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

Diff of /code/trunk/pcre_compile.c

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

revision 483 by ph10, Mon Jan 4 16:19:17 2010 UTC revision 552 by ph10, Wed Oct 13 10:15:41 2010 UTC
# Line 92  is 4 there is plenty of room. */ Line 92  is 4 there is plenty of room. */
92    
93  #define COMPILE_WORK_SIZE (4096)  #define COMPILE_WORK_SIZE (4096)
94    
95    /* The overrun tests check for a slightly smaller size so that they detect the
96    overrun before it actually does run off the end of the data block. */
97    
98    #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100)
99    
100    
101  /* Table for handling escaped characters in the range '0'-'z'. Positive returns  /* Table for handling escaped characters in the range '0'-'z'. Positive returns
102  are simple data values; negative values are for special things like \d and so  are simple data values; negative values are for special things like \d and so
# Line 119  static const short int escapes[] = { Line 124  static const short int escapes[] = {
124       -ESC_H,                  0,       -ESC_H,                  0,
125       0,                       -ESC_K,       0,                       -ESC_K,
126       0,                       0,       0,                       0,
127       0,                       0,       -ESC_N,                  0,
128       -ESC_P,                  -ESC_Q,       -ESC_P,                  -ESC_Q,
129       -ESC_R,                  -ESC_S,       -ESC_R,                  -ESC_S,
130       0,                       0,       0,                       0,
# Line 166  static const short int escapes[] = { Line 171  static const short int escapes[] = {
171  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',
172  /*  C0 */   '{',-ESC_A, -ESC_B,  -ESC_C, -ESC_D,-ESC_E,      0, -ESC_G,  /*  C0 */   '{',-ESC_A, -ESC_B,  -ESC_C, -ESC_D,-ESC_E,      0, -ESC_G,
173  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,
174  /*  D0 */   '}',     0, -ESC_K,       0,      0,     0,      0, -ESC_P,  /*  D0 */   '}',     0, -ESC_K,       0,      0,-ESC_N,      0, -ESC_P,
175  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,
176  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,
177  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,
# Line 183  string is built from string macros so th Line 188  string is built from string macros so th
188  platforms. */  platforms. */
189    
190  typedef struct verbitem {  typedef struct verbitem {
191    int   len;    int   len;                 /* Length of verb name */
192    int   op;    int   op;                  /* Op when no arg, or -1 if arg mandatory */
193      int   op_arg;              /* Op when arg present, or -1 if not allowed */
194  } verbitem;  } verbitem;
195    
196  static const char verbnames[] =  static const char verbnames[] =
197      "\0"                       /* Empty name is a shorthand for MARK */
198      STRING_MARK0
199    STRING_ACCEPT0    STRING_ACCEPT0
200    STRING_COMMIT0    STRING_COMMIT0
201    STRING_F0    STRING_F0
# Line 197  static const char verbnames[] = Line 205  static const char verbnames[] =
205    STRING_THEN;    STRING_THEN;
206    
207  static const verbitem verbs[] = {  static const verbitem verbs[] = {
208    { 6, OP_ACCEPT },    { 0, -1,        OP_MARK },
209    { 6, OP_COMMIT },    { 4, -1,        OP_MARK },
210    { 1, OP_FAIL },    { 6, OP_ACCEPT, -1 },
211    { 4, OP_FAIL },    { 6, OP_COMMIT, -1 },
212    { 5, OP_PRUNE },    { 1, OP_FAIL,   -1 },
213    { 4, OP_SKIP  },    { 4, OP_FAIL,   -1 },
214    { 4, OP_THEN  }    { 5, OP_PRUNE,  OP_PRUNE_ARG },
215      { 4, OP_SKIP,   OP_SKIP_ARG  },
216      { 4, OP_THEN,   OP_THEN_ARG  }
217  };  };
218    
219  static const int verbcount = sizeof(verbs)/sizeof(verbitem);  static const int verbcount = sizeof(verbs)/sizeof(verbitem);
# Line 251  static const int posix_class_maps[] = { Line 261  static const int posix_class_maps[] = {
261    cbit_xdigit,-1,          0              /* xdigit */    cbit_xdigit,-1,          0              /* xdigit */
262  };  };
263    
264    /* Table of substitutes for \d etc when PCRE_UCP is set. The POSIX class
265    substitutes must be in the order of the names, defined above, and there are
266    both positive and negative cases. NULL means no substitute. */
267    
268    #ifdef SUPPORT_UCP
269    static const uschar *substitutes[] = {
270      (uschar *)"\\P{Nd}",    /* \D */
271      (uschar *)"\\p{Nd}",    /* \d */
272      (uschar *)"\\P{Xsp}",   /* \S */       /* NOTE: Xsp is Perl space */
273      (uschar *)"\\p{Xsp}",   /* \s */
274      (uschar *)"\\P{Xwd}",   /* \W */
275      (uschar *)"\\p{Xwd}"    /* \w */
276    };
277    
278    static const uschar *posix_substitutes[] = {
279      (uschar *)"\\p{L}",     /* alpha */
280      (uschar *)"\\p{Ll}",    /* lower */
281      (uschar *)"\\p{Lu}",    /* upper */
282      (uschar *)"\\p{Xan}",   /* alnum */
283      NULL,                   /* ascii */
284      (uschar *)"\\h",        /* blank */
285      NULL,                   /* cntrl */
286      (uschar *)"\\p{Nd}",    /* digit */
287      NULL,                   /* graph */
288      NULL,                   /* print */
289      NULL,                   /* punct */
290      (uschar *)"\\p{Xps}",   /* space */    /* NOTE: Xps is POSIX space */
291      (uschar *)"\\p{Xwd}",   /* word */
292      NULL,                   /* xdigit */
293      /* Negated cases */
294      (uschar *)"\\P{L}",     /* ^alpha */
295      (uschar *)"\\P{Ll}",    /* ^lower */
296      (uschar *)"\\P{Lu}",    /* ^upper */
297      (uschar *)"\\P{Xan}",   /* ^alnum */
298      NULL,                   /* ^ascii */
299      (uschar *)"\\H",        /* ^blank */
300      NULL,                   /* ^cntrl */
301      (uschar *)"\\P{Nd}",    /* ^digit */
302      NULL,                   /* ^graph */
303      NULL,                   /* ^print */
304      NULL,                   /* ^punct */
305      (uschar *)"\\P{Xps}",   /* ^space */   /* NOTE: Xps is POSIX space */
306      (uschar *)"\\P{Xwd}",   /* ^word */
307      NULL                    /* ^xdigit */
308    };
309    #define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))
310    #endif
311    
312  #define STRING(a)  # a  #define STRING(a)  # a
313  #define XSTRING(s) STRING(s)  #define XSTRING(s) STRING(s)
# Line 263  the number of relocations needed when a Line 320  the number of relocations needed when a
320  it is now one long string. We cannot use a table of offsets, because the  it is now one long string. We cannot use a table of offsets, because the
321  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we
322  simply count through to the one we want - this isn't a performance issue  simply count through to the one we want - this isn't a performance issue
323  because these strings are used only when there is a compilation error. */  because these strings are used only when there is a compilation error.
