/[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 97 by ph10, Mon Mar 5 12:36:47 2007 UTC revision 286 by ph10, Mon Dec 17 14:46:11 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
1383      if (utf8) switch(c)      if (utf8) switch(c)
1384        {        {
1385        case OP_CHAR:        case OP_CHAR:
# Line 1266  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
1404      }      }
1405    }    }
1406  }  }
# Line 1301  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
1476      if (utf8) switch(c)      if (utf8) switch(c)
1477        {        {
1478        case OP_CHAR:        case OP_CHAR:
# Line 1329  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
1497      }      }
1498    }    }
1499  }  }
# Line 1343  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 1366  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 1382  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 1395  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 1461  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 1577  Returns:     a value representing the na Line 1784  Returns:     a value representing the na
1784  static int  static int
1785  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1786  {  {
1787    const char *pn = posix_names;
1788  register int yield = 0;  register int yield = 0;
1789  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1790    {    {
1791    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1792      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1793      pn += posix_name_lengths[yield] + 1;
1794    yield++;    yield++;
1795    }    }
1796  return -1;  return -1;
# Line 1623  adjust_recurse(uschar *group, int adjust Line 1832  adjust_recurse(uschar *group, int adjust
1832    uschar *save_hwm)    uschar *save_hwm)
1833  {  {
1834  uschar *ptr = group;  uschar *ptr = group;
1835    
1836  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1837    {    {
1838    int offset;    int offset;
# Line 1921  if (next >= 0) switch(op_code) Line 2131  if (next >= 0) switch(op_code)
2131    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2132    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2133    
2134      case OP_HSPACE:
2135      case OP_NOT_HSPACE:
2136      switch(next)
2137        {
2138        case 0x09:
2139        case 0x20:
2140        case 0xa0:
2141        case 0x1680:
2142        case 0x180e:
2143        case 0x2000:
2144        case 0x2001:
2145        case 0x2002:
2146        case 0x2003:
2147        case 0x2004:
2148        case 0x2005:
2149        case 0x2006:
2150        case 0x2007:
2151        case 0x2008:
2152        case 0x2009:
2153        case 0x200A:
2154        case 0x202f:
2155        case 0x205f:
2156        case 0x3000:
2157        return op_code != OP_HSPACE;
2158        default:
2159        return op_code == OP_HSPACE;
2160        }
2161    
2162      case OP_VSPACE:
2163      case OP_NOT_VSPACE:
2164      switch(next)
2165        {
2166        case 0x0a:
2167        case 0x0b:
2168        case 0x0c:
2169        case 0x0d:
2170        case 0x85:
2171        case 0x2028:
2172        case 0x2029:
2173        return op_code != OP_VSPACE;
2174        default:
2175        return op_code == OP_VSPACE;
2176        }
2177    
2178    default:    default:
2179    return FALSE;    return FALSE;
2180    }    }
# Line 1955  switch(op_code) Line 2209  switch(op_code)
2209      case ESC_W:      case ESC_W:
2210      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2211    
2212        case ESC_h:
2213        case ESC_H:
2214        switch(item)
2215          {
2216          case 0x09:
2217          case 0x20:
2218          case 0xa0:
2219          case 0x1680:
2220          case 0x180e:
2221          case 0x2000:
2222          case 0x2001:
2223          case 0x2002:
2224          case 0x2003:
2225          case 0x2004:
2226          case 0x2005:
2227          case 0x2006:
2228          case 0x2007:
2229          case 0x2008:
2230          case 0x2009:
2231          case 0x200A:
2232          case 0x202f:
2233          case 0x205f:
2234          case 0x3000:
2235          return -next != ESC_h;
2236          default:
2237          return -next == ESC_h;
2238          }
2239    
2240        case ESC_v:
2241        case ESC_V:
2242        switch(item)
2243          {
2244          case 0x0a:
2245          case 0x0b:
2246          case 0x0c:
2247          case 0x0d:
2248          case 0x85:
2249          case 0x2028:
2250          case 0x2029:
2251          return -next != ESC_v;
2252          default:
2253          return -next == ESC_v;
2254          }
2255    
2256      default:      default:
2257      return FALSE;      return FALSE;
2258      }      }
2259    
2260    case OP_DIGIT:    case OP_DIGIT:
2261    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2262             next == -ESC_h || next == -ESC_v;
2263    
2264    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2265    return next == -ESC_d;    return next == -ESC_d;
# Line 1969  switch(op_code) Line 2268  switch(op_code)
2268    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2269    
2270    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2271    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2272    
2273      case OP_HSPACE:
2274      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2275    
2276      case OP_NOT_HSPACE:
2277      return next == -ESC_h;
2278    
2279      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2280      case OP_VSPACE:
2281      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2282    
2283      case OP_NOT_VSPACE:
2284      return next == -ESC_v;
2285    
2286    case OP_WORDCHAR:    case OP_WORDCHAR:
2287    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2288    
2289    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2290    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2084  req_caseopt = ((options & PCRE_CASELESS) Line 2396  req_caseopt = ((options & PCRE_CASELESS)
2396  for (;; ptr++)  for (;; ptr++)
2397    {    {
2398    BOOL negate_class;    BOOL negate_class;
2399      BOOL should_flip_negation;
2400    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2401    BOOL is_quantifier;    BOOL is_quantifier;
2402    BOOL is_recurse;    BOOL is_recurse;
2403      BOOL reset_bracount;
2404    int class_charcount;    int class_charcount;
2405    int class_lastchar;    int class_lastchar;
2406    int newoptions;    int newoptions;
2407    int recno;    int recno;
2408      int refsign;
2409    int skipbytes;    int skipbytes;
2410    int subreqbyte;    int subreqbyte;
2411    int subfirstbyte;    int subfirstbyte;
# Line 2123  for (;; ptr++) Line 2438  for (;; ptr++)
2438      */      */
2439    
2440      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2441    
2442        /* Paranoid check for integer overflow */
2443    
2444        if (OFLOW_MAX - *lengthptr < code - last_code)
2445          {
2446          *errorcodeptr = ERR20;
2447          goto FAILED;
2448          }
2449    
2450      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2451      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));
2452    
# Line 2235  for (;; ptr++) Line 2559  for (;; ptr++)
2559      *ptrptr = ptr;      *ptrptr = ptr;
2560      if (lengthptr != NULL)      if (lengthptr != NULL)
