/[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 172 by ph10, Tue Jun 5 10:40:13 2007 UTC revision 341 by ph10, Sat Apr 19 16:41:04 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  /*************************************************  /*************************************************
# 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;
# 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 861  be terminated by '>' because that is che Line 976  be terminated by '>' because that is che
976    
977  Arguments:  Arguments:
978    ptr          current position in the pattern    ptr          current position in the pattern
979    count        current count of capturing parens so far encountered    cd           compile background data
980    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
981    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
982    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
# Line 870  Returns:       the number of the named s Line 985  Returns:       the number of the named s
985  */  */
986    
987  static int  static int
988  find_parens(const uschar *ptr, int count, const uschar *name, int lorn,  find_parens(const uschar *ptr, compile_data *cd, const uschar *name, int lorn,
989    BOOL xmode)    BOOL xmode)
990  {  {
991  const uschar *thisname;  const uschar *thisname;
992    int count = cd->bracount;
993    
994  for (; *ptr != 0; ptr++)  for (; *ptr != 0; ptr++)
995    {    {
# Line 893  for (; *ptr != 0; ptr++) Line 1009  for (; *ptr != 0; ptr++)
1009      continue;      continue;
1010      }      }
1011    
1012    /* Skip over character classes */    /* Skip over character classes; this logic must be similar to the way they
1013      are handled for real. If the first character is '^', skip it. Also, if the
1014      first few characters (either before or after ^) are \Q\E or \E we skip them
1015      too. This makes for compatibility with Perl. */
1016    
1017    if (*ptr == '[')    if (*ptr == '[')
1018      {      {
1019        BOOL negate_class = FALSE;
1020        for (;;)
1021          {
1022          int c = *(++ptr);
1023          if (c == '\\')
1024            {
1025            if (ptr[1] == 'E') ptr++;
1026              else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
1027                else break;
1028            }
1029          else if (!negate_class && c == '^')
1030            negate_class = TRUE;
1031          else break;
1032          }
1033    
1034        /* If the next character is ']', it is a data character that must be
1035        skipped, except in JavaScript compatibility mode. */
1036    
1037        if (ptr[1] == ']' && (cd->external_options & PCRE_JAVASCRIPT_COMPAT) == 0)
1038          ptr++;
1039    
1040      while (*(++ptr) != ']')      while (*(++ptr) != ']')
1041        {        {
1042          if (*ptr == 0) return -1;
1043        if (*ptr == '\\')        if (*ptr == '\\')
1044          {          {
1045          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 926  for (; *ptr != 0; ptr++) Line 1067  for (; *ptr != 0; ptr++)
1067    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
1068    
1069    if (*ptr != '(') continue;    if (*ptr != '(') continue;
1070    if (ptr[1] != '?')    if (ptr[1] != '?' && ptr[1] != '*')
1071      {      {
1072      count++;      count++;
1073      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 1054  for (;;) Line 1195  for (;;)
1195    {    {
1196    int d;    int d;
1197    register int op = *cc;    register int op = *cc;
   
1198    switch (op)    switch (op)
1199      {      {
1200      case OP_CBRA:      case OP_CBRA:
# Line 1143  for (;;) Line 1283  for (;;)
1283    
1284      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1285      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1286        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1287      cc += 4;      cc += 4;
1288      break;      break;
1289    
# Line 1251  for (;;) Line 1392  for (;;)
1392      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1393      }      }
1394    
1395    /* 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
1396    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
1397    arrange to skip the extra bytes. */    two bytes of parameters. */
1398    
1399    else    else
1400      {      {
1401        switch(c)
1402          {
1403          case OP_TYPESTAR:
1404          case OP_TYPEMINSTAR:
1405          case OP_TYPEPLUS:
1406          case OP_TYPEMINPLUS:
1407          case OP_TYPEQUERY:
1408          case OP_TYPEMINQUERY:
1409          case OP_TYPEPOSSTAR:
1410          case OP_TYPEPOSPLUS:
1411          case OP_TYPEPOSQUERY:
1412          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1413          break;
1414    
1415          case OP_TYPEUPTO:
1416          case OP_TYPEMINUPTO:
1417          case OP_TYPEEXACT:
1418          case OP_TYPEPOSUPTO:
1419          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1420          break;
1421          }
1422    
1423        /* Add in the fixed length from the table */
1424    
1425      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1426    
1427      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1428      a multi-byte character. The length in the table is a minimum, so we have to
1429      arrange to skip the extra bytes. */
1430    
1431  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1432      if (utf8) switch(c)      if (utf8) switch(c)
1433        {        {
# Line 1315  for (;;) Line 1485  for (;;)
1485    
1486    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1487    
1488    /* 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
1489    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
1490    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1491    
1492    else    else
1493      {      {
1494        switch(c)
1495          {
1496          case OP_TYPESTAR:
1497          case OP_TYPEMINSTAR:
1498          case OP_TYPEPLUS:
1499          case OP_TYPEMINPLUS:
1500          case OP_TYPEQUERY:
1501          case OP_TYPEMINQUERY:
1502          case OP_TYPEPOSSTAR:
1503          case OP_TYPEPOSPLUS:
1504          case OP_TYPEPOSQUERY:
1505          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1506          break;
1507    
1508          case OP_TYPEPOSUPTO:
1509          case OP_TYPEUPTO:
1510          case OP_TYPEMINUPTO:
1511          case OP_TYPEEXACT:
1512          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1513          break;
1514          }
1515    
1516        /* Add in the fixed length from the table */
1517    
1518      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1519    
1520        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1521        by a multi-byte character. The length in the table is a minimum, so we have
1522        to arrange to skip the extra bytes. */
1523    
1524  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1525      if (utf8) switch(c)      if (utf8) switch(c)
1526        {        {
# Line 1359  for (;;) Line 1557  for (;;)
1557  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()
1558  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
1559  group that can match nothing. Note that first_significant_code() skips over  group that can match nothing. Note that first_significant_code() skips over
1560  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
1561  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
1562    bracket whose current branch will already have been scanned.