324    
325    Each substring ends with \0 to insert a null character. This includes the final
326    substring, so that the whole string ends with \0\0, which can be detected when
327    counting through. */
328    
329  static const char error_texts[] =  static const char error_texts[] =
330    "no error\0"    "no error\0"
# Line 310  static const char error_texts[] = Line 371  static const char error_texts[] =
371    /* 35 */    /* 35 */
372    "invalid condition (?(0)\0"    "invalid condition (?(0)\0"
373    "\\C not allowed in lookbehind assertion\0"    "\\C not allowed in lookbehind assertion\0"
374    "PCRE does not support \\L, \\l, \\N, \\U, or \\u\0"    "PCRE does not support \\L, \\l, \\N{name}, \\U, or \\u\0"
375    "number after (?C is > 255\0"    "number after (?C is > 255\0"
376    "closing ) for (?C expected\0"    "closing ) for (?C expected\0"
377    /* 40 */    /* 40 */
# Line 336  static const char error_texts[] = Line 397  static const char error_texts[] =
397    "inconsistent NEWLINE options\0"    "inconsistent NEWLINE options\0"
398    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"
399    "a numbered reference must not be zero\0"    "a numbered reference must not be zero\0"
400    "(*VERB) with an argument is not supported\0"    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"
401    /* 60 */    /* 60 */
402    "(*VERB) not recognized\0"    "(*VERB) not recognized\0"
403    "number is too big\0"    "number is too big\0"
# Line 344  static const char error_texts[] = Line 405  static const char error_texts[] =
405    "digit expected after (?+\0"    "digit expected after (?+\0"
406    "] is an invalid data character in JavaScript compatibility mode\0"    "] is an invalid data character in JavaScript compatibility mode\0"
407    /* 65 */    /* 65 */
408    "different names for subpatterns of the same number are not allowed";    "different names for subpatterns of the same number are not allowed\0"
409      "(*MARK) must have an argument\0"
410      "this version of PCRE is not compiled with PCRE_UCP support\0"
411      ;
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 503  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 574  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 812  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 1050  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 1080  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 1202  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 1443  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 1602  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 1626  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          code += code[1];
1728          break;
1729    
1730          case OP_THEN_ARG:
1731          code += code[1+LINK_SIZE];
1732          break;
1733        }        }
1734    
1735      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
# Line 1697  for (;;) Line 1801  for (;;)
1801    
1802    /* 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
1803    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
1804    two bytes of parameters. */    two bytes of parameters, and for MARK/PRUNE/SKIP/THEN with an argument, we
1805      must add in its length. */
1806    
1807    else    else
1808      {      {
# Line 1721  for (;;) Line 1826  for (;;)
1826        case OP_TYPEEXACT:        case OP_TYPEEXACT:
1827        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1828        break;        break;
1829    
1830          case OP_MARK:
1831          case OP_PRUNE_ARG:
1832          case OP_SKIP_ARG:
1833          code += code[1];
1834          break;
1835    
1836          case OP_THEN_ARG:
1837          code += code[1+LINK_SIZE];
1838          break;
1839        }        }
1840    
1841      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
# Line 1777  Arguments: Line 1892  Arguments:
1892    code        points to start of search    code        points to start of search
1893    endcode     points to where to stop    endcode     points to where to stop
1894    utf8        TRUE if in UTF8 mode    utf8        TRUE if in UTF8 mode
1895      cd          contains pointers to tables etc.
1896    
1897  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
1898  */  */
1899    
1900  static BOOL  static BOOL
1901  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8)  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,
1902      compile_data *cd)
1903  {  {
1904  register int c;  register int c;
1905  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);
# Line 1813  for (code = first_significant_code(code Line 1930  for (code = first_significant_code(code
1930      continue;      continue;
1931      }      }
1932    
1933      /* For a recursion/subroutine call, if its end has been reached, which
1934      implies a subroutine call, we can scan it. */
1935    
1936      if (c == OP_RECURSE)
1937        {
1938        BOOL empty_branch = FALSE;
1939        const uschar *scode = cd->start_code + GET(code, 1);
1940        if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
1941        do
1942          {
1943          if (could_be_empty_branch(scode, endcode, utf8, cd))
1944            {
1945            empty_branch = TRUE;
1946            break;
1947            }
1948          scode += GET(scode, 1);
1949          }
1950        while (*scode == OP_ALT);
1951        if (!empty_branch) return FALSE;  /* All branches are non-empty */
1952        continue;
1953        }
1954    
1955    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1956    
1957    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
# Line 1831  for (code = first_significant_code(code Line 1970  for (code = first_significant_code(code
1970        empty_branch = FALSE;        empty_branch = FALSE;
1971        do        do
1972          {          {
1973          if (!empty_branch && could_be_empty_branch(code, endcode, utf8))          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))
1974            empty_branch = TRUE;            empty_branch = TRUE;
1975          code += GET(code, 1);          code += GET(code, 1);
1976          }          }
# Line 1965  for (code = first_significant_code(code Line 2104  for (code = first_significant_code(code
2104      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];
2105      break;      break;
2106  #endif  #endif
2107    
2108        /* MARK, and PRUNE/SKIP/THEN with an argument must skip over the argument
2109        string. */
2110    
2111        case OP_MARK:
2112        case OP_PRUNE_ARG:
2113        case OP_SKIP_ARG:
2114        code += code[1];
2115        break;
2116    
2117        case OP_THEN_ARG:
2118        code += code[1+LINK_SIZE];
2119        break;
2120    
2121        /* None of the remaining opcodes are required to match a character. */
2122    
2123        default:
2124        break;
2125      }      }
2126    }    }
2127    
# Line 1987  Arguments: Line 2144  Arguments:
2144    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2145    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2146    utf8        TRUE if in UTF-8 mode    utf8        TRUE if in UTF-8 mode
2147      cd          pointers to tables etc
2148    
2149  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2150  */  */
2151    
2152  static BOOL  static BOOL
2153  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,
2154    BOOL utf8)    BOOL utf8, compile_data *cd)
2155  {  {
2156  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2157    {    {
2158    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))
2159      return FALSE;      return FALSE;
2160    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2161    }    }
# Line 2180  auto_callout(uschar *code, const uschar Line 2338  auto_callout(uschar *code, const uschar
2338  {  {
2339  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2340  *code++ = 255;  *code++ = 255;
2341  PUT(code, 0, ptr - cd->start_pattern);  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2342  PUT(code, LINK_SIZE, 0);                /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2343  return code + 2*LINK_SIZE;  return code + 2*LINK_SIZE;
2344  }  }
2345    
# Line 2206  Returns:             nothing Line 2364  Returns:             nothing
2364  static void  static void
2365  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)
2366  {  {
2367  int length = ptr - cd->start_pattern - GET(previous_callout, 2);  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2368  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
2369  }  }
2370    
# Line 2256  for (++c; c <= d; c++) Line 2414  for (++c; c <= d; c++)
2414    
2415  return TRUE;  return TRUE;
2416  }  }
2417    
2418    
2419    
2420    /*************************************************
2421    *        Check a character and a property        *
2422    *************************************************/
2423    
2424    /* This function is called by check_auto_possessive() when a property item
2425    is adjacent to a fixed character.
2426    
2427    Arguments:
2428      c            the character
2429      ptype        the property type
2430      pdata        the data for the type
2431      negated      TRUE if it's a negated property (\P or \p{^)
2432    
2433    Returns:       TRUE if auto-possessifying is OK
2434    */
2435    
2436    static BOOL
2437    check_char_prop(int c, int ptype, int pdata, BOOL negated)
2438    {
2439    const ucd_record *prop = GET_UCD(c);
2440    switch(ptype)
2441      {
2442      case PT_LAMP:
2443      return (prop->chartype == ucp_Lu ||
2444              prop->chartype == ucp_Ll ||
2445              prop->chartype == ucp_Lt) == negated;
2446    
2447      case PT_GC:
2448      return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;
2449    
2450      case PT_PC:
2451      return (pdata == prop->chartype) == negated;
2452    
2453      case PT_SC:
2454      return (pdata == prop->script) == negated;
2455    
2456      /* These are specials */
2457    
2458      case PT_ALNUM:
2459      return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2460              _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;
2461    
2462      case PT_SPACE:    /* Perl space */
2463      return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2464              c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2465              == negated;
2466    
2467      case PT_PXSPACE:  /* POSIX space */
2468      return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||
2469              c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2470              c == CHAR_FF || c == CHAR_CR)
2471              == negated;
2472    
2473      case PT_WORD:
2474      return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||
2475              _pcre_ucp_gentype[prop->chartype] == ucp_N ||
2476              c == CHAR_UNDERSCORE) == negated;
2477      }
2478    return FALSE;
2479    }
2480  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
2481    
2482    
# Line 2269  whether the next thing could possibly ma Line 2490  whether the next thing could possibly ma
2490  sense to automatically possessify the repeated item.  sense to automatically possessify the repeated item.