2561        {        {
2562          if (OFLOW_MAX - *lengthptr < code - last_code)
2563            {
2564            *errorcodeptr = ERR20;
2565            goto FAILED;
2566            }
2567        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2568        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2569        }        }
# Line 2297  for (;; ptr++) Line 2626  for (;; ptr++)
2626        goto FAILED;        goto FAILED;
2627        }        }
2628    
2629      /* 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,
2630        if the first few characters (either before or after ^) are \Q\E or \E we
2631        skip them too. This makes for compatibility with Perl. */
2632    
2633      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2634        for (;;)
2635        {        {
       negate_class = TRUE;  
2636        c = *(++ptr);        c = *(++ptr);
2637          if (c == '\\')
2638            {
2639            if (ptr[1] == 'E') ptr++;
2640              else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2641                else break;
2642            }
2643          else if (!negate_class && c == '^')
2644            negate_class = TRUE;
2645          else break;
2646        }        }
2647      else  
2648        {      /* If a class contains a negative special such as \S, we need to flip the
2649        negate_class = FALSE;      negation flag at the end, so that support for characters > 255 works
2650        }      correctly (they are all included in the class). */
2651    
2652        should_flip_negation = FALSE;
2653    
2654      /* 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
2655      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 2381  for (;; ptr++) Line 2723  for (;; ptr++)
2723          if (*ptr == '^')          if (*ptr == '^')
2724            {            {
2725            local_negate = TRUE;            local_negate = TRUE;
2726              should_flip_negation = TRUE;  /* Note negative special */
2727            ptr++;            ptr++;
2728            }            }
2729    
# Line 2447  for (;; ptr++) Line 2790  for (;; ptr++)
2790        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
2791        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2792        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2793        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
2794        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2795    
2796        if (c == '\\')        if (c == '\\')
# Line 2455  for (;; ptr++) Line 2798  for (;; ptr++)
2798          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2799          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2800    
2801          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = '\b';       /* \b is backspace in a class */
2802          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 */
2803          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 */
2804          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
# Line 2467  for (;; ptr++) Line 2810  for (;; ptr++)
2810            else inescq = TRUE;            else inescq = TRUE;
2811            continue;            continue;
2812            }            }
2813            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2814    
2815          if (c < 0)          if (c < 0)
2816            {            {
# Line 2482  for (;; ptr++) Line 2826  for (;; ptr++)
2826              continue;              continue;
2827    
2828              case ESC_D:              case ESC_D:
2829                should_flip_negation = TRUE;
2830              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2831              continue;              continue;
2832    
# Line 2490  for (;; ptr++) Line 2835  for (;; ptr++)
2835              continue;              continue;
2836    
2837              case ESC_W:              case ESC_W:
2838                should_flip_negation = TRUE;
2839              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2840              continue;              continue;
2841    
# Line 2499  for (;; ptr++) Line 2845  for (;; ptr++)
2845              continue;              continue;
2846    
2847              case ESC_S:              case ESC_S:
2848                should_flip_negation = TRUE;
2849              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2850              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2851              continue;              continue;
2852    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
2853              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
2854              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
2855              }              }
# Line 2515  for (;; ptr++) Line 2859  for (;; ptr++)
2859            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
2860                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
2861    
2862              /* We need to deal with \H, \h, \V, and \v in both phases because
2863              they use extra memory. */
2864    
2865              if (-c == ESC_h)
2866                {
2867                SETBIT(classbits, 0x09); /* VT */
2868                SETBIT(classbits, 0x20); /* SPACE */
2869                SETBIT(classbits, 0xa0); /* NSBP */
2870    #ifdef SUPPORT_UTF8
2871                if (utf8)
2872                  {
2873                  class_utf8 = TRUE;
2874                  *class_utf8data++ = XCL_SINGLE;
2875                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
2876                  *class_utf8data++ = XCL_SINGLE;
2877                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
2878                  *class_utf8data++ = XCL_RANGE;
2879                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
2880                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
2881                  *class_utf8data++ = XCL_SINGLE;
2882                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
2883                  *class_utf8data++ = XCL_SINGLE;
2884                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
2885                  *class_utf8data++ = XCL_SINGLE;
2886                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
2887                  }
2888    #endif
2889                continue;
2890                }
2891    
2892              if (-c == ESC_H)
2893                {
2894                for (c = 0; c < 32; c++)
2895                  {
2896                  int x = 0xff;
2897                  switch (c)
2898                    {
2899                    case 0x09/8: x ^= 1 << (0x09%8); break;
2900                    case 0x20/8: x ^= 1 << (0x20%8); break;
2901                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
2902                    default: break;
2903                    }
2904                  classbits[c] |= x;
2905                  }
2906    
2907    #ifdef SUPPORT_UTF8
2908                if (utf8)
2909                  {
2910                  class_utf8 = TRUE;
2911                  *class_utf8data++ = XCL_RANGE;
2912                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2913                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
2914                  *class_utf8data++ = XCL_RANGE;
