/[pcre]/code/trunk/pcre_compile.c
ViewVC logotype

Diff of /code/trunk/pcre_compile.c

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

revision 172 by ph10, Tue Jun 5 10:40:13 2007 UTC revision 270 by ph10, Fri Nov 16 19:55:16 2007 UTC
# Line 42  POSSIBILITY OF SUCH DAMAGE. Line 42  POSSIBILITY OF SUCH DAMAGE.
42  supporting internal functions that are not used by other modules. */  supporting internal functions that are not used by other modules. */
43    
44    
45    #ifdef HAVE_CONFIG_H
46    #include "config.h"
47    #endif
48    
49  #define NLBLOCK cd             /* Block containing newline information */  #define NLBLOCK cd             /* Block containing newline information */
50  #define PSSTART start_pattern  /* Field containing processed string start */  #define PSSTART start_pattern  /* Field containing processed string start */
51  #define PSEND   end_pattern    /* Field containing processed string end */  #define PSEND   end_pattern    /* Field containing processed string end */
52    
   
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
# Line 58  used by pcretest. DEBUG is not defined w Line 61  used by pcretest. DEBUG is not defined w
61  #endif  #endif
62    
63    
64    /* Macro for setting individual bits in class bitmaps. */
65    
66    #define SETBIT(a,b) a[b/8] |= (1 << (b%8))
67    
68    /* Maximum length value to check against when making sure that the integer that
69    holds the compiled pattern length does not overflow. We make it a bit less than
70    INT_MAX to allow for adding in group terminating bytes, so that we don't have
71    to check them every time. */
72    
73    #define OFLOW_MAX (INT_MAX - 20)
74    
75    
76  /*************************************************  /*************************************************
77  *      Code parameters and static tables         *  *      Code parameters and static tables         *
78  *************************************************/  *************************************************/
# Line 87  static const short int escapes[] = { Line 102  static const short int escapes[] = {
102       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */
103       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */
104     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */
105       0,      0,      0, -ESC_K,      0,      0,      0,      0,   /* H - O */  -ESC_H,      0,      0, -ESC_K,      0,      0,      0,      0,   /* H - O */
106  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0,      0, -ESC_W,   /* P - W */  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0, -ESC_V, -ESC_W,   /* P - W */
107  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */
108     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */
109       0,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */  -ESC_h,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */
110  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0,      0, -ESC_w,   /* p - w */  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0, -ESC_v, -ESC_w,   /* p - w */
111       0,      0, -ESC_z                                            /* x - z */       0,      0, -ESC_z                                            /* x - z */
112  };  };
113    
# Line 106  static const short int escapes[] = { Line 121  static const short int escapes[] = {
121  /*  70 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  70 */     0,     0,      0,       0,      0,     0,      0,      0,
122  /*  78 */     0,   '`',    ':',     '#',    '@',  '\'',    '=',    '"',  /*  78 */     0,   '`',    ':',     '#',    '@',  '\'',    '=',    '"',
123  /*  80 */     0,     7, -ESC_b,       0, -ESC_d, ESC_e,  ESC_f,      0,  /*  80 */     0,     7, -ESC_b,       0, -ESC_d, ESC_e,  ESC_f,      0,
124  /*  88 */     0,     0,      0,     '{',      0,     0,      0,      0,  /*  88 */-ESC_h,     0,      0,     '{',      0,     0,      0,      0,
125  /*  90 */     0,     0, -ESC_k,     'l',      0, ESC_n,      0, -ESC_p,  /*  90 */     0,     0, -ESC_k,     'l',      0, ESC_n,      0, -ESC_p,
126  /*  98 */     0, ESC_r,      0,     '}',      0,     0,      0,      0,  /*  98 */     0, ESC_r,      0,     '}',      0,     0,      0,      0,
127  /*  A0 */     0,   '~', -ESC_s, ESC_tee,      0,     0, -ESC_w,      0,  /*  A0 */     0,   '~', -ESC_s, ESC_tee,      0,-ESC_v, -ESC_w,      0,
128  /*  A8 */     0,-ESC_z,      0,       0,      0,   '[',      0,      0,  /*  A8 */     0,-ESC_z,      0,       0,      0,   '[',      0,      0,
129  /*  B0 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  B0 */     0,     0,      0,       0,      0,     0,      0,      0,
130  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',  /*  B8 */     0,     0,      0,       0,      0,   ']',    '=',    '-',
131  /*  C0 */   '{',-ESC_A, -ESC_B,  -ESC_C, -ESC_D,-ESC_E,      0, -ESC_G,  /*  C0 */   '{',-ESC_A, -ESC_B,  -ESC_C, -ESC_D,-ESC_E,      0, -ESC_G,
132  /*  C8 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  C8 */-ESC_H,     0,      0,       0,      0,     0,      0,      0,
133  /*  D0 */   '}',     0,      0,       0,      0,     0,      0, -ESC_P,  /*  D0 */   '}',     0, -ESC_K,       0,      0,     0,      0, -ESC_P,
134  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,  /*  D8 */-ESC_Q,-ESC_R,      0,       0,      0,     0,      0,      0,
135  /*  E0 */  '\\',     0, -ESC_S,       0,      0,     0, -ESC_W, -ESC_X,  /*  E0 */  '\\',     0, -ESC_S,       0,      0,-ESC_V, -ESC_W, -ESC_X,
136  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,  /*  E8 */     0,-ESC_Z,      0,       0,      0,     0,      0,      0,
137  /*  F0 */     0,     0,      0,       0,      0,     0,      0,      0,  /*  F0 */     0,     0,      0,       0,      0,     0,      0,      0,
138  /*  F8 */     0,     0,      0,       0,      0,     0,      0,      0  /*  F8 */     0,     0,      0,       0,      0,     0,      0,      0
# Line 125  static const short int escapes[] = { Line 140  static const short int escapes[] = {
140  #endif  #endif
141    
142    
143  /* Tables of names of POSIX character classes and their lengths. The list is  /* Table of special "verbs" like (*PRUNE). This is a short table, so it is
144  terminated by a zero length entry. The first three must be alpha, lower, upper,  searched linearly. Put all the names into a single string, in order to reduce
145  as this is assumed for handling case independence. */  the number of relocations when a shared library is dynamically linked. */
146    
147  static const char *const posix_names[] = {  typedef struct verbitem {
148    "alpha", "lower", "upper",    int   len;
149    "alnum", "ascii", "blank", "cntrl", "digit", "graph",    int   op;
150    "print", "punct", "space", "word",  "xdigit" };  } verbitem;
151    
152    static const char verbnames[] =
153      "ACCEPT\0"
154      "COMMIT\0"
155      "F\0"
156      "FAIL\0"
157      "PRUNE\0"
158      "SKIP\0"
159      "THEN";
160    
161    static 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 a braced name or 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"    "(?+ 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 after (?&\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 374  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 400  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 485  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 496  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 525  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 620  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 696  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 725  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 899  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 926  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 1054  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 1143  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 1251  for (;;) Line 1343  for (;;)