2491    
2492  Arguments:  Arguments:
2493    op_code       the repeated op code    previous      pointer to the repeated opcode
   this          data for this item, depends on the opcode  
2494    utf8          TRUE in UTF-8 mode    utf8          TRUE in UTF-8 mode
   utf8_char     used for utf8 character bytes, NULL if not relevant  
2495    ptr           next character in pattern    ptr           next character in pattern
2496    options       options bits    options       options bits
2497    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2281  Returns:        TRUE if possessifying is Line 2500  Returns:        TRUE if possessifying is
2500  */  */
2501    
2502  static BOOL  static BOOL
2503  check_auto_possessive(int op_code, int item, BOOL utf8, uschar *utf8_char,  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,
2504    const uschar *ptr, int options, compile_data *cd)    int options, compile_data *cd)
2505  {  {
2506  int next;  int c, next;
2507    int op_code = *previous++;
2508    
2509  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
2510    
# Line 2345  if (*ptr == CHAR_ASTERISK || *ptr == CHA Line 2565  if (*ptr == CHAR_ASTERISK || *ptr == CHA
2565    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)
2566      return FALSE;      return FALSE;
2567    
2568  /* 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
2569  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. */  
2570    
2571  if (next >= 0) switch(op_code)  if (next >= 0) switch(op_code)
2572    {    {
2573    case OP_CHAR:    case OP_CHAR:
2574  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2575    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2576  #else  #else
2577    (void)(utf8_char);  /* Keep compiler happy by referencing function argument */    c = *previous;
2578  #endif  #endif
2579    return item != next;    return c != next;
2580    
2581    /* 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
2582    Unicode property support, we can use it to test the other case of    Unicode property support, we can use it to test the other case of
# Line 2369  if (next >= 0) switch(op_code) Line 2584  if (next >= 0) switch(op_code)
2584    
2585    case OP_CHARNC:    case OP_CHARNC:
2586  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2587    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2588    #else
2589      c = *previous;
2590  #endif  #endif
2591    if (item == next) return FALSE;    if (c == next) return FALSE;
2592  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2593    if (utf8)    if (utf8)
2594      {      {
# Line 2382  if (next >= 0) switch(op_code) Line 2599  if (next >= 0) switch(op_code)
2599  #else  #else
2600      othercase = NOTACHAR;      othercase = NOTACHAR;
2601  #endif  #endif
2602      return (unsigned int)item != othercase;      return (unsigned int)c != othercase;
2603      }      }
2604    else    else
2605  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2606    return (item != cd->fcc[next]);  /* Non-UTF-8 mode */    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */
2607    
2608    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, its data is always a single-byte character. */
2609    
2610    case OP_NOT:    case OP_NOT:
2611    if (item == next) return TRUE;    if ((c = *previous) == next) return TRUE;
2612    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2613  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2614    if (utf8)    if (utf8)
# Line 2403  if (next >= 0) switch(op_code) Line 2620  if (next >= 0) switch(op_code)
2620  #else  #else
2621      othercase = NOTACHAR;      othercase = NOTACHAR;
2622  #endif  #endif
2623      return (unsigned int)item == othercase;      return (unsigned int)c == othercase;
2624      }      }
2625    else    else
2626  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF8 */
2627    return (item == cd->fcc[next]);  /* Non-UTF-8 mode */    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */
2628    
2629      /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.
2630      When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
2631    
2632    case OP_DIGIT:    case OP_DIGIT:
2633    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;
# Line 2450  if (next >= 0) switch(op_code) Line 2670  if (next >= 0) switch(op_code)
2670      case 0x202f:      case 0x202f:
2671      case 0x205f:      case 0x205f:
2672      case 0x3000:      case 0x3000:
2673      return op_code != OP_HSPACE;      return op_code == OP_NOT_HSPACE;
2674      default:      default:
2675      return op_code == OP_HSPACE;      return op_code != OP_NOT_HSPACE;
2676      }      }
2677    
2678      case OP_ANYNL:
2679    case OP_VSPACE:    case OP_VSPACE:
2680    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
2681    switch(next)    switch(next)
# Line 2466  if (next >= 0) switch(op_code) Line 2687  if (next >= 0) switch(op_code)
2687      case 0x85:      case 0x85:
2688      case 0x2028:      case 0x2028:
2689      case 0x2029:      case 0x2029:
2690      return op_code != OP_VSPACE;      return op_code == OP_NOT_VSPACE;
2691      default:      default:
2692      return op_code == OP_VSPACE;      return op_code != OP_NOT_VSPACE;
2693      }      }
2694    
2695    #ifdef SUPPORT_UCP
2696      case OP_PROP:
2697      return check_char_prop(next, previous[0], previous[1], FALSE);
2698    
2699      case OP_NOTPROP:
2700      return check_char_prop(next, previous[0], previous[1], TRUE);
2701    #endif
2702    
2703    default:    default:
2704    return FALSE;    return FALSE;
2705    }    }
2706    
2707    
2708  /* 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
2709    is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are
2710    generated only when PCRE_UCP is *not* set, that is, when only ASCII
2711    characteristics are recognized. Similarly, the opcodes OP_DIGIT etc. are
2712    replaced by OP_PROP codes when PCRE_UCP is set. */
2713    
2714  switch(op_code)  switch(op_code)
2715    {    {
2716    case OP_CHAR:    case OP_CHAR:
2717    case OP_CHARNC:    case OP_CHARNC:
2718  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2719    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    GETCHARTEST(c, previous);
2720    #else
2721      c = *previous;
2722  #endif  #endif
2723    switch(-next)    switch(-next)
2724      {      {
2725      case ESC_d:      case ESC_d:
2726      return item > 127 || (cd->ctypes[item] & ctype_digit) == 0;      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;
2727    
2728      case ESC_D:      case ESC_D:
2729      return item <= 127 && (cd->ctypes[item] & ctype_digit) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;
2730    
2731      case ESC_s:      case ESC_s:
2732      return item > 127 || (cd->ctypes[item] & ctype_space) == 0;      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;
2733    
2734      case ESC_S:      case ESC_S:
2735      return item <= 127 && (cd->ctypes[item] & ctype_space) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;
2736    
2737      case ESC_w:      case ESC_w:
2738      return item > 127 || (cd->ctypes[item] & ctype_word) == 0;      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;
2739    
2740      case ESC_W:      case ESC_W:
2741      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;
2742    
2743      case ESC_h:      case ESC_h:
2744      case ESC_H:      case ESC_H:
2745      switch(item)      switch(c)
2746        {        {
2747        case 0x09:        case 0x09:
2748        case 0x20:        case 0x20:
# Line 2535  switch(op_code) Line 2770  switch(op_code)
2770    
2771      case ESC_v:      case ESC_v:
2772      case ESC_V:      case ESC_V:
2773      switch(item)      switch(c)
2774        {        {
2775        case 0x0a:        case 0x0a:
2776        case 0x0b:        case 0x0b:
# Line 2549  switch(op_code) Line 2784  switch(op_code)
2784        return -next == ESC_v;        return -next == ESC_v;
2785        }        }
2786    
2787        /* When PCRE_UCP is set, these values get generated for \d etc. Find
2788        their substitutions and process them. The result will always be either
