/[pcre]/code/tags/pcre-7.8/pcre_compile.c
ViewVC logotype

Diff of /code/tags/pcre-7.8/pcre_compile.c

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

revision 218 by ph10, Thu Aug 16 10:13:23 2007 UTC revision 333 by ph10, Thu Apr 10 19:55:57 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 43  supporting internal functions that are n Line 43  supporting internal functions that are n
43    
44    
45  #ifdef HAVE_CONFIG_H  #ifdef HAVE_CONFIG_H
46  #include <config.h>  #include "config.h"
47  #endif  #endif
48    
49  #define NLBLOCK cd             /* Block containing newline information */  #define NLBLOCK cd             /* Block containing newline information */
# Line 140  static const short int escapes[] = { Line 140  static const short int escapes[] = {
140  #endif  #endif
141    
142    
143  /* Table of special "verbs" like (*PRUNE) */  /* Table of special "verbs" like (*PRUNE). This is a short table, so it is
144    searched linearly. Put all the names into a single string, in order to reduce
145    the number of relocations when a shared library is dynamically linked. */
146    
147  typedef struct verbitem {  typedef struct verbitem {
   const char *name;  
148    int   len;    int   len;
149    int   op;    int   op;
150  } verbitem;  } verbitem;
151    
152  static verbitem verbs[] = {  static const char verbnames[] =
153    { "ACCEPT", 6, OP_ACCEPT },    "ACCEPT\0"
154    { "COMMIT", 6, OP_COMMIT },    "COMMIT\0"
155    { "F",      1, OP_FAIL },    "F\0"
156    { "FAIL",   4, OP_FAIL },    "FAIL\0"
157    { "PRUNE",  5, OP_PRUNE },    "PRUNE\0"
158    { "SKIP",   4, OP_SKIP  },    "SKIP\0"
159    { "THEN",   4, OP_THEN  }    "THEN";
160    
161    static const 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);  static const int verbcount = sizeof(verbs)/sizeof(verbitem);
172    
173    
174  /* Tables of names of POSIX character classes and their lengths. The list is  /* Tables of names of POSIX character classes and their lengths. The names are
175  terminated by a zero length entry. The first three must be alpha, lower, upper,  now all in a single string, to reduce the number of relocations when a shared
176  as this is assumed for handling case independence. */  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  static const char *const posix_names[] = {  for handling case independence. */
179    "alpha", "lower", "upper",  
180    "alnum", "ascii", "blank", "cntrl", "digit", "graph",  static const char posix_names[] =
181    "print", "punct", "space", "word",  "xdigit" };    "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 207  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 is 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",    /** DEAD **/    "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, angle-bracketed, or quoted name/number or by a plain number\0"
299    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number",    "a numbered reference must not be zero\0"
300    "(*VERB) with an argument is not supported",    "(*VERB) with an argument is not supported\0"
301    /* 60 */    /* 60 */
302    "(*VERB) not recognized",    "(*VERB) not recognized\0"
303    "number is too big"    "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 420  static BOOL Line 438  static BOOL
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 458  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 491  else Line 531  else
531      *errorcodeptr = ERR37;      *errorcodeptr = ERR37;
532      break;      break;
533    
534      /* \g must be followed by a number, either plain or braced. If positive, it      /* \g must be followed by one of a number of specific things:
535      is an absolute backreference. If negative, it is a relative backreference.  
536      This is a Perl 5.10 feature. Perl 5.10 also supports \g{name} as a      (1) A number, either plain or braced. If positive, it is an absolute
537      reference to a named group. This is part of Perl's movement towards a      backreference. If negative, it is a relative backreference. This is a Perl
538      unified syntax for back references. As this is synonymous with \k{name}, we      5.10 feature.
539      fudge it up by pretending it really was \k. */  
540        (2) Perl 5.10 also supports \g{name} as a reference to a named group. This
541        is part of Perl's movement towards a unified syntax for back references. As
542        this is synonymous with \k{name}, we fudge it up by pretending it really
543        was \k.
