/[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 107 by ph10, Wed Mar 7 11:02:28 2007 UTC revision 295 by ph10, Mon Dec 31 17:00:24 2007 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) 1997-2006 University of Cambridge             Copyright (c) 1997-2007 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,      0,      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 verbitem verbs[] = {
162      { 6, OP_ACCEPT },
163      { 6, OP_COMMIT },
164      { 1, OP_FAIL },
165      { 4, OP_FAIL },
166      { 5, OP_PRUNE },
167      { 4, OP_SKIP  },
168      { 4, OP_THEN  }
169    };
170    
171    static int verbcount = sizeof(verbs)/sizeof(verbitem);
172    
173    
174    /* Tables of names of POSIX character classes and their lengths. The names are
175    now all in a single string, to reduce the number of relocations when a shared
176    library is dynamically loaded. The list of lengths is terminated by a zero
177    length entry. The first three must be alpha, lower, upper, as this is assumed
178    for handling case independence. */
179    
180    static const char posix_names[] =
181      "alpha\0"  "lower\0"  "upper\0"  "alnum\0"  "ascii\0"  "blank\0"
182      "cntrl\0"  "digit\0"  "graph\0"  "print\0"  "punct\0"  "space\0"
183      "word\0"   "xdigit";
184    
185  static const uschar posix_name_lengths[] = {  static const uschar posix_name_lengths[] = {
186    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };
# Line 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 an (optionally braced) non-zero number"    "\\g is not followed by a braced name or an optionally braced non-zero number\0"
299  };    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number\0"
300      "(*VERB) with an argument is not supported\0"
301      /* 60 */
302      "(*VERB) not recognized\0"
303      "number is too big\0"
304      "subpattern name expected\0"
305      "digit expected after (?+";
306    
307    
308  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 373  static const unsigned char ebcdic_charta Line 432  static const unsigned char ebcdic_charta
432  /* Definition to allow mutual recursion */  /* Definition to allow mutual recursion */
433    
434  static BOOL  static BOOL
435    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, int, int *,    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,
436      int *, branch_chain *, compile_data *, int *);      int *, int *, branch_chain *, compile_data *, int *);
437    
438    
439    
440  /*************************************************  /*************************************************
441    *            Find an error text                  *
442    *************************************************/
443    
444    /* The error texts are now all in one long string, to save on relocations. As
445    some of the text is of unknown length, we can't use a table of offsets.
446    Instead, just count through the strings. This is not a performance issue
447    because it happens only when there has been a compilation error.
448    
449    Argument:   the error number
450    Returns:    pointer to the error string
451    */
452    
453    static const char *
454    find_error_text(int n)
455    {
456    const char *s = error_texts;
457    for (; n > 0; n--) while (*s++ != 0);
458    return s;
459    }
460    
461    
462    /*************************************************
463  *            Handle escapes                      *  *            Handle escapes                      *
464  *************************************************/  *************************************************/
465    
# Line 399  Arguments: Line 480  Arguments:
480    
481  Returns:         zero or positive => a data character  Returns:         zero or positive => a data character
482                   negative => a special escape sequence                   negative => a special escape sequence
483                   on error, errorptr is set                   on error, errorcodeptr is set
484  */  */
485    
486  static int  static int
# Line 417  ptr--; /* Set Line 498  ptr--; /* Set
498    
499  if (c == 0) *errorcodeptr = ERR1;  if (c == 0) *errorcodeptr = ERR1;
500    
501  /* Non-alphamerics are literals. For digits or letters, do an initial lookup in  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
502  a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
503  Otherwise further processing may be required. */  Otherwise further processing may be required. */
504    
505  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
506  else if (c < '0' || c > 'z') {}                           /* Not alphameric */  else if (c < '0' || c > 'z') {}                           /* Not alphanumeric */
507  else if ((i = escapes[c - '0']) != 0) c = i;  else if ((i = escapes[c - '0']) != 0) c = i;
508    
509  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
510  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphameric */  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */
511  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  c = i;
512  #endif  #endif
513    
# Line 452  else Line 533  else
533    
534      /* \g must be followed by a number, either plain or braced. If positive, it      /* \g must be followed by a number, either plain or braced. If positive, it
535      is an absolute backreference. If negative, it is a relative backreference.      is an absolute backreference. If negative, it is a relative backreference.
536      This is a Perl 5.10 feature. */      This is a Perl 5.10 feature. Perl 5.10 also supports \g{name} as a
537        reference to a named group. This is part of Perl's movement towards a
538        unified syntax for back references. As this is synonymous with \k{name}, we
539        fudge it up by pretending it really was \k. */
540    
541      case 'g':      case 'g':
542      if (ptr[1] == '{')      if (ptr[1] == '{')
543        {        {
544          const uschar *p;
545          for (p = ptr+2; *p != 0 && *p != '}'; p++)
546            if (*p != '-' && (digitab[*p] & ctype_digit) == 0) break;
547          if (*p != 0 && *p != '}')
548            {
549            c = -ESC_k;
550            break;
551            }
552        braced = TRUE;        braced = TRUE;
553        ptr++;        ptr++;
554        }        }
# Line 473  else Line 565  else
565      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
566        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - '0';
567    
568        if (c < 0)
569          {
570          *errorcodeptr = ERR61;
571          break;
572          }
573    
574      if (c == 0 || (braced && *(++ptr) != '}'))      if (c == 0 || (braced && *(++ptr) != '}'))
575        {        {
576        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
577        return 0;        break;
578        }        }
579    
580      if (negated)      if (negated)
# Line 484  else Line 582  else
582        if (c > bracount)        if (c > bracount)
583          {          {
584          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
585          return 0;          break;
586          }          }
587        c = bracount - (c - 1);        c = bracount - (c - 1);
588        }        }
# Line 513  else Line 611  else
611        c -= '0';        c -= '0';
612        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
613          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - '0';
614          if (c < 0)
615            {
616            *errorcodeptr = ERR61;
617            break;
618            }
619        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
620          {          {
621          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 608  else Line 711  else
711      if (c == 0)      if (c == 0)
712        {        {
713        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
714        return 0;        break;
715        }        }
716    
717  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
# Line 621  else Line 724  else
724      break;      break;
725    
726      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any
727      other alphameric following \ is an error if PCRE_EXTRA was set; otherwise,      other alphanumeric following \ is an error if PCRE_EXTRA was set;
728      for Perl compatibility, it is a literal. This code looks a bit odd, but      otherwise, for Perl compatibility, it is a literal. This code looks a bit
729      there used to be some cases other than the default, and there may be again      odd, but there used to be some cases other than the default, and there may
730      in future, so I haven't "optimized" it. */      be again in future, so I haven't "optimized" it. */
731    
732      default:      default:
733      if ((options & PCRE_EXTRA) != 0) switch(c)      if ((options & PCRE_EXTRA) != 0) switch(c)
# Line 684  if (c == '{') Line 787  if (c == '{')
787      *negptr = TRUE;      *negptr = TRUE;
788      ptr++;      ptr++;
789      }      }
790    for (i = 0; i < sizeof(name) - 1; i++)    for (i = 0; i < (int)sizeof(name) - 1; i++)
791      {      {
792      c = *(++ptr);      c = *(++ptr);
793      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 713  top = _pcre_utt_size; Line 816  top = _pcre_utt_size;
816  while (bot < top)  while (bot < top)
817    {    {
818    i = (bot + top) >> 1;    i = (bot + top) >> 1;
819    c = strcmp(name, _pcre_utt[i].name);    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);
820    if (c == 0)    if (c == 0)
821      {      {
822      *dptr = _pcre_utt[i].value;      *dptr = _pcre_utt[i].value;
# Line 887  for (; *ptr != 0; ptr++) Line 990  for (; *ptr != 0; ptr++)
990      {      {
991      while (*(++ptr) != ']')      while (*(++ptr) != ']')
992        {        {
993          if (*ptr == 0) return -1;
994        if (*ptr == '\\')        if (*ptr == '\\')
995          {          {
996          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 914  for (; *ptr != 0; ptr++) Line 1018  for (; *ptr != 0; ptr++)
