/[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 640 by ph10, Mon Jul 25 10:50:28 2011 UTC revision 1048 by ph10, Fri Sep 28 16:03:14 2012 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) 1997-2011 University of Cambridge             Copyright (c) 1997-2012 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 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is  /* When PCRE_DEBUG is defined, we need the pcre(16)_printint() function, which
57  also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. */  library. We do not need to select pcre16_printint.c specially, because the
59    COMPILE_PCREx macro will already be appropriately set. */
60    
61  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
62  #include "pcre_printint.src"  /* pcre_printint.c should not include any headers */
63    #define PCRE_INCLUDED
64    #include "pcre_printint.c"
65    #undef PCRE_INCLUDED
66  #endif  #endif
67    
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 73  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL,
88        int, int, int *, int *, branch_chain *, compile_data *, int *);
89    
90    
91    
92  /*************************************************  /*************************************************
93  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 88  so this number is very generous. Line 103  so this number is very generous.
103  The same workspace is used during the second, actual compile phase for  The same workspace is used during the second, actual compile phase for
104  remembering forward references to groups so that they can be filled in at the  remembering forward references to groups so that they can be filled in at the
105  end. Each entry in this list occupies LINK_SIZE bytes, so even when LINK_SIZE  end. Each entry in this list occupies LINK_SIZE bytes, so even when LINK_SIZE
106  is 4 there is plenty of room. */  is 4 there is plenty of room for most patterns. However, the memory can get
107    filled up by repetitions of forward references, for example patterns like
108    /(?1){0,1999}(b)/, and one user did hit the limit. The code has been changed so
109    that the workspace is expanded using malloc() in this situation. The value
110    below is therefore a minimum, and we put a maximum on it for safety. The
111    minimum is now also defined in terms of LINK_SIZE so that the use of malloc()
112    kicks in at the same number of forward references in all cases. */
113    
114  #define COMPILE_WORK_SIZE (4096)  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)
115    #define COMPILE_WORK_SIZE_MAX (100*COMPILE_WORK_SIZE)
116    
117  /* The overrun tests check for a slightly smaller size so that they detect the  /* The overrun tests check for a slightly smaller size so that they detect the
118  overrun before it actually does run off the end of the data block. */  overrun before it actually does run off the end of the data block. */
119    
120  #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100)  #define WORK_SIZE_SAFETY_MARGIN (100)
121    
122    /* Private flags added to firstchar and reqchar. */
123    
124    #define REQ_CASELESS   0x10000000l      /* Indicates caselessness */
125    #define REQ_VARY       0x20000000l      /* Reqchar followed non-literal item */
126    
127    /* Repeated character flags. */
128    
129    #define UTF_LENGTH     0x10000000l      /* The char contains its length. */
130    
131  /* Table for handling escaped characters in the range '0'-'z'. Positive returns  /* Table for handling escaped characters in the range '0'-'z'. Positive returns
132  are simple data values; negative values are for special things like \d and so  are simple data values; negative values are for special things like \d and so
# Line 231  static const char posix_names[] = Line 261  static const char posix_names[] =
261    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0
262    STRING_word0  STRING_xdigit;    STRING_word0  STRING_xdigit;
263    
264  static const uschar posix_name_lengths[] = {  static const pcre_uint8 posix_name_lengths[] = {
265    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 };
266    
267  /* Table of class bit maps for each POSIX class. Each class is formed from a  /* Table of class bit maps for each POSIX class. Each class is formed from a
# Line 266  substitutes must be in the order of the Line 296  substitutes must be in the order of the
296  both positive and negative cases. NULL means no substitute. */  both positive and negative cases. NULL means no substitute. */
297    
298  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
299  static const uschar *substitutes[] = {  static const pcre_uchar string_PNd[]  = {
300    (uschar *)"\\P{Nd}",    /* \D */    CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
301    (uschar *)"\\p{Nd}",    /* \d */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
302    (uschar *)"\\P{Xsp}",   /* \S */       /* NOTE: Xsp is Perl space */  static const pcre_uchar string_pNd[]  = {
303    (uschar *)"\\p{Xsp}",   /* \s */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
304    (uschar *)"\\P{Xwd}",   /* \W */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
305    (uschar *)"\\p{Xwd}"    /* \w */  static const pcre_uchar string_PXsp[] = {
306      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
307      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
308    static const pcre_uchar string_pXsp[] = {
309      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
310      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
311    static const pcre_uchar string_PXwd[] = {
312      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
313      CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
314    static const pcre_uchar string_pXwd[] = {
315      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
316      CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
317    
318    static const pcre_uchar *substitutes[] = {
319      string_PNd,           /* \D */
320      string_pNd,           /* \d */
321      string_PXsp,          /* \S */       /* NOTE: Xsp is Perl space */
322      string_pXsp,          /* \s */
323      string_PXwd,          /* \W */
324      string_pXwd           /* \w */
325  };  };
326    
327  static const uschar *posix_substitutes[] = {  static const pcre_uchar string_pL[] =   {
328    (uschar *)"\\p{L}",     /* alpha */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
329    (uschar *)"\\p{Ll}",    /* lower */    CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
330    (uschar *)"\\p{Lu}",    /* upper */  static const pcre_uchar string_pLl[] =  {
331    (uschar *)"\\p{Xan}",   /* alnum */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
332    NULL,                   /* ascii */    CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
333    (uschar *)"\\h",        /* blank */  static const pcre_uchar string_pLu[] =  {
334    NULL,                   /* cntrl */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
335    (uschar *)"\\p{Nd}",    /* digit */    CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
336    NULL,                   /* graph */  static const pcre_uchar string_pXan[] = {
337    NULL,                   /* print */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
338    NULL,                   /* punct */    CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
339    (uschar *)"\\p{Xps}",   /* space */    /* NOTE: Xps is POSIX space */  static const pcre_uchar string_h[] =    {
340    (uschar *)"\\p{Xwd}",   /* word */    CHAR_BACKSLASH, CHAR_h, '\0' };
341    NULL,                   /* xdigit */  static const pcre_uchar string_pXps[] = {
342      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
343      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
344    static const pcre_uchar string_PL[] =   {
345      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
346      CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
347    static const pcre_uchar string_PLl[] =  {
348      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
349      CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
350    static const pcre_uchar string_PLu[] =  {
351      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
352      CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
353    static const pcre_uchar string_PXan[] = {
354      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
355      CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
356    static const pcre_uchar string_H[] =    {
357      CHAR_BACKSLASH, CHAR_H, '\0' };
358    static const pcre_uchar string_PXps[] = {
359      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
360      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
361    
362    static const pcre_uchar *posix_substitutes[] = {
363      string_pL,            /* alpha */
364      string_pLl,           /* lower */
365      string_pLu,           /* upper */
366      string_pXan,          /* alnum */
367      NULL,                 /* ascii */
368      string_h,             /* blank */
369      NULL,                 /* cntrl */
370      string_pNd,           /* digit */
371      NULL,                 /* graph */
372      NULL,                 /* print */
373      NULL,                 /* punct */
374      string_pXps,          /* space */    /* NOTE: Xps is POSIX space */
375      string_pXwd,          /* word */
376      NULL,                 /* xdigit */
377    /* Negated cases */    /* Negated cases */
378    (uschar *)"\\P{L}",     /* ^alpha */    string_PL,            /* ^alpha */
379    (uschar *)"\\P{Ll}",    /* ^lower */    string_PLl,           /* ^lower */
380    (uschar *)"\\P{Lu}",    /* ^upper */    string_PLu,           /* ^upper */
381    (uschar *)"\\P{Xan}",   /* ^alnum */    string_PXan,          /* ^alnum */
382    NULL,                   /* ^ascii */    NULL,                 /* ^ascii */
383    (uschar *)"\\H",        /* ^blank */    string_H,             /* ^blank */
384    NULL,                   /* ^cntrl */    NULL,                 /* ^cntrl */
385    (uschar *)"\\P{Nd}",    /* ^digit */    string_PNd,           /* ^digit */
386    NULL,                   /* ^graph */    NULL,                 /* ^graph */
387    NULL,                   /* ^print */    NULL,                 /* ^print */
388    NULL,                   /* ^punct */    NULL,                 /* ^punct */
389    (uschar *)"\\P{Xps}",   /* ^space */   /* NOTE: Xps is POSIX space */    string_PXps,          /* ^space */   /* NOTE: Xps is POSIX space */
390    (uschar *)"\\P{Xwd}",   /* ^word */    string_PXwd,          /* ^word */
391    NULL                    /* ^xdigit */    NULL                  /* ^xdigit */
392  };  };
393  #define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))  #define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))
394  #endif  #endif
395    
396  #define STRING(a)  # a  #define STRING(a)  # a
# Line 365  static const char error_texts[] = Line 449  static const char error_texts[] =
449    /* 30 */    /* 30 */
450    "unknown POSIX class name\0"    "unknown POSIX class name\0"
451    "POSIX collating elements are not supported\0"    "POSIX collating elements are not supported\0"
452    "this version of PCRE is not compiled with PCRE_UTF8 support\0"    "this version of PCRE is compiled without UTF support\0"
453    "spare error\0"  /** DEAD **/    "spare error\0"  /** DEAD **/
454    "character value in \\x{...} sequence is too large\0"    "character value in \\x{...} sequence is too large\0"
455    /* 35 */    /* 35 */
# Line 388  static const char error_texts[] = Line 472  static const char error_texts[] =
472    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
473    /* 50 */    /* 50 */
474    "repeated subpattern is too long\0"    /** DEAD **/    "repeated subpattern is too long\0"    /** DEAD **/
475    "octal value is greater than \\377 (not in UTF-8 mode)\0"    "octal value is greater than \\377 in 8-bit non-UTF-8 mode\0"
476    "internal error: overran compiling workspace\0"    "internal error: overran compiling workspace\0"
477    "internal error: previously-checked referenced subpattern not found\0"    "internal error: previously-checked referenced subpattern not found\0"
478    "DEFINE group contains more than one branch\0"    "DEFINE group contains more than one branch\0"
# Line 407  static const char error_texts[] = Line 491  static const char error_texts[] =
491    /* 65 */    /* 65 */
492    "different names for subpatterns of the same number are not allowed\0"    "different names for subpatterns of the same number are not allowed\0"
493    "(*MARK) must have an argument\0"    "(*MARK) must have an argument\0"
494    "this version of PCRE is not compiled with PCRE_UCP support\0"    "this version of PCRE is not compiled with Unicode property support\0"
495    "\\c must be followed by an ASCII character\0"    "\\c must be followed by an ASCII character\0"
496    "\\k is not followed by a braced, angle-bracketed, or quoted name\0"    "\\k is not followed by a braced, angle-bracketed, or quoted name\0"
497      /* 70 */
498      "internal error: unknown opcode in find_fixedlength()\0"
499      "\\N is not supported in a class\0"
500      "too many forward references\0"
501      "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
502      "invalid UTF-16 string\0"
503      /* 75 */
504      "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
505      "character value in \\u.... sequence is too large\0"
506    ;    ;
507    
508  /* 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 428  For convenience, we use the same bit def Line 521  For convenience, we use the same bit def
521    
522  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
523    
524    /* Using a simple comparison for decimal numbers rather than a memory read
525    is much faster, and the resulting code is simpler (the compiler turns it
526    into a subtraction and unsigned comparison). */
527    
528    #define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9)
529    
530  #ifndef EBCDIC  #ifndef EBCDIC
531    
532  /* This is the "normal" case, for ASCII systems, and EBCDIC systems running in  /* This is the "normal" case, for ASCII systems, and EBCDIC systems running in
533  UTF-8 mode. */  UTF-8 mode. */
534    
535  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
536    {    {
537    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
538    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */
# Line 472  static const unsigned char digitab[] = Line 571  static const unsigned char digitab[] =
571    
572  /* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */  /* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */
573    
574  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
575    {    {
576    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
577    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */
# Line 507  static const unsigned char digitab[] = Line 606  static const unsigned char digitab[] =
606    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */
607    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */
608    
609  static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */  static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */
610    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */
611    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */
612    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */
# Line 543  static const unsigned char ebcdic_charta Line 642  static const unsigned char ebcdic_charta
642  #endif  #endif
643    
644    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int *,  
     int *, branch_chain *, compile_data *, int *);  
   
