/[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 175 by ph10, Mon Jun 11 13:38:38 2007 UTC revision 268 by ph10, Thu Nov 15 10:28:09 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 (?\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 (?&";
305    
306    
307  /* 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 431  static const unsigned char ebcdic_charta
431  /* Definition to allow mutual recursion */  /* Definition to allow mutual recursion */
432    
433  static BOOL  static BOOL
434    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,    compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,
435      int *, int *, branch_chain *, compile_data *, int *);      int *, int *, branch_chain *, compile_data *, int *);
436    
437    
438    
439  /*************************************************  /*************************************************
440    *            Find an error text                  *
441    *************************************************/
442    
443    /* The error texts are now all in one long string, to save on relocations. As
444    some of the text is of unknown length, we can't use a table of offsets.
445    Instead, just count through the strings. This is not a performance issue
446    because it happens only when there has been a compilation error.
447    
448    Argument:   the error number
449    Returns:    pointer to the error string
450    */
451    
452    static const char *
453    find_error_text(int n)
454    {
455    const char *s = error_texts;
456    for (; n > 0; n--) while (*s++ != 0);
457    return s;
458    }
459    
460    
461    /*************************************************
462  *            Handle escapes                      *  *            Handle escapes                      *
463  *************************************************/  *************************************************/
464    
# Line 400  Arguments: Line 479  Arguments:
479    
480  Returns:         zero or positive => a data character  Returns:         zero or positive => a data character
481                   negative => a special escape sequence                   negative => a special escape sequence
482                   on error, errorptr is set                   on error, errorcodeptr is set
483  */  */
484    
485  static int  static int
# Line 485  else Line 564  else
564      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
565        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - '0';
566    
567        if (c < 0)
568          {
569          *errorcodeptr = ERR61;
570          break;
571          }
572    
573      if (c == 0 || (braced && *(++ptr) != '}'))      if (c == 0 || (braced && *(++ptr) != '}'))
574        {        {
575        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
576        return 0;        break;
577        }        }
578    
579      if (negated)      if (negated)
# Line 496  else Line 581  else
581        if (c > bracount)        if (c > bracount)
582          {          {
583          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
584          return 0;          break;
585          }          }
586        c = bracount - (c - 1);        c = bracount - (c - 1);
587        }        }
# Line 525  else Line 610  else
610        c -= '0';        c -= '0';
611        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
612          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - '0';
613          if (c < 0)
614            {
615            *errorcodeptr = ERR61;
616            break;
617            }
618        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
619          {          {
620          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 620  else Line 710  else
710      if (c == 0)      if (c == 0)
711        {        {
712        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
713        return 0;        break;
714        }        }
715    
716  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII coding */
# Line 696  if (c == '{') Line 786  if (c == '{')
786      *negptr = TRUE;      *negptr = TRUE;
787      ptr++;      ptr++;
788      }      }
789    for (i = 0; i < sizeof(name) - 1; i++)    for (i = 0; i < (int)sizeof(name) - 1; i++)
790      {      {
791      c = *(++ptr);      c = *(++ptr);
792      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 725  top = _pcre_utt_size; Line 815  top = _pcre_utt_size;
815  while (bot < top)  while (bot < top)
816    {    {
817    i = (bot + top) >> 1;    i = (bot + top) >> 1;
818    c = strcmp(name, _pcre_utt[i].name);    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);
819    if (c == 0)    if (c == 0)
820      {      {
821      *dptr = _pcre_utt[i].value;      *dptr = _pcre_utt[i].value;
# Line 899  for (; *ptr != 0; ptr++) Line 989  for (; *ptr != 0; ptr++)
989      {      {
990      while (*(++ptr) != ']')      while (*(++ptr) != ']')
991        {        {
992          if (*ptr == 0) return -1;
993        if (*ptr == '\\')        if (*ptr == '\\')
994          {          {
995          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
# Line 926  for (; *ptr != 0; ptr++) Line 1017  for (; *ptr != 0; ptr++)
1017    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
1018    
1019    if (*ptr != '(') continue;    if (*ptr != '(') continue;
1020    if (ptr[1] != '?')    if (ptr[1] != '?' && ptr[1] != '*')
1021      {      {
1022      count++;      count++;
1023      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 1054  for (;;) Line 1145  for (;;)
1145    {    {
1146    int d;    int d;
1147    register int op = *cc;    register int op = *cc;
   
1148    switch (op)    switch (op)
1149      {      {
1150      case OP_CBRA:      case OP_CBRA:
# Line 1143  for (;;) Line 1233  for (;;)
1233    
1234      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1235      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1236        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1237      cc += 4;      cc += 4;
1238      break;      break;
1239    
# Line 1251  for (;;) Line 1342  for (;;)
1342      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1343      }      }
1344    
1345    /* 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
1346    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
1347    arrange to skip the extra bytes. */    two bytes of parameters. */
1348    
1349    else    else
1350      {      {
1351        switch(c)
1352          {
1353          case OP_TYPESTAR:
1354          case OP_TYPEMINSTAR:
1355          case OP_TYPEPLUS:
1356          case OP_TYPEMINPLUS:
1357          case OP_TYPEQUERY:
1358          case OP_TYPEMINQUERY:
1359          case OP_TYPEPOSSTAR:
1360          case OP_TYPEPOSPLUS:
1361          case OP_TYPEPOSQUERY:
1362          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1363          break;
1364    
1365          case OP_TYPEUPTO:
1366          case OP_TYPEMINUPTO:
1367          case OP_TYPEEXACT:
1368          case OP_TYPEPOSUPTO:
1369          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1370          break;
1371          }
1372    
1373        /* Add in the fixed length from the table */
1374    
1375      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1376    
1377      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1378      a multi-byte character. The length in the table is a minimum, so we have to
1379      arrange to skip the extra bytes. */
1380    
1381  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1382      if (utf8) switch(c)      if (utf8) switch(c)
1383        {        {
# Line 1315  for (;;) Line 1435  for (;;)
1435    
1436    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1437    
1438    /* 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
1439    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
1440    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1441    
1442    else    else
1443      {      {
1444        switch(c)
1445          {
1446          case OP_TYPESTAR:
1447          case OP_TYPEMINSTAR:
1448          case OP_TYPEPLUS:
1449          case OP_TYPEMINPLUS:
1450          case OP_TYPEQUERY:
1451          case OP_TYPEMINQUERY:
1452          case OP_TYPEPOSSTAR:
1453          case OP_TYPEPOSPLUS:
1454          case OP_TYPEPOSQUERY:
1455          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1456          break;
1457    
1458          case OP_TYPEPOSUPTO:
1459          case OP_TYPEUPTO:
1460          case OP_TYPEMINUPTO:
1461          case OP_TYPEEXACT:
1462          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1463          break;
1464          }
1465    
1466        /* Add in the fixed length from the table */
1467    
1468      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1469    
1470        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1471        by a multi-byte character. The length in the table is a minimum, so we have
1472        to arrange to skip the extra bytes. */
1473    
1474  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1475      if (utf8) switch(c)      if (utf8) switch(c)
1476        {        {
# Line 1394  for (code = first_significant_code(code Line 1542  for (code = first_significant_code(code
1542    
1543    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
1544    
1545    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE)    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)
1546      {      {
1547      BOOL empty_branch;      BOOL empty_branch;
1548      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 1566  for (code = first_significant_code(code
1566    
1567    switch (c)    switch (c)
1568      {      {
1569      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1570        cannot be represented just by a bit map. This includes negated single
1571        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1572        actual length is stored in the compiled code, so we must update "code"
1573        here. */
1574    
1575  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1576      case OP_XCLASS:      case OP_XCLASS:
1577      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1578      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1579  #endif  #endif
1580    
# Line 1484  for (code = first_significant_code(code Line 1636  for (code = first_significant_code(code
1636      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1637      return FALSE;      return FALSE;
1638    
1639        /* These are going to continue, as they may be empty, but we have to
1640        fudge the length for the \p and \P cases. */
1641    
1642        case OP_TYPESTAR:
1643        case OP_TYPEMINSTAR:
1644        case OP_TYPEPOSSTAR:
1645        case OP_TYPEQUERY:
1646        case OP_TYPEMINQUERY:
1647        case OP_TYPEPOSQUERY:
1648        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1649        break;
1650    
1651        /* Same for these */
1652    
1653        case OP_TYPEUPTO:
1654        case OP_TYPEMINUPTO:
1655        case OP_TYPEPOSUPTO:
1656        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1657        break;
1658    
1659      /* End of branch */      /* End of branch */
1660    
1661      case OP_KET:      case OP_KET:
# Line 1600  Returns:     a value representing the na Line 1772  Returns:     a value representing the na
1772  static int  static int
1773  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1774  {  {
1775    const char *pn = posix_names;
1776  register int yield = 0;  register int yield = 0;
1777  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1778    {    {
1779    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1780      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1781      pn += posix_name_lengths[yield] + 1;
1782    yield++;    yield++;
1783    }    }
1784  return -1;  return -1;
# Line 1646  adjust_recurse(uschar *group, int adjust Line 1820  adjust_recurse(uschar *group, int adjust
1820    uschar *save_hwm)    uschar *save_hwm)
1821  {  {
1822  uschar *ptr = group;  uschar *ptr = group;
1823    
1824  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1825    {    {
1826    int offset;    int offset;
# Line 1944  if (next >= 0) switch(op_code) Line 2119  if (next >= 0) switch(op_code)
2119    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2120    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;
2121    
2122      case OP_HSPACE:
2123      case OP_NOT_HSPACE:
2124      switch(next)
2125        {
2126        case 0x09:
2127        case 0x20:
2128        case 0xa0:
2129        case 0x1680:
2130        case 0x180e:
2131        case 0x2000:
2132        case 0x2001:
2133        case 0x2002:
2134        case 0x2003:
2135        case 0x2004:
2136        case 0x2005:
2137        case 0x2006:
2138        case 0x2007:
2139        case 0x2008:
2140        case 0x2009:
2141        case 0x200A:
2142        case 0x202f:
2143        case 0x205f:
2144        case 0x3000:
2145        return op_code != OP_HSPACE;
2146        default:
2147        return op_code == OP_HSPACE;
2148        }
2149    
2150      case OP_VSPACE:
2151      case OP_NOT_VSPACE:
2152      switch(next)
2153        {
2154        case 0x0a:
2155        case 0x0b:
2156        case 0x0c:
2157        case 0x0d:
2158        case 0x85:
2159        case 0x2028:
2160        case 0x2029:
2161        return op_code != OP_VSPACE;
2162        default:
2163        return op_code == OP_VSPACE;
2164        }
2165    
2166    default:    default:
2167    return FALSE;    return FALSE;
2168    }    }
# Line 1978  switch(op_code) Line 2197  switch(op_code)
2197      case ESC_W:      case ESC_W:
2198      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;      return item <= 127 && (cd->ctypes[item] & ctype_word) != 0;
2199    
2200        case ESC_h:
2201        case ESC_H:
2202        switch(item)
2203          {
2204          case 0x09:
2205          case 0x20:
2206          case 0xa0:
2207          case 0x1680:
2208          case 0x180e:
2209          case 0x2000:
2210          case 0x2001:
2211          case 0x2002:
2212          case 0x2003:
2213          case 0x2004:
2214          case 0x2005:
2215          case 0x2006:
2216          case 0x2007:
2217          case 0x2008:
2218          case 0x2009:
2219          case 0x200A:
2220          case 0x202f:
2221          case 0x205f:
2222          case 0x3000:
2223          return -next != ESC_h;
2224          default:
2225          return -next == ESC_h;
2226          }
2227    
2228        case ESC_v:
2229        case ESC_V:
2230        switch(item)
2231          {
2232          case 0x0a:
2233          case 0x0b:
2234          case 0x0c:
2235          case 0x0d:
2236          case 0x85:
2237          case 0x2028:
2238          case 0x2029:
2239          return -next != ESC_v;
2240          default:
2241          return -next == ESC_v;
2242          }
2243    
2244      default:      default:
2245      return FALSE;      return FALSE;
2246      }      }
2247    
2248    case OP_DIGIT:    case OP_DIGIT:
2249    return next == -ESC_D || next == -ESC_s || next == -ESC_W;    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||
2250             next == -ESC_h || next == -ESC_v;
2251    
2252    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
2253    return next == -ESC_d;    return next == -ESC_d;
# Line 1992  switch(op_code) Line 2256  switch(op_code)
2256    return next == -ESC_S || next == -ESC_d || next == -ESC_w;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
2257    
2258    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
2259    return next == -ESC_s;    return next == -ESC_s || next == -ESC_h || next == -ESC_v;
2260    
2261      case OP_HSPACE:
2262      return next == -ESC_S || next == -ESC_H || next == -ESC_d || next == -ESC_w;
2263    
2264      case OP_NOT_HSPACE:
2265      return next == -ESC_h;
2266    
2267      /* Can't have \S in here because VT matches \S (Perl anomaly) */
2268      case OP_VSPACE:
2269      return next == -ESC_V || next == -ESC_d || next == -ESC_w;
2270    
2271      case OP_NOT_VSPACE:
2272      return next == -ESC_v;
2273    
2274    case OP_WORDCHAR:    case OP_WORDCHAR:
2275    return next == -ESC_W || next == -ESC_s;    return next == -ESC_W || next == -ESC_s || next == -ESC_h || next == -ESC_v;
2276    
2277    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
2278    return next == -ESC_w || next == -ESC_d;    return next == -ESC_w || next == -ESC_d;
# Line 2107  req_caseopt = ((options & PCRE_CASELESS) Line 2384  req_caseopt = ((options & PCRE_CASELESS)
2384  for (;; ptr++)  for (;; ptr++)
2385    {    {
2386    BOOL negate_class;    BOOL negate_class;
2387      BOOL should_flip_negation;
2388    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2389    BOOL is_quantifier;    BOOL is_quantifier;
2390    BOOL is_recurse;    BOOL is_recurse;
2391    BOOL reset_bracount;    BOOL reset_bracount;
2392    int class_charcount;    int class_charcount;
2393    int class_lastchar;    int class_lastchar;
2394    int newoptions;    int newoptions;
# Line 2148  for (;; ptr++) Line 2426  for (;; ptr++)
2426      */      */
2427    
2428      if (code < last_code) code = last_code;      if (code < last_code) code = last_code;
2429    
2430        /* Paranoid check for integer overflow */
2431    
2432        if (OFLOW_MAX - *lengthptr < code - last_code)
2433          {
2434          *errorcodeptr = ERR20;
2435          goto FAILED;
2436          }
2437    
2438      *lengthptr += code - last_code;      *lengthptr += code - last_code;
2439      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));
2440    
# Line 2260  for (;; ptr++) Line 2547  for (;; ptr++)
2547      *ptrptr = ptr;      *ptrptr = ptr;
2548      if (lengthptr != NULL)      if (lengthptr != NULL)
2549        {        {
2550          if (OFLOW_MAX - *lengthptr < code - last_code)
2551            {
2552            *errorcodeptr = ERR20;
2553            goto FAILED;
2554            }
2555        *lengthptr += code - last_code;   /* To include callout length */        *lengthptr += code - last_code;   /* To include callout length */
2556        DPRINTF((">> end branch\n"));        DPRINTF((">> end branch\n"));
2557        }        }
# Line 2322  for (;; ptr++) Line 2614  for (;; ptr++)
2614        goto FAILED;        goto FAILED;
2615        }        }
2616    
2617      /* 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,
2618        if the first few characters (either before or after ^) are \Q\E or \E we
2619        skip them too. This makes for compatibility with Perl. */
2620    
2621      if ((c = *(++ptr)) == '^')      negate_class = FALSE;
2622        for (;;)
2623        {        {
       negate_class = TRUE;  
2624        c = *(++ptr);        c = *(++ptr);
2625          if (c == '\\')
2626            {
2627            if (ptr[1] == 'E') ptr++;
2628              else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;
2629                else break;
2630            }
2631          else if (!negate_class && c == '^')
2632            negate_class = TRUE;
2633          else break;
2634        }        }
2635      else  
2636        {      /* If a class contains a negative special such as \S, we need to flip the
2637        negate_class = FALSE;      negation flag at the end, so that support for characters > 255 works
2638        }      correctly (they are all included in the class). */
2639    
2640        should_flip_negation = FALSE;
2641    
2642      /* 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
2643      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 2406  for (;; ptr++) Line 2711  for (;; ptr++)
2711          if (*ptr == '^')          if (*ptr == '^')
2712            {            {
2713            local_negate = TRUE;            local_negate = TRUE;
2714              should_flip_negation = TRUE;  /* Note negative special */
2715            ptr++;            ptr++;
2716            }            }
2717    
# Line 2472  for (;; ptr++) Line 2778  for (;; ptr++)
2778        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
2779        case. Inside a class (and only there) it is treated as backspace.        case. Inside a class (and only there) it is treated as backspace.