1563    
1564  Arguments:  Arguments:
1565    code        points to start of search    code        points to start of search
# Line 1382  for (code = first_significant_code(code Line 1581  for (code = first_significant_code(code
1581    
1582    c = *code;    c = *code;
1583    
1584      /* Skip over forward assertions; the other assertions are skipped by
1585      first_significant_code() with a TRUE final argument. */
1586    
1587      if (c == OP_ASSERT)
1588        {
1589        do code += GET(code, 1); while (*code == OP_ALT);
1590        c = *code;
1591        continue;
1592        }
1593    
1594    /* Groups with zero repeats can of course be empty; skip them. */    /* Groups with zero repeats can of course be empty; skip them. */
1595    
1596    if (c == OP_BRAZERO || c == OP_BRAMINZERO)    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO)
1597      {      {
1598      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1599      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
# Line 1394  for (code = first_significant_code(code Line 1603  for (code = first_significant_code(code
1603    
1604    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1605    
1606    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1607      {      {
1608      BOOL empty_branch;      BOOL empty_branch;
1609      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 1418  for (code = first_significant_code(code Line 1627  for (code = first_significant_code(code
1627    
1628    switch (c)    switch (c)
1629      {      {
1630      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1631        cannot be represented just by a bit map. This includes negated single
1632        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1633        actual length is stored in the compiled code, so we must update "code"
1634        here. */
1635    
1636  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1637      case OP_XCLASS:      case OP_XCLASS:
1638      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1639      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1640  #endif  #endif
1641    
# Line 1484  for (code = first_significant_code(code Line 1697  for (code = first_significant_code(code
1697      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1698      return FALSE;      return FALSE;
1699    
1700        /* These are going to continue, as they may be empty, but we have to
1701        fudge the length for the \p and \P cases. */
1702    
1703        case OP_TYPESTAR:
1704        case OP_TYPEMINSTAR:
1705        case OP_TYPEPOSSTAR:
1706        case OP_TYPEQUERY:
1707        case OP_TYPEMINQUERY:
1708        case OP_TYPEPOSQUERY:
1709        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1710        break;
1711    
1712        /* Same for these */
1713    
1714        case OP_TYPEUPTO:
1715        case OP_TYPEMINUPTO:
1716        case OP_TYPEPOSUPTO:
1717        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1718        break;
1719    
1720      /* End of branch */      /* End of branch */
1721    
1722      case OP_KET:      case OP_KET:
# Line 1553  return TRUE; Line 1786  return TRUE;
1786  *************************************************/  *************************************************/
1787    
1788  /* This function is called when the sequence "[:" or "[." or "[=" is  /* This function is called when the sequence "[:" or "[." or "[=" is
1789  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
1790  optional ^ and then a sequence of letters, terminated by a matching ":]" or  sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
1791  ".]" or "=]".  reach an unescaped ']' without the special preceding character, return FALSE.
1792    
1793    Originally, this function only recognized a sequence of letters between the
1794    terminators, but it seems that Perl recognizes any sequence of characters,
1795    though of course unknown POSIX names are subsequently rejected. Perl gives an
1796    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
1797    didn't consider this to be a POSIX class. Likewise for [:1234:].
1798    
1799    The problem in trying to be exactly like Perl is in the handling of escapes. We
1800    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
1801    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
1802    below handles the special case of \], but does not try to do any other escape
1803    processing. This makes it different from Perl for cases such as [:l\ower:]
1804    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
1805    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
1806    I think.
1807    
1808  Argument:  Arguments:
1809    ptr      pointer to the initial [    ptr      pointer to the initial [
1810    endptr   where to return the end pointer    endptr   where to return the end pointer
   cd       pointer to compile data  
1811    
1812  Returns:   TRUE or FALSE  Returns:   TRUE or FALSE
1813  */  */
1814    
1815  static BOOL  static BOOL
1816  check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd)  check_posix_syntax(const uschar *ptr, const uschar **endptr)
1817  {  {
1818  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
1819  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
1820  if (*(++ptr) == '^') ptr++;  for (++ptr; *ptr != 0; ptr++)
 while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;  
 if (*ptr == terminator && ptr[1] == ']')  
1821    {    {
1822    *endptr = ptr;    if (*ptr == '\\' && ptr[1] == ']') ptr++; else
1823    return TRUE;      {
1824        if (*ptr == ']') return FALSE;
1825        if (*ptr == terminator && ptr[1] == ']')
1826          {
1827          *endptr = ptr;
1828          return TRUE;
1829          }
1830        }
1831    }    }
1832  return FALSE;  return FALSE;
1833  }  }
# Line 1600  Returns:     a value representing the na Line 1852  Returns:     a value representing the na
1852  static int  static int
1853  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1854  {  {
1855    const char *pn = posix_names;
1856  register int yield = 0;  register int yield = 0;
1857  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1858    {    {
1859    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1860      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1861      pn += posix_name_lengths[yield] + 1;
1862    yield++;    yield++;
1863    }    }
1864  return -1;  return -1;
# Line 1619  return -1; Line 1873  return -1;
1873  that is referenced. This means that groups can be replicated for fixed  that is referenced. This means that groups can be replicated for fixed
1874  repetition simply by copying (because the recursion is allowed to refer to  repetition simply by copying (because the recursion is allowed to refer to
1875  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
1876  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
1877  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
1878  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
1879  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
1880  the partially compiled regex must be temporarily terminated with OP_END.  is called, the partially compiled regex must be temporarily terminated with
1881    OP_END.
1882    
1883  This function has been extended with the possibility of forward references for  This function has been extended with the possibility of forward references for
1884  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 1646  adjust_recurse(uschar *group, int adjust Line 1901  adjust_recurse(uschar *group, int adjust
1901    uschar *save_hwm)    uschar *save_hwm)
1902  {  {
1903  uschar *ptr = group;  uschar *ptr = group;
1904    
1905  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1906    {    {
1907    int offset;    int offset;
# Line 1653  while ((ptr = (uschar *)find_recurse(ptr Line 1909  while ((ptr = (uschar *)find_recurse(ptr
1909    
1910    /* 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
1911    reference. */    reference. */
1912    
1913    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
1914      {      {
1915      offset = GET(hc, 0);      offset = GET(hc, 0);
# Line 1907  if (next >= 0) switch(op_code) Line 2163  if (next >= 0) switch(op_code)
2163    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, "item" must be a single-byte character. */
2164    
2165    case OP_NOT:    case OP_NOT:
   if (next < 0) return FALSE;  /* Not a character */  
2166    if (item == next) return TRUE;    if (item == next) return TRUE;
2167    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2168  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1944  if (next >= 0) switch(op_code) Line 2199  if (next >= 0) switch(op_code)
2199    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2200    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2201    
2202      case OP_HSPACE:
2203      case OP_NOT_HSPACE:
2204      switch(next)
2205        {
2206        case 0x09:
2207        case 0x20:
2208        case 0xa0:
2209        case 0x1680:
2210        case 0x180e:
2211        case 0x2000:
2212        case 0x2001:
2213        case 0x2002:
2214        case 0x2003:
2215        case 0x2004:
2216        case 0x2005:
2217        case 0x2006:
2218        case 0x2007:
2219        case 0x2008:
2220        case 0x2009:
2221        case 0x200A:
2222        case 0x202f:
2223        case 0x205f:
2224        case 0x3000:
2225        return op_code != OP_HSPACE;
2226        default:
2227        return op_code == OP_HSPACE;
2228        }
2229    
2230      case OP_VSPACE:
2231      case OP_NOT_VSPACE:
2232      switch(next)
2233        {
2234        case 0x0a:
2235        case 0x0b:
2236        case 0x0c:
2237        case 0x0d:
2238        case 0x85:
2239        case 0x2028:
2240        case 0x2029:
2241        return op_code != OP_VSPACE;
2242        default:
2243        return op_code == OP_VSPACE;
2244        }
2245    
2246    default:    default:
2247    return FALSE;    return FALSE;
2248    }    }
# Line 1978  switch(op_code) Line 2277  switch(op_code)
2277      case ESC_W:      case ESC_W:
2278      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2279    
2280        case ESC_h:
2281        case ESC_H:
2282        switch(item)
2283          {
2284          case 0x09:
2285          case 0x20:
2286          case 0xa0:
2287          case 0x1680:
2288          case 0x180e:
2289          case 0x2000:
2290          case 0x2001:
2291          case 0x2002:
2292          case 0x2003:
2293          case 0x2004:
2294          case 0x2005:
2295          case 0x2006:
2296          case 0x2007:
2297          case 0x2008:
2298          case 0x2009:
2299          case 0x200A:
2300          case 0x202f:
2301          case 0x205f:
2302          case 0x3000:
2303          return -next != ESC_h;
2304          default:
2305          return -next == ESC_h;
2306          }
2307    
2308        case ESC_v:
2309        case ESC_V:
2310        switch(item)
2311          {
2312          case 0x0a:
2313          case 0x0b:
2314          case 0x0c:
2315          case 0x0d:
2316          case 0x85:
2317          case 0x2028:
2318          case 0x2029:
2319          return -next != ESC_v;
2320          default:
2321          return -next == ESC_v;
2322          }
2323    
2324      default:      default:
2325      return FALSE;      return FALSE;
2326      }      }
2327    
2328    case OP_DIGIT:    case OP_DIGIT:
2329    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2330             next == -ESC_h || next == -ESC_v;
2331    
2332    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2333    return next == -ESC_d;    return next == -ESC_d;
# Line 1992  switch(op_code) Line 2336  switch(op_code)
2336    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2337    
2338    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2339    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2340    
2341      case OP_HSPACE:
2342      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2343    
2344      case OP_NOT_HSPACE:
2345      return next == -ESC_h;
2346    
2347      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2348      case OP_VSPACE:
2349      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2350    
2351      case OP_NOT_VSPACE:
2352      return next == -ESC_v;
2353    
2354    case OP_WORDCHAR:    case OP_WORDCHAR:
2355    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2356    
2357    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2358    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2068  uschar classbits[32]; Line 2425  uschar classbits[32];
2425  BOOL class_utf8;  BOOL class_utf8;
2426  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8 = (options & PCRE_UTF8) != 0;
2427  uschar *class_utf8data;  uschar *class_utf8data;
2428    uschar *class_utf8data_base;
2429  uschar utf8_char[6];  uschar utf8_char[6];
2430  #else  #else
2431  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
# Line 2107  req_caseopt = ((options & PCRE_CASELESS) Line 2465  req_caseopt = ((options & PCRE_CASELESS)
2465  for (;; ptr++)  for (;; ptr++)
2466    {    {
2467    BOOL negate_class;    BOOL negate_class;
2468      BOOL should_flip_negation;
2469    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2470    BOOL is_quantifier;    BOOL is_quantifier;
2471    BOOL is_recurse;    BOOL is_recurse;
2472      BOOL reset_bracount;
2473    int class_charcount;    int class_charcount;
2474    int class_lastchar;    int class_lastchar;
2475    int newoptions;    int newoptions;
# Line 2125  for (;; ptr++) Line 2485  for (;; ptr++)
2485    /* Get next byte in the pattern */    /* Get next byte in the pattern */
2486    
2487    c = *ptr;    c = *ptr;
2488    
2489    /* 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
2490    previous cycle of this loop. */    previous cycle of this loop. */
2491    
# Line 2147  for (;; ptr++) Line 2507  for (;; ptr++)
2507      */      */
2508    
2509      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2510    
2511        /* Paranoid check for integer overflow */
2512    
2513        if (OFLOW_MAX - *lengthptr < code - last_code)
2514          {
2515          *errorcodeptr = ERR20;
2516          goto FAILED;
2517          }
2518    
2519      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2520      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));
2521    
# Line 2259  for (;; ptr++) Line 2628  for (;; ptr++)
2628      *ptrptr = ptr;      *ptrptr = ptr;
2629      if (lengthptr != NULL)      if (lengthptr != NULL)
2630        {        {
2631          if (OFLOW_MAX - *lengthptr < code - last_code)
2632            {
2633            *errorcodeptr = ERR20;
2634            goto FAILED;
2635            }
2636        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2637        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2638        }        }
# Line 2306  for (;; ptr++) Line 2680  for (;; ptr++)
2680      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,
2681      but those above are are explicitly listed afterwards. A flag byte tells      but those above are are explicitly listed afterwards. A flag byte tells
2682      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.
2683      */  
2684        In JavaScript compatibility mode, an isolated ']' causes an error. In
2685        default (Perl) mode, it is treated as a data character. */
2686    
2687        case ']':
2688        if ((cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
2689          {
2690          *errorcodeptr = ERR64;
2691          goto FAILED;
2692          }
2693        goto NORMAL_CHAR;
2694    
2695      case '[':      case '[':
2696      previous = code;      previous = code;
# Line 2315  for (;; ptr++) Line 2699  for (;; ptr++)
2699      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. */
2700    
2701      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2702          check_posix_syntax(ptr, &tempptr, cd))          check_posix_syntax(ptr, &tempptr))
2703        {        {
2704        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;
2705        goto FAILED;        goto FAILED;
2706        }        }
2707    
2708      /* 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,
2709        if the first few characters (either before or after ^) are \Q\E or \E we
2710        skip them too. This makes for compatibility with Perl. */
2711    
2712      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2713        for (;;)
2714        {        {
       negate_class = TRUE;  
2715        c = *(++ptr);        c = *(++ptr);
2716        }        if (c == '\\')
2717      else          {
2718            if (ptr[1] == 'E') ptr++;
2719              else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2720                else break;
2721            }
2722          else if (!negate_class && c == '^')
2723            negate_class = TRUE;
2724          else break;
2725          }
2726    
2727        /* Empty classes are allowed in JavaScript compatibility mode. Otherwise,
2728        an initial ']' is taken as a data character -- the code below handles
2729        that. In JS mode, [] must always fail, so generate OP_FAIL, whereas
2730        [^] must match any character, so generate OP_ALLANY. */
2731    
2732        if (c ==']' && (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
2733        {        {
2734        negate_class = FALSE;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
2735        }        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
2736          zerofirstbyte = firstbyte;
2737          break;
2738          }
2739    
2740        /* If a class contains a negative special such as \S, we need to flip the
2741        negation flag at the end, so that support for characters > 255 works
2742        correctly (they are all included in the class). */
2743    
2744        should_flip_negation = FALSE;
2745    
2746      /* 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
2747      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 2350  for (;; ptr++) Line 2760  for (;; ptr++)
2760  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2761      class_utf8 = FALSE;                       /* No chars >= 256 */      class_utf8 = FALSE;                       /* No chars >= 256 */
2762      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */
2763        class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */
2764  #endif  #endif
2765    
2766      /* 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 2365  for (;; ptr++) Line 2776  for (;; ptr++)
2776          {                           /* Braces are required because the */          {                           /* Braces are required because the */
2777          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
2778          }          }
2779    
2780          /* In the pre-compile phase, accumulate the length of any UTF-8 extra
2781          data and reset the pointer. This is so that very large classes that
2782          contain a zillion UTF-8 characters no longer overwrite the work space
2783          (which is on the stack). */
2784    
2785          if (lengthptr != NULL)
2786            {
2787            *lengthptr += class_utf8data - class_utf8data_base;
2788            class_utf8data = class_utf8data_base;
2789            }
2790    
2791  #endif  #endif
2792    
2793        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 2388  for (;; ptr++) Line 2811  for (;; ptr++)
2811    
2812        if (c == '[' &&        if (c == '[' &&
2813            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2814            check_posix_syntax(ptr, &tempptr, cd))            check_posix_syntax(ptr, &tempptr))
2815          {          {
2816          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
2817          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
# Line 2405  for (;; ptr++) Line 2828  for (;; ptr++)
2828          if (*ptr == '^')          if (*ptr == '^')
2829            {            {
2830            local_negate = TRUE;            local_negate = TRUE;
2831              should_flip_negation = TRUE;  /* Note negative special */
2832            ptr++;            ptr++;
2833            }            }
2834    
# Line 2471  for (;; ptr++) Line 2895  for (;; ptr++)
2895        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
2896        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2897        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2898        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
2899        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2900    
2901        if (c == '\\')        if (c == '\\')
# Line 2479  for (;; ptr++) Line 2903  for (;; ptr++)
2903          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2904          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2905    
2906          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = '\b';       /* \b is backspace in a class */
2907          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 */
2908          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 */
2909          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
# Line 2491  for (;; ptr++) Line 2915  for (;; ptr++)
2915            else inescq = TRUE;            else inescq = TRUE;
2916            continue;            continue;
2917            }            }
2918            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2919    
2920          if (c < 0)          if (c < 0)
2921            {            {
# Line 2506  for (;; ptr++) Line 2931  for (;; ptr++)
2931              continue;              continue;
2932    
2933              case ESC_D:              case ESC_D:
2934                should_flip_negation = TRUE;
2935              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2936              continue;              continue;