2915                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
2916                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
2917                  *class_utf8data++ = XCL_RANGE;
2918                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
2919                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
2920                  *class_utf8data++ = XCL_RANGE;
2921                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
2922                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
2923                  *class_utf8data++ = XCL_RANGE;
2924                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
2925                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
2926                  *class_utf8data++ = XCL_RANGE;
2927                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
2928                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
2929                  *class_utf8data++ = XCL_RANGE;
2930                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
2931                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2932                  }
2933    #endif
2934                continue;
2935                }
2936    
2937              if (-c == ESC_v)
2938                {
2939                SETBIT(classbits, 0x0a); /* LF */
2940                SETBIT(classbits, 0x0b); /* VT */
2941                SETBIT(classbits, 0x0c); /* FF */
2942                SETBIT(classbits, 0x0d); /* CR */
2943                SETBIT(classbits, 0x85); /* NEL */
2944    #ifdef SUPPORT_UTF8
2945                if (utf8)
2946                  {
2947                  class_utf8 = TRUE;
2948                  *class_utf8data++ = XCL_RANGE;
2949                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
2950                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2951                  }
2952    #endif
2953                continue;
2954                }
2955    
2956              if (-c == ESC_V)
2957                {
2958                for (c = 0; c < 32; c++)
2959                  {
2960                  int x = 0xff;
2961                  switch (c)
2962                    {
2963                    case 0x0a/8: x ^= 1 << (0x0a%8);
2964                                 x ^= 1 << (0x0b%8);
2965                                 x ^= 1 << (0x0c%8);
2966                                 x ^= 1 << (0x0d%8);
2967                                 break;
2968                    case 0x85/8: x ^= 1 << (0x85%8); break;
2969                    default: break;
2970                    }
2971                  classbits[c] |= x;
2972                  }
2973    
2974    #ifdef SUPPORT_UTF8
2975                if (utf8)
2976                  {
2977                  class_utf8 = TRUE;
2978                  *class_utf8data++ = XCL_RANGE;
2979                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2980                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
2981                  *class_utf8data++ = XCL_RANGE;
2982                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2983                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2984                  }
2985    #endif
2986                continue;
2987                }
2988    
2989            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
2990    
2991  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2566  for (;; ptr++) Line 3037  for (;; ptr++)
3037    
3038        oldptr = ptr;        oldptr = ptr;
3039    
3040          /* Remember \r or \n */
3041    
3042          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3043    
3044          /* Check for range */
3045    
3046        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3047          {          {
3048          int d;          int d;
# Line 2607  for (;; ptr++) Line 3084  for (;; ptr++)
3084            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3085            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3086    
3087            /* \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
3088            special means the '-' was literal */            special means the '-' was literal */
3089    
3090            if (d < 0)            if (d < 0)
# Line 2633  for (;; ptr++) Line 3110  for (;; ptr++)
3110    
3111          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3112    
3113            /* Remember \r or \n */
3114    
3115            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3116    
3117          /* 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
3118          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3119          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2655  for (;; ptr++) Line 3136  for (;; ptr++)
3136              unsigned int origd = d;              unsigned int origd = d;
3137              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3138                {                {
3139                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3140                      ocd <= (unsigned int)d)
3141                    continue;                          /* Skip embedded ranges */
3142    
3143                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3144                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3145                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3146                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3147                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3148                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3149                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3150                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3151                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3152                  d = ocd;                  d = ocd;
3153                  continue;                  continue;
# Line 2782  for (;; ptr++) Line 3267  for (;; ptr++)
3267        goto FAILED;        goto FAILED;
3268        }        }
3269    
3270    
3271    /* This code has been disabled because it would mean that \s counts as
3272    an explicit \r or \n reference, and that's not really what is wanted. Now
3273    we set the flag only if there is a literal "\r" or "\n" in the class. */
3274    
3275    #if 0
3276        /* Remember whether \r or \n are in this class */
3277    
3278        if (negate_class)
3279          {
3280          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3281          }
3282        else
3283          {
3284          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3285          }
3286    #endif
3287    
3288    
3289      /* 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
3290      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
3291      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
3292      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3293      single-bytes only. This is an historical hangover. Maybe one day we can  
3294      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3295        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3296        operate on single-bytes only. This is an historical hangover. Maybe one day
3297        we can tidy these opcodes to handle multi-byte characters.
