/[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 178 by ph10, Wed Jun 13 08:44:34 2007 UTC revision 323 by ph10, Wed Mar 5 17:23:42 2008 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) 1997-2007 University of Cambridge             Copyright (c) 1997-2008 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 42  POSSIBILITY OF SUCH DAMAGE. Line 42  POSSIBILITY OF SUCH DAMAGE.
42  supporting internal functions that are not used by other modules. */  supporting internal functions that are not used by other modules. */
43    
44    
45    #ifdef HAVE_CONFIG_H
46    #include "config.h"
47    #endif
48    
49  #define NLBLOCK cd             /* Block containing newline information */  #define NLBLOCK cd             /* Block containing newline information */
50  #define PSSTART start_pattern  /* Field containing processed string start */  #define PSSTART start_pattern  /* Field containing processed string start */
51  #define PSEND   end_pattern    /* Field containing processed string end */  #define PSEND   end_pattern    /* Field containing processed string end */
52    
   
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
# Line 62  used by pcretest. DEBUG is not defined w Line 65  used by pcretest. DEBUG is not defined w
65    
66  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))
67    
68    /* Maximum length value to check against when making sure that the integer that
69    holds the compiled pattern length does not overflow. We make it a bit less than
70    INT_MAX to allow for adding in group terminating bytes, so that we don't have
71    to check them every time. */
72    
73    #define OFLOW_MAX (INT_MAX - 20)
74    
75    
76  /*************************************************  /*************************************************
77  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 120  static const short int escapes[] = { Line 130  static const short int escapes[] = {
130  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',
131  /*  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,
132  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,
133  /*  D0 */   '}',     0,      0,       0,      0,     0,      0, -ESC_P,  /*  D0 */   '}',     0, -ESC_K,       0,      0,     0,      0, -ESC_P,
134  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,
135  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,
136  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,
# Line 130  static const short int escapes[] = { Line 140  static const short int escapes[] = {
140  #endif  #endif
141    
142    
143  /* Tables of names of POSIX character classes and their lengths. The list is  /* Table of special "verbs" like (*PRUNE). This is a short table, so it is
144  terminated by a zero length entry. The first three must be alpha, lower, upper,  searched linearly. Put all the names into a single string, in order to reduce
145  as this is assumed for handling case independence. */  the number of relocations when a shared library is dynamically linked. */
146    
147  static const char *const posix_names[] = {  typedef struct verbitem {
148    "alpha", "lower", "upper",    int   len;
149    "alnum", "ascii", "blank", "cntrl", "digit", "graph",    int   op;
150    "print", "punct", "space", "word",  "xdigit" };  } verbitem;
151    
152    static const char verbnames[] =
153      "ACCEPT\0"
154      "COMMIT\0"
155      "F\0"
156      "FAIL\0"
157      "PRUNE\0"
158      "SKIP\0"
159      "THEN";
160    
161    static verbitem verbs[] = {
162      { 6, OP_ACCEPT },
163      { 6, OP_COMMIT },
164      { 1, OP_FAIL },
165      { 4, OP_FAIL },
166      { 5, OP_PRUNE },
167      { 4, OP_SKIP  },
168      { 4, OP_THEN  }
169    };
170    
171    static int verbcount = sizeof(verbs)/sizeof(verbitem);
172    
173    
174    /* Tables of names of POSIX character classes and their lengths. The names are
175    now all in a single string, to reduce the number of relocations when a shared
176    library is dynamically loaded. The list of lengths is terminated by a zero
177    length entry. The first three must be alpha, lower, upper, as this is assumed
178    for handling case independence. */
179    
180    static const char posix_names[] =
181      "alpha\0"  "lower\0"  "upper\0"  "alnum\0"  "ascii\0"  "blank\0"
182      "cntrl\0"  "digit\0"  "graph\0"  "print\0"  "punct\0"  "space\0"
183      "word\0"   "xdigit";
184    
185  static const uschar posix_name_lengths[] = {  static const uschar posix_name_lengths[] = {
186    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };
# Line 176  static const int posix_class_maps[] = { Line 219  static const int posix_class_maps[] = {
219  /* The texts of compile-time error messages. These are "char *" because they  /* The texts of compile-time error messages. These are "char *" because they
220  are passed to the outside world. Do not ever re-use any error number, because  are passed to the outside world. Do not ever re-use any error number, because
221  they are documented. Always add a new error instead. Messages marked DEAD below  they are documented. Always add a new error instead. Messages marked DEAD below
222  are no longer used. */  are no longer used. This used to be a table of strings, but in order to reduce
223    the number of relocations needed when a shared library is loaded dynamically,
224  static const char *error_texts[] = {  it is now one long string. We cannot use a table of offsets, because the
225    "no error",  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we
226    "\\ at end of pattern",  simply count through to the one we want - this isn't a performance issue
227    "\\c at end of pattern",  because these strings are used only when there is a compilation error. */
228    "unrecognized character follows \\",  
229    "numbers out of order in {} quantifier",  static const char error_texts[] =
230      "no error\0"
231      "\\ at end of pattern\0"
232      "\\c at end of pattern\0"
233      "unrecognized character follows \\\0"
234      "numbers out of order in {} quantifier\0"
235    /* 5 */    /* 5 */
236    "number too big in {} quantifier",    "number too big in {} quantifier\0"
237    "missing terminating ] for character class",    "missing terminating ] for character class\0"
238    "invalid escape sequence in character class",    "invalid escape sequence in character class\0"
239    "range out of order in character class",    "range out of order in character class\0"
240    "nothing to repeat",    "nothing to repeat\0"
241    /* 10 */    /* 10 */
242    "operand of unlimited repeat could match the empty string",  /** DEAD **/    "operand of unlimited repeat could match the empty string\0"  /** DEAD **/
243    "internal error: unexpected repeat",    "internal error: unexpected repeat\0"
244    "unrecognized character after (?",    "unrecognized character after (? or (?-\0"
245    "POSIX named classes are supported only within a class",    "POSIX named classes are supported only within a class\0"
246    "missing )",    "missing )\0"
247    /* 15 */    /* 15 */
248    "reference to non-existent subpattern",    "reference to non-existent subpattern\0"
249    "erroffset passed as NULL",    "erroffset passed as NULL\0"
250    "unknown option bit(s) set",    "unknown option bit(s) set\0"
251    "missing ) after comment",    "missing ) after comment\0"
252    "parentheses nested too deeply",  /** DEAD **/    "parentheses nested too deeply\0"  /** DEAD **/
253    /* 20 */    /* 20 */
254    "regular expression too large",    "regular expression is too large\0"
255    "failed to get memory",    "failed to get memory\0"
256    "unmatched parentheses",    "unmatched parentheses\0"
257    "internal error: code overflow",    "internal error: code overflow\0"
258    "unrecognized character after (?<",    "unrecognized character after (?<\0"
259    /* 25 */    /* 25 */
260    "lookbehind assertion is not fixed length",    "lookbehind assertion is not fixed length\0"
261    "malformed number or name after (?(",    "malformed number or name after (?(\0"
262    "conditional group contains more than two branches",    "conditional group contains more than two branches\0"
263    "assertion expected after (?(",    "assertion expected after (?(\0"
264    "(?R or (?[+-]digits must be followed by )",    "(?R or (?[+-]digits must be followed by )\0"
265    /* 30 */    /* 30 */
266    "unknown POSIX class name",    "unknown POSIX class name\0"
267    "POSIX collating elements are not supported",    "POSIX collating elements are not supported\0"
268    "this version of PCRE is not compiled with PCRE_UTF8 support",    "this version of PCRE is not compiled with PCRE_UTF8 support\0"
269    "spare error",  /** DEAD **/    "spare error\0"  /** DEAD **/
270    "character value in \\x{...} sequence is too large",    "character value in \\x{...} sequence is too large\0"
271    /* 35 */    /* 35 */
272    "invalid condition (?(0)",    "invalid condition (?(0)\0"
273    "\\C not allowed in lookbehind assertion",    "\\C not allowed in lookbehind assertion\0"
274    "PCRE does not support \\L, \\l, \\N, \\U, or \\u",    "PCRE does not support \\L, \\l, \\N, \\U, or \\u\0"
275    "number after (?C is > 255",    "number after (?C is > 255\0"
276    "closing ) for (?C expected",    "closing ) for (?C expected\0"
277    /* 40 */    /* 40 */