544    
545        (3) For Oniguruma compatibility we also support \g followed by a name or a
546        number either in angle brackets or in single quotes. However, these are
547        (possibly recursive) subroutine calls, _not_ backreferences. Just return
548        the -ESC_g code (cf \k). */
549    
550      case 'g':      case 'g':
551        if (ptr[1] == '<' || ptr[1] == '\'')
552          {
553          c = -ESC_g;
554          break;
555          }
556    
557        /* Handle the Perl-compatible cases */
558    
559      if (ptr[1] == '{')      if (ptr[1] == '{')
560        {        {
561        const uschar *p;        const uschar *p;
# Line 524  else Line 581  else
581      c = 0;      c = 0;
582      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
583        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - '0';
584    
585      if (c < 0)      if (c < 0)   /* Integer overflow */
586        {        {
587        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
588        break;        break;
589        }        }
590    
591      if (c == 0 || (braced && *(++ptr) != '}'))      if (braced && *(++ptr) != '}')
592        {        {
593        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
594        break;        break;
595        }        }
596    
597        if (c == 0)
598          {
599          *errorcodeptr = ERR58;
600          break;
601          }
602    
603      if (negated)      if (negated)
604        {        {
# Line 571  else Line 634  else
634        c -= '0';        c -= '0';
635        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
636          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - '0';
637        if (c < 0)        if (c < 0)    /* Integer overflow */
638          {          {
639          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
640          break;          break;
641          }          }
642        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
643          {          {
644          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 684  else Line 747  else
747      break;      break;
748    
749      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any
750      other alphameric following \ is an error if PCRE_EXTRA was set; otherwise,      other alphanumeric following \ is an error if PCRE_EXTRA was set;
751      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
752      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
753      in future, so I haven't "optimized" it. */      be again in future, so I haven't "optimized" it. */
754    
755      default:      default:
756      if ((options & PCRE_EXTRA) != 0) switch(c)      if ((options & PCRE_EXTRA) != 0) switch(c)
# Line 776  top = _pcre_utt_size; Line 839  top = _pcre_utt_size;
839  while (bot < top)  while (bot < top)
840    {    {
841    i = (bot + top) >> 1;    i = (bot + top) >> 1;
842    c = strcmp(name, _pcre_utt[i].name);    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);
843    if (c == 0)    if (c == 0)
844      {      {
845      *dptr = _pcre_utt[i].value;      *dptr = _pcre_utt[i].value;
# Line 950  for (; *ptr != 0; ptr++) Line 1013  for (; *ptr != 0; ptr++)
1013      {      {
1014      while (*(++ptr) != ']')      while (*(++ptr) != ']')
1015        {        {
1016        if (*ptr == 0) return -1;        if (*ptr == 0) return -1;
1017        if (*ptr == '\\')        if (*ptr == '\\')
1018          {          {
1019          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 1194  for (;;) Line 1257  for (;;)
1257    
1258      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1259      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1260      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1261      cc += 4;      cc += 4;
1262      break;      break;
1263    
# Line 1303  for (;;) Line 1366  for (;;)
1366      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1367      }      }
1368    
1369    /* Otherwise, we can get the item's length from the table, except that for    /* Otherwise, we can get the item's length from the table, except that for
1370    repeated character types, we have to test for \p and \P, which have an extra    repeated character types, we have to test for \p and \P, which have an extra
1371    two bytes of parameters. */    two bytes of parameters. */
1372    
1373    else    else
# Line 1317  for (;;) Line 1380  for (;;)
1380        case OP_TYPEMINPLUS:        case OP_TYPEMINPLUS:
1381        case OP_TYPEQUERY:        case OP_TYPEQUERY:
1382        case OP_TYPEMINQUERY:        case OP_TYPEMINQUERY:
       case OP_TYPEUPTO:  
       case OP_TYPEMINUPTO:  
       case OP_TYPEEXACT:  
1383        case OP_TYPEPOSSTAR:        case OP_TYPEPOSSTAR:
1384        case OP_TYPEPOSPLUS:        case OP_TYPEPOSPLUS:
1385        case OP_TYPEPOSQUERY:        case OP_TYPEPOSQUERY:
       case OP_TYPEPOSUPTO:  
1386        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1387        break;        break;
1388        }  
1389          case OP_TYPEUPTO:
1390          case OP_TYPEMINUPTO:
1391          case OP_TYPEEXACT:
1392          case OP_TYPEPOSUPTO:
1393          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1394          break;
1395          }
1396    
1397      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
1398    
1399      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1400    
1401    /* In UTF-8 mode, opcodes that are followed by a character may be followed by    /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1402    a multi-byte character. The length in the table is a minimum, so we have to    a multi-byte character. The length in the table is a minimum, so we have to
1403    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
1404    
1405  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1406      if (utf8) switch(c)      if (utf8) switch(c)
1407        {        {
# Line 1386  for (;;) Line 1452  for (;;)
1452    register int c = *code;    register int c = *code;
1453    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
1454    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
1455    
1456    /* XCLASS is used for classes that cannot be represented just by a bit    /* XCLASS is used for classes that cannot be represented just by a bit