1018    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
1019    
1020    if (*ptr != '(') continue;    if (*ptr != '(') continue;
1021    if (ptr[1] != '?')    if (ptr[1] != '?' && ptr[1] != '*')
1022      {      {
1023      count++;      count++;
1024      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 1042  for (;;) Line 1146  for (;;)
1146    {    {
1147    int d;    int d;
1148    register int op = *cc;    register int op = *cc;
   
1149    switch (op)    switch (op)
1150      {      {
1151      case OP_CBRA:      case OP_CBRA:
# Line 1131  for (;;) Line 1234  for (;;)
1234    
1235      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1236      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1237        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1238      cc += 4;      cc += 4;
1239      break;      break;
1240    
# Line 1239  for (;;) Line 1343  for (;;)
1343      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1344      }      }
1345    
1346    /* In UTF-8 mode, opcodes that are followed by a character may be followed by    /* Otherwise, we can get the item's length from the table, except that for
1347    a multi-byte character. The length in the table is a minimum, so we have to    repeated character types, we have to test for \p and \P, which have an extra
1348    arrange to skip the extra bytes. */    two bytes of parameters. */
1349    
1350    else    else
1351      {      {
1352        switch(c)
1353          {
1354          case OP_TYPESTAR:
1355          case OP_TYPEMINSTAR:
1356          case OP_TYPEPLUS:
1357          case OP_TYPEMINPLUS:
1358          case OP_TYPEQUERY:
1359          case OP_TYPEMINQUERY:
1360          case OP_TYPEPOSSTAR:
1361          case OP_TYPEPOSPLUS:
1362          case OP_TYPEPOSQUERY:
1363          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1364          break;
1365    
1366          case OP_TYPEUPTO:
1367          case OP_TYPEMINUPTO:
1368          case OP_TYPEEXACT:
1369          case OP_TYPEPOSUPTO:
1370          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1371          break;
1372          }
1373    
1374        /* Add in the fixed length from the table */
1375    
1376      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1377    
1378      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1379      a multi-byte character. The length in the table is a minimum, so we have to
1380      arrange to skip the extra bytes. */
1381    
1382  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1383      if (utf8) switch(c)      if (utf8) switch(c)
1384        {        {
# Line 1267  for (;;) Line 1400  for (;;)
1400        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1401        break;        break;
1402        }        }
1403  #endif  #endif
1404      }      }
1405    }    }
1406  }  }
# Line 1303  for (;;) Line 1436  for (;;)
1436    
1437    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1438    
1439    /* Otherwise, we get the item's length from the table. In UTF-8 mode, opcodes    /* Otherwise, we can get the item's length from the table, except that for
1440    that are followed by a character may be followed by a multi-byte character.    repeated character types, we have to test for \p and \P, which have an extra
1441    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1442    
1443    else    else
1444      {      {
1445        switch(c)
1446          {
1447          case OP_TYPESTAR:
1448          case OP_TYPEMINSTAR:
1449          case OP_TYPEPLUS:
1450          case OP_TYPEMINPLUS:
1451          case OP_TYPEQUERY:
1452          case OP_TYPEMINQUERY:
1453          case OP_TYPEPOSSTAR:
1454          case OP_TYPEPOSPLUS:
1455          case OP_TYPEPOSQUERY:
1456          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1457          break;
1458    
1459          case OP_TYPEPOSUPTO:
1460          case OP_TYPEUPTO:
1461          case OP_TYPEMINUPTO:
1462          case OP_TYPEEXACT:
1463          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1464          break;
1465          }
1466    
1467        /* Add in the fixed length from the table */
1468    
1469      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1470    
1471        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1472        by a multi-byte character. The length in the table is a minimum, so we have
1473        to arrange to skip the extra bytes. */
1474    
1475  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1476      if (utf8) switch(c)      if (utf8) switch(c)
1477        {        {
# Line 1332  for (;;) Line 1493  for (;;)
1493        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1494        break;        break;
1495        }        }
1496  #endif  #endif
1497      }      }
1498    }    }
1499  }  }
# Line 1347  for (;;) Line 1508  for (;;)
1508  can match the empty string or not. It is called from could_be_empty()  can match the empty string or not. It is called from could_be_empty()
1509  below and from compile_branch() when checking for an unlimited repeat of a  below and from compile_branch() when checking for an unlimited repeat of a
1510  group that can match nothing. Note that first_significant_code() skips over  group that can match nothing. Note that first_significant_code() skips over
1511  assertions. If we hit an unclosed bracket, we return "empty" - this means we've  backward and negative forward assertions when its final argument is TRUE. If we
1512  struck an inner bracket whose current branch will already have been scanned.  hit an unclosed bracket, we return "empty" - this means we've struck an inner
1513    bracket whose current branch will already have been scanned.
1514    
1515  Arguments:  Arguments:
1516    code        points to start of search    code        points to start of search
# Line 1370  for (code = first_significant_code(code Line 1532  for (code = first_significant_code(code
1532    
1533    c = *code;    c = *code;
1534    
1535    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    /* Skip over forward assertions; the other assertions are skipped by
1536      first_significant_code() with a TRUE final argument. */
1537    
1538      if (c == OP_ASSERT)
1539        {
1540        do code += GET(code, 1); while (*code == OP_ALT);
1541        c = *code;
1542        continue;
1543        }
1544    
1545      /* Groups with zero repeats can of course be empty; skip them. */
1546    
1547      if (c == OP_BRAZERO || c == OP_BRAMINZERO)
1548        {
1549        code += _pcre_OP_lengths[c];
1550        do code += GET(code, 1); while (*code == OP_ALT);
1551        c = *code;
1552        continue;
1553        }
1554    
1555      /* For other groups, scan the branches. */
1556    
1557      if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1558      {      {
1559      BOOL empty_branch;      BOOL empty_branch;
1560      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 1386  for (code = first_significant_code(code Line 1570  for (code = first_significant_code(code
1570        }        }
1571      while (*code == OP_ALT);      while (*code == OP_ALT);
1572      if (!empty_branch) return FALSE;   /* All branches are non-empty */      if (!empty_branch) return FALSE;   /* All branches are non-empty */
1573        c = *code;
     /* Move past the KET and fudge things so that the increment in the "for"  
     above has no effect. */  
   
     c = OP_END;  
     code += 1 + LINK_SIZE - _pcre_OP_lengths[c];  
1574      continue;      continue;
1575      }      }
1576    
# Line 1399  for (code = first_significant_code(code Line 1578  for (code = first_significant_code(code
1578    
1579    switch (c)    switch (c)
1580      {      {
1581      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1582        cannot be represented just by a bit map. This includes negated single
1583        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1584        actual length is stored in the compiled code, so we must update "code"
1585        here. */
1586    
1587  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1588      case OP_XCLASS:      case OP_XCLASS:
1589      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1590      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1591  #endif  #endif
1592    
# Line 1465  for (code = first_significant_code(code Line 1648  for (code = first_significant_code(code
1648      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1649      return FALSE;      return FALSE;
1650    
1651        /* These are going to continue, as they may be empty, but we have to
1652        fudge the length for the \p and \P cases. */
1653    
1654        case OP_TYPESTAR:
1655        case OP_TYPEMINSTAR:
1656        case OP_TYPEPOSSTAR:
1657        case OP_TYPEQUERY:
1658        case OP_TYPEMINQUERY:
1659        case OP_TYPEPOSQUERY:
1660        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1661        break;
1662    
1663        /* Same for these */
1664    
1665        case OP_TYPEUPTO:
1666        case OP_TYPEMINUPTO:
1667        case OP_TYPEPOSUPTO:
1668        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1669        break;
1670    
1671      /* End of branch */      /* End of branch */
1672    
1673      case OP_KET:      case OP_KET:
# Line 1534  return TRUE; Line 1737  return TRUE;
1737  *************************************************/  *************************************************/
1738    
1739  /* This function is called when the sequence "[:" or "[." or "[=" is  /* This function is called when the sequence "[:" or "[." or "[=" is
1740  encountered in a character class. It checks whether this is followed by an  encountered in a character class. It checks whether this is followed by a
1741  optional ^ and then a sequence of letters, terminated by a matching ":]" or  sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
1742  ".]" or "=]".  reach an unescaped ']' without the special preceding character, return FALSE.