278    "recursive call could loop indefinitely",    "recursive call could loop indefinitely\0"
279    "unrecognized character after (?P",    "unrecognized character after (?P\0"
280    "syntax error in subpattern name (missing terminator)",    "syntax error in subpattern name (missing terminator)\0"
281    "two named subpatterns have the same name",    "two named subpatterns have the same name\0"
282    "invalid UTF-8 string",    "invalid UTF-8 string\0"
283    /* 45 */    /* 45 */
284    "support for \\P, \\p, and \\X has not been compiled",    "support for \\P, \\p, and \\X has not been compiled\0"
285    "malformed \\P or \\p sequence",    "malformed \\P or \\p sequence\0"
286    "unknown property name after \\P or \\p",    "unknown property name after \\P or \\p\0"
287    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)\0"
288    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
289    /* 50 */    /* 50 */
290    "repeated subpattern is too long",    "repeated subpattern is too long\0"    /** DEAD **/
291    "octal value is greater than \\377 (not in UTF-8 mode)",    "octal value is greater than \\377 (not in UTF-8 mode)\0"
292    "internal error: overran compiling workspace",    "internal error: overran compiling workspace\0"
293    "internal error: previously-checked referenced subpattern not found",    "internal error: previously-checked referenced subpattern not found\0"
294    "DEFINE group contains more than one branch",    "DEFINE group contains more than one branch\0"
295    /* 55 */    /* 55 */
296    "repeating a DEFINE group is not allowed",    "repeating a DEFINE group is not allowed\0"
297    "inconsistent NEWLINE options",    "inconsistent NEWLINE options\0"
298    "\\g is not followed by a braced name or an optionally braced non-zero number",    "\\g is not followed by a braced name or an optionally braced non-zero number\0"
299    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number"    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number\0"
300  };    "(*VERB) with an argument is not supported\0"
301      /* 60 */
302      "(*VERB) not recognized\0"
303      "number is too big\0"
304      "subpattern name expected\0"
305      "digit expected after (?+";
306    
307    
308  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 379  static const unsigned char ebcdic_charta Line 432  static const unsigned char ebcdic_charta
432  /* Definition to allow mutual recursion */  /* Definition to allow mutual recursion */
433    
434  static BOOL  static BOOL
435    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,
436      int *, int *, branch_chain *, compile_data *, int *);      int *, int *, branch_chain *, compile_data *, int *);
437    
438    
439    
440  /*************************************************  /*************************************************
441    *            Find an error text                  *
442    *************************************************/
443    
444    /* The error texts are now all in one long string, to save on relocations. As
445    some of the text is of unknown length, we can't use a table of offsets.
446    Instead, just count through the strings. This is not a performance issue
447    because it happens only when there has been a compilation error.
448    
449    Argument:   the error number
450    Returns:    pointer to the error string
451    */
452    
453    static const char *
454    find_error_text(int n)
455    {
456    const char *s = error_texts;
457    for (; n > 0; n--) while (*s++ != 0);
458    return s;
459    }
460    
461    
462    /*************************************************
463  *            Handle escapes                      *  *            Handle escapes                      *
464  *************************************************/  *************************************************/
465    
# Line 405  Arguments: Line 480  Arguments:
480    
481  Returns:         zero or positive => a data character  Returns:         zero or positive => a data character
482                   negative => a special escape sequence                   negative => a special escape sequence
483                   on error, errorptr is set                   on error, errorcodeptr is set
484  */  */
485    
486  static int  static int
# Line 423  ptr--; /* Set Line 498  ptr--; /* Set
498    
499  if (c == 0) *errorcodeptr = ERR1;  if (c == 0) *errorcodeptr = ERR1;
500    
501  /* Non-alphamerics are literals. For digits or letters, do an initial lookup in  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
502  a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
503  Otherwise further processing may be required. */  Otherwise further processing may be required. */
504    
505  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
506  else if (c < '0' || c > 'z') {}                           /* Not alphameric */  else if (c < '0' || c > 'z') {}                           /* Not alphanumeric */
507  else if ((i = escapes[c - '0']) != 0) c = i;  else if ((i = escapes[c - '0']) != 0) c = i;
508    
509  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
510  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphameric */  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */
511  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  c = i;
512  #endif  #endif
513    
# Line 490  else Line 565  else
565      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
566        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - '0';
567    
568        if (c < 0)
569          {
570          *errorcodeptr = ERR61;
571          break;
572          }
573    
574      if (c == 0 || (braced && *(++ptr) != '}'))      if (c == 0 || (braced && *(++ptr) != '}'))
575        {        {
576        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
577        return 0;        break;
578        }        }
579    
580      if (negated)      if (negated)
# Line 501  else Line 582  else
582        if (c > bracount)        if (c > bracount)
583          {          {
584          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
585          return 0;          break;
586          }          }
587        c = bracount - (c - 1);        c = bracount - (c - 1);
588        }        }
# Line 530  else Line 611  else
611        c -= '0';        c -= '0';
612        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
613          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - '0';
614          if (c < 0)
615            {
616            *errorcodeptr = ERR61;
617            break;
618            }
619        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
620          {          {
621          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 625  else Line 711  else
711      if (c == 0)      if (c == 0)
712        {        {
713        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
714        return 0;        break;
715        }        }
716    
717  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
# Line 638  else Line 724  else
724      break;      break;
725    
726      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any
727      other alphameric following \ is an error if PCRE_EXTRA was set; otherwise,      other alphanumeric following \ is an error if PCRE_EXTRA was set;
728      for Perl compatibility, it is a literal. This code looks a bit odd, but      otherwise, for Perl compatibility, it is a literal. This code looks a bit
729      there used to be some cases other than the default, and there may be again      odd, but there used to be some cases other than the default, and there may
730      in future, so I haven't "optimized" it. */      be again in future, so I haven't "optimized" it. */
731    
732      default:      default:
733      if ((options & PCRE_EXTRA) != 0) switch(c)      if ((options & PCRE_EXTRA) != 0) switch(c)
# Line 701  if (c == '{') Line 787  if (c == '{')
787      *negptr = TRUE;      *negptr = TRUE;
788      ptr++;      ptr++;
789      }      }
790    for (i = 0; i < sizeof(name) - 1; i++)    for (i = 0; i < (int)sizeof(name) - 1; i++)
791      {      {
792      c = *(++ptr);      c = *(++ptr);
793      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 730  top = _pcre_utt_size; Line 816  top = _pcre_utt_size;
816  while (bot < top)  while (bot < top)
817    {    {
818    i = (bot + top) >> 1;    i = (bot + top) >> 1;
819    c = strcmp(name, _pcre_utt[i].name);    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);
820    if (c == 0)    if (c == 0)
821      {      {
822      *dptr = _pcre_utt[i].value;      *dptr = _pcre_utt[i].value;
# Line 904  for (; *ptr != 0; ptr++) Line 990  for (; *ptr != 0; ptr++)
990      {      {
991      while (*(++ptr) != ']')      while (*(++ptr) != ']')
992        {        {
993          if (*ptr == 0) return -1;
994        if (*ptr == '\\')        if (*ptr == '\\')
995          {          {
996          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 931  for (; *ptr != 0; ptr++) Line 1018  for (; *ptr != 0; ptr++)
1018    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
1019    
1020    if (*ptr != '(') continue;    if (*ptr != '(') continue;
1021    if (ptr[1] != '?')    if (ptr[1] != '?' && ptr[1] != '*')
1022      {      {
1023      count++;      count++;
1024      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 1059  for (;;) Line 1146  for (;;)
