/[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 171 by ph10, Mon Jun 4 14:28:58 2007 UTC revision 336 by ph10, Sat Apr 12 15:59:03 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 58  used by pcretest. DEBUG is not defined w Line 61  used by pcretest. DEBUG is not defined w
61  #endif  #endif
62    
63    
64    /* Macro for setting individual bits in class bitmaps. */
65    
66    #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         *
78  *************************************************/  *************************************************/
# Line 87  static const short int escapes[] = { Line 102  static const short int escapes[] = {
102       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */
103       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */
104     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */
105       0,      0,      0, -ESC_K,      0,      0,      0,      0,   /* H - O */  -ESC_H,      0,      0, -ESC_K,      0,      0,      0,      0,   /* H - O */
106  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0,      0, -ESC_W,   /* P - W */  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0, -ESC_V, -ESC_W,   /* P - W */
107  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */
108     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */
109       0,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */  -ESC_h,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */
110  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0,      0, -ESC_w,   /* p - w */  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0, -ESC_v, -ESC_w,   /* p - w */
111       0,      0, -ESC_z                                            /* x - z */       0,      0, -ESC_z                                            /* x - z */
112  };  };
113    
# Line 106  static const short int escapes[] = { Line 121  static const short int escapes[] = {
121  /*  70 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  70 */     0,     0,      0,       0,      0,     0,      0,      0,
122  /*  78 */     0,   '`',    ':',     '#',    '@',  '\'',    '=',    '"',  /*  78 */     0,   '`',    ':',     '#',    '@',  '\'',    '=',    '"',
123  /*  80 */     0,     7, -ESC_b,       0, -ESC_d, ESC_e,  ESC_f,      0,  /*  80 */     0,     7, -ESC_b,       0, -ESC_d, ESC_e,  ESC_f,      0,
124  /*  88 */     0,     0,      0,     '{',      0,     0,      0,      0,  /*  88 */-ESC_h,     0,      0,     '{',      0,     0,      0,      0,
125  /*  90 */     0,     0, -ESC_k,     'l',      0, ESC_n,      0, -ESC_p,  /*  90 */     0,     0, -ESC_k,     'l',      0, ESC_n,      0, -ESC_p,
126  /*  98 */     0, ESC_r,      0,     '}',      0,     0,      0,      0,  /*  98 */     0, ESC_r,      0,     '}',      0,     0,      0,      0,
127  /*  A0 */     0,   '~', -ESC_s, ESC_tee,      0,     0, -ESC_w,      0,  /*  A0 */     0,   '~', -ESC_s, ESC_tee,      0,-ESC_v, -ESC_w,      0,
128  /*  A8 */     0,-ESC_z,      0,       0,      0,   '[',      0,      0,  /*  A8 */     0,-ESC_z,      0,       0,      0,   '[',      0,      0,
129  /*  B0 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  B0 */     0,     0,      0,       0,      0,     0,      0,      0,
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 */     0,     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,     0, -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,
137  /*  F0 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  F0 */     0,     0,      0,       0,      0,     0,      0,      0,
138  /*  F8 */     0,     0,      0,       0,      0,     0,      0,      0  /*  F8 */     0,     0,      0,       0,      0,     0,      0,      0
# Line 125  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 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 const 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 171  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, 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\0"
301      /* 60 */
302      "(*VERB) not recognized\0"
303      "number is too big\0"
304      "subpattern name expected\0"
305      "digit expected after (?+\0"
306      "] is an invalid data character in JavaScript compatibility mode";
307    
308    
309  /* 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 374  static const unsigned char ebcdic_charta Line 433  static const unsigned char ebcdic_charta
433  /* Definition to allow mutual recursion */  /* Definition to allow mutual recursion */
434    
435  static BOOL  static BOOL
436    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, int, int *,    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,
437      int *, branch_chain *, compile_data *, int *);      int *, int *, branch_chain *, compile_data *, int *);
438    
439    
440    
441  /*************************************************  /*************************************************
442    *            Find an error text                  *
443    *************************************************/
444    
445    /* The error texts are now all in one long string, to save on relocations. As
446    some of the text is of unknown length, we can't use a table of offsets.
447    Instead, just count through the strings. This is not a performance issue
448    because it happens only when there has been a compilation error.
449    
450    Argument:   the error number
451    Returns:    pointer to the error string
452    */
453    
454    static const char *
455    find_error_text(int n)
456    {
457    const char *s = error_texts;
458    for (; n > 0; n--) while (*s++ != 0);
459    return s;
460    }
461    
462    
463    /*************************************************
464  *            Handle escapes                      *  *            Handle escapes                      *
465  *************************************************/  *************************************************/
466    
# Line 400  Arguments: Line 481  Arguments:
481    
482  Returns:         zero or positive => a data character  Returns:         zero or positive => a data character
483                   negative => a special escape sequence                   negative => a special escape sequence
484                   on error, errorptr is set                   on error, errorcodeptr is set
485  */  */
486    
487  static int  static int
# Line 418  ptr--;                            /* Set Line 499  ptr--;                            /* Set
499    
500  if (c == 0) *errorcodeptr = ERR1;  if (c == 0) *errorcodeptr = ERR1;
501    
502  /* 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
503  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.
504  Otherwise further processing may be required. */  Otherwise further processing may be required. */
505    
506  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
507  else if (c < '0' || c > 'z') {}                           /* Not alphameric */  else if (c < '0' || c > 'z') {}                           /* Not alphanumeric */
508  else if ((i = escapes[c - '0']) != 0) c = i;  else if ((i = escapes[c - '0']) != 0) c = i;
509    
510  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
511  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphameric */  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */
512  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  c = i;
513  #endif  #endif
514    
# Line 451  else Line 532  else
532      *errorcodeptr = ERR37;      *errorcodeptr = ERR37;
533      break;      break;
534    
535      /* \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:
536      is an absolute backreference. If negative, it is a relative backreference.  
537      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
538      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
539      unified syntax for back references. As this is synonymous with \k{name}, we      5.10 feature.
540      fudge it up by pretending it really was \k. */  
541        (2) Perl 5.10 also supports \g{name} as a reference to a named group. This
542        is part of Perl's movement towards a unified syntax for back references. As
543        this is synonymous with \k{name}, we fudge it up by pretending it really
544        was \k.
