/[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 206 by ph10, Fri Aug 3 14:53:04 2007 UTC revision 392 by ph10, Tue Mar 17 21:30:30 2009 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2007 University of Cambridge             Copyright (c) 1997-2009 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 43  supporting internal functions that are n Line 43  supporting internal functions that are n
43    
44    
45  #ifdef HAVE_CONFIG_H  #ifdef HAVE_CONFIG_H
46  #include <config.h>  #include "config.h"
47  #endif  #endif
48    
49  #define NLBLOCK cd             /* Block containing newline information */  #define NLBLOCK cd             /* Block containing newline information */
# Line 97  are simple data values; negative values Line 97  are simple data values; negative values
97  on. Zero means further processing is needed (for things like \x), or the escape  on. Zero means further processing is needed (for things like \x), or the escape
98  is invalid. */  is invalid. */
99    
100  #ifndef EBCDIC  /* This is the "normal" table for ASCII systems */  #ifndef EBCDIC
101    
102    /* This is the "normal" table for ASCII systems or for EBCDIC systems running
103    in UTF-8 mode. */
104    
105  static const short int escapes[] = {  static const short int escapes[] = {
106       0,      0,      0,      0,      0,      0,      0,      0,   /* 0 - 7 */       0,                       0,
107       0,      0,    ':',    ';',    '<',    '=',    '>',    '?',   /* 8 - ? */       0,                       0,
108     '@', -ESC_A, -ESC_B, -ESC_C, -ESC_D, -ESC_E,      0, -ESC_G,   /* @ - G */       0,                       0,
109  -ESC_H,      0,      0, -ESC_K,      0,      0,      0,      0,   /* H - O */       0,                       0,
110  -ESC_P, -ESC_Q, -ESC_R, -ESC_S,      0,      0, -ESC_V, -ESC_W,   /* P - W */       0,                       0,
111  -ESC_X,      0, -ESC_Z,    '[',   '\\',    ']',    '^',    '_',   /* X - _ */       CHAR_COLON,              CHAR_SEMICOLON,
112     '`',      7, -ESC_b,      0, -ESC_d,  ESC_e,  ESC_f,      0,   /* ` - g */       CHAR_LESS_THAN_SIGN,     CHAR_EQUALS_SIGN,
113  -ESC_h,      0,      0, -ESC_k,      0,      0,  ESC_n,      0,   /* h - o */       CHAR_GREATER_THAN_SIGN,  CHAR_QUESTION_MARK,
114  -ESC_p,      0,  ESC_r, -ESC_s,  ESC_tee,    0, -ESC_v, -ESC_w,   /* p - w */       CHAR_COMMERCIAL_AT,      -ESC_A,
115       0,      0, -ESC_z                                            /* x - z */       -ESC_B,                  -ESC_C,
116         -ESC_D,                  -ESC_E,
117         0,                       -ESC_G,
118         -ESC_H,                  0,
119         0,                       -ESC_K,
120         0,                       0,
121         0,                       0,
122         -ESC_P,                  -ESC_Q,
123         -ESC_R,                  -ESC_S,
124         0,                       0,
125         -ESC_V,                  -ESC_W,
126         -ESC_X,                  0,
127         -ESC_Z,                  CHAR_LEFT_SQUARE_BRACKET,
128         CHAR_BACKSLASH,          CHAR_RIGHT_SQUARE_BRACKET,
129         CHAR_CIRCUMFLEX_ACCENT,  CHAR_UNDERSCORE,
130         CHAR_GRAVE_ACCENT,       7,
131         -ESC_b,                  0,
132         -ESC_d,                  ESC_e,
133         ESC_f,                   0,
134         -ESC_h,                  0,
135         0,                       -ESC_k,
136         0,                       0,
137         ESC_n,                   0,
138         -ESC_p,                  0,
139         ESC_r,                   -ESC_s,
140         ESC_tee,                 0,
141         -ESC_v,                  -ESC_w,
142         0,                       0,
143         -ESC_z
144  };  };
145    
146  #else           /* This is the "abnormal" table for EBCDIC systems */  #else
147    
148    /* This is the "abnormal" table for EBCDIC systems without UTF-8 support. */
149    
150  static const short int escapes[] = {  static const short int escapes[] = {
151  /*  48 */     0,     0,      0,     '.',    '<',   '(',    '+',    '|',  /*  48 */     0,     0,      0,     '.',    '<',   '(',    '+',    '|',
152  /*  50 */   '&',     0,      0,       0,      0,     0,      0,      0,  /*  50 */   '&',     0,      0,       0,      0,     0,      0,      0,
# Line 140  static const short int escapes[] = { Line 175  static const short int escapes[] = {
175  #endif  #endif
176    
177    
178  /* 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
179  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
180  as this is assumed for handling case independence. */  the number of relocations when a shared library is dynamically linked. The
181    string is built from string macros so that it works in UTF-8 mode on EBCDIC
182  static const char *const posix_names[] = {  platforms. */
183    "alpha", "lower", "upper",  
184    "alnum", "ascii", "blank", "cntrl", "digit", "graph",  typedef struct verbitem {
185    "print", "punct", "space", "word",  "xdigit" };    int   len;
186      int   op;
187    } verbitem;
188    
189    static const char verbnames[] =
190      STRING_ACCEPT0
191      STRING_COMMIT0
192      STRING_F0
193      STRING_FAIL0
194      STRING_PRUNE0
195      STRING_SKIP0
196      STRING_THEN;
197    
198    static const verbitem verbs[] = {
199      { 6, OP_ACCEPT },
200      { 6, OP_COMMIT },
201      { 1, OP_FAIL },
202      { 4, OP_FAIL },
203      { 5, OP_PRUNE },
204      { 4, OP_SKIP  },
205      { 4, OP_THEN  }
206    };
207    
208    static const int verbcount = sizeof(verbs)/sizeof(verbitem);
209    
210    
211    /* Tables of names of POSIX character classes and their lengths. The names are
212    now all in a single string, to reduce the number of relocations when a shared
213    library is dynamically loaded. The list of lengths is terminated by a zero
214    length entry. The first three must be alpha, lower, upper, as this is assumed
215    for handling case independence. */
216    
217    static const char posix_names[] =
218      STRING_alpha0 STRING_lower0 STRING_upper0 STRING_alnum0
219      STRING_ascii0 STRING_blank0 STRING_cntrl0 STRING_digit0
220      STRING_graph0 STRING_print0 STRING_punct0 STRING_space0
221      STRING_word0  STRING_xdigit;
222    
223  static const uschar posix_name_lengths[] = {  static const uschar posix_name_lengths[] = {
224    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 186  static const int posix_class_maps[] = { Line 257  static const int posix_class_maps[] = {
257  /* The texts of compile-time error messages. These are "char *" because they  /* The texts of compile-time error messages. These are "char *" because they
258  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
259  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
260  are no longer used. */  are no longer used. This used to be a table of strings, but in order to reduce
261    the number of relocations needed when a shared library is loaded dynamically,
262  static const char *error_texts[] = {  it is now one long string. We cannot use a table of offsets, because the
263    "no error",  lengths of inserts such as XSTRING(MAX_NAME_SIZE) are not known. Instead, we
264    "\\ at end of pattern",  simply count through to the one we want - this isn't a performance issue
265    "\\c at end of pattern",  because these strings are used only when there is a compilation error. */
266    "unrecognized character follows \\",  
267    "numbers out of order in {} quantifier",  static const char error_texts[] =
268      "no error\0"
269      "\\ at end of pattern\0"
270      "\\c at end of pattern\0"
271      "unrecognized character follows \\\0"
272      "numbers out of order in {} quantifier\0"
273    /* 5 */    /* 5 */
274    "number too big in {} quantifier",    "number too big in {} quantifier\0"
275    "missing terminating ] for character class",    "missing terminating ] for character class\0"
276    "invalid escape sequence in character class",    "invalid escape sequence in character class\0"
277    "range out of order in character class",    "range out of order in character class\0"
278    "nothing to repeat",    "nothing to repeat\0"
279    /* 10 */    /* 10 */
280    "operand of unlimited repeat could match the empty string",  /** DEAD **/    "operand of unlimited repeat could match the empty string\0"  /** DEAD **/
281    "internal error: unexpected repeat",    "internal error: unexpected repeat\0"
282    "unrecognized character after (?",    "unrecognized character after (? or (?-\0"
283    "POSIX named classes are supported only within a class",    "POSIX named classes are supported only within a class\0"
284    "missing )",    "missing )\0"
285    /* 15 */    /* 15 */
286    "reference to non-existent subpattern",    "reference to non-existent subpattern\0"
287    "erroffset passed as NULL",    "erroffset passed as NULL\0"
288    "unknown option bit(s) set",    "unknown option bit(s) set\0"
289    "missing ) after comment",    "missing ) after comment\0"
290    "parentheses nested too deeply",  /** DEAD **/    "parentheses nested too deeply\0"  /** DEAD **/
291    /* 20 */    /* 20 */
292    "regular expression is too large",    "regular expression is too large\0"
293    "failed to get memory",    "failed to get memory\0"
294    "unmatched parentheses",    "unmatched parentheses\0"
295    "internal error: code overflow",    "internal error: code overflow\0"
296    "unrecognized character after (?<",    "unrecognized character after (?<\0"
297    /* 25 */    /* 25 */
298    "lookbehind assertion is not fixed length",    "lookbehind assertion is not fixed length\0"
299    "malformed number or name after (?(",    "malformed number or name after (?(\0"
300    "conditional group contains more than two branches",    "conditional group contains more than two branches\0"
301    "assertion expected after (?(",    "assertion expected after (?(\0"
302    "(?R or (?[+-]digits must be followed by )",    "(?R or (?[+-]digits must be followed by )\0"
303    /* 30 */    /* 30 */
304    "unknown POSIX class name",    "unknown POSIX class name\0"
305    "POSIX collating elements are not supported",    "POSIX collating elements are not supported\0"
306    "this version of PCRE is not compiled with PCRE_UTF8 support",    "this version of PCRE is not compiled with PCRE_UTF8 support\0"
307    "spare error",  /** DEAD **/    "spare error\0"  /** DEAD **/
308    "character value in \\x{...} sequence is too large",    "character value in \\x{...} sequence is too large\0"
309    /* 35 */    /* 35 */
310    "invalid condition (?(0)",    "invalid condition (?(0)\0"
311    "\\C not allowed in lookbehind assertion",    "\\C not allowed in lookbehind assertion\0"
312    "PCRE does not support \\L, \\l, \\N, \\U, or \\u",    "PCRE does not support \\L, \\l, \\N, \\U, or \\u\0"
313    "number after (?C is > 255",    "number after (?C is > 255\0"
314    "closing ) for (?C expected",    "closing ) for (?C expected\0"
315    /* 40 */    /* 40 */
316    "recursive call could loop indefinitely",    "recursive call could loop indefinitely\0"
317    "unrecognized character after (?P",    "unrecognized character after (?P\0"
318    "syntax error in subpattern name (missing terminator)",    "syntax error in subpattern name (missing terminator)\0"
319    "two named subpatterns have the same name",    "two named subpatterns have the same name\0"
320    "invalid UTF-8 string",    "invalid UTF-8 string\0"
321    /* 45 */    /* 45 */
322    "support for \\P, \\p, and \\X has not been compiled",    "support for \\P, \\p, and \\X has not been compiled\0"
323    "malformed \\P or \\p sequence",    "malformed \\P or \\p sequence\0"
324    "unknown property name after \\P or \\p",    "unknown property name after \\P or \\p\0"
325    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)",    "subpattern name is too long (maximum " XSTRING(MAX_NAME_SIZE) " characters)\0"
326    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")",    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
327    /* 50 */    /* 50 */
328    "repeated subpattern is too long",    /** DEAD **/    "repeated subpattern is too long\0"    /** DEAD **/
329    "octal value is greater than \\377 (not in UTF-8 mode)",    "octal value is greater than \\377 (not in UTF-8 mode)\0"
330    "internal error: overran compiling workspace",    "internal error: overran compiling workspace\0"
331    "internal error: previously-checked referenced subpattern not found",    "internal error: previously-checked referenced subpattern not found\0"
332    "DEFINE group contains more than one branch",    "DEFINE group contains more than one branch\0"
333    /* 55 */    /* 55 */
334    "repeating a DEFINE group is not allowed",    "repeating a DEFINE group is not allowed\0"
335    "inconsistent NEWLINE options",    "inconsistent NEWLINE options\0"
336    "\\g is not followed by a braced name or an optionally braced non-zero number",    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"
337    "(?+ or (?- or (?(+ or (?(- must be followed by a non-zero number"    "a numbered reference must not be zero\0"
338  };    "(*VERB) with an argument is not supported\0"
339      /* 60 */
340      "(*VERB) not recognized\0"
341      "number is too big\0"
342      "subpattern name expected\0"
343      "digit expected after (?+\0"
344      "] is an invalid data character in JavaScript compatibility mode";
345    
346    
347  /* 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 278  For convenience, we use the same bit def Line 360  For convenience, we use the same bit def
360    
361  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
362    
363  #ifndef EBCDIC  /* This is the "normal" case, for ASCII systems */  #ifndef EBCDIC
364    
365    /* This is the "normal" case, for ASCII systems, and EBCDIC systems running in
366    UTF-8 mode. */
367    
368  static const unsigned char digitab[] =  static const unsigned char digitab[] =
369    {    {
370    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
# Line 314  static const unsigned char digitab[] = Line 400  static const unsigned char digitab[] =
400    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /* 240-247 */
401    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00};/* 248-255 */
402    
403  #else           /* This is the "abnormal" case, for EBCDIC systems */  #else
404    
405    /* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */
406    
407  static const unsigned char digitab[] =  static const unsigned char digitab[] =
408    {    {
409    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
# Line 395  static BOOL Line 484  static BOOL
484    
485    
486  /*************************************************  /*************************************************
487    *            Find an error text                  *
488    *************************************************/
489    
490    /* The error texts are now all in one long string, to save on relocations. As
491    some of the text is of unknown length, we can't use a table of offsets.
492    Instead, just count through the strings. This is not a performance issue
493    because it happens only when there has been a compilation error.
494    
495    Argument:   the error number
496    Returns:    pointer to the error string
497    */
498    
499    static const char *
500    find_error_text(int n)
501    {
502    const char *s = error_texts;
503    for (; n > 0; n--) while (*s++ != 0) {};
504    return s;
505    }
506    
507    
508    /*************************************************
509  *            Handle escapes                      *  *            Handle escapes                      *
510  *************************************************/  *************************************************/
511    
# Line 415  Arguments: Line 526  Arguments:
526    
527  Returns:         zero or positive => a data character  Returns:         zero or positive => a data character
528                   negative => a special escape sequence                   negative => a special escape sequence
529                   on error, errorptr is set                   on error, errorcodeptr is set
530  */  */
531    
532  static int  static int
# Line 433  ptr--;                            /* Set Line 544  ptr--;                            /* Set
544    
545  if (c == 0) *errorcodeptr = ERR1;  if (c == 0) *errorcodeptr = ERR1;
546    
547  /* Non-alphamerics are literals. For digits or letters, do an initial lookup in  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
548  a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
549  Otherwise further processing may be required. */  Otherwise further processing may be required. */
550    
551  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
552  else if (c < '0' || c > 'z') {}                           /* Not alphameric */  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */
553  else if ((i = escapes[c - '0']) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0) c = i;
554    
555  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
556  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphameric */  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */
557  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  c = i;
558  #endif  #endif
559    
# Line 458  else Line 569  else
569      /* A number of Perl escapes are not handled by PCRE. We give an explicit      /* A number of Perl escapes are not handled by PCRE. We give an explicit
570      error. */      error. */
571    
572      case 'l':      case CHAR_l:
573      case 'L':      case CHAR_L:
574      case 'N':      case CHAR_N:
575      case 'u':      case CHAR_u:
576      case 'U':      case CHAR_U:
577      *errorcodeptr = ERR37;      *errorcodeptr = ERR37;
578      break;      break;
579    
580      /* \g must be followed by a number, either plain or braced. If positive, it      /* \g must be followed by one of a number of specific things:
581      is an absolute backreference. If negative, it is a relative backreference.  
582      This is a Perl 5.10 feature. Perl 5.10 also supports \g{name} as a      (1) A number, either plain or braced. If positive, it is an absolute
583      reference to a named group. This is part of Perl's movement towards a      backreference. If negative, it is a relative backreference. This is a Perl
584      unified syntax for back references. As this is synonymous with \k{name}, we      5.10 feature.
