 + LINK_SIZE;
6414      }      }
6415    
# Line 6346  int length = 1;  /* For final END opcode Line 6727  int length = 1;  /* For final END opcode
6727  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
6728  int errorcode = 0;  int errorcode = 0;
6729  int skipatstart = 0;  int skipatstart = 0;
6730  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8;
6731  size_t size;  size_t size;
6732  uschar *code;  uschar *code;
6733  const uschar *codestart;  const uschar *codestart;
# Line 6416  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6797  while (ptr[skipatstart] == CHAR_LEFT_PAR
6797    
6798    if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)
6799      { skipatstart += 7; options |= PCRE_UTF8; continue; }      { skipatstart += 7; options |= PCRE_UTF8; continue; }
6800      else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0)
6801        { skipatstart += 6; options |= PCRE_UCP; continue; }
6802    
6803    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)
6804      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
# Line 6440  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6823  while (ptr[skipatstart] == CHAR_LEFT_PAR
6823    else break;    else break;
6824    }    }
6825    
6826    utf8 = (options & PCRE_UTF8) != 0;
6827    
6828  /* 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. */
6829    
6830  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
6831  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 &&  if (utf8 && (options & PCRE_NO_UTF8_CHECK) == 0 &&
6832       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)       (*erroroffset = _pcre_valid_utf8((USPTR)pattern, -1)) >= 0)
6833    {    {
6834    errorcode = ERR44;    errorcode = ERR44;
6835    goto PCRE_EARLY_ERROR_RETURN2;    goto PCRE_EARLY_ERROR_RETURN2;
# Line 6457  if (utf8) Line 6842  if (utf8)
6842    }    }
6843  #endif  #endif
6844    
6845    /* Can't support UCP unless PCRE has been compiled to include the code. */
6846    
6847    #ifndef SUPPORT_UCP
6848    if ((options & PCRE_UCP) != 0)
6849      {
6850      errorcode = ERR67;
6851      goto PCRE_EARLY_ERROR_RETURN;
6852      }
6853    #endif
6854    
6855  /* Check validity of \R options. */  /* Check validity of \R options. */
6856    
6857  switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))  switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
# Line 6585  regex compiled on a system with 4-byte p Line 6980  regex compiled on a system with 4-byte p
6980  pointers. */  pointers. */
6981    
6982  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
6983  re->size = size;  re->size = (int)size;
6984  re->options = cd->external_options;  re->options = cd->external_options;
6985  re->flags = cd->external_flags;  re->flags = cd->external_flags;
6986  re->dummy1 = 0;  re->dummy1 = 0;
# Line 6641  if debugging, leave the test till after Line 7036  if debugging, leave the test till after
7036    
7037  *code++ = OP_END;  *code++ = OP_END;
7038    
7039  #ifndef DEBUG  #ifndef PCRE_DEBUG
7040  if (code - codestart > length) errorcode = ERR23;  if (code - codestart > length) errorcode = ERR23;
7041  #endif  #endif
7042    
# Line 6656  while (errorcode == 0 && cd->hwm > cwork Line 7051  while (errorcode == 0 && cd->hwm > cwork
7051    recno = GET(codestart, offset);    recno = GET(codestart, offset);
7052    groupptr = _pcre_find_bracket(codestart, utf8, recno);    groupptr = _pcre_find_bracket(codestart, utf8, recno);
7053    if (groupptr == NULL) errorcode = ERR53;    if (groupptr == NULL) errorcode = ERR53;
7054      else PUT(((uschar *)codestart), offset, groupptr - codestart);      else PUT(((uschar *)codestart), offset, (int)(groupptr - codestart));
7055    }    }
7056    
7057  /* 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 6711  if (errorcode != 0) Line 7106  if (errorcode != 0)
7106    {    {
7107    (pcre_free)(re);    (pcre_free)(re);
7108    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
7109    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = (int)(ptr - (const uschar *)pattern);
7110    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
7111    *errorptr = find_error_text(errorcode);    *errorptr = find_error_text(errorcode);
7112    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
# Line 6765  if (reqbyte >= 0 && Line 7160  if (reqbyte >= 0 &&
7160  /* Print out the compiled data if debugging is enabled. This is never the  /* Print out the compiled data if debugging is enabled. This is never the
7161  case when building a production library. */  case when building a production library. */
7162    
7163  #ifdef DEBUG  #ifdef PCRE_DEBUG
   
7164  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
7165    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
7166    
# Line 6803  if (code - codestart > length) Line 7197  if (code - codestart > length)
7197    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
7198    return NULL;    return NULL;
7199    }    }
7200  #endif   /* DEBUG */  #endif   /* PCRE_DEBUG */
7201    
7202  return (pcre *)re;  return (pcre *)re;
7203  }  }

Legend:
Removed from v.472  
changed lines
  Added in v.545

  ViewVC Help
Powered by ViewVC 1.1.5