545    
546        (3) For Oniguruma compatibility we also support \g followed by a name or a
547        number either in angle brackets or in single quotes. However, these are
548        (possibly recursive) subroutine calls, _not_ backreferences. Just return
549        the -ESC_g code (cf \k). */
550    
551      case 'g':      case 'g':
552        if (ptr[1] == '<' || ptr[1] == '\'')
553          {
554          c = -ESC_g;
555          break;
556          }
557    
558        /* Handle the Perl-compatible cases */
559    
560      if (ptr[1] == '{')      if (ptr[1] == '{')
561        {        {
562        const uschar *p;        const uschar *p;
563        for (p = ptr+2; *p != 0 && *p != '}'; p++)        for (p = ptr+2; *p != 0 && *p != '}'; p++)
564          if (*p != '-' && (digitab[*p] & ctype_digit) == 0) break;          if (*p != '-' && (digitab[*p] & ctype_digit) == 0) break;
565        if (*p != 0 && *p != '}')        if (*p != 0 && *p != '}')
566          {          {
567          c = -ESC_k;          c = -ESC_k;
568          break;          break;
569          }          }
570        braced = TRUE;        braced = TRUE;
571        ptr++;        ptr++;
572        }        }
# Line 485  else Line 583  else
583      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
584        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - '0';
585    
586      if (c == 0 || (braced && *(++ptr) != '}'))      if (c < 0)   /* Integer overflow */
587          {
588          *errorcodeptr = ERR61;
589          break;
590          }
591    
592        if (braced && *(++ptr) != '}')
593        {        {
594        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
595        return 0;        break;
596        }        }
597    
598        if (c == 0)
599          {
600          *errorcodeptr = ERR58;
601          break;
602          }
603    
604      if (negated)      if (negated)
605        {        {
606        if (c > bracount)        if (c > bracount)
607          {          {
608          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
609          return 0;          break;
610          }          }
611        c = bracount - (c - 1);        c = bracount - (c - 1);
612        }        }
# Line 525  else Line 635  else
635        c -= '0';        c -= '0';
636        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
637          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - '0';
638          if (c < 0)    /* Integer overflow */
639            {
640            *errorcodeptr = ERR61;
641            break;
642            }
643        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
644          {          {
645          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 620  else Line 735  else
735      if (c == 0)      if (c == 0)
736        {        {
737        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
738        return 0;        break;
739        }        }
740    
741  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
# Line 633  else Line 748  else
748      break;      break;
749    
750      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any
751      other alphameric following \ is an error if PCRE_EXTRA was set; otherwise,      other alphanumeric following \ is an error if PCRE_EXTRA was set;
752      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
753      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
754      in future, so I haven't "optimized" it. */      be again in future, so I haven't "optimized" it. */
755    
756      default:      default:
757      if ((options & PCRE_EXTRA) != 0) switch(c)      if ((options & PCRE_EXTRA) != 0) switch(c)
# Line 696  if (c == '{') Line 811  if (c == '{')
811      *negptr = TRUE;      *negptr = TRUE;
812      ptr++;      ptr++;
813      }      }
814    for (i = 0; i < sizeof(name) - 1; i++)    for (i = 0; i < (int)sizeof(name) - 1; i++)
815      {      {
816      c = *(++ptr);      c = *(++ptr);
817      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 725  top = _pcre_utt_size; Line 840  top = _pcre_utt_size;
840  while (bot < top)  while (bot < top)
841    {    {
842    i = (bot + top) >> 1;    i = (bot + top) >> 1;
843    c = strcmp(name, _pcre_utt[i].name);    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);
844    if (c == 0)    if (c == 0)
845      {      {
846      *dptr = _pcre_utt[i].value;      *dptr = _pcre_utt[i].value;
# Line 899  for (; *ptr != 0; ptr++) Line 1014  for (; *ptr != 0; ptr++)
1014      {      {
1015      while (*(++ptr) != ']')      while (*(++ptr) != ']')
1016        {        {
1017          if (*ptr == 0) return -1;
1018        if (*ptr == '\\')        if (*ptr == '\\')
1019          {          {
1020          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 926  for (; *ptr != 0; ptr++) Line 1042  for (; *ptr != 0; ptr++)
1042    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
1043    
1044    if (*ptr != '(') continue;    if (*ptr != '(') continue;
1045    if (ptr[1] != '?')    if (ptr[1] != '?' && ptr[1] != '*')
1046      {      {
1047      count++;      count++;
1048      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 1054  for (;;) Line 1170  for (;;)
1170    {    {
1171    int d;    int d;
1172    register int op = *cc;    register int op = *cc;
   
1173    switch (op)    switch (op)
1174      {      {
1175      case OP_CBRA:      case OP_CBRA:
# Line 1143  for (;;) Line 1258  for (;;)
1258    
1259      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1260      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1261        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1262      cc += 4;      cc += 4;
1263      break;      break;
1264    
# Line 1251  for (;;) Line 1367  for (;;)
1367      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1368      }      }
1369    
1370    /* 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
1371    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
1372    arrange to skip the extra bytes. */    two bytes of parameters. */
1373    
1374    else    else
1375      {      {
1376        switch(c)
1377          {
1378          case OP_TYPESTAR:
1379          case OP_TYPEMINSTAR:
1380          case OP_TYPEPLUS:
1381          case OP_TYPEMINPLUS:
1382          case OP_TYPEQUERY:
1383          case OP_TYPEMINQUERY:
1384          case OP_TYPEPOSSTAR:
1385          case OP_TYPEPOSPLUS:
1386          case OP_TYPEPOSQUERY:
1387          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1388          break;
1389    
1390          case OP_TYPEUPTO:
1391          case OP_TYPEMINUPTO:
1392          case OP_TYPEEXACT:
1393          case OP_TYPEPOSUPTO:
1394          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1395          break;
1396          }
1397    
1398        /* Add in the fixed length from the table */
1399    
1400      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1401    
1402      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1403      a multi-byte character. The length in the table is a minimum, so we have to
1404      arrange to skip the extra bytes. */
1405    
1406  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1407      if (utf8) switch(c)      if (utf8) switch(c)
1408        {        {
# Line 1315  for (;;) Line 1460  for (;;)
1460    
1461    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1462    
1463    /* 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
1464    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
1465    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1466    
1467    else    else
1468      {      {
1469        switch(c)
1470          {
1471          case OP_TYPESTAR:
1472          case OP_TYPEMINSTAR:
1473          case OP_TYPEPLUS:
1474          case OP_TYPEMINPLUS:
1475          case OP_TYPEQUERY:
1476          case OP_TYPEMINQUERY:
1477          case OP_TYPEPOSSTAR:
1478          case OP_TYPEPOSPLUS:
1479          case OP_TYPEPOSQUERY:
1480          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1481          break;
1482    
1483          case OP_TYPEPOSUPTO:
1484          case OP_TYPEUPTO:
1485          case OP_TYPEMINUPTO:
1486          case OP_TYPEEXACT:
1487          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1488          break;
1489          }
1490    
1491        /* Add in the fixed length from the table */
1492    
1493      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1494    
1495        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1496        by a multi-byte character. The length in the table is a minimum, so we have
1497        to arrange to skip the extra bytes. */
1498    
1499  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1500      if (utf8) switch(c)      if (utf8) switch(c)
1501        {        {
# Line 1359  for (;;) Line 1532  for (;;)
1532  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()
1533  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
1534  group that can match nothing. Note that first_significant_code() skips over  group that can match nothing. Note that first_significant_code() skips over
1535  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
1536  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
1537    bracket whose current branch will already have been scanned.
1538    
1539  Arguments:  Arguments:
1540    code        points to start of search    code        points to start of search
# Line 1381  for (code = first_significant_code(code Line 1555  for (code = first_significant_code(code
1555    const uschar *ccode;    const uschar *ccode;
1556    
1557    c = *code;    c = *code;
1558    
1559      /* Skip over forward assertions; the other assertions are skipped by
1560      first_significant_code() with a TRUE final argument. */
1561    
1562      if (c == OP_ASSERT)
1563        {
1564        do code += GET(code, 1); while (*code == OP_ALT);
1565        c = *code;
1566        continue;
1567        }
1568    
1569    /* Groups with zero repeats can of course be empty; skip them. */    /* Groups with zero repeats can of course be empty; skip them. */
1570    
1571    if (c == OP_BRAZERO || c == OP_BRAMINZERO)    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO)
1572      {      {
1573        code += _pcre_OP_lengths[c];
1574      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1575      c = *code;      c = *code;
1576      continue;      continue;
1577      }      }
1578    
1579    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1580    
1581    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1582      {      {
1583      BOOL empty_branch;      BOOL empty_branch;
1584      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 1409  for (code = first_significant_code(code Line 1594  for (code = first_significant_code(code
1594        }        }
1595      while (*code == OP_ALT);      while (*code == OP_ALT);
1596      if (!empty_branch) return FALSE;   /* All branches are non-empty */      if (!empty_branch) return FALSE;   /* All branches are non-empty */
1597      c = *code;      c = *code;
1598      continue;      continue;
1599      }      }
1600    
# Line 1417  for (code = first_significant_code(code Line 1602  for (code = first_significant_code(code
1602    
1603    switch (c)    switch (c)
1604      {      {
1605      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1606        cannot be represented just by a bit map. This includes negated single
1607        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1608        actual length is stored in the compiled code, so we must update "code"
1609        here. */
1610    
1611  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1612      case OP_XCLASS:      case OP_XCLASS:
1613      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1614      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1615  #endif  #endif
1616    
# Line 1483  for (code = first_significant_code(code Line 1672  for (code = first_significant_code(code
1672      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1673      return FALSE;      return FALSE;
1674    
1675        /* These are going to continue, as they may be empty, but we have to
1676        fudge the length for the \p and \P cases. */
1677    
1678        case OP_TYPESTAR:
1679        case OP_TYPEMINSTAR:
1680        case OP_TYPEPOSSTAR:
1681        case OP_TYPEQUERY:
1682        case OP_TYPEMINQUERY:
1683        case OP_TYPEPOSQUERY:
1684        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1685        break;
1686    
1687        /* Same for these */
1688    
1689        case OP_TYPEUPTO:
1690        case OP_TYPEMINUPTO:
1691        case OP_TYPEPOSUPTO:
1692        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1693        break;
1694    
1695      /* End of branch */      /* End of branch */
1696    
1697      case OP_KET:      case OP_KET:
# Line 1552  return TRUE; Line 1761  return TRUE;
1761  *************************************************/  *************************************************/
1762    
1763  /* This function is called when the sequence "[:" or "[." or "[=" is  /* This function is called when the sequence "[:" or "[." or "[=" is
1764  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
1765  optional ^ and then a sequence of letters, terminated by a matching ":]" or  sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
1766  ".]" or "=]".  reach an unescaped ']' without the special preceding character, return FALSE.
1767    
1768    Originally, this function only recognized a sequence of letters between the
1769    terminators, but it seems that Perl recognizes any sequence of characters,
1770    though of course unknown POSIX names are subsequently rejected. Perl gives an
1771    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
1772    didn't consider this to be a POSIX class. Likewise for [:1234:].