1343      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1344      }      }
1345    
1346    /* In UTF-8 mode, opcodes that are followed by a character may be followed by    /* Otherwise, we can get the item's length from the table, except that for
1347    a multi-byte character. The length in the table is a minimum, so we have to    repeated character types, we have to test for \p and \P, which have an extra
1348    arrange to skip the extra bytes. */    two bytes of parameters. */
1349    
1350    else    else
1351      {      {
1352        switch(c)
1353          {
1354          case OP_TYPESTAR:
1355          case OP_TYPEMINSTAR:
1356          case OP_TYPEPLUS:
1357          case OP_TYPEMINPLUS:
1358          case OP_TYPEQUERY:
1359          case OP_TYPEMINQUERY:
1360          case OP_TYPEPOSSTAR:
1361          case OP_TYPEPOSPLUS:
1362          case OP_TYPEPOSQUERY:
1363          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1364          break;
1365    
1366          case OP_TYPEUPTO:
1367          case OP_TYPEMINUPTO:
1368          case OP_TYPEEXACT:
1369          case OP_TYPEPOSUPTO:
1370          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1371          break;
1372          }
1373    
1374        /* Add in the fixed length from the table */
1375    
1376      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1377    
1378      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1379      a multi-byte character. The length in the table is a minimum, so we have to
1380      arrange to skip the extra bytes. */
1381    
1382  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1383      if (utf8) switch(c)      if (utf8) switch(c)
1384        {        {
# Line 1315  for (;;) Line 1436  for (;;)
1436    
1437    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1438    
1439    /* Otherwise, we get the item's length from the table. In UTF-8 mode, opcodes    /* Otherwise, we can get the item's length from the table, except that for
1440    that are followed by a character may be followed by a multi-byte character.    repeated character types, we have to test for \p and \P, which have an extra
1441    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1442    
1443    else    else
1444      {      {
1445        switch(c)
1446          {
1447          case OP_TYPESTAR:
1448          case OP_TYPEMINSTAR:
1449          case OP_TYPEPLUS:
1450          case OP_TYPEMINPLUS:
1451          case OP_TYPEQUERY:
1452          case OP_TYPEMINQUERY:
1453          case OP_TYPEPOSSTAR:
1454          case OP_TYPEPOSPLUS:
1455          case OP_TYPEPOSQUERY:
1456          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1457          break;
1458    
1459          case OP_TYPEPOSUPTO:
1460          case OP_TYPEUPTO:
1461          case OP_TYPEMINUPTO:
1462          case OP_TYPEEXACT:
1463          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1464          break;
1465          }
1466    
1467        /* Add in the fixed length from the table */
1468    
1469      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1470    
1471        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1472        by a multi-byte character. The length in the table is a minimum, so we have
1473        to arrange to skip the extra bytes. */
1474    
1475  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1476      if (utf8) switch(c)      if (utf8) switch(c)
1477        {        {
# Line 1394  for (code = first_significant_code(code Line 1543  for (code = first_significant_code(code
1543    
1544    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1545    
1546    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1547      {      {
1548      BOOL empty_branch;      BOOL empty_branch;
1549      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 1418  for (code = first_significant_code(code Line 1567  for (code = first_significant_code(code
1567    
1568    switch (c)    switch (c)
1569      {      {
1570      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1571        cannot be represented just by a bit map. This includes negated single
1572        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1573        actual length is stored in the compiled code, so we must update "code"
1574        here. */
1575    
1576  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1577      case OP_XCLASS:      case OP_XCLASS:
1578      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1579      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1580  #endif  #endif
1581    
# Line 1484  for (code = first_significant_code(code Line 1637  for (code = first_significant_code(code
1637      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1638      return FALSE;      return FALSE;
1639    
1640        /* These are going to continue, as they may be empty, but we have to
1641        fudge the length for the \p and \P cases. */
1642    
1643        case OP_TYPESTAR:
1644        case OP_TYPEMINSTAR:
1645        case OP_TYPEPOSSTAR:
1646        case OP_TYPEQUERY:
1647        case OP_TYPEMINQUERY:
1648        case OP_TYPEPOSQUERY:
1649        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1650        break;
1651    
1652        /* Same for these */
1653    
1654        case OP_TYPEUPTO:
1655        case OP_TYPEMINUPTO:
1656        case OP_TYPEPOSUPTO:
1657        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1658        break;
1659    
1660      /* End of branch */      /* End of branch */
1661    
1662      case OP_KET:      case OP_KET:
# Line 1600  Returns:     a value representing the na Line 1773  Returns:     a value representing the na
1773  static int  static int
1774  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1775  {  {
1776    const char *pn = posix_names;
1777  register int yield = 0;  register int yield = 0;
1778  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1779    {    {
1780    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1781      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1782      pn += posix_name_lengths[yield] + 1;
1783    yield++;    yield++;
1784    }    }
1785  return -1;  return -1;
# Line 1646  adjust_recurse(uschar *group, int adjust Line 1821  adjust_recurse(uschar *group, int adjust
1821    uschar *save_hwm)    uschar *save_hwm)
1822  {  {
1823  uschar *ptr = group;  uschar *ptr = group;
1824    
1825  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1826    {    {
1827    int offset;    int offset;
# Line 1944  if (next >= 0) switch(op_code) Line 2120  if (next >= 0) switch(op_code)
2120    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2121    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2122    
2123      case OP_HSPACE:
2124      case OP_NOT_HSPACE:
2125      switch(next)
2126        {
2127        case 0x09:
2128        case 0x20:
2129        case 0xa0:
2130        case 0x1680:
2131        case 0x180e:
2132        case 0x2000:
2133        case 0x2001:
2134        case 0x2002:
2135        case 0x2003:
2136        case 0x2004:
2137        case 0x2005:
2138        case 0x2006:
2139        case 0x2007:
2140        case 0x2008:
2141        case 0x2009:
2142        case 0x200A:
2143        case 0x202f:
2144        case 0x205f:
2145        case 0x3000:
2146        return op_code != OP_HSPACE;
2147        default:
2148        return op_code == OP_HSPACE;
2149        }
2150    
2151      case OP_VSPACE:
2152      case OP_NOT_VSPACE:
2153      switch(next)
2154        {
2155        case 0x0a:
2156        case 0x0b:
2157        case 0x0c:
2158        case 0x0d:
2159        case 0x85:
2160        case 0x2028:
2161        case 0x2029:
2162        return op_code != OP_VSPACE;
2163        default:
2164        return op_code == OP_VSPACE;
2165        }
2166    
2167    default:    default:
2168    return FALSE;    return FALSE;
2169    }    }
# Line 1978  switch(op_code) Line 2198  switch(op_code)
2198      case ESC_W:      case ESC_W:
2199      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2200    
2201        case ESC_h:
2202        case ESC_H:
2203        switch(item)
2204          {
2205          case 0x09:
2206          case 0x20:
2207          case 0xa0:
2208          case 0x1680:
2209          case 0x180e:
2210          case 0x2000:
2211          case 0x2001:
2212          case 0x2002:
2213          case 0x2003:
2214          case 0x2004:
2215          case 0x2005:
2216          case 0x2006:
2217          case 0x2007:
2218          case 0x2008:
2219          case 0x2009:
2220          case 0x200A:
2221          case 0x202f:
2222          case 0x205f:
2223          case 0x3000:
2224          return -next != ESC_h;
2225          default:
2226          return -next == ESC_h;
2227          }
2228    
2229        case ESC_v:
2230        case ESC_V:
2231        switch(item)
2232          {
2233          case 0x0a:
2234          case 0x0b:
2235          case 0x0c:
2236          case 0x0d:
2237          case 0x85:
2238          case 0x2028:
2239          case 0x2029:
2240          return -next != ESC_v;
2241          default:
2242          return -next == ESC_v;
2243          }
2244    
2245      default:      default:
2246      return FALSE;      return FALSE;
2247      }      }
2248    
2249    case OP_DIGIT:    case OP_DIGIT:
2250    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2251             next == -ESC_h || next == -ESC_v;
2252    
2253    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2254    return next == -ESC_d;    return next == -ESC_d;
# Line 1992  switch(op_code) Line 2257  switch(op_code)
2257    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2258    
2259    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2260    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2261    
2262      case OP_HSPACE:
2263      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2264    
2265      case OP_NOT_HSPACE:
2266      return next == -ESC_h;
2267    
2268      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2269      case OP_VSPACE:
2270      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2271    
2272      case OP_NOT_VSPACE:
2273      return next == -ESC_v;
2274    
2275    case OP_WORDCHAR:    case OP_WORDCHAR:
2276    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2277    
2278    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2279    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2107  req_caseopt = ((options & PCRE_CASELESS) Line 2385  req_caseopt = ((options & PCRE_CASELESS)
2385  for (;; ptr++)  for (;; ptr++)
2386    {    {
2387    BOOL negate_class;    BOOL negate_class;
2388      BOOL should_flip_negation;
2389    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2390    BOOL is_quantifier;    BOOL is_quantifier;
2391    BOOL is_recurse;    BOOL is_recurse;
2392      BOOL reset_bracount;
2393    int class_charcount;    int class_charcount;
2394    int class_lastchar;    int class_lastchar;
2395    int newoptions;    int newoptions;
# Line 2147  for (;; ptr++) Line 2427  for (;; ptr++)
2427      */      */
2428    
2429      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2430    
2431        /* Paranoid check for integer overflow */
2432    
2433        if (OFLOW_MAX - *lengthptr < code - last_code)
2434          {
2435          *errorcodeptr = ERR20;
2436          goto FAILED;
2437          }
2438    
2439      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2440      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));
2441    
# Line 2259  for (;; ptr++) Line 2548  for (;; ptr++)
2548      *ptrptr = ptr;      *ptrptr = ptr;
2549      if (lengthptr != NULL)      if (lengthptr != NULL)
2550        {        {
2551          if (OFLOW_MAX - *lengthptr < code - last_code)
2552            {
2553            *errorcodeptr = ERR20;
2554            goto FAILED;
2555            }
2556        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2557        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2558        }        }
# Line 2321  for (;; ptr++) Line 2615  for (;; ptr++)
2615        goto FAILED;        goto FAILED;
2616        }        }
2617    
2618      /* 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,
2619        if the first few characters (either before or after ^) are \Q\E or \E we
2620        skip them too. This makes for compatibility with Perl. */
2621    
2622      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2623        for (;;)
2624        {        {
       negate_class = TRUE;  
2625        c = *(++ptr);        c = *(++ptr);
2626          if (c == '\\')
2627            {
2628            if (ptr[1] == 'E') ptr++;
2629              else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2630                else break;
2631            }
2632          else if (!negate_class && c == '^')
2633            negate_class = TRUE;
2634          else break;
2635        }        }
2636      else  
2637        {      /* If a class contains a negative special such as \S, we need to flip the
2638        negate_class = FALSE;      negation flag at the end, so that support for characters > 255 works
2639        }      correctly (they are all included in the class). */
2640    
2641        should_flip_negation = FALSE;
2642    
2643      /* 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
2644      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 2405  for (;; ptr++) Line 2712  for (;; ptr++)
2712          if (*ptr == '^')          if (*ptr == '^')
2713            {            {
2714            local_negate = TRUE;            local_negate = TRUE;
2715              should_flip_negation = TRUE;  /* Note negative special */
2716            ptr++;            ptr++;
2717            }            }
2718    
# Line 2471  for (;; ptr++) Line 2779  for (;; ptr++)
2779        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
2780        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2781        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2782        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
2783        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2784    
2785        if (c == '\\')        if (c == '\\')
# Line 2491  for (;; ptr++) Line 2799  for (;; ptr++)
2799            else inescq = TRUE;            else inescq = TRUE;
2800            continue;            continue;
2801            }            }
2802            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2803    
2804          if (c < 0)          if (c < 0)
2805            {            {
# Line 2506  for (;; ptr++) Line 2815  for (;; ptr++)
2815              continue;              continue;
2816    
2817              case ESC_D:              case ESC_D:
2818                should_flip_negation = TRUE;
2819              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2820              continue;              continue;
2821    
# Line 2514  for (;; ptr++) Line 2824  for (;; ptr++)
2824              continue;              continue;
2825    
2826              case ESC_W:              case ESC_W:
2827                should_flip_negation = TRUE;
2828              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2829              continue;              continue;
2830    