1743    
1744    Originally, this function only recognized a sequence of letters between the
1745    terminators, but it seems that Perl recognizes any sequence of characters,
1746    though of course unknown POSIX names are subsequently rejected. Perl gives an
1747    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
1748    didn't consider this to be a POSIX class. Likewise for [:1234:].
1749    
1750    The problem in trying to be exactly like Perl is in the handling of escapes. We
1751    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
1752    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
1753    below handles the special case of \], but does not try to do any other escape
1754    processing. This makes it different from Perl for cases such as [:l\ower:]
1755    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
1756    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
1757    I think.
1758    
1759  Argument:  Arguments:
1760    ptr      pointer to the initial [    ptr      pointer to the initial [
1761    endptr   where to return the end pointer    endptr   where to return the end pointer
   cd       pointer to compile data  
1762    
1763  Returns:   TRUE or FALSE  Returns:   TRUE or FALSE
1764  */  */
1765    
1766  static BOOL  static BOOL
1767  check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd)  check_posix_syntax(const uschar *ptr, const uschar **endptr)
1768  {  {
1769  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
1770  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
1771  if (*(++ptr) == '^') ptr++;  for (++ptr; *ptr != 0; ptr++)
 while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;  
 if (*ptr == terminator && ptr[1] == ']')  
1772    {    {
1773    *endptr = ptr;    if (*ptr == '\\' && ptr[1] == ']') ptr++; else
1774    return TRUE;      {
1775    }      if (*ptr == ']') return FALSE;
1776        if (*ptr == terminator && ptr[1] == ']')
1777          {
1778          *endptr = ptr;
1779          return TRUE;
1780          }
1781        }
1782      }
1783  return FALSE;  return FALSE;
1784  }  }
1785    
# Line 1581  Returns: a value representing the na Line 1803  Returns: a value representing the na
1803  static int  static int
1804  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1805  {  {
1806    const char *pn = posix_names;
1807  register int yield = 0;  register int yield = 0;
1808  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1809    {    {
1810    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1811      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1812      pn += posix_name_lengths[yield] + 1;
1813    yield++;    yield++;
1814    }    }
1815  return -1;  return -1;
# Line 1627  adjust_recurse(uschar *group, int adjust Line 1851  adjust_recurse(uschar *group, int adjust
1851    uschar *save_hwm)    uschar *save_hwm)
1852  {  {
1853  uschar *ptr = group;  uschar *ptr = group;
1854    
1855  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1856    {    {
1857    int offset;    int offset;
# Line 1925  if (next >= 0) switch(op_code) Line 2150  if (next >= 0) switch(op_code)
2150    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2151    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2152    
2153      case OP_HSPACE:
2154      case OP_NOT_HSPACE:
2155      switch(next)
2156        {
2157        case 0x09:
2158        case 0x20:
2159        case 0xa0:
2160        case 0x1680:
2161        case 0x180e:
2162        case 0x2000:
2163        case 0x2001:
2164        case 0x2002:
2165        case 0x2003:
2166        case 0x2004:
2167        case 0x2005:
2168        case 0x2006:
2169        case 0x2007:
2170        case 0x2008:
2171        case 0x2009:
2172        case 0x200A:
2173        case 0x202f:
2174        case 0x205f:
2175        case 0x3000:
2176        return op_code != OP_HSPACE;
2177        default:
2178        return op_code == OP_HSPACE;
2179        }
2180    
2181      case OP_VSPACE:
2182      case OP_NOT_VSPACE:
2183      switch(next)
2184        {
2185        case 0x0a:
2186        case 0x0b:
2187        case 0x0c:
2188        case 0x0d:
2189        case 0x85:
2190        case 0x2028:
2191        case 0x2029:
2192        return op_code != OP_VSPACE;
2193        default:
2194        return op_code == OP_VSPACE;
2195        }
2196    
2197    default:    default:
2198    return FALSE;    return FALSE;
2199    }    }
# Line 1959  switch(op_code) Line 2228  switch(op_code)
2228      case ESC_W:      case ESC_W:
2229      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2230    
2231        case ESC_h:
2232        case ESC_H:
2233        switch(item)
2234          {
2235          case 0x09:
2236          case 0x20:
2237          case 0xa0:
2238          case 0x1680:
2239          case 0x180e:
2240          case 0x2000:
2241          case 0x2001:
2242          case 0x2002:
2243          case 0x2003:
2244          case 0x2004:
2245          case 0x2005:
2246          case 0x2006:
2247          case 0x2007:
2248          case 0x2008:
2249          case 0x2009:
2250          case 0x200A:
2251          case 0x202f:
2252          case 0x205f:
2253          case 0x3000:
2254          return -next != ESC_h;
2255          default:
2256          return -next == ESC_h;
2257          }
2258    
2259        case ESC_v:
2260        case ESC_V:
2261        switch(item)
2262          {
2263          case 0x0a:
2264          case 0x0b:
2265          case 0x0c:
2266          case 0x0d:
2267          case 0x85:
2268          case 0x2028:
2269          case 0x2029:
2270          return -next != ESC_v;
2271          default:
2272          return -next == ESC_v;
2273          }
2274    
2275      default:      default:
2276      return FALSE;      return FALSE;
2277      }      }
2278    
2279    case OP_DIGIT:    case OP_DIGIT:
2280    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2281             next == -ESC_h || next == -ESC_v;
2282    
2283    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2284    return next == -ESC_d;    return next == -ESC_d;
# Line 1973  switch(op_code) Line 2287  switch(op_code)
2287    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2288    
2289    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2290    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2291    
2292      case OP_HSPACE:
2293      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2294    
2295      case OP_NOT_HSPACE:
2296      return next == -ESC_h;
2297    
2298      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2299      case OP_VSPACE:
2300      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2301    
2302      case OP_NOT_VSPACE:
2303      return next == -ESC_v;
2304    
2305    case OP_WORDCHAR:    case OP_WORDCHAR:
2306    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2307    
2308    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2309    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2088  req_caseopt = ((options & PCRE_CASELESS) Line 2415  req_caseopt = ((options & PCRE_CASELESS)
2415  for (;; ptr++)  for (;; ptr++)
2416    {    {
2417    BOOL negate_class;    BOOL negate_class;
2418      BOOL should_flip_negation;
2419    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2420    BOOL is_quantifier;    BOOL is_quantifier;
2421    BOOL is_recurse;    BOOL is_recurse;
2422      BOOL reset_bracount;
2423    int class_charcount;    int class_charcount;
2424    int class_lastchar;    int class_lastchar;
2425    int newoptions;    int newoptions;
2426    int recno;    int recno;
2427      int refsign;
2428    int skipbytes;    int skipbytes;
2429    int subreqbyte;    int subreqbyte;
2430    int subfirstbyte;    int subfirstbyte;
# Line 2127  for (;; ptr++) Line 2457  for (;; ptr++)
2457      */      */
2458    
2459      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2460    
2461        /* Paranoid check for integer overflow */
2462    
2463        if (OFLOW_MAX - *lengthptr < code - last_code)
2464          {
2465          *errorcodeptr = ERR20;
2466          goto FAILED;
2467          }
2468    
2469      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2470      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));
2471    
# Line 2239  for (;; ptr++) Line 2578  for (;; ptr++)
2578      *ptrptr = ptr;      *ptrptr = ptr;
2579      if (lengthptr != NULL)      if (lengthptr != NULL)
2580        {        {
2581          if (OFLOW_MAX - *lengthptr < code - last_code)
2582            {
2583            *errorcodeptr = ERR20;
2584            goto FAILED;
2585            }
2586        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2587        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2588        }        }
# Line 2295  for (;; ptr++) Line 2639  for (;; ptr++)
2639      they are encountered at the top level, so we'll do that too. */      they are encountered at the top level, so we'll do that too. */
2640    
2641      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2642          check_posix_syntax(ptr, &tempptr, cd))          check_posix_syntax(ptr, &tempptr))
2643        {        {
2644        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;
2645        goto FAILED;        goto FAILED;
2646        }        }
2647    
2648      /* If the first character is '^', set the negation flag and skip it. */      /* If the first character is '^', set the negation flag and skip it. Also,
2649        if the first few characters (either before or after ^) are \Q\E or \E we
2650        skip them too. This makes for compatibility with Perl. */
2651    
2652      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2653        for (;;)
2654        {        {
       negate_class = TRUE;  
2655        c = *(++ptr);        c = *(++ptr);
2656          if (c == '\\')
2657            {
2658            if (ptr[1] == 'E') ptr++;
2659              else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2660                else break;
2661            }
2662          else if (!negate_class && c == '^')
2663            negate_class = TRUE;
2664          else break;
2665        }        }
2666      else  
2667        {      /* If a class contains a negative special such as \S, we need to flip the
2668        negate_class = FALSE;      negation flag at the end, so that support for characters > 255 works
2669        }      correctly (they are all included in the class). */
2670    
2671        should_flip_negation = FALSE;
2672    
2673      /* Keep a count of chars with values < 256 so that we can optimize the case      /* Keep a count of chars with values < 256 so that we can optimize the case
2674      of just a single character (as long as it's < 256). However, For higher      of just a single character (as long as it's < 256). However, For higher
# Line 2368  for (;; ptr++) Line 2725  for (;; ptr++)
2725    
2726        if (c == '[' &&        if (c == '[' &&