3298    
3299      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
3300      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 2797  for (;; ptr++) Line 3304  for (;; ptr++)
3304      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3305    
3306  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3307      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3308            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3309  #else  #else
3310      if (class_charcount == 1)      if (class_charcount == 1)
3311  #endif  #endif
# Line 2843  for (;; ptr++) Line 3348  for (;; ptr++)
3348      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3349    
3350      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3351      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3352      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
3353        the class, so any that were explicitly given as well can be ignored. If
3354        (when there are explicit characters > 255 that must be listed) there are no
3355        characters < 256, we can omit the bitmap in the actual compiled code. */
3356    
3357  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3358      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3359        {        {
3360        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3361        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 2873  for (;; ptr++) Line 3381  for (;; ptr++)
3381        }        }
3382  #endif  #endif
3383    
3384      /* 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
3385      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
3386      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
3387      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3388    
3389        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3390      if (negate_class)      if (negate_class)
3391        {        {
       *code++ = OP_NCLASS;  
3392        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3393          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3394        }        }
3395      else      else
3396        {        {
       *code++ = OP_CLASS;  
3397        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3398        }        }
3399      code += 32;      code += 32;
# Line 3072  for (;; ptr++) Line 3579  for (;; ptr++)
3579        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3580        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3581    
3582        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3583    
3584        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3585    
# Line 3222  for (;; ptr++) Line 3729  for (;; ptr++)
3729        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3730        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3731    
3732        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3733    
3734        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3735          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3258  for (;; ptr++) Line 3765  for (;; ptr++)
3765          goto FAILED;          goto FAILED;
3766          }          }
3767    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3768        /* 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
3769        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
3770        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 3354  for (;; ptr++) Line 3853  for (;; ptr++)
3853          if (repeat_min > 1)          if (repeat_min > 1)
3854            {            {
3855            /* 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
3856            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3857              potential integer overflow. */
3858    
3859            if (lengthptr != NULL)            if (lengthptr != NULL)
3860              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3861                int delta = (repeat_min - 1)*length_prevgroup;
3862                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3863                                                                (double)INT_MAX ||
3864                    OFLOW_MAX - *lengthptr < delta)
3865                  {
3866                  *errorcodeptr = ERR20;
3867                  goto FAILED;
3868                  }
3869                *lengthptr += delta;
3870                }
3871    
3872            /* This is compiling for real */            /* This is compiling for real */
3873    
# Line 3395  for (;; ptr++) Line 3905  for (;; ptr++)
3905          /* 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
3906          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
3907          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
3908          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3909            paranoid checks to avoid integer overflow. */
3910    
3911          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3912            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3913              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3914                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3915              if ((double)repeat_max *
3916                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3917                      > (double)INT_MAX ||
3918                  OFLOW_MAX - *lengthptr < delta)
3919                {
3920                *errorcodeptr = ERR20;
3921                goto FAILED;
3922                }
3923              *lengthptr += delta;
3924              }
3925    
3926          /* This is compiling for real */          /* This is compiling for real */
3927    
# Line 3507  for (;; ptr++) Line 4029  for (;; ptr++)
4029        int len;        int len;
4030        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||
4031            *tempcode == OP_NOTEXACT)            *tempcode == OP_NOTEXACT)
4032          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode] +
4033              ((*tempcode == OP_TYPEEXACT &&
4034                 (tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0);
4035        len = code - tempcode;        len = code - tempcode;
4036        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4037          {          {
# Line 3551  for (;; ptr++) Line 4075  for (;; ptr++)
4075      /* ===================================================================*/      /* ===================================================================*/
4076      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4077      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4078      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. */  
4079    
4080      case '(':      case '(':
4081      newoptions = options;      newoptions = options;
4082      skipbytes = 0;      skipbytes = 0;
4083      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4084      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4085        reset_bracount = FALSE;
4086    
4087        /* First deal with various "verbs" that can be introduced by '*'. */
4088    
4089        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4090          {
4091          int i, namelen;
4092          const char *vn = verbnames;
4093          const uschar *name = ++ptr;
4094          previous = NULL;
4095          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
4096          if (*ptr == ':')
4097            {
4098            *errorcodeptr = ERR59;   /* Not supported */
4099            goto FAILED;
4100            }
4101          if (*ptr != ')')
4102            {
4103            *errorcodeptr = ERR60;
4104            goto FAILED;
4105            }
4106          namelen = ptr - name;
4107          for (i = 0; i < verbcount; i++)
4108            {
4109            if (namelen == verbs[i].len &&
4110                strncmp((char *)name, vn, namelen) == 0)
4111              {
4112              *code = verbs[i].op;
4113              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4114              break;
4115              }
4116            vn += verbs[i].len + 1;
4117            }
4118          if (i < verbcount) continue;
4119          *errorcodeptr = ERR60;
4120          goto FAILED;
4121          }
4122    
4123        /* Deal with the extended parentheses; all are introduced by '?', and the
4124        appearance of any of them means that this is not a capturing group. */
4125    
4126      if (*(++ptr) == '?')      else if (*ptr == '?')
4127        {        {
4128        int i, set, unset, namelen;        int i, set, unset, namelen;
4129        int *optset;        int *optset;
# Line 3582  for (;; ptr++) Line 4144  for (;; ptr++)
4144    
4145    
4146          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4147            case '|':                 /* Reset capture count for each branch */
4148            reset_bracount = TRUE;
4149            /* Fall through */
4150    
4151            /* ------------------------------------------------------------ */
4152          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
4153          bravalue = OP_BRA;          bravalue = OP_BRA;
4154          ptr++;          ptr++;
# Line 3617  for (;; ptr++) Line 4184  for (;; ptr++)
4184    
4185          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
4186          skipbytes = 3;          skipbytes = 3;
4187            refsign = -1;
4188    
4189          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4190    
# Line 3640  for (;; ptr++) Line 4208  for (;; ptr++)
4208            terminator = '\'';            terminator = '\'';
4209            ptr++;            ptr++;
4210            }            }
4211          else terminator = 0;          else
4212              {
4213              terminator = 0;
4214              if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);
4215              }
4216    
4217          /* 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 */
4218    
# Line 3676  for (;; ptr++) Line 4248  for (;; ptr++)
4248          if (lengthptr != NULL) break;          if (lengthptr != NULL) break;
4249    
4250          /* 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
4251          reference. */          reference. If the string started with "+" or "-" we require the rest to
4252            be digits, in which case recno will be set. */
4253    
4254            if (refsign > 0)
4255              {
4256              if (recno <= 0)
4257                {
4258                *errorcodeptr = ERR58;
4259                goto FAILED;
4260                }
4261              recno = (refsign == '-')?