1457    map. This includes negated single high-valued characters. The length in    map. This includes negated single high-valued characters. The length in
1458    the table is zero; the actual length is stored in the compiled code. */    the table is zero; the actual length is stored in the compiled code. */
1459    
1460    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1461    
1462    /* Otherwise, we can get the item's length from the table, except that for    /* Otherwise, we can get the item's length from the table, except that for
1463    repeated character types, we have to test for \p and \P, which have an extra    repeated character types, we have to test for \p and \P, which have an extra
1464    two bytes of parameters. */    two bytes of parameters. */
1465    
1466    else    else
# Line 1407  for (;;) Line 1473  for (;;)
1473        case OP_TYPEMINPLUS:        case OP_TYPEMINPLUS:
1474        case OP_TYPEQUERY:        case OP_TYPEQUERY:
1475        case OP_TYPEMINQUERY:        case OP_TYPEMINQUERY:
       case OP_TYPEUPTO:  
       case OP_TYPEMINUPTO:  
       case OP_TYPEEXACT:  
1476        case OP_TYPEPOSSTAR:        case OP_TYPEPOSSTAR:
1477        case OP_TYPEPOSPLUS:        case OP_TYPEPOSPLUS:
1478        case OP_TYPEPOSQUERY:        case OP_TYPEPOSQUERY:
       case OP_TYPEPOSUPTO:  
1479        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1480        break;        break;
1481        }  
1482          case OP_TYPEPOSUPTO:
1483          case OP_TYPEUPTO:
1484          case OP_TYPEMINUPTO:
1485          case OP_TYPEEXACT:
1486          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1487          break;
1488          }
1489    
1490      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
1491    
1492      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1493    
1494      /* In UTF-8 mode, opcodes that are followed by a character may be followed      /* In UTF-8 mode, opcodes that are followed by a character may be followed
1495      by a multi-byte character. The length in the table is a minimum, so we have      by a multi-byte character. The length in the table is a minimum, so we have
1496      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
1497    
1498  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1499      if (utf8) switch(c)      if (utf8) switch(c)
1500        {        {
# Line 1462  for (;;) Line 1531  for (;;)
1531  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()
1532  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
1533  group that can match nothing. Note that first_significant_code() skips over  group that can match nothing. Note that first_significant_code() skips over
1534  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
1535  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
1536    bracket whose current branch will already have been scanned.
1537    
1538  Arguments:  Arguments:
1539    code        points to start of search    code        points to start of search
# Line 1485  for (code = first_significant_code(code Line 1555  for (code = first_significant_code(code
1555    
1556    c = *code;    c = *code;
1557    
1558      /* Skip over forward assertions; the other assertions are skipped by
1559      first_significant_code() with a TRUE final argument. */
1560    
1561      if (c == OP_ASSERT)
1562        {
1563        do code += GET(code, 1); while (*code == OP_ALT);
1564        c = *code;
1565        continue;
1566        }
1567    
1568    /* Groups with zero repeats can of course be empty; skip them. */    /* Groups with zero repeats can of course be empty; skip them. */
1569    
1570    if (c == OP_BRAZERO || c == OP_BRAMINZERO)    if (c == OP_BRAZERO || c == OP_BRAMINZERO)
# Line 1524  for (code = first_significant_code(code Line 1604  for (code = first_significant_code(code
1604      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
1605      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
1606      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1607      actual length is stored in the compiled code, so we must update "code"      actual length is stored in the compiled code, so we must update "code"
1608      here. */      here. */
1609    
1610  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1591  for (code = first_significant_code(code Line 1671  for (code = first_significant_code(code
1671      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1672      return FALSE;      return FALSE;
1673    
1674        /* These are going to continue, as they may be empty, but we have to
1675        fudge the length for the \p and \P cases. */
1676    
1677        case OP_TYPESTAR:
1678        case OP_TYPEMINSTAR:
1679        case OP_TYPEPOSSTAR:
1680        case OP_TYPEQUERY:
1681        case OP_TYPEMINQUERY:
1682        case OP_TYPEPOSQUERY:
1683        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1684        break;
1685    
1686        /* Same for these */
1687    
1688        case OP_TYPEUPTO:
1689        case OP_TYPEMINUPTO:
1690        case OP_TYPEPOSUPTO:
1691        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1692        break;
1693    
1694      /* End of branch */      /* End of branch */
1695    
1696      case OP_KET:      case OP_KET:
# Line 1660  return TRUE; Line 1760  return TRUE;
1760  *************************************************/  *************************************************/
1761    
1762  /* This function is called when the sequence "[:" or "[." or "[=" is  /* This function is called when the sequence "[:" or "[." or "[=" is
1763  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
1764  optional ^ and then a sequence of letters, terminated by a matching ":]" or  sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
1765  ".]" or "=]".  reach an unescaped ']' without the special preceding character, return FALSE.
1766    
1767    Originally, this function only recognized a sequence of letters between the
1768    terminators, but it seems that Perl recognizes any sequence of characters,
1769    though of course unknown POSIX names are subsequently rejected. Perl gives an
1770    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
1771    didn't consider this to be a POSIX class. Likewise for [:1234:].