1773    
1774    The problem in trying to be exactly like Perl is in the handling of escapes. We
1775    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
1776    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
1777    below handles the special case of \], but does not try to do any other escape
1778    processing. This makes it different from Perl for cases such as [:l\ower:]
1779    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
1780    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
1781    I think.
1782    
1783  Argument:  Arguments:
1784    ptr      pointer to the initial [    ptr      pointer to the initial [
1785    endptr   where to return the end pointer    endptr   where to return the end pointer
   cd       pointer to compile data  
1786    
1787  Returns:   TRUE or FALSE  Returns:   TRUE or FALSE
1788  */  */
1789    
1790  static BOOL  static BOOL
1791  check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd)  check_posix_syntax(const uschar *ptr, const uschar **endptr)
1792  {  {
1793  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
1794  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
1795  if (*(++ptr) == '^') ptr++;  for (++ptr; *ptr != 0; ptr++)
 while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;  
 if (*ptr == terminator && ptr[1] == ']')  
1796    {    {
1797    *endptr = ptr;    if (*ptr == '\\' && ptr[1] == ']') ptr++; else
1798    return TRUE;      {
1799        if (*ptr == ']') return FALSE;
1800        if (*ptr == terminator && ptr[1] == ']')
1801          {
1802          *endptr = ptr;
1803          return TRUE;
1804          }
1805        }
1806    }    }
1807  return FALSE;  return FALSE;
1808  }  }
# Line 1599  Returns:     a value representing the na Line 1827  Returns:     a value representing the na
1827  static int  static int
1828  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1829  {  {
1830    const char *pn = posix_names;
1831  register int yield = 0;  register int yield = 0;
1832  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1833    {    {
1834    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1835      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1836      pn += posix_name_lengths[yield] + 1;
1837    yield++;    yield++;
1838    }    }
1839  return -1;  return -1;
# Line 1618  return -1; Line 1848  return -1;
1848  that is referenced. This means that groups can be replicated for fixed  that is referenced. This means that groups can be replicated for fixed
1849  repetition simply by copying (because the recursion is allowed to refer to  repetition simply by copying (because the recursion is allowed to refer to
1850  earlier groups that are outside the current group). However, when a group is  earlier groups that are outside the current group). However, when a group is
1851  optional (i.e. the minimum quantifier is zero), OP_BRAZERO is inserted before  optional (i.e. the minimum quantifier is zero), OP_BRAZERO or OP_SKIPZERO is
1852  it, after it has been compiled. This means that any OP_RECURSE items within it  inserted before it, after it has been compiled. This means that any OP_RECURSE
1853  that refer to the group itself or any contained groups have to have their  items within it that refer to the group itself or any contained groups have to
1854  offsets adjusted. That one of the jobs of this function. Before it is called,  have their offsets adjusted. That one of the jobs of this function. Before it
1855  the partially compiled regex must be temporarily terminated with OP_END.  is called, the partially compiled regex must be temporarily terminated with
1856    OP_END.
1857    
1858  This function has been extended with the possibility of forward references for  This function has been extended with the possibility of forward references for
1859  recursions and subroutine calls. It must also check the list of such references  recursions and subroutine calls. It must also check the list of such references
# Line 1645  adjust_recurse(uschar *group, int adjust Line 1876  adjust_recurse(uschar *group, int adjust
1876    uschar *save_hwm)    uschar *save_hwm)
1877  {  {
1878  uschar *ptr = group;  uschar *ptr = group;
1879    
1880  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1881    {    {
1882    int offset;    int offset;
# Line 1652  while ((ptr = (uschar *)find_recurse(ptr Line 1884  while ((ptr = (uschar *)find_recurse(ptr
1884    
1885    /* See if this recursion is on the forward reference list. If so, adjust the    /* See if this recursion is on the forward reference list. If so, adjust the
1886    reference. */    reference. */
1887    
1888    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
1889      {      {
1890      offset = GET(hc, 0);      offset = GET(hc, 0);
# Line 1906  if (next >= 0) switch(op_code) Line 2138  if (next >= 0) switch(op_code)
2138    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, "item" must be a single-byte character. */
2139    
2140    case OP_NOT:    case OP_NOT:
   if (next < 0) return FALSE;  /* Not a character */  
2141    if (item == next) return TRUE;    if (item == next) return TRUE;
2142    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2143  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1943  if (next >= 0) switch(op_code) Line 2174  if (next >= 0) switch(op_code)
2174    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2175    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2176    
2177      case OP_HSPACE:
2178      case OP_NOT_HSPACE:
2179      switch(next)
2180        {
2181        case 0x09:
2182        case 0x20:
2183        case 0xa0:
2184        case 0x1680:
2185        case 0x180e:
2186        case 0x2000:
2187        case 0x2001:
2188        case 0x2002:
2189        case 0x2003:
2190        case 0x2004:
2191        case 0x2005:
2192        case 0x2006:
2193        case 0x2007:
2194        case 0x2008:
2195        case 0x2009:
2196        case 0x200A:
2197        case 0x202f:
2198        case 0x205f:
2199        case 0x3000:
2200        return op_code != OP_HSPACE;
2201        default:
2202        return op_code == OP_HSPACE;
2203        }
2204    
2205      case OP_VSPACE:
2206      case OP_NOT_VSPACE:
2207      switch(next)
2208        {
2209        case 0x0a:
2210        case 0x0b:
2211        case 0x0c:
2212        case 0x0d:
2213        case 0x85:
2214        case 0x2028:
2215        case 0x2029:
2216        return op_code != OP_VSPACE;
2217        default:
2218        return op_code == OP_VSPACE;
2219        }
2220    
2221    default:    default:
2222    return FALSE;    return FALSE;
2223    }    }
# Line 1977  switch(op_code) Line 2252  switch(op_code)
2252      case ESC_W:      case ESC_W:
2253      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2254    
2255        case ESC_h:
2256        case ESC_H:
2257        switch(item)
2258          {
2259          case 0x09:
2260          case 0x20:
2261          case 0xa0:
2262          case 0x1680:
2263          case 0x180e:
2264          case 0x2000:
2265          case 0x2001:
2266          case 0x2002:
2267          case 0x2003:
2268          case 0x2004:
2269          case 0x2005:
2270          case 0x2006:
2271          case 0x2007:
2272          case 0x2008:
2273          case 0x2009:
2274          case 0x200A:
2275          case 0x202f:
2276          case 0x205f:
2277          case 0x3000:
2278          return -next != ESC_h;
2279          default:
2280          return -next == ESC_h;
2281          }
2282    
2283        case ESC_v:
2284        case ESC_V:
2285        switch(item)
2286          {
2287          case 0x0a:
2288          case 0x0b:
2289          case 0x0c:
2290          case 0x0d:
2291          case 0x85:
2292          case 0x2028:
2293          case 0x2029:
2294          return -next != ESC_v;
2295          default:
2296          return -next == ESC_v;
2297          }
2298    
2299      default:      default:
2300      return FALSE;      return FALSE;
2301      }      }
2302    
2303    case OP_DIGIT:    case OP_DIGIT:
2304    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2305             next == -ESC_h || next == -ESC_v;
2306    
2307    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2308    return next == -ESC_d;    return next == -ESC_d;
# Line 1991  switch(op_code) Line 2311  switch(op_code)
2311    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2312    
2313    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2314    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2315    
2316      case OP_HSPACE:
2317      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2318    
2319      case OP_NOT_HSPACE:
2320      return next == -ESC_h;
2321    
2322      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2323      case OP_VSPACE:
2324      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2325    
2326      case OP_NOT_VSPACE:
2327      return next == -ESC_v;
2328    
2329    case OP_WORDCHAR:    case OP_WORDCHAR:
2330    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2331    
2332    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2333    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2067  uschar classbits[32]; Line 2400  uschar classbits[32];
2400  BOOL class_utf8;  BOOL class_utf8;
2401  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8 = (options & PCRE_UTF8) != 0;
2402  uschar *class_utf8data;  uschar *class_utf8data;
2403    uschar *class_utf8data_base;
2404  uschar utf8_char[6];  uschar utf8_char[6];
2405  #else  #else
2406  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
# Line 2106  req_caseopt = ((options & PCRE_CASELESS) Line 2440  req_caseopt = ((options & PCRE_CASELESS)
2440  for (;; ptr++)  for (;; ptr++)
2441    {    {
2442    BOOL negate_class;    BOOL negate_class;
2443      BOOL should_flip_negation;
2444    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2445    BOOL is_quantifier;    BOOL is_quantifier;
2446    BOOL is_recurse;    BOOL is_recurse;
2447      BOOL reset_bracount;
2448    int class_charcount;    int class_charcount;
2449    int class_lastchar;    int class_lastchar;
2450    int newoptions;    int newoptions;
2451    int recno;    int recno;
2452    int refsign;    int refsign;
2453    int skipbytes;    int skipbytes;
2454    int subreqbyte;    int subreqbyte;
2455    int subfirstbyte;    int subfirstbyte;
# Line 2124  for (;; ptr++) Line 2460  for (;; ptr++)
2460    /* Get next byte in the pattern */    /* Get next byte in the pattern */
2461    
2462    c = *ptr;    c = *ptr;
2463    
2464    /* If we are in the pre-compile phase, accumulate the length used for the    /* If we are in the pre-compile phase, accumulate the length used for the
2465    previous cycle of this loop. */    previous cycle of this loop. */
2466    
# Line 2146  for (;; ptr++) Line 2482  for (;; ptr++)
2482      */      */
2483    
2484      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2485    
2486        /* Paranoid check for integer overflow */
2487    
2488        if (OFLOW_MAX - *lengthptr < code - last_code)
2489          {
2490          *errorcodeptr = ERR20;
2491          goto FAILED;
2492          }
2493    
2494      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2495      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));
2496    
# Line 2258  for (;; ptr++) Line 2603  for (;; ptr++)
2603      *ptrptr = ptr;      *ptrptr = ptr;
2604      if (lengthptr != NULL)      if (lengthptr != NULL)
2605        {        {
2606          if (OFLOW_MAX - *lengthptr < code - last_code)
2607            {
2608            *errorcodeptr = ERR20;
2609            goto FAILED;
2610            }
2611        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2612        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2613        }        }
# Line 2305  for (;; ptr++) Line 2655  for (;; ptr++)
2655      opcode is compiled. It may optionally have a bit map for characters < 256,      opcode is compiled. It may optionally have a bit map for characters < 256,
2656      but those above are are explicitly listed afterwards. A flag byte tells      but those above are are explicitly listed afterwards. A flag byte tells
2657      whether the bitmap is present, and whether this is a negated class or not.      whether the bitmap is present, and whether this is a negated class or not.