645    
646    
647  /*************************************************  /*************************************************
# Line 578  return s; Line 671  return s;
671    
672    
673  /*************************************************  /*************************************************
674    *           Expand the workspace                 *
675    *************************************************/
676    
677    /* This function is called during the second compiling phase, if the number of
678    forward references fills the existing workspace, which is originally a block on
679    the stack. A larger block is obtained from malloc() unless the ultimate limit
680    has been reached or the increase will be rather small.
681    
682    Argument: pointer to the compile data block
683    Returns:  0 if all went well, else an error number
684    */
685    
686    static int
687    expand_workspace(compile_data *cd)
688    {
689    pcre_uchar *newspace;
690    int newsize = cd->workspace_size * 2;
691    
692    if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;
693    if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX ||
694        newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)
695     return ERR72;
696    
697    newspace = (PUBL(malloc))(IN_UCHARS(newsize));
698    if (newspace == NULL) return ERR21;
699    memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar));
700    cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace);
701    if (cd->workspace_size > COMPILE_WORK_SIZE)
702      (PUBL(free))((void *)cd->start_workspace);
703    cd->start_workspace = newspace;
704    cd->workspace_size = newsize;
705    return 0;
706    }
707    
708    
709    
710    /*************************************************
711  *            Check for counted repeat            *  *            Check for counted repeat            *
712  *************************************************/  *************************************************/
713    
# Line 593  Returns:    TRUE or FALSE Line 723  Returns:    TRUE or FALSE
723  */  */
724    
725  static BOOL  static BOOL
726  is_counted_repeat(const uschar *p)  is_counted_repeat(const pcre_uchar *p)
727  {  {
728  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
729  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
730    while (IS_DIGIT(*p)) p++;
731  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
732    
733  if (*p++ != CHAR_COMMA) return FALSE;  if (*p++ != CHAR_COMMA) return FALSE;
734  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
735    
736  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
737  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
738    while (IS_DIGIT(*p)) p++;
739    
740  return (*p == CHAR_RIGHT_CURLY_BRACKET);  return (*p == CHAR_RIGHT_CURLY_BRACKET);
741  }  }
# Line 635  Returns:         zero or positive => a d Line 767  Returns:         zero or positive => a d
767  */  */
768    
769  static int  static int
770  check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount,
771    int options, BOOL isclass)    int options, BOOL isclass)
772  {  {
773  BOOL utf8 = (options & PCRE_UTF8) != 0;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
774  const uschar *ptr = *ptrptr + 1;  BOOL utf = (options & PCRE_UTF8) != 0;
775  int c, i;  const pcre_uchar *ptr = *ptrptr + 1;
776    pcre_int32 c;
777    int i;
778    
779  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
780  ptr--;                            /* Set pointer back to the last byte */  ptr--;                            /* Set pointer back to the last byte */
# Line 654  in a table. A non-zero result is somethi Line 788  in a table. A non-zero result is somethi
788  Otherwise further processing may be required. */  Otherwise further processing may be required. */
789    
790  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
791  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */  /* Not alphanumeric */
792    else if (c < CHAR_0 || c > CHAR_z) {}
793  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0) c = i;
794    
795  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
796  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */  /* Not alphanumeric */
797    else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
798  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  c = i;
799  #endif  #endif
800    
# Line 666  else if ((i = escapes[c - 0x48]) != 0) Line 802  else if ((i = escapes[c - 0x48]) != 0)
802    
803  else  else
804    {    {
805    const uschar *oldptr;    const pcre_uchar *oldptr;
806    BOOL braced, negated;    BOOL braced, negated;
807    
808    switch (c)    switch (c)
# Line 676  else Line 812  else
812    
813      case CHAR_l:      case CHAR_l:
814      case CHAR_L:      case CHAR_L:
815        *errorcodeptr = ERR37;
816        break;
817    
818      case CHAR_u:      case CHAR_u:
819        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
820          {
821          /* In JavaScript, \u must be followed by four hexadecimal numbers.
822          Otherwise it is a lowercase u letter. */
823          if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
824            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
825            && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
826            && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
827            {
828            c = 0;
829            for (i = 0; i < 4; ++i)
830              {
831              register int cc = *(++ptr);
832    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
833              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
834              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
835    #else           /* EBCDIC coding */
836              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
837              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
838    #endif
839              }
840    
841    #ifdef COMPILE_PCRE8
842            if (c > (utf ? 0x10ffff : 0xff))
843    #else
844    #ifdef COMPILE_PCRE16
845            if (c > (utf ? 0x10ffff : 0xffff))
846    #endif
847    #endif
848              {
849              *errorcodeptr = ERR76;
850              }
851            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
852            }
853          }
854        else
855          *errorcodeptr = ERR37;
856        break;
857    
858      case CHAR_U:      case CHAR_U:
859      *errorcodeptr = ERR37;      /* In JavaScript, \U is an uppercase U letter. */
860        if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37;
861      break;      break;
862    
863      /* In a character class, \g is just a literal "g". Outside a character      /* In a character class, \g is just a literal "g". Outside a character
864      class, \g must be followed by one of a number of specific things:      class, \g must be followed by one of a number of specific things:
865    
866      (1) A number, either plain or braced. If positive, it is an absolute      (1) A number, either plain or braced. If positive, it is an absolute
# Line 710  else Line 889  else
889    
890      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
891        {        {
892        const uschar *p;        const pcre_uchar *p;
893        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
894          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
895        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
896          {          {
897          c = -ESC_k;          c = -ESC_k;
# Line 730  else Line 909  else
909        }        }
910      else negated = FALSE;      else negated = FALSE;
911    
912        /* The integer range is limited by the machine's int representation. */
913      c = 0;      c = 0;
914      while ((digitab[ptr[1]] & ctype_digit) != 0)      while (IS_DIGIT(ptr[1]))
915          {
916          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */
917            {
918            c = -1;
919            break;
920            }
921        c = c * 10 + *(++ptr) - CHAR_0;        c = c * 10 + *(++ptr) - CHAR_0;
922          }
923      if (c < 0)   /* Integer overflow */      if (((unsigned int)c) > INT_MAX) /* Integer overflow */
924        {        {
925          while (IS_DIGIT(ptr[1]))
926            ptr++;
927        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
928        break;        break;
929        }        }
# Line 783  else Line 971  else
971      if (!isclass)      if (!isclass)
972        {        {
973        oldptr = ptr;        oldptr = ptr;
974          /* The integer range is limited by the machine's int representation. */
975        c -= CHAR_0;        c -= CHAR_0;
976        while ((digitab[ptr[1]] & ctype_digit) != 0)        while (IS_DIGIT(ptr[1]))
977            {
978            if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */
979              {
980              c = -1;
981              break;
982              }
983          c = c * 10 + *(++ptr) - CHAR_0;          c = c * 10 + *(++ptr) - CHAR_0;
984        if (c < 0)    /* Integer overflow */          }
985          if (((unsigned int)c) > INT_MAX) /* Integer overflow */
986          {          {
987            while (IS_DIGIT(ptr[1]))
988              ptr++;
989          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
990          break;          break;
991          }          }
# Line 813  else Line 1011  else
1011      /* \0 always starts an octal number, but we may drop through to here with a      /* \0 always starts an octal number, but we may drop through to here with a
1012      larger first octal digit. The original code used just to take the least      larger first octal digit. The original code used just to take the least
1013      significant 8 bits of octal numbers (I think this is what early Perls used      significant 8 bits of octal numbers (I think this is what early Perls used
1014      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 and 16-bit mode,
1015      than 3 octal digits. */      but no more than 3 octal digits. */
1016    
1017      case CHAR_0:      case CHAR_0:
1018      c -= CHAR_0;      c -= CHAR_0;
1019      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1020          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1021      if (!utf8 && c > 255) *errorcodeptr = ERR51;  #ifdef COMPILE_PCRE8
1022        if (!utf && c > 0xff) *errorcodeptr = ERR51;
1023    #endif
1024      break;      break;
1025    
1026      /* \x is complicated. \x{ddd} is a character number which can be greater      /* \x is complicated. \x{ddd} is a character number which can be greater
1027      than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is      than 0xff in utf or non-8bit mode, but only if the ddd are hex digits.
1028      treated as a data character. */      If not, { is treated as a data character. */
1029    
1030      case CHAR_x:      case CHAR_x:
1031        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1032          {
1033          /* In JavaScript, \x must be followed by two hexadecimal numbers.
1034          Otherwise it is a lowercase x letter. */
1035          if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1036            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1037            {
1038            c = 0;
1039            for (i = 0; i < 2; ++i)
1040              {
1041              register int cc = *(++ptr);
1042    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1043              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1044              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
1045    #else           /* EBCDIC coding */
1046              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
1047              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1048    #endif
1049              }
1050            }
1051          break;
1052          }
1053    
1054      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1055        {        {
1056        const uschar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
       int count = 0;  
1057    
1058        c = 0;        c = 0;
1059        while ((digitab[*pt] & ctype_xdigit) != 0)        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1060          {          {
1061          register int cc = *pt++;          register int cc = *pt++;
1062          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
         count++;  
1063    
1064  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1065          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
# Line 847  else Line 1068  else
1068          if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */          if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
1069          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1070  #endif  #endif
1071    
1072    #ifdef COMPILE_PCRE8
1073            if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }
1074    #else
1075    #ifdef COMPILE_PCRE16
1076            if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }
1077    #endif
1078    #endif
1079            }
1080    
1081          if (c < 0)
1082            {
1083            while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1084            *errorcodeptr = ERR34;
1085          }          }
1086    
1087        if (*pt == CHAR_RIGHT_CURLY_BRACKET)        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
1088          {          {
1089          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1090          ptr = pt;          ptr = pt;
1091          break;          break;
1092          }          }
# Line 863  else Line 1098  else
1098      /* Read just a single-byte hex-defined char */      /* Read just a single-byte hex-defined char */
1099    
1100      c = 0;      c = 0;
1101      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1102        {        {
1103        int cc;                                  /* Some compilers don't like */        int cc;                                  /* Some compilers don't like */
1104        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
# Line 921  else Line 1156  else
1156    }    }
1157    
1158  /* Perl supports \N{name} for character names, as well as plain \N for "not  /* Perl supports \N{name} for character names, as well as plain \N for "not
1159  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1160  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1161    
1162  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
# Line 961  Returns:         type value from ucp_typ Line 1196  Returns:         type value from ucp_typ
1196  */  */
1197    
1198  static int  static int
1199  get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)
1200  {  {
1201  int c, i, bot, top;  int c, i, bot, top;
1202  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1203  char name[32];  pcre_uchar name[32];
1204    
1205  c = *(++ptr);  c = *(++ptr);
1206  if (c == 0) goto ERROR_RETURN;  if (c == 0) goto ERROR_RETURN;
# Line 982  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1217  if (c == CHAR_LEFT_CURLY_BRACKET)
1217      *negptr = TRUE;      *negptr = TRUE;
1218      ptr++;      ptr++;
1219      }      }
1220    for (i = 0; i < (int)sizeof(name) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1221      {      {
1222      c = *(++ptr);      c = *(++ptr);
1223      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 1006  else Line 1241  else
1241  /* Search for a recognized property name using binary chop */  /* Search for a recognized property name using binary chop */
1242    
1243  bot = 0;  bot = 0;
1244  top = _pcre_utt_size;  top = PRIV(utt_size);
1245    
1246  while (bot < top)  while (bot < top)
1247    {    {
1248    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1249    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);    c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1250    if (c == 0)    if (c == 0)
1251      {      {
1252      *dptr = _pcre_utt[i].value;      *dptr = PRIV(utt)[i].value;
1253      return _pcre_utt[i].type;      return PRIV(utt)[i].type;
1254      }      }
1255    if (c > 0) bot = i + 1; else top = i;    if (c > 0) bot = i + 1; else top = i;
1256    }    }
# Line 1053  Returns:         pointer to '}' on succe Line 1288  Returns:         pointer to '}' on succe
1288                   current ptr on error, with errorcodeptr set non-zero                   current ptr on error, with errorcodeptr set non-zero
1289  */  */
1290    
1291  static const uschar *  static const pcre_uchar *
1292  read_repeat_counts(const uschar *p, int *minp, int *maxp, int *errorcodeptr)  read_repeat_counts(const pcre_uchar *p, int *minp, int *maxp, int *errorcodeptr)
1293  {  {
1294  int min = 0;  int min = 0;
1295  int max = -1;  int max = -1;
# Line 1062  int max = -1; Line 1297  int max = -1;
1297  /* 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
1298  an integer overflow. */  an integer overflow. */
1299    
1300  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;
1301  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1302    {    {
1303    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1077  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1312  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1312    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1313      {      {
1314      max = 0;      max = 0;
1315      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;
1316      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1317        {        {
1318        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1132  Arguments: Line 1367  Arguments:
1367    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1368    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1369    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1370    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 mode
1371    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1372    
1373  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1374  */  */
1375    
1376  static int  static int
1377  find_parens_sub(uschar **ptrptr, compile_data *cd, const uschar *name, int lorn,  find_parens_sub(pcre_uchar **ptrptr, compile_data *cd, const pcre_uchar *name, int lorn,
1378    BOOL xmode, BOOL utf8, int *count)    BOOL xmode, BOOL utf, int *count)
1379  {  {
1380  uschar *ptr = *ptrptr;  pcre_uchar *ptr = *ptrptr;
1381  int start_count = *count;  int start_count = *count;
1382  int hwm_count = start_count;  int hwm_count = start_count;
1383  BOOL dup_parens = FALSE;  BOOL dup_parens = FALSE;
# Line 1209  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1444  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1444          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1445        {        {
1446        int term;        int term;
1447        const uschar *thisname;        const pcre_uchar *thisname;
1448        *count += 1;        *count += 1;
1449        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1450        term = *ptr++;        term = *ptr++;
# Line 1217  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1452  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1452        thisname = ptr;        thisname = ptr;
1453        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1454        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == ptr - thisname &&
1455            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, lorn) == 0)
1456          return *count;          return *count;
1457        term++;        term++;
1458        }        }
# Line 1260  for (; ptr < cd->end_pattern; ptr++) Line 1495  for (; ptr < cd->end_pattern; ptr++)
1495          {          {
1496          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1497            ptr+= 2;            ptr+= 2;
1498          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1499                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1500            ptr += 4;            ptr += 4;
1501          else          else
# Line 1308  for (; ptr < cd->end_pattern; ptr++) Line 1543  for (; ptr < cd->end_pattern; ptr++)
1543        {        {
1544        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1545        ptr++;        ptr++;
1546  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1547        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1548  #endif  #endif
1549        }        }
1550      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
# Line 1320  for (; ptr < cd->end_pattern; ptr++) Line 1555  for (; ptr < cd->end_pattern; ptr++)
1555    
1556    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1557      {      {
1558      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1559      if (rc > 0) return rc;      if (rc > 0) return rc;
1560      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1561      }      }
# Line 1366  Arguments: Line 1601  Arguments:
1601    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1602    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1603    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1604    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 mode
1605    
1606  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1607  */  */
1608    
1609  static int  static int
1610  find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode,  find_parens(compile_data *cd, const pcre_uchar *name, int lorn, BOOL xmode,
1611    BOOL utf8)    BOOL utf)
1612  {  {
1613  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1614  int count = 0;  int count = 0;
1615  int rc;  int rc;
1616    
# Line 1386  matching closing parens. That is why we Line 1621  matching closing parens. That is why we
1621    
1622  for (;;)  for (;;)
1623    {    {
1624    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1625    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == 0) break;
1626    }    }
1627    
# Line 1413  Arguments: Line 1648  Arguments:
1648  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1649  */  */
1650    
1651  static const uschar*  static const pcre_uchar*
1652  first_significant_code(const uschar *code, BOOL skipassert)  first_significant_code(const pcre_uchar *code, BOOL skipassert)
1653  {  {
1654  for (;;)  for (;;)
1655    {    {
# Line 1425  for (;;) Line 1660  for (;;)
1660      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1661      if (!skipassert) return code;      if (!skipassert) return code;
1662      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1663      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1664      break;      break;
1665    
1666      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1439  for (;;) Line 1674  for (;;)
1674      case OP_RREF:      case OP_RREF:
1675      case OP_NRREF:      case OP_NRREF:
1676      case OP_DEF:      case OP_DEF:
1677      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1678      break;      break;
1679    
1680      default:      default:
# Line 1469  and doing the check at the end; a flag s Line 1704  and doing the check at the end; a flag s
1704    
1705  Arguments:  Arguments:
1706    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1707    utf8     TRUE in UTF-8 mode    utf      TRUE in UTF-8 / UTF-16 mode
1708    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1709    cd       the "compile data" structure    cd       the "compile data" structure
1710    
1711  Returns:   the fixed length,  Returns:   the fixed length,
1712               or -1 if there is no fixed length,               or -1 if there is no fixed length,
1713               or -2 if \C was encountered               or -2 if \C was encountered (in UTF-8 mode only)
1714               or -3 if an OP_RECURSE item was encountered and atend is FALSE               or -3 if an OP_RECURSE item was encountered and atend is FALSE
1715                 or -4 if an unknown opcode was encountered (internal error)
1716  */  */
1717    
1718  static int  static int
1719  find_fixedlength(uschar *code, BOOL utf8, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1720  {  {
1721  int length = -1;  int length = -1;
1722    
1723  register int branchlength = 0;  register int branchlength = 0;
1724  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1725    
1726  /* Scan along the opcodes for this branch. If we get to the end of the  /* Scan along the opcodes for this branch. If we get to the end of the
1727  branch, check the length against that of the other branches. */  branch, check the length against that of the other branches. */
# Line 1493  branch, check the length against that of Line 1729  branch, check the length against that of
1729  for (;;)  for (;;)
1730    {    {
1731    int d;    int d;
1732    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1733    register int op = *cc;    register int op = *cc;
1734    
1735    switch (op)    switch (op)
1736      {      {
1737      /* We only need to continue for OP_CBRA (normal capturing bracket) and      /* We only need to continue for OP_CBRA (normal capturing bracket) and
1738      OP_BRA (normal non-capturing bracket) because the other variants of these      OP_BRA (normal non-capturing bracket) because the other variants of these
1739      opcodes are all concerned with unlimited repeated groups, which of course      opcodes are all concerned with unlimited repeated groups, which of course
1740      are not of fixed length. They will cause a -1 response from the default      are not of fixed length. */
     case of this switch. */  
1741    
1742      case OP_CBRA:      case OP_CBRA:
1743      case OP_BRA:      case OP_BRA:
1744      case OP_ONCE:      case OP_ONCE:
1745        case OP_ONCE_NC:
1746      case OP_COND:      case OP_COND:
1747      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd);      d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd);
1748      if (d < 0) return d;      if (d < 0) return d;
1749      branchlength += d;      branchlength += d;
1750      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1751      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
1752      break;      break;
1753    
1754      /* Reached end of a branch; if it's a ket it is the end of a nested      /* Reached end of a branch; if it's a ket it is the end of a nested call.
1755      call. If it's ALT it is an alternation in a nested call. If it is      If it's ALT it is an alternation in a nested call. An ACCEPT is effectively
1756      END it's the end of the outer call. All can be handled by the same code.      an ALT. If it is END it's the end of the outer call. All can be handled by
1757      Note that we must not include the OP_KETRxxx opcodes here, because they      the same code. Note that we must not include the OP_KETRxxx opcodes here,
1758      all imply an unlimited repeat. */      because they all imply an unlimited repeat. */
1759    
1760      case OP_ALT:      case OP_ALT:
1761      case OP_KET:      case OP_KET:
1762      case OP_END:      case OP_END:
1763        case OP_ACCEPT:
1764        case OP_ASSERT_ACCEPT:
1765      if (length < 0) length = branchlength;      if (length < 0) length = branchlength;
1766        else if (length != branchlength) return -1;        else if (length != branchlength) return -1;
1767      if (*cc != OP_ALT) return length;      if (*cc != OP_ALT) return length;
# Line 1536  for (;;) Line 1775  for (;;)
1775    
1776      case OP_RECURSE:      case OP_RECURSE:
1777      if (!atend) return -3;      if (!atend) return -3;
1778      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1779      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1780      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1781      d = find_fixedlength(cs + 2, utf8, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1782      if (d < 0) return d;      if (d < 0) return d;
1783      branchlength += d;      branchlength += d;
1784      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1552  for (;;) Line 1791  for (;;)
1791      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1792      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1793      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1794      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1795        break;
1796    
1797      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1798    
1799      case OP_REVERSE:      case OP_MARK:
1800      case OP_CREF:      case OP_PRUNE_ARG:
1801      case OP_NCREF:      case OP_SKIP_ARG:
1802      case OP_RREF:      case OP_THEN_ARG:
1803      case OP_NRREF:      cc += cc[1] + PRIV(OP_lengths)[*cc];
1804      case OP_DEF:      break;
1805    
1806      case OP_CALLOUT:      case OP_CALLOUT:
     case OP_SOD:  
     case OP_SOM:  
     case OP_SET_SOM:  
     case OP_EOD:  
     case OP_EODN:  
1807      case OP_CIRC:      case OP_CIRC:
1808      case OP_CIRCM:      case OP_CIRCM:
1809        case OP_CLOSE:
1810        case OP_COMMIT:
1811        case OP_CREF:
1812        case OP_DEF:
1813      case OP_DOLL:      case OP_DOLL:
1814      case OP_DOLLM:      case OP_DOLLM:
1815        case OP_EOD:
1816        case OP_EODN:
1817        case OP_FAIL:
1818        case OP_NCREF:
1819        case OP_NRREF:
1820      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1821        case OP_PRUNE:
1822        case OP_REVERSE:
1823        case OP_RREF:
1824        case OP_SET_SOM:
1825        case OP_SKIP:
1826        case OP_SOD:
1827        case OP_SOM:
1828        case OP_THEN:
1829      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1830      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1831      break;      break;
1832    
1833      /* Handle literal characters */      /* Handle literal characters */
# Line 1585  for (;;) Line 1838  for (;;)
1838      case OP_NOTI:      case OP_NOTI:
1839      branchlength++;      branchlength++;
1840      cc += 2;      cc += 2;
1841  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1842      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1843  #endif  #endif
1844      break;      break;
1845    
# Line 1594  for (;;) Line 1847  for (;;)
1847      need to skip over a multibyte character in UTF8 mode.  */      need to skip over a multibyte character in UTF8 mode.  */
1848    
1849      case OP_EXACT:      case OP_EXACT:
1850        case OP_EXACTI:
1851        case OP_NOTEXACT:
1852        case OP_NOTEXACTI:
1853      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1854      cc += 4;      cc += 2 + IMM2_SIZE;
1855  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1856      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1857  #endif  #endif
1858      break;      break;
1859    
1860      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1861      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1862      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1863      cc += 4;        cc += 2;
1864        cc += 1 + IMM2_SIZE + 1;
1865      break;      break;
1866    
1867      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1614  for (;;) Line 1871  for (;;)
1871      cc += 2;      cc += 2;
1872      /* Fall through */      /* Fall through */
1873    
1874        case OP_HSPACE:
1875        case OP_VSPACE:
1876        case OP_NOT_HSPACE:
1877        case OP_NOT_VSPACE:
1878      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
1879      case OP_DIGIT:      case OP_DIGIT:
1880      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
# Line 1626  for (;;) Line 1887  for (;;)
1887      cc++;      cc++;
1888      break;      break;
1889    
1890      /* The single-byte matcher isn't allowed */      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;
1891        otherwise \C is coded as OP_ALLANY. */
1892    
1893      case OP_ANYBYTE:      case OP_ANYBYTE:
1894      return -2;      return -2;
1895    
1896      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1897    
1898  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || defined COMPILE_PCRE16
1899      case OP_XCLASS:      case OP_XCLASS:
1900      cc += GET(cc, 1) - 33;      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
1901      /* Fall through */      /* Fall through */
1902  #endif  #endif
1903    
1904      case OP_CLASS:      case OP_CLASS:
1905      case OP_NCLASS:      case OP_NCLASS:
1906      cc += 33;      cc += PRIV(OP_lengths)[OP_CLASS];
1907    
1908      switch (*cc)      switch (*cc)
1909        {        {
1910          case OP_CRPLUS:
1911          case OP_CRMINPLUS:
1912        case OP_CRSTAR:        case OP_CRSTAR:
1913        case OP_CRMINSTAR:        case OP_CRMINSTAR:
1914        case OP_CRQUERY:        case OP_CRQUERY:
# Line 1653  for (;;) Line 1917  for (;;)
1917    
1918        case OP_CRRANGE:        case OP_CRRANGE:
1919        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1920        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1921        branchlength += GET2(cc,1);        branchlength += GET2(cc,1);
1922        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1923        break;        break;
1924    
1925        default:        default:
# Line 1665  for (;;) Line 1929  for (;;)
1929    