# Line 2523  for (;; ptr++) Line 2834  for (;; ptr++)
2834              continue;              continue;
2835    
2836              case ESC_S:              case ESC_S:
2837                should_flip_negation = TRUE;
2838              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2839              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2840              continue;              continue;
# Line 2539  for (;; ptr++) Line 2851  for (;; ptr++)
2851            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
2852                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
2853    
2854              /* We need to deal with \H, \h, \V, and \v in both phases because
2855              they use extra memory. */
2856    
2857              if (-c == ESC_h)
2858                {
2859                SETBIT(classbits, 0x09); /* VT */
2860                SETBIT(classbits, 0x20); /* SPACE */
2861                SETBIT(classbits, 0xa0); /* NSBP */
2862    #ifdef SUPPORT_UTF8
2863                if (utf8)
2864                  {
2865                  class_utf8 = TRUE;
2866                  *class_utf8data++ = XCL_SINGLE;
2867                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
2868                  *class_utf8data++ = XCL_SINGLE;
2869                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
2870                  *class_utf8data++ = XCL_RANGE;
2871                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
2872                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
2873                  *class_utf8data++ = XCL_SINGLE;
2874                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
2875                  *class_utf8data++ = XCL_SINGLE;
2876                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
2877                  *class_utf8data++ = XCL_SINGLE;
2878                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
2879                  }
2880    #endif
2881                continue;
2882                }
2883    
2884              if (-c == ESC_H)
2885                {
2886                for (c = 0; c < 32; c++)
2887                  {
2888                  int x = 0xff;
2889                  switch (c)
2890                    {
2891                    case 0x09/8: x ^= 1 << (0x09%8); break;
2892                    case 0x20/8: x ^= 1 << (0x20%8); break;
2893                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
2894                    default: break;
2895                    }
2896                  classbits[c] |= x;
2897                  }
2898    
2899    #ifdef SUPPORT_UTF8
2900                if (utf8)
2901                  {
2902                  class_utf8 = TRUE;
2903                  *class_utf8data++ = XCL_RANGE;
2904                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2905                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
2906                  *class_utf8data++ = XCL_RANGE;
2907                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
2908                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
2909                  *class_utf8data++ = XCL_RANGE;
2910                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
2911                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
2912                  *class_utf8data++ = XCL_RANGE;
2913                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
2914                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
2915                  *class_utf8data++ = XCL_RANGE;
2916                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
2917                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
2918                  *class_utf8data++ = XCL_RANGE;
2919                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
2920                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
2921                  *class_utf8data++ = XCL_RANGE;
2922                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
2923                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2924                  }
2925    #endif
2926                continue;
2927                }
2928    
2929              if (-c == ESC_v)
2930                {
2931                SETBIT(classbits, 0x0a); /* LF */
2932                SETBIT(classbits, 0x0b); /* VT */
2933                SETBIT(classbits, 0x0c); /* FF */
2934                SETBIT(classbits, 0x0d); /* CR */
2935                SETBIT(classbits, 0x85); /* NEL */
2936    #ifdef SUPPORT_UTF8
2937                if (utf8)
2938                  {
2939                  class_utf8 = TRUE;
2940                  *class_utf8data++ = XCL_RANGE;
2941                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
2942                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2943                  }
2944    #endif
2945                continue;
2946                }
2947    
2948              if (-c == ESC_V)
2949                {
2950                for (c = 0; c < 32; c++)
2951                  {
2952                  int x = 0xff;
2953                  switch (c)
2954                    {
2955                    case 0x0a/8: x ^= 1 << (0x0a%8);
2956                                 x ^= 1 << (0x0b%8);
2957                                 x ^= 1 << (0x0c%8);
2958                                 x ^= 1 << (0x0d%8);
2959                                 break;
2960                    case 0x85/8: x ^= 1 << (0x85%8); break;
2961                    default: break;
2962                    }
2963                  classbits[c] |= x;
2964                  }
2965    
2966    #ifdef SUPPORT_UTF8
2967                if (utf8)
2968                  {
2969                  class_utf8 = TRUE;
2970                  *class_utf8data++ = XCL_RANGE;
2971                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2972                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
2973                  *class_utf8data++ = XCL_RANGE;
2974                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2975                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2976                  }
2977    #endif
2978                continue;
2979                }
2980    
2981            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
2982    
2983  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2590  for (;; ptr++) Line 3029  for (;; ptr++)
3029    
3030        oldptr = ptr;        oldptr = ptr;
3031    
3032          /* Remember \r or \n */
3033    
3034          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3035    
3036          /* Check for range */
3037    
3038        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3039          {          {
3040          int d;          int d;
# Line 2657  for (;; ptr++) Line 3102  for (;; ptr++)
3102    
3103          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3104    
3105            /* Remember \r or \n */
3106    
3107            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3108    
3109          /* 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
3110          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3111          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2679  for (;; ptr++) Line 3128  for (;; ptr++)
3128              unsigned int origd = d;              unsigned int origd = d;
3129              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3130                {                {
3131                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3132                      ocd <= (unsigned int)d)
3133                    continue;                          /* Skip embedded ranges */
3134    