2780        Elsewhere it marks a word boundary. Other escapes have preset maps ready        Elsewhere it marks a word boundary. Other escapes have preset maps ready
2781        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
2782        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2783    
2784        if (c == '\\')        if (c == '\\')
# Line 2492  for (;; ptr++) Line 2798  for (;; ptr++)
2798            else inescq = TRUE;            else inescq = TRUE;
2799            continue;            continue;
2800            }            }
2801            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2802    
2803          if (c < 0)          if (c < 0)
2804            {            {
# Line 2507  for (;; ptr++) Line 2814  for (;; ptr++)
2814              continue;              continue;
2815    
2816              case ESC_D:              case ESC_D:
2817                should_flip_negation = TRUE;
2818              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
2819              continue;              continue;
2820    
# Line 2515  for (;; ptr++) Line 2823  for (;; ptr++)
2823              continue;              continue;
2824    
2825              case ESC_W:              case ESC_W:
2826                should_flip_negation = TRUE;
2827              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
2828              continue;              continue;
2829    
# Line 2524  for (;; ptr++) Line 2833  for (;; ptr++)
2833              continue;              continue;
2834    
2835              case ESC_S:              case ESC_S:
2836                should_flip_negation = TRUE;
2837              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
2838              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
2839              continue;              continue;
# Line 2540  for (;; ptr++) Line 2850  for (;; ptr++)
2850            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||            else if (c == -ESC_d || c == -ESC_D || c == -ESC_w ||
2851                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;                     c == -ESC_W || c == -ESC_s || c == -ESC_S) continue;
2852    
2853              /* We need to deal with \H, \h, \V, and \v in both phases because
2854              they use extra memory. */
2855    
2856              if (-c == ESC_h)
2857                {
2858                SETBIT(classbits, 0x09); /* VT */
2859                SETBIT(classbits, 0x20); /* SPACE */
2860                SETBIT(classbits, 0xa0); /* NSBP */
2861    #ifdef SUPPORT_UTF8
2862                if (utf8)
2863                  {
2864                  class_utf8 = TRUE;
2865                  *class_utf8data++ = XCL_SINGLE;
2866                  class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);
2867                  *class_utf8data++ = XCL_SINGLE;
2868                  class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);
2869                  *class_utf8data++ = XCL_RANGE;
2870                  class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);
2871                  class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);
2872                  *class_utf8data++ = XCL_SINGLE;
2873                  class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);
2874                  *class_utf8data++ = XCL_SINGLE;
2875                  class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);
2876                  *class_utf8data++ = XCL_SINGLE;
2877                  class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);
2878                  }
2879    #endif
2880                continue;
2881                }
2882    
2883              if (-c == ESC_H)
2884                {
2885                for (c = 0; c < 32; c++)
2886                  {
2887                  int x = 0xff;
2888                  switch (c)
2889                    {
2890                    case 0x09/8: x ^= 1 << (0x09%8); break;
2891                    case 0x20/8: x ^= 1 << (0x20%8); break;
2892                    case 0xa0/8: x ^= 1 << (0xa0%8); break;
2893                    default: break;
2894                    }
2895                  classbits[c] |= x;
2896                  }
2897    
2898    #ifdef SUPPORT_UTF8
2899                if (utf8)
2900                  {
2901                  class_utf8 = TRUE;
2902                  *class_utf8data++ = XCL_RANGE;
2903                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2904                  class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);
2905                  *class_utf8data++ = XCL_RANGE;
2906                  class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);
2907                  class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);
2908                  *class_utf8data++ = XCL_RANGE;
2909                  class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);
2910                  class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);
2911                  *class_utf8data++ = XCL_RANGE;
2912                  class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);
2913                  class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);
2914                  *class_utf8data++ = XCL_RANGE;
2915                  class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);
2916                  class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);
2917                  *class_utf8data++ = XCL_RANGE;
2918                  class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);
2919                  class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);
2920                  *class_utf8data++ = XCL_RANGE;
2921                  class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);
2922                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2923                  }
2924    #endif
2925                continue;
2926                }
2927    
2928              if (-c == ESC_v)
2929                {
2930                SETBIT(classbits, 0x0a); /* LF */
2931                SETBIT(classbits, 0x0b); /* VT */
2932                SETBIT(classbits, 0x0c); /* FF */
2933                SETBIT(classbits, 0x0d); /* CR */
2934                SETBIT(classbits, 0x85); /* NEL */
2935    #ifdef SUPPORT_UTF8
2936                if (utf8)
2937                  {
2938                  class_utf8 = TRUE;
2939                  *class_utf8data++ = XCL_RANGE;
2940                  class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);
2941                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2942                  }
2943    #endif
2944                continue;
2945                }
2946    
2947              if (-c == ESC_V)
2948                {
2949                for (c = 0; c < 32; c++)
2950                  {
2951                  int x = 0xff;
2952                  switch (c)
2953                    {
2954                    case 0x0a/8: x ^= 1 << (0x0a%8);
2955                                 x ^= 1 << (0x0b%8);
2956                                 x ^= 1 << (0x0c%8);
2957                                 x ^= 1 << (0x0d%8);
2958                                 break;
2959                    case 0x85/8: x ^= 1 << (0x85%8); break;
2960                    default: break;
2961                    }
2962                  classbits[c] |= x;
2963                  }
2964    
2965    #ifdef SUPPORT_UTF8
2966                if (utf8)
2967                  {
2968                  class_utf8 = TRUE;
2969                  *class_utf8data++ = XCL_RANGE;
2970                  class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);
2971                  class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);
2972                  *class_utf8data++ = XCL_RANGE;
2973                  class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);
2974                  class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);
2975                  }