585      fudge it up by pretending it really was \k. */  
586        (2) Perl 5.10 also supports \g{name} as a reference to a named group. This
587        is part of Perl's movement towards a unified syntax for back references. As
588        this is synonymous with \k{name}, we fudge it up by pretending it really
589        was \k.
590    
591        (3) For Oniguruma compatibility we also support \g followed by a name or a
592        number either in angle brackets or in single quotes. However, these are
593        (possibly recursive) subroutine calls, _not_ backreferences. Just return
594        the -ESC_g code (cf \k). */
595    
596        case CHAR_g:
597        if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
598          {
599          c = -ESC_g;
600          break;
601          }
602    
603        /* Handle the Perl-compatible cases */
604    
605      case 'g':      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
     if (ptr[1] == '{')  
606        {        {
607        const uschar *p;        const uschar *p;
608        for (p = ptr+2; *p != 0 && *p != '}'; p++)        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
609          if (*p != '-' && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;
610        if (*p != 0 && *p != '}')        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
611          {          {
612          c = -ESC_k;          c = -ESC_k;
613          break;          break;
# Line 489  else Line 617  else
617        }        }
618      else braced = FALSE;      else braced = FALSE;
619    
620      if (ptr[1] == '-')      if (ptr[1] == CHAR_MINUS)
621        {        {
622        negated = TRUE;        negated = TRUE;
623        ptr++;        ptr++;
# Line 498  else Line 626  else
626    
627      c = 0;      c = 0;
628      while ((digitab[ptr[1]] & ctype_digit) != 0)      while ((digitab[ptr[1]] & ctype_digit) != 0)
629        c = c * 10 + *(++ptr) - '0';        c = c * 10 + *(++ptr) - CHAR_0;
630    
631        if (c < 0)   /* Integer overflow */
632          {
633          *errorcodeptr = ERR61;
634          break;
635          }
636    
637      if (c == 0 || (braced && *(++ptr) != '}'))      if (braced && *(++ptr) != CHAR_RIGHT_CURLY_BRACKET)
638        {        {
639        *errorcodeptr = ERR57;        *errorcodeptr = ERR57;
640        return 0;        break;
641          }
642    
643        if (c == 0)
644          {
645          *errorcodeptr = ERR58;
646          break;
647        }        }
648    
649      if (negated)      if (negated)
# Line 511  else Line 651  else
651        if (c > bracount)        if (c > bracount)
652          {          {
653          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
654          return 0;          break;
655          }          }
656        c = bracount - (c - 1);        c = bracount - (c - 1);
657        }        }
# Line 531  else Line 671  else
671      value is greater than 377, the least significant 8 bits are taken. Inside a      value is greater than 377, the least significant 8 bits are taken. Inside a
672      character class, \ followed by a digit is always an octal number. */      character class, \ followed by a digit is always an octal number. */
673    
674      case '1': case '2': case '3': case '4': case '5':      case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4: case CHAR_5:
675      case '6': case '7': case '8': case '9':      case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:
676    
677      if (!isclass)      if (!isclass)
678        {        {
679        oldptr = ptr;        oldptr = ptr;
680        c -= '0';        c -= CHAR_0;
681        while ((digitab[ptr[1]] & ctype_digit) != 0)        while ((digitab[ptr[1]] & ctype_digit) != 0)
682          c = c * 10 + *(++ptr) - '0';          c = c * 10 + *(++ptr) - CHAR_0;
683          if (c < 0)    /* Integer overflow */
684            {
685            *errorcodeptr = ERR61;
686            break;
687            }
688        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
689          {          {
690          c = -(ESC_REF + c);          c = -(ESC_REF + c);
# Line 552  else Line 697  else
697      generates a binary zero byte and treats the digit as a following literal.      generates a binary zero byte and treats the digit as a following literal.
698      Thus we have to pull back the pointer by one. */      Thus we have to pull back the pointer by one. */
699    
700      if ((c = *ptr) >= '8')      if ((c = *ptr) >= CHAR_8)
701        {        {
702        ptr--;        ptr--;
703        c = 0;        c = 0;
# Line 565  else Line 710  else
710      to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more      to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more
711      than 3 octal digits. */      than 3 octal digits. */
712    
713      case '0':      case CHAR_0:
714      c -= '0';      c -= CHAR_0;
715      while(i++ < 2 && ptr[1] >= '0' && ptr[1] <= '7')      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
716          c = c * 8 + *(++ptr) - '0';          c = c * 8 + *(++ptr) - CHAR_0;
717      if (!utf8 && c > 255) *errorcodeptr = ERR51;      if (!utf8 && c > 255) *errorcodeptr = ERR51;
718      break;      break;
719    
# Line 576  else Line 721  else
721      than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is      than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is
722      treated as a data character. */      treated as a data character. */
723    
724      case 'x':      case CHAR_x:
725      if (ptr[1] == '{')      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
726        {        {
727        const uschar *pt = ptr + 2;        const uschar *pt = ptr + 2;
728        int count = 0;        int count = 0;
# Line 586  else Line 731  else
731        while ((digitab[*pt] & ctype_xdigit) != 0)        while ((digitab[*pt] & ctype_xdigit) != 0)
732          {          {
733          register int cc = *pt++;          register int cc = *pt++;
734          if (c == 0 && cc == '0') continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
735          count++;          count++;
736    
737  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
738          if (cc >= 'a') cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
739          c = (c << 4) + cc - ((cc < 'A')? '0' : ('A' - 10));          c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
740  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
741          if (cc >= 'a' && cc <= 'z') cc += 64;  /* Convert to upper case */          if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
742          c = (c << 4) + cc - ((cc >= '0')? '0' : ('A' - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
743  #endif  #endif
744          }          }
745    
746        if (*pt == '}')        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
747          {          {
748          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;
749          ptr = pt;          ptr = pt;
# Line 614  else Line 759  else
759      c = 0;      c = 0;
760      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)
761        {        {
762        int cc;                               /* Some compilers don't like ++ */        int cc;                                  /* Some compilers don't like */
763        cc = *(++ptr);                        /* in initializers */        cc = *(++ptr);                           /* ++ in initializers */
764  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
765        if (cc >= 'a') cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
766        c = c * 16 + cc - ((cc < 'A')? '0' : ('A' - 10));        c = c * 16 + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
767  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
768        if (cc <= 'z') cc += 64;              /* Convert to upper case */        if (cc <= CHAR_z) cc += 64;              /* Convert to upper case */
769        c = c * 16 + cc - ((cc >= '0')? '0' : ('A' - 10));        c = c * 16 + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
770  #endif  #endif
771        }        }
772      break;      break;
# Line 630  else Line 775  else
775      This coding is ASCII-specific, but then the whole concept of \cx is      This coding is ASCII-specific, but then the whole concept of \cx is
776      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */      ASCII-specific. (However, an EBCDIC equivalent has now been added.) */
777    
778      case 'c':      case CHAR_c:
779      c = *(++ptr);      c = *(++ptr);
780      if (c == 0)      if (c == 0)
781        {        {
782        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
783        return 0;        break;
784        }        }
785    
786  #ifndef EBCDIC  /* ASCII coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
787      if (c >= 'a' && c <= 'z') c -= 32;      if (c >= CHAR_a && c <= CHAR_z) c -= 32;
788      c ^= 0x40;      c ^= 0x40;
789  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
790      if (c >= 'a' && c <= 'z') c += 64;      if (c >= CHAR_a && c <= CHAR_z) c += 64;
791      c ^= 0xC0;      c ^= 0xC0;
792  #endif  #endif
793      break;      break;
794    
795      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any      /* PCRE_EXTRA enables extensions to Perl in the matter of escapes. Any
796      other alphameric following \ is an error if PCRE_EXTRA was set; otherwise,      other alphanumeric following \ is an error if PCRE_EXTRA was set;
797      for Perl compatibility, it is a literal. This code looks a bit odd, but      otherwise, for Perl compatibility, it is a literal. This code looks a bit
798      there used to be some cases other than the default, and there may be again      odd, but there used to be some cases other than the default, and there may
799      in future, so I haven't "optimized" it. */      be again in future, so I haven't "optimized" it. */
800    
801      default:      default:
802      if ((options & PCRE_EXTRA) != 0) switch(c)      if ((options & PCRE_EXTRA) != 0) switch(c)
# Line 704  if (c == 0) goto ERROR_RETURN; Line 849  if (c == 0) goto ERROR_RETURN;
849  /* \P or \p can be followed by a name in {}, optionally preceded by ^ for  /* \P or \p can be followed by a name in {}, optionally preceded by ^ for
850  negation. */  negation. */
851    
852  if (c == '{')  if (c == CHAR_LEFT_CURLY_BRACKET)
853    {    {
854    if (ptr[1] == '^')    if (ptr[1] == CHAR_CIRCUMFLEX_ACCENT)
855      {      {
856      *negptr = TRUE;      *negptr = TRUE;
857      ptr++;      ptr++;
# Line 715  if (c == '{') Line 860  if (c == '{')
860      {      {
861      c = *(++ptr);      c = *(++ptr);
862      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
863      if (c == '}') break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
864      name[i] = c;      name[i] = c;
865      }      }
866    if (c !='}') goto ERROR_RETURN;    if (c != CHAR_RIGHT_CURLY_BRACKET) goto ERROR_RETURN;
867    name[i] = 0;    name[i] = 0;
868    }    }
869    
# Line 740  top = _pcre_utt_size; Line 885  top = _pcre_utt_size;
885  while (bot < top)  while (bot < top)
886    {    {
887    i = (bot + top) >> 1;    i = (bot + top) >> 1;
888    c = strcmp(name, _pcre_utt[i].name);    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);
889    if (c == 0)    if (c == 0)
890      {      {
891      *dptr = _pcre_utt[i].value;      *dptr = _pcre_utt[i].value;
# Line 783  is_counted_repeat(const uschar *p) Line 928  is_counted_repeat(const uschar *p)
928  {  {
929  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;
930  while ((digitab[*p] & ctype_digit) != 0) p++;  while ((digitab[*p] & ctype_digit) != 0) p++;
931  if (*p == '}') return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
932    
933  if (*p++ != ',') return FALSE;  if (*p++ != CHAR_COMMA) return FALSE;
934  if (*p == '}') return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
935    
936  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;
937  while ((digitab[*p] & ctype_digit) != 0) p++;  while ((digitab[*p] & ctype_digit) != 0) p++;
938    
939  return (*p == '}');  return (*p == CHAR_RIGHT_CURLY_BRACKET);
940  }  }
941    
942    
# Line 824  int max = -1; Line 969  int max = -1;
969  /* Read the minimum value and do a paranoid check: a negative value indicates  /* Read the minimum value and do a paranoid check: a negative value indicates
970  an integer overflow. */  an integer overflow. */
971    
972  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - '0';  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;
973  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
974    {    {
975    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 834  if (min < 0 || min > 65535) Line 979  if (min < 0 || min > 65535)
979  /* Read the maximum value if there is one, and again do a paranoid on its size.  /* Read the maximum value if there is one, and again do a paranoid on its size.
980  Also, max must not be less than min. */  Also, max must not be less than min. */
981    
982  if (*p == '}') max = min; else  if (*p == CHAR_RIGHT_CURLY_BRACKET) max = min; else
983    {    {
984    if (*(++p) != '}')    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
985      {      {
986      max = 0;      max = 0;
987      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - '0';      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;
988      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
989        {        {
990        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 876  be terminated by '>' because that is che Line 1021  be terminated by '>' because that is che
1021    
1022  Arguments:  Arguments:
1023    ptr          current position in the pattern    ptr          current position in the pattern
1024    count        current count of capturing parens so far encountered    cd           compile background data
1025    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1026    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1027    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
# Line 885  Returns:       the number of the named s Line 1030  Returns:       the number of the named s
1030  */  */
1031    
1032  static int  static int
1033  find_parens(const uschar *ptr, int count, const uschar *name, int lorn,  find_parens(const uschar *ptr, compile_data *cd, const uschar *name, int lorn,
1034    BOOL xmode)    BOOL xmode)
1035  {  {
1036  const uschar *thisname;  const uschar *thisname;
1037    int count = cd->bracount;
1038    
1039  for (; *ptr != 0; ptr++)  for (; *ptr != 0; ptr++)
1040    {    {
# Line 896  for (; *ptr != 0; ptr++) Line 1042  for (; *ptr != 0; ptr++)
1042    
1043    /* Skip over backslashed characters and also entire \Q...\E */    /* Skip over backslashed characters and also entire \Q...\E */
1044    
1045    if (*ptr == '\\')    if (*ptr == CHAR_BACKSLASH)
1046      {      {
1047      if (*(++ptr) == 0) return -1;      if (*(++ptr) == 0) return -1;
1048      if (*ptr == 'Q') for (;;)      if (*ptr == CHAR_Q) for (;;)
1049        {        {
1050        while (*(++ptr) != 0 && *ptr != '\\');        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};
1051        if (*ptr == 0) return -1;        if (*ptr == 0) return -1;
1052        if (*(++ptr) == 'E') break;        if (*(++ptr) == CHAR_E) break;
1053        }        }
1054      continue;      continue;
1055      }      }
1056    
1057    /* Skip over character classes */    /* Skip over character classes; this logic must be similar to the way they
1058      are handled for real. If the first character is '^', skip it. Also, if the
1059      first few characters (either before or after ^) are \Q\E or \E we skip them
1060      too. This makes for compatibility with Perl. Note the use of STR macros to
1061      encode "Q\\E" so that it works in UTF-8 on EBCDIC platforms. */
1062    
1063    if (*ptr == '[')    if (*ptr == CHAR_LEFT_SQUARE_BRACKET)
1064      {      {
1065      while (*(++ptr) != ']')      BOOL negate_class = FALSE;
1066        for (;;)
1067        {        {
1068        if (*ptr == '\\')        int c = *(++ptr);
1069          if (c == CHAR_BACKSLASH)
1070            {
1071            if (ptr[1] == CHAR_E)
1072              ptr++;
1073            else if (strncmp((const char *)ptr+1,
1074                     STR_Q STR_BACKSLASH STR_E, 3) == 0)
1075              ptr += 3;
1076            else
1077              break;
1078            }
1079          else if (!negate_class && c == CHAR_CIRCUMFLEX_ACCENT)
1080            negate_class = TRUE;
1081          else break;
1082          }
1083    
1084        /* If the next character is ']', it is a data character that must be
1085        skipped, except in JavaScript compatibility mode. */
1086    
1087        if (ptr[1] == CHAR_RIGHT_SQUARE_BRACKET &&
1088            (cd->external_options & PCRE_JAVASCRIPT_COMPAT) == 0)
1089          ptr++;