2789        -ESC_p or -ESC_P. Then fall through to process those values. */
2790    
2791    #ifdef SUPPORT_UCP
2792        case ESC_du:
2793        case ESC_DU:
2794        case ESC_wu:
2795        case ESC_WU:
2796        case ESC_su:
2797        case ESC_SU:
2798          {
2799          int temperrorcode = 0;
2800          ptr = substitutes[-next - ESC_DU];
2801          next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);
2802          if (temperrorcode != 0) return FALSE;
2803          ptr++;    /* For compatibility */
2804          }
2805        /* Fall through */
2806    
2807        case ESC_p:
2808        case ESC_P:
2809          {
2810          int ptype, pdata, errorcodeptr;
2811          BOOL negated;
2812    
2813          ptr--;      /* Make ptr point at the p or P */
2814          ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);
2815          if (ptype < 0) return FALSE;
2816          ptr++;      /* Point past the final curly ket */
2817    
2818          /* If the property item is optional, we have to give up. (When generated
2819          from \d etc by PCRE_UCP, this test will have been applied much earlier,
2820          to the original \d etc. At this point, ptr will point to a zero byte. */
2821    
2822          if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
2823            strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
2824              return FALSE;
2825    
2826          /* Do the property check. */
2827    
2828          return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);
2829          }
2830    #endif
2831    
2832      default:      default:
2833      return FALSE;      return FALSE;
2834      }      }
2835    
2836      /* In principle, support for Unicode properties should be integrated here as
2837      well. It means re-organizing the above code so as to get hold of the property
2838      values before switching on the op-code. However, I wonder how many patterns
2839      combine ASCII \d etc with Unicode properties? (Note that if PCRE_UCP is set,
2840      these op-codes are never generated.) */
2841    
2842    case OP_DIGIT:    case OP_DIGIT:
2843    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2844           next == -ESC_h || next == -ESC_v;           next == -ESC_h || next == -ESC_v || next == -ESC_R;
2845    
2846    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2847    return next == -ESC_d;    return next == -ESC_d;
2848    
2849    case OP_WHITESPACE:    case OP_WHITESPACE:
2850    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;
2851    
2852    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2853    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2854    
2855    case OP_HSPACE:    case OP_HSPACE:
2856    return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||
2857             next == -ESC_w || next == -ESC_v || next == -ESC_R;
2858    
2859    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
2860    return next == -ESC_h;    return next == -ESC_h;
2861    
2862    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
2863      case OP_ANYNL:
2864    case OP_VSPACE:    case OP_VSPACE:
2865    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2866    
2867    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
2868    return next == -ESC_v;    return next == -ESC_v || next == -ESC_R;
2869    
2870    case OP_WORDCHAR:    case OP_WORDCHAR:
2871    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||
2872             next == -ESC_v || next == -ESC_R;
2873    
2874    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2875    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2644  BOOL inescq = FALSE; Line 2933  BOOL inescq = FALSE;
2933  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstbyte = FALSE;
2934  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
2935  const uschar *tempptr;  const uschar *tempptr;
2936    const uschar *nestptr = NULL;
2937  uschar *previous = NULL;  uschar *previous = NULL;
2938  uschar *previous_callout = NULL;  uschar *previous_callout = NULL;
2939  uschar *save_hwm = NULL;  uschar *save_hwm = NULL;
# Line 2714  for (;; ptr++) Line 3004  for (;; ptr++)
3004    
3005    c = *ptr;    c = *ptr;
3006    
3007      /* If we are at the end of a nested substitution, revert to the outer level
3008      string. Nesting only happens one level deep. */
3009    
3010      if (c == 0 && nestptr != NULL)
3011        {
3012        ptr = nestptr;
3013        nestptr = NULL;
3014        c = *ptr;
3015        }
3016    
3017    /* 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
3018    previous cycle of this loop. */    previous cycle of this loop. */
3019    
# Line 2722  for (;; ptr++) Line 3022  for (;; ptr++)
3022  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3023      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3024  #endif  #endif
3025      if (code > cd->start_workspace + COMPILE_WORK_SIZE) /* Check for overrun */      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */
3026        {        {
3027        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3028        goto FAILED;        goto FAILED;
# Line 2744  for (;; ptr++) Line 3044  for (;; ptr++)
3044        goto FAILED;        goto FAILED;
3045        }        }
3046    
3047      *lengthptr += code - last_code;      *lengthptr += (int)(code - last_code);
3048      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));
3049    
3050      /* If "previous" is set and it is not at the start of the work space, move      /* If "previous" is set and it is not at the start of the work space, move
# Line 2771  for (;; ptr++) Line 3071  for (;; ptr++)
3071    /* 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
3072    reference list. */    reference list. */
3073    
3074    else if (cd->hwm > cd->start_workspace + COMPILE_WORK_SIZE)    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)
3075      {      {
3076      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3077      goto FAILED;      goto FAILED;
# Line 2862  for (;; ptr++) Line 3162  for (;; ptr++)
3162          *errorcodeptr = ERR20;          *errorcodeptr = ERR20;
3163          goto FAILED;          goto FAILED;
3164          }          }
3165        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += (int)(code - last_code);   /* To include callout length */
3166        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
3167        }        }
3168      return TRUE;      return TRUE;
# Line 3067  for (;; ptr++) Line 3367  for (;; ptr++)
3367            ptr++;            ptr++;
3368            }            }
3369    
3370          posix_class = check_posix_name(ptr, tempptr - ptr);          posix_class = check_posix_name(ptr, (int)(tempptr - ptr));
3371          if (posix_class < 0)          if (posix_class < 0)
3372            {            {
3373            *errorcodeptr = ERR30;            *errorcodeptr = ERR30;
# Line 3081  for (;; ptr++) Line 3381  for (;; ptr++)
3381          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
3382            posix_class = 0;            posix_class = 0;
3383    
3384          /* 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
3385          because we may be adding and subtracting from it, and we don't want to          different escape sequences that use Unicode properties. */
3386          subtract bits that may be in the main map already. At the end we or the  
3387          result into the bit map that is being built. */  #ifdef SUPPORT_UCP
3388            if ((options & PCRE_UCP) != 0)
3389              {
3390              int pc = posix_class + ((local_negate)? POSIX_SUBSIZE/2 : 0);
3391              if (posix_substitutes[pc] != NULL)
3392                {
3393                nestptr = tempptr + 1;
3394                ptr = posix_substitutes[pc] - 1;
3395                continue;
3396                }
3397              }
3398    #endif
3399            /* In the non-UCP case, we build the bit map for the POSIX class in a
3400            chunk of local store because we may be adding and subtracting from it,
3401            and we don't want to subtract bits that may be in the main map already.