1930      /* Anything else is variable length */      /* Anything else is variable length */
1931    
1932      default:      case OP_ANYNL:
1933        case OP_BRAMINZERO:
1934        case OP_BRAPOS:
1935        case OP_BRAPOSZERO:
1936        case OP_BRAZERO:
1937        case OP_CBRAPOS:
1938        case OP_EXTUNI:
1939        case OP_KETRMAX:
1940        case OP_KETRMIN:
1941        case OP_KETRPOS:
1942        case OP_MINPLUS:
1943        case OP_MINPLUSI:
1944        case OP_MINQUERY:
1945        case OP_MINQUERYI:
1946        case OP_MINSTAR:
1947        case OP_MINSTARI:
1948        case OP_MINUPTO:
1949        case OP_MINUPTOI:
1950        case OP_NOTMINPLUS:
1951        case OP_NOTMINPLUSI:
1952        case OP_NOTMINQUERY:
1953        case OP_NOTMINQUERYI:
1954        case OP_NOTMINSTAR:
1955        case OP_NOTMINSTARI:
1956        case OP_NOTMINUPTO:
1957        case OP_NOTMINUPTOI:
1958        case OP_NOTPLUS:
1959        case OP_NOTPLUSI:
1960        case OP_NOTPOSPLUS:
1961        case OP_NOTPOSPLUSI:
1962        case OP_NOTPOSQUERY:
1963        case OP_NOTPOSQUERYI:
1964        case OP_NOTPOSSTAR:
1965        case OP_NOTPOSSTARI:
1966        case OP_NOTPOSUPTO:
1967        case OP_NOTPOSUPTOI:
1968        case OP_NOTQUERY:
1969        case OP_NOTQUERYI:
1970        case OP_NOTSTAR:
1971        case OP_NOTSTARI:
1972        case OP_NOTUPTO:
1973        case OP_NOTUPTOI:
1974        case OP_PLUS:
1975        case OP_PLUSI:
1976        case OP_POSPLUS:
1977        case OP_POSPLUSI:
1978        case OP_POSQUERY:
1979        case OP_POSQUERYI:
1980        case OP_POSSTAR:
1981        case OP_POSSTARI:
1982        case OP_POSUPTO:
1983        case OP_POSUPTOI:
1984        case OP_QUERY:
1985        case OP_QUERYI:
1986        case OP_REF:
1987        case OP_REFI:
1988        case OP_SBRA:
1989        case OP_SBRAPOS:
1990        case OP_SCBRA:
1991        case OP_SCBRAPOS:
1992        case OP_SCOND:
1993        case OP_SKIPZERO:
1994        case OP_STAR:
1995        case OP_STARI:
1996        case OP_TYPEMINPLUS:
1997        case OP_TYPEMINQUERY:
1998        case OP_TYPEMINSTAR:
1999        case OP_TYPEMINUPTO:
2000        case OP_TYPEPLUS:
2001        case OP_TYPEPOSPLUS:
2002        case OP_TYPEPOSQUERY:
2003        case OP_TYPEPOSSTAR:
2004        case OP_TYPEPOSUPTO:
2005        case OP_TYPEQUERY:
2006        case OP_TYPESTAR:
2007        case OP_TYPEUPTO:
2008        case OP_UPTO:
2009        case OP_UPTOI:
2010      return -1;      return -1;
2011    
2012        /* Catch unrecognized opcodes so that when new ones are added they
2013        are not forgotten, as has happened in the past. */
2014    
2015        default:
2016        return -4;
2017      }      }
2018    }    }
2019  /* Control never gets here */  /* Control never gets here */
# Line 1687  length. Line 2034  length.
2034    
2035  Arguments:  Arguments:
2036    code        points to start of expression    code        points to start of expression
2037    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 mode
2038    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2039    
2040  Returns:      pointer to the opcode for the bracket, or NULL if not found  Returns:      pointer to the opcode for the bracket, or NULL if not found
2041  */  */
2042    
2043  const uschar *  const pcre_uchar *
2044  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2045  {  {
2046  for (;;)  for (;;)
2047    {    {
# Line 1712  for (;;) Line 2059  for (;;)
2059    
2060    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2061      {      {
2062      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2063      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2064      }      }
2065    
2066    /* Handle capturing bracket */    /* Handle capturing bracket */
# Line 1722  for (;;) Line 2069  for (;;)
2069             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2070      {      {
2071      int n = GET2(code, 1+LINK_SIZE);      int n = GET2(code, 1+LINK_SIZE);
2072      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2073      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2074      }      }
2075    
2076    /* Otherwise, we can get the item's length from the table, except that for    /* Otherwise, we can get the item's length from the table, except that for
# Line 1751  for (;;) Line 2098  for (;;)
2098        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2099        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2100        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2101        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2102            code += 2;
2103        break;        break;
2104    
2105        case OP_MARK:        case OP_MARK:
# Line 1761  for (;;) Line 2109  for (;;)
2109        break;        break;
2110    
2111        case OP_THEN_ARG:        case OP_THEN_ARG:
2112        code += code[1+LINK_SIZE];        code += code[1];
2113        break;        break;
2114        }        }
2115    
2116      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2117    
2118      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2119    
2120    /* In UTF-8 mode, opcodes that are followed by a character may be followed by    /* In UTF-8 mode, opcodes that are followed by a character may be followed by
2121    a multi-byte character. The length in the table is a minimum, so we have to    a multi-byte character. The length in the table is a minimum, so we have to
2122    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2123    
2124  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2125      if (utf8) switch(c)      if (utf) switch(c)
2126        {        {
2127        case OP_CHAR:        case OP_CHAR:
2128        case OP_CHARI:        case OP_CHARI:
# Line 1804  for (;;) Line 2152  for (;;)
2152        case OP_MINQUERYI:        case OP_MINQUERYI:
2153        case OP_POSQUERY:        case OP_POSQUERY:
2154        case OP_POSQUERYI:        case OP_POSQUERYI:
2155        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2156        break;        break;
2157        }        }
2158  #else  #else
2159      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2160  #endif  #endif
2161      }      }
2162    }    }
# Line 1825  instance of OP_RECURSE. Line 2173  instance of OP_RECURSE.
2173    
2174  Arguments:  Arguments:
2175    code        points to start of expression    code        points to start of expression
2176    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 mode
2177    
2178  Returns:      pointer to the opcode for OP_RECURSE, or NULL if not found  Returns:      pointer to the opcode for OP_RECURSE, or NULL if not found
2179  */  */
2180    
2181  static const uschar *  static const pcre_uchar *
2182  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2183  {  {
2184  for (;;)  for (;;)
2185    {    {
# Line 1870  for (;;) Line 2218  for (;;)
2218        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2219        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2220        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2221        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2222            code += 2;
2223        break;        break;
2224    
2225        case OP_MARK:        case OP_MARK:
# Line 1880  for (;;) Line 2229  for (;;)
2229        break;        break;
2230    
2231        case OP_THEN_ARG:        case OP_THEN_ARG:
2232        code += code[1+LINK_SIZE];        code += code[1];
2233        break;        break;
2234        }        }
2235    
2236      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2237    
2238      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2239    
2240      /* In UTF-8 mode, opcodes that are followed by a character may be followed      /* In UTF-8 mode, opcodes that are followed by a character may be followed
2241      by a multi-byte character. The length in the table is a minimum, so we have      by a multi-byte character. The length in the table is a minimum, so we have
2242      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2243    
2244  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2245      if (utf8) switch(c)      if (utf) switch(c)
2246        {        {
2247        case OP_CHAR:        case OP_CHAR:
2248        case OP_CHARI:        case OP_CHARI:
2249          case OP_NOT:
2250          case OP_NOTI:
2251        case OP_EXACT:        case OP_EXACT:
2252        case OP_EXACTI:        case OP_EXACTI:
2253          case OP_NOTEXACT:
2254          case OP_NOTEXACTI:
2255        case OP_UPTO:        case OP_UPTO:
2256        case OP_UPTOI:        case OP_UPTOI:
2257          case OP_NOTUPTO:
2258          case OP_NOTUPTOI:
2259        case OP_MINUPTO:        case OP_MINUPTO:
2260        case OP_MINUPTOI:        case OP_MINUPTOI:
2261          case OP_NOTMINUPTO:
2262          case OP_NOTMINUPTOI:
2263        case OP_POSUPTO:        case OP_POSUPTO:
2264        case OP_POSUPTOI:        case OP_POSUPTOI:
2265          case OP_NOTPOSUPTO:
2266          case OP_NOTPOSUPTOI:
2267        case OP_STAR:        case OP_STAR:
2268        case OP_STARI:        case OP_STARI:
2269          case OP_NOTSTAR:
2270          case OP_NOTSTARI:
2271        case OP_MINSTAR:        case OP_MINSTAR:
2272        case OP_MINSTARI:        case OP_MINSTARI:
2273          case OP_NOTMINSTAR:
2274          case OP_NOTMINSTARI:
2275        case OP_POSSTAR:        case OP_POSSTAR:
2276        case OP_POSSTARI:        case OP_POSSTARI:
2277          case OP_NOTPOSSTAR:
2278          case OP_NOTPOSSTARI:
2279        case OP_PLUS:        case OP_PLUS:
2280        case OP_PLUSI:        case OP_PLUSI:
2281          case OP_NOTPLUS:
2282          case OP_NOTPLUSI:
2283        case OP_MINPLUS:        case OP_MINPLUS:
2284        case OP_MINPLUSI:        case OP_MINPLUSI:
2285          case OP_NOTMINPLUS:
2286          case OP_NOTMINPLUSI:
2287        case OP_POSPLUS:        case OP_POSPLUS:
2288        case OP_POSPLUSI:        case OP_POSPLUSI:
2289          case OP_NOTPOSPLUS:
2290          case OP_NOTPOSPLUSI:
2291        case OP_QUERY:        case OP_QUERY:
2292        case OP_QUERYI:        case OP_QUERYI:
2293          case OP_NOTQUERY:
2294          case OP_NOTQUERYI:
2295        case OP_MINQUERY:        case OP_MINQUERY:
2296        case OP_MINQUERYI:        case OP_MINQUERYI:
2297          case OP_NOTMINQUERY:
2298          case OP_NOTMINQUERYI:
2299        case OP_POSQUERY:        case OP_POSQUERY:
2300        case OP_POSQUERYI:        case OP_POSQUERYI:
2301        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_NOTPOSQUERY:
2302          case OP_NOTPOSQUERYI:
2303          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2304        break;        break;
2305        }        }
2306  #else  #else
2307      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2308  #endif  #endif
2309      }      }
2310    }    }
# Line 1950  bracket whose current branch will alread Line 2327  bracket whose current branch will alread
2327  Arguments:  Arguments:
2328    code        points to start of search    code        points to start of search
2329    endcode     points to where to stop    endcode     points to where to stop
2330    utf8        TRUE if in UTF8 mode    utf         TRUE if in UTF-8 / UTF-16 mode
2331    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2332    
2333  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2334  */  */
2335    
2336  static BOOL  static BOOL
2337  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2338    compile_data *cd)    BOOL utf, compile_data *cd)
2339  {  {
2340  register int c;  register int c;
2341  for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2342       code < endcode;       code < endcode;
2343       code = first_significant_code(code + _pcre_OP_lengths[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2344    {    {
2345    const uschar *ccode;    const pcre_uchar *ccode;
2346    
2347    c = *code;    c = *code;
2348    
# Line 1982  for (code = first_significant_code(code Line 2359  for (code = first_significant_code(code
2359    /* For a recursion/subroutine call, if its end has been reached, which    /* For a recursion/subroutine call, if its end has been reached, which
2360    implies a backward reference subroutine call, we can scan it. If it's a    implies a backward reference subroutine call, we can scan it. If it's a
2361    forward reference subroutine call, we can't. To detect forward reference    forward reference subroutine call, we can't. To detect forward reference
2362    we have to scan up the list that is kept in the workspace. This function is    we have to scan up the list that is kept in the workspace. This function is
2363    called only when doing the real compile, not during the pre-compile that    called only when doing the real compile, not during the pre-compile that
2364    measures the size of the compiled pattern. */    measures the size of the compiled pattern. */
2365    
2366    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2367      {      {
2368      const uschar *scode;      const pcre_uchar *scode;
2369      BOOL empty_branch;      BOOL empty_branch;
2370    
2371      /* Test for forward reference */      /* Test for forward reference */
2372    
2373      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2374        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;
2375    
2376      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2377    
2378      empty_branch = FALSE;      empty_branch = FALSE;
2379      scode = cd->start_code + GET(code, 1);      scode = cd->start_code + GET(code, 1);
2380      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2381    
2382      /* Completed backwards reference */      /* Completed backwards reference */
2383    
2384      do      do
2385        {        {
2386        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd))
2387          {          {
2388          empty_branch = TRUE;          empty_branch = TRUE;
2389          break;          break;
# Line 2014  for (code = first_significant_code(code Line 2391  for (code = first_significant_code(code
2391        scode += GET(scode, 1);        scode += GET(scode, 1);
2392        }        }
2393      while (*scode == OP_ALT);      while (*scode == OP_ALT);
2394    
2395      if (!empty_branch) return FALSE;  /* All branches are non-empty */      if (!empty_branch) return FALSE;  /* All branches are non-empty */
2396      continue;      continue;
2397      }      }
# Line 2024  for (code = first_significant_code(code Line 2401  for (code = first_significant_code(code
2401    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2402        c == OP_BRAPOSZERO)        c == OP_BRAPOSZERO)
2403      {      {
2404      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2405      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
2406      c = *code;      c = *code;
2407      continue;      continue;
# Line 2045  for (code = first_significant_code(code Line 2422  for (code = first_significant_code(code
2422    
2423    if (c == OP_BRA  || c == OP_BRAPOS ||    if (c == OP_BRA  || c == OP_BRAPOS ||
2424        c == OP_CBRA || c == OP_CBRAPOS ||        c == OP_CBRA || c == OP_CBRAPOS ||
2425        c == OP_ONCE || c == OP_COND)        c == OP_ONCE || c == OP_ONCE_NC ||
2426          c == OP_COND)
2427      {      {
2428      BOOL empty_branch;      BOOL empty_branch;
2429      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 2061  for (code = first_significant_code(code Line 2439  for (code = first_significant_code(code
2439        empty_branch = FALSE;        empty_branch = FALSE;
2440        do        do
2441          {          {
2442          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
2443            empty_branch = TRUE;            empty_branch = TRUE;
2444          code += GET(code, 1);          code += GET(code, 1);
2445          }          }
# Line 2079  for (code = first_significant_code(code Line 2457  for (code = first_significant_code(code
2457      {      {
2458      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2459      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2460      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2461      actual length is stored in the compiled code, so we must update "code"      actual length is stored in the compiled code, so we must update "code"
2462      here. */      here. */
2463    
2464  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2465      case OP_XCLASS:      case OP_XCLASS:
2466      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2467      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2091  for (code = first_significant_code(code Line 2469  for (code = first_significant_code(code
2469    
2470      case OP_CLASS:      case OP_CLASS:
2471      case OP_NCLASS:      case OP_NCLASS:
2472      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2473    
2474  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2475      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2476  #endif  #endif
2477    
# Line 2166  for (code = first_significant_code(code Line 2544  for (code = first_significant_code(code
2544      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2545      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2546      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2547      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2548          code += 2;
2549      break;      break;
2550    
2551      /* End of branch */      /* End of branch */
# Line 2181  for (code = first_significant_code(code Line 2560  for (code = first_significant_code(code
2560      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2561      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2562    
2563  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2564      case OP_STAR:      case OP_STAR:
2565      case OP_STARI:      case OP_STARI:
2566      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2194  for (code = first_significant_code(code Line 2573  for (code = first_significant_code(code
2573      case OP_MINQUERYI:      case OP_MINQUERYI:
2574      case OP_POSQUERY:      case OP_POSQUERY:
2575      case OP_POSQUERYI:      case OP_POSQUERYI:
2576      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2577      break;      break;
2578    
2579      case OP_UPTO:      case OP_UPTO:
# Line 2203  for (code = first_significant_code(code Line 2582  for (code = first_significant_code(code
2582      case OP_MINUPTOI:      case OP_MINUPTOI:
2583      case OP_POSUPTO:      case OP_POSUPTO:
2584      case OP_POSUPTOI:      case OP_POSUPTOI:
2585      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);
2586      break;      break;
2587  #endif  #endif
2588    
# Line 2217  for (code = first_significant_code(code Line 2596  for (code = first_significant_code(code
2596      break;      break;
2597    
2598      case OP_THEN_ARG:      case OP_THEN_ARG:
2599      code += code[1+LINK_SIZE];      code += code[1];
2600      break;      break;
2601    
2602      /* None of the remaining opcodes are required to match a character. */      /* None of the remaining opcodes are required to match a character. */
# Line 2240  return TRUE; Line 2619  return TRUE;
2619  the current branch of the current pattern to see if it could match the empty  the current branch of the current pattern to see if it could match the empty
2620  string. If it could, we must look outwards for branches at other levels,  string. If it could, we must look outwards for branches at other levels,
2621  stopping when we pass beyond the bracket which is the subject of the recursion.  stopping when we pass beyond the bracket which is the subject of the recursion.
2622  This function is called only during the real compile, not during the  This function is called only during the real compile, not during the
2623  pre-compile.  pre-compile.
2624    
2625  Arguments:  Arguments:
2626    code        points to start of the recursion    code        points to start of the recursion
2627    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2628    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2629    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 mode
2630    cd          pointers to tables etc    cd          pointers to tables etc
2631    
2632  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2633  */  */
2634    
2635  static BOOL  static BOOL
2636  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2637    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2638  {  {
2639  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2640    {    {
2641    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
2642      return FALSE;      return FALSE;
2643    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2644    }    }
# Line 2295  I think. Line 2674  I think.
2674  A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.  A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.
2675  It seems that the appearance of a nested POSIX class supersedes an apparent  It seems that the appearance of a nested POSIX class supersedes an apparent
2676  external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or  external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or
2677  a digit. Also, unescaped square brackets may also appear as part of class  a digit.
2678  names. For example, [:a[:abc]b:] gives unknown class "[:abc]b:]"in Perl.  
2679    In Perl, unescaped square brackets may also appear as part of class names. For
2680    example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for
2681    [:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not
2682    seem right at all. PCRE does not allow closing square brackets in POSIX class
2683    names.
2684    
2685  Arguments:  Arguments:
2686    ptr      pointer to the initial [    ptr      pointer to the initial [
# Line 2306  Returns:   TRUE or FALSE Line 2690  Returns:   TRUE or FALSE
2690  */  */
2691    
2692  static BOOL  static BOOL
2693  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2694  {  {
2695  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
2696  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2697  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != 0; ptr++)
2698    {    {
2699    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2700      ptr++;      ptr++;
2701      else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
2702    else    else
2703      {      {
2704      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
# Line 2325  for (++ptr; *ptr != 0; ptr++) Line 2710  for (++ptr; *ptr != 0; ptr++)
2710           (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||           (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2711            ptr[1] == CHAR_EQUALS_SIGN) &&            ptr[1] == CHAR_EQUALS_SIGN) &&
2712          check_posix_syntax(ptr, endptr))          check_posix_syntax(ptr, endptr))
2713        return FALSE;        return FALSE;
2714      }      }
2715    }    }
2716  return FALSE;  return FALSE;
# Line 2349  Returns:     a value representing the na Line 2734  Returns:     a value representing the na
2734  */  */
2735    
2736  static int  static int
2737  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2738  {  {
2739  const char *pn = posix_names;  const char *pn = posix_names;
2740  register int yield = 0;  register int yield = 0;
2741  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2742    {    {
2743    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2744      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;
2745    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2746    yield++;    yield++;
2747    }    }
# Line 2388  value in the reference (which is a group Line 2773  value in the reference (which is a group
2773  Arguments:  Arguments:
2774    group      points to the start of the group    group      points to the start of the group
2775    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2776    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 mode
2777    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2778    save_hwm   the hwm forward reference pointer at the start of the group    save_hwm   the hwm forward reference pointer at the start of the group
2779    
# Line 2396  Returns:     nothing Line 2781  Returns:     nothing
2781  */  */
2782    
2783  static void  static void
2784  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2785    uschar *save_hwm)    pcre_uchar *save_hwm)
2786  {  {
2787  uschar *ptr = group;  pcre_uchar *ptr = group;
2788    
2789  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2790    {    {
2791    int offset;    int offset;
2792    uschar *hc;    pcre_uchar *hc;
2793    
2794    /* See if this recursion is on the forward reference list. If so, adjust the    /* See if this recursion is on the forward reference list. If so, adjust the
2795    reference. */    reference. */
# Line 2449  Arguments: Line 2834  Arguments:
2834  Returns:         new code pointer  Returns:         new code pointer
2835  */  */
2836    
2837  static uschar *  static pcre_uchar *
2838  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2839  {  {
2840  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2841  *code++ = 255;  *code++ = 255;
2842  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2843  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2844  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2845  }  }
2846    
2847    
# Line 2478  Returns:             nothing Line 2863  Returns:             nothing
2863  */  */
2864    
2865  static void  static void
2866  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2867  {  {
2868  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2869  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2492  PUT(previous_callout, 2 + LINK_SIZE, len Line 2877  PUT(previous_callout, 2 + LINK_SIZE, len
2877  *************************************************/  *************************************************/
2878    
2879  /* This function is passed the start and end of a class range, in UTF-8 mode  /* This function is passed the start and end of a class range, in UTF-8 mode
2880  with UCP support. It searches up the characters, looking for internal ranges of  with UCP support. It searches up the characters, looking for ranges of
2881  characters in the "other" case. Each call returns the next one, updating the  characters in the "other" case. Each call returns the next one, updating the
2882  start address.  start address. A character with multiple other cases is returned on its own
2883    with a special return value.
2884    
2885  Arguments:  Arguments:
2886    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2502  Arguments: Line 2888  Arguments:
2888    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2889    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2890    
2891  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2892                   0 when a range is returned
2893                  >0 the CASESET offset for char with multiple other cases
2894                    in this case, ocptr contains the original
2895  */  */
2896    
2897  static BOOL  static int
2898  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,
2899    unsigned int *odptr)    unsigned int *odptr)
2900  {  {
2901  unsigned int c, othercase, next;  unsigned int c, othercase, next;
2902    int co;
2903    
2904    /* Find the first character that has an other case. If it has multiple other
2905    cases, return its case offset value. */
2906    
2907  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2908    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2909      if ((co = UCD_CASESET(c)) != 0)
2910        {
2911        *ocptr = c++;   /* Character that has the set */
2912        *cptr = c;      /* Rest of input range */
2913        return co;
2914        }
2915      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2916      }
2917    
2918  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2919    
2920  *ocptr = othercase;  *ocptr = othercase;
2921  next = othercase + 1;  next = othercase + 1;
# Line 2525  for (++c; c <= d; c++) Line 2926  for (++c; c <= d; c++)
2926    next++;    next++;
2927    }    }
2928    
2929  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2930  *cptr = c;  *cptr = c;             /* Rest of input range */
2931    return 0;
 return TRUE;  
2932  }  }
2933    
2934    
# Line 2552  Returns:       TRUE if auto-possessifyin Line 2952  Returns:       TRUE if auto-possessifyin
2952  static BOOL  static BOOL
2953  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(int c, int ptype, int pdata, BOOL negated)
2954  {  {
2955    #ifdef SUPPORT_UCP
2956    const pcre_uint32 *p;
2957    #endif
2958    
2959  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2960    
2961  switch(ptype)  switch(ptype)
2962    {    {
2963    case PT_LAMP:    case PT_LAMP:
# Line 2561  switch(ptype) Line 2966  switch(ptype)
2966            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
2967    
2968    case PT_GC:    case PT_GC:
2969    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2970    
2971    case PT_PC:    case PT_PC:
2972    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2572  switch(ptype) Line 2977  switch(ptype)
2977    /* These are specials */    /* These are specials */
2978    
2979    case PT_ALNUM:    case PT_ALNUM:
2980    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2981            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
2982    
2983    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
2984    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2985            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2986            == negated;            == negated;
2987    
2988    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
2989    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2990            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2991            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
2992            == negated;            == negated;
2993    
2994    case PT_WORD:    case PT_WORD:
2995    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2996            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2997            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
2998    
2999    #ifdef SUPPORT_UCP
3000      case PT_CLIST:
3001      p = PRIV(ucd_caseless_sets) + prop->caseset;
3002      for (;;)
3003        {
3004        if ((unsigned int)c < *p) return !negated;
3005        if ((unsigned int)c == *p++) return negated;
3006        }
3007      break;  /* Control never reaches here */
3008    #endif
3009    }    }
3010    
3011  return FALSE;  return FALSE;
3012  }  }
3013  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2607  sense to automatically possessify the re Line 3024  sense to automatically possessify the re
3024    
3025  Arguments:  Arguments:
3026    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3027    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 mode
3028    ptr           next character in pattern    ptr           next character in pattern
3029    options       options bits    options       options bits
3030    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2616  Returns:        TRUE if possessifying is Line 3033  Returns:        TRUE if possessifying is
3033  */  */
3034    
3035  static BOOL  static BOOL
3036  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3037    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3038  {  {
3039  int c, next;  pcre_int32 c = NOTACHAR;
3040    pcre_int32 next;
3041  int op_code = *previous++;  int op_code = *previous++;
3042    
3043  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2628  if ((options & PCRE_EXTENDED) != 0) Line 3046  if ((options & PCRE_EXTENDED) != 0)
3046    {    {
3047    for (;;)    for (;;)
3048      {      {
3049      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3050      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3051        {        {
3052        ptr++;        ptr++;
# Line 2636  if ((options & PCRE_EXTENDED) != 0) Line 3054  if ((options & PCRE_EXTENDED) != 0)
3054          {          {
3055          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3056          ptr++;          ptr++;
3057  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3058          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3059  #endif  #endif
3060          }          }
3061        }        }
# Line 2655  if (*ptr == CHAR_BACKSLASH) Line 3073  if (*ptr == CHAR_BACKSLASH)
3073    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3074    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3075    }    }
3076    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
3077    {    {
3078  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3079    if (utf8) { GETCHARINC(next, ptr); } else    if (utf) { GETCHARINC(next, ptr); } else
3080  #endif  #endif
3081    next = *ptr++;    next = *ptr++;
3082    }    }
   