1146    {    {
1147    int d;    int d;
1148    register int op = *cc;    register int op = *cc;
   
1149    switch (op)    switch (op)
1150      {      {
1151      case OP_CBRA:      case OP_CBRA:
# Line 1148  for (;;) Line 1234  for (;;)
1234    
1235      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1236      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1237        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1238      cc += 4;      cc += 4;
1239      break;      break;
1240    
# Line 1256  for (;;) Line 1343  for (;;)
1343      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1344      }      }
1345    
1346    /* In UTF-8 mode, opcodes that are followed by a character may be followed by    /* Otherwise, we can get the item's length from the table, except that for
1347    a multi-byte character. The length in the table is a minimum, so we have to    repeated character types, we have to test for \p and \P, which have an extra
1348    arrange to skip the extra bytes. */    two bytes of parameters. */
1349    
1350    else    else
1351      {      {
1352        switch(c)
1353          {
1354          case OP_TYPESTAR:
1355          case OP_TYPEMINSTAR:
1356          case OP_TYPEPLUS:
1357          case OP_TYPEMINPLUS:
1358          case OP_TYPEQUERY:
1359          case OP_TYPEMINQUERY:
1360          case OP_TYPEPOSSTAR:
1361          case OP_TYPEPOSPLUS:
1362          case OP_TYPEPOSQUERY:
1363          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1364          break;
1365    
1366          case OP_TYPEUPTO:
1367          case OP_TYPEMINUPTO:
1368          case OP_TYPEEXACT:
1369          case OP_TYPEPOSUPTO:
1370          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1371          break;
1372          }
1373    
1374        /* Add in the fixed length from the table */
1375    
1376      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1377    
1378      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1379      a multi-byte character. The length in the table is a minimum, so we have to
1380      arrange to skip the extra bytes. */
1381    
1382  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1383      if (utf8) switch(c)      if (utf8) switch(c)
1384        {        {
# Line 1320  for (;;) Line 1436  for (;;)
1436    
1437    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1438    
1439    /* Otherwise, we get the item's length from the table. In UTF-8 mode, opcodes    /* Otherwise, we can get the item's length from the table, except that for
1440    that are followed by a character may be followed by a multi-byte character.    repeated character types, we have to test for \p and \P, which have an extra
1441    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1442    
1443    else    else
1444      {      {
1445        switch(c)
1446          {
1447          case OP_TYPESTAR:
1448          case OP_TYPEMINSTAR:
1449          case OP_TYPEPLUS:
1450          case OP_TYPEMINPLUS:
1451          case OP_TYPEQUERY:
1452          case OP_TYPEMINQUERY:
1453          case OP_TYPEPOSSTAR:
1454          case OP_TYPEPOSPLUS:
1455          case OP_TYPEPOSQUERY:
1456          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1457          break;
1458    
1459          case OP_TYPEPOSUPTO:
1460          case OP_TYPEUPTO:
1461          case OP_TYPEMINUPTO:
1462          case OP_TYPEEXACT:
1463          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1464          break;
1465          }
1466    
1467        /* Add in the fixed length from the table */
1468    
1469      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1470    
1471        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1472        by a multi-byte character. The length in the table is a minimum, so we have
1473        to arrange to skip the extra bytes. */
1474    
1475  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1476      if (utf8) switch(c)      if (utf8) switch(c)
1477        {        {
# Line 1364  for (;;) Line 1508  for (;;)
1508  can match the empty string or not. It is called from could_be_empty()  can match the empty string or not. It is called from could_be_empty()
1509  below and from compile_branch() when checking for an unlimited repeat of a  below and from compile_branch() when checking for an unlimited repeat of a
1510  group that can match nothing. Note that first_significant_code() skips over  group that can match nothing. Note that first_significant_code() skips over
1511  assertions. If we hit an unclosed bracket, we return "empty" - this means we've  backward and negative forward assertions when its final argument is TRUE. If we
1512  struck an inner bracket whose current branch will already have been scanned.  hit an unclosed bracket, we return "empty" - this means we've struck an inner
1513    bracket whose current branch will already have been scanned.
1514    
1515  Arguments:  Arguments:
1516    code        points to start of search    code        points to start of search
# Line 1387  for (code = first_significant_code(code Line 1532  for (code = first_significant_code(code
1532    
1533    c = *code;    c = *code;
1534    
1535      /* Skip over forward assertions; the other assertions are skipped by
1536      first_significant_code() with a TRUE final argument. */
1537    
1538      if (c == OP_ASSERT)
1539        {
1540        do code += GET(code, 1); while (*code == OP_ALT);
1541        c = *code;
1542        continue;
1543        }
1544    
1545    /* Groups with zero repeats can of course be empty; skip them. */    /* Groups with zero repeats can of course be empty; skip them. */
1546    
1547    if (c == OP_BRAZERO || c == OP_BRAMINZERO)    if (c == OP_BRAZERO || c == OP_BRAMINZERO)
# Line 1399  for (code = first_significant_code(code Line 1554  for (code = first_significant_code(code
1554    
1555    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1556    
1557    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1558      {      {
1559      BOOL empty_branch;      BOOL empty_branch;
1560      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 1423  for (code = first_significant_code(code Line 1578  for (code = first_significant_code(code
1578    
1579    switch (c)    switch (c)
1580      {      {
1581      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1582        cannot be represented just by a bit map. This includes negated single
1583        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1584        actual length is stored in the compiled code, so we must update "code"
1585        here. */
1586    
1587  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1588      case OP_XCLASS:      case OP_XCLASS:
1589      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1590      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1591  #endif  #endif
1592    
# Line 1489  for (code = first_significant_code(code Line 1648  for (code = first_significant_code(code
1648      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1649      return FALSE;      return FALSE;
1650    
1651        /* These are going to continue, as they may be empty, but we have to
1652        fudge the length for the \p and \P cases. */
1653    
1654        case OP_TYPESTAR:
1655        case OP_TYPEMINSTAR:
1656        case OP_TYPEPOSSTAR:
1657        case OP_TYPEQUERY:
1658        case OP_TYPEMINQUERY:
1659        case OP_TYPEPOSQUERY:
1660        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1661        break;
1662    
1663        /* Same for these */
1664    
1665        case OP_TYPEUPTO:
1666        case OP_TYPEMINUPTO:
1667        case OP_TYPEPOSUPTO:
1668        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1669        break;
1670    
1671      /* End of branch */      /* End of branch */
1672    
1673      case OP_KET:      case OP_KET:
# Line 1558  return TRUE; Line 1737  return TRUE;
1737  *************************************************/  *************************************************/
1738    
1739  /* This function is called when the sequence "[:" or "[." or "[=" is  /* This function is called when the sequence "[:" or "[." or "[=" is
1740  encountered in a character class. It checks whether this is followed by an  encountered in a character class. It checks whether this is followed by a
1741  optional ^ and then a sequence of letters, terminated by a matching ":]" or  sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
1742  ".]" or "=]".  reach an unescaped ']' without the special preceding character, return FALSE.
1743    
1744    Originally, this function only recognized a sequence of letters between the
1745    terminators, but it seems that Perl recognizes any sequence of characters,
1746    though of course unknown POSIX names are subsequently rejected. Perl gives an
1747    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
1748    didn't consider this to be a POSIX class. Likewise for [:1234:].
1749    
1750    The problem in trying to be exactly like Perl is in the handling of escapes. We
1751    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
1752    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
1753    below handles the special case of \], but does not try to do any other escape
1754    processing. This makes it different from Perl for cases such as [:l\ower:]