3135                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3136                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3137                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3138                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3139                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3140                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3141                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3142                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3143                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3144                  d = ocd;                  d = ocd;
3145                  continue;                  continue;
# Line 2806  for (;; ptr++) Line 3259  for (;; ptr++)
3259        goto FAILED;        goto FAILED;
3260        }        }
3261    
3262    
3263    /* This code has been disabled because it would mean that \s counts as
3264    an explicit \r or \n reference, and that's not really what is wanted. Now
3265    we set the flag only if there is a literal "\r" or "\n" in the class. */
3266    
3267    #if 0
3268        /* Remember whether \r or \n are in this class */
3269    
3270        if (negate_class)
3271          {
3272          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3273          }
3274        else
3275          {
3276          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3277          }
3278    #endif
3279    
3280    
3281      /* 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
3282      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
3283      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
3284      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3285      single-bytes only. This is an historical hangover. Maybe one day we can  
3286      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3287        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3288        operate on single-bytes only. This is an historical hangover. Maybe one day
3289        we can tidy these opcodes to handle multi-byte characters.
3290    
3291      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
3292      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note      1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note
# Line 2821  for (;; ptr++) Line 3296  for (;; ptr++)
3296      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3297    
3298  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3299      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3300            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3301  #else  #else
3302      if (class_charcount == 1)      if (class_charcount == 1)
3303  #endif  #endif
# Line 2867  for (;; ptr++) Line 3340  for (;; ptr++)
3340      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3341    
3342      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3343      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3344      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
3345        the class, so any that were explicitly given as well can be ignored. If
3346        (when there are explicit characters > 255 that must be listed) there are no
3347        characters < 256, we can omit the bitmap in the actual compiled code. */
3348    
3349  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3350      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3351        {        {
3352        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3353        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 2897  for (;; ptr++) Line 3373  for (;; ptr++)
3373        }        }
3374  #endif  #endif
3375    
3376      /* 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
3377      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
3378      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
3379      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3380    
3381        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3382      if (negate_class)      if (negate_class)
3383        {        {
       *code++ = OP_NCLASS;  
3384        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3385          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3386        }        }
3387      else      else
3388        {        {
       *code++ = OP_CLASS;  
3389        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3390        }        }
3391      code += 32;      code += 32;
# Line 3096  for (;; ptr++) Line 3571  for (;; ptr++)
3571        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3572        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3573    
3574        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3575    
3576        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3577    
# Line 3246  for (;; ptr++) Line 3721  for (;; ptr++)
3721        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3722        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3723    
3724        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3725    
3726        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3727          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3282  for (;; ptr++) Line 3757  for (;; ptr++)
3757          goto FAILED;          goto FAILED;
3758          }          }
3759    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3760        /* 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
3761        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
3762        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 3378  for (;; ptr++) Line 3845  for (;; ptr++)
3845          if (repeat_min > 1)          if (repeat_min > 1)
3846            {            {
3847            /* 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
3848            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3849              potential integer overflow. */
3850    
3851            if (lengthptr != NULL)            if (lengthptr != NULL)
3852              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3853                int delta = (repeat_min - 1)*length_prevgroup;
3854                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3855                                                                (double)INT_MAX ||
3856                    OFLOW_MAX - *lengthptr < delta)
3857                  {
3858                  *errorcodeptr = ERR20;
3859                  goto FAILED;
3860                  }
3861                *lengthptr += delta;
3862                }
3863    
3864            /* This is compiling for real */            /* This is compiling for real */
3865    
# Line 3419  for (;; ptr++) Line 3897  for (;; ptr++)
3897          /* 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
3898          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
3899          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
3900          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3901            paranoid checks to avoid integer overflow. */
3902    
3903          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3904            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3905              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3906                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3907              if ((double)repeat_max *
3908                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3909                      > (double)INT_MAX ||
3910                  OFLOW_MAX - *lengthptr < delta)
3911                {
3912                *errorcodeptr = ERR20;
3913                goto FAILED;
3914                }
3915              *lengthptr += delta;
3916              }
3917    