3083  else return FALSE;  else return FALSE;
3084    
3085  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2672  if ((options & PCRE_EXTENDED) != 0) Line 3088  if ((options & PCRE_EXTENDED) != 0)
3088    {    {
3089    for (;;)    for (;;)
3090      {      {
3091      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3092      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3093        {        {
3094        ptr++;        ptr++;
# Line 2680  if ((options & PCRE_EXTENDED) != 0) Line 3096  if ((options & PCRE_EXTENDED) != 0)
3096          {          {
3097          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3098          ptr++;          ptr++;
3099  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3100          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3101  #endif  #endif
3102          }          }
3103        }        }
# Line 2692  if ((options & PCRE_EXTENDED) != 0) Line 3108  if ((options & PCRE_EXTENDED) != 0)
3108  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3109    
3110  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3111    strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)    STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
3112      return FALSE;      return FALSE;
3113    
3114  /* Now compare the next item with the previous opcode. First, handle cases when  /* If the previous item is a character, get its value. */
 the next item is a character. */  
3115    
3116  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3117        op_code == OP_NOT || op_code == OP_NOTI)
3118    {    {
3119    case OP_CHAR:  #ifdef SUPPORT_UTF
 #ifdef SUPPORT_UTF8  
3120    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3121  #else  #else
3122    c = *previous;    c = *previous;
3123  #endif  #endif
3124    return c != next;    }
3125    
3126    /* For CHARI (caseless character) we must check the other case. If we have  /* Now compare the next item with the previous opcode. First, handle cases when
3127    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3128    
3129    case OP_CHARI:  if (next >= 0)
3130  #ifdef SUPPORT_UTF8    {
3131    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3132  #else    case, which maps to the special PT_CLIST property. Check this first. */
3133    c = *previous;  
3134    #ifdef SUPPORT_UCP
3135      if (utf && (unsigned int)c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3136        {
3137        int ocs = UCD_CASESET(next);
3138        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3139        }
3140  #endif  #endif
3141    if (c == next) return FALSE;  
3142  #ifdef SUPPORT_UTF8    switch(op_code)
   if (utf8)  
3143      {      {
3144      unsigned int othercase;      case OP_CHAR:
3145      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3146    
3147        /* For CHARI (caseless character) we must check the other case. If we have
3148        Unicode property support, we can use it to test the other case of
3149        high-valued characters. We know that next can have only one other case,
3150        because multi-other-case characters are dealt with above. */
3151    
3152        case OP_CHARI:
3153        if (c == next) return FALSE;
3154    #ifdef SUPPORT_UTF
3155        if (utf)
3156          {
3157          unsigned int othercase;
3158          if (next < 128) othercase = cd->fcc[next]; else
3159  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3160      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE((unsigned int)next);
3161  #else  #else
3162      othercase = NOTACHAR;        othercase = NOTACHAR;
3163  #endif  #endif
3164      return (unsigned int)c != othercase;        return (unsigned int)c != othercase;
3165      }        }
3166    else      else
3167  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3168    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */      return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3169    
3170    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These      case OP_NOT:
3171    opcodes are not used for multi-byte characters, because they are coded using      return c == next;
3172    an XCLASS instead. */  
3173        case OP_NOTI:
3174    case OP_NOT:      if (c == next) return TRUE;
3175    return (c = *previous) == next;  #ifdef SUPPORT_UTF
3176        if (utf)
3177    case OP_NOTI:        {
3178    if ((c = *previous) == next) return TRUE;        unsigned int othercase;
3179  #ifdef SUPPORT_UTF8        if (next < 128) othercase = cd->fcc[next]; else
   if (utf8)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3180  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3181      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE((unsigned int)next);
3182  #else  #else
3183      othercase = NOTACHAR;        othercase = NOTACHAR;
3184  #endif  #endif
3185      return (unsigned int)c == othercase;        return (unsigned int)c == othercase;
3186      }        }
3187    else      else
3188  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3189    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */      return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3190    
3191    /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.      /* Note that OP_DIGIT etc. are generated only when PCRE_UCP is *not* set.
3192    When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */      When it is set, \d etc. are converted into OP_(NOT_)PROP codes. */
3193    
3194    case OP_DIGIT:      case OP_DIGIT:
3195    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3196    
3197    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3198    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3199    
3200    case OP_WHITESPACE:      case OP_WHITESPACE:
3201    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3202    
3203    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3204    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3205    
3206    case OP_WORDCHAR:      case OP_WORDCHAR:
3207    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3208    
3209    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3210    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3211    
3212    case OP_HSPACE:      case OP_HSPACE:
3213    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3214    switch(next)      switch(next)
3215      {        {
3216      case 0x09:        HSPACE_CASES:
3217      case 0x20:        return op_code == OP_NOT_HSPACE;
     case 0xa0:  
     case 0x1680:  
     case 0x180e:  
     case 0x2000:  
     case 0x2001:  
     case 0x2002:  
     case 0x2003:  
     case 0x2004:  
     case 0x2005:  
     case 0x2006:  
     case 0x2007:  
     case 0x2008:  
     case 0x2009:  
     case 0x200A:  
     case 0x202f:  
     case 0x205f:  
     case 0x3000:  
     return op_code == OP_NOT_HSPACE;  
     default:  
     return op_code != OP_NOT_HSPACE;  
     }  
3218    
3219    case OP_ANYNL:        default:
3220    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3221    case OP_NOT_VSPACE:        }
3222    switch(next)  
3223      {      case OP_ANYNL:
3224      case 0x0a:      case OP_VSPACE:
3225      case 0x0b:      case OP_NOT_VSPACE:
3226      case 0x0c:      switch(next)
3227      case 0x0d:        {
3228      case 0x85:        VSPACE_CASES:
3229      case 0x2028:        return op_code == OP_NOT_VSPACE;
3230      case 0x2029:  
3231      return op_code == OP_NOT_VSPACE;        default:
3232      default:        return op_code != OP_NOT_VSPACE;
3233      return op_code != OP_NOT_VSPACE;        }
     }  
3234    
3235  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3236    case OP_PROP:      case OP_PROP:
3237    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3238    
3239    case OP_NOTPROP:      case OP_NOTPROP:
3240    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3241  #endif  #endif
3242    
3243    default:      default:
3244    return FALSE;      return FALSE;
3245        }
3246    }    }
3247    
   