4262                cd->bracount - recno + 1 : recno +cd->bracount;
4263              if (recno <= 0 || recno > cd->final_bracount)
4264                {
4265                *errorcodeptr = ERR15;
4266                goto FAILED;
4267                }
4268              PUT2(code, 2+LINK_SIZE, recno);
4269              break;
4270              }
4271    
4272            /* Otherwise (did not start with "+" or "-"), start by looking for the
4273            name. */
4274    
4275          slot = cd->name_table;          slot = cd->name_table;
4276          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
# Line 3742  for (;; ptr++) Line 4336  for (;; ptr++)
4336            skipbytes = 1;            skipbytes = 1;
4337            }            }
4338    
4339          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4340            in the second pass here, so final_bracount is set. */
4341    
4342          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4343            {            {
4344            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4345            }            }
# Line 3768  for (;; ptr++) Line 4363  for (;; ptr++)
4363    
4364          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4365          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4366          ptr++;          ptr++;
4367            if (*ptr == ')')          /* Optimize (?!) */
4368              {
4369              *code++ = OP_FAIL;
4370              previous = NULL;
4371              continue;
4372              }
4373            bravalue = OP_ASSERT_NOT;
4374          break;          break;
4375    
4376    
# Line 3932  for (;; ptr++) Line 4533  for (;; ptr++)
4533    
4534          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4535          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4536          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4537            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4538            .NET syntax. */
4539    
4540          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4541          name = ++ptr;          name = ++ptr;
# Line 3944  for (;; ptr++) Line 4547  for (;; ptr++)
4547    
4548          if (lengthptr != NULL)          if (lengthptr != NULL)
4549            {            {
4550              if (namelen == 0)
4551                {
4552                *errorcodeptr = ERR62;
4553                goto FAILED;
4554                }
4555            if (*ptr != terminator)            if (*ptr != terminator)
4556              {              {
4557              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 3957  for (;; ptr++) Line 4565  for (;; ptr++)
4565            recno = 0;            recno = 0;
4566            }            }
4567    
4568          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4569            first, and then check that we have reached the end of the name in the
4570            table. That way, if the name that is longer than any in the table,
4571            the comparison will fail without reading beyond the table entry. */
4572    
4573          else          else
4574            {            {
4575            slot = cd->name_table;            slot = cd->name_table;
4576            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4577              {              {
4578              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4579                    slot[2+namelen] == 0)
4580                  break;
4581              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4582              }              }
4583    
# Line 3995  for (;; ptr++) Line 4608  for (;; ptr++)
4608    
4609    
4610          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4611            case '-': case '+':
4612          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */
4613          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case '5': case '6': case '7': case '8': case '9':   /* subroutine */
4614            {            {
4615            const uschar *called;            const uschar *called;
4616    
4617              if ((refsign = *ptr) == '+')
4618                {
4619                ptr++;
4620                if ((digitab[*ptr] & ctype_digit) == 0)
4621                  {
4622                  *errorcodeptr = ERR63;
4623                  goto FAILED;
4624                  }
4625                }
4626              else if (refsign == '-')
4627                {
4628                if ((digitab[ptr[1]] & ctype_digit) == 0)
4629                  goto OTHER_CHAR_AFTER_QUERY;
4630                ptr++;
4631                }
4632    
4633            recno = 0;            recno = 0;
4634            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4635              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - '0';
4636    
4637            if (*ptr != ')')            if (*ptr != ')')
4638              {              {
4639              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4640              goto FAILED;              goto FAILED;
4641              }              }
4642    
4643              if (refsign == '-')
4644                {
4645                if (recno == 0)
4646                  {
4647                  *errorcodeptr = ERR58;
4648                  goto FAILED;
4649                  }
4650                recno = cd->bracount - recno + 1;
4651                if (recno <= 0)
4652                  {
4653                  *errorcodeptr = ERR15;
4654                  goto FAILED;
4655                  }
4656                }
4657              else if (refsign == '+')
4658                {
4659                if (recno == 0)
4660                  {
4661                  *errorcodeptr = ERR58;
4662                  goto FAILED;
4663                  }
4664                recno += cd->bracount;
4665                }
4666    
4667            /* Come here from code above that handles a named recursion */            /* Come here from code above that handles a named recursion */
4668    
4669            HANDLE_RECURSION:            HANDLE_RECURSION:
# Line 4080  for (;; ptr++) Line 4736  for (;; ptr++)
4736    
4737          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4738          default:              /* Other characters: check option setting */          default:              /* Other characters: check option setting */
4739            OTHER_CHAR_AFTER_QUERY:
4740          set = unset = 0;          set = unset = 0;
4741          optset = &set;          optset = &set;
4742    
# Line 4091  for (;; ptr++) Line 4748  for (;; ptr++)
4748    
4749              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4750              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4751              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4752              break;              break;
4753    
4754              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4214  for (;; ptr++) Line 4871  for (;; ptr++)
4871           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
4872           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
4873            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
4874             reset_bracount,               /* True if (?| group */