3918          /* This is compiling for real */          /* This is compiling for real */
3919    
# Line 3575  for (;; ptr++) Line 4065  for (;; ptr++)
4065      /* ===================================================================*/      /* ===================================================================*/
4066      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4067      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4068      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. */  
4069    
4070      case '(':      case '(':
4071      newoptions = options;      newoptions = options;
4072      skipbytes = 0;      skipbytes = 0;
4073      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4074      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4075        reset_bracount = FALSE;
4076    
4077      if (*(++ptr) == '?')      /* First deal with various "verbs" that can be introduced by '*'. */
4078    
4079        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4080          {
4081          int i, namelen;
4082          const char *vn = verbnames;
4083          const uschar *name = ++ptr;
4084          previous = NULL;
4085          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
4086          if (*ptr == ':')
4087            {
4088            *errorcodeptr = ERR59;   /* Not supported */
4089            goto FAILED;
4090            }
4091          if (*ptr != ')')
4092            {
4093            *errorcodeptr = ERR60;
4094            goto FAILED;
4095            }
4096          namelen = ptr - name;
4097          for (i = 0; i < verbcount; i++)
4098            {
4099            if (namelen == verbs[i].len &&
4100                strncmp((char *)name, vn, namelen) == 0)
4101              {
4102              *code = verbs[i].op;
4103              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4104              break;
4105              }
4106            vn += verbs[i].len + 1;
4107            }
4108          if (i < verbcount) continue;
4109          *errorcodeptr = ERR60;
4110          goto FAILED;
4111          }
4112    
4113        /* Deal with the extended parentheses; all are introduced by '?', and the
4114        appearance of any of them means that this is not a capturing group. */
4115    
4116        else if (*ptr == '?')
4117        {        {
4118        int i, set, unset, namelen;        int i, set, unset, namelen;
4119        int *optset;        int *optset;
# Line 3606  for (;; ptr++) Line 4134  for (;; ptr++)
4134    
4135    
4136          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4137            case '|':                 /* Reset capture count for each branch */
4138            reset_bracount = TRUE;
4139            /* Fall through */
4140    
4141            /* ------------------------------------------------------------ */
4142          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
4143          bravalue = OP_BRA;          bravalue = OP_BRA;
4144          ptr++;          ptr++;
# Line 3715  for (;; ptr++) Line 4248  for (;; ptr++)
4248              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4249              goto FAILED;              goto FAILED;
4250              }              }
4251            if (refsign == '-')            recno = (refsign == '-')?
4252                cd->bracount - recno + 1 : recno +cd->bracount;
4253              if (recno <= 0 || recno > cd->final_bracount)
4254              {              {
4255              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4256              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4257              }              }
           else recno += cd->bracount;  
4258            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4259            break;            break;
4260            }            }
# Line 3796  for (;; ptr++) Line 4326  for (;; ptr++)
4326            skipbytes = 1;            skipbytes = 1;
4327            }            }
4328    
4329          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4330            in the second pass here, so final_bracount is set. */
4331    
4332          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4333            {            {
4334            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4335            }            }
# Line 3822  for (;; ptr++) Line 4353  for (;; ptr++)
4353    
4354          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4355          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4356          ptr++;          ptr++;
4357            if (*ptr == ')')          /* Optimize (?!) */
4358              {
4359              *code++ = OP_FAIL;
4360              previous = NULL;
4361              continue;
4362              }
4363            bravalue = OP_ASSERT_NOT;
4364          break;          break;
4365    
4366    
# Line 3998  for (;; ptr++) Line 4535  for (;; ptr++)
4535    
4536          if (lengthptr != NULL)          if (lengthptr != NULL)
4537            {            {
4538              if (namelen == 0)
4539                {
4540                *errorcodeptr = ERR62;
4541                goto FAILED;
4542                }
4543            if (*ptr != terminator)            if (*ptr != terminator)
4544              {              {
4545              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4011  for (;; ptr++) Line 4553  for (;; ptr++)
4553            recno = 0;            recno = 0;
4554            }            }
4555    
4556          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4557            first, and then check that we have reached the end of the name in the
4558            table. That way, if the name that is longer than any in the table,
4559            the comparison will fail without reading beyond the table entry. */
4560    
4561          else          else
4562            {            {
4563            slot = cd->name_table;            slot = cd->name_table;
4564            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4565              {              {
4566              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4567                    slot[2+namelen] == 0)
4568                  break;
4569              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4570              }              }
4571    
# Line 4055  for (;; ptr++) Line 4602  for (;; ptr++)
4602            {            {
4603            const uschar *called;            const uschar *called;
4604    
4605            if ((refsign = *ptr) == '+') ptr++;            if ((refsign = *ptr) == '+')
4606                {
4607                ptr++;
4608                if ((digitab[*ptr] & ctype_digit) == 0)
4609                  {
4610                  *errorcodeptr = ERR63;
4611                  goto FAILED;
4612                  }
4613                }
4614            else if (refsign == '-')            else if (refsign == '-')
4615              {              {
4616              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
# Line 4181  for (;; ptr++) Line 4736  for (;; ptr++)
4736    
4737              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4738              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4739              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4740              break;              break;
4741    
4742              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4304  for (;; ptr++) Line 4859  for (;; ptr++)
4859           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
4860           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
4861            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
4862             reset_bracount,               /* True if (?| group */