2727            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&
2728            check_posix_syntax(ptr, &tempptr, cd))            check_posix_syntax(ptr, &tempptr))
2729          {          {
2730          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
2731          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
# Line 2385  for (;; ptr++) Line 2742  for (;; ptr++)
2742          if (*ptr == '^')          if (*ptr == '^')
2743            {            {
2744            local_negate = TRUE;            local_negate = TRUE;
2745              should_flip_negation = TRUE;  /* Note negative special */
2746            ptr++;            ptr++;
2747            }            }
2748    
# Line 2451  for (;; ptr++) Line 2809  for (;; ptr++)
2809        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
2810        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2811        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2812        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
2813        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2814    
2815        if (c == '\\')        if (c == '\\')
# Line 2459  for (;; ptr++) Line 2817  for (;; ptr++)
2817          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2818          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2819    
2820          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = '\b';       /* \b is backspace in a class */
2821          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 */
2822          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 */
2823          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
# Line 2471  for (;; ptr++) Line 2829  for (;; ptr++)
2829            else inescq = TRUE;            else inescq = TRUE;
2830            continue;            continue;
2831            }            }
2832            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2833    
2834          if (c < 0)          if (c < 0)
2835            {            {
# Line 2486  for (;; ptr++) Line 2845  for (;; ptr++)
2845              continue;              continue;
2846    
2847              case ESC_D:              case ESC_D:
2848                should_flip_negation = TRUE;
2849              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2850              continue;              continue;
2851    
# Line 2494  for (;; ptr++) Line 2854  for (;; ptr++)
2854              continue;              continue;
2855    
2856              case ESC_W:              case ESC_W:
2857                should_flip_negation = TRUE;
2858              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2859              continue;              continue;
2860    
# Line 2503  for (;; ptr++) Line 2864  for (;; ptr++)
2864              continue;              continue;
2865    
2866              case ESC_S:              case ESC_S:
2867                should_flip_negation = TRUE;
2868              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2869              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2870              continue;              continue;
2871    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
2872              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
2873              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
2874              }              }
# Line 2519  for (;; ptr++) Line 2878  for (;; ptr++)
2878            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
2879                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
2880    
2881              /* We need to deal with \H, \h, \V, and \v in both phases because
2882              they use extra memory. */
2883    
2884              if (-c == ESC_h)
2885                {
2886                SETBIT(classbits, 0x09); /* VT */
2887                SETBIT(classbits, 0x20); /* SPACE */
2888                SETBIT(classbits, 0xa0); /* NSBP */
2889    #ifdef SUPPORT_UTF8
2890                if (utf8)
2891                  {
2892                  class_utf8 = TRUE;
2893                  *class_utf8data++ = XCL_SINGLE;
2894                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
2895                  *class_utf8data++ = XCL_SINGLE;
2896                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
2897                  *class_utf8data++ = XCL_RANGE;
2898                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
2899                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
2900                  *class_utf8data++ = XCL_SINGLE;
2901                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
2902                  *class_utf8data++ = XCL_SINGLE;
2903                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
2904                  *class_utf8data++ = XCL_SINGLE;
2905                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
2906                  }
2907    #endif
2908                continue;
2909                }
2910    
2911              if (-c == ESC_H)
2912                {
2913                for (c = 0; c < 32; c++)
2914                  {
2915                  int x = 0xff;
2916                  switch (c)
2917                    {
2918                    case 0x09/8: x ^= 1 << (0x09%8); break;
2919                    case 0x20/8: x ^= 1 << (0x20%8); break;
2920                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
2921                    default: break;
2922                    }
2923                  classbits[c] |= x;
2924                  }
2925    
2926    #ifdef SUPPORT_UTF8
2927                if (utf8)
2928                  {
2929                  class_utf8 = TRUE;
2930                  *class_utf8data++ = XCL_RANGE;
2931                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2932                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
2933                  *class_utf8data++ = XCL_RANGE;
2934                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
2935                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
2936                  *class_utf8data++ = XCL_RANGE;
2937                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
2938                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
2939                  *class_utf8data++ = XCL_RANGE;
2940                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
2941                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
2942                  *class_utf8data++ = XCL_RANGE;
2943                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
2944                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
2945                  *class_utf8data++ = XCL_RANGE;
2946                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
2947                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
2948                  *class_utf8data++ = XCL_RANGE;
2949                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
2950                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2951                  }
2952    #endif
2953                continue;
2954                }
2955    
2956              if (-c == ESC_v)
2957                {
2958                SETBIT(classbits, 0x0a); /* LF */
2959                SETBIT(classbits, 0x0b); /* VT */
2960                SETBIT(classbits, 0x0c); /* FF */
2961                SETBIT(classbits, 0x0d); /* CR */
2962                SETBIT(classbits, 0x85); /* NEL */
2963    #ifdef SUPPORT_UTF8
2964                if (utf8)
2965                  {
2966                  class_utf8 = TRUE;
2967                  *class_utf8data++ = XCL_RANGE;
2968                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
2969                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2970                  }
2971    #endif
2972                continue;
2973                }
2974    
2975              if (-c == ESC_V)
2976                {
2977                for (c = 0; c < 32; c++)
2978                  {
2979                  int x = 0xff;
2980                  switch (c)
2981                    {
2982                    case 0x0a/8: x ^= 1 << (0x0a%8);
2983                                 x ^= 1 << (0x0b%8);
2984                                 x ^= 1 << (0x0c%8);
2985                                 x ^= 1 << (0x0d%8);
2986                                 break;
2987                    case 0x85/8: x ^= 1 << (0x85%8); break;
2988                    default: break;
2989                    }
2990                  classbits[c] |= x;
2991                  }
2992    
2993    #ifdef SUPPORT_UTF8
2994                if (utf8)
2995                  {
2996                  class_utf8 = TRUE;
2997                  *class_utf8data++ = XCL_RANGE;
2998                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2999                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
3000                  *class_utf8data++ = XCL_RANGE;
3001                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
3002                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
3003                  }
3004    #endif
3005                continue;
3006                }
3007    
3008            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
3009    
3010  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2570  for (;; ptr++) Line 3056  for (;; ptr++)
3056    
3057        oldptr = ptr;        oldptr = ptr;
3058    
3059          /* Remember \r or \n */
3060    
3061          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3062    
3063          /* Check for range */
3064    
3065        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3066          {          {
3067          int d;          int d;
# Line 2611  for (;; ptr++) Line 3103  for (;; ptr++)
3103            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3104            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3105    
3106            /* \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
3107            special means the '-' was literal */            special means the '-' was literal */
3108    
3109            if (d < 0)            if (d < 0)
# Line 2637  for (;; ptr++) Line 3129  for (;; ptr++)
3129    
3130          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3131    
3132            /* Remember \r or \n */
3133    
3134            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3135    
3136          /* 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
3137          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3138          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2659  for (;; ptr++) Line 3155  for (;; ptr++)
3155              unsigned int origd = d;              unsigned int origd = d;
3156              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3157                {                {