1090    
1091        while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1092          {
1093          if (*ptr == 0) return -1;
1094          if (*ptr == CHAR_BACKSLASH)
1095          {          {
1096          if (*(++ptr) == 0) return -1;          if (*(++ptr) == 0) return -1;
1097          if (*ptr == 'Q') for (;;)          if (*ptr == CHAR_Q) for (;;)
1098            {            {
1099            while (*(++ptr) != 0 && *ptr != '\\');            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};
1100            if (*ptr == 0) return -1;            if (*ptr == 0) return -1;
1101            if (*(++ptr) == 'E') break;            if (*(++ptr) == CHAR_E) break;
1102            }            }
1103          continue;          continue;
1104          }          }
# Line 931  for (; *ptr != 0; ptr++) Line 1108  for (; *ptr != 0; ptr++)
1108    
1109    /* Skip comments in /x mode */    /* Skip comments in /x mode */
1110    
1111    if (xmode && *ptr == '#')    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1112      {      {
1113      while (*(++ptr) != 0 && *ptr != '\n');      while (*(++ptr) != 0 && *ptr != CHAR_NL) {};
1114      if (*ptr == 0) return -1;      if (*ptr == 0) return -1;
1115      continue;      continue;
1116      }      }
1117    
1118    /* An opening parens must now be a real metacharacter */    /* An opening parens must now be a real metacharacter */
1119    
1120    if (*ptr != '(') continue;    if (*ptr != CHAR_LEFT_PARENTHESIS) continue;
1121    if (ptr[1] != '?')    if (ptr[1] != CHAR_QUESTION_MARK && ptr[1] != CHAR_ASTERISK)
1122      {      {
1123      count++;      count++;
1124      if (name == NULL && count == lorn) return count;      if (name == NULL && count == lorn) return count;
# Line 949  for (; *ptr != 0; ptr++) Line 1126  for (; *ptr != 0; ptr++)
1126      }      }
1127    
1128    ptr += 2;    ptr += 2;
1129    if (*ptr == 'P') ptr++;                      /* Allow optional P */    if (*ptr == CHAR_P) ptr++;                      /* Allow optional P */
1130    
1131    /* We have to disambiguate (?<! and (?<= from (?<name> */    /* We have to disambiguate (?<! and (?<= from (?<name> */
1132    
1133    if ((*ptr != '<' || ptr[1] == '!' || ptr[1] == '=') &&    if ((*ptr != CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_EXCLAMATION_MARK ||
1134         *ptr != '\'')        ptr[1] == CHAR_EQUALS_SIGN) && *ptr != CHAR_APOSTROPHE)
1135      continue;      continue;
1136    
1137    count++;    count++;
1138    
1139    if (name == NULL && count == lorn) return count;    if (name == NULL && count == lorn) return count;
1140    term = *ptr++;    term = *ptr++;
1141    if (term == '<') term = '>';    if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1142    thisname = ptr;    thisname = ptr;
1143    while (*ptr != term) ptr++;    while (*ptr != term) ptr++;
1144    if (name != NULL && lorn == ptr - thisname &&    if (name != NULL && lorn == ptr - thisname &&
# Line 1069  for (;;) Line 1246  for (;;)
1246    {    {
1247    int d;    int d;
1248    register int op = *cc;    register int op = *cc;
   
1249    switch (op)    switch (op)
1250      {      {
1251      case OP_CBRA:      case OP_CBRA:
# Line 1158  for (;;) Line 1334  for (;;)
1334    
1335      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1336      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1337        if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;
1338      cc += 4;      cc += 4;
1339      break;      break;
1340    
# Line 1175  for (;;) Line 1352  for (;;)
1352      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
1353      case OP_WORDCHAR:      case OP_WORDCHAR:
1354      case OP_ANY:      case OP_ANY:
1355        case OP_ALLANY:
1356      branchlength++;      branchlength++;
1357      cc++;      cc++;
1358      break;      break;
# Line 1266  for (;;) Line 1444  for (;;)
1444      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1445      }      }
1446    
1447    /* 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
1448    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
1449    arrange to skip the extra bytes. */    two bytes of parameters. */
1450    
1451    else    else
1452      {      {
1453        switch(c)
1454          {
1455          case OP_TYPESTAR:
1456          case OP_TYPEMINSTAR:
1457          case OP_TYPEPLUS:
1458          case OP_TYPEMINPLUS:
1459          case OP_TYPEQUERY:
1460          case OP_TYPEMINQUERY:
1461          case OP_TYPEPOSSTAR:
1462          case OP_TYPEPOSPLUS:
1463          case OP_TYPEPOSQUERY:
1464          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1465          break;
1466    
1467          case OP_TYPEUPTO:
1468          case OP_TYPEMINUPTO:
1469          case OP_TYPEEXACT:
1470          case OP_TYPEPOSUPTO:
1471          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1472          break;
1473          }
1474    
1475        /* Add in the fixed length from the table */
1476    
1477      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1478    
1479      /* In UTF-8 mode, opcodes that are followed by a character may be followed by
1480      a multi-byte character. The length in the table is a minimum, so we have to
1481      arrange to skip the extra bytes. */
1482    
1483  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1484      if (utf8) switch(c)      if (utf8) switch(c)
1485        {        {
# Line 1294  for (;;) Line 1501  for (;;)
1501        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1502        break;        break;
1503        }        }
1504    #else
1505        (void)(utf8);  /* Keep compiler happy by referencing function argument */
1506  #endif  #endif
1507      }      }
1508    }    }
# Line 1330  for (;;) Line 1539  for (;;)
1539    
1540    if (c == OP_XCLASS) code += GET(code, 1);    if (c == OP_XCLASS) code += GET(code, 1);
1541    
1542    /* 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
1543    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
1544    The length in the table is a minimum, so we have to arrange to skip the extra    two bytes of parameters. */
   bytes. */  
1545    
1546    else    else
1547      {      {
1548        switch(c)
1549          {
1550          case OP_TYPESTAR:
1551          case OP_TYPEMINSTAR:
1552          case OP_TYPEPLUS:
1553          case OP_TYPEMINPLUS:
1554          case OP_TYPEQUERY:
1555          case OP_TYPEMINQUERY:
1556          case OP_TYPEPOSSTAR:
1557          case OP_TYPEPOSPLUS:
1558          case OP_TYPEPOSQUERY:
1559          if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1560          break;
1561    
1562          case OP_TYPEPOSUPTO:
1563          case OP_TYPEUPTO:
1564          case OP_TYPEMINUPTO:
1565          case OP_TYPEEXACT:
1566          if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1567          break;
1568          }
1569    
1570        /* Add in the fixed length from the table */
1571    
1572      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1573    
1574        /* In UTF-8 mode, opcodes that are followed by a character may be followed
1575        by a multi-byte character. The length in the table is a minimum, so we have
1576        to arrange to skip the extra bytes. */
1577    
1578  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1579      if (utf8) switch(c)      if (utf8) switch(c)
1580        {        {
# Line 1359  for (;;) Line 1596  for (;;)
1596        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];
1597        break;        break;
1598        }        }
1599    #else
1600        (void)(utf8);  /* Keep compiler happy by referencing function argument */
1601  #endif  #endif
1602      }      }
1603    }    }
# Line 1374  for (;;) Line 1613  for (;;)
1613  can match the empty string or not. It is called from could_be_empty()  can match the empty string or not. It is called from could_be_empty()
1614  below and from compile_branch() when checking for an unlimited repeat of a  below and from compile_branch() when checking for an unlimited repeat of a
1615  group that can match nothing. Note that first_significant_code() skips over  group that can match nothing. Note that first_significant_code() skips over
1616  assertions. If we hit an unclosed bracket, we return "empty" - this means we've  backward and negative forward assertions when its final argument is TRUE. If we
1617  struck an inner bracket whose current branch will already have been scanned.  hit an unclosed bracket, we return "empty" - this means we've struck an inner
1618    bracket whose current branch will already have been scanned.
1619    
1620  Arguments:  Arguments:
1621    code        points to start of search    code        points to start of search
# Line 1397  for (code = first_significant_code(code Line 1637  for (code = first_significant_code(code
1637    
1638    c = *code;    c = *code;
1639    
1640      /* Skip over forward assertions; the other assertions are skipped by
1641      first_significant_code() with a TRUE final argument. */
1642    
1643      if (c == OP_ASSERT)
1644        {
1645        do code += GET(code, 1); while (*code == OP_ALT);
1646        c = *code;
1647        continue;
1648        }
1649    
1650    /* Groups with zero repeats can of course be empty; skip them. */    /* Groups with zero repeats can of course be empty; skip them. */
1651    
1652    if (c == OP_BRAZERO || c == OP_BRAMINZERO)    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO)
1653      {      {
1654      code += _pcre_OP_lengths[c];      code += _pcre_OP_lengths[c];
1655      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
# Line 1433  for (code = first_significant_code(code Line 1683  for (code = first_significant_code(code
1683    
1684    switch (c)    switch (c)
1685      {      {
1686      /* Check for quantifiers after a class */      /* Check for quantifiers after a class. XCLASS is used for classes that
1687        cannot be represented just by a bit map. This includes negated single
1688        high-valued characters. The length in _pcre_OP_lengths[] is zero; the
1689        actual length is stored in the compiled code, so we must update "code"
1690        here. */
1691    
1692  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
1693      case OP_XCLASS:      case OP_XCLASS:
1694      ccode = code + GET(code, 1);      ccode = code += GET(code, 1);
1695      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
1696  #endif  #endif
1697    
# Line 1481  for (code = first_significant_code(code Line 1735  for (code = first_significant_code(code
1735      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
1736      case OP_WORDCHAR:      case OP_WORDCHAR:
1737      case OP_ANY:      case OP_ANY:
1738        case OP_ALLANY:
1739      case OP_ANYBYTE:      case OP_ANYBYTE:
1740      case OP_CHAR:      case OP_CHAR:
1741      case OP_CHARNC:      case OP_CHARNC:
# Line 1499  for (code = first_significant_code(code Line 1754  for (code = first_significant_code(code
1754      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1755      return FALSE;      return FALSE;
1756    
1757        /* These are going to continue, as they may be empty, but we have to
1758        fudge the length for the \p and \P cases. */
1759    
1760        case OP_TYPESTAR:
1761        case OP_TYPEMINSTAR:
1762        case OP_TYPEPOSSTAR:
1763        case OP_TYPEQUERY:
1764        case OP_TYPEMINQUERY:
1765        case OP_TYPEPOSQUERY:
1766        if (code[1] == OP_PROP || code[1] == OP_NOTPROP) code += 2;
1767        break;
1768    
1769        /* Same for these */
1770    
1771        case OP_TYPEUPTO:
1772        case OP_TYPEMINUPTO:
1773        case OP_TYPEPOSUPTO:
1774        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;
1775        break;
1776    
1777      /* End of branch */      /* End of branch */
1778    
1779      case OP_KET:      case OP_KET:
# Line 1568  return TRUE; Line 1843  return TRUE;
1843  *************************************************/  *************************************************/
1844    
1845  /* This function is called when the sequence "[:" or "[." or "[=" is  /* This function is called when the sequence "[:" or "[." or "[=" is
1846  encountered in a character class. It checks whether this is followed by an  encountered in a character class. It checks whether this is followed by a
1847  optional ^ and then a sequence of letters, terminated by a matching ":]" or  sequence of characters terminated by a matching ":]" or ".]" or "=]". If we
1848  ".]" or "=]".  reach an unescaped ']' without the special preceding character, return FALSE.
1849    
1850    Originally, this function only recognized a sequence of letters between the
1851    terminators, but it seems that Perl recognizes any sequence of characters,
1852    though of course unknown POSIX names are subsequently rejected. Perl gives an
1853    "Unknown POSIX class" error for [:f\oo:] for example, where previously PCRE
1854    didn't consider this to be a POSIX class. Likewise for [:1234:].
1855    
1856    The problem in trying to be exactly like Perl is in the handling of escapes. We
1857    have to be sure that [abc[:x\]pqr] is *not* treated as containing a POSIX
1858    class, but [abc[:x\]pqr:]] is (so that an error can be generated). The code
1859    below handles the special case of \], but does not try to do any other escape
1860    processing. This makes it different from Perl for cases such as [:l\ower:]
1861    where Perl recognizes it as the POSIX class "lower" but PCRE does not recognize
1862    "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
1863    I think.
1864    
1865  Argument:  Arguments:
1866    ptr      pointer to the initial [    ptr      pointer to the initial [
1867    endptr   where to return the end pointer    endptr   where to return the end pointer
   cd       pointer to compile data  
1868    
1869  Returns:   TRUE or FALSE  Returns:   TRUE or FALSE
1870  */  */
1871    
1872  static BOOL  static BOOL
1873  check_posix_syntax(const uschar *ptr, const uschar **endptr, compile_data *cd)  check_posix_syntax(const uschar *ptr, const uschar **endptr)
1874  {  {
1875  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
1876  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
1877  if (*(++ptr) == '^') ptr++;  for (++ptr; *ptr != 0; ptr++)
 while ((cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;  
 if (*ptr == terminator && ptr[1] == ']')  
1878    {    {
1879    *endptr = ptr;    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else
1880    return TRUE;      {
1881        if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
1882        if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
1883          {
1884          *endptr = ptr;
1885          return TRUE;
1886          }
1887        }
1888    }    }
1889  return FALSE;  return FALSE;
1890  }  }
# Line 1615  Returns:     a value representing the na Line 1909  Returns:     a value representing the na
1909  static int  static int
1910  check_posix_name(const uschar *ptr, int len)  check_posix_name(const uschar *ptr, int len)
1911  {  {
1912    const char *pn = posix_names;
1913  register int yield = 0;  register int yield = 0;
1914  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
1915    {    {
1916    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
1917      strncmp((const char *)ptr, posix_names[yield], len) == 0) return yield;      strncmp((const char *)ptr, pn, len) == 0) return yield;
1918      pn += posix_name_lengths[yield] + 1;
1919    yield++;    yield++;
1920    }    }
1921  return -1;  return -1;
# Line 1634  return -1; Line 1930  return -1;
1930  that is referenced. This means that groups can be replicated for fixed  that is referenced. This means that groups can be replicated for fixed
1931  repetition simply by copying (because the recursion is allowed to refer to  repetition simply by copying (because the recursion is allowed to refer to
1932  earlier groups that are outside the current group). However, when a group is  earlier groups that are outside the current group). However, when a group is
1933  optional (i.e. the minimum quantifier is zero), OP_BRAZERO is inserted before  optional (i.e. the minimum quantifier is zero), OP_BRAZERO or OP_SKIPZERO is
1934  it, after it has been compiled. This means that any OP_RECURSE items within it  inserted before it, after it has been compiled. This means that any OP_RECURSE
1935  that refer to the group itself or any contained groups have to have their  items within it that refer to the group itself or any contained groups have to
1936  offsets adjusted. That one of the jobs of this function. Before it is called,  have their offsets adjusted. That one of the jobs of this function. Before it
1937  the partially compiled regex must be temporarily terminated with OP_END.  is called, the partially compiled regex must be temporarily terminated with