2976    #endif
2977                continue;
2978                }
2979    
2980            /* We need to deal with \P and \p in both phases. */            /* We need to deal with \P and \p in both phases. */
2981    
2982  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 2591  for (;; ptr++) Line 3028  for (;; ptr++)
3028    
3029        oldptr = ptr;        oldptr = ptr;
3030    
3031          /* Remember \r or \n */
3032    
3033          if (c == '\r' || c == '\n') cd->external_flags |= PCRE_HASCRORLF;
3034    
3035          /* Check for range */
3036    
3037        if (!inescq && ptr[1] == '-')        if (!inescq && ptr[1] == '-')
3038          {          {
3039          int d;          int d;
# Line 2658  for (;; ptr++) Line 3101  for (;; ptr++)
3101    
3102          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3103    
3104            /* Remember \r or \n */
3105    
3106            if (d == '\r' || d == '\n') cd->external_flags |= PCRE_HASCRORLF;
3107    
3108          /* 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
3109          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3110          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 2680  for (;; ptr++) Line 3127  for (;; ptr++)
3127              unsigned int origd = d;              unsigned int origd = d;
3128              while (get_othercase_range(&cc, origd, &occ, &ocd))              while (get_othercase_range(&cc, origd, &occ, &ocd))
3129                {                {
3130                if (occ >= c && ocd <= d) continue;  /* Skip embedded ranges */                if (occ >= (unsigned int)c &&
3131                      ocd <= (unsigned int)d)
3132                    continue;                          /* Skip embedded ranges */
3133    
3134                if (occ < c  && ocd >= c - 1)        /* Extend the basic range */                if (occ < (unsigned int)c  &&
3135                      ocd >= (unsigned int)c - 1)      /* Extend the basic range */
3136                  {                                  /* if there is overlap,   */                  {                                  /* if there is overlap,   */
3137                  c = occ;                           /* noting that if occ < c */                  c = occ;                           /* noting that if occ < c */
3138                  continue;                          /* we can't have ocd > d  */                  continue;                          /* we can't have ocd > d  */
3139                  }                                  /* because a subrange is  */                  }                                  /* because a subrange is  */
3140                if (ocd > d && occ <= d + 1)         /* always shorter than    */                if (ocd > (unsigned int)d &&
3141                      occ <= (unsigned int)d + 1)      /* always shorter than    */
3142                  {                                  /* the basic range.       */                  {                                  /* the basic range.       */
3143                  d = ocd;                  d = ocd;
3144                  continue;                  continue;
# Line 2807  for (;; ptr++) Line 3258  for (;; ptr++)
3258        goto FAILED;        goto FAILED;
3259        }        }
3260    
3261    
3262    /* This code has been disabled because it would mean that \s counts as
3263    an explicit \r or \n reference, and that's not really what is wanted. Now
3264    we set the flag only if there is a literal "\r" or "\n" in the class. */
3265    
3266    #if 0
3267        /* Remember whether \r or \n are in this class */
3268    
3269        if (negate_class)
3270          {
3271          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3272          }
3273        else
3274          {
3275          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3276          }
3277    #endif
3278    
3279    
3280      /* 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
3281      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
3282      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
3283      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3284      single-bytes only. This is an historical hangover. Maybe one day we can  
3285      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3286        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3287        operate on single-bytes only. This is an historical hangover. Maybe one day
3288        we can tidy these opcodes to handle multi-byte characters.
3289    
3290      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
3291      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 2822  for (;; ptr++) Line 3295  for (;; ptr++)
3295      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3296    
3297  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3298      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3299            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3300  #else  #else
3301      if (class_charcount == 1)      if (class_charcount == 1)
3302  #endif  #endif
# Line 2868  for (;; ptr++) Line 3339  for (;; ptr++)
3339      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3340    
3341      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3342      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3343      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
3344        the class, so any that were explicitly given as well can be ignored. If
3345        (when there are explicit characters > 255 that must be listed) there are no
3346        characters < 256, we can omit the bitmap in the actual compiled code. */
3347    
3348  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3349      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3350        {        {
3351        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3352        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 2898  for (;; ptr++) Line 3372  for (;; ptr++)
3372        }        }
3373  #endif  #endif
3374    
3375      /* 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
3376      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
3377      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
3378      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3379    
3380        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3381      if (negate_class)      if (negate_class)
3382        {        {
       *code++ = OP_NCLASS;  
3383        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3384          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3385        }        }
3386      else      else
3387        {        {
       *code++ = OP_CLASS;  
3388        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3389        }        }
3390      code += 32;      code += 32;
# Line 3097  for (;; ptr++) Line 3570  for (;; ptr++)
3570        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3571        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3572    
3573        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3574    
3575        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3576    
# Line 3247  for (;; ptr++) Line 3720  for (;; ptr++)
3720        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3721        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3722    
3723        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3724    
3725        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3726          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3283  for (;; ptr++) Line 3756  for (;; ptr++)
3756          goto FAILED;          goto FAILED;
3757          }          }
3758    
       /* This is a paranoid check to stop integer overflow later on */  
   
       if (len > MAX_DUPLENGTH)  
         {  
         *errorcodeptr = ERR50;  
         goto FAILED;  
         }  
   
3759        /* 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