3158                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3159                      ocd <= (unsigned int)d)
3160                    continue;                          /* Skip embedded ranges */
3161    
3162                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3163                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3164                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3165                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3166                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3167                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3168                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3169                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3170                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3171                  d = ocd;                  d = ocd;
3172                  continue;                  continue;
# Line 2786  for (;; ptr++) Line 3286  for (;; ptr++)
3286        goto FAILED;        goto FAILED;
3287        }        }
3288    
3289    
3290    /* This code has been disabled because it would mean that \s counts as
3291    an explicit \r or \n reference, and that's not really what is wanted. Now
3292    we set the flag only if there is a literal "\r" or "\n" in the class. */
3293    
3294    #if 0
3295        /* Remember whether \r or \n are in this class */
3296    
3297        if (negate_class)
3298          {
3299          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3300          }
3301        else
3302          {
3303          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3304          }
3305    #endif
3306    
3307    
3308      /* 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
3309      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
3310      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
3311      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3312      single-bytes only. This is an historical hangover. Maybe one day we can  
3313      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3314        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3315        operate on single-bytes only. This is an historical hangover. Maybe one day
3316        we can tidy these opcodes to handle multi-byte characters.
3317    
3318      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
3319      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 2801  for (;; ptr++) Line 3323  for (;; ptr++)
3323      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3324    
3325  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3326      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3327            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3328  #else  #else
3329      if (class_charcount == 1)      if (class_charcount == 1)
3330  #endif  #endif
# Line 2847  for (;; ptr++) Line 3367  for (;; ptr++)
3367      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3368    
3369      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3370      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3371      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
3372        the class, so any that were explicitly given as well can be ignored. If
3373        (when there are explicit characters > 255 that must be listed) there are no
3374        characters < 256, we can omit the bitmap in the actual compiled code. */
3375    
3376  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3377      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3378        {        {
3379        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3380        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 2877  for (;; ptr++) Line 3400  for (;; ptr++)
3400        }        }
3401  #endif  #endif
3402    
3403      /* 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
3404      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
3405      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
3406      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3407    
3408        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3409      if (negate_class)      if (negate_class)
3410        {        {
       *code++ = OP_NCLASS;  
3411        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3412          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3413        }        }
3414      else      else
3415        {        {
       *code++ = OP_CLASS;  
3416        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3417        }        }
3418      code += 32;      code += 32;
# Line 3076  for (;; ptr++) Line 3598  for (;; ptr++)
3598        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3599        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3600    
3601        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3602    
3603        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3604    
# Line 3226  for (;; ptr++) Line 3748  for (;; ptr++)
3748        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3749        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3750    
3751        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3752    
3753        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3754          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3262  for (;; ptr++) Line 3784  for (;; ptr++)
3784          goto FAILED;          goto FAILED;
3785          }          }
3786    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3787        /* 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
3788        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
3789        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 3358  for (;; ptr++) Line 3872  for (;; ptr++)
3872          if (repeat_min > 1)          if (repeat_min > 1)
3873            {            {
3874            /* 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
3875            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3876              potential integer overflow. */
3877    
3878            if (lengthptr != NULL)            if (lengthptr != NULL)
3879              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3880                int delta = (repeat_min - 1)*length_prevgroup;
3881                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3882                                                                (double)INT_MAX ||
3883                    OFLOW_MAX - *lengthptr < delta)
3884                  {
3885                  *errorcodeptr = ERR20;
3886                  goto FAILED;
3887                  }
3888                *lengthptr += delta;
3889                }
3890    
3891            /* This is compiling for real */            /* This is compiling for real */
3892    
# Line 3399  for (;; ptr++) Line 3924  for (;; ptr++)
3924          /* 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
3925          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
3926          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
3927          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3928            paranoid checks to avoid integer overflow. */
3929    
3930          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3931            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3932              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3933                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3934              if ((double)repeat_max *
3935                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3936                      > (double)INT_MAX ||
3937                  OFLOW_MAX - *lengthptr < delta)
3938                {
3939                *errorcodeptr = ERR20;
3940                goto FAILED;
3941                }
3942              *lengthptr += delta;
3943              }
3944    
3945          /* This is compiling for real */          /* This is compiling for real */
3946    
# Line 3511  for (;; ptr++) Line 4048  for (;; ptr++)
4048        int len;        int len;
4049        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||
4050            *tempcode == OP_NOTEXACT)            *tempcode == OP_NOTEXACT)
4051          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode] +
4052              ((*tempcode == OP_TYPEEXACT &&
4053                 (tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0);
4054        len = code - tempcode;        len = code - tempcode;
4055        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4056          {          {
# Line 3555  for (;; ptr++) Line 4094  for (;; ptr++)
4094      /* ===================================================================*/      /* ===================================================================*/
4095      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4096      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4097      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. */  
4098    
4099      case '(':      case '(':
4100      newoptions = options;      newoptions = options;
4101      skipbytes = 0;      skipbytes = 0;
4102      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4103      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4104        reset_bracount = FALSE;
4105    
4106        /* First deal with various "verbs" that can be introduced by '*'. */
4107    
4108        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4109          {
4110          int i, namelen;
4111          const char *vn = verbnames;
4112          const uschar *name = ++ptr;
4113          previous = NULL;
4114          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
4115          if (*ptr == ':')
4116            {
4117            *errorcodeptr = ERR59;   /* Not supported */
4118            goto FAILED;
4119            }
4120          if (*ptr != ')')
4121            {
4122            *errorcodeptr = ERR60;
4123            goto FAILED;
4124            }
4125          namelen = ptr - name;
4126          for (i = 0; i < verbcount; i++)
4127            {
4128            if (namelen == verbs[i].len &&
4129                strncmp((char *)name, vn, namelen) == 0)
4130              {
4131              *code = verbs[i].op;
4132              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4133              break;
4134              }
4135            vn += verbs[i].len + 1;
4136            }
4137          if (i < verbcount) continue;
4138          *errorcodeptr = ERR60;
4139          goto FAILED;
4140          }
4141    
4142        /* Deal with the extended parentheses; all are introduced by '?', and the
4143        appearance of any of them means that this is not a capturing group. */
4144    
4145      if (*(++ptr) == '?')      else if (*ptr == '?')