1755    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
1756    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
1757    I think.
1758    
1759  Argument:  Arguments:
1760    ptr      pointer to the initial [    ptr      pointer to the initial [
1761    endptr   where to return the end pointer    endptr   where to return the end pointer
   cd       pointer to compile data  
1762    
1763  Returns:   TRUE or FALSE  Returns:   TRUE or FALSE
1764  */  */
1765    
1766  static BOOL  static BOOL
1767  check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd)  check_posix_syntax(const uschar *ptr, const uschar **endptr)
1768  {  {
1769  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
1770  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
1771  if (*(++ptr) == '^') ptr++;  for (++ptr; *ptr != 0; ptr++)
 while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;  
 if (*ptr == terminator && ptr[1] == ']')  
1772    {    {
1773    *endptr = ptr;    if (*ptr == '\\' && ptr[1] == ']') ptr++; else
1774    return TRUE;      {
1775        if (*ptr == ']') return FALSE;
1776        if (*ptr == terminator && ptr[1] == ']')
1777          {
1778          *endptr = ptr;
1779          return TRUE;
1780          }
1781        }
1782    }    }
1783  return FALSE;  return FALSE;
1784  }  }
# Line 1605  Returns: a value representing the na Line 1803  Returns: a value representing the na
1803  static int  static int
1804  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1805  {  {
1806    const char *pn = posix_names;
1807  register int yield = 0;  register int yield = 0;
1808  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1809    {    {
1810    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1811      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1812      pn += posix_name_lengths[yield] + 1;
1813    yield++;    yield++;
1814    }    }
1815  return -1;  return -1;
# Line 1651  adjust_recurse(uschar *group, int adjust Line 1851  adjust_recurse(uschar *group, int adjust
1851    uschar *save_hwm)    uschar *save_hwm)
1852  {  {
1853  uschar *ptr = group;  uschar *ptr = group;
1854    
1855  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1856    {    {
1857    int offset;    int offset;
# Line 1912  if (next >= 0) switch(op_code) Line 2113  if (next >= 0) switch(op_code)
2113    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, "item" must be a single-byte character. */
2114    
2115    case OP_NOT:    case OP_NOT:
   if (next < 0) return FALSE;  /* Not a character */  
2116    if (item == next) return TRUE;    if (item == next) return TRUE;
2117    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2118  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1949  if (next >= 0) switch(op_code) Line 2149  if (next >= 0) switch(op_code)
2149    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2150    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2151    
2152      case OP_HSPACE:
2153      case OP_NOT_HSPACE:
2154      switch(next)
2155        {
2156        case 0x09:
2157        case 0x20:
2158        case 0xa0:
2159        case 0x1680:
2160        case 0x180e:
2161        case 0x2000:
2162        case 0x2001:
2163        case 0x2002:
2164        case 0x2003:
2165        case 0x2004:
2166        case 0x2005:
2167        case 0x2006:
2168        case 0x2007:
2169        case 0x2008:
2170        case 0x2009:
2171        case 0x200A:
2172        case 0x202f:
2173        case 0x205f:
2174        case 0x3000:
2175        return op_code != OP_HSPACE;
2176        default:
2177        return op_code == OP_HSPACE;
2178        }
2179    
2180      case OP_VSPACE:
2181      case OP_NOT_VSPACE:
2182      switch(next)
2183        {
2184        case 0x0a:
2185        case 0x0b:
2186        case 0x0c:
2187        case 0x0d:
2188        case 0x85:
2189        case 0x2028:
2190        case 0x2029:
2191        return op_code != OP_VSPACE;
2192        default:
2193        return op_code == OP_VSPACE;
2194        }
2195    
2196    default:    default:
2197    return FALSE;    return FALSE;
2198    }    }
# Line 1983  switch(op_code) Line 2227  switch(op_code)
2227      case ESC_W:      case ESC_W:
2228      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2229    
2230        case ESC_h:
2231        case ESC_H:
2232        switch(item)
2233          {
2234          case 0x09:
2235          case 0x20:
2236          case 0xa0:
2237          case 0x1680:
2238          case 0x180e:
2239          case 0x2000:
2240          case 0x2001:
2241          case 0x2002:
2242          case 0x2003:
2243          case 0x2004:
2244          case 0x2005:
2245          case 0x2006:
2246          case 0x2007:
2247          case 0x2008:
2248          case 0x2009:
2249          case 0x200A:
2250          case 0x202f:
2251          case 0x205f:
2252          case 0x3000:
2253          return -next != ESC_h;
2254          default:
2255          return -next == ESC_h;
2256          }
2257    
2258        case ESC_v:
2259        case ESC_V:
2260        switch(item)
2261          {
2262          case 0x0a:
2263          case 0x0b:
2264          case 0x0c:
2265          case 0x0d:
2266          case 0x85:
2267          case 0x2028:
2268          case 0x2029:
2269          return -next != ESC_v;
2270          default:
2271          return -next == ESC_v;
2272          }
2273    
2274      default:      default:
2275      return FALSE;      return FALSE;
2276      }      }
2277    
2278    case OP_DIGIT:    case OP_DIGIT:
2279    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2280             next == -ESC_h || next == -ESC_v;
2281    
2282    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2283    return next == -ESC_d;    return next == -ESC_d;
# Line 1997  switch(op_code) Line 2286  switch(op_code)
2286    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2287    
2288    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2289    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2290    
2291      case OP_HSPACE:
2292      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2293    
2294      case OP_NOT_HSPACE:
2295      return next == -ESC_h;
2296    
2297      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2298      case OP_VSPACE:
2299      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2300    
2301      case OP_NOT_VSPACE:
2302      return next == -ESC_v;
2303    
2304    case OP_WORDCHAR:    case OP_WORDCHAR:
2305    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2306    
2307    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2308    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2073  uschar classbits[32]; Line 2375  uschar classbits[32];
2375  BOOL class_utf8;  BOOL class_utf8;
2376  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8 = (options & PCRE_UTF8) != 0;
2377  uschar *class_utf8data;  uschar *class_utf8data;
2378    uschar *class_utf8data_base;
2379  uschar utf8_char[6];  uschar utf8_char[6];
2380  #else  #else
2381  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
# Line 2112  req_caseopt = ((options & PCRE_CASELESS) Line 2415  req_caseopt = ((options & PCRE_CASELESS)
2415  for (;; ptr++)  for (;; ptr++)
2416    {    {
2417    BOOL negate_class;    BOOL negate_class;
2418      BOOL should_flip_negation;
2419    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2420    BOOL is_quantifier;    BOOL is_quantifier;
2421    BOOL is_recurse;    BOOL is_recurse;
2422    BOOL reset_bracount;    BOOL reset_bracount;
2423    int class_charcount;    int class_charcount;
2424    int class_lastchar;    int class_lastchar;
2425    int newoptions;    int newoptions;
# Line 2153  for (;; ptr++) Line 2457  for (;; ptr++)
2457      */      */
2458    
2459      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2460    
2461        /* Paranoid check for integer overflow */
2462    
2463        if (OFLOW_MAX - *lengthptr < code - last_code)
2464          {
2465          *errorcodeptr = ERR20;
2466          goto FAILED;
2467          }
2468    
2469      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2470      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));
2471    
# Line 2265  for (;; ptr++) Line 2578  for (;; ptr++)
2578      *ptrptr = ptr;      *ptrptr = ptr;
2579      if (lengthptr != NULL)      if (lengthptr != NULL)
2580        {        {
2581          if (OFLOW_MAX - *lengthptr < code - last_code)
2582            {
2583            *errorcodeptr = ERR20;
2584            goto FAILED;
2585            }
2586        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2587        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2588        }        }
# Line 2321  for (;; ptr++) Line 2639  for (;; ptr++)
2639      they are encountered at the top level, so we'll do that too. */      they are encountered at the top level, so we'll do that too. */
2640    
2641      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2642          check_posix_syntax(ptr, &tempptr, cd))          check_posix_syntax(ptr, &tempptr))
2643        {        {
2644        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;
2645        goto FAILED;        goto FAILED;
2646        }        }
2647    
2648      /* If the first character is '^', set the negation flag and skip it. */      /* If the first character is '^', set the negation flag and skip it. Also,
2649        if the first few characters (either before or after ^) are \Q\E or \E we
2650        skip them too. This makes for compatibility with Perl. */
2651    
2652      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2653        for (;;)
2654        {        {
       negate_class = TRUE;  
2655        c = *(++ptr);        c = *(++ptr);
2656        }        if (c == '\\')
2657      else          {
2658        {          if (ptr[1] == 'E') ptr++;
2659        negate_class = FALSE;            else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2660                else break;
2661            }
2662          else if (!negate_class && c == '^')
2663            negate_class = TRUE;
2664          else break;
2665        }        }
2666    
2667        /* If a class contains a negative special such as \S, we need to flip the
2668        negation flag at the end, so that support for characters > 255 works
2669        correctly (they are all included in the class). */
2670    
2671        should_flip_negation = FALSE;
2672    
2673      /* Keep a count of chars with values < 256 so that we can optimize the case      /* Keep a count of chars with values < 256 so that we can optimize the case
2674      of just a single character (as long as it's < 256). However, For higher      of just a single character (as long as it's < 256). However, For higher
2675      valued UTF-8 characters, we don't yet do any optimization. */      valued UTF-8 characters, we don't yet do any optimization. */
# Line 2356  for (;; ptr++) Line 2687  for (;; ptr++)
2687  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2688      class_utf8 = FALSE;                       /* No chars >= 256 */      class_utf8 = FALSE;                       /* No chars >= 256 */
2689      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */
2690        class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */
2691  #endif  #endif
2692    
2693      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
# Line 2371  for (;; ptr++) Line 2703  for (;; ptr++)
2703          {                           /* Braces are required because the */          {                           /* Braces are required because the */
2704          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
2705          }          }
2706    
2707          /* In the pre-compile phase, accumulate the length of any UTF-8 extra
2708          data and reset the pointer. This is so that very large classes that
2709          contain a zillion UTF-8 characters no longer overwrite the work space
2710          (which is on the stack). */
2711    
2712          if (lengthptr != NULL)
2713            {
2714            *lengthptr += class_utf8data - class_utf8data_base;
2715            class_utf8data = class_utf8data_base;
2716            }
2717    
2718  #endif  #endif
2719    
2720        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 2394  for (;; ptr++) Line 2738  for (;; ptr++)
2738    
2739        if (c == '[' &&        if (c == '[' &&
2740            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2741            check_posix_syntax(ptr, &tempptr, cd))            check_posix_syntax(ptr, &tempptr))
2742          {          {
2743          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
2744          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
# Line 2411  for (;; ptr++) Line 2755  for (;; ptr++)
2755          if (*ptr == '^')          if (*ptr == '^')
2756            {            {