3760        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
3761        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 3379  for (;; ptr++) Line 3844  for (;; ptr++)
3844          if (repeat_min > 1)          if (repeat_min > 1)
3845            {            {
3846            /* 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
3847            just adjust the length as if we had. */            just adjust the length as if we had. Do some paranoid checks for
3848              potential integer overflow. */
3849    
3850            if (lengthptr != NULL)            if (lengthptr != NULL)
3851              *lengthptr += (repeat_min - 1)*length_prevgroup;              {
3852                int delta = (repeat_min - 1)*length_prevgroup;
3853                if ((double)(repeat_min - 1)*(double)length_prevgroup >
3854                                                                (double)INT_MAX ||
3855                    OFLOW_MAX - *lengthptr < delta)
3856                  {
3857                  *errorcodeptr = ERR20;
3858                  goto FAILED;
3859                  }
3860                *lengthptr += delta;
3861                }
3862    
3863            /* This is compiling for real */            /* This is compiling for real */
3864    
# Line 3420  for (;; ptr++) Line 3896  for (;; ptr++)
3896          /* 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
3897          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
3898          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
3899          add 2 + 2*LINKSIZE to allow for the nesting that occurs. */          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
3900            paranoid checks to avoid integer overflow. */
3901    
3902          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
3903            *lengthptr += repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            {
3904              2 - 2*LINK_SIZE;  /* Last one doesn't nest */            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
3905                          2 - 2*LINK_SIZE;   /* Last one doesn't nest */
3906              if ((double)repeat_max *
3907                    (double)(length_prevgroup + 1 + 2 + 2*LINK_SIZE)
3908                      > (double)INT_MAX ||
3909                  OFLOW_MAX - *lengthptr < delta)
3910                {
3911                *errorcodeptr = ERR20;
3912                goto FAILED;
3913                }
3914              *lengthptr += delta;
3915              }
3916    
3917          /* This is compiling for real */          /* This is compiling for real */
3918    
# Line 3576  for (;; ptr++) Line 4064  for (;; ptr++)
4064      /* ===================================================================*/      /* ===================================================================*/
4065      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4066      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4067      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. */  
4068    
4069      case '(':      case '(':
4070      newoptions = options;      newoptions = options;
4071      skipbytes = 0;      skipbytes = 0;
4072      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4073      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4074      reset_bracount = FALSE;      reset_bracount = FALSE;
4075    
4076      if (*(++ptr) == '?')      /* First deal with various "verbs" that can be introduced by '*'. */
4077    
4078        if (*(++ptr) == '*' && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4079          {
4080          int i, namelen;
4081          const char *vn = verbnames;
4082          const uschar *name = ++ptr;
4083          previous = NULL;
4084          while ((cd->ctypes[*++ptr] & ctype_letter) != 0);
4085          if (*ptr == ':')
4086            {
4087            *errorcodeptr = ERR59;   /* Not supported */
4088            goto FAILED;
4089            }
4090          if (*ptr != ')')
4091            {
4092            *errorcodeptr = ERR60;
4093            goto FAILED;
4094            }
4095          namelen = ptr - name;
4096          for (i = 0; i < verbcount; i++)
4097            {
4098            if (namelen == verbs[i].len &&
4099                strncmp((char *)name, vn, namelen) == 0)
4100              {
4101              *code = verbs[i].op;
4102              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4103              break;
4104              }
4105            vn += verbs[i].len + 1;
4106            }
4107          if (i < verbcount) continue;
4108          *errorcodeptr = ERR60;
4109          goto FAILED;
4110          }
4111    
4112        /* Deal with the extended parentheses; all are introduced by '?', and the
4113        appearance of any of them means that this is not a capturing group. */
4114    
4115        else if (*ptr == '?')
4116        {        {
4117        int i, set, unset, namelen;        int i, set, unset, namelen;
4118        int *optset;        int *optset;
# Line 3610  for (;; ptr++) Line 4135  for (;; ptr++)
4135          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4136          case '|':                 /* Reset capture count for each branch */          case '|':                 /* Reset capture count for each branch */
4137          reset_bracount = TRUE;          reset_bracount = TRUE;
4138          /* Fall through */          /* Fall through */
4139    
4140          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4141          case ':':                 /* Non-capturing bracket */          case ':':                 /* Non-capturing bracket */
# Line 3829  for (;; ptr++) Line 4354  for (;; ptr++)
4354    
4355          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4356          case '!':                 /* Negative lookahead */          case '!':                 /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4357          ptr++;          ptr++;
4358            if (*ptr == ')')          /* Optimize (?!) */
4359              {
4360              *code++ = OP_FAIL;
4361              previous = NULL;
4362              continue;
4363              }
4364            bravalue = OP_ASSERT_NOT;
4365          break;          break;
4366    
4367    
# Line 4005  for (;; ptr++) Line 4536  for (;; ptr++)
4536    
4537          if (lengthptr != NULL)          if (lengthptr != NULL)
4538            {            {
4539              if (namelen == 0)
4540                {
4541                *errorcodeptr = ERR62;
4542                goto FAILED;
4543                }
4544            if (*ptr != terminator)            if (*ptr != terminator)
4545              {              {
4546              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4018  for (;; ptr++) Line 4554  for (;; ptr++)
4554            recno = 0;            recno = 0;
4555            }            }
4556    
4557          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4558            first, and then check that we have reached the end of the name in the
4559            table. That way, if the name that is longer than any in the table,
4560            the comparison will fail without reading beyond the table entry. */
4561    
4562          else          else
4563            {            {
4564            slot = cd->name_table;            slot = cd->name_table;
4565            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4566              {              {
4567              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4568                    slot[2+namelen] == 0)
4569                  break;
4570              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4571              }              }
4572    
# Line 4188  for (;; ptr++) Line 4729  for (;; ptr++)
4729    
4730              case 'J':    /* Record that it changed in the external options */              case 'J':    /* Record that it changed in the external options */