3402            At the end we or the result into the bit map that is being built. */
3403    
3404          posix_class *= 3;          posix_class *= 3;
3405    
# Line 3128  for (;; ptr++) Line 3443  for (;; ptr++)
3443    
3444        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
3445        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
3446        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
3447        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
3448        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
3449        character in them, so set class_charcount bigger than one. */        are either treated as literal characters (by default), or are faulted if
3450          PCRE_EXTRA is set. */
3451    
3452        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
3453          {          {
3454          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3455          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
3456    
3457          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 */  
3458          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
3459            {            {
3460            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3157  for (;; ptr++) Line 3471  for (;; ptr++)
3471            register const uschar *cbits = cd->cbits;            register const uschar *cbits = cd->cbits;
3472            class_charcount += 2;     /* Greater than 1 is what matters */            class_charcount += 2;     /* Greater than 1 is what matters */
3473    
3474            /* Save time by not doing this in the pre-compile phase. */            switch (-c)
   
           if (lengthptr == NULL) switch (-c)  
3475              {              {
3476    #ifdef SUPPORT_UCP
3477                case ESC_du:     /* These are the values given for \d etc */
3478                case ESC_DU:     /* when PCRE_UCP is set. We replace the */
3479                case ESC_wu:     /* escape sequence with an appropriate \p */
3480                case ESC_WU:     /* or \P to test Unicode properties instead */
3481                case ESC_su:     /* of the default ASCII testing. */
3482                case ESC_SU:
3483                nestptr = ptr;
3484                ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
3485                class_charcount -= 2;                /* Undo! */
3486                continue;
3487    #endif
3488              case ESC_d:              case ESC_d:
3489              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_digit];
3490              continue;              continue;
# Line 3179  for (;; ptr++) Line 3503  for (;; ptr++)
3503              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
3504              continue;              continue;
3505    
3506                /* Perl 5.004 onwards omits VT from \s, but we must preserve it
3507                if it was previously set by something earlier in the character
3508                class. */
3509    
3510              case ESC_s:              case ESC_s:
3511              for (c = 0; c < 32; c++) classbits[c] |= cbits[c+cbit_space];              classbits[0] |= cbits[cbit_space];
3512              classbits[1] &= ~0x08;   /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= cbits[cbit_space+1] & ~0x08;
3513                for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];
3514              continue;              continue;
3515    
3516              case ESC_S:              case ESC_S:
# Line 3190  for (;; ptr++) Line 3519  for (;; ptr++)
3519              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
3520              continue;              continue;
3521    
3522              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)  
             {  
3523              SETBIT(classbits, 0x09); /* VT */              SETBIT(classbits, 0x09); /* VT */
3524              SETBIT(classbits, 0x20); /* SPACE */              SETBIT(classbits, 0x20); /* SPACE */
3525              SETBIT(classbits, 0xa0); /* NSBP */              SETBIT(classbits, 0xa0); /* NSBP */
# Line 3227  for (;; ptr++) Line 3543  for (;; ptr++)
3543                }                }
3544  #endif  #endif
3545              continue;              continue;
             }  
3546    
3547            if (-c == ESC_H)              case ESC_H:
             {  
3548              for (c = 0; c < 32; c++)              for (c = 0; c < 32; c++)
3549                {                {
3550                int x = 0xff;                int x = 0xff;
# Line 3272  for (;; ptr++) Line 3586  for (;; ptr++)
3586                }                }
3587  #endif  #endif
3588              continue;              continue;
             }  
3589    
3590            if (-c == ESC_v)              case ESC_v:
             {  
3591              SETBIT(classbits, 0x0a); /* LF */              SETBIT(classbits, 0x0a); /* LF */
3592              SETBIT(classbits, 0x0b); /* VT */              SETBIT(classbits, 0x0b); /* VT */
3593              SETBIT(classbits, 0x0c); /* FF */              SETBIT(classbits, 0x0c); /* FF */
# Line 3291  for (;; ptr++) Line 3603  for (;; ptr++)
3603                }                }
3604  #endif  #endif
3605              continue;              continue;
             }  
3606    
3607            if (-c == ESC_V)              case ESC_V:
             {  
3608              for (c = 0; c < 32; c++)              for (c = 0; c < 32; c++)
3609                {                {
3610                int x = 0xff;                int x = 0xff;
# Line 3324  for (;; ptr++) Line 3634  for (;; ptr++)
3634                }                }
3635  #endif  #endif
3636              continue;              continue;
             }  
   
           /* We need to deal with \P and \p in both phases. */  
3637    
3638  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3639            if (-c == ESC_p || -c == ESC_P)              case ESC_p:
3640              {              case ESC_P:
3641              BOOL negated;                {
3642              int pdata;                BOOL negated;
3643              int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int pdata;
3644              if (ptype < 0) goto FAILED;                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
3645              class_utf8 = TRUE;                if (ptype < 0) goto FAILED;
3646              *class_utf8data++ = ((-c == ESC_p) != negated)?                class_utf8 = TRUE;
3647                XCL_PROP : XCL_NOTPROP;                *class_utf8data++ = ((-c == ESC_p) != negated)?
3648              *class_utf8data++ = ptype;                  XCL_PROP : XCL_NOTPROP;
3649              *class_utf8data++ = pdata;                *class_utf8data++ = ptype;
3650              class_charcount -= 2;   /* Not a < 256 character */                *class_utf8data++ = pdata;
3651              continue;                class_charcount -= 2;   /* Not a < 256 character */
3652              }                continue;
3653                  }
3654  #endif  #endif
3655            /* Unrecognized escapes are faulted if PCRE is running in its              /* Unrecognized escapes are faulted if PCRE is running in its
3656            strict mode. By default, for compatibility with Perl, they are              strict mode. By default, for compatibility with Perl, they are
3657            treated as literals. */              treated as literals. */
3658    
3659            if ((options & PCRE_EXTRA) != 0)              default:
3660              {              if ((options & PCRE_EXTRA) != 0)
3661              *errorcodeptr = ERR7;                {
3662              goto FAILED;                *errorcodeptr = ERR7;
3663                  goto FAILED;
3664                  }
3665                class_charcount -= 2;  /* Undo the default count from above */
3666                c = *ptr;              /* Get the final character and fall through */
3667                break;
3668              }              }
   
           class_charcount -= 2;  /* Undo the default count from above */  
           c = *ptr;              /* Get the final character and fall through */  
3669            }            }
3670    
3671          /* Fall through if we have a single character (c >= 0). This may be          /* Fall through if we have a single character (c >= 0). This may be
# Line 3425  for (;; ptr++) Line 3735  for (;; ptr++)
3735            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3736            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3737    
3738            /* \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 */  
3739    
3740            if (d < 0)            if (d < 0)
3741              {              {
3742              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  
3743                {                {
3744                ptr = oldptr;                ptr = oldptr;
3745                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto LONE_SINGLE_CHARACTER;  /* A few lines below */
# Line 3598  for (;; ptr++) Line 3905  for (;; ptr++)
3905          }          }
3906        }        }
3907    
3908      /* 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.