1772    
1773    The problem in trying to be exactly like Perl is in the handling of escapes. We
1774    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
1775    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
1776    below handles the special case of \], but does not try to do any other escape
1777    processing. This makes it different from Perl for cases such as [:l\ower:]
1778    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
1779    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
1780    I think.
1781    
1782  Argument:  Arguments:
1783    ptr      pointer to the initial [    ptr      pointer to the initial [
1784    endptr   where to return the end pointer    endptr   where to return the end pointer
   cd       pointer to compile data  
1785    
1786  Returns:   TRUE or FALSE  Returns:   TRUE or FALSE
1787  */  */
1788    
1789  static BOOL  static BOOL
1790  check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd)  check_posix_syntax(const uschar *ptr, const uschar **endptr)
1791  {  {
1792  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
1793  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
1794  if (*(++ptr) == '^') ptr++;  for (++ptr; *ptr != 0; ptr++)
 while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;  
 if (*ptr == terminator && ptr[1] == ']')  
1795    {    {
1796    *endptr = ptr;    if (*ptr == '\\' && ptr[1] == ']') ptr++; else
1797    return TRUE;      {
1798        if (*ptr == ']') return FALSE;
1799        if (*ptr == terminator && ptr[1] == ']')
1800          {
1801          *endptr = ptr;
1802          return TRUE;
1803          }
1804        }
1805    }    }
1806  return FALSE;  return FALSE;
1807  }  }
# Line 1707  Returns: a value representing the na Line 1826  Returns: a value representing the na
1826  static int  static int
1827  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1828  {  {
1829    const char *pn = posix_names;
1830  register int yield = 0;  register int yield = 0;
1831  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1832    {    {
1833    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1834      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1835      pn += posix_name_lengths[yield] + 1;
1836    yield++;    yield++;
1837    }    }
1838  return -1;  return -1;
# Line 1753  adjust_recurse(uschar *group, int adjust Line 1874  adjust_recurse(uschar *group, int adjust
1874    uschar *save_hwm)    uschar *save_hwm)
1875  {  {
1876  uschar *ptr = group;  uschar *ptr = group;
1877    
1878  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1879    {    {
1880    int offset;    int offset;
# Line 2014  if (next >= 0) switch(op_code) Line 2136  if (next >= 0) switch(op_code)
2136    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, "item" must be a single-byte character. */
2137    
2138    case OP_NOT:    case OP_NOT:
   if (next < 0) return FALSE;  /* Not a character */  
2139    if (item == next) return TRUE;    if (item == next) return TRUE;
2140    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2141  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 2277  uschar classbits[32]; Line 2398  uschar classbits[32];
2398  BOOL class_utf8;  BOOL class_utf8;
2399  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8 = (options & PCRE_UTF8) != 0;
2400  uschar *class_utf8data;  uschar *class_utf8data;
2401    uschar *class_utf8data_base;
2402  uschar utf8_char[6];  uschar utf8_char[6];
2403  #else  #else
2404  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
# Line 2316  req_caseopt = ((options & PCRE_CASELESS) Line 2438  req_caseopt = ((options & PCRE_CASELESS)
2438  for (;; ptr++)  for (;; ptr++)
2439    {    {
2440    BOOL negate_class;    BOOL negate_class;
2441      BOOL should_flip_negation;
2442    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2443    BOOL is_quantifier;    BOOL is_quantifier;
2444    BOOL is_recurse;    BOOL is_recurse;
# Line 2539  for (;; ptr++) Line 2662  for (;; ptr++)
2662      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. */
2663    
2664      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2665          check_posix_syntax(ptr, &tempptr, cd))          check_posix_syntax(ptr, &tempptr))
2666        {        {
2667        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;
2668        goto FAILED;        goto FAILED;
# Line 2564  for (;; ptr++) Line 2687  for (;; ptr++)
2687        else break;        else break;
2688        }        }
2689    
2690        /* If a class contains a negative special such as \S, we need to flip the
2691        negation flag at the end, so that support for characters > 255 works
2692        correctly (they are all included in the class). */
2693    
2694        should_flip_negation = FALSE;
2695    
2696      /* 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
2697      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
2698      valued UTF-8 characters, we don't yet do any optimization. */      valued UTF-8 characters, we don't yet do any optimization. */
# Line 2581  for (;; ptr++) Line 2710  for (;; ptr++)
2710  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2711      class_utf8 = FALSE;                       /* No chars >= 256 */      class_utf8 = FALSE;                       /* No chars >= 256 */
2712      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */
2713        class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */
2714  #endif  #endif
2715    
2716      /* 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 2596  for (;; ptr++) Line 2726  for (;; ptr++)
2726          {                           /* Braces are required because the */          {                           /* Braces are required because the */
2727          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
2728          }          }
2729    
2730          /* In the pre-compile phase, accumulate the length of any UTF-8 extra
2731          data and reset the pointer. This is so that very large classes that
2732          contain a zillion UTF-8 characters no longer overwrite the work space