4863           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
4864           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
4865           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4371  for (;; ptr++) Line 4927  for (;; ptr++)
4927        goto FAILED;        goto FAILED;
4928        }        }
4929    
4930      /* 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,
4931      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
4932      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
4933      a quantifier. */      duplicated by a quantifier.*/
4934    
4935      if (lengthptr != NULL)      if (lengthptr != NULL)
4936        {        {
4937          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
4938            {
4939            *errorcodeptr = ERR20;
4940            goto FAILED;
4941            }
4942        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4943        code++;        *code++ = OP_BRA;
4944        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4945        *code++ = OP_KET;        *code++ = OP_KET;
4946        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4947          break;    /* No need to waste time with special character handling */
4948        }        }
4949    
4950      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4951    
4952      else code = tempcode;      code = tempcode;
4953    
4954      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4955      relevant. */      relevant. */
# Line 4591  for (;; ptr++) Line 5153  for (;; ptr++)
5153      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5154      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5155    
5156        /* Remember if \r or \n were seen */
5157    
5158        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5159          cd->external_flags |= PCRE_HASCRORLF;
5160    
5161      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5162      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.
5163      Otherwise, leave the firstbyte value alone, and don't change it on a zero      Otherwise, leave the firstbyte value alone, and don't change it on a zero
# Line 4663  Arguments: Line 5230  Arguments:
5230    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5231    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5232    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5233      reset_bracount TRUE to reset the count for each branch
5234    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5235    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5236    reqbyteptr     place to put the last required character, or a negative number    reqbyteptr     place to put the last required character, or a negative number
# Line 4676  Returns:         TRUE on success Line 5244  Returns:         TRUE on success
5244    
5245  static BOOL  static BOOL
5246  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5247    int *errorcodeptr, BOOL lookbehind, int skipbytes, int *firstbyteptr,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5248    int *reqbyteptr, branch_chain *bcptr, compile_data *cd, int *lengthptr)    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5249      int *lengthptr)
5250  {  {
5251  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
5252  uschar *code = *codeptr;  uschar *code = *codeptr;
# Line 4687  uschar *reverse_count = NULL; Line 5256  uschar *reverse_count = NULL;
5256  int firstbyte, reqbyte;  int firstbyte, reqbyte;
5257  int branchfirstbyte, branchreqbyte;  int branchfirstbyte, branchreqbyte;
5258  int length;  int length;
5259    int orig_bracount;
5260    int max_bracount;
5261  branch_chain bc;  branch_chain bc;
5262    
5263  bc.outer = bcptr;  bc.outer = bcptr;
# Line 4715  code += 1 + LINK_SIZE + skipbytes; Line 5286  code += 1 + LINK_SIZE + skipbytes;
5286    
5287  /* Loop for each alternative branch */  /* Loop for each alternative branch */
5288    
5289    orig_bracount = max_bracount = cd->bracount;
5290  for (;;)  for (;;)
5291    {    {
5292      /* For a (?| group, reset the capturing bracket count so that each branch
5293      uses the same numbers. */
5294    
5295      if (reset_bracount) cd->bracount = orig_bracount;
5296    
5297    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5298    
5299    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4746  for (;;) Line 5323  for (;;)
5323      return FALSE;      return FALSE;
5324      }      }
5325    
5326      /* Keep the highest bracket count in case (?| was used and some branch
5327      has fewer than the rest. */
5328    
5329      if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5330    
5331    /* 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. */
5332    
5333    if (lengthptr == NULL)    if (lengthptr == NULL)
# Line 4848  for (;;) Line 5430  for (;;)
5430        length += 2;        length += 2;
5431        }        }
5432    
5433        /* Retain the highest bracket number, in case resetting was used. */
5434    
5435        cd->bracount = max_bracount;
5436    
5437      /* Set values to pass back */      /* Set values to pass back */
5438    
5439      *codeptr = code;      *codeptr = code;
5440      *ptrptr = ptr;      *ptrptr = ptr;
5441      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5442      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5443      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5444          {
5445          if (OFLOW_MAX - *lengthptr < length)
5446            {
5447            *errorcodeptr = ERR20;
5448            return FALSE;
5449            }
5450          *lengthptr += length;
5451          }
5452      return TRUE;      return TRUE;
5453      }      }
5454    
# Line 5163  real_pcre *re; Line 5757  real_pcre *re;
5757  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5758  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5759  int errorcode = 0;  int errorcode = 0;
5760    int skipatstart = 0;
5761  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5762  BOOL utf8;  BOOL utf8;
5763  #endif  #endif
# Line 5241  cd->fcc = tables + fcc_offset; Line 5836  cd->fcc = tables + fcc_offset;
5836  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5837  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5838    
5839    /* Check for global one-time settings at the start of the pattern, and remember
5840    the offset for later. */
5841    
5842    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5843      {
5844      int newnl = 0;
5845      int newbsr = 0;
5846    
5847      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5848        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5849      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5850        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5851      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5852        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5853      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5854        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5855      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5856        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5857    
5858      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5859        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5860      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5861        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5862    
5863      if (newnl != 0)
5864        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5865      else if (newbsr != 0)
5866        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5867      else break;
5868      }
5869    
5870    /* Check validity of \R options. */
5871    