3909        If we are at the end of an internal nested string, revert to the outer
3910        string. */
3911    
3912        while (((c = *(++ptr)) != 0 ||
3913               (nestptr != NULL &&
3914                 (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&
3915               (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
3916    
3917      while ((c = *(++ptr)) != 0 && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));      /* Check for missing terminating ']' */
3918    
3919      if (c == 0)                          /* Missing terminating ']' */      if (c == 0)
3920        {        {
3921        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
3922        goto FAILED;        goto FAILED;
3923        }        }
3924    
   
 /* 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  
   
   
3925      /* 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
3926      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
3927      use of \p or \P, in other words, no use of any XCLASS features, we can      use of \p or \P, in other words, no use of any XCLASS features, we can
# Line 3690  we set the flag only if there is a liter Line 3985  we set the flag only if there is a liter
3985    
3986      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3987      extended class, with its own opcode, unless there was a negated special      extended class, with its own opcode, unless there was a negated special
3988      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
3989      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
3990      (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
3991      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
3992        actual compiled code. */
3993    
3994  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3995      if (class_utf8 && !should_flip_negation)      if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))
3996        {        {
3997        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3998        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3722  we set the flag only if there is a liter Line 4018  we set the flag only if there is a liter
4018        }        }
4019  #endif  #endif
4020    
4021      /* 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
4022      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
4023      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
4024      map into the code vector, negating it if necessary. */      (non-UCP) in the class. Then copy the 32-byte map into the code vector,
4025        negating it if necessary. */
4026    
4027      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
4028      if (negate_class)      if (negate_class)
# Line 3849  we set the flag only if there is a liter Line 4146  we set the flag only if there is a liter
4146    
4147        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4148            repeat_max < 0 &&            repeat_max < 0 &&
4149            check_auto_possessive(*previous, c, utf8, utf8_char, ptr + 1,            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
             options, cd))  
4150          {          {
4151          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4152          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 3871  we set the flag only if there is a liter Line 4167  we set the flag only if there is a liter
4167        c = previous[1];        c = previous[1];
4168        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4169            repeat_max < 0 &&            repeat_max < 0 &&
4170            check_auto_possessive(OP_NOT, c, utf8, NULL, ptr + 1, options, cd))            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
4171          {          {
4172          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4173          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 3895  we set the flag only if there is a liter Line 4191  we set the flag only if there is a liter
4191    
4192        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4193            repeat_max < 0 &&            repeat_max < 0 &&
4194            check_auto_possessive(c, 0, utf8, NULL, ptr + 1, options, cd))            check_auto_possessive(previous, utf8, ptr + 1, options, cd))
4195          {          {
4196          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4197          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 4105  we set the flag only if there is a liter Line 4401  we set the flag only if there is a liter
4401        {        {
4402        register int i;        register int i;
4403        int ketoffset = 0;        int ketoffset = 0;
4404        int len = code - previous;        int len = (int)(code - previous);
4405        uschar *bralink = NULL;        uschar *bralink = NULL;
4406    
4407        /* Repeating a DEFINE group is pointless */        /* Repeating a DEFINE group is pointless */
# Line 4126  we set the flag only if there is a liter Line 4422  we set the flag only if there is a liter
4422          {          {
4423          register uschar *ket = previous;          register uschar *ket = previous;
4424          do ket += GET(ket, 1); while (*ket != OP_KET);          do ket += GET(ket, 1); while (*ket != OP_KET);
4425          ketoffset = code - ket;          ketoffset = (int)(code - ket);
4426          }          }
4427    
4428        /* The case of a zero minimum is special because of the need to stick        /* The case of a zero minimum is special because of the need to stick
# Line 4194  we set the flag only if there is a liter Line 4490  we set the flag only if there is a liter
4490            /* We chain together the bracket offset fields that have to be            /* We chain together the bracket offset fields that have to be
4491            filled in later when the ends of the brackets are reached. */            filled in later when the ends of the brackets are reached. */
4492    
4493            offset = (bralink == NULL)? 0 : previous - bralink;            offset = (bralink == NULL)? 0 : (int)(previous - bralink);
4494            bralink = previous;            bralink = previous;
4495            PUTINC(previous, 0, offset);            PUTINC(previous, 0, offset);
4496            }            }
# Line 4222  we set the flag only if there is a liter Line 4518  we set the flag only if there is a liter
4518              {              {
4519              int delta = (repeat_min - 1)*length_prevgroup;              int delta = (repeat_min - 1)*length_prevgroup;
4520              if ((INT64_OR_DOUBLE)(repeat_min - 1)*              if ((INT64_OR_DOUBLE)(repeat_min - 1)*
4521                    (INT64_OR_DOUBLE)length_prevgroup >                    (INT64_OR_DOUBLE)length_prevgroup >
4522                      (INT64_OR_DOUBLE)INT_MAX ||                      (INT64_OR_DOUBLE)INT_MAX ||
4523                  OFLOW_MAX - *lengthptr < delta)                  OFLOW_MAX - *lengthptr < delta)
4524                {                {
# Line 4269  we set the flag only if there is a liter Line 4565  we set the flag only if there is a liter
4565          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
4566          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
4567          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
4568          paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is          paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is
4569          a 64-bit integer type when available, otherwise double. */          a 64-bit integer type when available, otherwise double. */
4570    
4571          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
# Line 4303  we set the flag only if there is a liter Line 4599  we set the flag only if there is a liter
4599              {              {
4600              int offset;              int offset;
4601              *code++ = OP_BRA;              *code++ = OP_BRA;
4602              offset = (bralink == NULL)? 0 : code - bralink;              offset = (bralink == NULL)? 0 : (int)(code - bralink);
4603              bralink = code;              bralink = code;
4604              PUTINC(code, 0, offset);              PUTINC(code, 0, offset);
4605              }              }
# Line 4324  we set the flag only if there is a liter Line 4620  we set the flag only if there is a liter
4620          while (bralink != NULL)          while (bralink != NULL)
4621            {            {
4622            int oldlinkoffset;            int oldlinkoffset;
4623            int offset = code - bralink + 1;            int offset = (int)(code - bralink + 1);
4624            uschar *bra = code - offset;            uschar *bra = code - offset;
4625            oldlinkoffset = GET(bra, 1);            oldlinkoffset = GET(bra, 1);
4626            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;
# Line 4355  we set the flag only if there is a liter Line 4651  we set the flag only if there is a liter
4651            uschar *scode = bracode;            uschar *scode = bracode;
4652            do            do
4653              {              {
4654              if (could_be_empty_branch(scode, ketcode, utf8))              if (could_be_empty_branch(scode, ketcode, utf8, cd))
4655                {                {
4656                *bracode += OP_SBRA - OP_BRA;                *bracode += OP_SBRA - OP_BRA;
4657                break;                break;
# Line 4412  we set the flag only if there is a liter Line 4708  we set the flag only if there is a liter
4708  #endif  #endif
4709          }          }
4710    
4711        len = code - tempcode;        len = (int)(code - tempcode);
4712        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4713          {          {
4714          case OP_STAR:  *tempcode = OP_POSSTAR; break;          case OP_STAR:  *tempcode = OP_POSSTAR; break;