2658      */  
2659        In JavaScript compatibility mode, an isolated ']' causes an error. In
2660        default (Perl) mode, it is treated as a data character. */
2661    
2662        case ']':
2663        if ((cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
2664          {
2665          *errorcodeptr = ERR64;
2666          goto FAILED;
2667          }
2668        goto NORMAL_CHAR;
2669    
2670      case '[':      case '[':
2671      previous = code;      previous = code;
# Line 2314  for (;; ptr++) Line 2674  for (;; ptr++)
2674      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. */
2675    
2676      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2677          check_posix_syntax(ptr, &tempptr, cd))          check_posix_syntax(ptr, &tempptr))
2678        {        {
2679        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;
2680        goto FAILED;        goto FAILED;
2681        }        }
2682    
2683      /* 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,
2684        if the first few characters (either before or after ^) are \Q\E or \E we
2685        skip them too. This makes for compatibility with Perl. */
2686    
2687      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2688        for (;;)
2689        {        {
       negate_class = TRUE;  
2690        c = *(++ptr);        c = *(++ptr);
2691          if (c == '\\')
2692            {
2693            if (ptr[1] == 'E') ptr++;
2694              else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2695                else break;
2696            }
2697          else if (!negate_class && c == '^')
2698            negate_class = TRUE;
2699          else break;
2700        }        }
2701      else  
2702        {      /* If a class contains a negative special such as \S, we need to flip the
2703        negate_class = FALSE;      negation flag at the end, so that support for characters > 255 works
2704        }      correctly (they are all included in the class). */
2705    
2706        should_flip_negation = FALSE;
2707    
2708      /* 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
2709      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
# Line 2349  for (;; ptr++) Line 2722  for (;; ptr++)
2722  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2723      class_utf8 = FALSE;                       /* No chars >= 256 */      class_utf8 = FALSE;                       /* No chars >= 256 */
2724      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */
2725        class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */
2726  #endif  #endif
2727    
2728      /* 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 2364  for (;; ptr++) Line 2738  for (;; ptr++)
2738          {                           /* Braces are required because the */          {                           /* Braces are required because the */
2739          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
2740          }          }
2741    
2742          /* In the pre-compile phase, accumulate the length of any UTF-8 extra
2743          data and reset the pointer. This is so that very large classes that
2744          contain a zillion UTF-8 characters no longer overwrite the work space
2745          (which is on the stack). */
2746    
2747          if (lengthptr != NULL)
2748            {
2749            *lengthptr += class_utf8data - class_utf8data_base;
2750            class_utf8data = class_utf8data_base;
2751            }
2752    
2753  #endif  #endif
2754    
2755        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 2387  for (;; ptr++) Line 2773  for (;; ptr++)
2773    
2774        if (c == '[' &&        if (c == '[' &&
2775            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2776            check_posix_syntax(ptr, &tempptr, cd))            check_posix_syntax(ptr, &tempptr))
2777          {          {
2778          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
2779          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
# Line 2404  for (;; ptr++) Line 2790  for (;; ptr++)
2790          if (*ptr == '^')          if (*ptr == '^')
2791            {            {
2792            local_negate = TRUE;            local_negate = TRUE;
2793              should_flip_negation = TRUE;  /* Note negative special */
2794            ptr++;            ptr++;
2795            }            }
2796    
# Line 2470  for (;; ptr++) Line 2857  for (;; ptr++)
2857        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
2858        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2859        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2860        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
2861        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2862    
2863        if (c == '\\')        if (c == '\\')
# Line 2478  for (;; ptr++) Line 2865  for (;; ptr++)
2865          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2866          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2867    
2868          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = '\b';       /* \b is backspace in a class */
2869          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 */
2870          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 */
2871          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
# Line 2490  for (;; ptr++) Line 2877  for (;; ptr++)
2877            else inescq = TRUE;            else inescq = TRUE;
2878            continue;            continue;
2879            }            }
2880            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2881    
2882          if (c < 0)          if (c < 0)
2883            {            {
# Line 2505  for (;; ptr++) Line 2893  for (;; ptr++)
2893              continue;              continue;
2894    
2895              case ESC_D:              case ESC_D:
2896                should_flip_negation = TRUE;
2897              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2898              continue;              continue;
2899    
# Line 2513  for (;; ptr++) Line 2902  for (;; ptr++)
2902              continue;              continue;
2903    
2904              case ESC_W:              case ESC_W:
2905                should_flip_negation = TRUE;
2906              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2907              continue;              continue;
2908    
# Line 2522  for (;; ptr++) Line 2912  for (;; ptr++)
2912              continue;              continue;
2913    
2914              case ESC_S:              case ESC_S:
2915                should_flip_negation = TRUE;
2916              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2917              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2918              continue;              continue;
2919    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
2920              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
2921              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
2922              }              }
# Line 2538  for (;; ptr++) Line 2926  for (;; ptr++)
2926            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
2927                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
2928    
2929              /* We need to deal with \H, \h, \V, and \v in both phases because
2930              they use extra memory. */
2931    
2932              if (-c == ESC_h)
2933                {
2934                SETBIT(classbits, 0x09); /* VT */
2935                SETBIT(classbits, 0x20); /* SPACE */
2936                SETBIT(classbits, 0xa0); /* NSBP */
2937    #ifdef SUPPORT_UTF8
2938                if (utf8)
2939                  {
2940                  class_utf8 = TRUE;
2941                  *class_utf8data++ = XCL_SINGLE;
2942                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
2943                  *class_utf8data++ = XCL_SINGLE;
2944                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
2945                  *class_utf8data++ = XCL_RANGE;
2946                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
2947                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
2948                  *class_utf8data++ = XCL_SINGLE;
2949                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
2950                  *class_utf8data++ = XCL_SINGLE;
2951                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
2952                  *class_utf8data++ = XCL_SINGLE;
2953                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
2954                  }
2955    #endif
2956                continue;
2957                }
2958    
2959              if (-c == ESC_H)
2960                {
2961                for (c = 0; c < 32; c++)
2962                  {
2963                  int x = 0xff;
2964                  switch (c)
2965                    {
2966                    case 0x09/8: x ^= 1 << (0x09%8); break;
2967                    case 0x20/8: x ^= 1 << (0x20%8); break;
2968                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
2969                    default: break;
2970                    }
2971                  classbits[c] |= x;
2972                  }
2973    
2974    #ifdef SUPPORT_UTF8
2975                if (utf8)
2976                  {
2977                  class_utf8 = TRUE;
2978                  *class_utf8data++ = XCL_RANGE;
2979                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2980                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
2981                  *class_utf8data++ = XCL_RANGE;
2982                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
2983                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
2984                  *class_utf8data++ = XCL_RANGE;
2985                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
2986                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
2987                  *class_utf8data++ = XCL_RANGE;
2988                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
2989                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
2990                  *class_utf8data++ = XCL_RANGE;
2991                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
2992                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
2993                  *class_utf8data++ = XCL_RANGE;
2994                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
2995                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
2996                  *class_utf8data++ = XCL_RANGE;
2997                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
2998                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2999                  }
3000    #endif
3001                continue;
3002                }
3003    
3004              if (-c == ESC_v)
3005                {
3006                SETBIT(classbits, 0x0a); /* LF */
3007                SETBIT(classbits, 0x0b); /* VT */
3008                SETBIT(classbits, 0x0c); /* FF */
3009                SETBIT(classbits, 0x0d); /* CR */
3010                SETBIT(classbits, 0x85); /* NEL */
3011    #ifdef SUPPORT_UTF8
3012                if (utf8)
3013                  {
3014                  class_utf8 = TRUE;
3015                  *class_utf8data++ = XCL_RANGE;
3016                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
3017                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
3018                  }
3019    #endif
3020                continue;
3021                }
3022    
3023              if (-c == ESC_V)
3024                {
3025                for (c = 0; c < 32; c++)
3026                  {
3027                  int x = 0xff;
3028                  switch (c)
3029                    {
3030                    case 0x0a/8: x ^= 1 << (0x0a%8);
3031                                 x ^= 1 << (0x0b%8);
3032                                 x ^= 1 << (0x0c%8);
3033                                 x ^= 1 << (0x0d%8);
3034                                 break;
3035                    case 0x85/8: x ^= 1 << (0x85%8); break;
3036                    default: break;
3037                    }
3038                  classbits[c] |= x;
3039                  }
3040    
3041    #ifdef SUPPORT_UTF8
3042                if (utf8)
3043                  {
3044                  class_utf8 = TRUE;
3045                  *class_utf8data++ = XCL_RANGE;
3046                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
3047                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
3048                  *class_utf8data++ = XCL_RANGE;
3049                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
3050                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
3051                  }
3052    #endif
3053                continue;
3054                }
3055    
3056            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
3057    
3058  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2589  for (;; ptr++) Line 3104  for (;; ptr++)
3104    
3105        oldptr = ptr;        oldptr = ptr;
3106    
3107          /* Remember \r or \n */
3108    
3109          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3110    
3111          /* Check for range */
3112    
3113        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3114          {          {
3115          int d;          int d;
# Line 2630  for (;; ptr++) Line 3151  for (;; ptr++)
3151            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3152            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3153    
3154            /* \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
3155            special means the '-' was literal */            special means the '-' was literal */
3156    
3157            if (d < 0)            if (d < 0)
# Line 2656  for (;; ptr++) Line 3177  for (;; ptr++)
3177    
3178          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3179    