1938    OP_END.
1939    
1940  This function has been extended with the possibility of forward references for  This function has been extended with the possibility of forward references for
1941  recursions and subroutine calls. It must also check the list of such references  recursions and subroutine calls. It must also check the list of such references
# Line 1661  adjust_recurse(uschar *group, int adjust Line 1958  adjust_recurse(uschar *group, int adjust
1958    uschar *save_hwm)    uschar *save_hwm)
1959  {  {
1960  uschar *ptr = group;  uschar *ptr = group;
1961    
1962  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)
1963    {    {
1964    int offset;    int offset;
# Line 1772  get_othercase_range(unsigned int *cptr, Line 2070  get_othercase_range(unsigned int *cptr,
2070  unsigned int c, othercase, next;  unsigned int c, othercase, next;
2071    
2072  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2073    { if ((othercase = _pcre_ucp_othercase(c)) != NOTACHAR) break; }    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }
2074    
2075  if (c > d) return FALSE;  if (c > d) return FALSE;
2076    
# Line 1781  next = othercase + 1; Line 2079  next = othercase + 1;
2079    
2080  for (++c; c <= d; c++)  for (++c; c <= d; c++)
2081    {    {
2082    if (_pcre_ucp_othercase(c) != next) break;    if (UCD_OTHERCASE(c) != next) break;
2083    next++;    next++;
2084    }    }
2085    
# Line 1827  if ((options & PCRE_EXTENDED) != 0) Line 2125  if ((options & PCRE_EXTENDED) != 0)
2125    for (;;)    for (;;)
2126      {      {
2127      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2128      if (*ptr == '#')      if (*ptr == CHAR_NUMBER_SIGN)
2129        {        {
2130        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2131          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
# Line 1839  if ((options & PCRE_EXTENDED) != 0) Line 2137  if ((options & PCRE_EXTENDED) != 0)
2137  /* If the next item is one that we can handle, get its value. A non-negative  /* If the next item is one that we can handle, get its value. A non-negative
2138  value is a character, a negative value is an escape value. */  value is a character, a negative value is an escape value. */
2139    
2140  if (*ptr == '\\')  if (*ptr == CHAR_BACKSLASH)
2141    {    {
2142    int temperrorcode = 0;    int temperrorcode = 0;
2143    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);
# Line 1864  if ((options & PCRE_EXTENDED) != 0) Line 2162  if ((options & PCRE_EXTENDED) != 0)
2162    for (;;)    for (;;)
2163      {      {
2164      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2165      if (*ptr == '#')      if (*ptr == CHAR_NUMBER_SIGN)
2166        {        {
2167        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2168          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
# Line 1875  if ((options & PCRE_EXTENDED) != 0) Line 2173  if ((options & PCRE_EXTENDED) != 0)
2173    
2174  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
2175    
2176  if (*ptr == '*' || *ptr == '?' || strncmp((char *)ptr, "{0,", 3) == 0)  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
2177    return FALSE;    strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
2178        return FALSE;
2179    
2180  /* Now compare the next item with the previous opcode. If the previous is a  /* Now compare the next item with the previous opcode. If the previous is a
2181  positive single character match, "item" either contains the character or, if  positive single character match, "item" either contains the character or, if
# Line 1891  if (next >= 0) switch(op_code) Line 2190  if (next >= 0) switch(op_code)
2190    case OP_CHAR:    case OP_CHAR:
2191  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2192    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }    if (utf8 && item > 127) { GETCHAR(item, utf8_char); }
2193    #else
2194      (void)(utf8_char);  /* Keep compiler happy by referencing function argument */
2195  #endif  #endif
2196    return item != next;    return item != next;
2197    
# Line 1909  if (next >= 0) switch(op_code) Line 2210  if (next >= 0) switch(op_code)
2210      unsigned int othercase;      unsigned int othercase;
2211      if (next < 128) othercase = cd->fcc[next]; else      if (next < 128) othercase = cd->fcc[next]; else
2212  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2213      othercase = _pcre_ucp_othercase((unsigned int)next);      othercase = UCD_OTHERCASE((unsigned int)next);
2214  #else  #else
2215      othercase = NOTACHAR;      othercase = NOTACHAR;
2216  #endif  #endif
# Line 1922  if (next >= 0) switch(op_code) Line 2223  if (next >= 0) switch(op_code)
2223    /* For OP_NOT, "item" must be a single-byte character. */    /* For OP_NOT, "item" must be a single-byte character. */
2224    
2225    case OP_NOT:    case OP_NOT:
   if (next < 0) return FALSE;  /* Not a character */  
2226    if (item == next) return TRUE;    if (item == next) return TRUE;
2227    if ((options & PCRE_CASELESS) == 0) return FALSE;    if ((options & PCRE_CASELESS) == 0) return FALSE;
2228  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
# Line 1931  if (next >= 0) switch(op_code) Line 2231  if (next >= 0) switch(op_code)
2231      unsigned int othercase;      unsigned int othercase;
2232      if (next < 128) othercase = cd->fcc[next]; else      if (next < 128) othercase = cd->fcc[next]; else
2233  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
2234      othercase = _pcre_ucp_othercase(next);      othercase = UCD_OTHERCASE(next);
2235  #else  #else
2236      othercase = NOTACHAR;      othercase = NOTACHAR;
2237  #endif  #endif
# Line 2185  uschar classbits[32]; Line 2485  uschar classbits[32];
2485  BOOL class_utf8;  BOOL class_utf8;
2486  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf8 = (options & PCRE_UTF8) != 0;
2487  uschar *class_utf8data;  uschar *class_utf8data;
2488    uschar *class_utf8data_base;
2489  uschar utf8_char[6];  uschar utf8_char[6];
2490  #else  #else
2491  BOOL utf8 = FALSE;  BOOL utf8 = FALSE;
# Line 2224  req_caseopt = ((options & PCRE_CASELESS) Line 2525  req_caseopt = ((options & PCRE_CASELESS)
2525  for (;; ptr++)  for (;; ptr++)
2526    {    {
2527    BOOL negate_class;    BOOL negate_class;
2528      BOOL should_flip_negation;
2529    BOOL possessive_quantifier;    BOOL possessive_quantifier;
2530    BOOL is_quantifier;    BOOL is_quantifier;
2531    BOOL is_recurse;    BOOL is_recurse;
# Line 2311  for (;; ptr++) Line 2613  for (;; ptr++)
2613    
2614    if (inescq && c != 0)    if (inescq && c != 0)
2615      {      {
2616      if (c == '\\' && ptr[1] == 'E')      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
2617        {        {
2618        inescq = FALSE;        inescq = FALSE;
2619        ptr++;        ptr++;
# Line 2337  for (;; ptr++) Line 2639  for (;; ptr++)
2639    /* Fill in length of a previous callout, except when the next thing is    /* Fill in length of a previous callout, except when the next thing is
2640    a quantifier. */    a quantifier. */
2641    
2642    is_quantifier = c == '*' || c == '+' || c == '?' ||    is_quantifier =
2643      (c == '{' && is_counted_repeat(ptr+1));      c == CHAR_ASTERISK || c == CHAR_PLUS || c == CHAR_QUESTION_MARK ||
2644        (c == CHAR_LEFT_CURLY_BRACKET && is_counted_repeat(ptr+1));
2645    
2646    if (!is_quantifier && previous_callout != NULL &&    if (!is_quantifier && previous_callout != NULL &&
2647         after_manual_callout-- <= 0)         after_manual_callout-- <= 0)
# Line 2353  for (;; ptr++) Line 2656  for (;; ptr++)
2656    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
2657      {      {
2658      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if ((cd->ctypes[c] & ctype_space) != 0) continue;
2659      if (c == '#')      if (c == CHAR_NUMBER_SIGN)
2660        {        {
2661        while (*(++ptr) != 0)        while (*(++ptr) != 0)
2662          {          {
# Line 2378  for (;; ptr++) Line 2681  for (;; ptr++)
2681      {      {
2682      /* ===================================================================*/      /* ===================================================================*/
2683      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
2684      case '|':                      /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
2685      case ')':      case CHAR_RIGHT_PARENTHESIS:
2686      *firstbyteptr = firstbyte;      *firstbyteptr = firstbyte;
2687      *reqbyteptr = reqbyte;      *reqbyteptr = reqbyte;
2688      *codeptr = code;      *codeptr = code;
# Line 2401  for (;; ptr++) Line 2704  for (;; ptr++)
2704      /* Handle single-character metacharacters. In multiline mode, ^ disables      /* Handle single-character metacharacters. In multiline mode, ^ disables
2705      the setting of any following char as a first character. */      the setting of any following char as a first character. */
2706    
2707      case '^':      case CHAR_CIRCUMFLEX_ACCENT:
2708      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
2709        {        {
2710        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
# Line 2410  for (;; ptr++) Line 2713  for (;; ptr++)
2713      *code++ = OP_CIRC;      *code++ = OP_CIRC;
2714      break;      break;
2715    
2716      case '$':      case CHAR_DOLLAR_SIGN:
2717      previous = NULL;      previous = NULL;
2718      *code++ = OP_DOLL;      *code++ = OP_DOLL;
2719      break;      break;
# Line 2418  for (;; ptr++) Line 2721  for (;; ptr++)
2721      /* There can never be a first char if '.' is first, whatever happens about      /* There can never be a first char if '.' is first, whatever happens about
2722      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqbyte doesn't change either. */
2723    
2724      case '.':      case CHAR_DOT:
2725      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
2726      zerofirstbyte = firstbyte;      zerofirstbyte = firstbyte;
2727      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
2728      previous = code;      previous = code;
2729      *code++ = OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
2730      break;      break;
2731    
2732    
# Line 2438  for (;; ptr++) Line 2741  for (;; ptr++)
2741      opcode is compiled. It may optionally have a bit map for characters < 256,      opcode is compiled. It may optionally have a bit map for characters < 256,
2742      but those above are are explicitly listed afterwards. A flag byte tells      but those above are are explicitly listed afterwards. A flag byte tells
2743      whether the bitmap is present, and whether this is a negated class or not.      whether the bitmap is present, and whether this is a negated class or not.
     */  
2744    
2745      case '[':      In JavaScript compatibility mode, an isolated ']' causes an error. In
2746        default (Perl) mode, it is treated as a data character. */
2747    
2748        case CHAR_RIGHT_SQUARE_BRACKET:
2749        if ((cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
2750          {
2751          *errorcodeptr = ERR64;
2752          goto FAILED;
2753          }
2754        goto NORMAL_CHAR;
2755    
2756        case CHAR_LEFT_SQUARE_BRACKET:
2757      previous = code;      previous = code;
2758    
2759      /* PCRE supports POSIX class stuff inside a class. Perl gives an error if      /* PCRE supports POSIX class stuff inside a class. Perl gives an error if
2760      they are encountered at the top level, so we'll do that too. */      they are encountered at the top level, so we'll do that too. */
2761    
2762      if ((ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&      if ((ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2763          check_posix_syntax(ptr, &tempptr, cd))           ptr[1] == CHAR_EQUALS_SIGN) &&
2764            check_posix_syntax(ptr, &tempptr))
2765        {        {
2766        *errorcodeptr = (ptr[1] == ':')? ERR13 : ERR31;        *errorcodeptr = (ptr[1] == CHAR_COLON)? ERR13 : ERR31;
2767        goto FAILED;        goto FAILED;
2768        }        }
2769    
2770      /* If the first character is '^', set the negation flag and skip it. Also,      /* If the first character is '^', set the negation flag and skip it. Also,
2771      if the first few characters (either before or after ^) are \Q\E or \E we      if the first few characters (either before or after ^) are \Q\E or \E we
2772      skip them too. This makes for compatibility with Perl. */      skip them too. This makes for compatibility with Perl. */
2773    
2774      negate_class = FALSE;      negate_class = FALSE;
2775      for (;;)      for (;;)
2776        {        {
2777        c = *(++ptr);        c = *(++ptr);
2778        if (c == '\\')        if (c == CHAR_BACKSLASH)
2779          {          {
2780          if (ptr[1] == 'E') ptr++;          if (ptr[1] == CHAR_E)
2781            else if (strncmp((const char *)ptr+1, "Q\\E", 3) == 0) ptr += 3;            ptr++;
2782              else break;          else if (strncmp((const char *)ptr+1,
2783                              STR_Q STR_BACKSLASH STR_E, 3) == 0)
2784              ptr += 3;
2785            else
2786              break;
2787          }          }
2788        else if (!negate_class && c == '^')        else if (!negate_class && c == CHAR_CIRCUMFLEX_ACCENT)
2789          negate_class = TRUE;          negate_class = TRUE;
2790        else break;        else break;
2791        }        }
2792    
2793        /* Empty classes are allowed in JavaScript compatibility mode. Otherwise,
2794        an initial ']' is taken as a data character -- the code below handles
2795        that. In JS mode, [] must always fail, so generate OP_FAIL, whereas
2796        [^] must match any character, so generate OP_ALLANY. */
2797    
2798        if (c == CHAR_RIGHT_SQUARE_BRACKET &&
2799            (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
2800          {
2801          *code++ = negate_class? OP_ALLANY : OP_FAIL;
2802          if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;
2803          zerofirstbyte = firstbyte;
2804          break;
2805          }
2806    
2807        /* If a class contains a negative special such as \S, we need to flip the
2808        negation flag at the end, so that support for characters > 255 works
2809        correctly (they are all included in the class). */
2810    
2811        should_flip_negation = FALSE;
2812    
2813      /* 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
2814      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 2489  for (;; ptr++) Line 2827  for (;; ptr++)
2827  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
2828      class_utf8 = FALSE;                       /* No chars >= 256 */      class_utf8 = FALSE;                       /* No chars >= 256 */
2829      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */
2830        class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */
2831  #endif  #endif
2832    
2833      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
# Line 2504  for (;; ptr++) Line 2843  for (;; ptr++)
2843          {                           /* Braces are required because the */          {                           /* Braces are required because the */
2844          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
2845          }          }
2846    
2847          /* In the pre-compile phase, accumulate the length of any UTF-8 extra
2848          data and reset the pointer. This is so that very large classes that
2849          contain a zillion UTF-8 characters no longer overwrite the work space
2850          (which is on the stack). */
2851    
2852          if (lengthptr != NULL)
2853            {
2854            *lengthptr += class_utf8data - class_utf8data_base;
2855            class_utf8data = class_utf8data_base;
2856            }
2857    
2858  #endif  #endif
2859    
2860        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
2861    
2862        if (inescq)        if (inescq)
2863          {          {
2864          if (c == '\\' && ptr[1] == 'E')     /* If we are at \E */          if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)  /* If we are at \E */
2865            {            {
2866            inescq = FALSE;                   /* Reset literal state */            inescq = FALSE;                   /* Reset literal state */
2867            ptr++;                            /* Skip the 'E' */            ptr++;                            /* Skip the 'E' */
# Line 2525  for (;; ptr++) Line 2876  for (;; ptr++)
2876        [.ch.] and [=ch=] ("collating elements") and fault them, as Perl        [.ch.] and [=ch=] ("collating elements") and fault them, as Perl
2877        5.6 and 5.8 do. */        5.6 and 5.8 do. */
2878    
2879        if (c == '[' &&        if (c == CHAR_LEFT_SQUARE_BRACKET &&
2880            (ptr[1] == ':' || ptr[1] == '.' || ptr[1] == '=') &&            (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2881            check_posix_syntax(ptr, &tempptr, cd))             ptr[1] == CHAR_EQUALS_SIGN) && check_posix_syntax(ptr, &tempptr))
2882          {          {
2883          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
2884          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
2885          register const uschar *cbits = cd->cbits;          register const uschar *cbits = cd->cbits;
2886          uschar pbits[32];          uschar pbits[32];
2887    
2888          if (ptr[1] != ':')          if (ptr[1] != CHAR_COLON)
2889            {            {
2890            *errorcodeptr = ERR31;            *errorcodeptr = ERR31;
2891            goto FAILED;            goto FAILED;
2892            }            }
2893    
2894          ptr += 2;          ptr += 2;
2895          if (*ptr == '^')          if (*ptr == CHAR_CIRCUMFLEX_ACCENT)
2896            {            {
2897            local_negate = TRUE;            local_negate = TRUE;
2898              should_flip_negation = TRUE;  /* Note negative special */
2899            ptr++;            ptr++;
2900            }            }
2901    
# Line 2613  for (;; ptr++) Line 2965  for (;; ptr++)
2965        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
2966        character in them, so set class_charcount bigger than one. */        character in them, so set class_charcount bigger than one. */
2967    
2968        if (c == '\\')        if (c == CHAR_BACKSLASH)
2969          {          {
2970          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
2971          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
2972    
2973          if (-c == ESC_b) c = '\b';       /* \b is backslash in a class */          if (-c == ESC_b) c = CHAR_BS;       /* \b is backspace in a class */
2974          else if (-c == ESC_X) c = 'X';   /* \X is literal X in a class */          else if (-c == ESC_X) c = CHAR_X;   /* \X is literal X in a class */
2975          else if (-c == ESC_R) c = 'R';   /* \R is literal R in a class */          else if (-c == ESC_R) c = CHAR_R;   /* \R is literal R in a class */
2976          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
2977            {            {
2978            if (ptr[1] == '\\' && ptr[2] == 'E')            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
2979              {              {
2980              ptr += 2; /* avoid empty string */              ptr += 2; /* avoid empty string */
2981              }              }
2982            else inescq = TRUE;            else inescq = TRUE;
2983            continue;            continue;
2984            }            }
2985            else if (-c == ESC_E) continue;  /* Ignore orphan \E */
2986    
2987          if (c < 0)          if (c < 0)
2988            {            {
# Line 2645  for (;; ptr++) Line 2998  for (;; ptr++)
2998              continue;              continue;
2999    
3000              case ESC_D:              case ESC_D:
3001                should_flip_negation = TRUE;
3002              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_digit];
3003              continue;              continue;
3004    
# Line 2653  for (;; ptr++) Line 3007  for (;; ptr++)
3007              continue;              continue;
3008    
3009              case ESC_W:              case ESC_W:
3010                should_flip_negation = TRUE;
3011              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_word];
3012              continue;              continue;
3013    
# Line 2662  for (;; ptr++) Line 3017  for (;; ptr++)
3017              continue;              continue;
3018    
3019              case ESC_S:              case ESC_S:
3020                should_flip_negation = TRUE;
3021              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
3022              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
3023              continue;              continue;
3024    
             case ESC_E: /* Perl ignores an orphan \E */  
             continue;  
   
3025              default:    /* Not recognized; fall through */              default:    /* Not recognized; fall through */
3026              break;      /* Need "default" setting to stop compiler warning. */              break;      /* Need "default" setting to stop compiler warning. */
3027              }              }
# Line 2848  for (;; ptr++) Line 3201  for (;; ptr++)
3201        entirely. The code for handling \Q and \E is messy. */        entirely. The code for handling \Q and \E is messy. */
3202    
3203        CHECK_RANGE:        CHECK_RANGE:
3204        while (ptr[1] == '\\' && ptr[2] == 'E')        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
3205          {          {
3206          inescq = FALSE;          inescq = FALSE;
3207          ptr += 2;          ptr += 2;
# Line 2856  for (;; ptr++) Line 3209  for (;; ptr++)
3209    
3210        oldptr = ptr;        oldptr = ptr;
3211    
3212        if (!inescq && ptr[1] == '-')        /* Remember \r or \n */
3213    
3214          if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
3215    
3216          /* Check for range */
3217    
3218          if (!inescq && ptr[1] == CHAR_MINUS)
3219          {          {
3220          int d;          int d;
3221          ptr += 2;          ptr += 2;
3222          while (*ptr == '\\' && ptr[1] == 'E') ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
3223    
3224          /* If we hit \Q (not followed by \E) at this point, go into escaped          /* If we hit \Q (not followed by \E) at this point, go into escaped
3225          mode. */          mode. */
3226    
3227          while (*ptr == '\\' && ptr[1] == 'Q')          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_Q)
3228            {            {
3229            ptr += 2;            ptr += 2;
3230            if (*ptr == '\\' && ptr[1] == 'E') { ptr += 2; continue; }            if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3231                { ptr += 2; continue; }
3232            inescq = TRUE;            inescq = TRUE;
3233            break;            break;
3234            }            }
3235    
3236          if (*ptr == 0 || (!inescq && *ptr == ']'))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
3237            {            {
3238            ptr = oldptr;            ptr = oldptr;
3239            goto LONE_SINGLE_CHARACTER;            goto LONE_SINGLE_CHARACTER;
# Line 2892  for (;; ptr++) Line 3252  for (;; ptr++)
3252          not any of the other escapes. Perl 5.6 treats a hyphen as a literal          not any of the other escapes. Perl 5.6 treats a hyphen as a literal
3253          in such circumstances. */          in such circumstances. */
3254    
3255          if (!inescq && d == '\\')          if (!inescq && d == CHAR_BACKSLASH)
3256            {            {
3257            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
3258            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
3259    
3260            /* \b is backslash; \X is literal X; \R is literal R; any other            /* \b is backspace; \X is literal X; \R is literal R; any other
3261            special means the '-' was literal */            special means the '-' was literal */
3262    
3263            if (d < 0)            if (d < 0)
3264              {              {
3265              if (d == -ESC_b) d = '\b';              if (d == -ESC_b) d = CHAR_BS;
3266              else if (d == -ESC_X) d = 'X';              else if (d == -ESC_X) d = CHAR_X;
3267              else if (d == -ESC_R) d = 'R'; else              else if (d == -ESC_R) d = CHAR_R; else
3268                {                {
3269                ptr = oldptr;                ptr = oldptr;
3270                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto LONE_SINGLE_CHARACTER;  /* A few lines below */
# Line 2923  for (;; ptr++) Line 3283  for (;; ptr++)
3283    
3284          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */
3285    
3286            /* Remember \r or \n */
3287    
3288            if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
3289    
3290          /* 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
3291          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
3292          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
# Line 3041  for (;; ptr++) Line 3405  for (;; ptr++)
3405          if ((options & PCRE_CASELESS) != 0)          if ((options & PCRE_CASELESS) != 0)
3406            {            {
3407            unsigned int othercase;            unsigned int othercase;
3408            if ((othercase = _pcre_ucp_othercase(c)) != NOTACHAR)            if ((othercase = UCD_OTHERCASE(c)) != c)
3409              {              {
3410              *class_utf8data++ = XCL_SINGLE;              *class_utf8data++ = XCL_SINGLE;
3411              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);
# Line 3068  for (;; ptr++) Line 3432  for (;; ptr++)
3432    
3433      /* Loop until ']' reached. This "while" is the end of the "do" above. */      /* Loop until ']' reached. This "while" is the end of the "do" above. */
3434    
3435      while ((c = *(++ptr)) != 0 && (c != ']' || inescq));      while ((c = *(++ptr)) != 0 && (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
3436    
3437      if (c == 0)                          /* Missing terminating ']' */      if (c == 0)                          /* Missing terminating ']' */
3438        {        {
3439        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
3440        goto FAILED;        goto FAILED;
3441        }        }
3442    
3443    
3444    /* This code has been disabled because it would mean that \s counts as
3445    an explicit \r or \n reference, and that's not really what is wanted. Now
3446    we set the flag only if there is a literal "\r" or "\n" in the class. */
3447    
3448    #if 0
3449        /* Remember whether \r or \n are in this class */
3450    
3451        if (negate_class)
3452          {
3453          if ((classbits[1] & 0x24) != 0x24) cd->external_flags |= PCRE_HASCRORLF;
3454          }
3455        else
3456          {
3457          if ((classbits[1] & 0x24) != 0) cd->external_flags |= PCRE_HASCRORLF;
3458          }
3459    #endif
3460    
3461    
3462      /* 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
3463      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
3464      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
3465      because OP_NOT and the related opcodes like OP_NOTSTAR operate on      optimize.