4875           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
4876           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
4877           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4230  for (;; ptr++) Line 4888  for (;; ptr++)
4888      is on the bracket. */      is on the bracket. */
4889    
4890      /* 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
4891      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
4892        in the real compile phase, not in the pre-pass, where the whole group may
4893        not be available. */
4894    
4895      if (bravalue == OP_COND)      if (bravalue == OP_COND && lengthptr == NULL)
4896        {        {
4897        uschar *tc = code;        uschar *tc = code;
4898        int condcount = 0;        int condcount = 0;
# Line 4279  for (;; ptr++) Line 4939  for (;; ptr++)
4939        goto FAILED;        goto FAILED;
4940        }        }
4941    
4942      /* 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,
4943      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
4944      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
4945      a quantifier. */      duplicated by a quantifier.*/
4946    
4947      if (lengthptr != NULL)      if (lengthptr != NULL)
4948        {        {
4949          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
4950            {
4951            *errorcodeptr = ERR20;
4952            goto FAILED;
4953            }
4954        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4955        code++;        *code++ = OP_BRA;
4956        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4957        *code++ = OP_KET;        *code++ = OP_KET;
4958        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4959          break;    /* No need to waste time with special character handling */
4960        }        }
4961    
4962      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4963    
4964      else code = tempcode;      code = tempcode;
4965    
4966      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4967      relevant. */      relevant. */
# Line 4392  for (;; ptr++) Line 5058  for (;; ptr++)
5058        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
5059        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
5060    
5061        /* \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).
5062          We also support \k{name} (.NET syntax) */
5063    
5064        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\''))        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\'' || ptr[1] == '{'))
5065          {          {
5066          is_recurse = FALSE;          is_recurse = FALSE;
5067          terminator = (*(++ptr) == '<')? '>' : '\'';          terminator = (*(++ptr) == '<')? '>' : (*ptr == '\'')? '\'' : '}';
5068          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
5069          }          }
5070    
# Line 4498  for (;; ptr++) Line 5165  for (;; ptr++)
5165      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5166      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5167    
5168        /* Remember if \r or \n were seen */
5169    
5170        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5171          cd->external_flags |= PCRE_HASCRORLF;
5172    
5173      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5174      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.
5175      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 4563  This function is used during the pre-com Line 5235  This function is used during the pre-com
5235  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
5236  value of lengthptr distinguishes the two phases.  value of lengthptr distinguishes the two phases.
5237    
5238  Argument:  Arguments:
5239    options        option bits, including any changes for this subpattern    options        option bits, including any changes for this subpattern
5240    oldims         previous settings of ims option bits    oldims         previous settings of ims option bits
5241    codeptr        -> the address of the current code pointer    codeptr        -> the address of the current code pointer
5242    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5243    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5244    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5245      reset_bracount TRUE to reset the count for each branch
5246    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5247    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5248    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 4583  Returns:         TRUE on success Line 5256  Returns:         TRUE on success
5256    
5257  static BOOL  static BOOL
5258  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5259    int *errorcodeptr, BOOL lookbehind, int skipbytes, int *firstbyteptr,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5260    int *reqbyteptr, branch_chain *bcptr, compile_data *cd, int *lengthptr)    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5261      int *lengthptr)
5262  {  {
5263  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
5264  uschar *code = *codeptr;  uschar *code = *codeptr;
# Line 4594  uschar *reverse_count = NULL; Line 5268  uschar *reverse_count = NULL;
5268  int firstbyte, reqbyte;  int firstbyte, reqbyte;
5269  int branchfirstbyte, branchreqbyte;  int branchfirstbyte, branchreqbyte;
5270  int length;  int length;
5271    int orig_bracount;
5272    int max_bracount;
5273  branch_chain bc;  branch_chain bc;
5274    
5275  bc.outer = bcptr;  bc.outer = bcptr;
# Line 4622  code += 1 + LINK_SIZE + skipbytes; Line 5298  code += 1 + LINK_SIZE + skipbytes;
5298    
5299  /* Loop for each alternative branch */  /* Loop for each alternative branch */
5300    
5301    orig_bracount = max_bracount = cd->bracount;
5302  for (;;)  for (;;)
5303    {    {
5304      /* For a (?| group, reset the capturing bracket count so that each branch
5305      uses the same numbers. */
5306    
5307      if (reset_bracount) cd->bracount = orig_bracount;
5308    
5309    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5310    
5311    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4653  for (;;) Line 5335  for (;;)
5335      return FALSE;      return FALSE;
5336      }      }
5337    
5338      /* Keep the highest bracket count in case (?| was used and some branch
5339      has fewer than the rest. */
5340    
5341      if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5342    
5343    /* 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. */
5344    
5345    if (lengthptr == NULL)    if (lengthptr == NULL)
# Line 4716  for (;;) Line 5403  for (;;)
5403        }        }
5404      }      }
5405    
5406    /* Reached end of expression, either ')' or end of pattern. Go back through    /* Reached end of expression, either ')' or end of pattern. In the real
5407    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
5408    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
5409    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
5410    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
5411    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,
5412    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.