3180            /* Remember \r or \n */
3181    
3182            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3183    
3184          /* 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
3185          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3186          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2678  for (;; ptr++) Line 3203  for (;; ptr++)
3203              unsigned int origd = d;              unsigned int origd = d;
3204              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3205                {                {
3206                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3207                      ocd <= (unsigned int)d)
3208                    continue;                          /* Skip embedded ranges */
3209    
3210                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3211                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3212                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3213                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3214                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3215                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3216                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3217                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3218                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3219                  d = ocd;                  d = ocd;
3220                  continue;                  continue;
# Line 2805  for (;; ptr++) Line 3334  for (;; ptr++)
3334        goto FAILED;        goto FAILED;
3335        }        }
3336    
3337    
3338    /* This code has been disabled because it would mean that \s counts as
3339    an explicit \r or \n reference, and that's not really what is wanted. Now
3340    we set the flag only if there is a literal "\r" or "\n" in the class. */
3341    
3342    #if 0
3343        /* Remember whether \r or \n are in this class */
3344    
3345        if (negate_class)
3346          {
3347          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3348          }
3349        else
3350          {
3351          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3352          }
3353    #endif
3354    
3355    
3356      /* 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
3357      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
3358      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
3359      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3360      single-bytes only. This is an historical hangover. Maybe one day we can  
3361      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3362        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3363        operate on single-bytes only. This is an historical hangover. Maybe one day
3364        we can tidy these opcodes to handle multi-byte characters.
3365    
3366      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
3367      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 2820  for (;; ptr++) Line 3371  for (;; ptr++)
3371      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3372    
3373  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3374      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3375            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3376  #else  #else
3377      if (class_charcount == 1)      if (class_charcount == 1)
3378  #endif  #endif
# Line 2866  for (;; ptr++) Line 3415  for (;; ptr++)
3415      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3416    
3417      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3418      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3419      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
3420        the class, so any that were explicitly given as well can be ignored. If
3421        (when there are explicit characters > 255 that must be listed) there are no
3422        characters < 256, we can omit the bitmap in the actual compiled code. */
3423    
3424  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3425      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3426        {        {
3427        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3428        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 2896  for (;; ptr++) Line 3448  for (;; ptr++)
3448        }        }
3449  #endif  #endif
3450    
3451      /* 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
3452      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
3453      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
3454      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3455    
3456        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3457      if (negate_class)      if (negate_class)
3458        {        {
       *code++ = OP_NCLASS;  
3459        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3460          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3461        }        }
3462      else      else
3463        {        {
       *code++ = OP_CLASS;  
3464        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3465        }        }
3466      code += 32;      code += 32;
# Line 3095  for (;; ptr++) Line 3646  for (;; ptr++)
3646        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3647        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3648    
3649        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3650    
3651        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3652    
# Line 3245  for (;; ptr++) Line 3796  for (;; ptr++)
3796        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3797        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3798    
3799        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3800    
3801        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3802          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3281  for (;; ptr++) Line 3832  for (;; ptr++)
3832          goto FAILED;          goto FAILED;
3833          }          }
3834    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3835        /* 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
3836        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
3837        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 3311  for (;; ptr++) Line 3854  for (;; ptr++)
3854    
3855        if (repeat_min == 0)        if (repeat_min == 0)
3856          {          {
3857          /* If the maximum is also zero, we just omit the group from the output          /* If the maximum is also zero, we used to just omit the group from the
3858          altogether. */          output altogether, like this:
   
         if (repeat_max == 0)  
           {  
           code = previous;  
           goto END_REPEAT;  
           }  
3859    
3860          /* If the maximum is 1 or unlimited, we just have to stick in the          ** if (repeat_max == 0)
3861          BRAZERO and do no more at this point. However, we do need to adjust          **   {
3862          any OP_RECURSE calls inside the group that refer to the group itself or          **   code = previous;
3863          any internal or forward referenced group, because the offset is from          **   goto END_REPEAT;
3864          the start of the whole regex. Temporarily terminate the pattern while          **   }
3865          doing this. */  
3866            However, that fails when a group is referenced as a subroutine from
3867            elsewhere in the pattern, so now we stick in OP_SKIPZERO in front of it
3868            so that it is skipped on execution. As we don't have a list of which
3869            groups are referenced, we cannot do this selectively.
3870    
3871            If the maximum is 1 or unlimited, we just have to stick in the BRAZERO
3872            and do no more at this point. However, we do need to adjust any
3873            OP_RECURSE calls inside the group that refer to the group itself or any
3874            internal or forward referenced group, because the offset is from the
3875            start of the whole regex. Temporarily terminate the pattern while doing
3876            this. */
3877    
3878          if (repeat_max <= 1)          if (repeat_max <= 1)    /* Covers 0, 1, and unlimited */
3879            {            {
3880            *code = OP_END;            *code = OP_END;
3881            adjust_recurse(previous, 1, utf8, cd, save_hwm);            adjust_recurse(previous, 1, utf8, cd, save_hwm);
3882            memmove(previous+1, previous, len);            memmove(previous+1, previous, len);
3883            code++;            code++;
3884              if (repeat_max == 0)
3885                {
3886                *previous++ = OP_SKIPZERO;
3887                goto END_REPEAT;
3888                }
3889            *previous++ = OP_BRAZERO + repeat_type;            *previous++ = OP_BRAZERO + repeat_type;
3890            }            }
3891    
# Line 3377  for (;; ptr++) Line 3930  for (;; ptr++)
3930          if (repeat_min > 1)          if (repeat_min > 1)
3931            {            {
3932            /* 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
3933            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3934              potential integer overflow. */
3935    
3936            if (lengthptr != NULL)            if (lengthptr != NULL)
3937              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3938                int delta = (repeat_min - 1)*length_prevgroup;
3939                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3940                                                                (double)INT_MAX ||
3941                    OFLOW_MAX - *lengthptr < delta)
3942                  {
3943                  *errorcodeptr = ERR20;
3944                  goto FAILED;
3945                  }
3946                *lengthptr += delta;
3947                }
3948    
3949            /* This is compiling for real */            /* This is compiling for real */
3950    
# Line 3418  for (;; ptr++) Line 3982  for (;; ptr++)
3982          /* 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
3983          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
3984          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
3985          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3986            paranoid checks to avoid integer overflow. */
3987    
3988          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3989            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3990              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3991                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3992              if ((double)repeat_max *
3993                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3994                      > (double)INT_MAX ||
3995                  OFLOW_MAX - *lengthptr < delta)
3996                {
3997                *errorcodeptr = ERR20;
3998                goto FAILED;
3999                }
4000              *lengthptr += delta;
4001              }
4002    
4003          /* This is compiling for real */          /* This is compiling for real */
4004    
# Line 3530  for (;; ptr++) Line 4106  for (;; ptr++)
4106        int len;        int len;
4107        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||
4108            *tempcode == OP_NOTEXACT)            *tempcode == OP_NOTEXACT)
4109          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode] +
4110              ((*tempcode == OP_TYPEEXACT &&
4111                 (tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0);
4112        len = code - tempcode;        len = code - tempcode;
4113        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4114          {          {
# Line 3574  for (;; ptr++) Line 4152  for (;; ptr++)
4152      /* ===================================================================*/      /* ===================================================================*/
4153      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4154      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4155      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. */  
4156    
4157      case '(':      case '(':
4158      newoptions = options;      newoptions = options;
4159      skipbytes = 0;      skipbytes = 0;
4160      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4161      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4162        reset_bracount = FALSE;
4163    
4164        /* First deal with various "verbs" that can be introduced by '*'. */
4165    
4166        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4167          {
4168          int i, namelen;
4169          const char *vn = verbnames;
4170          const uschar *name = ++ptr;
4171          previous = NULL;
4172          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
4173          if (*ptr == ':')
4174            {
4175            *errorcodeptr = ERR59;   /* Not supported */
4176            goto FAILED;
4177            }
4178          if (*ptr != ')')
4179            {
4180            *errorcodeptr = ERR60;
4181            goto FAILED;
4182            }
4183          namelen = ptr - name;
4184          for (i = 0; i < verbcount; i++)
4185            {
4186            if (namelen == verbs[i].len &&
4187                strncmp((char *)name, vn, namelen) == 0)
4188              {
4189              *code = verbs[i].op;
4190              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4191              break;
4192              }
4193            vn += verbs[i].len + 1;
4194            }
4195          if (i < verbcount) continue;
4196          *errorcodeptr = ERR60;
4197          goto FAILED;
4198          }
4199    
4200        /* Deal with the extended parentheses; all are introduced by '?', and the
4201        appearance of any of them means that this is not a capturing group. */
4202    
4203      if (*(++ptr) == '?')      else if (*ptr == '?')