3466      single-bytes only. This is an historical hangover. Maybe one day we can  
3467      tidy these opcodes to handle multi-byte characters.      In UTF-8 mode, we can optimize the negative case only if there were no
3468        characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
3469        operate on single-bytes only. This is an historical hangover. Maybe one day
3470        we can tidy these opcodes to handle multi-byte characters.
3471    
3472      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
3473      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 3091  for (;; ptr++) Line 3477  for (;; ptr++)
3477      reqbyte, save the previous value for reinstating. */      reqbyte, save the previous value for reinstating. */
3478    
3479  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3480      if (class_charcount == 1 &&      if (class_charcount == 1 && !class_utf8 &&
3481            (!utf8 ||        (!utf8 || !negate_class || class_lastchar < 128))
           (!class_utf8 && (!negate_class || class_lastchar < 128))))  
   
3482  #else  #else
3483      if (class_charcount == 1)      if (class_charcount == 1)
3484  #endif  #endif
# Line 3137  for (;; ptr++) Line 3521  for (;; ptr++)
3521      zeroreqbyte = reqbyte;      zeroreqbyte = reqbyte;
3522    
3523      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
3524      extended class, with its own opcode. If there are no characters < 256,      extended class, with its own opcode, unless there was a negated special
3525      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
3526        the class, so any that were explicitly given as well can be ignored. If
3527        (when there are explicit characters > 255 that must be listed) there are no
3528        characters < 256, we can omit the bitmap in the actual compiled code. */
3529    
3530  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
3531      if (class_utf8)      if (class_utf8 && !should_flip_negation)
3532        {        {
3533        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */
3534        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
# Line 3167  for (;; ptr++) Line 3554  for (;; ptr++)
3554        }        }
3555  #endif  #endif
3556    
3557      /* 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
3558      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
3559      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
3560      setting must remain unchanged after any kind of repeat. */      map into the code vector, negating it if necessary. */
3561    
3562        *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
3563      if (negate_class)      if (negate_class)
3564        {        {
       *code++ = OP_NCLASS;  
3565        if (lengthptr == NULL)    /* Save time in the pre-compile phase */        if (lengthptr == NULL)    /* Save time in the pre-compile phase */
3566          for (c = 0; c < 32; c++) code[c] = ~classbits[c];          for (c = 0; c < 32; c++) code[c] = ~classbits[c];
3567        }        }
3568      else      else
3569        {        {
       *code++ = OP_CLASS;  
3570        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
3571        }        }
3572      code += 32;      code += 32;
# Line 3191  for (;; ptr++) Line 3577  for (;; ptr++)
3577      /* Various kinds of repeat; '{' is not necessarily a quantifier, but this      /* Various kinds of repeat; '{' is not necessarily a quantifier, but this
3578      has been tested above. */      has been tested above. */
3579    
3580      case '{':      case CHAR_LEFT_CURLY_BRACKET:
3581      if (!is_quantifier) goto NORMAL_CHAR;      if (!is_quantifier) goto NORMAL_CHAR;
3582      ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr);      ptr = read_repeat_counts(ptr+1, &repeat_min, &repeat_max, errorcodeptr);
3583      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
3584      goto REPEAT;      goto REPEAT;
3585    
3586      case '*':      case CHAR_ASTERISK:
3587      repeat_min = 0;      repeat_min = 0;
3588      repeat_max = -1;      repeat_max = -1;
3589      goto REPEAT;      goto REPEAT;
3590    
3591      case '+':      case CHAR_PLUS:
3592      repeat_min = 1;      repeat_min = 1;
3593      repeat_max = -1;      repeat_max = -1;
3594      goto REPEAT;      goto REPEAT;
3595    
3596      case '?':      case CHAR_QUESTION_MARK:
3597      repeat_min = 0;      repeat_min = 0;
3598      repeat_max = 1;      repeat_max = 1;
3599    
# Line 3242  for (;; ptr++) Line 3628  for (;; ptr++)
3628      but if PCRE_UNGREEDY is set, it works the other way round. We change the      but if PCRE_UNGREEDY is set, it works the other way round. We change the
3629      repeat type to the non-default. */      repeat type to the non-default. */
3630    
3631      if (ptr[1] == '+')      if (ptr[1] == CHAR_PLUS)
3632        {        {
3633        repeat_type = 0;                  /* Force greedy */        repeat_type = 0;                  /* Force greedy */
3634        possessive_quantifier = TRUE;        possessive_quantifier = TRUE;
3635        ptr++;        ptr++;
3636        }        }
3637      else if (ptr[1] == '?')      else if (ptr[1] == CHAR_QUESTION_MARK)
3638        {        {
3639        repeat_type = greedy_non_default;        repeat_type = greedy_non_default;
3640        ptr++;        ptr++;
# Line 3366  for (;; ptr++) Line 3752  for (;; ptr++)
3752        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3753        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3754    
3755        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3756    
3757        /* Combine the op_type with the repeat_type */        /* Combine the op_type with the repeat_type */
3758    
# Line 3516  for (;; ptr++) Line 3902  for (;; ptr++)
3902        /* All real repeats make it impossible to handle partial matching (maybe        /* All real repeats make it impossible to handle partial matching (maybe
3903        one day we will be able to remove this restriction). */        one day we will be able to remove this restriction). */
3904    
3905        if (repeat_max != 1) cd->nopartial = TRUE;        if (repeat_max != 1) cd->external_flags |= PCRE_NOPARTIAL;
3906    
3907        if (repeat_min == 0 && repeat_max == -1)        if (repeat_min == 0 && repeat_max == -1)
3908          *code++ = OP_CRSTAR + repeat_type;          *code++ = OP_CRSTAR + repeat_type;
# Line 3574  for (;; ptr++) Line 3960  for (;; ptr++)
3960    
3961        if (repeat_min == 0)        if (repeat_min == 0)
3962          {          {
3963          /* If the maximum is also zero, we just omit the group from the output          /* If the maximum is also zero, we used to just omit the group from the
3964          altogether. */          output altogether, like this:
   
         if (repeat_max == 0)  
           {  
           code = previous;  
           goto END_REPEAT;  
           }  
3965    
3966          /* If the maximum is 1 or unlimited, we just have to stick in the          ** if (repeat_max == 0)
3967          BRAZERO and do no more at this point. However, we do need to adjust          **   {
3968          any OP_RECURSE calls inside the group that refer to the group itself or          **   code = previous;
3969          any internal or forward referenced group, because the offset is from          **   goto END_REPEAT;
3970          the start of the whole regex. Temporarily terminate the pattern while          **   }
3971          doing this. */  
3972            However, that fails when a group is referenced as a subroutine from
3973            elsewhere in the pattern, so now we stick in OP_SKIPZERO in front of it
3974            so that it is skipped on execution. As we don't have a list of which
3975            groups are referenced, we cannot do this selectively.
3976    
3977            If the maximum is 1 or unlimited, we just have to stick in the BRAZERO
3978            and do no more at this point. However, we do need to adjust any
3979            OP_RECURSE calls inside the group that refer to the group itself or any
3980            internal or forward referenced group, because the offset is from the
3981            start of the whole regex. Temporarily terminate the pattern while doing
3982            this. */
3983    
3984          if (repeat_max <= 1)          if (repeat_max <= 1)    /* Covers 0, 1, and unlimited */
3985            {            {
3986            *code = OP_END;            *code = OP_END;
3987            adjust_recurse(previous, 1, utf8, cd, save_hwm);            adjust_recurse(previous, 1, utf8, cd, save_hwm);
3988            memmove(previous+1, previous, len);            memmove(previous+1, previous, len);
3989            code++;            code++;
3990              if (repeat_max == 0)
3991                {
3992                *previous++ = OP_SKIPZERO;
3993                goto END_REPEAT;
3994                }
3995            *previous++ = OP_BRAZERO + repeat_type;            *previous++ = OP_BRAZERO + repeat_type;
3996            }            }
3997    
# Line 3790  for (;; ptr++) Line 4186  for (;; ptr++)
4186          }          }
4187        }        }
4188    
4189        /* If previous is OP_FAIL, it was generated by an empty class [] in
4190        JavaScript mode. The other ways in which OP_FAIL can be generated, that is
4191        by (*FAIL) or (?!) set previous to NULL, which gives a "nothing to repeat"
4192        error above. We can just ignore the repeat in JS case. */
4193    
4194        else if (*previous == OP_FAIL) goto END_REPEAT;
4195    
4196      /* Else there's some kind of shambles */      /* Else there's some kind of shambles */
4197    
4198      else      else
# Line 3816  for (;; ptr++) Line 4219  for (;; ptr++)
4219        int len;        int len;
4220        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||        if (*tempcode == OP_EXACT || *tempcode == OP_TYPEEXACT ||
4221            *tempcode == OP_NOTEXACT)            *tempcode == OP_NOTEXACT)
4222          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += _pcre_OP_lengths[*tempcode] +
4223              ((*tempcode == OP_TYPEEXACT &&
4224                 (tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP))? 2:0);
4225        len = code - tempcode;        len = code - tempcode;
4226        if (len > 0) switch (*tempcode)        if (len > 0) switch (*tempcode)
4227          {          {
# Line 3860  for (;; ptr++) Line 4265  for (;; ptr++)
4265      /* ===================================================================*/      /* ===================================================================*/
4266      /* Start of nested parenthesized sub-expression, or comment or lookahead or      /* Start of nested parenthesized sub-expression, or comment or lookahead or
4267      lookbehind or option setting or condition or all the other extended      lookbehind or option setting or condition or all the other extended
4268      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. */  
4269    
4270      case '(':      case CHAR_LEFT_PARENTHESIS:
4271      newoptions = options;      newoptions = options;
4272      skipbytes = 0;      skipbytes = 0;
4273      bravalue = OP_CBRA;      bravalue = OP_CBRA;
4274      save_hwm = cd->hwm;      save_hwm = cd->hwm;
4275      reset_bracount = FALSE;      reset_bracount = FALSE;
4276    
4277      if (*(++ptr) == '?')      /* First deal with various "verbs" that can be introduced by '*'. */
4278    
4279        if (*(++ptr) == CHAR_ASTERISK && (cd->ctypes[ptr[1]] & ctype_letter) != 0)
4280          {
4281          int i, namelen;
4282          const char *vn = verbnames;
4283          const uschar *name = ++ptr;
4284          previous = NULL;
4285          while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};
4286          if (*ptr == CHAR_COLON)
4287            {
4288            *errorcodeptr = ERR59;   /* Not supported */
4289            goto FAILED;
4290            }
4291          if (*ptr != CHAR_RIGHT_PARENTHESIS)
4292            {
4293            *errorcodeptr = ERR60;
4294            goto FAILED;
4295            }
4296          namelen = ptr - name;
4297          for (i = 0; i < verbcount; i++)
4298            {
4299            if (namelen == verbs[i].len &&
4300                strncmp((char *)name, vn, namelen) == 0)
4301              {
4302              *code = verbs[i].op;
4303              if (*code++ == OP_ACCEPT) cd->had_accept = TRUE;
4304              break;
4305              }
4306            vn += verbs[i].len + 1;
4307            }
4308          if (i < verbcount) continue;
4309          *errorcodeptr = ERR60;
4310          goto FAILED;
4311          }
4312    
4313        /* Deal with the extended parentheses; all are introduced by '?', and the
4314        appearance of any of them means that this is not a capturing group. */
4315    
4316        else if (*ptr == CHAR_QUESTION_MARK)
4317        {        {
4318        int i, set, unset, namelen;        int i, set, unset, namelen;
4319        int *optset;        int *optset;
# Line 3880  for (;; ptr++) Line 4322  for (;; ptr++)
4322    
4323        switch (*(++ptr))        switch (*(++ptr))
4324          {          {
4325          case '#':                 /* Comment; skip to ket */          case CHAR_NUMBER_SIGN:                 /* Comment; skip to ket */
4326          ptr++;          ptr++;
4327          while (*ptr != 0 && *ptr != ')') ptr++;          while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
4328          if (*ptr == 0)          if (*ptr == 0)
4329            {            {
4330            *errorcodeptr = ERR18;            *errorcodeptr = ERR18;
# Line 3892  for (;; ptr++) Line 4334  for (;; ptr++)
4334    
4335    
4336          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4337          case '|':                 /* Reset capture count for each branch */          case CHAR_VERTICAL_LINE:  /* Reset capture count for each branch */
4338          reset_bracount = TRUE;          reset_bracount = TRUE;
4339          /* Fall through */          /* Fall through */
4340    
4341          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4342          case ':':                 /* Non-capturing bracket */          case CHAR_COLON:          /* Non-capturing bracket */
4343          bravalue = OP_BRA;          bravalue = OP_BRA;
4344          ptr++;          ptr++;
4345          break;          break;
4346    
4347    
4348          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4349          case '(':          case CHAR_LEFT_PARENTHESIS:
4350          bravalue = OP_COND;       /* Conditional group */          bravalue = OP_COND;       /* Conditional group */
4351    
4352          /* A condition can be an assertion, a number (referring to a numbered          /* A condition can be an assertion, a number (referring to a numbered
# Line 3924  for (;; ptr++) Line 4366  for (;; ptr++)
4366          the switch. This will take control down to where bracketed groups,          the switch. This will take control down to where bracketed groups,
4367          including assertions, are processed. */          including assertions, are processed. */
4368    
4369          if (ptr[1] == '?' && (ptr[2] == '=' || ptr[2] == '!' || ptr[2] == '<'))          if (ptr[1] == CHAR_QUESTION_MARK && (ptr[2] == CHAR_EQUALS_SIGN ||
4370                ptr[2] == CHAR_EXCLAMATION_MARK || ptr[2] == CHAR_LESS_THAN_SIGN))
4371            break;            break;
4372    
4373          /* Most other conditions use OP_CREF (a couple change to OP_RREF          /* Most other conditions use OP_CREF (a couple change to OP_RREF
# Line 3936  for (;; ptr++) Line 4379  for (;; ptr++)
4379    
4380          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
4381    
4382          if (ptr[1] == 'R' && ptr[2] == '&')          if (ptr[1] == CHAR_R && ptr[2] == CHAR_AMPERSAND)
4383            {            {
4384            terminator = -1;            terminator = -1;
4385            ptr += 2;            ptr += 2;
# Line 3946  for (;; ptr++) Line 4389  for (;; ptr++)
4389          /* Check for a test for a named group's having been set, using the Perl          /* Check for a test for a named group's having been set, using the Perl
4390          syntax (?(<name>) or (?('name') */          syntax (?(<name>) or (?('name') */
4391    
4392          else if (ptr[1] == '<')          else if (ptr[1] == CHAR_LESS_THAN_SIGN)
4393            {            {
4394            terminator = '>';            terminator = CHAR_GREATER_THAN_SIGN;
4395            ptr++;            ptr++;
4396            }            }
4397          else if (ptr[1] == '\'')          else if (ptr[1] == CHAR_APOSTROPHE)
4398            {            {
4399            terminator = '\'';            terminator = CHAR_APOSTROPHE;
4400            ptr++;            ptr++;
4401            }            }
4402          else          else
4403            {            {
4404            terminator = 0;            terminator = 0;
4405            if (ptr[1] == '-' || ptr[1] == '+') refsign = *(++ptr);            if (ptr[1] == CHAR_MINUS || ptr[1] == CHAR_PLUS) refsign = *(++ptr);
4406            }            }
4407    
4408          /* We now expect to read a name; any thing else is an error */          /* We now expect to read a name; any thing else is an error */
# Line 3979  for (;; ptr++) Line 4422  for (;; ptr++)
4422            {            {
4423            if (recno >= 0)            if (recno >= 0)
4424              recno = ((digitab[*ptr] & ctype_digit) != 0)?              recno = ((digitab[*ptr] & ctype_digit) != 0)?