2757            local_negate = TRUE;            local_negate = TRUE;
2758              should_flip_negation = TRUE;  /* Note negative special */
2759            ptr++;            ptr++;
2760            }            }
2761    
# Line 2477  for (;; ptr++) Line 2822  for (;; ptr++)
2822        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
2823        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2824        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2825        to or into the one we are building. We assume they have more than one        to 'or' into the one we are building. We assume they have more than one
2826        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2827    
2828        if (c == '\\')        if (c == '\\')
# Line 2485  for (;; ptr++) Line 2830  for (;; ptr++)
2830          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2831          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2832    
2833          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = '\b';       /* \b is backspace in a class */
2834          else if (-c == ESC_X) c = 'X';   /* \X is literal X in a class */          else if (-c == ESC_X) c = 'X';   /* \X is literal X in a class */
2835          else if (-c == ESC_R) c = 'R';   /* \R is literal R in a class */          else if (-c == ESC_R) c = 'R';   /* \R is literal R in a class */
2836          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
# Line 2497  for (;; ptr++) Line 2842  for (;; ptr++)
2842            else inescq = TRUE;            else inescq = TRUE;
2843            continue;            continue;
2844            }            }
2845            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2846    
2847          if (c < 0)          if (c < 0)
2848            {            {
# Line 2512  for (;; ptr++) Line 2858  for (;; ptr++)
2858              continue;              continue;
2859    
2860              case ESC_D:              case ESC_D:
2861                should_flip_negation = TRUE;
2862              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2863              continue;              continue;
2864    
# Line 2520  for (;; ptr++) Line 2867  for (;; ptr++)
2867              continue;              continue;
2868    
2869              case ESC_W:              case ESC_W:
2870                should_flip_negation = TRUE;
2871              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2872              continue;              continue;
2873    
# Line 2529  for (;; ptr++) Line 2877  for (;; ptr++)
2877              continue;              continue;
2878    
2879              case ESC_S:              case ESC_S:
2880                should_flip_negation = TRUE;
2881              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2882              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2883              continue;              continue;
2884    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
2885              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
2886              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
2887              }              }
# Line 2544  for (;; ptr++) Line 2890  for (;; ptr++)
2890    
2891            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
2892                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
2893    
2894            /* We need to deal with \H, \h, \V, and \v in both phases because            /* We need to deal with \H, \h, \V, and \v in both phases because
2895            they use extra memory. */            they use extra memory. */
2896    
2897            if (-c == ESC_h)            if (-c == ESC_h)
2898              {              {
2899              SETBIT(classbits, 0x09); /* VT */              SETBIT(classbits, 0x09); /* VT */
2900              SETBIT(classbits, 0x20); /* SPACE */              SETBIT(classbits, 0x20); /* SPACE */
2901              SETBIT(classbits, 0xa0); /* NSBP */              SETBIT(classbits, 0xa0); /* NSBP */
2902  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2903              if (utf8)              if (utf8)
2904                {                {
2905                class_utf8 = TRUE;                class_utf8 = TRUE;
2906                *class_utf8data++ = XCL_SINGLE;                *class_utf8data++ = XCL_SINGLE;
2907                class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
2908                *class_utf8data++ = XCL_SINGLE;                *class_utf8data++ = XCL_SINGLE;
2909                class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
2910                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
2911                class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
2912                class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
2913                *class_utf8data++ = XCL_SINGLE;                *class_utf8data++ = XCL_SINGLE;
2914                class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
2915                *class_utf8data++ = XCL_SINGLE;                *class_utf8data++ = XCL_SINGLE;
2916                class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
2917                *class_utf8data++ = XCL_SINGLE;                *class_utf8data++ = XCL_SINGLE;
2918                class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
2919                }                }
2920  #endif  #endif
2921              continue;              continue;
2922              }              }
2923    
2924            if (-c == ESC_H)            if (-c == ESC_H)
2925              {              {
# Line 2581  for (;; ptr++) Line 2927  for (;; ptr++)
2927                {                {
2928                int x = 0xff;                int x = 0xff;
2929                switch (c)                switch (c)
2930                  {                  {
2931                  case 0x09/8: x ^= 1 << (0x09%8); break;                  case 0x09/8: x ^= 1 << (0x09%8); break;
2932                  case 0x20/8: x ^= 1 << (0x20%8); break;                  case 0x20/8: x ^= 1 << (0x20%8); break;
2933                  case 0xa0/8: x ^= 1 << (0xa0%8); break;                  case 0xa0/8: x ^= 1 << (0xa0%8); break;
2934                  default: break;                  default: break;
2935                  }                  }
2936                classbits[c] |= x;                classbits[c] |= x;
2937                }                }
2938    
2939  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2940              if (utf8)              if (utf8)
2941                {                {
2942                class_utf8 = TRUE;                class_utf8 = TRUE;
2943                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
2944                class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2945                class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
2946                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
2947                class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
2948                class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
2949                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
2950                class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
2951                class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
2952                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
2953                class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
2954                class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
2955                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
2956                class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
2957                class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
2958                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
2959                class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
2960                class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
2961                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
2962                class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
2963                class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2964                }                }
2965  #endif  #endif
2966              continue;              continue;
2967              }              }
2968    
2969            if (-c == ESC_v)            if (-c == ESC_v)
2970              {              {
2971              SETBIT(classbits, 0x0a); /* LF */              SETBIT(classbits, 0x0a); /* LF */
2972              SETBIT(classbits, 0x0b); /* VT */              SETBIT(classbits, 0x0b); /* VT */
2973              SETBIT(classbits, 0x0c); /* FF */              SETBIT(classbits, 0x0c); /* FF */
2974              SETBIT(classbits, 0x0d); /* CR */              SETBIT(classbits, 0x0d); /* CR */
2975              SETBIT(classbits, 0x85); /* NEL */              SETBIT(classbits, 0x85); /* NEL */
2976  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2977              if (utf8)              if (utf8)
2978                {                {
2979                class_utf8 = TRUE;                class_utf8 = TRUE;
2980                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
2981                class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
2982                class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2983                }                }
2984  #endif  #endif
2985              continue;              continue;
2986              }              }
2987    
2988            if (-c == ESC_V)            if (-c == ESC_V)
2989              {              {
# Line 2645  for (;; ptr++) Line 2991  for (;; ptr++)
2991                {                {
2992                int x = 0xff;                int x = 0xff;
2993                switch (c)                switch (c)
2994                  {                  {
2995                  case 0x0a/8: x ^= 1 << (0x0a%8);                  case 0x0a/8: x ^= 1 << (0x0a%8);
2996                               x ^= 1 << (0x0b%8);                               x ^= 1 << (0x0b%8);
2997                               x ^= 1 << (0x0c%8);                               x ^= 1 << (0x0c%8);
2998                               x ^= 1 << (0x0d%8);                               x ^= 1 << (0x0d%8);
2999                               break;                               break;
3000                  case 0x85/8: x ^= 1 << (0x85%8); break;                  case 0x85/8: x ^= 1 << (0x85%8); break;
3001                  default: break;                  default: break;
3002                  }                  }
3003                classbits[c] |= x;                classbits[c] |= x;
3004                }                }
3005    
3006  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3007              if (utf8)              if (utf8)
3008                {                {
3009                class_utf8 = TRUE;                class_utf8 = TRUE;
3010                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
3011                class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
3012                class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
3013                *class_utf8data++ = XCL_RANGE;                *class_utf8data++ = XCL_RANGE;
3014                class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
3015                class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);                class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
3016                }                }
3017  #endif  #endif
3018              continue;              continue;
3019              }              }
3020    
3021            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
3022    
# Line 2723  for (;; ptr++) Line 3069  for (;; ptr++)
3069    
3070        oldptr = ptr;        oldptr = ptr;
3071    
3072          /* Remember \r or \n */
3073    
3074          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3075    
3076          /* Check for range */
3077    
3078        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3079          {          {
3080          int d;          int d;
# Line 2764  for (;; ptr++) Line 3116  for (;; ptr++)
3116            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3117            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3118    
3119            /* \b is backslash; \X is literal X; \R is literal R; any other            /* \b is backspace; \X is literal X; \R is literal R; any other
3120            special means the '-' was literal */            special means the '-' was literal */
3121    
3122            if (d < 0)            if (d < 0)
# Line 2790  for (;; ptr++) Line 3142  for (;; ptr++)
3142    