3248  /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP  /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP
3249  is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are  is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are
3250  generated only when PCRE_UCP is *not* set, that is, when only ASCII  generated only when PCRE_UCP is *not* set, that is, when only ASCII
# Line 2849  switch(op_code) Line 3255  switch(op_code)
3255    {    {
3256    case OP_CHAR:    case OP_CHAR:
3257    case OP_CHARI:    case OP_CHARI:
 #ifdef SUPPORT_UTF8  
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
3258    switch(-next)    switch(-next)
3259      {      {
3260      case ESC_d:      case ESC_d:
3261      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3262    
3263      case ESC_D:      case ESC_D:
3264      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3265    
3266      case ESC_s:      case ESC_s:
3267      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3268    
3269      case ESC_S:      case ESC_S:
3270      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3271    
3272      case ESC_w:      case ESC_w:
3273      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3274    
3275      case ESC_W:      case ESC_W:
3276      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3277    
3278      case ESC_h:      case ESC_h:
3279      case ESC_H:      case ESC_H:
3280      switch(c)      switch(c)
3281        {        {
3282        case 0x09:        HSPACE_CASES:
       case 0x20:  
       case 0xa0:  
       case 0x1680:  
       case 0x180e:  
       case 0x2000:  
       case 0x2001:  
       case 0x2002:  
       case 0x2003:  
       case 0x2004:  
       case 0x2005:  
       case 0x2006:  
       case 0x2007:  
       case 0x2008:  
       case 0x2009:  
       case 0x200A:  
       case 0x202f:  
       case 0x205f:  
       case 0x3000:  
3283        return -next != ESC_h;        return -next != ESC_h;
3284    
3285        default:        default:
3286        return -next == ESC_h;        return -next == ESC_h;
3287        }        }
# Line 2906  switch(op_code) Line 3290  switch(op_code)
3290      case ESC_V:      case ESC_V:
3291      switch(c)      switch(c)
3292        {        {
3293        case 0x0a:        VSPACE_CASES:
       case 0x0b:  
       case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
3294        return -next != ESC_v;        return -next != ESC_v;
3295    
3296        default:        default:
3297        return -next == ESC_v;        return -next == ESC_v;
3298        }        }
# Line 2954  switch(op_code) Line 3333  switch(op_code)
3333        to the original \d etc. At this point, ptr will point to a zero byte. */        to the original \d etc. At this point, ptr will point to a zero byte. */
3334    
3335        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3336          strncmp((char *)ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)          STRNCMP_UC_C8(ptr, STR_LEFT_CURLY_BRACKET STR_0 STR_COMMA, 3) == 0)
3337            return FALSE;            return FALSE;
3338    
3339        /* Do the property check. */        /* Do the property check. */
# Line 2981  switch(op_code) Line 3360  switch(op_code)
3360    return next == -ESC_d;    return next == -ESC_d;
3361    
3362    case OP_WHITESPACE:    case OP_WHITESPACE:
3363    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return next == -ESC_S || next == -ESC_d || next == -ESC_w;
3364    
3365    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3366    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return next == -ESC_s || next == -ESC_h || next == -ESC_v || next == -ESC_R;
3367    
3368    case OP_HSPACE:    case OP_HSPACE:
3369    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||
# Line 3018  switch(op_code) Line 3397  switch(op_code)
3397    
3398    
3399  /*************************************************  /*************************************************
3400    *        Add a character or range to a class     *
3401    *************************************************/
3402    
3403    /* This function packages up the logic of adding a character or range of
3404    characters to a class. The character values in the arguments will be within the
3405    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3406    mutually recursive with the function immediately below.
3407    
3408    Arguments:
3409      classbits     the bit map for characters < 256
3410      uchardptr     points to the pointer for extra data
3411      options       the options word
3412      cd            contains pointers to tables etc.
3413      start         start of range character
3414      end           end of range character
3415    
3416    Returns:        the number of < 256 characters added
3417                    the pointer to extra data is updated
3418    */
3419    
3420    static int
3421    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3422      compile_data *cd, unsigned int start, unsigned int end)
3423    {
3424    unsigned int c;
3425    int n8 = 0;
3426    
3427    /* If caseless matching is required, scan the range and process alternate
3428    cases. In Unicode, there are 8-bit characters that have alternate cases that
3429    are greater than 255 and vice-versa. Sometimes we can just extend the original
3430    range. */
3431    
3432    if ((options & PCRE_CASELESS) != 0)
3433      {
3434    #ifdef SUPPORT_UCP
3435      if ((options & PCRE_UTF8) != 0)
3436        {
3437        int rc;
3438        unsigned int oc, od;
3439    
3440        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3441        c = start;
3442    
3443        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3444          {
3445          /* Handle a single character that has more than one other case. */
3446    
3447          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3448            PRIV(ucd_caseless_sets) + rc, oc);
3449    
3450          /* Do nothing if the other case range is within the original range. */
3451    
3452          else if (oc >= start && od <= end) continue;
3453    
3454          /* Extend the original range if there is overlap, noting that if oc < c, we
3455          can't have od > end because a subrange is always shorter than the basic
3456          range. Otherwise, use a recursive call to add the additional range. */
3457    
3458          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3459          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3460          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3461          }
3462        }
3463      else
3464    #endif  /* SUPPORT_UCP */
3465    
3466      /* Not UTF-mode, or no UCP */
3467    
3468      for (c = start; c <= end && c < 256; c++)
3469        {
3470        SETBIT(classbits, cd->fcc[c]);
3471        n8++;
3472        }
3473      }
3474    
3475    /* Now handle the original range. Adjust the final value according to the bit
3476    length - this means that the same lists of (e.g.) horizontal spaces can be used
3477    in all cases. */
3478    
3479    #ifdef COMPILE_PCRE8
3480    #ifdef SUPPORT_UTF
3481      if ((options & PCRE_UTF8) == 0)
3482    #endif
3483      if (end > 0xff) end = 0xff;
3484    #endif
3485    
3486    #ifdef COMPILE_PCRE16
3487    #ifdef SUPPORT_UTF
3488      if ((options & PCRE_UTF16) == 0)
3489    #endif
3490      if (end > 0xffff) end = 0xffff;
3491    #endif
3492    
3493    /* If all characters are less than 256, use the bit map. Otherwise use extra
3494    data. */
3495    
3496    if (end < 0x100)
3497      {
3498      for (c = start; c <= end; c++)
3499        {
3500        n8++;
3501        SETBIT(classbits, c);
3502        }
3503      }
3504    
3505    else
3506      {
3507      pcre_uchar *uchardata = *uchardptr;
3508    
3509    #ifdef SUPPORT_UTF
3510      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3511        {
3512        if (start < end)
3513          {
3514          *uchardata++ = XCL_RANGE;
3515          uchardata += PRIV(ord2utf)(start, uchardata);
3516          uchardata += PRIV(ord2utf)(end, uchardata);
3517          }
3518        else if (start == end)
3519          {
3520          *uchardata++ = XCL_SINGLE;
3521          uchardata += PRIV(ord2utf)(start, uchardata);
3522          }
3523        }
3524      else
3525    #endif  /* SUPPORT_UTF */
3526    
3527      /* Without UTF support, character values are constrained by the bit length,
3528      and can only be > 256 for 16-bit and 32-bit libraries. */
3529    
3530    #ifdef COMPILE_PCRE8
3531        {}
3532    #else
3533      if (start < end)
3534        {
3535        *uchardata++ = XCL_RANGE;
3536        *uchardata++ = start;
3537        *uchardata++ = end;
3538        }
3539      else if (start == end)
3540        {
3541        *uchardata++ = XCL_SINGLE;
3542        *uchardata++ = start;
3543        }
3544    #endif
3545    
3546      *uchardptr = uchardata;   /* Updata extra data pointer */
3547      }
3548    
3549    return n8;    /* Number of 8-bit characters */
3550    }
3551    
3552    
3553    
3554    
3555    /*************************************************
3556    *        Add a list of characters to a class     *
3557    *************************************************/
3558    
3559    /* This function is used for adding a list of case-equivalent characters to a
3560    class, and also for adding a list of horizontal or vertical whitespace. If the
3561    list is in order (which it should be), ranges of characters are detected and
3562    handled appropriately. This function is mutually recursive with the function
3563    above.
3564    
3565    Arguments:
3566      classbits     the bit map for characters < 256
3567      uchardptr     points to the pointer for extra data
3568      options       the options word
3569      cd            contains pointers to tables etc.
3570      p             points to row of 32-bit values, terminated by NOTACHAR
3571      except        character to omit; this is used when adding lists of
3572                      case-equivalent characters to avoid including the one we
3573                      already know about
3574    
3575    Returns:        the number of < 256 characters added
3576                    the pointer to extra data is updated
3577    */
3578    
3579    static int
3580    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3581      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3582    {
3583    int n8 = 0;
3584    while (p[0] < NOTACHAR)
3585      {
3586      int n = 0;
3587      if (p[0] != except)
3588        {
3589        while(p[n+1] == p[0] + n + 1) n++;
3590        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3591        }
3592      p += n + 1;
3593      }
3594    return n8;
3595    }
3596    
3597    
3598    
3599    /*************************************************
3600    *    Add characters not in a list to a class     *
3601    *************************************************/
3602    
3603    /* This function is used for adding the complement of a list of horizontal or
3604    vertical whitespace to a class. The list must be in order.
3605    
3606    Arguments:
3607      classbits     the bit map for characters < 256
3608      uchardptr     points to the pointer for extra data
3609      options       the options word
3610      cd            contains pointers to tables etc.
3611      p             points to row of 32-bit values, terminated by NOTACHAR
3612    
3613    Returns:        the number of < 256 characters added
3614                    the pointer to extra data is updated
3615    */
3616    
3617    static int
3618    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3619      int options, compile_data *cd, const pcre_uint32 *p)
3620    {
3621    int n8 = 0;
3622    if (p[0] > 0)
3623      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3624    while (p[0] < NOTACHAR)
3625      {
3626      while (p[1] == p[0] + 1) p++;
3627      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3628        (p[1] == NOTACHAR)? 0x10ffff : p[1] - 1);
3629      p++;
3630      }
3631    return n8;
3632    }
3633    
3634    
3635    
3636    /*************************************************
3637  *           Compile one branch                   *  *           Compile one branch                   *
3638  *************************************************/  *************************************************/
3639    
# Line 3032  Arguments: Line 3648  Arguments:
3648    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3649    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3650    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3651    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)
3652    reqbyteptr     set to the last literal character required, else < 0    reqcharptr     set to the last literal character required, else < 0
3653    bcptr          points to current branch chain    bcptr          points to current branch chain
3654      cond_depth     conditional nesting depth
3655    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
3656    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
3657                   points to length accumulator during pre-compile phase                   points to length accumulator during pre-compile phase
# Line 3044  Returns:         TRUE on success Line 3661  Returns:         TRUE on success
3661  */  */
3662    
3663  static BOOL  static BOOL
3664  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3665    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,
3666      pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,
3667    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3668  {  {
3669  int repeat_type, op_type;  int repeat_type, op_type;
3670  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3671  int bravalue = 0;  int bravalue = 0;
3672  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3673  int firstbyte, reqbyte;  pcre_int32 firstchar, reqchar;
3674  int zeroreqbyte, zerofirstbyte;  pcre_int32 zeroreqchar, zerofirstchar;
3675  int req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3676  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3677  int after_manual_callout = 0;  int after_manual_callout = 0;
3678  int length_prevgroup = 0;  int length_prevgroup = 0;
3679  register int c;  register int c;
3680  register uschar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3681  uschar *last_code = code;  pcre_uchar *last_code = code;
3682  uschar *orig_code = code;  pcre_uchar *orig_code = code;
3683  uschar *tempcode;  pcre_uchar *tempcode;
3684  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3685  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3686  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3687  const uschar *tempptr;  const pcre_uchar *tempptr;
3688  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3689  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3690  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3691  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3692  uschar classbits[32];  pcre_uint8 classbits[32];
3693    
3694  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we
3695  must not do this for other options (e.g. PCRE_EXTENDED) because they may change  must not do this for other options (e.g. PCRE_EXTENDED) because they may change
3696  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3697    
3698  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3699  BOOL class_utf8;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
3700  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3701  uschar *class_utf8data;  pcre_uchar utf_chars[6];
 uschar *class_utf8data_base;  
 uschar utf8_char[6];  
3702  #else  #else
3703  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3704  uschar *utf8_char = NULL;  #endif
3705    
3706    /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3707    class_uchardata always so that it can be passed to add_to_class() always,
3708    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3709    alternative calls for the different cases. */
3710    
3711    pcre_uchar *class_uchardata;
3712    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3713    BOOL xclass;
3714    pcre_uchar *class_uchardata_base;
3715  #endif  #endif
3716    
3717  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3099  greedy_non_default = greedy_default ^ 1; Line 3725  greedy_non_default = greedy_default ^ 1;
3725    
3726  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3727  matching encountered yet". It gets changed to REQ_NONE if we hit something that  matching encountered yet". It gets changed to REQ_NONE if we hit something that
3728  matches a non-fixed char first char; reqbyte just remains unset if we never  matches a non-fixed char first char; reqchar just remains unset if we never
3729  find one.  find one.
3730    
3731  When we hit a repeat whose minimum is zero, we may have to adjust these values  When we hit a repeat whose minimum is zero, we may have to adjust these values
3732  to take the zero repeat into account. This is implemented by setting them to  to take the zero repeat into account. This is implemented by setting them to
3733  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3734  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3735    
3736  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;
3737    
3738  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3739  according to the current setting of the caseless flag. REQ_CASELESS is a bit  or zero, according to the current setting of the caseless flag. The
3740  value > 255. It is added into the firstbyte or reqbyte variables to record the  REQ_CASELESS leaves the lower 28 bit empty. It is added into the
3741  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3742    value. This is used only for ASCII characters. */
3743    
3744  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3745    
3746  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3747    
# Line 3126  for (;; ptr++) Line 3753  for (;; ptr++)
3753    BOOL is_quantifier;    BOOL is_quantifier;
3754    BOOL is_recurse;    BOOL is_recurse;
3755    BOOL reset_bracount;    BOOL reset_bracount;
3756    int class_charcount;    int class_has_8bitchar;
3757    int class_lastchar;    int class_one_char;
3758    int newoptions;    int newoptions;
3759    int recno;    int recno;
3760    int refsign;    int refsign;
3761    int skipbytes;    int skipbytes;
3762    int subreqbyte;    int subreqchar;
3763    int subfirstbyte;    int subfirstchar;
3764    int terminator;    int terminator;
3765    int mclength;    int mclength;
3766    uschar mcbuffer[8];    int tempbracount;
3767      pcre_uchar mcbuffer[8];
3768    
3769    /* Get next byte in the pattern */    /* Get next character in the pattern */
3770    
3771    c = *ptr;    c = *ptr;
3772    
# Line 3160  for (;; ptr++) Line 3788  for (;; ptr++)
3788  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3789      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3790  #endif  #endif
3791      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */      if (code > cd->start_workspace + cd->workspace_size -
3792            WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3793        {        {
3794        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3795        goto FAILED;        goto FAILED;
# Line 3183  for (;; ptr++) Line 3812  for (;; ptr++)
3812        }        }
3813    
3814      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3815      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3816          (int)(code - last_code), c, c));
3817    
3818      /* If "previous" is set and it is not at the start of the work space, move      /* If "previous" is set and it is not at the start of the work space, move
3819      it back to there, in order to avoid filling up the work space. Otherwise,      it back to there, in order to avoid filling up the work space. Otherwise,
# Line 3193  for (;; ptr++) Line 3823  for (;; ptr++)
3823        {        {
3824        if (previous > orig_code)        if (previous > orig_code)
3825          {          {
3826          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3827          code -= previous - orig_code;          code -= previous - orig_code;
3828          previous = orig_code;          previous = orig_code;
3829          }          }
# Line 3209  for (;; ptr++) Line 3839  for (;; ptr++)
3839    /* In the real compile phase, just check the workspace used by the forward    /* In the real compile phase, just check the workspace used by the forward
3840    reference list. */    reference list. */
3841    
3842    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3843               WORK_SIZE_SAFETY_MARGIN)
3844      {      {
3845      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3846      goto FAILED;      goto FAILED;
# Line 3261  for (;; ptr++) Line 3892  for (;; ptr++)
3892    
3893    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
3894      {      {
3895      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
3896      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3897        {        {
3898        ptr++;        ptr++;
# Line 3269  for (;; ptr++) Line 3900  for (;; ptr++)
3900          {          {
3901          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3902          ptr++;          ptr++;
3903  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3904          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3905  #endif  #endif
3906          }          }
3907        if (*ptr != 0) continue;        if (*ptr != 0) continue;
# Line 3294  for (;; ptr++) Line 3925  for (;; ptr++)
3925      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
3926      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3927      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3928      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
3929      *reqbyteptr = reqbyte;      *reqcharptr = reqchar;
3930      *codeptr = code;      *codeptr = code;
3931      *ptrptr = ptr;      *ptrptr = ptr;
3932      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3319  for (;; ptr++) Line 3950  for (;; ptr++)
3950      previous = NULL;      previous = NULL;
3951      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3952        {        {
3953        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3954        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3955        }        }
3956      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3331  for (;; ptr++) Line 3962  for (;; ptr++)
3962      break;      break;
3963    
3964      /* 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
3965      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
3966    
3967      case CHAR_DOT:      case CHAR_DOT:
3968      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3969      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
3970      zeroreqbyte = reqbyte;      zeroreqchar = reqchar;
3971      previous = code;      previous = code;
3972      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
3973      break;      break;
# Line 3391  for (;; ptr++) Line 4022  for (;; ptr++)
4022          {          {
4023          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
4024            ptr++;            ptr++;
4025          else if (strncmp((const char *)ptr+1,          else if (STRNCMP_UC_C8(ptr + 1, STR_Q STR_BACKSLASH STR_E, 3) == 0)
                           STR_Q STR_BACKSLASH STR_E, 3) == 0)  
4026            ptr += 3;            ptr += 3;
4027          else          else
4028            break;            break;
# Line 3411  for (;; ptr++) Line 4041  for (;; ptr++)
4041          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4042        {        {
4043        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4044        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4045        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
4046        break;        break;
4047        }        }
4048    
# Line 3422  for (;; ptr++) Line 4052  for (;; ptr++)
4052    
4053      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4054    
4055      /* Keep a count of chars with values < 256 so that we can optimize the case      /* For optimization purposes, we track some properties of the class:
4056      of just a single character (as long as it's < 256). However, For higher      class_has_8bitchar will be non-zero if the class contains at least one <
4057      valued UTF-8 characters, we don't yet do any optimization. */      256 character; class_one_char will be 1 if the class contains just one
4058        character. */
4059    
4060      class_charcount = 0;      class_has_8bitchar = 0;
4061      class_lastchar = -1;      class_one_char = 0;
4062    
4063      /* Initialize the 32-char bit map to all zeros. We build the map in a      /* Initialize the 32-char bit map to all zeros. We build the map in a
4064      temporary bit of memory, in case the class contains only 1 character (less      temporary bit of memory, in case the class contains fewer than two
4065      than 256), because in that case the compiled code doesn't use the bit map.      8-bit characters because in that case the compiled code doesn't use the bit
4066      */      map. */
4067    
4068      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4069    
4070  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4071      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;
4072      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4073      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4074  #endif  #endif
4075    
4076      /* 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 3448  for (;; ptr++) Line 4079  for (;; ptr++)
4079    
4080      if (c != 0) do      if (c != 0) do
4081        {        {
4082        const uschar *oldptr;        const pcre_uchar *oldptr;
4083    
4084  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4085        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
4086          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4087          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
4088          }          }
4089    #endif
4090    
4091        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4092          /* In the pre-compile phase, accumulate the length of any extra
4093        data and reset the pointer. This is so that very large classes that        data and reset the pointer. This is so that very large classes that
4094        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4095        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4096          however. */
4097        if (lengthptr != NULL)  
4098          {        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4099          *lengthptr += class_utf8data - class_utf8data_base;          {
4100          class_utf8data = class_utf8data_base;          xclass = TRUE;
4101            *lengthptr += class_uchardata - class_uchardata_base;
4102            class_uchardata = class_uchardata_base;
4103          }          }
   