5413    at the terminating char. */    Return leaving the pointer at the terminating char. */
5414    
5415    if (*ptr != '|')    if (*ptr != '|')
5416      {      {
5417      int branch_length = code - last_branch;      if (lengthptr == NULL)
     do  
5418        {        {
5419        int prev_length = GET(last_branch, 1);        int branch_length = code - last_branch;
5420        PUT(last_branch, 1, branch_length);        do
5421        branch_length = prev_length;          {
5422        last_branch -= branch_length;          int prev_length = GET(last_branch, 1);
5423            PUT(last_branch, 1, branch_length);
5424            branch_length = prev_length;
5425            last_branch -= branch_length;
5426            }
5427          while (branch_length > 0);
5428        }        }
     while (branch_length > 0);  
5429    
5430      /* Fill in the ket */      /* Fill in the ket */
5431    
# Line 4752  for (;;) Line 5442  for (;;)
5442        length += 2;        length += 2;
5443        }        }
5444    
5445        /* Retain the highest bracket number, in case resetting was used. */
5446    
5447        cd->bracount = max_bracount;
5448    
5449      /* Set values to pass back */      /* Set values to pass back */
5450    
5451      *codeptr = code;      *codeptr = code;
5452      *ptrptr = ptr;      *ptrptr = ptr;
5453      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5454      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5455      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5456          {
5457          if (OFLOW_MAX - *lengthptr < length)
5458            {
5459            *errorcodeptr = ERR20;
5460            return FALSE;
5461            }
5462          *lengthptr += length;
5463          }
5464      return TRUE;      return TRUE;
5465      }      }
5466    
5467    /* 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
5468      pointer back to where it was for the start of the first branch. (That is,
5469      pretend that each branch is the only one.)
5470    
5471      In the real compile phase, insert an ALT node. Its length field points back
5472    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
5473    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
5474    zero offset until it is closed, making it possible to detect recursion. */    zero offset until it is closed, making it possible to detect recursion. */
5475    
5476    *code = OP_ALT;    if (lengthptr != NULL)
5477    PUT(code, 1, code - last_branch);      {
5478    bc.current = last_branch = code;      code = *codeptr + 1 + LINK_SIZE + skipbytes;
5479    code += 1 + LINK_SIZE;      length += 1 + LINK_SIZE;
5480        }
5481      else
5482        {
5483        *code = OP_ALT;
5484        PUT(code, 1, code - last_branch);
5485        bc.current = last_branch = code;
5486        code += 1 + LINK_SIZE;
5487        }
5488    
5489    ptr++;    ptr++;
   length += 1 + LINK_SIZE;  
5490    }    }
5491  /* Control never reaches here */  /* Control never reaches here */
5492  }  }
# Line 5039  Returns:        pointer to compiled data Line 5753  Returns:        pointer to compiled data
5753                  with errorptr and erroroffset set                  with errorptr and erroroffset set
5754  */  */
5755    
5756  PCRE_DATA_SCOPE pcre *  PCRE_EXP_DEFN pcre *
5757  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
5758    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
5759  {  {
# Line 5047  return pcre_compile2(pattern, options, N Line 5761  return pcre_compile2(pattern, options, N
5761  }  }
5762    
5763    
5764  PCRE_DATA_SCOPE pcre *  PCRE_EXP_DEFN pcre *
5765  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
5766    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
5767  {  {
# Line 5055  real_pcre *re; Line 5769  real_pcre *re;
5769  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5770  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5771  int errorcode = 0;  int errorcode = 0;
5772    int skipatstart = 0;
5773  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5774  BOOL utf8;  BOOL utf8;
5775  #endif  #endif
# Line 5096  if (errorcodeptr != NULL) *errorcodeptr Line 5811  if (errorcodeptr != NULL) *errorcodeptr
5811  if (erroroffset == NULL)  if (erroroffset == NULL)
5812    {    {
5813    errorcode = ERR16;    errorcode = ERR16;
5814    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN2;
5815    }    }
5816    
5817  *erroroffset = 0;  *erroroffset = 0;
# Line 5109  if (utf8 && (options & PCRE_NO_UTF8_CHEC Line 5824  if (utf8 && (options & PCRE_NO_UTF8_CHEC
5824       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)       (*erroroffset = _pcre_valid_utf8((uschar *)pattern, -1)) >= 0)
5825    {    {
5826    errorcode = ERR44;    errorcode = ERR44;
5827    goto PCRE_UTF8_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN2;
5828    }    }
5829  #else  #else
5830  if ((options & PCRE_UTF8) != 0)  if ((options & PCRE_UTF8) != 0)
# Line 5133  cd->fcc = tables + fcc_offset; Line 5848  cd->fcc = tables + fcc_offset;
5848  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5849  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5850    
5851    /* Check for global one-time settings at the start of the pattern, and remember
5852    the offset for later. */
5853    
5854    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5855      {
5856      int newnl = 0;
5857      int newbsr = 0;
5858    
5859      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5860        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5861      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5862        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5863      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5864        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5865      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5866        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5867      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5868        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5869    
5870      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5871        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5872      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5873        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5874    
5875      if (newnl != 0)
5876        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5877      else if (newbsr != 0)
5878        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5879      else break;
5880      }
5881    
5882    /* Check validity of \R options. */
5883    
5884    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5885      {
5886      case 0:
5887      case PCRE_BSR_ANYCRLF:
5888      case PCRE_BSR_UNICODE:
5889      break;
5890      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5891      }
5892    
5893  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5894  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
5895    "anycrlf". */
5896    
5897  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5898    {    {
5899    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5900    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5901    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5902    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
5903         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;
5904    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
5905      case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