4146        {        {
4147        int i, set, unset, namelen;        int i, set, unset, namelen;
4148        int *optset;        int *optset;
# Line 3586  for (;; ptr++) Line 4163  for (;; ptr++)
4163    
4164    
4165          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4166            case '|':                 /* Reset capture count for each branch */
4167            reset_bracount = TRUE;
4168            /* Fall through */
4169    
4170            /* ------------------------------------------------------------ */
4171          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
4172          bravalue = OP_BRA;          bravalue = OP_BRA;
4173          ptr++;          ptr++;
# Line 3621  for (;; ptr++) Line 4203  for (;; ptr++)
4203    
4204          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
4205          skipbytes = 3;          skipbytes = 3;
4206            refsign = -1;
4207    
4208          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4209    
# Line 3644  for (;; ptr++) Line 4227  for (;; ptr++)
4227            terminator = '\'';            terminator = '\'';
4228            ptr++;            ptr++;
4229            }            }
4230          else terminator = 0;          else
4231              {
4232              terminator = 0;
4233              if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);
4234              }
4235    
4236          /* We now expect to read a name; any thing else is an error */          /* We now expect to read a name; any thing else is an error */
4237    
# Line 3680  for (;; ptr++) Line 4267  for (;; ptr++)
4267          if (lengthptr != NULL) break;          if (lengthptr != NULL) break;
4268    
4269          /* In the real compile we do the work of looking for the actual          /* In the real compile we do the work of looking for the actual
4270          reference. */          reference. If the string started with "+" or "-" we require the rest to
4271            be digits, in which case recno will be set. */
4272    
4273            if (refsign > 0)
4274              {
4275              if (recno <= 0)
4276                {
4277                *errorcodeptr = ERR58;
4278                goto FAILED;
4279                }
4280              recno = (refsign == '-')?
4281                cd->bracount - recno + 1 : recno +cd->bracount;
4282              if (recno <= 0 || recno > cd->final_bracount)
4283                {
4284                *errorcodeptr = ERR15;
4285                goto FAILED;
4286                }
4287              PUT2(code, 2+LINK_SIZE, recno);
4288              break;
4289              }
4290    
4291            /* Otherwise (did not start with "+" or "-"), start by looking for the
4292            name. */
4293    
4294          slot = cd->name_table;          slot = cd->name_table;
4295          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
# Line 3746  for (;; ptr++) Line 4355  for (;; ptr++)
4355            skipbytes = 1;            skipbytes = 1;
4356            }            }
4357    
4358          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4359            in the second pass here, so final_bracount is set. */
4360    
4361          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4362            {            {
4363            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4364            }            }
# Line 3772  for (;; ptr++) Line 4382  for (;; ptr++)
4382    
4383          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4384          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4385          ptr++;          ptr++;
4386            if (*ptr == ')')          /* Optimize (?!) */
4387              {
4388              *code++ = OP_FAIL;
4389              previous = NULL;
4390              continue;
4391              }
4392            bravalue = OP_ASSERT_NOT;
4393          break;          break;
4394    
4395    
# Line 3936  for (;; ptr++) Line 4552  for (;; ptr++)
4552    
4553          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4554          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4555          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4556            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4557            .NET syntax. */
4558    
4559          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4560          name = ++ptr;          name = ++ptr;
# Line 3948  for (;; ptr++) Line 4566  for (;; ptr++)
4566    
4567          if (lengthptr != NULL)          if (lengthptr != NULL)
4568            {            {
4569              if (namelen == 0)
4570                {
4571                *errorcodeptr = ERR62;
4572                goto FAILED;
4573                }
4574            if (*ptr != terminator)            if (*ptr != terminator)
4575              {              {
4576              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 3961  for (;; ptr++) Line 4584  for (;; ptr++)
4584            recno = 0;            recno = 0;
4585            }            }
4586    
4587          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4588            first, and then check that we have reached the end of the name in the
4589            table. That way, if the name that is longer than any in the table,
4590            the comparison will fail without reading beyond the table entry. */
4591    
4592          else          else
4593            {            {
4594            slot = cd->name_table;            slot = cd->name_table;
4595            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4596              {              {
4597              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4598                    slot[2+namelen] == 0)
4599                  break;
4600              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4601              }              }
4602    
# Line 3999  for (;; ptr++) Line 4627  for (;; ptr++)
4627    
4628    
4629          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4630            case '-': case '+':
4631          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */
4632          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case '5': case '6': case '7': case '8': case '9':   /* subroutine */
4633            {            {
4634            const uschar *called;            const uschar *called;
4635    
4636              if ((refsign = *ptr) == '+')
4637                {
4638                ptr++;
4639                if ((digitab[*ptr] & ctype_digit) == 0)
4640                  {
4641                  *errorcodeptr = ERR63;
4642                  goto FAILED;
4643                  }
4644                }
4645              else if (refsign == '-')
4646                {
4647                if ((digitab[ptr[1]] & ctype_digit) == 0)
4648                  goto OTHER_CHAR_AFTER_QUERY;
4649                ptr++;
4650                }
4651    
4652            recno = 0;            recno = 0;
4653            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4654              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - '0';
4655    
4656            if (*ptr != ')')            if (*ptr != ')')
4657              {              {
4658              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4659              goto FAILED;              goto FAILED;
4660              }              }
4661    
4662              if (refsign == '-')
4663                {
4664                if (recno == 0)
4665                  {
4666                  *errorcodeptr = ERR58;
4667                  goto FAILED;
4668                  }
4669                recno = cd->bracount - recno + 1;
4670                if (recno <= 0)
4671                  {
4672                  *errorcodeptr = ERR15;
4673                  goto FAILED;
4674                  }
4675                }
4676              else if (refsign == '+')
4677                {
4678                if (recno == 0)
4679                  {
4680                  *errorcodeptr = ERR58;
4681                  goto FAILED;
4682                  }
4683                recno += cd->bracount;
4684                }
4685    
4686            /* Come here from code above that handles a named recursion */            /* Come here from code above that handles a named recursion */
4687    
4688            HANDLE_RECURSION:            HANDLE_RECURSION:
# Line 4084  for (;; ptr++) Line 4755  for (;; ptr++)
4755    
4756          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4757          default:              /* Other characters: check option setting */          default:              /* Other characters: check option setting */
4758            OTHER_CHAR_AFTER_QUERY:
4759          set = unset = 0;          set = unset = 0;
4760          optset = &set;          optset = &set;
4761    
# Line 4095  for (;; ptr++) Line 4767  for (;; ptr++)
4767    
4768              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4769              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4770              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4771              break;              break;
4772    
4773              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4218  for (;; ptr++) Line 4890  for (;; ptr++)
4890           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
4891           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
4892            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
4893             reset_bracount,               /* True if (?| group */
4894           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
4895           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
4896           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4234  for (;; ptr++) Line 4907  for (;; ptr++)
4907      is on the bracket. */      is on the bracket. */
4908    
4909      /* If this is a conditional bracket, check that there are no more than      /* If this is a conditional bracket, check that there are no more than
4910      two branches in the group, or just one if it's a DEFINE group. */      two branches in the group, or just one if it's a DEFINE group. We do this
4911        in the real compile phase, not in the pre-pass, where the whole group may
4912        not be available. */
4913    
4914      if (bravalue == OP_COND)      if (bravalue == OP_COND && lengthptr == NULL)
4915        {        {
4916        uschar *tc = code;        uschar *tc = code;
4917        int condcount = 0;        int condcount = 0;
# Line 4283  for (;; ptr++) Line 4958  for (;; ptr++)
4958        goto FAILED;        goto FAILED;
4959        }        }
4960    
4961      /* 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,
4962      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
4963      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
4964      a quantifier. */      duplicated by a quantifier.*/
4965    
4966      if (lengthptr != NULL)      if (lengthptr != NULL)
4967        {        {
4968          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
4969            {
4970            *errorcodeptr = ERR20;
4971            goto FAILED;
4972            }
4973        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4974        code++;        *code++ = OP_BRA;
4975        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4976        *code++ = OP_KET;        *code++ = OP_KET;
4977        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4978          break;    /* No need to waste time with special character handling */
4979        }        }
4980    
4981      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4982    
4983      else code = tempcode;      code = tempcode;
4984    
4985      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4986      relevant. */      relevant. */
# Line 4396  for (;; ptr++) Line 5077  for (;; ptr++)
5077        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
5078        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
5079    
5080        /* \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).