2733          (which is on the stack). */
2734    
2735          if (lengthptr != NULL)
2736            {
2737            *lengthptr += class_utf8data - class_utf8data_base;
2738            class_utf8data = class_utf8data_base;
2739            }
2740    
2741  #endif  #endif
2742    
2743        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 2619  for (;; ptr++) Line 2761  for (;; ptr++)
2761    
2762        if (c == '[' &&        if (c == '[' &&
2763            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2764            check_posix_syntax(ptr, &tempptr, cd))            check_posix_syntax(ptr, &tempptr))
2765          {          {
2766          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
2767          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
# Line 2636  for (;; ptr++) Line 2778  for (;; ptr++)
2778          if (*ptr == '^')          if (*ptr == '^')
2779            {            {
2780            local_negate = TRUE;            local_negate = TRUE;
2781              should_flip_negation = TRUE;  /* Note negative special */
2782            ptr++;            ptr++;
2783            }            }
2784    
# Line 2710  for (;; ptr++) Line 2853  for (;; ptr++)
2853          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2854          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2855    
2856          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = '\b';       /* \b is backspace in a class */
2857          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 */
2858          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 */
2859          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
# Line 2722  for (;; ptr++) Line 2865  for (;; ptr++)
2865            else inescq = TRUE;            else inescq = TRUE;
2866            continue;            continue;
2867            }            }
2868          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2869    
2870          if (c < 0)          if (c < 0)
2871            {            {
# Line 2738  for (;; ptr++) Line 2881  for (;; ptr++)
2881              continue;              continue;
2882    
2883              case ESC_D:              case ESC_D:
2884                should_flip_negation = TRUE;
2885              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2886              continue;              continue;
2887    
# Line 2746  for (;; ptr++) Line 2890  for (;; ptr++)
2890              continue;              continue;
2891    
2892              case ESC_W:              case ESC_W:
2893                should_flip_negation = TRUE;
2894              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2895              continue;              continue;
2896    
# Line 2755  for (;; ptr++) Line 2900  for (;; ptr++)
2900              continue;              continue;
2901    
2902              case ESC_S:              case ESC_S:
2903                should_flip_negation = TRUE;
2904              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2905              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2906              continue;              continue;
2907    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
2908              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
2909              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
2910              }              }
# Line 2949  for (;; ptr++) Line 3092  for (;; ptr++)
3092    
3093        oldptr = ptr;        oldptr = ptr;
3094    
3095          /* Remember \r or \n */
3096    
3097          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3098    
3099          /* Check for range */
3100    
3101        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3102          {          {
3103          int d;          int d;
# Line 2990  for (;; ptr++) Line 3139  for (;; ptr++)
3139            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3140            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3141    
3142            /* \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
3143            special means the '-' was literal */            special means the '-' was literal */
3144    
3145            if (d < 0)            if (d < 0)
# Line 3016  for (;; ptr++) Line 3165  for (;; ptr++)
3165    
3166          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3167    
3168            /* Remember \r or \n */
3169    
3170            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3171    
3172          /* 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
3173          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3174          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 3169  for (;; ptr++) Line 3322  for (;; ptr++)
3322        goto FAILED;        goto FAILED;
3323        }        }
3324    
3325    
3326    /* This code has been disabled because it would mean that \s counts as
3327    an explicit \r or \n reference, and that's not really what is wanted. Now
3328    we set the flag only if there is a literal "\r" or "\n" in the class. */
3329    
3330    #if 0
3331        /* Remember whether \r or \n are in this class */
3332    
3333        if (negate_class)
3334          {
3335          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3336          }
3337        else
3338          {
3339          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3340          }
3341    #endif
3342    
3343    
3344      /* 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
3345      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
3346      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
3347      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3348      single-bytes only. This is an historical hangover. Maybe one day we can  
3349      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3350        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3351        operate on single-bytes only. This is an historical hangover. Maybe one day
3352        we can tidy these opcodes to handle multi-byte characters.