3143          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3144    
3145            /* Remember \r or \n */
3146    
3147            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3148    
3149          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless
3150          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3151          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2812  for (;; ptr++) Line 3168  for (;; ptr++)
3168              unsigned int origd = d;              unsigned int origd = d;
3169              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3170                {                {
3171                if (occ >= (unsigned int)c &&                if (occ >= (unsigned int)c &&
3172                    ocd <= (unsigned int)d)                    ocd <= (unsigned int)d)
3173                  continue;                          /* Skip embedded ranges */                  continue;                          /* Skip embedded ranges */
3174    
3175                if (occ < (unsigned int)c  &&                if (occ < (unsigned int)c  &&
3176                    ocd >= (unsigned int)c - 1)      /* Extend the basic range */                    ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3177                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3178                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3179                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3180                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3181                if (ocd > (unsigned int)d &&                if (ocd > (unsigned int)d &&
3182                    occ <= (unsigned int)d + 1)      /* always shorter than    */                    occ <= (unsigned int)d + 1)      /* always shorter than    */
3183                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3184                  d = ocd;                  d = ocd;
# Line 2943  for (;; ptr++) Line 3299  for (;; ptr++)
3299        goto FAILED;        goto FAILED;
3300        }        }
3301    
3302    
3303    /* This code has been disabled because it would mean that \s counts as
3304    an explicit \r or \n reference, and that's not really what is wanted. Now
3305    we set the flag only if there is a literal "\r" or "\n" in the class. */
3306    
3307    #if 0
3308        /* Remember whether \r or \n are in this class */
3309    
3310        if (negate_class)
3311          {
3312          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3313          }
3314        else
3315          {
3316          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3317          }
3318    #endif
3319    
3320    
3321      /* 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
3322      less than 256. In non-UTF-8 mode we can always optimize. In UTF-8 mode, we      less than 256. As long as there were no characters >= 128 and there was no
3323      can optimize the negative case only if there were no characters >= 128      use of \p or \P, in other words, no use of any XCLASS features, we can
3324      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3325      single-bytes only. This is an historical hangover. Maybe one day we can  
3326      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3327        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3328        operate on single-bytes only. This is an historical hangover. Maybe one day
3329        we can tidy these opcodes to handle multi-byte characters.
3330    
3331      The optimization throws away the bit map. We turn the item into a      The optimization throws away the bit map. We turn the item into a
3332      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note
# Line 2958  for (;; ptr++) Line 3336  for (;; ptr++)
3336      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3337    
3338  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3339      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3340            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3341  #else  #else
3342      if (class_charcount == 1)      if (class_charcount == 1)
3343  #endif  #endif
# Line 3004  for (;; ptr++) Line 3380  for (;; ptr++)
3380      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3381    
3382      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3383      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3384      we can omit the bitmap in the actual compiled code. */      such as \S in the class, because in that case all characters > 255 are in
3385        the class, so any that were explicitly given as well can be ignored. If
3386        (when there are explicit characters > 255 that must be listed) there are no
3387        characters < 256, we can omit the bitmap in the actual compiled code. */
3388    
3389  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3390      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3391        {        {
3392        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3393        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3034  for (;; ptr++) Line 3413  for (;; ptr++)
3413        }        }
3414  #endif  #endif
3415    
3416      /* If there are no characters > 255, negate the 32-byte map if necessary,      /* If there are no characters > 255, set the opcode to OP_CLASS or
3417      and copy it into the code vector. If this is the first thing in the branch,      OP_NCLASS, depending on whether the whole class was negated and whether
3418      there can be no first char setting, whatever the repeat count. Any reqbyte      there were negative specials such as \S in the class. Then copy the 32-byte
3419      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3420    
3421        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3422      if (negate_class)      if (negate_class)
3423        {        {
       *code++ = OP_NCLASS;  
3424        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3425          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3426        }        }
3427      else      else
3428        {        {
       *code++ = OP_CLASS;  
3429        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3430        }        }
3431      code += 32;      code += 32;
# Line 3233  for (;; ptr++) Line 3611  for (;; ptr++)
3611        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3612        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3613    
3614        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3615    
3616        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3617    
# Line 3383  for (;; ptr++) Line 3761  for (;; ptr++)
3761        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3762        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3763    
3764        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3765    
3766        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3767          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3419  for (;; ptr++) Line 3797  for (;; ptr++)
3797          goto FAILED;          goto FAILED;
3798          }          }
3799    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3800        /* If the maximum repeat count is unlimited, find the end of the bracket        /* If the maximum repeat count is unlimited, find the end of the bracket
3801        by scanning through from the start, and compute the offset back to it        by scanning through from the start, and compute the offset back to it
3802        from the current code pointer. There may be an OP_OPT setting following        from the current code pointer. There may be an OP_OPT setting following
# Line 3515  for (;; ptr++) Line 3885  for (;; ptr++)
3885          if (repeat_min > 1)          if (repeat_min > 1)
3886            {            {
3887            /* 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
3888            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3889              potential integer overflow. */
3890    
3891            if (lengthptr != NULL)            if (lengthptr != NULL)
3892              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3893                int delta = (repeat_min - 1)*length_prevgroup;
3894                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3895                                                                (double)INT_MAX ||
3896                    OFLOW_MAX - *lengthptr < delta)
3897                  {
3898                  *errorcodeptr = ERR20;
3899                  goto FAILED;
3900                  }
3901                *lengthptr += delta;
3902                }
3903    
3904            /* This is compiling for real */            /* This is compiling for real */
3905    
# Line 3556  for (;; ptr++) Line 3937  for (;; ptr++)
3937          /* 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
3938          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
3939          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
3940          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3941            paranoid checks to avoid integer overflow. */
3942    
3943          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3944            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3945              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3946                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3947              if ((double)repeat_max *
3948                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3949                      > (double)INT_MAX ||
3950                  OFLOW_MAX - *lengthptr < delta)
3951                {
3952                *errorcodeptr = ERR20;
3953                goto FAILED;
3954                }
3955              *lengthptr += delta;
3956              }
3957    
3958          /* This is compiling for real */          /* This is compiling for real */
3959    
# Line 3668  for (;; ptr++) Line 4061  for (;; ptr++)
4061        int len;        int len;
4062        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||
4063            *tempcode == OP_NOTEXACT)            *tempcode == OP_NOTEXACT)
4064          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode] +
4065              ((*tempcode == OP_TYPEEXACT &&
4066                 (tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0);
4067        len = code - tempcode;        len = code - tempcode;
4068        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4069          {          {
# Line 3712  for (;; ptr++) Line 4107  for (;; ptr++)
4107      /* ===================================================================*/      /* ===================================================================*/
4108      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4109      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4110      parenthesis forms. First deal with the specials; all are introduced by ?,      parenthesis forms.  */
     and the appearance of any of them means that this is not a capturing  
     group. */  
4111    
4112      case '(':      case '(':
4113      newoptions = options;      newoptions = options;
4114      skipbytes = 0;      skipbytes = 0;
4115      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4116      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4117      reset_bracount = FALSE;      reset_bracount = FALSE;
4118    
4119        /* First deal with various "verbs" that can be introduced by '*'. */
4120    
4121        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4122          {
4123          int i, namelen;
4124          const char *vn = verbnames;
4125          const uschar *name = ++ptr;
4126          previous = NULL;
4127          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
4128          if (*ptr == ':')
4129            {
4130            *errorcodeptr = ERR59;   /* Not supported */
4131            goto FAILED;
4132            }
4133          if (*ptr != ')')
4134            {
4135            *errorcodeptr = ERR60;
4136            goto FAILED;
4137            }
4138          namelen = ptr - name;
4139          for (i = 0; i < verbcount; i++)
4140            {
4141            if (namelen == verbs[i].len &&
4142                strncmp((char *)name, vn, namelen) == 0)
4143              {
4144              *code = verbs[i].op;
4145              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4146              break;
4147              }
4148            vn += verbs[i].len + 1;
4149            }
4150          if (i < verbcount) continue;
4151          *errorcodeptr = ERR60;
4152          goto FAILED;
4153          }
4154    
4155      if (*(++ptr) == '?')      /* Deal with the extended parentheses; all are introduced by '?', and the
4156        appearance of any of them means that this is not a capturing group. */
4157    
4158        else if (*ptr == '?')