4104  #endif  #endif
4105    
4106        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3494  for (;; ptr++) Line 4128  for (;; ptr++)
4128          {          {
4129          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
4130          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
4131          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
4132          uschar pbits[32];          pcre_uint8 pbits[32];
4133    
4134          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
4135            {            {
# Line 3522  for (;; ptr++) Line 4156  for (;; ptr++)
4156          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4157          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4158    
4159          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4160            posix_class = 0;            posix_class = 0;
4161    
4162          /* When PCRE_UCP is set, some of the POSIX classes are converted to          /* When PCRE_UCP is set, some of the POSIX classes are converted to
# Line 3550  for (;; ptr++) Line 4184  for (;; ptr++)
4184          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
4185    
4186          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
4187            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
4188    
4189          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
4190    
# Line 3565  for (;; ptr++) Line 4199  for (;; ptr++)
4199              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4200            }            }
4201    
4202          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4203          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4204    
4205          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3581  for (;; ptr++) Line 4215  for (;; ptr++)
4215            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4216    
4217          ptr = tempptr + 1;          ptr = tempptr + 1;
4218          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 character. */
4219            class_has_8bitchar = 1;
4220            /* Every class contains at least two characters. */
4221            class_one_char = 2;
4222          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4223          }          }
4224    
4225        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
4226        of the specials, which just set a flag. The sequence \b is a special        of the specials, which just set a flag. The sequence \b is a special
4227        case. Inside a class (and only there) it is treated as backspace. We        case. Inside a class (and only there) it is treated as backspace. We
4228        assume that other escapes have more than one character in them, so set        assume that other escapes have more than one character in them, so
4229        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_one_char bigger
4230        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
4231          as literal characters (by default), or are faulted if
4232        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4233    
4234        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
# Line 3599  for (;; ptr++) Line 4237  for (;; ptr++)
4237          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4238    
4239          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */
4240            else if (-c == ESC_N)            /* \N is not supported in a class */
4241              {
4242              *errorcodeptr = ERR71;
4243              goto FAILED;
4244              }
4245          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
4246            {            {
4247            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3612  for (;; ptr++) Line 4255  for (;; ptr++)
4255    
4256          if (c < 0)          if (c < 0)
4257            {            {
4258            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4259            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
4260              class_has_8bitchar++;
4261              /* Every class contains at least two characters. */
4262              class_one_char += 2;
4263    
4264            switch (-c)            switch (-c)
4265              {              {
# Line 3626  for (;; ptr++) Line 4272  for (;; ptr++)
4272              case ESC_SU:              case ESC_SU:
4273              nestptr = ptr;              nestptr = ptr;
4274              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
4275              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4276              continue;              continue;
4277  #endif  #endif
4278              case ESC_d:              case ESC_d:
# Line 3649  for (;; ptr++) Line 4295  for (;; ptr++)
4295    
4296              /* Perl 5.004 onwards omits VT from \s, but we must preserve it              /* Perl 5.004 onwards omits VT from \s, but we must preserve it
4297              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4298              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4299                EBCDIC, so we lazily just adjust the appropriate bit. */
4300    
4301              case ESC_s:              case ESC_s:
4302              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 3662  for (;; ptr++) Line 4309  for (;; ptr++)
4309              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4310              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4311              continue;              continue;
4312    
4313                /* The rest apply in both UCP and non-UCP cases. */
4314    
4315              case ESC_h:              case ESC_h:
4316              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4317              SETBIT(classbits, 0x20); /* SPACE */                PRIV(hspace_list), NOTACHAR);
             SETBIT(classbits, 0xa0); /* NSBP */  
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);  
               }  
 #endif  
4318              continue;              continue;
4319    
4320              case ESC_H:              case ESC_H:
4321              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4322                {                cd, PRIV(hspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x09/8: x ^= 1 << (0x09%8); break;  
                 case 0x20/8: x ^= 1 << (0x20%8); break;  
                 case 0xa0/8: x ^= 1 << (0xa0%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
   