2937    
# Line 2514  for (;; ptr++) Line 2940  for (;; ptr++)
2940              continue;              continue;
2941    
2942              case ESC_W:              case ESC_W:
2943                should_flip_negation = TRUE;
2944              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2945              continue;              continue;
2946    
# Line 2523  for (;; ptr++) Line 2950  for (;; ptr++)
2950              continue;              continue;
2951    
2952              case ESC_S:              case ESC_S:
2953                should_flip_negation = TRUE;
2954              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2955              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2956              continue;              continue;
2957    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
2958              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
2959              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
2960              }              }
# Line 2539  for (;; ptr++) Line 2964  for (;; ptr++)
2964            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
2965                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
2966    
2967              /* We need to deal with \H, \h, \V, and \v in both phases because
2968              they use extra memory. */
2969    
2970              if (-c == ESC_h)
2971                {
2972                SETBIT(classbits, 0x09); /* VT */
2973                SETBIT(classbits, 0x20); /* SPACE */
2974                SETBIT(classbits, 0xa0); /* NSBP */
2975    #ifdef SUPPORT_UTF8
2976                if (utf8)
2977                  {
2978                  class_utf8 = TRUE;
2979                  *class_utf8data++ = XCL_SINGLE;
2980                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
2981                  *class_utf8data++ = XCL_SINGLE;
2982                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
2983                  *class_utf8data++ = XCL_RANGE;
2984                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
2985                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
2986                  *class_utf8data++ = XCL_SINGLE;
2987                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
2988                  *class_utf8data++ = XCL_SINGLE;
2989                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
2990                  *class_utf8data++ = XCL_SINGLE;
2991                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
2992                  }
2993    #endif
2994                continue;
2995                }
2996    
2997              if (-c == ESC_H)
2998                {
2999                for (c = 0; c < 32; c++)
3000                  {
3001                  int x = 0xff;
3002                  switch (c)
3003                    {
3004                    case 0x09/8: x ^= 1 << (0x09%8); break;
3005                    case 0x20/8: x ^= 1 << (0x20%8); break;
3006                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
3007                    default: break;
3008                    }
3009                  classbits[c] |= x;
3010                  }
3011    
3012    #ifdef SUPPORT_UTF8
3013                if (utf8)
3014                  {
3015                  class_utf8 = TRUE;
3016                  *class_utf8data++ = XCL_RANGE;
3017                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
3018                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
3019                  *class_utf8data++ = XCL_RANGE;
3020                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
3021                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
3022                  *class_utf8data++ = XCL_RANGE;
3023                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
3024                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
3025                  *class_utf8data++ = XCL_RANGE;
3026                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
3027                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
3028                  *class_utf8data++ = XCL_RANGE;
3029                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
3030                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
3031                  *class_utf8data++ = XCL_RANGE;
3032                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
3033                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
3034                  *class_utf8data++ = XCL_RANGE;
3035                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
3036                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
3037                  }
3038    #endif
3039                continue;
3040                }
3041    
3042              if (-c == ESC_v)
3043                {
3044                SETBIT(classbits, 0x0a); /* LF */
3045                SETBIT(classbits, 0x0b); /* VT */
3046                SETBIT(classbits, 0x0c); /* FF */
3047                SETBIT(classbits, 0x0d); /* CR */
3048                SETBIT(classbits, 0x85); /* NEL */
3049    #ifdef SUPPORT_UTF8
3050                if (utf8)
3051                  {
3052                  class_utf8 = TRUE;
3053                  *class_utf8data++ = XCL_RANGE;
3054                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
3055                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
3056                  }
3057    #endif
3058                continue;
3059                }
3060    
3061              if (-c == ESC_V)
3062                {
3063                for (c = 0; c < 32; c++)
3064                  {
3065                  int x = 0xff;
3066                  switch (c)
3067                    {
3068                    case 0x0a/8: x ^= 1 << (0x0a%8);
3069                                 x ^= 1 << (0x0b%8);
3070                                 x ^= 1 << (0x0c%8);
3071                                 x ^= 1 << (0x0d%8);
3072                                 break;
3073                    case 0x85/8: x ^= 1 << (0x85%8); break;
3074                    default: break;
3075                    }
3076                  classbits[c] |= x;
3077                  }
3078    
3079    #ifdef SUPPORT_UTF8
3080                if (utf8)
3081                  {
3082                  class_utf8 = TRUE;
3083                  *class_utf8data++ = XCL_RANGE;
3084                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
3085                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
3086                  *class_utf8data++ = XCL_RANGE;
3087                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
3088                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
3089                  }
3090    #endif
3091                continue;
3092                }
3093    
3094            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
3095    
3096  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2590  for (;; ptr++) Line 3142  for (;; ptr++)
3142    
3143        oldptr = ptr;        oldptr = ptr;
3144    
3145          /* Remember \r or \n */
3146    
3147          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3148    
3149          /* Check for range */
3150    
3151        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3152          {          {
3153          int d;          int d;
# Line 2631  for (;; ptr++) Line 3189  for (;; ptr++)
3189            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3190            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3191    
3192            /* \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
3193            special means the '-' was literal */            special means the '-' was literal */
3194    
3195            if (d < 0)            if (d < 0)
# Line 2657  for (;; ptr++) Line 3215  for (;; ptr++)
3215    
3216          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3217    
3218            /* Remember \r or \n */
3219    
3220            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3221    
3222          /* 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
3223          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3224          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2679  for (;; ptr++) Line 3241  for (;; ptr++)
3241              unsigned int origd = d;              unsigned int origd = d;
3242              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3243                {                {
3244                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3245                      ocd <= (unsigned int)d)
3246                    continue;                          /* Skip embedded ranges */
3247    
3248                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3249                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3250                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3251                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3252                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3253                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3254                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3255                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3256                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3257                  d = ocd;                  d = ocd;
3258                  continue;                  continue;
# Line 2806  for (;; ptr++) Line 3372  for (;; ptr++)
3372        goto FAILED;        goto FAILED;
3373        }        }
3374    
3375    
3376    /* This code has been disabled because it would mean that \s counts as
3377    an explicit \r or \n reference, and that's not really what is wanted. Now
3378    we set the flag only if there is a literal "\r" or "\n" in the class. */
3379    
3380    #if 0
3381        /* Remember whether \r or \n are in this class */
3382    
3383        if (negate_class)
3384          {
3385          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3386          }
3387        else
3388          {
3389          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3390          }
3391    #endif
3392    
3393    
3394      /* 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
3395      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
3396      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
3397      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3398      single-bytes only. This is an historical hangover. Maybe one day we can  
3399      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3400        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3401        operate on single-bytes only. This is an historical hangover. Maybe one day
3402        we can tidy these opcodes to handle multi-byte characters.