3353    
3354      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
3355      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 3184  for (;; ptr++) Line 3359  for (;; ptr++)
3359      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3360    
3361  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3362      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3363            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3364  #else  #else
3365      if (class_charcount == 1)      if (class_charcount == 1)
3366  #endif  #endif
# Line 3230  for (;; ptr++) Line 3403  for (;; ptr++)
3403      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3404    
3405      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3406      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3407      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
3408        the class, so any that were explicitly given as well can be ignored. If
3409        (when there are explicit characters > 255 that must be listed) there are no
3410        characters < 256, we can omit the bitmap in the actual compiled code. */
3411    
3412  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3413      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3414        {        {
3415        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3416        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3260  for (;; ptr++) Line 3436  for (;; ptr++)
3436        }        }
3437  #endif  #endif
3438    
3439      /* 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
3440      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
3441      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
3442      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3443    
3444        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3445      if (negate_class)      if (negate_class)
3446        {        {
       *code++ = OP_NCLASS;  
3447        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3448          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3449        }        }
3450      else      else
3451        {        {
       *code++ = OP_CLASS;  
3452        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3453        }        }
3454      code += 32;      code += 32;
# Line 3459  for (;; ptr++) Line 3634  for (;; ptr++)
3634        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3635        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3636    
3637        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3638    
3639        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3640    
# Line 3609  for (;; ptr++) Line 3784  for (;; ptr++)
3784        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3785        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3786    
3787        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3788    
3789        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3790          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3909  for (;; ptr++) Line 4084  for (;; ptr++)
4084        int len;        int len;
4085        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||
4086            *tempcode == OP_NOTEXACT)            *tempcode == OP_NOTEXACT)
4087          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode] +
4088              ((*tempcode == OP_TYPEEXACT &&
4089                 (tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0);
4090        len = code - tempcode;        len = code - tempcode;
4091        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4092          {          {
# Line 3967  for (;; ptr++) Line 4144  for (;; ptr++)
4144      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)      if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4145        {        {
4146        int i, namelen;        int i, namelen;
4147          const char *vn = verbnames;
4148        const uschar *name = ++ptr;        const uschar *name = ++ptr;
4149        previous = NULL;        previous = NULL;
4150        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);        while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
# Line 3984  for (;; ptr++) Line 4162  for (;; ptr++)
4162        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
4163          {          {
4164          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
4165              strncmp((char *)name, verbs[i].name, namelen) == 0)              strncmp((char *)name, vn, namelen) == 0)
4166            {            {
4167            *code = verbs[i].op;            *code = verbs[i].op;
4168            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;            if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4169            break;            break;
4170            }            }
4171            vn += verbs[i].len + 1;
4172          }          }
4173        if (i < verbcount) continue;        if (i < verbcount) continue;
4174        *errorcodeptr = ERR60;        *errorcodeptr = ERR60;
# Line 4134  for (;; ptr++) Line 4313  for (;; ptr++)
4313              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4314              goto FAILED;              goto FAILED;
4315              }              }
4316            if (refsign == '-')            recno = (refsign == '-')?
4317                cd->bracount - recno + 1 : recno +cd->bracount;
4318              if (recno <= 0 || recno > cd->final_bracount)
4319              {              {
4320              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4321              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4322              }              }
           else recno += cd->bracount;  
4323            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4324            break;            break;
4325            }            }
# Line 4215  for (;; ptr++) Line 4391  for (;; ptr++)
4391            skipbytes = 1;            skipbytes = 1;
4392            }            }
4393    
4394          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4395            in the second pass here, so final_bracount is set. */
4396    
4397          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4398            {            {
4399            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4400            }            }
# Line 4411  for (;; ptr++) Line 4588  for (;; ptr++)
4588    
4589          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4590          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4591          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4592            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4593            .NET syntax, and the Oniguruma \g<...> and \g'...' subroutine syntax. */
4594    
4595          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4596          name = ++ptr;          name = ++ptr;
# Line 4423  for (;; ptr++) Line 4602  for (;; ptr++)
4602    
4603          if (lengthptr != NULL)          if (lengthptr != NULL)
4604            {            {
4605              if (namelen == 0)
4606                {
4607                *errorcodeptr = ERR62;
4608                goto FAILED;
4609                }
4610            if (*ptr != terminator)            if (*ptr != terminator)
4611              {              {
4612              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4436  for (;; ptr++) Line 4620  for (;; ptr++)
4620            recno = 0;            recno = 0;
4621            }            }
4622    
4623          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4624            first, and then check that we have reached the end of the name in the
4625            table. That way, if the name that is longer than any in the table,
4626            the comparison will fail without reading beyond the table entry. */
4627    
4628          else          else
4629            {            {
4630            slot = cd->name_table;            slot = cd->name_table;
4631            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4632              {              {
4633              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4634                    slot[2+namelen] == 0)
4635                  break;
4636              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4637              }              }
4638    
# Line 4479  for (;; ptr++) Line 4668  for (;; ptr++)
4668          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case '5': case '6': case '7': case '8': case '9':   /* subroutine */
4669            {            {
4670            const uschar *called;            const uschar *called;
4671              terminator = ')';
4672    
4673              /* Come here from the \g<...> and \g'...' code (Oniguruma
4674              compatibility). However, the syntax has been checked to ensure that
4675              the ... are a (signed) number, so that neither ERR63 nor ERR29 will
4676              be called on this path, nor with the jump to OTHER_CHAR_AFTER_QUERY
4677              ever be taken. */
4678    
4679              HANDLE_NUMERICAL_RECURSION:
4680    
4681            if ((refsign = *ptr) == '+') ptr++;            if ((refsign = *ptr) == '+')
4682                {
4683                ptr++;
4684                if ((digitab[*ptr] & ctype_digit) == 0)
4685                  {
4686                  *errorcodeptr = ERR63;
4687                  goto FAILED;
4688                  }
4689                }
4690            else if (refsign == '-')            else if (refsign == '-')
4691              {              {
4692              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
# Line 4492  for (;; ptr++) Line 4698  for (;; ptr++)
4698            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4699              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - '0';
4700    
4701            if (*ptr != ')')            if (*ptr != terminator)
4702              {              {
4703              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4704              goto FAILED;              goto FAILED;
# Line 4606  for (;; ptr++) Line 4812  for (;; ptr++)
4812    
4813              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4814              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4815              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4816              break;              break;
4817    
4818              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4798  for (;; ptr++) Line 5004  for (;; ptr++)
5004        }        }
5005    
5006      /* In the pre-compile phase, update the length by the length of the group,      /* In the pre-compile phase, update the length by the length of the group,
5007      less the brackets at either end. Then reduce the compiled code to just a      less the brackets at either end. Then reduce the compiled code to just a
5008      set of non-capturing brackets so that it doesn't use much memory if it is      set of non-capturing brackets so that it doesn't use much memory if it is
5009      duplicated by a quantifier.*/      duplicated by a quantifier.*/
5010    
# Line 4810  for (;; ptr++) Line 5016  for (;; ptr++)
5016          goto FAILED;          goto FAILED;
5017          }          }
5018        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
5019        *code++ = OP_BRA;        *code++ = OP_BRA;
5020        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
5021        *code++ = OP_KET;        *code++ = OP_KET;
5022        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
# Line 4888  for (;; ptr++) Line 5094  for (;; ptr++)
5094      back references and those types that consume a character may be repeated.      back references and those types that consume a character may be repeated.