4159        {        {
4160        int i, set, unset, namelen;        int i, set, unset, namelen;
4161        int *optset;        int *optset;
# Line 3746  for (;; ptr++) Line 4178  for (;; ptr++)
4178          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4179          case '|':                 /* Reset capture count for each branch */          case '|':                 /* Reset capture count for each branch */
4180          reset_bracount = TRUE;          reset_bracount = TRUE;
4181          /* Fall through */          /* Fall through */
4182    
4183          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4184          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
# Line 3858  for (;; ptr++) Line 4290  for (;; ptr++)
4290              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4291              goto FAILED;              goto FAILED;
4292              }              }
4293            if (refsign == '-')            recno = (refsign == '-')?
4294                cd->bracount - recno + 1 : recno +cd->bracount;
4295              if (recno <= 0 || recno > cd->final_bracount)
4296              {              {
4297              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4298              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4299              }              }
           else recno += cd->bracount;  
4300            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4301            break;            break;
4302            }            }
# Line 3939  for (;; ptr++) Line 4368  for (;; ptr++)
4368            skipbytes = 1;            skipbytes = 1;
4369            }            }
4370    
4371          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4372            in the second pass here, so final_bracount is set. */
4373    
4374          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4375            {            {
4376            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4377            }            }
# Line 3965  for (;; ptr++) Line 4395  for (;; ptr++)
4395    
4396          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4397          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4398          ptr++;          ptr++;
4399            if (*ptr == ')')          /* Optimize (?!) */
4400              {
4401              *code++ = OP_FAIL;
4402              previous = NULL;
4403              continue;
4404              }
4405            bravalue = OP_ASSERT_NOT;
4406          break;          break;
4407    
4408    
# Line 4129  for (;; ptr++) Line 4565  for (;; ptr++)
4565    
4566          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4567          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4568          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4569            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4570            .NET syntax. */
4571    
4572          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4573          name = ++ptr;          name = ++ptr;
# Line 4141  for (;; ptr++) Line 4579  for (;; ptr++)
4579    
4580          if (lengthptr != NULL)          if (lengthptr != NULL)
4581            {            {
4582              if (namelen == 0)
4583                {
4584                *errorcodeptr = ERR62;
4585                goto FAILED;
4586                }
4587            if (*ptr != terminator)            if (*ptr != terminator)
4588              {              {
4589              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4154  for (;; ptr++) Line 4597  for (;; ptr++)
4597            recno = 0;            recno = 0;
4598            }            }
4599    
4600          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4601            first, and then check that we have reached the end of the name in the
4602            table. That way, if the name that is longer than any in the table,
4603            the comparison will fail without reading beyond the table entry. */
4604    
4605          else          else
4606            {            {
4607            slot = cd->name_table;            slot = cd->name_table;
4608            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4609              {              {
4610              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4611                    slot[2+namelen] == 0)
4612                  break;
4613              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4614              }              }
4615    
# Line 4198  for (;; ptr++) Line 4646  for (;; ptr++)
4646            {            {
4647            const uschar *called;            const uschar *called;
4648    
4649            if ((refsign = *ptr) == '+') ptr++;            if ((refsign = *ptr) == '+')
4650                {
4651                ptr++;
4652                if ((digitab[*ptr] & ctype_digit) == 0)
4653                  {
4654                  *errorcodeptr = ERR63;
4655                  goto FAILED;
4656                  }
4657                }
4658            else if (refsign == '-')            else if (refsign == '-')
4659              {              {
4660              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
# Line 4324  for (;; ptr++) Line 4780  for (;; ptr++)
4780    
4781              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4782              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4783              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4784              break;              break;
4785    
4786              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4447  for (;; ptr++) Line 4903  for (;; ptr++)
4903           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
4904           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
4905            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
4906           reset_bracount,               /* True if (?| group */           reset_bracount,               /* True if (?| group */
4907           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
4908           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
4909           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4515  for (;; ptr++) Line 4971  for (;; ptr++)
4971        goto FAILED;        goto FAILED;
4972        }        }
4973    
4974      /* In the pre-compile phase, update the length by the length of the nested      /* In the pre-compile phase, update the length by the length of the group,
4975      group, less the brackets at either end. Then reduce the compiled code to      less the brackets at either end. Then reduce the compiled code to just a
4976      just the brackets so that it doesn't use much memory if it is duplicated by      set of non-capturing brackets so that it doesn't use much memory if it is
4977      a quantifier. */      duplicated by a quantifier.*/
4978    
4979      if (lengthptr != NULL)      if (lengthptr != NULL)
4980        {        {
4981          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
4982            {
4983            *errorcodeptr = ERR20;
4984            goto FAILED;
4985            }
4986        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4987        code++;        *code++ = OP_BRA;
4988        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4989        *code++ = OP_KET;        *code++ = OP_KET;
4990        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4991          break;    /* No need to waste time with special character handling */
4992        }        }
4993    
4994      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4995    
4996      else code = tempcode;      code = tempcode;
4997    
4998      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4999      relevant. */      relevant. */
# Line 4735  for (;; ptr++) Line 5197  for (;; ptr++)
5197      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5198      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5199    
5200        /* Remember if \r or \n were seen */
5201    
5202        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5203          cd->external_flags |= PCRE_HASCRORLF;
5204    
5205      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5206      byte, set it from this character, but revert to none on a zero repeat.      byte, set it from this character, but revert to none on a zero repeat.