# Line 4430  we set the flag only if there is a liter Line 4726  we set the flag only if there is a liter
4726          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;          case OP_NOTQUERY: *tempcode = OP_NOTPOSQUERY; break;
4727          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;          case OP_NOTUPTO:  *tempcode = OP_NOTPOSUPTO; break;
4728    
4729            /* Because we are moving code along, we must ensure that any
4730            pending recursive references are updated. */
4731    
4732          default:          default:
4733            *code = OP_END;
4734            adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);
4735          memmove(tempcode + 1+LINK_SIZE, tempcode, len);          memmove(tempcode + 1+LINK_SIZE, tempcode, len);
4736          code += 1 + LINK_SIZE;          code += 1 + LINK_SIZE;
4737          len += 1 + LINK_SIZE;          len += 1 + LINK_SIZE;
# Line 4466  we set the flag only if there is a liter Line 4767  we set the flag only if there is a liter
4767    
4768      /* First deal with various "verbs" that can be introduced by '*'. */      /* First deal with various "verbs" that can be introduced by '*'. */
4769    
4770      if (*(++ptr) == CHAR_ASTERISK && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == CHAR_ASTERISK &&
4771             ((cd->ctypes[ptr[1]] & ctype_letter) != 0 || ptr[1] == ':'))
4772        {        {
4773        int i, namelen;        int i, namelen;
4774          int arglen = 0;
4775        const char *vn = verbnames;        const char *vn = verbnames;
4776        const uschar *name = ++ptr;        const uschar *name = ptr + 1;
4777          const uschar *arg = NULL;
4778        previous = NULL;        previous = NULL;
4779        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
4780          namelen = (int)(ptr - name);
4781    
4782        if (*ptr == CHAR_COLON)        if (*ptr == CHAR_COLON)
4783          {          {
4784          *errorcodeptr = ERR59;   /* Not supported */          arg = ++ptr;
4785          goto FAILED;          while ((cd->ctypes[*ptr] & (ctype_letter|ctype_digit)) != 0
4786              || *ptr == '_') ptr++;
4787            arglen = (int)(ptr - arg);
4788          }          }
4789    
4790        if (*ptr != CHAR_RIGHT_PARENTHESIS)        if (*ptr != CHAR_RIGHT_PARENTHESIS)
4791          {          {
4792          *errorcodeptr = ERR60;          *errorcodeptr = ERR60;
4793          goto FAILED;          goto FAILED;
4794          }          }
4795        namelen = ptr - name;  
4796          /* Scan the table of verb names */
4797    
4798        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4799          {          {
4800          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
# Line 4501  we set the flag only if there is a liter Line 4812  we set the flag only if there is a liter
4812                PUT2INC(code, 0, oc->number);                PUT2INC(code, 0, oc->number);
4813                }                }
4814              }              }
4815            *code++ = verbs[i].op;  
4816            break;            /* Handle the cases with/without an argument */
4817    
4818              if (arglen == 0)
4819                {
4820                if (verbs[i].op < 0)   /* Argument is mandatory */
4821                  {
4822                  *errorcodeptr = ERR66;
4823                  goto FAILED;
4824                  }
4825                *code = verbs[i].op;
4826                if (*code++ == OP_THEN)
4827                  {
4828                  PUT(code, 0, code - bcptr->current_branch - 1);
4829                  code += LINK_SIZE;
4830                  }
4831                }
4832    
4833              else
4834                {
4835                if (verbs[i].op_arg < 0)   /* Argument is forbidden */
4836                  {
4837                  *errorcodeptr = ERR59;
4838                  goto FAILED;
4839                  }
4840                *code = verbs[i].op_arg;
4841                if (*code++ == OP_THEN_ARG)
4842                  {
4843                  PUT(code, 0, code - bcptr->current_branch - 1);
4844                  code += LINK_SIZE;
4845                  }
4846                *code++ = arglen;
4847                memcpy(code, arg, arglen);
4848                code += arglen;
4849                *code++ = 0;
4850                }
4851    
4852              break;  /* Found verb, exit loop */
4853            }            }
4854    
4855          vn += verbs[i].len + 1;          vn += verbs[i].len + 1;
4856          }          }
4857        if (i < verbcount) continue;  
4858        *errorcodeptr = ERR60;        if (i < verbcount) continue;    /* Successfully handled a verb */
4859          *errorcodeptr = ERR60;          /* Verb not recognized */
4860        goto FAILED;        goto FAILED;
4861        }        }
4862    
# Line 4626  we set the flag only if there is a liter Line 4975  we set the flag only if there is a liter
4975                recno * 10 + *ptr - CHAR_0 : -1;                recno * 10 + *ptr - CHAR_0 : -1;
4976            ptr++;            ptr++;
4977            }            }
4978          namelen = ptr - name;          namelen = (int)(ptr - name);
4979    
4980          if ((terminator > 0 && *ptr++ != terminator) ||          if ((terminator > 0 && *ptr++ != terminator) ||
4981              *ptr++ != CHAR_RIGHT_PARENTHESIS)              *ptr++ != CHAR_RIGHT_PARENTHESIS)
# Line 4822  we set the flag only if there is a liter Line 5171  we set the flag only if there is a liter
5171              goto FAILED;              goto FAILED;
5172              }              }
5173            *code++ = n;            *code++ = n;
5174            PUT(code, 0, ptr - cd->start_pattern + 1);  /* Pattern offset */            PUT(code, 0, (int)(ptr - cd->start_pattern + 1)); /* Pattern offset */
5175            PUT(code, LINK_SIZE, 0);                    /* Default length */            PUT(code, LINK_SIZE, 0);                          /* Default length */
5176            code += 2 * LINK_SIZE;            code += 2 * LINK_SIZE;
5177            }            }
5178          previous = NULL;          previous = NULL;
# Line 4856  we set the flag only if there is a liter Line 5205  we set the flag only if there is a liter
5205            name = ++ptr;            name = ++ptr;
5206    
5207            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
5208            namelen = ptr - name;            namelen = (int)(ptr - name);
5209    
5210            /* In the pre-compile phase, just do a syntax check. */            /* In the pre-compile phase, just do a syntax check. */
5211    
# Line 4986  we set the flag only if there is a liter Line 5335  we set the flag only if there is a liter
5335          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
5336          name = ++ptr;          name = ++ptr;
5337          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
5338          namelen = ptr - name;          namelen = (int)(ptr - name);
5339    
5340          /* 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
5341          reference number. */          reference number. */
# Line 5149  we set the flag only if there is a liter Line 5498  we set the flag only if there is a liter
5498                  *errorcodeptr = ERR15;                  *errorcodeptr = ERR15;
5499                  goto FAILED;                  goto FAILED;
5500                  }                  }
5501    
5502                  /* Fudge the value of "called" so that when it is inserted as an
5503                  offset below, what it actually inserted is the reference number
5504                  of the group. */
5505    
5506                called = cd->start_code + recno;                called = cd->start_code + recno;
5507                PUTINC(cd->hwm, 0, code + 2 + LINK_SIZE - cd->start_code);                PUTINC(cd->hwm, 0, (int)(code + 2 + LINK_SIZE - cd->start_code));
5508                }                }
5509    
5510              /* If not a forward reference, and the subpattern is still open,              /* If not a forward reference, and the subpattern is still open,
# Line 5158  we set the flag only if there is a liter Line 5512  we set the flag only if there is a liter
5512              recursion that could loop for ever, and diagnose that case. */              recursion that could loop for ever, and diagnose that case. */
5513    
5514              else if (GET(called, 1) == 0 &&              else if (GET(called, 1) == 0 &&
5515                       could_be_empty(called, code, bcptr, utf8))                       could_be_empty(called, code, bcptr, utf8, cd))
5516                {                {
5517                *errorcodeptr = ERR40;                *errorcodeptr = ERR40;
5518                goto FAILED;                goto FAILED;
# Line 5174  we set the flag only if there is a liter Line 5528  we set the flag only if there is a liter
5528            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
5529    
5530            *code = OP_RECURSE;            *code = OP_RECURSE;
5531            PUT(code, 1, called - cd->start_code);            PUT(code, 1, (int)(called - cd->start_code));
5532            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
5533    
5534            *code = OP_KET;            *code = OP_KET;
# Line 5285  we set the flag only if there is a liter Line 5639  we set the flag only if there is a liter
5639          }     /* End of switch for character following (? */          }     /* End of switch for character following (? */
5640        }       /* End of (? handling */        }       /* End of (? handling */
5641    
5642      /* Opening parenthesis not followed by '?'. If PCRE_NO_AUTO_CAPTURE is set,      /* Opening parenthesis not followed by '*' or '?'. If PCRE_NO_AUTO_CAPTURE
5643      all unadorned brackets become non-capturing and behave like (?:...)      is set, all unadorned brackets become non-capturing and behave like (?:...)