5095      We can test for values between ESC_b and ESC_Z for the latter; this may      We can test for values between ESC_b and ESC_Z for the latter; this may
5096      have to change if any new ones are ever created. */      have to change if any new ones are ever created. */
5097    
5098      case '\\':      case '\\':
5099      tempptr = ptr;      tempptr = ptr;
5100      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);
# Line 4915  for (;; ptr++) Line 5121  for (;; ptr++)
5121    
5122        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
5123        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
5124    
5125          /* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'
5126          is a subroutine call by number (Oniguruma syntax). In fact, the value
5127          -ESC_g is returned only for these cases. So we don't need to check for <
5128          or ' if the value is -ESC_g. For the Perl syntax \g{n} the value is
5129          -ESC_REF+n, and for the Perl syntax \g{name} the result is -ESC_k (as
5130          that is a synonym). */
5131    
5132          if (-c == ESC_g)
5133            {
5134            const uschar *p;
5135            terminator = (*(++ptr) == '<')? '>' : '\'';
5136    
5137            /* These two statements stop the compiler for warning about possibly
5138            unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In
5139            fact, because we actually check for a number below, the paths that
5140            would actually be in error are never taken. */
5141    
5142            skipbytes = 0;
5143            reset_bracount = FALSE;
5144    
5145            /* Test for a name */
5146    
5147            if (ptr[1] != '+' && ptr[1] != '-')
5148              {
5149              BOOL isnumber = TRUE;
5150              for (p = ptr + 1; *p != 0 && *p != terminator; p++)
5151                {
5152                if ((cd->ctypes[*p] & ctype_digit) == 0) isnumber = FALSE;
5153                if ((cd->ctypes[*p] & ctype_word) == 0) break;
5154                }
5155              if (*p != terminator)
5156                {
5157                *errorcodeptr = ERR57;
5158                break;
5159                }
5160              if (isnumber)
5161                {
5162                ptr++;
5163                goto HANDLE_NUMERICAL_RECURSION;
5164                }
5165              is_recurse = TRUE;
5166              goto NAMED_REF_OR_RECURSE;
5167              }
5168    
5169            /* Test a signed number in angle brackets or quotes. */
5170    
5171            p = ptr + 2;
5172            while ((digitab[*p] & ctype_digit) != 0) p++;
5173            if (*p != terminator)
5174              {
5175              *errorcodeptr = ERR57;
5176              break;
5177              }
5178            ptr++;
5179            goto HANDLE_NUMERICAL_RECURSION;
5180            }
5181    
5182        /* \k<name> or \k'name' is a back reference by name (Perl syntax).        /* \k<name> or \k'name' is a back reference by name (Perl syntax).
5183        We also support \k{name} (.NET syntax) */        We also support \k{name} (.NET syntax) */
# Line 5023  for (;; ptr++) Line 5286  for (;; ptr++)
5286      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5287      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5288    
5289        /* Remember if \r or \n were seen */
5290    
5291        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5292          cd->external_flags |= PCRE_HASCRORLF;
5293    
5294      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5295      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.