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);  
               }  
 #endif  
4323              continue;              continue;
4324    
4325              case ESC_v:              case ESC_v:
4326              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4327              SETBIT(classbits, 0x0b); /* VT */                PRIV(vspace_list), NOTACHAR);
             SETBIT(classbits, 0x0c); /* FF */  
             SETBIT(classbits, 0x0d); /* CR */  
             SETBIT(classbits, 0x85); /* NEL */  
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);  
               }  
 #endif  
4328              continue;              continue;
4329    
4330              case ESC_V:              case ESC_V:
4331              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4332                {                cd, PRIV(vspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x0a/8: x ^= 1 << (0x0a%8);  
                              x ^= 1 << (0x0b%8);  
                              x ^= 1 << (0x0c%8);  
                              x ^= 1 << (0x0d%8);  
                              break;  
                 case 0x85/8: x ^= 1 << (0x85%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
   
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);  
               }  
 #endif  
4333              continue;              continue;
4334    
4335  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3787  for (;; ptr++) Line 4340  for (;; ptr++)
4340                int pdata;                int pdata;
4341                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
4342                if (ptype < 0) goto FAILED;                if (ptype < 0) goto FAILED;
4343                class_utf8 = TRUE;                *class_uchardata++ = ((-c == ESC_p) != negated)?
               *class_utf8data++ = ((-c == ESC_p) != negated)?  
4344                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4345                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4346                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4347                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4348                continue;                continue;
4349                }                }
4350  #endif  #endif
# Line 3806  for (;; ptr++) Line 4358  for (;; ptr++)
4358                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4359                goto FAILED;                goto FAILED;
4360                }                }
4361              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4362              c = *ptr;              /* Get the final character and fall through */              class_one_char -= 2;     /* Undo the speculative increase. */
4363                c = *ptr;                /* Get the final character and fall through */
4364              break;              break;
4365              }              }
4366            }            }
4367    
4368          /* Fall through if we have a single character (c >= 0). This may be          /* Fall through if the escape just defined a single character (c >= 0).
4369          greater than 256 in UTF-8 mode. */          This may be greater than 256. */
4370    
4371          }   /* End of backslash handling */          }   /* End of backslash handling */
4372    
4373        /* A single character may be followed by '-' to form a range. However,        /* A character may be followed by '-' to form a range. However, Perl does
4374        Perl does not permit ']' to be the end of the range. A '-' character        not permit ']' to be the end of the range. A '-' character at the end is
4375        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4376        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4377    
4378        CHECK_RANGE:        CHECK_RANGE:
4379        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3828  for (;; ptr++) Line 4381  for (;; ptr++)
4381          inescq = FALSE;          inescq = FALSE;
4382          ptr += 2;          ptr += 2;
4383          }          }
   