5081          We also support \k{name} (.NET syntax) */
5082    
5083        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\''))        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\'' || ptr[1] == '{'))
5084          {          {
5085          is_recurse = FALSE;          is_recurse = FALSE;
5086          terminator = (*(++ptr) == '<')? '>' : '\'';          terminator = (*(++ptr) == '<')? '>' : (*ptr == '\'')? '\'' : '}';
5087          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
5088          }          }
5089    
# Line 4502  for (;; ptr++) Line 5184  for (;; ptr++)
5184      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5185      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5186    
5187        /* Remember if \r or \n were seen */
5188    
5189        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5190          cd->external_flags |= PCRE_HASCRORLF;
5191    
5192      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5193      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.
5194      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 4567  This function is used during the pre-com Line 5254  This function is used during the pre-com
5254  out the amount of memory needed, as well as during the real compile phase. The  out the amount of memory needed, as well as during the real compile phase. The
5255  value of lengthptr distinguishes the two phases.  value of lengthptr distinguishes the two phases.
5256    
5257  Argument:  Arguments:
5258    options        option bits, including any changes for this subpattern    options        option bits, including any changes for this subpattern
5259    oldims         previous settings of ims option bits    oldims         previous settings of ims option bits
5260    codeptr        -> the address of the current code pointer    codeptr        -> the address of the current code pointer
5261    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5262    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5263    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5264      reset_bracount TRUE to reset the count for each branch
5265    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5266    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5267    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 4587  Returns: TRUE on success Line 5275  Returns: TRUE on success
5275    
5276  static BOOL  static BOOL
5277  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5278    int *errorcodeptr, BOOL lookbehind, int skipbytes, int *firstbyteptr,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5279    int *reqbyteptr, branch_chain *bcptr, compile_data *cd, int *lengthptr)    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5280      int *lengthptr)
5281  {  {
5282  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
5283  uschar *code = *codeptr;  uschar *code = *codeptr;
# Line 4598  uschar *reverse_count = NULL; Line 5287  uschar *reverse_count = NULL;
5287  int firstbyte, reqbyte;  int firstbyte, reqbyte;
5288  int branchfirstbyte, branchreqbyte;  int branchfirstbyte, branchreqbyte;
5289  int length;  int length;
5290    int orig_bracount;
5291    int max_bracount;
5292  branch_chain bc;  branch_chain bc;
5293    
5294  bc.outer = bcptr;  bc.outer = bcptr;
# Line 4626  code += 1 + LINK_SIZE + skipbytes; Line 5317  code += 1 + LINK_SIZE + skipbytes;
5317    
5318  /* Loop for each alternative branch */  /* Loop for each alternative branch */
5319    
5320    orig_bracount = max_bracount = cd->bracount;
5321  for (;;)  for (;;)
5322    {    {
5323      /* For a (?| group, reset the capturing bracket count so that each branch
5324      uses the same numbers. */
5325    
5326      if (reset_bracount) cd->bracount = orig_bracount;
5327    
5328    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5329    
5330    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4657  for (;;) Line 5354  for (;;)
5354      return FALSE;      return FALSE;
5355      }      }
5356    
5357      /* Keep the highest bracket count in case (?| was used and some branch
5358      has fewer than the rest. */
5359    
5360      if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5361    
5362    /* 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. */
5363    
5364    if (lengthptr == NULL)    if (lengthptr == NULL)
# Line 4720  for (;;) Line 5422  for (;;)
5422        }        }
5423      }      }
5424    
5425    /* Reached end of expression, either ')' or end of pattern. Go back through    /* Reached end of expression, either ')' or end of pattern. In the real
5426    the alternative branches and reverse the chain of offsets, with the field in    compile phase, go back through the alternative branches and reverse the chain
5427    the BRA item now becoming an offset to the first alternative. If there are    of offsets, with the field in the BRA item now becoming an offset to the
5428    no alternatives, it points to the end of the group. The length in the    first alternative. If there are no alternatives, it points to the end of the
5429    terminating ket is always the length of the whole bracketed item. If any of    group. The length in the terminating ket is always the length of the whole
5430    the ims options were changed inside the group, compile a resetting op-code    bracketed item. If any of the ims options were changed inside the group,
5431    following, except at the very end of the pattern. Return leaving the pointer    compile a resetting op-code following, except at the very end of the pattern.
5432    at the terminating char. */    Return leaving the pointer at the terminating char. */
5433    
5434    if (*ptr != '|')    if (*ptr != '|')
5435      {      {
5436      int branch_length = code - last_branch;      if (lengthptr == NULL)
     do  
5437        {        {
5438        int prev_length = GET(last_branch, 1);        int branch_length = code - last_branch;
5439        PUT(last_branch, 1, branch_length);        do
5440        branch_length = prev_length;          {
5441        last_branch -= branch_length;          int prev_length = GET(last_branch, 1);
5442            PUT(last_branch, 1, branch_length);
5443            branch_length = prev_length;
5444            last_branch -= branch_length;
5445            }
5446          while (branch_length > 0);
5447        }        }
     while (branch_length > 0);  
5448    
5449      /* Fill in the ket */      /* Fill in the ket */
5450    
# Line 4756  for (;;) Line 5461  for (;;)
5461        length += 2;        length += 2;
5462        }        }
5463    
5464        /* Retain the highest bracket number, in case resetting was used. */
5465    
5466        cd->bracount = max_bracount;
5467    
5468      /* Set values to pass back */      /* Set values to pass back */
5469    
5470      *codeptr = code;      *codeptr = code;
5471      *ptrptr = ptr;      *ptrptr = ptr;
5472      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5473      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5474      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5475          {
5476          if (OFLOW_MAX - *lengthptr < length)
5477            {
5478            *errorcodeptr = ERR20;
5479            return FALSE;
5480            }
5481          *lengthptr += length;
5482          }
5483      return TRUE;      return TRUE;
5484      }      }
5485    
5486    /* Another branch follows; insert an "or" node. Its length field points back    /* Another branch follows. In the pre-compile phase, we can move the code
5487      pointer back to where it was for the start of the first branch. (That is,
5488      pretend that each branch is the only one.)