3403    
3404      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
3405      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 2821  for (;; ptr++) Line 3409  for (;; ptr++)
3409      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3410    
3411  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3412      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3413            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3414  #else  #else
3415      if (class_charcount == 1)      if (class_charcount == 1)
3416  #endif  #endif
# Line 2867  for (;; ptr++) Line 3453  for (;; ptr++)
3453      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3454    
3455      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3456      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3457      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
3458        the class, so any that were explicitly given as well can be ignored. If
3459        (when there are explicit characters > 255 that must be listed) there are no
3460        characters < 256, we can omit the bitmap in the actual compiled code. */
3461    
3462  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3463      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3464        {        {
3465        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3466        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 2897  for (;; ptr++) Line 3486  for (;; ptr++)
3486        }        }
3487  #endif  #endif
3488    
3489      /* 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
3490      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
3491      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
3492      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3493    
3494        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3495      if (negate_class)      if (negate_class)
3496        {        {
       *code++ = OP_NCLASS;  
3497        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3498          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3499        }        }
3500      else      else
3501        {        {
       *code++ = OP_CLASS;  
3502        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3503        }        }
3504      code += 32;      code += 32;
# Line 3096  for (;; ptr++) Line 3684  for (;; ptr++)
3684        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3685        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3686    
3687        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3688    
3689        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3690    
# Line 3246  for (;; ptr++) Line 3834  for (;; ptr++)
3834        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3835        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3836    
3837        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3838    
3839        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3840          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3282  for (;; ptr++) Line 3870  for (;; ptr++)
3870          goto FAILED;          goto FAILED;
3871          }          }
3872    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3873        /* 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
3874        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
3875        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 3312  for (;; ptr++) Line 3892  for (;; ptr++)
3892    
3893        if (repeat_min == 0)        if (repeat_min == 0)
3894          {          {
3895          /* 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
3896          altogether. */          output altogether, like this:
3897    
3898          if (repeat_max == 0)          ** if (repeat_max == 0)
3899            {          **   {
3900            code = previous;          **   code = previous;
3901            goto END_REPEAT;          **   goto END_REPEAT;
3902            }          **   }
3903    
3904            However, that fails when a group is referenced as a subroutine from
3905            elsewhere in the pattern, so now we stick in OP_SKIPZERO in front of it
3906            so that it is skipped on execution. As we don't have a list of which
3907            groups are referenced, we cannot do this selectively.
3908    
3909            If the maximum is 1 or unlimited, we just have to stick in the BRAZERO
3910            and do no more at this point. However, we do need to adjust any
3911            OP_RECURSE calls inside the group that refer to the group itself or any
3912            internal or forward referenced group, because the offset is from the
3913            start of the whole regex. Temporarily terminate the pattern while doing
3914            this. */
3915    
3916          /* If the maximum is 1 or unlimited, we just have to stick in the          if (repeat_max <= 1)    /* Covers 0, 1, and unlimited */
         BRAZERO and do no more at this point. However, we do need to adjust  
         any OP_RECURSE calls inside the group that refer to the group itself or  
         any internal or forward referenced group, because the offset is from  
         the start of the whole regex. Temporarily terminate the pattern while  
         doing this. */  
   
         if (repeat_max <= 1)  
3917            {            {
3918            *code = OP_END;            *code = OP_END;
3919            adjust_recurse(previous, 1, utf8, cd, save_hwm);            adjust_recurse(previous, 1, utf8, cd, save_hwm);
3920            memmove(previous+1, previous, len);            memmove(previous+1, previous, len);
3921            code++;            code++;
3922              if (repeat_max == 0)
3923                {
3924                *previous++ = OP_SKIPZERO;
3925                goto END_REPEAT;
3926                }
3927            *previous++ = OP_BRAZERO + repeat_type;            *previous++ = OP_BRAZERO + repeat_type;
3928            }            }
3929    
# Line 3378  for (;; ptr++) Line 3968  for (;; ptr++)
3968          if (repeat_min > 1)          if (repeat_min > 1)
3969            {            {
3970            /* 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
3971            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3972              potential integer overflow. */
3973    
3974            if (lengthptr != NULL)            if (lengthptr != NULL)
3975              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3976                int delta = (repeat_min - 1)*length_prevgroup;
3977                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3978                                                                (double)INT_MAX ||
3979                    OFLOW_MAX - *lengthptr < delta)
3980                  {
3981                  *errorcodeptr = ERR20;
3982                  goto FAILED;
3983                  }
3984                *lengthptr += delta;
3985                }
3986    
3987            /* This is compiling for real */            /* This is compiling for real */
3988    
# Line 3419  for (;; ptr++) Line 4020  for (;; ptr++)
4020          /* 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
4021          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
4022          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
4023          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
4024            paranoid checks to avoid integer overflow. */
4025    
4026          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
4027            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
4028              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
4029                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
4030              if ((double)repeat_max *
4031                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
4032                      > (double)INT_MAX ||
4033                  OFLOW_MAX - *lengthptr < delta)
4034                {
4035                *errorcodeptr = ERR20;
4036                goto FAILED;
4037                }
4038              *lengthptr += delta;
4039              }
4040    
4041          /* This is compiling for real */          /* This is compiling for real */
4042    
# Line 3504  for (;; ptr++) Line 4117  for (;; ptr++)
4117            }            }
4118          }          }
4119        }        }
4120    
4121        /* If previous is OP_FAIL, it was generated by an empty class [] in
4122        JavaScript mode. The other ways in which OP_FAIL can be generated, that is
4123        by (*FAIL) or (?!) set previous to NULL, which gives a "nothing to repeat"
4124        error above. We can just ignore the repeat in JS case. */
4125    
4126        else if (*previous == OP_FAIL) goto END_REPEAT;
4127    
4128      /* Else there's some kind of shambles */      /* Else there's some kind of shambles */
4129    
# Line 3531  for (;; ptr++) Line 4151  for (;; ptr++)
4151        int len;        int len;
4152        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||
4153            *tempcode == OP_NOTEXACT)            *tempcode == OP_NOTEXACT)
4154          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode] +
4155              ((*tempcode == OP_TYPEEXACT &&
4156                 (tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0);
4157        len = code - tempcode;        len = code - tempcode;
4158        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4159          {          {
# Line 3575  for (;; ptr++) Line 4197  for (;; ptr++)
4197      /* ===================================================================*/      /* ===================================================================*/
4198      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4199      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4200      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. */  
4201    
4202      case '(':      case '(':
4203      newoptions = options;      newoptions = options;
4204      skipbytes = 0;      skipbytes = 0;
4205      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4206      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4207        reset_bracount = FALSE;
4208    
4209        /* First deal with various "verbs" that can be introduced by '*'. */
4210    
4211        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4212          {
4213          int i, namelen;
4214          const char *vn = verbnames;
4215          const uschar *name = ++ptr;
4216          previous = NULL;
4217          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
4218          if (*ptr == ':')
4219            {
4220            *errorcodeptr = ERR59;   /* Not supported */
4221            goto FAILED;
4222            }
4223          if (*ptr != ')')
4224            {
4225            *errorcodeptr = ERR60;
4226            goto FAILED;
4227            }
4228          namelen = ptr - name;
4229          for (i = 0; i < verbcount; i++)
4230            {
4231            if (namelen == verbs[i].len &&
4232                strncmp((char *)name, vn, namelen) == 0)
4233              {
4234              *code = verbs[i].op;
4235              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4236              break;
4237              }
4238            vn += verbs[i].len + 1;
4239            }
4240          if (i < verbcount) continue;
4241          *errorcodeptr = ERR60;
4242          goto FAILED;
4243          }
4244    
4245      if (*(++ptr) == '?')      /* Deal with the extended parentheses; all are introduced by '?', and the
4246        appearance of any of them means that this is not a capturing group. */
4247    
4248        else if (*ptr == '?')