4425                recno * 10 + *ptr - '0' : -1;                recno * 10 + *ptr - CHAR_0 : -1;
4426            ptr++;            ptr++;
4427            }            }
4428          namelen = ptr - name;          namelen = ptr - name;
4429    
4430          if ((terminator > 0 && *ptr++ != terminator) || *ptr++ != ')')          if ((terminator > 0 && *ptr++ != terminator) ||
4431                *ptr++ != CHAR_RIGHT_PARENTHESIS)
4432            {            {
4433            ptr--;      /* Error offset */            ptr--;      /* Error offset */
4434            *errorcodeptr = ERR26;            *errorcodeptr = ERR26;
# Line 4006  for (;; ptr++) Line 4450  for (;; ptr++)
4450              *errorcodeptr = ERR58;              *errorcodeptr = ERR58;
4451              goto FAILED;              goto FAILED;
4452              }              }
4453            if (refsign == '-')            recno = (refsign == CHAR_MINUS)?
4454                cd->bracount - recno + 1 : recno +cd->bracount;
4455              if (recno <= 0 || recno > cd->final_bracount)
4456              {              {
4457              recno = cd->bracount - recno + 1;              *errorcodeptr = ERR15;
4458              if (recno <= 0)              goto FAILED;
               {  
               *errorcodeptr = ERR15;  
               goto FAILED;  
               }  
4459              }              }
           else recno += cd->bracount;  
4460            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4461            break;            break;
4462            }            }
# Line 4040  for (;; ptr++) Line 4481  for (;; ptr++)
4481    
4482          /* Search the pattern for a forward reference */          /* Search the pattern for a forward reference */
4483    
4484          else if ((i = find_parens(ptr, cd->bracount, name, namelen,          else if ((i = find_parens(ptr, cd, name, namelen,
4485                          (options & PCRE_EXTENDED) != 0)) > 0)                          (options & PCRE_EXTENDED) != 0)) > 0)
4486            {            {
4487            PUT2(code, 2+LINK_SIZE, i);            PUT2(code, 2+LINK_SIZE, i);
# Line 4061  for (;; ptr++) Line 4502  for (;; ptr++)
4502          /* Check for (?(R) for recursion. Allow digits after R to specify a          /* Check for (?(R) for recursion. Allow digits after R to specify a
4503          specific group number. */          specific group number. */
4504    
4505          else if (*name == 'R')          else if (*name == CHAR_R)
4506            {            {
4507            recno = 0;            recno = 0;
4508            for (i = 1; i < namelen; i++)            for (i = 1; i < namelen; i++)
# Line 4071  for (;; ptr++) Line 4512  for (;; ptr++)
4512                *errorcodeptr = ERR15;                *errorcodeptr = ERR15;
4513                goto FAILED;                goto FAILED;
4514                }                }
4515              recno = recno * 10 + name[i] - '0';              recno = recno * 10 + name[i] - CHAR_0;
4516              }              }
4517            if (recno == 0) recno = RREF_ANY;            if (recno == 0) recno = RREF_ANY;
4518            code[1+LINK_SIZE] = OP_RREF;      /* Change test type */            code[1+LINK_SIZE] = OP_RREF;      /* Change test type */
# Line 4081  for (;; ptr++) Line 4522  for (;; ptr++)
4522          /* Similarly, check for the (?(DEFINE) "condition", which is always          /* Similarly, check for the (?(DEFINE) "condition", which is always
4523          false. */          false. */
4524    
4525          else if (namelen == 6 && strncmp((char *)name, "DEFINE", 6) == 0)          else if (namelen == 6 && strncmp((char *)name, STRING_DEFINE, 6) == 0)
4526            {            {
4527            code[1+LINK_SIZE] = OP_DEF;            code[1+LINK_SIZE] = OP_DEF;
4528            skipbytes = 1;            skipbytes = 1;
4529            }            }
4530    
4531          /* Check for the "name" actually being a subpattern number. */          /* Check for the "name" actually being a subpattern number. We are
4532            in the second pass here, so final_bracount is set. */
4533    
4534          else if (recno > 0)          else if (recno > 0 && recno <= cd->final_bracount)
4535            {            {
4536            PUT2(code, 2+LINK_SIZE, recno);            PUT2(code, 2+LINK_SIZE, recno);
4537            }            }
# Line 4105  for (;; ptr++) Line 4547  for (;; ptr++)
4547    
4548    
4549          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4550          case '=':                 /* Positive lookahead */          case CHAR_EQUALS_SIGN:                 /* Positive lookahead */
4551          bravalue = OP_ASSERT;          bravalue = OP_ASSERT;
4552          ptr++;          ptr++;
4553          break;          break;
4554    
4555    
4556          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4557          case '!':                 /* Negative lookahead */          case CHAR_EXCLAMATION_MARK:            /* Negative lookahead */
         bravalue = OP_ASSERT_NOT;  
4558          ptr++;          ptr++;
4559            if (*ptr == CHAR_RIGHT_PARENTHESIS)    /* Optimize (?!) */
4560              {
4561              *code++ = OP_FAIL;
4562              previous = NULL;
4563              continue;
4564              }
4565            bravalue = OP_ASSERT_NOT;
4566          break;          break;
4567    
4568    
4569          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4570          case '<':                 /* Lookbehind or named define */          case CHAR_LESS_THAN_SIGN:              /* Lookbehind or named define */
4571          switch (ptr[1])          switch (ptr[1])
4572            {            {
4573            case '=':               /* Positive lookbehind */            case CHAR_EQUALS_SIGN:               /* Positive lookbehind */
4574            bravalue = OP_ASSERTBACK;            bravalue = OP_ASSERTBACK;
4575            ptr += 2;            ptr += 2;
4576            break;            break;
4577    
4578            case '!':               /* Negative lookbehind */            case CHAR_EXCLAMATION_MARK:          /* Negative lookbehind */
4579            bravalue = OP_ASSERTBACK_NOT;            bravalue = OP_ASSERTBACK_NOT;
4580            ptr += 2;            ptr += 2;
4581            break;            break;
# Line 4142  for (;; ptr++) Line 4590  for (;; ptr++)
4590    
4591    
4592          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4593          case '>':                 /* One-time brackets */          case CHAR_GREATER_THAN_SIGN:           /* One-time brackets */
4594          bravalue = OP_ONCE;          bravalue = OP_ONCE;
4595          ptr++;          ptr++;
4596          break;          break;
4597    
4598    
4599          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4600          case 'C':                 /* Callout - may be followed by digits; */          case CHAR_C:                 /* Callout - may be followed by digits; */
4601          previous_callout = code;  /* Save for later completion */          previous_callout = code;  /* Save for later completion */
4602          after_manual_callout = 1; /* Skip one item before completing */          after_manual_callout = 1; /* Skip one item before completing */
4603          *code++ = OP_CALLOUT;          *code++ = OP_CALLOUT;
4604            {            {
4605            int n = 0;            int n = 0;
4606            while ((digitab[*(++ptr)] & ctype_digit) != 0)            while ((digitab[*(++ptr)] & ctype_digit) != 0)
4607              n = n * 10 + *ptr - '0';              n = n * 10 + *ptr - CHAR_0;
4608            if (*ptr != ')')            if (*ptr != CHAR_RIGHT_PARENTHESIS)
4609              {              {
4610              *errorcodeptr = ERR39;              *errorcodeptr = ERR39;
4611              goto FAILED;              goto FAILED;
# Line 4177  for (;; ptr++) Line 4625  for (;; ptr++)
4625    
4626    
4627          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4628          case 'P':                 /* Python-style named subpattern handling */          case CHAR_P:              /* Python-style named subpattern handling */
4629          if (*(++ptr) == '=' || *ptr == '>')  /* Reference or recursion */          if (*(++ptr) == CHAR_EQUALS_SIGN ||
4630                *ptr == CHAR_GREATER_THAN_SIGN)  /* Reference or recursion */
4631            {            {
4632            is_recurse = *ptr == '>';            is_recurse = *ptr == CHAR_GREATER_THAN_SIGN;
4633            terminator = ')';            terminator = CHAR_RIGHT_PARENTHESIS;
4634            goto NAMED_REF_OR_RECURSE;            goto NAMED_REF_OR_RECURSE;
4635            }            }
4636          else if (*ptr != '<')    /* Test for Python-style definition */          else if (*ptr != CHAR_LESS_THAN_SIGN)  /* Test for Python-style defn */
4637            {            {
4638            *errorcodeptr = ERR41;            *errorcodeptr = ERR41;
4639            goto FAILED;            goto FAILED;
# Line 4194  for (;; ptr++) Line 4643  for (;; ptr++)
4643    
4644          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4645          DEFINE_NAME:    /* Come here from (?< handling */          DEFINE_NAME:    /* Come here from (?< handling */
4646          case '\'':          case CHAR_APOSTROPHE:
4647            {            {
4648            terminator = (*ptr == '<')? '>' : '\'';            terminator = (*ptr == CHAR_LESS_THAN_SIGN)?