5644      brackets. */      brackets. */
5645    
5646      else if ((options & PCRE_NO_AUTO_CAPTURE) != 0)      else if ((options & PCRE_NO_AUTO_CAPTURE) != 0)
# Line 5478  we set the flag only if there is a liter Line 5832  we set the flag only if there is a liter
5832    
5833      /* ===================================================================*/      /* ===================================================================*/
5834      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values      /* Handle metasequences introduced by \. For ones like \d, the ESC_ values
5835      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
5836      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
5837      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
5838      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
5839      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
5840        ever created. */
5841    
5842      case CHAR_BACKSLASH:      case CHAR_BACKSLASH:
5843      tempptr = ptr;      tempptr = ptr;
# Line 5590  we set the flag only if there is a liter Line 5945  we set the flag only if there is a liter
5945    
5946        if (-c >= ESC_REF)        if (-c >= ESC_REF)
5947          {          {
5948            open_capitem *oc;
5949          recno = -c - ESC_REF;          recno = -c - ESC_REF;
5950    
5951          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
# Line 5599  we set the flag only if there is a liter Line 5955  we set the flag only if there is a liter
5955          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
5956          cd->backref_map |= (recno < 32)? (1 << recno) : 1;          cd->backref_map |= (recno < 32)? (1 << recno) : 1;
5957          if (recno > cd->top_backref) cd->top_backref = recno;          if (recno > cd->top_backref) cd->top_backref = recno;
5958    
5959            /* Check to see if this back reference is recursive, that it, it
5960            is inside the group that it references. A flag is set so that the
5961            group can be made atomic. */
5962    
5963            for (oc = cd->open_caps; oc != NULL; oc = oc->next)
5964              {
5965              if (oc->number == recno)
5966                {
5967                oc->flag = TRUE;
5968                break;
5969                }
5970              }
5971          }          }
5972    
5973        /* So are Unicode property matches, if supported. */        /* So are Unicode property matches, if supported. */
# Line 5628  we set the flag only if there is a liter Line 5997  we set the flag only if there is a liter
5997  #endif  #endif
5998    
5999        /* For the rest (including \X when Unicode properties are supported), we        /* For the rest (including \X when Unicode properties are supported), we
6000        can obtain the OP value by negating the escape value. */        can obtain the OP value by negating the escape value in the default
6001          situation when PCRE_UCP is not set. When it *is* set, we substitute
6002          Unicode property tests. */
6003    
6004        else        else
6005          {          {
6006          previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;  #ifdef SUPPORT_UCP
6007          *code++ = -c;          if (-c >= ESC_DU && -c <= ESC_wu)
6008              {
6009              nestptr = ptr + 1;                   /* Where to resume */
6010              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
6011              }
6012            else
6013    #endif
6014              {
6015              previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;
6016              *code++ = -c;
6017              }
6018          }          }
6019        continue;        continue;
6020        }        }
# Line 5811  them global. It tests the value of lengt Line 6192  them global. It tests the value of lengt
6192  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. */
6193    
6194  /* 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
6195  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
6196    detect groups that contain recursive back references to themselves. */
6197    
6198  if (*code == OP_CBRA)  if (*code == OP_CBRA)
6199    {    {
6200    capnumber = GET2(code, 1 + LINK_SIZE);    capnumber = GET2(code, 1 + LINK_SIZE);
6201    capitem.number = capnumber;    capitem.number = capnumber;
6202    capitem.next = cd->open_caps;    capitem.next = cd->open_caps;
6203      capitem.flag = FALSE;
6204    cd->open_caps = &capitem;    cd->open_caps = &capitem;
6205    }    }
6206    
# Line 5963  for (;;) Line 6346  for (;;)
6346      {      {
6347      if (lengthptr == NULL)      if (lengthptr == NULL)
6348        {        {
6349        int branch_length = code - last_branch;        int branch_length = (int)(code - last_branch);
6350        do        do
6351          {          {
6352          int prev_length = GET(last_branch, 1);          int prev_length = GET(last_branch, 1);
# Line 5974  for (;;) Line 6357  for (;;)
6357        while (branch_length > 0);        while (branch_length > 0);
6358        }        }
6359    
     /* If it was a capturing subpattern, remove it from the chain. */  
   
     if (capnumber > 0) cd->open_caps = cd->open_caps->next;  
   
6360      /* Fill in the ket */      /* Fill in the ket */
6361    
6362      *code = OP_KET;      *code = OP_KET;
6363      PUT(code, 1, code - start_bracket);      PUT(code, 1, (int)(code - start_bracket));
6364      code += 1 + LINK_SIZE;      code += 1 + LINK_SIZE;
6365    
6366        /* If it was a capturing subpattern, check to see if it contained any
6367        recursive back references. If so, we must wrap it in atomic brackets.
6368        In any event, remove the block from the chain. */
6369    
6370        if (capnumber > 0)
6371          {
6372          if (cd->open_caps->flag)
6373            {
6374            memmove(start_bracket + 1 + LINK_SIZE, start_bracket,
6375              code - start_bracket);
6376            *start_bracket = OP_ONCE;
6377            code += 1 + LINK_SIZE;
6378            PUT(start_bracket, 1, (int)(code - start_bracket));
6379            *code = OP_KET;
6380            PUT(code, 1, (int)(code - start_bracket));
6381            code += 1 + LINK_SIZE;
6382            length += 2 + 2*LINK_SIZE;
6383            }
6384          cd->open_caps = cd->open_caps->next;
6385          }
6386    
6387      /* Reset options if needed. */      /* Reset options if needed. */
6388    
6389      if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS)      if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS)
# Line 6032  for (;;) Line 6432  for (;;)
6432    else    else
6433      {      {
6434      *code = OP_ALT;      *code = OP_ALT;
6435      PUT(code, 1, code - last_branch);      PUT(code, 1, (int)(code - last_branch));
6436      bc.current_branch = last_branch = code;      bc.current_branch = last_branch = code;
6437      code += 1 + LINK_SIZE;      code += 1 + LINK_SIZE;
6438      }      }
# Line 6351  int length = 1;  /* For final END opcode Line 6751  int length = 1;  /* For final END opcode
6751  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
6752  int errorcode = 0;  int errorcode = 0;
6753  int skipatstart = 0;  int skipatstart = 0;
6754  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8;
6755  size_t size;  size_t size;
6756  uschar *code;  uschar *code;
6757  const uschar *codestart;  const uschar *codestart;
# Line 6421  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6821  while (ptr[skipatstart] == CHAR_LEFT_PAR
6821    
6822    if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_UTF8_RIGHTPAR, 5) == 0)
6823      { skipatstart += 7; options |= PCRE_UTF8; continue; }      { skipatstart += 7; options |= PCRE_UTF8; continue; }
6824      else if (strncmp((char *)(ptr+skipatstart+2), STRING_UCP_RIGHTPAR, 4) == 0)
6825        { skipatstart += 6; options |= PCRE_UCP; continue; }
6826    
6827    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)    if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)
6828      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }      { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
# Line 6445  while (ptr[skipatstart] == CHAR_LEFT_PAR Line 6847  while (ptr[skipatstart] == CHAR_LEFT_PAR
6847    else break;    else break;
6848    }    }
6849    
6850    utf8 = (options & PCRE_UTF8) != 0;
6851    
6852  /* 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. */
6853    
6854  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 6462  if (utf8) Line 6866  if (utf8)
6866    }    }
6867  #endif  #endif
6868    
6869    /* Can't support UCP unless PCRE has been compiled to include the code. */
6870    
6871    #ifndef SUPPORT_UCP
6872    if ((options & PCRE_UCP) != 0)
6873      {
6874      errorcode = ERR67;
6875      goto PCRE_EARLY_ERROR_RETURN;
6876      }
6877    #endif
6878    
6879  /* Check validity of \R options. */  /* Check validity of \R options. */
6880    
6881  switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))  switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
# Line 6590  regex compiled on a system with 4-byte p Line 7004  regex compiled on a system with 4-byte p
7004  pointers. */  pointers. */
7005    
7006  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
7007  re->size = size;  re->size = (int)size;
7008  re->options = cd->external_options;  re->options = cd->external_options;
7009  re->flags = cd->external_flags;  re->flags = cd->external_flags;
7010  re->dummy1 = 0;  re->dummy1 = 0;
# Line 6661  while (errorcode == 0 && cd->hwm > cwork Line 7075  while (errorcode == 0 && cd->hwm > cwork
7075    recno = GET(codestart, offset);    recno = GET(codestart, offset);
7076    groupptr = _pcre_find_bracket(codestart, utf8, recno);    groupptr = _pcre_find_bracket(codestart, utf8, recno);
7077    if (groupptr == NULL) errorcode = ERR53;    if (groupptr == NULL) errorcode = ERR53;
7078      else PUT(((uschar *)codestart), offset, groupptr - codestart);      else PUT(((uschar *)codestart), offset, (int)(groupptr - codestart));
7079    }    }
7080    
7081  /* 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 6716  if (errorcode != 0) Line 7130  if (errorcode != 0)
7130    {    {
7131    (pcre_free)(re);    (pcre_free)(re);
7132    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
7133    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = (int)(ptr - (const uschar *)pattern);
7134    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
7135    *errorptr = find_error_text(errorcode);    *errorptr = find_error_text(errorcode);
7136    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
# Line 6771  if (reqbyte >= 0 && Line 7185  if (reqbyte >= 0 &&
7185  case when building a production library. */  case when building a production library. */
7186    
7187  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
   
7188  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
7189    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
7190    

Legend:
Removed from v.483  
changed lines
  Added in v.552

  ViewVC Help
Powered by ViewVC 1.1.5