5489    
5490      In the real compile phase, insert an ALT node. Its length field points back
5491    to the previous branch while the bracket remains open. At the end the chain    to the previous branch while the bracket remains open. At the end the chain
5492    is reversed. It's done like this so that the start of the bracket has a    is reversed. It's done like this so that the start of the bracket has a
5493    zero offset until it is closed, making it possible to detect recursion. */    zero offset until it is closed, making it possible to detect recursion. */
5494    
5495    *code = OP_ALT;    if (lengthptr != NULL)
5496    PUT(code, 1, code - last_branch);      {
5497    bc.current = last_branch = code;      code = *codeptr + 1 + LINK_SIZE + skipbytes;
5498    code += 1 + LINK_SIZE;      length += 1 + LINK_SIZE;
5499        }
5500      else
5501        {
5502        *code = OP_ALT;
5503        PUT(code, 1, code - last_branch);
5504        bc.current = last_branch = code;
5505        code += 1 + LINK_SIZE;
5506        }
5507    
5508    ptr++;    ptr++;
   length += 1 + LINK_SIZE;  
5509    }    }
5510  /* Control never reaches here */  /* Control never reaches here */
5511  }  }
# Line 5043  Returns: pointer to compiled data Line 5772  Returns: pointer to compiled data
5772                  with errorptr and erroroffset set                  with errorptr and erroroffset set
5773  */  */
5774    
5775  PCRE_DATA_SCOPE pcre *  PCRE_EXP_DEFN pcre *
5776  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
5777    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
5778  {  {
# Line 5051  return pcre_compile2(pattern, options, N Line 5780  return pcre_compile2(pattern, options, N
5780  }  }
5781    
5782    
5783  PCRE_DATA_SCOPE pcre *  PCRE_EXP_DEFN pcre *
5784  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
5785    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
5786  {  {
# Line 5059  real_pcre *re; Line 5788  real_pcre *re;
5788  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5789  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5790  int errorcode = 0;  int errorcode = 0;
5791    int skipatstart = 0;
5792  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5793  BOOL utf8;  BOOL utf8;
5794  #endif  #endif
# Line 5100  if (errorcodeptr != NULL) *errorcodeptr Line 5830  if (errorcodeptr != NULL) *errorcodeptr
5830  if (erroroffset == NULL)  if (erroroffset == NULL)
5831    {    {
5832    errorcode = ERR16;    errorcode = ERR16;
5833    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN2;
5834    }    }
5835    
5836  *erroroffset = 0;  *erroroffset = 0;
# Line 5113  if (utf8 && (options & PCRE_NO_UTF8_CHEC Line 5843  if (utf8 && (options & PCRE_NO_UTF8_CHEC
5843       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)
5844    {    {
5845    errorcode = ERR44;    errorcode = ERR44;
5846    goto PCRE_UTF8_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN2;
5847    }    }
5848  #else  #else
5849  if ((options & PCRE_UTF8) != 0)  if ((options & PCRE_UTF8) != 0)
# Line 5137  cd->fcc = tables + fcc_offset; Line 5867  cd->fcc = tables + fcc_offset;
5867  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5868  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5869    
5870    /* Check for global one-time settings at the start of the pattern, and remember
5871    the offset for later. */
5872    
5873    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5874      {
5875      int newnl = 0;
5876      int newbsr = 0;
5877    
5878      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5879        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5880      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5881        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5882      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5883        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5884      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5885        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5886      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5887        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5888    
5889      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5890        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5891      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5892        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5893    
5894      if (newnl != 0)
5895        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5896      else if (newbsr != 0)
5897        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5898      else break;
5899      }
5900    
5901    /* Check validity of \R options. */
5902    
5903    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5904      {
5905      case 0:
5906      case PCRE_BSR_ANYCRLF:
5907      case PCRE_BSR_UNICODE:
5908      break;
5909      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5910      }
5911    
5912  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5913  current code allows for fixed one- or two-byte sequences, plus "any". */  current code allows for fixed one- or two-byte sequences, plus "any" and
5914    "anycrlf". */
5915    
5916  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5917    {    {
5918    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5919    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5920    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5921    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
5922         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;
5923    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
5924      case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
5925    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5926    }    }
5927    
5928  if (newline < 0)  if (newline == -2)
5929      {
5930      cd->nltype = NLTYPE_ANYCRLF;
5931      }
5932    else if (newline < 0)
5933    {    {
5934    cd->nltype = NLTYPE_ANY;    cd->nltype = NLTYPE_ANY;
5935    }    }
# Line 5190  to compile parts of the pattern into; th Line 5968  to compile parts of the pattern into; th
5968  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
5969  is a test for its doing so. */  is a test for its doing so. */
5970    
5971  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
5972  cd->names_found = 0;  cd->names_found = 0;
5973  cd->name_entry_size = 0;  cd->name_entry_size = 0;
5974  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5200  cd->hwm = cworkspace; Line 5978  cd->hwm = cworkspace;
5978  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5979  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5980  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5981  cd->external_options = options;  cd->external_options = options;
5982    cd->external_flags = 0;
5983    
5984  /* 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
5985  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 5209  been put into the cd block so that they Line 5987  been put into the cd block so that they
5987  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5988  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5989    
5990    ptr += skipatstart;
5991  code = cworkspace;  code = cworkspace;
5992  *code = OP_BRA;  *code = OP_BRA;
5993  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
5994    &code, &ptr, &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length);    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
5995      &length);
5996  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
5997    
5998  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 5238  if (re == NULL) Line 6018  if (re == NULL)
6018    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
6019    }    }
6020    
6021  /* 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
6022  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
6023  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
6024  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
6025    pointers. */
6026    
6027  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
6028  re->size = size;  re->size = size;
6029  re->options = cd->external_options;  re->options = cd->external_options;
6030    re->flags = cd->external_flags;
6031  re->dummy1 = 0;  re->dummy1 = 0;
6032  re->first_byte = 0;  re->first_byte = 0;
6033  re->req_byte = 0;  re->req_byte = 0;
# Line 5263  field. Reset the bracket count and the n Line 6045  field. Reset the bracket count and the n
6045  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6046  */  */
6047    
6048    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6049  cd->bracount = 0;  cd->bracount = 0;
6050  cd->names_found = 0;  cd->names_found = 0;
6051  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5270  codestart = cd->name_table + re->name_en Line 6053  codestart = cd->name_table + re->name_en
6053  cd->start_code = codestart;  cd->start_code = codestart;
6054  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6055  cd->req_varyopt = 0;  cd->req_varyopt = 0;
6056  cd->nopartial = FALSE;  cd->had_accept = FALSE;
6057    
6058  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6059  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
6060  of the function here. */  of the function here. */
6061    
6062  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6063  code = (uschar *)codestart;  code = (uschar *)codestart;
6064  *code = OP_BRA;  *code = OP_BRA;
6065  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6066    &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6067  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6068  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6069    re->flags = cd->external_flags;
6070    
6071  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6072    
6073  /* 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. */
6074    
# Line 5325  if (errorcode != 0) Line 6109  if (errorcode != 0)
6109    (pcre_free)(re);    (pcre_free)(re);
6110    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6111    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6112  #ifdef SUPPORT_UTF8    PCRE_EARLY_ERROR_RETURN2:
6113    PCRE_UTF8_ERROR_RETURN:    *errorptr = find_error_text(errorcode);
 #endif  
   *errorptr = error_texts[errorcode];  
6114    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6115    return NULL;    return NULL;
6116    }    }
# Line 5357  if ((re->options & PCRE_ANCHORED) == 0) Line 6139  if ((re->options & PCRE_ANCHORED) == 0)
6139        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6140        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6141           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6142        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6143        }        }
6144      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6145        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6146      }      }
6147    }    }
6148    
# Line 5374  if (reqbyte >= 0 && Line 6156  if (reqbyte >= 0 &&
6156    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6157    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6158      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6159    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6160    }    }
6161    
6162  /* 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 5385  case when building a production library. Line 6167  case when building a production library.
6167  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6168    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6169    
6170  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 " : "");  
   }  
6171    
6172  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6173    {    {
6174    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6175    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5408  if ((re->options & PCRE_FIRSTSET) != 0) Line 6178  if ((re->options & PCRE_FIRSTSET) != 0)
6178      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6179    }    }
6180    
6181  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6182    {    {
6183    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6184    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5417  if ((re->options & PCRE_REQCHSET) != 0) Line 6187  if ((re->options & PCRE_REQCHSET) != 0)
6187      else printf("Req char = \\x%02x%s\n", ch, caseless);      else printf("Req char = \\x%02x%s\n", ch, caseless);
6188    }    }
6189    
6190  pcre_printint(re, stdout);  pcre_printint(re, stdout, TRUE);
6191    
6192  /* This check is done here in the debugging case so that the code that  /* This check is done here in the debugging case so that the code that
6193  was compiled can be seen. */  was compiled can be seen. */
# Line 5425  was compiled can be seen. */ Line 6195  was compiled can be seen. */
6195  if (code - codestart > length)  if (code - codestart > length)
6196    {    {
6197    (pcre_free)(re);    (pcre_free)(re);
6198    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6199    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6200    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6201    return NULL;    return NULL;

Legend:
Removed from v.107  
changed lines
  Added in v.295

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12