4731              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4732              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4733              break;              break;
4734    
4735              case 'i': *optset |= PCRE_CASELESS; break;              case 'i': *optset |= PCRE_CASELESS; break;
# Line 4311  for (;; ptr++) Line 4852  for (;; ptr++)
4852           errorcodeptr,                 /* Where to put an error message */           errorcodeptr,                 /* Where to put an error message */
4853           (bravalue == OP_ASSERTBACK ||           (bravalue == OP_ASSERTBACK ||
4854            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */            bravalue == OP_ASSERTBACK_NOT), /* TRUE if back assert */
4855           reset_bracount,               /* True if (?| group */           reset_bracount,               /* True if (?| group */
4856           skipbytes,                    /* Skip over bracket number */           skipbytes,                    /* Skip over bracket number */
4857           &subfirstbyte,                /* For possible first char */           &subfirstbyte,                /* For possible first char */
4858           &subreqbyte,                  /* For possible last char */           &subreqbyte,                  /* For possible last char */
# Line 4379  for (;; ptr++) Line 4920  for (;; ptr++)
4920        goto FAILED;        goto FAILED;
4921        }        }
4922    
4923      /* 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,
4924      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
4925      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
4926      a quantifier. */      duplicated by a quantifier.*/
4927    
4928      if (lengthptr != NULL)      if (lengthptr != NULL)
4929        {        {
4930          if (OFLOW_MAX - *lengthptr < length_prevgroup - 2 - 2*LINK_SIZE)
4931            {
4932            *errorcodeptr = ERR20;
4933            goto FAILED;
4934            }
4935        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
4936        code++;        *code++ = OP_BRA;
4937        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4938        *code++ = OP_KET;        *code++ = OP_KET;
4939        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
4940          break;    /* No need to waste time with special character handling */
4941        }        }
4942    
4943      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
4944    
4945      else code = tempcode;      code = tempcode;
4946    
4947      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
4948      relevant. */      relevant. */
# Line 4599  for (;; ptr++) Line 5146  for (;; ptr++)
5146      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5147      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5148    
5149        /* Remember if \r or \n were seen */
5150    
5151        if (mcbuffer[0] == '\r' || mcbuffer[0] == '\n')
5152          cd->external_flags |= PCRE_HASCRORLF;
5153    
5154      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5155      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.
5156      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 4671  Arguments: Line 5223  Arguments:
5223    ptrptr         -> the address of the current pattern pointer    ptrptr         -> the address of the current pattern pointer
5224    errorcodeptr   -> pointer to error code variable    errorcodeptr   -> pointer to error code variable
5225    lookbehind     TRUE if this is a lookbehind assertion    lookbehind     TRUE if this is a lookbehind assertion
5226    reset_bracount TRUE to reset the count for each branch    reset_bracount TRUE to reset the count for each branch
5227    skipbytes      skip this many bytes at start (for brackets and OP_COND)    skipbytes      skip this many bytes at start (for brackets and OP_COND)
5228    firstbyteptr   place to put the first required character, or a negative number    firstbyteptr   place to put the first required character, or a negative number
5229    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 4685  Returns:         TRUE on success Line 5237  Returns:         TRUE on success
5237    
5238  static BOOL  static BOOL
5239  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,  compile_regex(int options, int oldims, uschar **codeptr, const uschar **ptrptr,
5240    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,    int *errorcodeptr, BOOL lookbehind, BOOL reset_bracount, int skipbytes,
5241    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,    int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr, compile_data *cd,
5242    int *lengthptr)    int *lengthptr)
5243  {  {
5244  const uschar *ptr = *ptrptr;  const uschar *ptr = *ptrptr;
# Line 4731  orig_bracount = max_bracount = cd->braco Line 5283  orig_bracount = max_bracount = cd->braco
5283  for (;;)  for (;;)
5284    {    {
5285    /* For a (?| group, reset the capturing bracket count so that each branch    /* For a (?| group, reset the capturing bracket count so that each branch
5286    uses the same numbers. */    uses the same numbers. */
5287    
5288    if (reset_bracount) cd->bracount = orig_bracount;    if (reset_bracount) cd->bracount = orig_bracount;
5289    
5290    /* Handle a change of ims options at the start of the branch */    /* Handle a change of ims options at the start of the branch */
5291    
5292    if ((options & PCRE_IMS) != oldims)    if ((options & PCRE_IMS) != oldims)
# Line 4763  for (;;) Line 5315  for (;;)
5315      *ptrptr = ptr;      *ptrptr = ptr;
5316      return FALSE;      return FALSE;
5317      }      }
5318    
5319    /* Keep the highest bracket count in case (?| was used and some branch    /* Keep the highest bracket count in case (?| was used and some branch
5320    has fewer than the rest. */    has fewer than the rest. */
5321    
5322    if (cd->bracount > max_bracount) max_bracount = cd->bracount;    if (cd->bracount > max_bracount) max_bracount = cd->bracount;
5323    
5324    /* 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. */
# Line 4870  for (;;) Line 5422  for (;;)
5422        *code++ = oldims;        *code++ = oldims;
5423        length += 2;        length += 2;
5424        }        }
5425    
5426      /* Retain the highest bracket number, in case resetting was used. */      /* Retain the highest bracket number, in case resetting was used. */
5427    
5428      cd->bracount = max_bracount;      cd->bracount = max_bracount;
5429    
5430      /* Set values to pass back */      /* Set values to pass back */
# Line 4881  for (;;) Line 5433  for (;;)
5433      *ptrptr = ptr;      *ptrptr = ptr;
5434      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
5435      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
5436      if (lengthptr != NULL) *lengthptr += length;      if (lengthptr != NULL)
5437          {
5438          if (OFLOW_MAX - *lengthptr < length)
5439            {
5440            *errorcodeptr = ERR20;
5441            return FALSE;
5442            }
5443          *lengthptr += length;
5444          }
5445      return TRUE;      return TRUE;
5446      }      }
5447    
# Line 5190  real_pcre *re; Line 5750  real_pcre *re;
5750  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
5751  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
5752  int errorcode = 0;  int errorcode = 0;
5753    int skipatstart = 0;
5754  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
5755  BOOL utf8;  BOOL utf8;
5756  #endif  #endif
# Line 5268  cd->fcc = tables + fcc_offset; Line 5829  cd->fcc = tables + fcc_offset;
5829  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
5830  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
5831    
5832    /* Check for global one-time settings at the start of the pattern, and remember
5833    the offset for later. */
5834    
5835    while (ptr[skipatstart] == '(' && ptr[skipatstart+1] == '*')
5836      {