4249        {        {
4250        int i, set, unset, namelen;        int i, set, unset, namelen;
4251        int *optset;        int *optset;
# Line 3606  for (;; ptr++) Line 4266  for (;; ptr++)
4266    
4267    
4268          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4269            case '|':                 /* Reset capture count for each branch */
4270            reset_bracount = TRUE;
4271            /* Fall through */
4272    
4273            /* ------------------------------------------------------------ */
4274          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
4275          bravalue = OP_BRA;          bravalue = OP_BRA;
4276          ptr++;          ptr++;
# Line 3715  for (;; ptr++) Line 4380  for (;; ptr++)
4380              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4381              goto FAILED;              goto FAILED;
4382              }              }
4383            if (refsign == '-')            recno = (refsign == '-')?
4384                cd->bracount - recno + 1 : recno +cd->bracount;
4385              if (recno <= 0 || recno > cd->final_bracount)
4386              {              {
4387              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4388              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4389              }              }
           else recno += cd->bracount;  
4390            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4391            break;            break;
4392            }            }
# Line 3749  for (;; ptr++) Line 4411  for (;; ptr++)
4411    
4412          /* Search the pattern for a forward reference */          /* Search the pattern for a forward reference */
4413    
4414          else if ((i = find_parens(ptr, cd->bracount, name, namelen,          else if ((i = find_parens(ptr, cd, name, namelen,
4415                          (options & PCRE_EXTENDED) != 0)) > 0)                          (options & PCRE_EXTENDED) != 0)) > 0)
4416            {            {
4417            PUT2(code, 2+LINK_SIZE, i);            PUT2(code, 2+LINK_SIZE, i);
# Line 3796  for (;; ptr++) Line 4458  for (;; ptr++)
4458            skipbytes = 1;            skipbytes = 1;
4459            }            }
4460    
4461          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4462            in the second pass here, so final_bracount is set. */
4463    
4464          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4465            {            {
4466            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4467            }            }
# Line 3822  for (;; ptr++) Line 4485  for (;; ptr++)
4485    
4486          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4487          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4488          ptr++;          ptr++;
4489            if (*ptr == ')')          /* Optimize (?!) */
4490              {
4491              *code++ = OP_FAIL;
4492              previous = NULL;
4493              continue;
4494              }
4495            bravalue = OP_ASSERT_NOT;
4496          break;          break;
4497    
4498    
# Line 3986  for (;; ptr++) Line 4655  for (;; ptr++)
4655    
4656          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4657          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4658          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4659            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4660            .NET syntax, and the Oniguruma \g<...> and \g'...' subroutine syntax. */
4661    
4662          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4663          name = ++ptr;          name = ++ptr;
# Line 3998  for (;; ptr++) Line 4669  for (;; ptr++)
4669    
4670          if (lengthptr != NULL)          if (lengthptr != NULL)
4671            {            {
4672              if (namelen == 0)
4673                {
4674                *errorcodeptr = ERR62;
4675                goto FAILED;
4676                }
4677            if (*ptr != terminator)            if (*ptr != terminator)
4678              {              {
4679              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4011  for (;; ptr++) Line 4687  for (;; ptr++)
4687            recno = 0;            recno = 0;
4688            }            }
4689    
4690          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4691            first, and then check that we have reached the end of the name in the
4692            table. That way, if the name that is longer than any in the table,
4693            the comparison will fail without reading beyond the table entry. */
4694    
4695          else          else
4696            {            {
4697            slot = cd->name_table;            slot = cd->name_table;
4698            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4699              {              {
4700              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4701                    slot[2+namelen] == 0)
4702                  break;
4703              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4704              }              }
4705    
# Line 4027  for (;; ptr++) Line 4708  for (;; ptr++)
4708              recno = GET2(slot, 0);              recno = GET2(slot, 0);
4709              }              }
4710            else if ((recno =                /* Forward back reference */            else if ((recno =                /* Forward back reference */
4711                      find_parens(ptr, cd->bracount, name, namelen,                      find_parens(ptr, cd, name, namelen,
4712                        (options & PCRE_EXTENDED) != 0)) <= 0)                        (options & PCRE_EXTENDED) != 0)) <= 0)
4713              {              {
4714              *errorcodeptr = ERR15;              *errorcodeptr = ERR15;
# Line 4054  for (;; ptr++) Line 4735  for (;; ptr++)
4735          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case '5': case '6': case '7': case '8': case '9':   /* subroutine */
4736            {            {
4737            const uschar *called;            const uschar *called;
4738              terminator = ')';
4739    
4740              /* Come here from the \g<...> and \g'...' code (Oniguruma
4741              compatibility). However, the syntax has been checked to ensure that
4742              the ... are a (signed) number, so that neither ERR63 nor ERR29 will
4743              be called on this path, nor with the jump to OTHER_CHAR_AFTER_QUERY
4744              ever be taken. */
4745    
4746              HANDLE_NUMERICAL_RECURSION:
4747    
4748            if ((refsign = *ptr) == '+') ptr++;            if ((refsign = *ptr) == '+')
4749                {
4750                ptr++;
4751                if ((digitab[*ptr] & ctype_digit) == 0)
4752                  {
4753                  *errorcodeptr = ERR63;
4754                  goto FAILED;
4755                  }
4756                }
4757            else if (refsign == '-')            else if (refsign == '-')
4758              {              {
4759              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