4649                CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
4650            name = ++ptr;            name = ++ptr;
4651    
4652            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
# Line 4270  for (;; ptr++) Line 4720  for (;; ptr++)
4720    
4721    
4722          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4723          case '&':                 /* Perl recursion/subroutine syntax */          case CHAR_AMPERSAND:            /* Perl recursion/subroutine syntax */
4724          terminator = ')';          terminator = CHAR_RIGHT_PARENTHESIS;
4725          is_recurse = TRUE;          is_recurse = TRUE;
4726          /* Fall through */          /* Fall through */
4727    
4728          /* We come here from the Python syntax above that handles both          /* We come here from the Python syntax above that handles both
4729          references (?P=name) and recursion (?P>name), as well as falling          references (?P=name) and recursion (?P>name), as well as falling
4730          through from the Perl recursion syntax (?&name). */          through from the Perl recursion syntax (?&name). We also come here from
4731            the Perl \k<name> or \k'name' back reference syntax and the \k{name}
4732            .NET syntax, and the Oniguruma \g<...> and \g'...' subroutine syntax. */
4733    
4734          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
4735          name = ++ptr;          name = ++ptr;
# Line 4289  for (;; ptr++) Line 4741  for (;; ptr++)
4741    
4742          if (lengthptr != NULL)          if (lengthptr != NULL)
4743            {            {
4744              if (namelen == 0)
4745                {
4746                *errorcodeptr = ERR62;
4747                goto FAILED;
4748                }
4749            if (*ptr != terminator)            if (*ptr != terminator)
4750              {              {
4751              *errorcodeptr = ERR42;              *errorcodeptr = ERR42;
# Line 4302  for (;; ptr++) Line 4759  for (;; ptr++)
4759            recno = 0;            recno = 0;
4760            }            }
4761    
4762          /* In the real compile, seek the name in the table */          /* In the real compile, seek the name in the table. We check the name
4763            first, and then check that we have reached the end of the name in the
4764            table. That way, if the name that is longer than any in the table,
4765            the comparison will fail without reading beyond the table entry. */
4766    
4767          else          else
4768            {            {
4769            slot = cd->name_table;            slot = cd->name_table;
4770            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
4771              {              {
4772              if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&
4773                    slot[2+namelen] == 0)
4774                  break;
4775              slot += cd->name_entry_size;              slot += cd->name_entry_size;
4776              }              }
4777    
# Line 4318  for (;; ptr++) Line 4780  for (;; ptr++)
4780              recno = GET2(slot, 0);              recno = GET2(slot, 0);
4781              }              }
4782            else if ((recno =                /* Forward back reference */            else if ((recno =                /* Forward back reference */
4783                      find_parens(ptr, cd->bracount, name, namelen,                      find_parens(ptr, cd, name, namelen,
4784                        (options & PCRE_EXTENDED) != 0)) <= 0)                        (options & PCRE_EXTENDED) != 0)) <= 0)
4785              {              {
4786              *errorcodeptr = ERR15;              *errorcodeptr = ERR15;
# Line 4334  for (;; ptr++) Line 4796  for (;; ptr++)
4796    
4797    
4798          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4799          case 'R':                 /* Recursion */          case CHAR_R:              /* Recursion */
4800          ptr++;                    /* Same as (?0)      */          ptr++;                    /* Same as (?0)      */
4801          /* Fall through */          /* Fall through */
4802    
4803    
4804          /* ------------------------------------------------------------ */          /* ------------------------------------------------------------ */
4805          case '-': case '+':          case CHAR_MINUS: case CHAR_PLUS:  /* Recursion or subroutine */
4806          case '0': case '1': case '2': case '3': case '4':   /* Recursion or */          case CHAR_0: case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4:
4807          case '5': case '6': case '7': case '8': case '9':   /* subroutine */          case CHAR_5: case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:
4808            {            {
4809            const uschar *called;            const uschar *called;
4810              terminator = CHAR_RIGHT_PARENTHESIS;
4811    
4812              /* Come here from the \g<...> and \g'...' code (Oniguruma
4813              compatibility). However, the syntax has been checked to ensure that
4814              the ... are a (signed) number, so that neither ERR63 nor ERR29 will
4815              be called on this path, nor with the jump to OTHER_CHAR_AFTER_QUERY
4816              ever be taken. */
4817    
4818              HANDLE_NUMERICAL_RECURSION:
4819    
4820            if ((refsign = *ptr) == '+') ptr++;            if ((refsign = *ptr) == CHAR_PLUS)
4821            else if (refsign == '-')              {
4822                ptr++;
4823                if ((digitab[*ptr] & ctype_digit) == 0)
4824                  {
4825                  *errorcodeptr = ERR63;
4826                  goto FAILED;
4827                  }
4828                }
4829              else if (refsign == CHAR_MINUS)
4830              {              {
4831              if ((digitab[ptr[1]] & ctype_digit) == 0)              if ((digitab[ptr[1]] & ctype_digit) == 0)
4832                goto OTHER_CHAR_AFTER_QUERY;                goto OTHER_CHAR_AFTER_QUERY;
# Line 4356  for (;; ptr++) Line 4835  for (;; ptr++)
4835    
4836            recno = 0;            recno = 0;
4837            while((digitab[*ptr] & ctype_digit) != 0)            while((digitab[*ptr] & ctype_digit) != 0)
4838              recno = recno * 10 + *ptr++ - '0';              recno = recno * 10 + *ptr++ - CHAR_0;
4839    
4840            if (*ptr != ')')            if (*ptr != terminator)
4841              {              {
4842              *errorcodeptr = ERR29;              *errorcodeptr = ERR29;
4843              goto FAILED;              goto FAILED;
4844              }              }
4845    
4846            if (refsign == '-')            if (refsign == CHAR_MINUS)
4847              {              {
4848              if (recno == 0)              if (recno == 0)
4849                {                {
# Line 4378  for (;; ptr++) Line 4857  for (;; ptr++)
4857                goto FAILED;                goto FAILED;
4858                }                }
4859              }              }
4860            else if (refsign == '+')            else if (refsign == CHAR_PLUS)
4861              {              {
4862              if (recno == 0)              if (recno == 0)
4863                {                {
# Line 4411  for (;; ptr++) Line 4890  for (;; ptr++)
4890    
4891              if (called == NULL)              if (called == NULL)
4892                {                {
4893                if (find_parens(ptr, cd->bracount, NULL, recno,                if (find_parens(ptr, cd, NULL, recno,
4894                     (options & PCRE_EXTENDED) != 0) < 0)                      (options & PCRE_EXTENDED) != 0) < 0)
4895                  {                  {
4896                  *errorcodeptr = ERR15;                  *errorcodeptr = ERR15;
4897                  goto FAILED;                  goto FAILED;
# Line 4464  for (;; ptr++) Line 4943  for (;; ptr++)
4943          set = unset = 0;          set = unset = 0;
4944          optset = &set;          optset = &set;
4945    
4946          while (*ptr != ')' && *ptr != ':')          while (*ptr != CHAR_RIGHT_PARENTHESIS && *ptr != CHAR_COLON)
4947            {            {
4948            switch (*ptr++)            switch (*ptr++)
4949              {              {
4950              case '-': optset = &unset; break;              case CHAR_MINUS: optset = &unset; break;
4951    
4952              case 'J':    /* Record that it changed in the external options */              case CHAR_J:    /* Record that it changed in the external options */
4953              *optset |= PCRE_DUPNAMES;              *optset |= PCRE_DUPNAMES;
4954              cd->external_options |= PCRE_JCHANGED;              cd->external_flags |= PCRE_JCHANGED;
4955              break;              break;
4956    
4957              case 'i': *optset |= PCRE_CASELESS; break;              case CHAR_i: *optset |= PCRE_CASELESS; break;
4958              case 'm': *optset |= PCRE_MULTILINE; break;              case CHAR_m: *optset |= PCRE_MULTILINE; break;
4959              case 's': *optset |= PCRE_DOTALL; break;              case CHAR_s: *optset |= PCRE_DOTALL; break;
4960              case 'x': *optset |= PCRE_EXTENDED; break;              case CHAR_x: *optset |= PCRE_EXTENDED; break;
4961              case 'U': *optset |= PCRE_UNGREEDY; break;              case CHAR_U: *optset |= PCRE_UNGREEDY; break;
4962              case 'X': *optset |= PCRE_EXTRA; break;              case CHAR_X: *optset |= PCRE_EXTRA; break;
4963    
4964              default:  *errorcodeptr = ERR12;              default:  *errorcodeptr = ERR12;
4965                        ptr--;    /* Correct the offset */                        ptr--;    /* Correct the offset */
# Line 4511  for (;; ptr++) Line 4990  for (;; ptr++)
4990          both phases.          both phases.
4991    
4992          If we are not at the pattern start, compile code to change the ims          If we are not at the pattern start, compile code to change the ims
4993          options if this setting actually changes any of them. We also pass the          options if this setting actually changes any of them, and reset the
4994          new setting back so that it can be put at the start of any following          greedy defaults and the case value for firstbyte and reqbyte. */
         branches, and when this group ends (if we are in a group), a resetting  
         item can be compiled. */  
4995    
4996          if (*ptr == ')')          if (*ptr == CHAR_RIGHT_PARENTHESIS)
4997            {            {
4998            if (code == cd->start_code + 1 + LINK_SIZE &&            if (code == cd->start_code + 1 + LINK_SIZE &&
4999                 (lengthptr == NULL || *lengthptr == 2 + 2*LINK_SIZE))                 (lengthptr == NULL || *lengthptr == 2 + 2*LINK_SIZE))
5000              {              {
5001              cd->external_options = newoptions;              cd->external_options = newoptions;
             options = newoptions;  
5002              }              }
5003           else           else
5004              {              {
# Line 4531  for (;; ptr++) Line 5007  for (;; ptr++)
5007                *code++ = OP_OPT;                *code++ = OP_OPT;
5008                *code++ = newoptions & PCRE_IMS;                *code++ = newoptions & PCRE_IMS;
5009                }                }
   
             /* Change options at this level, and pass them back for use  
             in subsequent branches. Reset the greedy defaults and the case  
             value for firstbyte and reqbyte. */  
   
             *optionsptr = options = newoptions;  
5010              greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);              greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);
5011              greedy_non_default = greedy_default ^ 1;              greedy_non_default = greedy_default ^ 1;
5012              req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;              req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;
5013              }              }
5014    
5015              /* Change options at this level, and pass them back for use
5016              in subsequent branches. When not at the start of the pattern, this
5017              information is also necessary so that a resetting item can be
5018              compiled at the end of a group (if we are in a group). */
5019    
5020              *optionsptr = options = newoptions;
5021            previous = NULL;       /* This item can't be repeated */            previous = NULL;       /* This item can't be repeated */
5022            continue;              /* It is complete */            continue;              /* It is complete */
5023            }            }
# Line 4657  for (;; ptr++) Line 5133  for (;; ptr++)
5133    
5134      /* Error if hit end of pattern */      /* Error if hit end of pattern */
5135    
5136      if (*ptr != ')')      if (*ptr != CHAR_RIGHT_PARENTHESIS)
5137        {        {
5138        *errorcodeptr = ERR14;        *errorcodeptr = ERR14;
5139        goto FAILED;        goto FAILED;
5140        }        }
5141    
5142      /* 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,
5143      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
5144      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
5145      a quantifier. */      duplicated by a quantifier.*/
5146    
5147      if (lengthptr != NULL)      if (lengthptr != NULL)
5148        {        {
# Line 4676  for (;; ptr++) Line 5152  for (;; ptr++)
5152          goto FAILED;          goto FAILED;
5153          }          }
5154        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;        *lengthptr += length_prevgroup - 2 - 2*LINK_SIZE;
5155        code++;        *code++ = OP_BRA;
5156        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
5157        *code++ = OP_KET;        *code++ = OP_KET;
5158        PUTINC(code, 0, 1 + LINK_SIZE);        PUTINC(code, 0, 1 + LINK_SIZE);
5159          break;    /* No need to waste time with special character handling */
5160        }        }
5161    
5162      /* Otherwise update the main code pointer to the end of the group. */      /* Otherwise update the main code pointer to the end of the group. */
5163    
5164      else code = tempcode;      code = tempcode;
5165    
5166      /* For a DEFINE group, required and first character settings are not      /* For a DEFINE group, required and first character settings are not
5167      relevant. */      relevant. */
# Line 4754  for (;; ptr++) Line 5231  for (;; ptr++)
5231      We can test for values between ESC_b and ESC_Z for the latter; this may      We can test for values between ESC_b and ESC_Z for the latter; this may
5232      have to change if any new ones are ever created. */      have to change if any new ones are ever created. */
5233    
5234      case '\\':      case CHAR_BACKSLASH:
5235      tempptr = ptr;      tempptr = ptr;
5236      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);      c = check_escape(&ptr, errorcodeptr, cd->bracount, options, FALSE);
5237      if (*errorcodeptr != 0) goto FAILED;      if (*errorcodeptr != 0) goto FAILED;
# Line 4763  for (;; ptr++) Line 5240  for (;; ptr++)
5240        {        {
5241        if (-c == ESC_Q)            /* Handle start of quoted string */        if (-c == ESC_Q)            /* Handle start of quoted string */
5242          {          {
5243          if (ptr[1] == '\\' && ptr[2] == 'E') ptr += 2; /* avoid empty string */          if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
5244            else inescq = TRUE;            ptr += 2;               /* avoid empty string */
5245                else inescq = TRUE;
5246          continue;          continue;
5247          }          }
5248    
# Line 4781  for (;; ptr++) Line 5259  for (;; ptr++)
5259        zerofirstbyte = firstbyte;        zerofirstbyte = firstbyte;
5260        zeroreqbyte = reqbyte;        zeroreqbyte = reqbyte;
5261    
5262          /* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'
5263          is a subroutine call by number (Oniguruma syntax). In fact, the value
5264          -ESC_g is returned only for these cases. So we don't need to check for <
5265          or ' if the value is -ESC_g. For the Perl syntax \g{n} the value is
5266          -ESC_REF+n, and for the Perl syntax \g{name} the result is -ESC_k (as
5267          that is a synonym for a named back reference). */
5268    
5269          if (-c == ESC_g)
5270            {
5271            const uschar *p;
5272            save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
5273            terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?
5274              CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
5275    
5276            /* These two statements stop the compiler for warning about possibly
5277            unset variables caused by the jump to HANDLE_NUMERICAL_RECURSION. In
5278            fact, because we actually check for a number below, the paths that
5279            would actually be in error are never taken. */
5280    
5281            skipbytes = 0;
5282            reset_bracount = FALSE;
5283    
5284            /* Test for a name */
5285    
5286            if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)
5287              {
5288              BOOL isnumber = TRUE;
5289              for (p = ptr + 1; *p != 0 && *p != terminator; p++)
5290                {
5291                if ((cd->ctypes[*p] & ctype_digit) == 0) isnumber = FALSE;
5292                if ((cd->ctypes[*p] & ctype_word) == 0) break;
5293                }
5294              if (*p != terminator)
5295                {
5296                *errorcodeptr = ERR57;
5297                break;
5298                }
5299              if (isnumber)
5300                {
5301                ptr++;
5302                goto HANDLE_NUMERICAL_RECURSION;
5303                }
5304              is_recurse = TRUE;
5305              goto NAMED_REF_OR_RECURSE;
5306              }
5307    
5308            /* Test a signed number in angle brackets or quotes. */
5309    
5310            p = ptr + 2;
5311            while ((digitab[*p] & ctype_digit) != 0) p++;
5312            if (*p != terminator)
5313              {
5314              *errorcodeptr = ERR57;
5315              break;
5316              }
5317            ptr++;
5318            goto HANDLE_NUMERICAL_RECURSION;
5319            }
5320    
5321        /* \k<name> or \k'name' is a back reference by name (Perl syntax).        /* \k<name> or \k'name' is a back reference by name (Perl syntax).
5322        We also support \k{name} (.NET syntax) */        We also support \k{name} (.NET syntax) */
5323    
5324        if (-c == ESC_k && (ptr[1] == '<' || ptr[1] == '\'' || ptr[1] == '{'))        if (-c == ESC_k && (ptr[1] == CHAR_LESS_THAN_SIGN ||
5325              ptr[1] == CHAR_APOSTROPHE || ptr[1] == CHAR_LEFT_CURLY_BRACKET))
5326          {          {
5327          is_recurse = FALSE;          is_recurse = FALSE;
5328          terminator = (*(++ptr) == '<')? '>' : (*ptr == '\'')? '\'' : '}';          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?
5329              CHAR_GREATER_THAN_SIGN : (*ptr == CHAR_APOSTROPHE)?
5330              CHAR_APOSTROPHE : CHAR_RIGHT_CURLY_BRACKET;
5331          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
5332          }          }
5333    
# Line 4888  for (;; ptr++) Line 5428  for (;; ptr++)
5428      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;      *code++ = ((options & PCRE_CASELESS) != 0)? OP_CHARNC : OP_CHAR;
5429      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];      for (c = 0; c < mclength; c++) *code++ = mcbuffer[c];
5430    
5431        /* Remember if \r or \n were seen */
5432    
5433        if (mcbuffer[0] == CHAR_CR || mcbuffer[0] == CHAR_NL)
5434          cd->external_flags |= PCRE_HASCRORLF;
5435    
5436      /* Set the first and required bytes appropriately. If no previous first      /* Set the first and required bytes appropriately. If no previous first
5437      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.
5438      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 5130  for (;;) Line 5675  for (;;)
5675    compile a resetting op-code following, except at the very end of the pattern.    compile a resetting op-code following, except at the very end of the pattern.