4204        {        {
4205        int i, set, unset, namelen;        int i, set, unset, namelen;
4206        int *optset;        int *optset;
# Line 3605  for (;; ptr++) Line 4221  for (;; ptr++)
4221    
4222    
4223          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4224            case '|':                 /* Reset capture count for each branch */
4225            reset_bracount = TRUE;
4226            /* Fall through */
4227    
4228            /* ------------------------------------------------------------ */
4229          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
4230          bravalue = OP_BRA;          bravalue = OP_BRA;
4231          ptr++;          ptr++;
# Line 3640  for (;; ptr++) Line 4261  for (;; ptr++)
4261    
4262          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
4263          skipbytes = 3;          skipbytes = 3;
4264          refsign = -1;          refsign = -1;
4265    
4266          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4267    
# Line 3664  for (;; ptr++) Line 4285  for (;; ptr++)
4285            terminator = '\'';            terminator = '\'';
4286            ptr++;            ptr++;
4287            }            }
4288          else          else
4289            {            {
4290            terminator = 0;            terminator = 0;
4291            if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);            if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);
4292            }            }
4293    
4294          /* We now expect to read a name; any thing else is an error */          /* We now expect to read a name; any thing else is an error */
4295    
# Line 3706  for (;; ptr++) Line 4327  for (;; ptr++)
4327          /* In the real compile we do the work of looking for the actual          /* In the real compile we do the work of looking for the actual
4328          reference. If the string started with "+" or "-" we require the rest to          reference. If the string started with "+" or "-" we require the rest to
4329          be digits, in which case recno will be set. */          be digits, in which case recno will be set. */
4330    
4331          if (refsign > 0)          if (refsign > 0)
4332            {            {
4333            if (recno <= 0)            if (recno <= 0)
4334              {              {
4335              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4336              goto FAILED;              goto FAILED;
4337              }              }
4338            if (refsign == '-')            recno = (refsign == '-')?
4339                cd->bracount - recno + 1 : recno +cd->bracount;
4340              if (recno <= 0 || recno > cd->final_bracount)
4341              {              {
4342              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4343              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4344              }              }
           else recno += cd->bracount;  
4345            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4346            break;            break;
4347            }            }
4348    
4349          /* Otherwise (did not start with "+" or "-"), start by looking for the          /* Otherwise (did not start with "+" or "-"), start by looking for the
4350          name. */          name. */
4351    
4352          slot = cd->name_table;          slot = cd->name_table;
4353          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
4354            {            {
# Line 3795  for (;; ptr++) Line 4413  for (;; ptr++)
4413            skipbytes = 1;            skipbytes = 1;
4414            }            }
4415    
4416          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4417            in the second pass here, so final_bracount is set. */
4418    
4419          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4420            {            {
4421            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4422            }            }
# Line 3821  for (;; ptr++) Line 4440  for (;; ptr++)
4440    
4441          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4442          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4443          ptr++;          ptr++;
4444            if (*ptr == ')')          /* Optimize (?!) */
4445              {
4446              *code++ = OP_FAIL;
4447              previous = NULL;
4448              continue;
4449              }
4450            bravalue = OP_ASSERT_NOT;
4451          break;          break;
4452    
4453    
# Line 3985  for (;; ptr++) Line 4610  for (;; ptr++)
4610    
4611          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4612          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4613          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4614            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4615            .NET syntax, and the Oniguruma \g<...> and \g'...' subroutine syntax. */
4616    
4617          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4618          name = ++ptr;          name = ++ptr;
# Line 3997  for (;; ptr++) Line 4624  for (;; ptr++)
4624    
4625          if (lengthptr != NULL)          if (lengthptr != NULL)
4626            {            {
4627              if (namelen == 0)
4628                {
4629                *errorcodeptr = ERR62;
4630                goto FAILED;
4631                }
4632            if (*ptr != terminator)            if (*ptr != terminator)
4633              {              {
4634              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4010  for (;; ptr++) Line 4642  for (;; ptr++)
4642            recno = 0;            recno = 0;
4643            }            }
4644    
4645          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4646            first, and then check that we have reached the end of the name in the
4647            table. That way, if the name that is longer than any in the table,
4648            the comparison will fail without reading beyond the table entry. */
4649    
4650          else          else
4651            {            {
4652            slot = cd->name_table;            slot = cd->name_table;
4653            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4654              {              {
4655              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4656                    slot[2+namelen] == 0)
4657                  break;
4658              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4659              }              }
4660    
# Line 4053  for (;; ptr++) Line 4690  for (;; ptr++)
4690          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case '5': case '6': case '7': case '8': case '9':   /* subroutine */
4691            {            {
4692            const uschar *called;            const uschar *called;
4693              terminator = ')';
4694    
4695              /* Come here from the \g<...> and \g'...' code (Oniguruma
4696              compatibility). However, the syntax has been checked to ensure that
4697              the ... are a (signed) number, so that neither ERR63 nor ERR29 will
4698              be called on this path, nor with the jump to OTHER_CHAR_AFTER_QUERY
4699              ever be taken. */
4700    
4701              HANDLE_NUMERICAL_RECURSION:
4702    
4703            if ((refsign = *ptr) == '+') ptr++;            if ((refsign = *ptr) == '+')
4704            else if (refsign == '-')              {
4705                ptr++;
4706                if ((digitab[*ptr] & ctype_digit) == 0)
4707                  {
4708                  *errorcodeptr = ERR63;
4709                  goto FAILED;
4710                  }
4711                }
4712              else if (refsign == '-')
4713              {              {
4714              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
4715                goto OTHER_CHAR_AFTER_QUERY;                goto OTHER_CHAR_AFTER_QUERY;
4716              ptr++;              ptr++;
4717              }              }
4718    
4719            recno = 0;            recno = 0;
4720            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4721              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - '0';
4722    
4723            if (*ptr != ')')            if (*ptr != terminator)
4724              {              {
4725              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4726              goto FAILED;              goto FAILED;
4727              }              }
4728    
4729            if (refsign == '-')            if (refsign == '-')
4730              {              {
4731              if (recno == 0)              if (recno == 0)
4732                {                {
4733                *errorcodeptr = ERR58;                *errorcodeptr = ERR58;
4734                goto FAILED;                goto FAILED;
4735                }                }
4736              recno = cd->bracount - recno + 1;              recno = cd->bracount - recno + 1;
4737              if (recno <= 0)              if (recno <= 0)
4738                {                {
4739                *errorcodeptr = ERR15;                *errorcodeptr = ERR15;
4740                goto FAILED;                goto FAILED;
4741                }                }
4742              }              }
4743            else if (refsign == '+')            else if (refsign == '+')
4744              {              {
# Line 4092  for (;; ptr++) Line 4746  for (;; ptr++)
4746                {                {
4747                *errorcodeptr = ERR58;                *errorcodeptr = ERR58;
4748                goto FAILED;                goto FAILED;
4749                }                }
4750              recno += cd->bracount;              recno += cd->bracount;
4751              }              }
4752    
4753            /* Come here from code above that handles a named recursion */            /* Come here from code above that handles a named recursion */
4754    
# Line 4168  for (;; ptr++) Line 4822  for (;; ptr++)
4822    
4823          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4824          default:              /* Other characters: check option setting */          default:              /* Other characters: check option setting */
4825          OTHER_CHAR_AFTER_QUERY:          OTHER_CHAR_AFTER_QUERY:
4826          set = unset = 0;          set = unset = 0;
4827          optset = &set;          optset = &set;
4828    
# Line 4180  for (;; ptr++) Line 4834  for (;; ptr++)
4834    
4835              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4836              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4837              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4838              break;              break;
4839    
4840              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4303  for (;; ptr++) Line 4957  for (;; ptr++)
4957           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
4958           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
4959            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
4960             reset_bracount,               /* True if (?| group */
4961           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
4962           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
4963           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4319  for (;; ptr++) Line 4974  for (;; ptr++)
4974      is on the bracket. */      is on the bracket. */
4975    
4976      /* If this is a conditional bracket, check that there are no more than      /* If this is a conditional bracket, check that there are no more than
4977      two branches in the group, or just one if it's a DEFINE group. */      two branches in the group, or just one if it's a DEFINE group. We do this
4978        in the real compile phase, not in the pre-pass, where the whole group may
4979        not be available. */
4980    
4981      if (bravalue == OP_COND)      if (bravalue == OP_COND && lengthptr == NULL)
4982        {        {
4983        uschar *tc = code;        uschar *tc = code;
4984        int condcount = 0;        int condcount = 0;
# Line 4368  for (;; ptr++) Line 5025  for (;; ptr++)
5025        goto FAILED;        goto FAILED;
5026        }        }
5027    
5028      /* 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,
5029      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
5030      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
5031      a quantifier. */      duplicated by a quantifier.*/
5032    
5033      if (lengthptr != NULL)      if (lengthptr != NULL)
5034        {        {
5035          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
5036            {
5037            *errorcodeptr = ERR20;
5038            goto FAILED;
5039            }
5040        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
5041        code++;        *code++ = OP_BRA;
5042        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
5043        *code++ = OP_KET;        *code++ = OP_KET;
5044        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
5045          break;    /* No need to waste time with special character handling */
5046        }        }
5047    
5048      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
5049    
5050      else code = tempcode;      code = tempcode;
5051    
5052      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
5053      relevant. */      relevant. */
# Line 4453  for (;; ptr++) Line 5116  for (;; ptr++)
5116      back references and those types that consume a character may be repeated.      back references and those types that consume a character may be repeated.