5872    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5873      {
5874      case 0:
5875      case PCRE_BSR_ANYCRLF:
5876      case PCRE_BSR_UNICODE:
5877      break;
5878      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5879      }
5880    
5881  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5882  current code allows for fixed one- or two-byte sequences, plus "any" and  current code allows for fixed one- or two-byte sequences, plus "any" and
5883  "anycrlf". */  "anycrlf". */
5884    
5885  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5886    {    {
5887    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5888    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5889    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5890    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5300  to compile parts of the pattern into; th Line 5937  to compile parts of the pattern into; th
5937  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
5938  is a test for its doing so. */  is a test for its doing so. */
5939    
5940  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
5941  cd->names_found = 0;  cd->names_found = 0;
5942  cd->name_entry_size = 0;  cd->name_entry_size = 0;
5943  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5310  cd->hwm = cworkspace; Line 5947  cd->hwm = cworkspace;
5947  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5948  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5949  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5950  cd->external_options = options;  cd->external_options = options;
5951    cd->external_flags = 0;
5952    
5953  /* 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
5954  don't need to look at the result of the function here. The initial options have  don't need to look at the result of the function here. The initial options have
# Line 5319  been put into the cd block so that they Line 5956  been put into the cd block so that they
5956  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5957  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5958    
5959    ptr += skipatstart;
5960  code = cworkspace;  code = cworkspace;
5961  *code = OP_BRA;  *code = OP_BRA;
5962  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
5963    &code, &ptr, &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, &length);    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
5964      &length);
5965  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
5966    
5967  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,  DPRINTF(("end pre-compile: length=%d workspace=%d\n", length,
# Line 5348  if (re == NULL) Line 5987  if (re == NULL)
5987    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
5988    }    }
5989    
5990  /* 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
5991  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
5992  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
5993  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
5994    pointers. */
5995    
5996  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
5997  re->size = size;  re->size = size;
5998  re->options = cd->external_options;  re->options = cd->external_options;
5999    re->flags = cd->external_flags;
6000  re->dummy1 = 0;  re->dummy1 = 0;
6001  re->first_byte = 0;  re->first_byte = 0;
6002  re->req_byte = 0;  re->req_byte = 0;
# Line 5373  field. Reset the bracket count and the n Line 6014  field. Reset the bracket count and the n
6014  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6015  */  */
6016    
6017    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6018  cd->bracount = 0;  cd->bracount = 0;
6019  cd->names_found = 0;  cd->names_found = 0;
6020  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5380  codestart = cd->name_table + re->name_en Line 6022  codestart = cd->name_table + re->name_en
6022  cd->start_code = codestart;  cd->start_code = codestart;
6023  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6024  cd->req_varyopt = 0;  cd->req_varyopt = 0;
6025  cd->nopartial = FALSE;  cd->had_accept = FALSE;
6026    
6027  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6028  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
6029  of the function here. */  of the function here. */
6030    
6031  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6032  code = (uschar *)codestart;  code = (uschar *)codestart;
6033  *code = OP_BRA;  *code = OP_BRA;
6034  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6035    &errorcode, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6036  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6037  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6038    re->flags = cd->external_flags;
6039    
6040  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6041    
6042  /* 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. */
6043    
# Line 5436  if (errorcode != 0) Line 6079  if (errorcode != 0)
6079    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6080    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6081    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6082    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6083    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6084    return NULL;    return NULL;
6085    }    }
# Line 5465  if ((re->options & PCRE_ANCHORED) == 0) Line 6108  if ((re->options & PCRE_ANCHORED) == 0)
6108        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6109        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6110           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6111        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6112        }        }
6113      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6114        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6115      }      }
6116    }    }
6117    
# Line 5482  if (reqbyte >= 0 && Line 6125  if (reqbyte >= 0 &&
6125    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6126    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6127      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6128    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6129    }    }
6130    
6131  /* Print out the compiled data if debugging is enabled. This is never the  /* Print out the compiled data if debugging is enabled. This is never the
# Line 5493  case when building a production library. Line 6136  case when building a production library.
6136  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6137    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6138    
6139  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 " : "");  
   }  
6140    
6141  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6142    {    {
6143    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6144    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5516  if ((re->options & PCRE_FIRSTSET) != 0) Line 6147  if ((re->options & PCRE_FIRSTSET) != 0)
6147      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6148    }    }
6149    
6150  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6151    {    {
6152    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6153    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5533  was compiled can be seen. */ Line 6164  was compiled can be seen. */
6164  if (code - codestart > length)  if (code - codestart > length)
6165    {    {
6166    (pcre_free)(re);    (pcre_free)(re);
6167    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6168    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6169    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6170    return NULL;    return NULL;

Legend:
Removed from v.172  
changed lines
  Added in v.270

  ViewVC Help
Powered by ViewVC 1.1.5