# Line 4067  for (;; ptr++) Line 4765  for (;; ptr++)
4765            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4766              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - '0';
4767    
4768            if (*ptr != ')')            if (*ptr != terminator)
4769              {              {
4770              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4771              goto FAILED;              goto FAILED;
# Line 4120  for (;; ptr++) Line 4818  for (;; ptr++)
4818    
4819              if (called == NULL)              if (called == NULL)
4820                {                {
4821                if (find_parens(ptr, cd->bracount, NULL, recno,                if (find_parens(ptr, cd, NULL, recno,
4822                     (options & PCRE_EXTENDED) != 0) < 0)                      (options & PCRE_EXTENDED) != 0) < 0)
4823                  {                  {
4824                  *errorcodeptr = ERR15;                  *errorcodeptr = ERR15;
4825                  goto FAILED;                  goto FAILED;
# Line 4181  for (;; ptr++) Line 4879  for (;; ptr++)
4879    
4880              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4881              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4882              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4883              break;              break;
4884    
4885              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4304  for (;; ptr++) Line 5002  for (;; ptr++)
5002           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
5003           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
5004            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
5005             reset_bracount,               /* True if (?| group */
5006           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
5007           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
5008           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4371  for (;; ptr++) Line 5070  for (;; ptr++)
5070        goto FAILED;        goto FAILED;
5071        }        }
5072    
5073      /* 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,
5074      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
5075      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
5076      a quantifier. */      duplicated by a quantifier.*/
5077    
5078      if (lengthptr != NULL)      if (lengthptr != NULL)
5079        {        {
5080          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
5081            {
5082            *errorcodeptr = ERR20;
5083            goto FAILED;
5084            }
5085        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
5086        code++;        *code++ = OP_BRA;
5087        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
5088        *code++ = OP_KET;        *code++ = OP_KET;
5089        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
5090          break;    /* No need to waste time with special character handling */
5091        }        }
5092    
5093      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
5094    
5095      else code = tempcode;      code = tempcode;
5096    
5097      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
5098      relevant. */      relevant. */
# Line 4456  for (;; ptr++) Line 5161  for (;; ptr++)
5161      back references and those types that consume a character may be repeated.      back references and those types that consume a character may be repeated.
5162      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
5163      have to change if any new ones are ever created. */      have to change if any new ones are ever created. */
5164    
5165      case '\\':      case '\\':
5166      tempptr = ptr;      tempptr = ptr;
5167      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);
# Line 4483  for (;; ptr++) Line 5188  for (;; ptr++)
5188    
5189        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
5190        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
5191    
5192          /* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'
5193          is a subroutine call by number (Oniguruma syntax). In fact, the value
5194          -ESC_g is returned only for these cases. So we don't need to check for <
5195          or ' if the value is -ESC_g. For the Perl syntax \g{n} the value is
5196          -ESC_REF+n, and for the Perl syntax \g{name} the result is -ESC_k (as
5197          that is a synonym for a named back reference). */
5198    
5199          if (-c == ESC_g)
5200            {
5201            const uschar *p;
5202            save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
5203            terminator = (*(++ptr) == '<')? '>' : '\'';
5204    
5205            /* These two statements stop the compiler for warning about possibly
5206            unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In
5207            fact, because we actually check for a number below, the paths that
5208            would actually be in error are never taken. */
5209    
5210            skipbytes = 0;
5211            reset_bracount = FALSE;
5212    
5213            /* Test for a name */
5214    
5215            if (ptr[1] != '+' && ptr[1] != '-')
5216              {
5217              BOOL isnumber = TRUE;
5218              for (p = ptr + 1; *p != 0 && *p != terminator; p++)
5219                {
5220                if ((cd->ctypes[*p] & ctype_digit) == 0) isnumber = FALSE;
5221                if ((cd->ctypes[*p] & ctype_word) == 0) break;
5222                }
5223              if (*p != terminator)
5224                {
5225                *errorcodeptr = ERR57;
5226                break;
5227                }
5228              if (isnumber)
5229                {
5230                ptr++;
5231                goto HANDLE_NUMERICAL_RECURSION;
5232                }
5233              is_recurse = TRUE;
5234              goto NAMED_REF_OR_RECURSE;
5235              }
5236    
5237            /* Test a signed number in angle brackets or quotes. */
5238    
5239            p = ptr + 2;
5240            while ((digitab[*p] & ctype_digit) != 0) p++;
5241            if (*p != terminator)
5242              {
5243              *errorcodeptr = ERR57;
5244              break;
5245              }
5246            ptr++;
5247            goto HANDLE_NUMERICAL_RECURSION;
5248            }
5249    
5250        /* \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).
5251        We also support \k{name} (.NET syntax) */        We also support \k{name} (.NET syntax) */
# Line 4591  for (;; ptr++) Line 5354  for (;; ptr++)
5354      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5355      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5356    
5357        /* Remember if \r or \n were seen */
5358    
5359        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5360          cd->external_flags |= PCRE_HASCRORLF;
5361    
5362      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5363      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.