5296      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 5622  real_pcre *re; Line 5890  real_pcre *re;
5890  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5891  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5892  int errorcode = 0;  int errorcode = 0;
5893    int skipatstart = 0;
5894  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5895  BOOL utf8;  BOOL utf8;
5896  #endif  #endif
# Line 5640  to fill in forward references to subpatt Line 5909  to fill in forward references to subpatt
5909    
5910  uschar cworkspace[COMPILE_WORK_SIZE];  uschar cworkspace[COMPILE_WORK_SIZE];
5911    
   
5912  /* Set this early so that early errors get offset 0. */  /* Set this early so that early errors get offset 0. */
5913    
5914  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern;
# Line 5700  cd->fcc = tables + fcc_offset; Line 5968  cd->fcc = tables + fcc_offset;
5968  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5969  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5970    
5971    /* Check for global one-time settings at the start of the pattern, and remember
5972    the offset for later. */
5973    
5974    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5975      {
5976      int newnl = 0;
5977      int newbsr = 0;
5978    
5979      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5980        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5981      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5982        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5983      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5984        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5985      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5986        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5987      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5988        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5989    
5990      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5991        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5992      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5993        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5994    
5995      if (newnl != 0)
5996        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5997      else if (newbsr != 0)
5998        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5999      else break;
6000      }
6001    
6002    /* Check validity of \R options. */
6003    
6004    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
6005      {
6006      case 0:
6007      case PCRE_BSR_ANYCRLF:
6008      case PCRE_BSR_UNICODE:
6009      break;
6010      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
6011      }
6012    
6013  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
6014  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
6015  "anycrlf". */  "anycrlf". */
6016    
6017  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
6018    {    {
6019    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
6020    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
6021    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
6022    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5759  to compile parts of the pattern into; th Line 6069  to compile parts of the pattern into; th
6069  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
6070  is a test for its doing so. */  is a test for its doing so. */
6071    
6072  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
6073  cd->names_found = 0;  cd->names_found = 0;
6074  cd->name_entry_size = 0;  cd->name_entry_size = 0;
6075  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5769  cd->hwm = cworkspace; Line 6079  cd->hwm = cworkspace;
6079  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
6080  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
6081  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
6082  cd->external_options = options;  cd->external_options = options;
6083    cd->external_flags = 0;
6084    
6085  /* 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
6086  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 5778  been put into the cd block so that they Line 6088  been put into the cd block so that they
6088  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
6089  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
6090    
6091    ptr += skipatstart;
6092  code = cworkspace;  code = cworkspace;
6093  *code = OP_BRA;  *code = OP_BRA;
6094  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
# Line 5808  if (re == NULL) Line 6119  if (re == NULL)
6119    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
6120    }    }
6121    
6122  /* 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
6123  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
6124  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
6125  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
6126    pointers. */
6127    
6128  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
6129  re->size = size;  re->size = size;
6130  re->options = cd->external_options;  re->options = cd->external_options;
6131    re->flags = cd->external_flags;
6132  re->dummy1 = 0;  re->dummy1 = 0;
6133  re->first_byte = 0;  re->first_byte = 0;
6134  re->req_byte = 0;  re->req_byte = 0;
# Line 5833  field. Reset the bracket count and the n Line 6146  field. Reset the bracket count and the n
6146  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6147  */  */
6148    
6149    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6150  cd->bracount = 0;  cd->bracount = 0;
6151  cd->names_found = 0;  cd->names_found = 0;
6152  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5840  codestart = cd->name_table + re->name_en Line 6154  codestart = cd->name_table + re->name_en
6154  cd->start_code = codestart;  cd->start_code = codestart;
6155  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6156  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
6157  cd->had_accept = FALSE;  cd->had_accept = FALSE;
6158    
6159  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6160  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
6161  of the function here. */  of the function here. */
6162    
6163  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6164  code = (uschar *)codestart;  code = (uschar *)codestart;
6165  *code = OP_BRA;  *code = OP_BRA;
6166  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6167    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6168  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6169  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6170    re->flags = cd->external_flags;
6171    
 if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  
6172  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6173    
6174  /* 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. */
# Line 5898  if (errorcode != 0) Line 6211  if (errorcode != 0)
6211    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6212    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6213    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6214    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6215    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6216    return NULL;    return NULL;
6217    }    }
# Line 5927  if ((re->options & PCRE_ANCHORED) == 0) Line 6240  if ((re->options & PCRE_ANCHORED) == 0)
6240        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6241        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6242           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6243        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6244        }        }
6245      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6246        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6247      }      }
6248    }    }
6249    
# Line 5944  if (reqbyte >= 0 && Line 6257  if (reqbyte >= 0 &&
6257    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6258    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6259      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6260    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6261    }    }
6262    
6263  /* 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 5955  case when building a production library. Line 6268  case when building a production library.
6268  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6269    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6270    
6271  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 " : "");  
   }  
6272    
6273  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6274    {    {
6275    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6276    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5978  if ((re->options & PCRE_FIRSTSET) != 0) Line 6279  if ((re->options & PCRE_FIRSTSET) != 0)
6279      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6280    }    }
6281    
6282  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6283    {    {
6284    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6285    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5995  was compiled can be seen. */ Line 6296  was compiled can be seen. */
6296  if (code - codestart > length)  if (code - codestart > length)
6297    {    {
6298    (pcre_free)(re);    (pcre_free)(re);
6299    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6300    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6301    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6302    return NULL;    return NULL;

Legend:
Removed from v.218  
changed lines
  Added in v.333

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12