5676    Return leaving the pointer at the terminating char. */    Return leaving the pointer at the terminating char. */
5677    
5678    if (*ptr != '|')    if (*ptr != CHAR_VERTICAL_LINE)
5679      {      {
5680      if (lengthptr == NULL)      if (lengthptr == NULL)
5681        {        {
# Line 5153  for (;;) Line 5698  for (;;)
5698    
5699      /* Resetting option if needed */      /* Resetting option if needed */
5700    
5701      if ((options & PCRE_IMS) != oldims && *ptr == ')')      if ((options & PCRE_IMS) != oldims && *ptr == CHAR_RIGHT_PARENTHESIS)
5702        {        {
5703        *code++ = OP_OPT;        *code++ = OP_OPT;
5704        *code++ = oldims;        *code++ = oldims;
# Line 5282  do { Line 5827  do {
5827       if (!is_anchored(scode, options, bracket_map, backref_map)) return FALSE;       if (!is_anchored(scode, options, bracket_map, backref_map)) return FALSE;
5828       }       }
5829    
5830     /* .* is not anchored unless DOTALL is set and it isn't in brackets that     /* .* is not anchored unless DOTALL is set (which generates OP_ALLANY) and
5831     are or may be referenced. */     it isn't in brackets that are or may be referenced. */
5832    
5833     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||     else if ((op == OP_TYPESTAR || op == OP_TYPEMINSTAR ||
5834               op == OP_TYPEPOSSTAR) &&               op == OP_TYPEPOSSTAR))
             (*options & PCRE_DOTALL) != 0)  
5835       {       {
5836       if (scode[1] != OP_ANY || (bracket_map & backref_map) != 0) return FALSE;       if (scode[1] != OP_ALLANY || (bracket_map & backref_map) != 0)
5837           return FALSE;
5838       }       }
5839    
5840     /* Check for explicit anchoring */     /* Check for explicit anchoring */
# Line 5335  do { Line 5880  do {
5880       NULL, 0, FALSE);       NULL, 0, FALSE);
5881     register int op = *scode;     register int op = *scode;
5882    
5883       /* If we are at the start of a conditional assertion group, *both* the
5884       conditional assertion *and* what follows the condition must satisfy the test
5885       for start of line. Other kinds of condition fail. Note that there may be an
5886       auto-callout at the start of a condition. */
5887    
5888       if (op == OP_COND)
5889         {
5890         scode += 1 + LINK_SIZE;
5891         if (*scode == OP_CALLOUT) scode += _pcre_OP_lengths[OP_CALLOUT];
5892         switch (*scode)
5893           {
5894           case OP_CREF:
5895           case OP_RREF:
5896           case OP_DEF:
5897           return FALSE;
5898    
5899           default:     /* Assertion */
5900           if (!is_startline(scode, bracket_map, backref_map)) return FALSE;
5901           do scode += GET(scode, 1); while (*scode == OP_ALT);
5902           scode += 1 + LINK_SIZE;
5903           break;
5904           }
5905         scode = first_significant_code(scode, NULL, 0, FALSE);
5906         op = *scode;
5907         }
5908    
5909     /* Non-capturing brackets */     /* Non-capturing brackets */
5910    
5911     if (op == OP_BRA)     if (op == OP_BRA)
# Line 5353  do { Line 5924  do {
5924    
5925     /* Other brackets */     /* Other brackets */
5926    
5927     else if (op == OP_ASSERT || op == OP_ONCE || op == OP_COND)     else if (op == OP_ASSERT || op == OP_ONCE)
5928       { if (!is_startline(scode, bracket_map, backref_map)) return FALSE; }       {
5929         if (!is_startline(scode, bracket_map, backref_map)) return FALSE;
5930         }
5931    
5932     /* .* means "start at start or after \n" if it isn't in brackets that     /* .* means "start at start or after \n" if it isn't in brackets that
5933     may be referenced. */     may be referenced. */
# Line 5471  Returns:        pointer to compiled data Line 6044  Returns:        pointer to compiled data
6044                  with errorptr and erroroffset set                  with errorptr and erroroffset set
6045  */  */
6046    
6047  PCRE_EXP_DEFN pcre *  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
6048  pcre_compile(const char *pattern, int options, const char **errorptr,  pcre_compile(const char *pattern, int options, const char **errorptr,
6049    int *erroroffset, const unsigned char *tables)    int *erroroffset, const unsigned char *tables)
6050  {  {
# Line 5479  return pcre_compile2(pattern, options, N Line 6052  return pcre_compile2(pattern, options, N
6052  }  }
6053    
6054    
6055  PCRE_EXP_DEFN pcre *  PCRE_EXP_DEFN pcre * PCRE_CALL_CONVENTION
6056  pcre_compile2(const char *pattern, int options, int *errorcodeptr,  pcre_compile2(const char *pattern, int options, int *errorcodeptr,
6057    const char **errorptr, int *erroroffset, const unsigned char *tables)    const char **errorptr, int *erroroffset, const unsigned char *tables)
6058  {  {
# Line 5487  real_pcre *re; Line 6060  real_pcre *re;
6060  int length = 1;  /* For final END opcode */  int length = 1;  /* For final END opcode */
6061  int firstbyte, reqbyte, newline;  int firstbyte, reqbyte, newline;
6062  int errorcode = 0;  int errorcode = 0;
6063    int skipatstart = 0;
6064  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF8
6065  BOOL utf8;  BOOL utf8;
6066  #endif  #endif
# Line 5505  to fill in forward references to subpatt Line 6079  to fill in forward references to subpatt
6079    
6080  uschar cworkspace[COMPILE_WORK_SIZE];  uschar cworkspace[COMPILE_WORK_SIZE];
6081    
   
6082  /* Set this early so that early errors get offset 0. */  /* Set this early so that early errors get offset 0. */
6083    
6084  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern;
# Line 5551  if ((options & PCRE_UTF8) != 0) Line 6124  if ((options & PCRE_UTF8) != 0)
6124    }    }
6125  #endif  #endif
6126    
6127  if ((options & ~PUBLIC_OPTIONS) != 0)  if ((options & ~PUBLIC_COMPILE_OPTIONS) != 0)
6128    {    {
6129    errorcode = ERR17;    errorcode = ERR17;
6130    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
# Line 5565  cd->fcc = tables + fcc_offset; Line 6138  cd->fcc = tables + fcc_offset;
6138  cd->cbits = tables + cbits_offset;  cd->cbits = tables + cbits_offset;
6139  cd->ctypes = tables + ctypes_offset;  cd->ctypes = tables + ctypes_offset;
6140    
6141    /* Check for global one-time settings at the start of the pattern, and remember
6142    the offset for later. */
6143    
6144    while (ptr[skipatstart] == CHAR_LEFT_PARENTHESIS &&
6145           ptr[skipatstart+1] == CHAR_ASTERISK)
6146      {
6147      int newnl = 0;
6148      int newbsr = 0;
6149    
6150      if (strncmp((char *)(ptr+skipatstart+2), STRING_CR_RIGHTPAR, 3) == 0)
6151        { skipatstart += 5; newnl = PCRE_NEWLINE_CR; }
6152      else if (strncmp((char *)(ptr+skipatstart+2), STRING_LF_RIGHTPAR, 3)  == 0)
6153        { skipatstart += 5; newnl = PCRE_NEWLINE_LF; }
6154      else if (strncmp((char *)(ptr+skipatstart+2), STRING_CRLF_RIGHTPAR, 5)  == 0)
6155        { skipatstart += 7; newnl = PCRE_NEWLINE_CR + PCRE_NEWLINE_LF; }
6156      else if (strncmp((char *)(ptr+skipatstart+2), STRING_ANY_RIGHTPAR, 4) == 0)
6157        { skipatstart += 6; newnl = PCRE_NEWLINE_ANY; }
6158      else if (strncmp((char *)(ptr+skipatstart+2), STRING_ANYCRLF_RIGHTPAR, 8) == 0)
6159        { skipatstart += 10; newnl = PCRE_NEWLINE_ANYCRLF; }
6160    
6161      else if (strncmp((char *)(ptr+skipatstart+2), STRING_BSR_ANYCRLF_RIGHTPAR, 12) == 0)
6162        { skipatstart += 14; newbsr = PCRE_BSR_ANYCRLF; }
6163      else if (strncmp((char *)(ptr+skipatstart+2), STRING_BSR_UNICODE_RIGHTPAR, 12) == 0)
6164        { skipatstart += 14; newbsr = PCRE_BSR_UNICODE; }
6165    
6166      if (newnl != 0)
6167        options = (options & ~PCRE_NEWLINE_BITS) | newnl;
6168      else if (newbsr != 0)
6169        options = (options & ~(PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)) | newbsr;
6170      else break;
6171      }
6172    
6173    /* Check validity of \R options. */
6174    
6175    switch (options & (PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE))
6176      {
6177      case 0:
6178      case PCRE_BSR_ANYCRLF:
6179      case PCRE_BSR_UNICODE:
6180      break;
6181      default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
6182      }
6183    
6184  /* Handle different types of newline. The three bits give seven cases. The  /* Handle different types of newline. The three bits give seven cases. The
6185  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
6186  "anycrlf". */  "anycrlf". */
6187    
6188  switch (options & (PCRE_NEWLINE_CRLF | PCRE_NEWLINE_ANY))  switch (options & PCRE_NEWLINE_BITS)
6189    {    {
6190    case 0: newline = NEWLINE; break;   /* Compile-time default */    case 0: newline = NEWLINE; break;   /* Build-time default */
6191    case PCRE_NEWLINE_CR: newline = '\r'; break;    case PCRE_NEWLINE_CR: newline = CHAR_CR; break;
6192    case PCRE_NEWLINE_LF: newline = '\n'; break;    case PCRE_NEWLINE_LF: newline = CHAR_NL; break;
6193    case PCRE_NEWLINE_CR+    case PCRE_NEWLINE_CR+
6194         PCRE_NEWLINE_LF: newline = ('\r' << 8) | '\n'; break;         PCRE_NEWLINE_LF: newline = (CHAR_CR << 8) | CHAR_NL; break;
6195    case PCRE_NEWLINE_ANY: newline = -1; break;    case PCRE_NEWLINE_ANY: newline = -1; break;
6196    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;    case PCRE_NEWLINE_ANYCRLF: newline = -2; break;
6197    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;    default: errorcode = ERR56; goto PCRE_EARLY_ERROR_RETURN;
# Line 5624  to compile parts of the pattern into; th Line 6240  to compile parts of the pattern into; th
6240  no longer needed, so hopefully this workspace will never overflow, though there  no longer needed, so hopefully this workspace will never overflow, though there
6241  is a test for its doing so. */  is a test for its doing so. */
6242    
6243  cd->bracount = 0;  cd->bracount = cd->final_bracount = 0;
6244  cd->names_found = 0;  cd->names_found = 0;
6245  cd->name_entry_size = 0;  cd->name_entry_size = 0;
6246  cd->name_table = NULL;  cd->name_table = NULL;
# Line 5634  cd->hwm = cworkspace; Line 6250  cd->hwm = cworkspace;
6250  cd->start_pattern = (const uschar *)pattern;  cd->start_pattern = (const uschar *)pattern;
6251  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));  cd->end_pattern = (const uschar *)(pattern + strlen(pattern));
6252  cd->req_varyopt = 0;  cd->req_varyopt = 0;
 cd->nopartial = FALSE;  
6253  cd->external_options = options;  cd->external_options = options;
6254    cd->external_flags = 0;
6255    
6256  /* 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
6257  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 5643  been put into the cd block so that they Line 6259  been put into the cd block so that they
6259  found within the regex right at the beginning. Bringing initial option settings  found within the regex right at the beginning. Bringing initial option settings
6260  outside can help speed up starting point checks. */  outside can help speed up starting point checks. */
6261    
6262    ptr += skipatstart;
6263  code = cworkspace;  code = cworkspace;
6264  *code = OP_BRA;  *code = OP_BRA;
6265  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,  (void)compile_regex(cd->external_options, cd->external_options & PCRE_IMS,
# Line 5673  if (re == NULL) Line 6290  if (re == NULL)
6290    goto PCRE_EARLY_ERROR_RETURN;    goto PCRE_EARLY_ERROR_RETURN;
6291    }    }
6292    
6293  /* 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
6294  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
6295  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
6296  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
6297    pointers. */
6298    
6299  re->magic_number = MAGIC_NUMBER;  re->magic_number = MAGIC_NUMBER;
6300  re->size = size;  re->size = size;
6301  re->options = cd->external_options;  re->options = cd->external_options;
6302    re->flags = cd->external_flags;
6303  re->dummy1 = 0;  re->dummy1 = 0;
6304  re->first_byte = 0;  re->first_byte = 0;
6305  re->req_byte = 0;  re->req_byte = 0;
# Line 5698  field. Reset the bracket count and the n Line 6317  field. Reset the bracket count and the n
6317  field; this time it's used for remembering forward references to subpatterns.  field; this time it's used for remembering forward references to subpatterns.
6318  */  */
6319    
6320    cd->final_bracount = cd->bracount;  /* Save for checking forward references */
6321  cd->bracount = 0;  cd->bracount = 0;
6322  cd->names_found = 0;  cd->names_found = 0;
6323  cd->name_table = (uschar *)re + re->name_table_offset;  cd->name_table = (uschar *)re + re->name_table_offset;
# Line 5705  codestart = cd->name_table + re->name_en Line 6325  codestart = cd->name_table + re->name_en
6325  cd->start_code = codestart;  cd->start_code = codestart;
6326  cd->hwm = cworkspace;  cd->hwm = cworkspace;
6327  cd->req_varyopt = 0;  cd->req_varyopt = 0;
6328  cd->nopartial = FALSE;  cd->had_accept = FALSE;
6329    
6330  /* Set up a starting, non-extracting bracket, then compile the expression. On  /* Set up a starting, non-extracting bracket, then compile the expression. On
6331  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
6332  of the function here. */  of the function here. */
6333    
6334  ptr = (const uschar *)pattern;  ptr = (const uschar *)pattern + skipatstart;
6335  code = (uschar *)codestart;  code = (uschar *)codestart;
6336  *code = OP_BRA;  *code = OP_BRA;
6337  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,  (void)compile_regex(re->options, re->options & PCRE_IMS, &code, &ptr,
6338    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);    &errorcode, FALSE, FALSE, 0, &firstbyte, &reqbyte, NULL, cd, NULL);
6339  re->top_bracket = cd->bracount;  re->top_bracket = cd->bracount;
6340  re->top_backref = cd->top_backref;  re->top_backref = cd->top_backref;
6341    re->flags = cd->external_flags;
6342    
6343  if (cd->nopartial) re->options |= PCRE_NOPARTIAL;  if (cd->had_accept) reqbyte = -1;   /* Must disable after (*ACCEPT) */
6344    
6345  /* 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. */
6346    
# Line 5761  if (errorcode != 0) Line 6382  if (errorcode != 0)
6382    PCRE_EARLY_ERROR_RETURN:    PCRE_EARLY_ERROR_RETURN:
6383    *erroroffset = ptr - (const uschar *)pattern;    *erroroffset = ptr - (const uschar *)pattern;
6384    PCRE_EARLY_ERROR_RETURN2:    PCRE_EARLY_ERROR_RETURN2:
6385    *errorptr = error_texts[errorcode];    *errorptr = find_error_text(errorcode);
6386    if (errorcodeptr != NULL) *errorcodeptr = errorcode;    if (errorcodeptr != NULL) *errorcodeptr = errorcode;
6387    return NULL;    return NULL;
6388    }    }
# Line 5790  if ((re->options & PCRE_ANCHORED) == 0) Line 6411  if ((re->options & PCRE_ANCHORED) == 0)
6411        int ch = firstbyte & 255;        int ch = firstbyte & 255;
6412        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&        re->first_byte = ((firstbyte & REQ_CASELESS) != 0 &&
6413           cd->fcc[ch] == ch)? ch : firstbyte;           cd->fcc[ch] == ch)? ch : firstbyte;
6414        re->options |= PCRE_FIRSTSET;        re->flags |= PCRE_FIRSTSET;
6415        }        }
6416      else if (is_startline(codestart, 0, cd->backref_map))      else if (is_startline(codestart, 0, cd->backref_map))
6417        re->options |= PCRE_STARTLINE;        re->flags |= PCRE_STARTLINE;
6418      }      }
6419    }    }
6420    
# Line 5807  if (reqbyte >= 0 && Line 6428  if (reqbyte >= 0 &&
6428    int ch = reqbyte & 255;    int ch = reqbyte & 255;
6429    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&    re->req_byte = ((reqbyte & REQ_CASELESS) != 0 &&
6430      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;      cd->fcc[ch] == ch)? (reqbyte & ~REQ_CASELESS) : reqbyte;
6431    re->options |= PCRE_REQCHSET;    re->flags |= PCRE_REQCHSET;
6432    }    }
6433    
6434  /* 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 5818  case when building a production library. Line 6439  case when building a production library.
6439  printf("Length = %d top_bracket = %d top_backref = %d\n",  printf("Length = %d top_bracket = %d top_backref = %d\n",
6440    length, re->top_bracket, re->top_backref);    length, re->top_bracket, re->top_backref);
6441    
6442  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 " : "");  
   }  
6443    
6444  if ((re->options & PCRE_FIRSTSET) != 0)  if ((re->flags & PCRE_FIRSTSET) != 0)
6445    {    {
6446    int ch = re->first_byte & 255;    int ch = re->first_byte & 255;
6447    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->first_byte & REQ_CASELESS) == 0)?
# Line 5841  if ((re->options & PCRE_FIRSTSET) != 0) Line 6450  if ((re->options & PCRE_FIRSTSET) != 0)
6450      else printf("First char = \\x%02x%s\n", ch, caseless);      else printf("First char = \\x%02x%s\n", ch, caseless);
6451    }    }
6452    
6453  if ((re->options & PCRE_REQCHSET) != 0)  if ((re->flags & PCRE_REQCHSET) != 0)
6454    {    {
6455    int ch = re->req_byte & 255;    int ch = re->req_byte & 255;
6456    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?    const char *caseless = ((re->req_byte & REQ_CASELESS) == 0)?
# Line 5858  was compiled can be seen. */ Line 6467  was compiled can be seen. */
6467  if (code - codestart > length)  if (code - codestart > length)
6468    {    {
6469    (pcre_free)(re);    (pcre_free)(re);
6470    *errorptr = error_texts[ERR23];    *errorptr = find_error_text(ERR23);
6471    *erroroffset = ptr - (uschar *)pattern;    *erroroffset = ptr - (uschar *)pattern;
6472    if (errorcodeptr != NULL) *errorcodeptr = ERR23;    if (errorcodeptr != NULL) *errorcodeptr = ERR23;
6473    return NULL;    return NULL;

Legend:
Removed from v.206  
changed lines
  Added in v.392

  ViewVC Help
Powered by ViewVC 1.1.5