5364      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 4663  Arguments: Line 5431  Arguments:
5431    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5432    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5433    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5434      reset_bracount TRUE to reset the count for each branch
5435    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5436    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5437    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 4676  Returns:         TRUE on success Line 5445  Returns:         TRUE on success
5445    
5446  static BOOL  static BOOL
5447  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5448    int *errorcodeptr, BOOL lookbehind, int skipbytes, int *firstbyteptr,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5449    int *reqbyteptr, branch_chain *bcptr, compile_data *cd, int *lengthptr)    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5450      int *lengthptr)
5451  {  {
5452  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
5453  uschar *code = *codeptr;  uschar *code = *codeptr;
# Line 4687  uschar *reverse_count = NULL; Line 5457  uschar *reverse_count = NULL;
5457  int firstbyte, reqbyte;  int firstbyte, reqbyte;
5458  int branchfirstbyte, branchreqbyte;  int branchfirstbyte, branchreqbyte;
5459  int length;  int length;
5460    int orig_bracount;
5461    int max_bracount;
5462  branch_chain bc;  branch_chain bc;
5463    
5464  bc.outer = bcptr;  bc.outer = bcptr;
# Line 4715  code += 1 + LINK_SIZE + skipbytes; Line 5487  code += 1 + LINK_SIZE + skipbytes;
5487    
5488  /* Loop for each alternative branch */  /* Loop for each alternative branch */
5489    
5490    orig_bracount = max_bracount = cd->bracount;
5491  for (;;)  for (;;)
5492    {    {
5493      /* For a (?| group, reset the capturing bracket count so that each branch
5494      uses the same numbers. */
5495    
5496      if (reset_bracount) cd->bracount = orig_bracount;
5497    
5498    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5499    
5500    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4746  for (;;) Line 5524  for (;;)
5524      return FALSE;      return FALSE;
5525      }      }
5526    
5527      /* Keep the highest bracket count in case (?| was used and some branch
5528      has fewer than the rest. */
5529    
5530      if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5531    
5532    /* 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. */
5533    
5534    if (lengthptr == NULL)    if (lengthptr == NULL)
# Line 4848  for (;;) Line 5631  for (;;)
5631        length += 2;        length += 2;
5632        }        }
5633    
5634        /* Retain the highest bracket number, in case resetting was used. */
5635    
5636        cd->bracount = max_bracount;
5637    
5638      /* Set values to pass back */      /* Set values to pass back */
5639    
5640      *codeptr = code;      *codeptr = code;
5641      *ptrptr = ptr;      *ptrptr = ptr;
5642      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5643      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5644      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5645          {
5646          if (OFLOW_MAX - *lengthptr < length)
5647            {
5648            *errorcodeptr = ERR20;
5649            return FALSE;
5650            }
5651          *lengthptr += length;
5652          }
5653      return TRUE;      return TRUE;
5654      }      }
5655    
# Line 5163  real_pcre *re; Line 5958  real_pcre *re;
5958  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5959  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5960  int errorcode = 0;  int errorcode = 0;
5961    int skipatstart = 0;
5962  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5963  BOOL utf8;  BOOL utf8;
5964  #endif  #endif
# Line 5181  to fill in forward references to subpatt Line 5977  to fill in forward references to subpatt
5977    
5978  uschar cworkspace[COMPILE_WORK_SIZE];  uschar cworkspace[COMPILE_WORK_SIZE];
5979    
   
5980  /* Set this early so that early errors get offset 0. */  /* Set this early so that early errors get offset 0. */
5981    
5982  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern;
# Line 5241  cd->fcc = tables + fcc_offset; Line 6036  cd->fcc = tables + fcc_offset;
6036  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
6037  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
6038    
6039    /* Check for global one-time settings at the start of the pattern, and remember
6040    the offset for later. */
6041    
6042    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
6043      {
6044      int newnl = 0;
6045      int newbsr = 0;
6046    
6047      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
6048        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
6049      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
6050        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
6051      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
6052        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
6053      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
6054        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
6055      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
6056        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
6057    
6058      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
6059        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
6060      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
6061        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
6062    
6063      if (newnl != 0)
6064        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
6065      else if (newbsr != 0)
6066        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
6067      else break;
6068      }
6069    
6070    /* Check validity of \R options. */
6071    
6072    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
6073      {
6074      case 0:
6075      case PCRE_BSR_ANYCRLF:
6076      case PCRE_BSR_UNICODE:
6077      break;
6078      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
6079      }
6080    
6081  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
6082  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
6083  "anycrlf". */  "anycrlf". */
6084    
6085  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
6086    {    {
6087    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
6088    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
6089    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
6090    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5300  to compile parts of the pattern into; th Line 6137  to compile parts of the pattern into; th
6137  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
6138  is a test for its doing so. */  is a test for its doing so. */
6139    
6140  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
6141  cd->names_found = 0;  cd->names_found = 0;
6142  cd->name_entry_size = 0;  cd->name_entry_size = 0;
6143  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5310  cd->hwm = cworkspace; Line 6147  cd->hwm = cworkspace;
6147  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
6148  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
6149  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
6150  cd->external_options = options;  cd->external_options = options;
6151    cd->external_flags = 0;
6152    
6153  /* 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
6154  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 5319  been put into the cd block so that they Line 6156  been put into the cd block so that they
6156  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
6157  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
6158    
6159    ptr += skipatstart;
6160  code = cworkspace;  code = cworkspace;
6161  *code = OP_BRA;  *code = OP_BRA;
6162  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
6163    &code, &ptr, &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length);    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
6164      &length);
6165  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
6166    
6167  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 5348  if (re == NULL) Line 6187  if (re == NULL)
6187    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
6188    }    }
6189    
6190  /* 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
6191  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
6192  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
6193  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
6194    pointers. */
6195    
6196  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
6197  re->size = size;  re->size = size;
6198  re->options = cd->external_options;  re->options = cd->external_options;
6199    re->flags = cd->external_flags;
6200  re->dummy1 = 0;  re->dummy1 = 0;
6201  re->first_byte = 0;  re->first_byte = 0;
6202  re->req_byte = 0;  re->req_byte = 0;
# Line 5373  field. Reset the bracket count and the n Line 6214  field. Reset the bracket count and the n
6214  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6215  */  */
6216    
6217    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6218  cd->bracount = 0;  cd->bracount = 0;
6219  cd->names_found = 0;  cd->names_found = 0;
6220  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5380  codestart = cd->name_table + re->name_en Line 6222  codestart = cd->name_table + re->name_en
6222  cd->start_code = codestart;  cd->start_code = codestart;
6223  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6224  cd->req_varyopt = 0;  cd->req_varyopt = 0;
6225  cd->nopartial = FALSE;  cd->had_accept = FALSE;
6226    
6227  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6228  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
6229  of the function here. */  of the function here. */
6230    
6231  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6232  code = (uschar *)codestart;  code = (uschar *)codestart;
6233  *code = OP_BRA;  *code = OP_BRA;
6234  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6235    &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6236  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6237  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6238    re->flags = cd->external_flags;
6239    
6240  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6241    
6242  /* 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. */
6243    
# Line 5422  while (errorcode == 0 && cd->hwm > cwork Line 6265  while (errorcode == 0 && cd->hwm > cwork
6265    if (groupptr == NULL) errorcode = ERR53;    if (groupptr == NULL) errorcode = ERR53;
6266      else PUT(((uschar *)codestart), offset, groupptr - codestart);      else PUT(((uschar *)codestart), offset, groupptr - codestart);
6267    }    }
6268    
6269  /* 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
6270  subpattern. */  subpattern. */
6271    
# Line 5436  if (errorcode != 0) Line 6279  if (errorcode != 0)
6279    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6280    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6281    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6282    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6283    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6284    return NULL;    return NULL;
6285    }    }
# Line 5465  if ((re->options & PCRE_ANCHORED) == 0) Line 6308  if ((re->options & PCRE_ANCHORED) == 0)
6308        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6309        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6310           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6311        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6312        }        }
6313      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6314        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6315      }      }
6316    }    }
6317    
# Line 5482  if (reqbyte >= 0 && Line 6325  if (reqbyte >= 0 &&
6325    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6326    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6327      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6328    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6329    }    }
6330    
6331  /* 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 5493  case when building a production library. Line 6336  case when building a production library.
6336  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6337    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6338    
6339  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 " : "");  
   }  
6340    
6341  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6342    {    {
6343    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6344    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5516  if ((re->options & PCRE_FIRSTSET) != 0) Line 6347  if ((re->options & PCRE_FIRSTSET) != 0)
6347      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6348    }    }
6349    
6350  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6351    {    {
6352    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6353    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5533  was compiled can be seen. */ Line 6364  was compiled can be seen. */
6364  if (code - codestart > length)  if (code - codestart > length)
6365    {    {
6366    (pcre_free)(re);    (pcre_free)(re);
6367    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6368    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6369    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6370    return NULL;    return NULL;

Legend:
Removed from v.172  
changed lines
  Added in v.341

  ViewVC Help
Powered by ViewVC 1.1.5