5117      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
5118      have to change if any new ones are ever created. */      have to change if any new ones are ever created. */
5119    
5120      case '\\':      case '\\':
5121      tempptr = ptr;      tempptr = ptr;
5122      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);
# Line 4480  for (;; ptr++) Line 5143  for (;; ptr++)
5143    
5144        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
5145        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
5146    
5147          /* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'
5148          is a subroutine call by number (Oniguruma syntax). In fact, the value
5149          -ESC_g is returned only for these cases. So we don't need to check for <
5150          or ' if the value is -ESC_g. For the Perl syntax \g{n} the value is
5151          -ESC_REF+n, and for the Perl syntax \g{name} the result is -ESC_k (as
5152          that is a synonym for a named back reference). */
5153    
5154          if (-c == ESC_g)
5155            {
5156            const uschar *p;
5157            save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
5158            terminator = (*(++ptr) == '<')? '>' : '\'';
5159    
5160            /* These two statements stop the compiler for warning about possibly
5161            unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In
5162            fact, because we actually check for a number below, the paths that
5163            would actually be in error are never taken. */
5164    
5165            skipbytes = 0;
5166            reset_bracount = FALSE;
5167    
5168            /* Test for a name */
5169    
5170            if (ptr[1] != '+' && ptr[1] != '-')
5171              {
5172              BOOL isnumber = TRUE;
5173              for (p = ptr + 1; *p != 0 && *p != terminator; p++)
5174                {
5175                if ((cd->ctypes[*p] & ctype_digit) == 0) isnumber = FALSE;
5176                if ((cd->ctypes[*p] & ctype_word) == 0) break;
5177                }
5178              if (*p != terminator)
5179                {
5180                *errorcodeptr = ERR57;
5181                break;
5182                }
5183              if (isnumber)
5184                {
5185                ptr++;
5186                goto HANDLE_NUMERICAL_RECURSION;
5187                }
5188              is_recurse = TRUE;
5189              goto NAMED_REF_OR_RECURSE;
5190              }
5191    
5192            /* Test a signed number in angle brackets or quotes. */
5193    
5194            p = ptr + 2;
5195            while ((digitab[*p] & ctype_digit) != 0) p++;
5196            if (*p != terminator)
5197              {
5198              *errorcodeptr = ERR57;
5199              break;
5200              }
5201            ptr++;
5202            goto HANDLE_NUMERICAL_RECURSION;
5203            }
5204    
5205        /* \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).
5206        We also support \k{name} (.NET syntax) */        We also support \k{name} (.NET syntax) */
# Line 4588  for (;; ptr++) Line 5309  for (;; ptr++)
5309      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5310      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5311    
5312        /* Remember if \r or \n were seen */
5313    
5314        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5315          cd->external_flags |= PCRE_HASCRORLF;
5316    
5317      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5318      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.
5319      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 4653  This function is used during the pre-com Line 5379  This function is used during the pre-com
5379  out the amount of memory needed, as well as during the real compile phase. The  out the amount of memory needed, as well as during the real compile phase. The
5380  value of lengthptr distinguishes the two phases.  value of lengthptr distinguishes the two phases.
5381    
5382  Argument:  Arguments:
5383    options        option bits, including any changes for this subpattern    options        option bits, including any changes for this subpattern
5384    oldims         previous settings of ims option bits    oldims         previous settings of ims option bits
5385    codeptr        -> the address of the current code pointer    codeptr        -> the address of the current code pointer
5386    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5387    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5388    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5389      reset_bracount TRUE to reset the count for each branch
5390    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5391    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5392    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 4673  Returns:         TRUE on success Line 5400  Returns:         TRUE on success
5400    
5401  static BOOL  static BOOL
5402  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5403    int *errorcodeptr, BOOL lookbehind, int skipbytes, int *firstbyteptr,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5404    int *reqbyteptr, branch_chain *bcptr, compile_data *cd, int *lengthptr)    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5405      int *lengthptr)
5406  {  {
5407  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
5408  uschar *code = *codeptr;  uschar *code = *codeptr;
# Line 4684  uschar *reverse_count = NULL; Line 5412  uschar *reverse_count = NULL;
5412  int firstbyte, reqbyte;  int firstbyte, reqbyte;
5413  int branchfirstbyte, branchreqbyte;  int branchfirstbyte, branchreqbyte;
5414  int length;  int length;
5415    int orig_bracount;
5416    int max_bracount;
5417  branch_chain bc;  branch_chain bc;
5418    
5419  bc.outer = bcptr;  bc.outer = bcptr;
# Line 4712  code += 1 + LINK_SIZE + skipbytes; Line 5442  code += 1 + LINK_SIZE + skipbytes;
5442    
5443  /* Loop for each alternative branch */  /* Loop for each alternative branch */
5444    
5445    orig_bracount = max_bracount = cd->bracount;
5446  for (;;)  for (;;)
5447    {    {
5448      /* For a (?| group, reset the capturing bracket count so that each branch
5449      uses the same numbers. */
5450    
5451      if (reset_bracount) cd->bracount = orig_bracount;
5452    
5453    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5454    
5455    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4743  for (;;) Line 5479  for (;;)
5479      return FALSE;      return FALSE;
5480      }      }
5481    
5482      /* Keep the highest bracket count in case (?| was used and some branch
5483      has fewer than the rest. */
5484    
5485      if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5486    
5487    /* 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. */
5488    
5489    if (lengthptr == NULL)    if (lengthptr == NULL)
# Line 4806  for (;;) Line 5547  for (;;)
5547        }        }
5548      }      }
5549    
5550    /* Reached end of expression, either ')' or end of pattern. Go back through    /* Reached end of expression, either ')' or end of pattern. In the real
5551    the alternative branches and reverse the chain of offsets, with the field in    compile phase, go back through the alternative branches and reverse the chain
5552    the BRA item now becoming an offset to the first alternative. If there are    of offsets, with the field in the BRA item now becoming an offset to the
5553    no alternatives, it points to the end of the group. The length in the    first alternative. If there are no alternatives, it points to the end of the
5554    terminating ket is always the length of the whole bracketed item. If any of    group. The length in the terminating ket is always the length of the whole
5555    the ims options were changed inside the group, compile a resetting op-code    bracketed item. If any of the ims options were changed inside the group,
5556    following, except at the very end of the pattern. Return leaving the pointer    compile a resetting op-code following, except at the very end of the pattern.