5837      int newnl = 0;
5838      int newbsr = 0;
5839    
5840      if (strncmp((char *)(ptr+skipatstart+2), "CR)", 3) == 0)
5841        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
5842      else if (strncmp((char *)(ptr+skipatstart+2), "LF)", 3)  == 0)
5843        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
5844      else if (strncmp((char *)(ptr+skipatstart+2), "CRLF)", 5)  == 0)
5845        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
5846      else if (strncmp((char *)(ptr+skipatstart+2), "ANY)", 4) == 0)
5847        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
5848      else if (strncmp((char *)(ptr+skipatstart+2), "ANYCRLF)", 8)  == 0)
5849        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
5850    
5851      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_ANYCRLF)", 12) == 0)
5852        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
5853      else if (strncmp((char *)(ptr+skipatstart+2), "BSR_UNICODE)", 12) == 0)
5854        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
5855    
5856      if (newnl != 0)
5857        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
5858      else if (newbsr != 0)
5859        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
5860      else break;
5861      }
5862    
5863    /* Check validity of \R options. */
5864    
5865    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
5866      {
5867      case 0:
5868      case PCRE_BSR_ANYCRLF:
5869      case PCRE_BSR_UNICODE:
5870      break;
5871      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
5872      }
5873    
5874  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
5875  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
5876  "anycrlf". */  "anycrlf". */
5877    
5878  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
5879    {    {
5880    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
5881    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = '\r'; break;
5882    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = '\n'; break;
5883    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
# Line 5337  cd->hwm = cworkspace; Line 5940  cd->hwm = cworkspace;
5940  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
5941  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
5942  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
5943  cd->external_options = options;  cd->external_options = options;
5944    cd->external_flags = 0;
5945    
5946  /* 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
5947  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 5346  been put into the cd block so that they Line 5949  been put into the cd block so that they
5949  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
5950  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
5951    
5952    ptr += skipatstart;
5953  code = cworkspace;  code = cworkspace;
5954  *code = OP_BRA;  *code = OP_BRA;
5955  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
5956    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,    &code, &ptr, &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd,
5957    &length);    &length);
5958  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;  if (errorcode != 0) goto PCRE_EARLY_ERROR_RETURN;
5959    
# Line 5376  if (re == NULL) Line 5980  if (re == NULL)
5980    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
5981    }    }
5982    
5983  /* 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
5984  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
5985  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
5986  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
5987    pointers. */
5988    
5989  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
5990  re->size = size;  re->size = size;
5991  re->options = cd->external_options;  re->options = cd->external_options;
5992    re->flags = cd->external_flags;
5993  re->dummy1 = 0;  re->dummy1 = 0;
5994  re->first_byte = 0;  re->first_byte = 0;
5995  re->req_byte = 0;  re->req_byte = 0;
# Line 5408  codestart = cd->name_table + re->name_en Line 6014  codestart = cd->name_table + re->name_en
6014  cd->start_code = codestart;  cd->start_code = codestart;
6015  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6016  cd->req_varyopt = 0;  cd->req_varyopt = 0;
6017  cd->nopartial = FALSE;  cd->had_accept = FALSE;
6018    
6019  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6020  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
6021  of the function here. */  of the function here. */
6022    
6023  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6024  code = (uschar *)codestart;  code = (uschar *)codestart;
6025  *code = OP_BRA;  *code = OP_BRA;
6026  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6027    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6028  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6029  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6030    re->flags = cd->external_flags;
6031    
6032  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6033    
6034  /* 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. */
6035    
# Line 5464  if (errorcode != 0) Line 6071  if (errorcode != 0)
6071    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6072    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6073    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6074    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6075    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6076    return NULL;    return NULL;
6077    }    }
# Line 5493  if ((re->options & PCRE_ANCHORED) == 0) Line 6100  if ((re->options & PCRE_ANCHORED) == 0)
6100        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6101        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6102           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6103        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6104        }        }
6105      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6106        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6107      }      }
6108    }    }
6109    
# Line 5510  if (reqbyte >= 0 && Line 6117  if (reqbyte >= 0 &&
6117    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6118    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6119      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6120    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6121    }    }
6122    
6123  /* 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 5521  case when building a production library. Line 6128  case when building a production library.
6128  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6129    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6130    
6131  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 " : "");  
   }  
6132    
6133  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6134    {    {
6135    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6136    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5544  if ((re->options & PCRE_FIRSTSET) != 0) Line 6139  if ((re->options & PCRE_FIRSTSET) != 0)
6139      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6140    }    }
6141    
6142  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6143    {    {
6144    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6145    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5561  was compiled can be seen. */ Line 6156  was compiled can be seen. */
6156  if (code - codestart > length)  if (code - codestart > length)
6157    {    {
6158    (pcre_free)(re);    (pcre_free)(re);
6159    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6160    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6161    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6162    return NULL;    return NULL;

Legend:
Removed from v.175  
changed lines
  Added in v.268

  ViewVC Help
Powered by ViewVC 1.1.5