5906    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5907    }    }
5908    
5909  if (newline < 0)  if (newline == -2)
5910      {
5911      cd->nltype = NLTYPE_ANYCRLF;
5912      }
5913    else if (newline < 0)
5914    {    {
5915    cd->nltype = NLTYPE_ANY;    cd->nltype = NLTYPE_ANY;
5916    }    }
# Line 5186  to compile parts of the pattern into; th Line 5949  to compile parts of the pattern into; th
5949  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
5950  is a test for its doing so. */  is a test for its doing so. */
5951    
5952  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
5953  cd->names_found = 0;  cd->names_found = 0;
5954  cd->name_entry_size = 0;  cd->name_entry_size = 0;
5955  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5196  cd->hwm = cworkspace; Line 5959  cd->hwm = cworkspace;
5959  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5960  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5961  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5962  cd->external_options = options;  cd->external_options = options;
5963    cd->external_flags = 0;
5964    
5965  /* 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
5966  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 5205  been put into the cd block so that they Line 5968  been put into the cd block so that they
5968  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5969  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5970    
5971    ptr += skipatstart;
5972  code = cworkspace;  code = cworkspace;
5973  *code = OP_BRA;  *code = OP_BRA;
5974  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
5975    &code, &ptr, &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length);    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
5976      &length);
5977  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
5978    
5979  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 5234  if (re == NULL) Line 5999  if (re == NULL)
5999    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
6000    }    }
6001    
6002  /* 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
6003  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
6004  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
6005  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
6006    pointers. */
6007    
6008  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
6009  re->size = size;  re->size = size;
6010  re->options = cd->external_options;  re->options = cd->external_options;
6011    re->flags = cd->external_flags;
6012  re->dummy1 = 0;  re->dummy1 = 0;
6013  re->first_byte = 0;  re->first_byte = 0;
6014  re->req_byte = 0;  re->req_byte = 0;
# Line 5259  field. Reset the bracket count and the n Line 6026  field. Reset the bracket count and the n
6026  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6027  */  */
6028    
6029    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6030  cd->bracount = 0;  cd->bracount = 0;
6031  cd->names_found = 0;  cd->names_found = 0;
6032  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5266  codestart = cd->name_table + re->name_en Line 6034  codestart = cd->name_table + re->name_en
6034  cd->start_code = codestart;  cd->start_code = codestart;
6035  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6036  cd->req_varyopt = 0;  cd->req_varyopt = 0;
6037  cd->nopartial = FALSE;  cd->had_accept = FALSE;
6038    
6039  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6040  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
6041  of the function here. */  of the function here. */
6042    
6043  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6044  code = (uschar *)codestart;  code = (uschar *)codestart;
6045  *code = OP_BRA;  *code = OP_BRA;
6046  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6047    &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6048  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6049  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6050    re->flags = cd->external_flags;
6051    
6052  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6053    
6054  /* 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. */
6055    
# Line 5321  if (errorcode != 0) Line 6090  if (errorcode != 0)
6090    (pcre_free)(re);    (pcre_free)(re);
6091    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6092    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6093  #ifdef SUPPORT_UTF8    PCRE_EARLY_ERROR_RETURN2:
6094    PCRE_UTF8_ERROR_RETURN:    *errorptr = find_error_text(errorcode);
 #endif  
   *errorptr = error_texts[errorcode];  
6095    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6096    return NULL;    return NULL;
6097    }    }
# Line 5353  if ((re->options & PCRE_ANCHORED) == 0) Line 6120  if ((re->options & PCRE_ANCHORED) == 0)
6120        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6121        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6122           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6123        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6124        }        }
6125      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6126        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6127      }      }
6128    }    }
6129    
# Line 5370  if (reqbyte >= 0 && Line 6137  if (reqbyte >= 0 &&
6137    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6138    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6139      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6140    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6141    }    }
6142    
6143  /* 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 5381  case when building a production library. Line 6148  case when building a production library.
6148  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6149    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6150    
6151  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 " : "");  
   }  
6152    
6153  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6154    {    {
6155    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6156    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5404  if ((re->options & PCRE_FIRSTSET) != 0) Line 6159  if ((re->options & PCRE_FIRSTSET) != 0)
6159      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6160    }    }
6161    
6162  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6163    {    {
6164    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6165    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5413  if ((re->options & PCRE_REQCHSET) != 0) Line 6168  if ((re->options & PCRE_REQCHSET) != 0)
6168      else printf("Req char = \\x%02x%s\n", ch, caseless);      else printf("Req char = \\x%02x%s\n", ch, caseless);
6169    }    }
6170    
6171  pcre_printint(re, stdout);  pcre_printint(re, stdout, TRUE);
6172    
6173  /* 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
6174  was compiled can be seen. */  was compiled can be seen. */
# Line 5421  was compiled can be seen. */ Line 6176  was compiled can be seen. */
6176  if (code - codestart > length)  if (code - codestart > length)
6177    {    {
6178    (pcre_free)(re);    (pcre_free)(re);
6179    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6180    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6181    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6182    return NULL;    return NULL;

Legend:
Removed from v.97  
changed lines
  Added in v.286

  ViewVC Help
Powered by ViewVC 1.1.5