5557    at the terminating char. */    Return leaving the pointer at the terminating char. */
5558    
5559    if (*ptr != '|')    if (*ptr != '|')
5560      {      {
5561      int branch_length = code - last_branch;      if (lengthptr == NULL)
     do  
5562        {        {
5563        int prev_length = GET(last_branch, 1);        int branch_length = code - last_branch;
5564        PUT(last_branch, 1, branch_length);        do
5565        branch_length = prev_length;          {
5566        last_branch -= branch_length;          int prev_length = GET(last_branch, 1);
5567            PUT(last_branch, 1, branch_length);
5568            branch_length = prev_length;
5569            last_branch -= branch_length;
5570            }
5571          while (branch_length > 0);
5572        }        }
     while (branch_length > 0);  
5573    
5574      /* Fill in the ket */      /* Fill in the ket */
5575    
# Line 4842  for (;;) Line 5586  for (;;)
5586        length += 2;        length += 2;
5587        }        }
5588    
5589        /* Retain the highest bracket number, in case resetting was used. */
5590    
5591        cd->bracount = max_bracount;
5592    
5593      /* Set values to pass back */      /* Set values to pass back */
5594    
5595      *codeptr = code;      *codeptr = code;
5596      *ptrptr = ptr;      *ptrptr = ptr;
5597      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5598      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5599      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5600          {
5601          if (OFLOW_MAX - *lengthptr < length)
5602            {
5603            *errorcodeptr = ERR20;
5604            return FALSE;
5605            }
5606          *lengthptr += length;
5607          }
5608      return TRUE;      return TRUE;
5609      }      }
5610    
5611    /* Another branch follows; insert an "or" node. Its length field points back    /* Another branch follows. In the pre-compile phase, we can move the code
5612      pointer back to where it was for the start of the first branch. (That is,
5613      pretend that each branch is the only one.)
5614    
5615      In the real compile phase, insert an ALT node. Its length field points back
5616    to the previous branch while the bracket remains open. At the end the chain    to the previous branch while the bracket remains open. At the end the chain
5617    is reversed. It's done like this so that the start of the bracket has a    is reversed. It's done like this so that the start of the bracket has a
5618    zero offset until it is closed, making it possible to detect recursion. */    zero offset until it is closed, making it possible to detect recursion. */
5619    
5620    *code = OP_ALT;    if (lengthptr != NULL)
5621    PUT(code, 1, code - last_branch);      {
5622    bc.current = last_branch = code;      code = *codeptr + 1 + LINK_SIZE + skipbytes;
5623    code += 1 + LINK_SIZE;      length += 1 + LINK_SIZE;
5624        }
5625      else
5626        {
5627        *code = OP_ALT;
5628        PUT(code, 1, code - last_branch);
5629        bc.current = last_branch = code;
5630        code += 1 + LINK_SIZE;
5631        }
5632    
5633    ptr++;    ptr++;
   length += 1 + LINK_SIZE;  
5634    }    }
5635  /* Control never reaches here */  /* Control never reaches here */
5636  }  }
# Line 5145  real_pcre *re; Line 5913  real_pcre *re;
5913  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5914  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5915  int errorcode = 0;  int errorcode = 0;
5916    int skipatstart = 0;
5917  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5918  BOOL utf8;  BOOL utf8;
5919  #endif  #endif
# Line 5163  to fill in forward references to subpatt Line 5932  to fill in forward references to subpatt
5932    
5933  uschar cworkspace[COMPILE_WORK_SIZE];  uschar cworkspace[COMPILE_WORK_SIZE];
5934    
   
5935  /* Set this early so that early errors get offset 0. */  /* Set this early so that early errors get offset 0. */
5936    
5937  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern;
# Line 5223  cd->fcc = tables + fcc_offset; Line 5991  cd->fcc = tables + fcc_offset;
5991  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5992  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5993    
5994    /* Check for global one-time settings at the start of the pattern, and remember
5995    the offset for later. */
5996    
5997    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5998      {
5999      int newnl = 0;
6000      int newbsr = 0;
6001    
6002      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
6003        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
6004      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
6005        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
6006      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
6007        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
6008      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
6009        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
6010      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
6011        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
6012    
6013      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
6014        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
6015      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
6016        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
6017    
6018      if (newnl != 0)
6019        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
6020      else if (newbsr != 0)
6021        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
6022      else break;
6023      }
6024    
6025    /* Check validity of \R options. */
6026    
6027    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
6028      {
6029      case 0:
6030      case PCRE_BSR_ANYCRLF:
6031      case PCRE_BSR_UNICODE:
6032      break;
6033      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
6034      }
6035    
6036  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
6037  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
6038  "anycrlf". */  "anycrlf". */
6039    
6040  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
6041    {    {
6042    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
6043    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
6044    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
6045    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5282  to compile parts of the pattern into; th Line 6092  to compile parts of the pattern into; th
6092  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
6093  is a test for its doing so. */  is a test for its doing so. */
6094    
6095  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
6096  cd->names_found = 0;  cd->names_found = 0;
6097  cd->name_entry_size = 0;  cd->name_entry_size = 0;
6098  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5292  cd->hwm = cworkspace; Line 6102  cd->hwm = cworkspace;
6102  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
6103  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
6104  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
6105  cd->external_options = options;  cd->external_options = options;
6106    cd->external_flags = 0;
6107    
6108  /* 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
6109  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 5301  been put into the cd block so that they Line 6111  been put into the cd block so that they
6111  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
6112  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
6113    
6114    ptr += skipatstart;
6115  code = cworkspace;  code = cworkspace;
6116  *code = OP_BRA;  *code = OP_BRA;
6117  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
6118    &code, &ptr, &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length);    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
6119      &length);
6120  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
6121    
6122  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 5330  if (re == NULL) Line 6142  if (re == NULL)
6142    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
6143    }    }
6144    
6145  /* 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
6146  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
6147  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
6148  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
6149    pointers. */
6150    
6151  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
6152  re->size = size;  re->size = size;
6153  re->options = cd->external_options;  re->options = cd->external_options;
6154    re->flags = cd->external_flags;
6155  re->dummy1 = 0;  re->dummy1 = 0;
6156  re->first_byte = 0;  re->first_byte = 0;
6157  re->req_byte = 0;  re->req_byte = 0;
# Line 5355  field. Reset the bracket count and the n Line 6169  field. Reset the bracket count and the n
6169  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6170  */  */
6171    
6172    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6173  cd->bracount = 0;  cd->bracount = 0;
6174  cd->names_found = 0;  cd->names_found = 0;
6175  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5362  codestart = cd->name_table + re->name_en Line 6177  codestart = cd->name_table + re->name_en
6177  cd->start_code = codestart;  cd->start_code = codestart;
6178  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6179  cd->req_varyopt = 0;  cd->req_varyopt = 0;
6180  cd->nopartial = FALSE;  cd->had_accept = FALSE;
6181    
6182  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6183  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
6184  of the function here. */  of the function here. */
6185    
6186  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6187  code = (uschar *)codestart;  code = (uschar *)codestart;
6188  *code = OP_BRA;  *code = OP_BRA;
6189  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6190    &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6191  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6192  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6193    re->flags = cd->external_flags;
6194    
6195  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6196    
6197  /* 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. */
6198    
# Line 5404  while (errorcode == 0 && cd->hwm > cwork Line 6220  while (errorcode == 0 && cd->hwm > cwork
6220    if (groupptr == NULL) errorcode = ERR53;    if (groupptr == NULL) errorcode = ERR53;
6221      else PUT(((uschar *)codestart), offset, groupptr - codestart);      else PUT(((uschar *)codestart), offset, groupptr - codestart);
6222    }    }
6223    
6224  /* Give an error if there's back reference to a non-existent capturing  /* Give an error if there's back reference to a non-existent capturing
6225  subpattern. */  subpattern. */
6226    
# Line 5418  if (errorcode != 0) Line 6234  if (errorcode != 0)
6234    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6235    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6236    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6237    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6238    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6239    return NULL;    return NULL;
6240    }    }
# Line 5447  if ((re->options & PCRE_ANCHORED) == 0) Line 6263  if ((re->options & PCRE_ANCHORED) == 0)
6263        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6264        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6265           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6266        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6267        }        }
6268      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6269        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6270      }      }
6271    }    }
6272    
# Line 5464  if (reqbyte >= 0 && Line 6280  if (reqbyte >= 0 &&
6280    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6281    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6282      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6283    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6284    }    }
6285    
6286  /* 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 5475  case when building a production library. Line 6291  case when building a production library.
6291  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6292    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6293    
6294  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 " : "");  
   }  
6295    
6296  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6297    {    {
6298    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6299    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5498  if ((re->options & PCRE_FIRSTSET) != 0) Line 6302  if ((re->options & PCRE_FIRSTSET) != 0)
6302      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6303    }    }
6304    
6305  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6306    {    {
6307    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6308    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5515  was compiled can be seen. */ Line 6319  was compiled can be seen. */
6319  if (code - codestart > length)  if (code - codestart > length)
6320    {    {
6321    (pcre_free)(re);    (pcre_free)(re);
6322    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6323    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6324    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6325    return NULL;    return NULL;

Legend:
Removed from v.171  
changed lines
  Added in v.336

  ViewVC Help
Powered by ViewVC 1.1.5