5207      Otherwise, leave the firstbyte value alone, and don't change it on a zero      Otherwise, leave the firstbyte value alone, and don't change it on a zero
# Line 4807  Arguments: Line 5274  Arguments:
5274    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5275    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5276    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5277    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
5278    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5279    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5280    reqbyteptr     place to put the last required character, or a negative number    reqbyteptr     place to put the last required character, or a negative number
# Line 4821  Returns: TRUE on success Line 5288  Returns: TRUE on success
5288    
5289  static BOOL  static BOOL
5290  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5291    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5292    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5293    int *lengthptr)    int *lengthptr)
5294  {  {
5295  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
# Line 4867  orig_bracount = max_bracount = cd->braco Line 5334  orig_bracount = max_bracount = cd->braco
5334  for (;;)  for (;;)
5335    {    {
5336    /* For a (?| group, reset the capturing bracket count so that each branch    /* For a (?| group, reset the capturing bracket count so that each branch
5337    uses the same numbers. */    uses the same numbers. */
5338    
5339    if (reset_bracount) cd->bracount = orig_bracount;    if (reset_bracount) cd->bracount = orig_bracount;
5340    
5341    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5342    
5343    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4899  for (;;) Line 5366  for (;;)
5366      *ptrptr = ptr;      *ptrptr = ptr;
5367      return FALSE;      return FALSE;
5368      }      }
5369    
5370    /* Keep the highest bracket count in case (?| was used and some branch    /* Keep the highest bracket count in case (?| was used and some branch
5371    has fewer than the rest. */    has fewer than the rest. */
5372    
5373    if (cd->bracount > max_bracount) max_bracount = cd->bracount;    if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5374    
5375    /* In the real compile phase, there is some post-processing to be done. */    /* In the real compile phase, there is some post-processing to be done. */
# Line 5006  for (;;) Line 5473  for (;;)
5473        *code++ = oldims;        *code++ = oldims;
5474        length += 2;        length += 2;
5475        }        }
5476    
5477      /* Retain the highest bracket number, in case resetting was used. */      /* Retain the highest bracket number, in case resetting was used. */
5478    
5479      cd->bracount = max_bracount;      cd->bracount = max_bracount;
5480    
5481      /* Set values to pass back */      /* Set values to pass back */
# Line 5017  for (;;) Line 5484  for (;;)
5484      *ptrptr = ptr;      *ptrptr = ptr;
5485      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5486      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5487      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5488          {
5489          if (OFLOW_MAX - *lengthptr < length)
5490            {
5491            *errorcodeptr = ERR20;
5492            return FALSE;
5493            }
5494          *lengthptr += length;
5495          }
5496      return TRUE;      return TRUE;
5497      }      }
5498    
# Line 5326  real_pcre *re; Line 5801  real_pcre *re;
5801  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5802  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5803  int errorcode = 0;  int errorcode = 0;
5804    int skipatstart = 0;
5805  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5806  BOOL utf8;  BOOL utf8;
5807  #endif  #endif
# Line 5344  to fill in forward references to subpatt Line 5820  to fill in forward references to subpatt
5820    
5821  uschar cworkspace[COMPILE_WORK_SIZE];  uschar cworkspace[COMPILE_WORK_SIZE];
5822    
   
5823  /* Set this early so that early errors get offset 0. */  /* Set this early so that early errors get offset 0. */
5824    
5825  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern;
# Line 5404  cd->fcc = tables + fcc_offset; Line 5879  cd->fcc = tables + fcc_offset;
5879  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5880  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5881    
5882    /* Check for global one-time settings at the start of the pattern, and remember
5883    the offset for later. */
5884    
5885    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5886      {
5887      int newnl = 0;
5888      int newbsr = 0;
5889    
5890      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5891        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5892      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5893        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5894      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5895        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5896      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5897        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5898      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5899        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5900    
5901      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5902        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5903      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5904        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5905    
5906      if (newnl != 0)
5907        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5908      else if (newbsr != 0)
5909        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5910      else break;
5911      }
5912    
5913    /* Check validity of \R options. */
5914    
5915    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5916      {
5917      case 0:
5918      case PCRE_BSR_ANYCRLF:
5919      case PCRE_BSR_UNICODE:
5920      break;
5921      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5922      }
5923    
5924  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5925  current code allows for fixed one- or two-byte sequences, plus "any" and  current code allows for fixed one- or two-byte sequences, plus "any" and
5926  "anycrlf". */  "anycrlf". */
5927    
5928  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5929    {    {
5930    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5931    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5932    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5933    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5463  to compile parts of the pattern into; th Line 5980  to compile parts of the pattern into; th
5980  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
5981  is a test for its doing so. */  is a test for its doing so. */
5982    
5983  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
5984  cd->names_found = 0;  cd->names_found = 0;
5985  cd->name_entry_size = 0;  cd->name_entry_size = 0;
5986  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5473  cd->hwm = cworkspace; Line 5990  cd->hwm = cworkspace;
5990  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5991  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5992  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5993  cd->external_options = options;  cd->external_options = options;
5994    cd->external_flags = 0;
5995    
5996  /* Now do the pre-compile. On error, errorcode will be set non-zero, so we  /* Now do the pre-compile. On error, errorcode will be set non-zero, so we
5997  don't need to look at the result of the function here. The initial options have  don't need to look at the result of the function here. The initial options have
# Line 5482  been put into the cd block so that they Line 5999  been put into the cd block so that they
5999  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
6000  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
6001    
6002    ptr += skipatstart;
6003  code = cworkspace;  code = cworkspace;
6004  *code = OP_BRA;  *code = OP_BRA;
6005  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
6006    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
6007    &length);    &length);
6008  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
6009    
# Line 5512  if (re == NULL) Line 6030  if (re == NULL)
6030    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
6031    }    }
6032    
6033  /* Put in the magic number, and save the sizes, initial options, and character  /* Put in the magic number, and save the sizes, initial options, internal
6034  table pointer. NULL is used for the default character tables. The nullpad field  flags, and character table pointer. NULL is used for the default character
6035  is at the end; it's there to help in the case when a regex compiled on a system  tables. The nullpad field is at the end; it's there to help in the case when a
6036  with 4-byte pointers is run on another with 8-byte pointers. */  regex compiled on a system with 4-byte pointers is run on another with 8-byte
6037    pointers. */
6038    
6039  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
6040  re->size = size;  re->size = size;
6041  re->options = cd->external_options;  re->options = cd->external_options;
6042    re->flags = cd->external_flags;
6043  re->dummy1 = 0;  re->dummy1 = 0;
6044  re->first_byte = 0;  re->first_byte = 0;
6045  re->req_byte = 0;  re->req_byte = 0;
# Line 5537  field. Reset the bracket count and the n Line 6057  field. Reset the bracket count and the n
6057  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6058  */  */
6059    
6060    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6061  cd->bracount = 0;  cd->bracount = 0;
6062  cd->names_found = 0;  cd->names_found = 0;
6063  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5544  codestart = cd->name_table + re->name_en Line 6065  codestart = cd->name_table + re->name_en
6065  cd->start_code = codestart;  cd->start_code = codestart;
6066  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6067  cd->req_varyopt = 0;  cd->req_varyopt = 0;
6068  cd->nopartial = FALSE;  cd->had_accept = FALSE;
6069    
6070  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6071  error, errorcode will be set non-zero, so we don't need to look at the result  error, errorcode will be set non-zero, so we don't need to look at the result
6072  of the function here. */  of the function here. */
6073    
6074  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6075  code = (uschar *)codestart;  code = (uschar *)codestart;
6076  *code = OP_BRA;  *code = OP_BRA;
6077  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6078    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6079  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6080  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6081    re->flags = cd->external_flags;
6082    
6083  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6084    
6085  /* If not reached end of pattern on success, there's an excess bracket. */  /* If not reached end of pattern on success, there's an excess bracket. */
6086    
# Line 5600  if (errorcode != 0) Line 6122  if (errorcode != 0)
6122    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6123    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6124    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6125    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6126    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6127    return NULL;    return NULL;
6128    }    }
# Line 5629  if ((re->options & PCRE_ANCHORED) == 0) Line 6151  if ((re->options & PCRE_ANCHORED) == 0)
6151        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6152        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6153           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6154        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6155        }        }
6156      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6157        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6158      }      }
6159    }    }
6160    
# Line 5646  if (reqbyte >= 0 && Line 6168  if (reqbyte >= 0 &&
6168    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6169    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6170      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6171    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6172    }    }
6173    
6174  /* 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
# Line 5657  case when building a production library. Line 6179  case when building a production library.
6179  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6180    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6181    
6182  if (re->options != 0)  printf("Options=%08x\n", re->options);
   {  
   printf("%s%s%s%s%s%s%s%s%s\n",  
     ((re->options & PCRE_NOPARTIAL) != 0)? "nopartial " : "",  
     ((re->options & PCRE_ANCHORED) != 0)? "anchored " : "",  
     ((re->options & PCRE_CASELESS) != 0)? "caseless " : "",  
     ((re->options & PCRE_EXTENDED) != 0)? "extended " : "",  
     ((re->options & PCRE_MULTILINE) != 0)? "multiline " : "",  
     ((re->options & PCRE_DOTALL) != 0)? "dotall " : "",  
     ((re->options & PCRE_DOLLAR_ENDONLY) != 0)? "endonly " : "",  
     ((re->options & PCRE_EXTRA) != 0)? "extra " : "",  
     ((re->options & PCRE_UNGREEDY) != 0)? "ungreedy " : "");  
   }  
6183    
6184  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6185    {    {
6186    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6187    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5680  if ((re->options & PCRE_FIRSTSET) != 0) Line 6190  if ((re->options & PCRE_FIRSTSET) != 0)
6190      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6191    }    }
6192    
6193  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6194    {    {
6195    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6196    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5697  was compiled can be seen. */ Line 6207  was compiled can be seen. */
6207  if (code - codestart > length)  if (code - codestart > length)
6208    {    {
6209    (pcre_free)(re);    (pcre_free)(re);
6210    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6211    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6212    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6213    return NULL;    return NULL;

Legend:
Removed from v.178  
changed lines
  Added in v.323

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12