4384        oldptr = ptr;        oldptr = ptr;
4385    
4386        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4387    
4388        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4389    
# Line 3854  for (;; ptr++) Line 4406  for (;; ptr++)
4406            inescq = TRUE;            inescq = TRUE;
4407            break;            break;
4408            }            }
4409    
4410            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4411            back the pointer and jump to handle the character that preceded it. */
4412    
4413          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4414            {            {
4415            ptr = oldptr;            ptr = oldptr;
4416            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4417            }            }
4418    
4419            /* Otherwise, we have a potential range; pick up the next character */
4420    
4421  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4422          if (utf8)          if (utf)
4423            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4424            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4425            }            }
# Line 3874  for (;; ptr++) Line 4431  for (;; ptr++)
4431          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
4432          in such circumstances. */          in such circumstances. */
4433    
4434          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4435            {            {
4436            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);
4437            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
   
           /* \b is backspace; any other special means the '-' was literal */  
   
           if (d < 0)  
             {  
             if (d == -ESC_b) d = CHAR_BS; else  
               {  
               ptr = oldptr;  
               goto LONE_SINGLE_CHARACTER;  /* A few lines below */  
               }  
             }  
           }  
   
         /* Check that the two values are in the correct order. Optimize  
         one-character ranges */  
   
         if (d < c)  
           {  
           *errorcodeptr = ERR8;  
           goto FAILED;  
           }  
   
         if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */  
   
         /* Remember \r or \n */  
   
         if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;  
   
         /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless  
         matching, we have to use an XCLASS with extra data items. Caseless  
         matching for characters > 127 is available only if UCP support is  
         available. */  
   
 #ifdef SUPPORT_UTF8  
         if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))  
           {  
           class_utf8 = TRUE;  
   
           /* With UCP support, we can find the other case equivalents of  
           the relevant characters. There may be several ranges. Optimize how  
           they fit with the basic range. */  
   
 #ifdef SUPPORT_UCP  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             unsigned int occ, ocd;  
             unsigned int cc = c;  
             unsigned int origd = d;  
             while (get_othercase_range(&cc, origd, &occ, &ocd))  
               {  
               if (occ >= (unsigned int)c &&  
                   ocd <= (unsigned int)d)  
                 continue;                          /* Skip embedded ranges */  
   
               if (occ < (unsigned int)c  &&  
                   ocd >= (unsigned int)c - 1)      /* Extend the basic range */  
                 {                                  /* if there is overlap,   */  
                 c = occ;                           /* noting that if occ < c */  
                 continue;                          /* we can't have ocd > d  */  
                 }                                  /* because a subrange is  */  
               if (ocd > (unsigned int)d &&  
                   occ <= (unsigned int)d + 1)      /* always shorter than    */  
                 {                                  /* the basic range.       */  
                 d = ocd;  
                 continue;  
                 }  
   
               if (occ == ocd)  
                 {  
                 *class_utf8data++ = XCL_SINGLE;  
                 }  
               else  
                 {  
                 *class_utf8data++ = XCL_RANGE;  
                 class_utf8data += _pcre_ord2utf8(occ, class_utf8data);  
                 }  
               class_utf8data += _pcre_ord2utf8(ocd, class_utf8data);  
               }  
             }  
 #endif  /* SUPPORT_UCP */  
   
           /* Now record the original range, possibly modified for UCP caseless  
           overlapping ranges. */  
   
           *class_utf8data++ = XCL_RANGE;  
           class_utf8data += _pcre_ord2utf8(c, class_utf8data);  
           class_utf8data += _pcre_ord2utf8(d, class_utf8data);  
   
           /* With UCP support, we are done. Without UCP support, there is no  
           caseless matching for UTF-8 characters > 127; we can use the bit map  
           for the smaller ones. */  
   
 #ifdef SUPPORT_UCP  
           continue;    /* With next character in the class */  
 #else  
           if ((options & PCRE_CASELESS) == 0 || c > 127) continue;  
   
           /* Adjust upper limit and fall through to set up the map */  
4438    
4439            d = 127;            /* \b is backspace; any other special means the '-' was literal. */
4440    
4441  #endif  /* SUPPORT_UCP */            if (d < 0)
4442                {
4443                if (d == -ESC_b) d = CHAR_BS; else
4444                  {
4445                  ptr = oldptr;
4446                  goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4447                  }
4448                }
4449            }            }
 #endif  /* SUPPORT_UTF8 */  
4450    
4451          /* We use the bit map for all cases when not in UTF-8 mode; else          /* Check that the two values are in the correct order. Optimize
4452          ranges that lie entirely within 0-127 when there is UCP support; else          one-character ranges. */
         for partial ranges without UCP support. */  
   
         class_charcount += d - c + 1;  
         class_lastchar = d;  
   
         /* We can save a bit of time by skipping this in the pre-compile. */  
4453    
4454          if (lengthptr == NULL) for (; c <= d; c++)          if (d < c)
4455            {            {
4456            classbits[c/8] |= (1 << (c&7));            *errorcodeptr = ERR8;
4457            if ((options & PCRE_CASELESS) != 0)            goto FAILED;
             {  
             int uc = cd->fcc[c];           /* flip case */  
             classbits[uc/8] |= (1 << (uc&7));  
             }  
4458            }            }
4459            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4460    
4461          continue;   /* Go get the next char in the class */          /* We have found a character range, so single character optimizations
4462          }          cannot be done anymore. Any value greater than 1 indicates that there
4463            is more than one character. */
4464    
4465            class_one_char = 2;
4466    
4467        /* Handle a lone single character - we can get here for a normal          /* Remember an explicit \r or \n, and add the range to the class. */
       non-escape char, or after \ that introduces a single character or for an  
       apparent range that isn't. */  
4468    
4469        LONE_SINGLE_CHARACTER:          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4470    
4471            class_has_8bitchar +=
4472              add_to_class(classbits, &class_uchardata, options, cd, c, d);
4473    
4474            continue;   /* Go get the next char in the class */
4475            }
4476    
4477        /* Handle a character that cannot go in the bit map */        /* Handle a single character - we can get here for a normal non-escape
4478          char, or after \ that introduces a single character or for an apparent
4479          range that isn't. Only the value 1 matters for class_one_char, so don't
4480          increase it if it is already 2 or more ... just in case there's a class
4481          with a zillion characters in it. */
4482    
4483          CLASS_SINGLE_CHARACTER:
4484          if (class_one_char < 2) class_one_char++;
4485    
4486          /* If class_one_char is 1, we have the first single character in the
4487          class, and there have been no prior ranges, or XCLASS items generated by
4488          escapes. If this is the final character in the class, we can optimize by
4489          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4490          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4491          to be set. Otherwise, there can be no first char if this item is first,
4492          whatever repeat count may follow. In the case of reqchar, save the
4493          previous value for reinstating. */
4494    
4495  #ifdef SUPPORT_UTF8        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
       if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
4496          {          {
4497          class_utf8 = TRUE;          ptr++;
4498          *class_utf8data++ = XCL_SINGLE;          zeroreqchar = reqchar;
         class_utf8data += _pcre_ord2utf8(c, class_utf8data);  
4499    
4500  #ifdef SUPPORT_UCP          if (negate_class)
         if ((options & PCRE_CASELESS) != 0)  
4501            {            {
4502            unsigned int othercase;  #ifdef SUPPORT_UCP
4503            if ((othercase = UCD_OTHERCASE(c)) != c)            int d;
4504    #endif
4505              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4506              zerofirstchar = firstchar;
4507    
4508              /* For caseless UTF-8 mode when UCP support is available, check
4509              whether this character has more than one other case. If so, generate
4510              a special OP_NOTPROP item instead of OP_NOTI. */
4511    
4512    #ifdef SUPPORT_UCP
4513              if (utf && (options & PCRE_CASELESS) != 0 &&
4514                  (d = UCD_CASESET(c)) != 0)
4515                {
4516                *code++ = OP_NOTPROP;
4517                *code++ = PT_CLIST;
4518                *code++ = d;
4519                }
4520              else
4521    #endif
4522              /* Char has only one other case, or UCP not available */
4523    
4524              {              {
4525              *class_utf8data++ = XCL_SINGLE;              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4526              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);  #ifdef SUPPORT_UTF
4527                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4528                  code += PRIV(ord2utf)(c, code);
4529                else
4530    #endif
4531                  *code++ = c;
4532              }              }
4533    
4534              /* We are finished with this character class */
4535    
4536              goto END_CLASS;
4537            }            }
 #endif  /* SUPPORT_UCP */  
4538    
4539          }          /* For a single, positive character, get the value into mcbuffer, and
4540        else          then we can handle this with the normal one-character code. */
 #endif  /* SUPPORT_UTF8 */  
4541    
4542        /* Handle a single-byte character */  #ifdef SUPPORT_UTF
4543          {          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4544          classbits[c/8] |= (1 << (c&7));            mclength = PRIV(ord2utf)(c, mcbuffer);
4545          if ((options & PCRE_CASELESS) != 0)          else
4546    #endif
4547            {            {
4548            c = cd->fcc[c];   /* flip case */            mcbuffer[0] = c;
4549            classbits[c/8] |= (1 << (c&7));            mclength = 1;
4550            }            }
4551          class_charcount++;          goto ONE_CHAR;
4552          class_lastchar = c;          }       /* End of 1-char optimization */
4553          }  
4554          /* There is more than one character in the class, or an XCLASS item
4555          has been generated. Add this character to the class. */
4556    
4557          class_has_8bitchar +=
4558            add_to_class(classbits, &class_uchardata, options, cd, c, c);
4559        }        }
4560    
4561      /* Loop until ']' reached. This "while" is the end of the "do" far above.      /* Loop until ']' reached. This "while" is the end of the "do" far above.
# Line 4066  for (;; ptr++) Line 4575  for (;; ptr++)
4575        goto FAILED;        goto FAILED;
4576        }        }
4577    
4578      /* If class_charcount is 1, we saw precisely one character whose value is      /* We will need an XCLASS if data has been placed in class_uchardata. In
4579      less than 256. As long as there were no characters >= 128 and there was no      the second phase this is a sufficient test. However, in the pre-compile
4580      use of \p or \P, in other words, no use of any XCLASS features, we can      phase, class_uchardata gets emptied to prevent workspace overflow, so it
4581      optimize.      only if the very last character in the class needs XCLASS will it contain
4582        anything at this point. For this reason, xclass gets set TRUE above when
4583      In UTF-8 mode, we can optimize the negative case only if there were no      uchar_classdata is emptied, and that's why this code is the way it is here
4584      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      instead of just doing a test on class_uchardata below. */
4585      operate on single-bytes characters only. This is an historical hangover.  
4586      Maybe one day we can tidy these opcodes to handle multi-byte characters.  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4587        if (class_uchardata > class_uchardata_base) xclass = TRUE;
     The optimization throws away the bit map. We turn the item into a  
     1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
     Note that OP_NOT[I] does not support multibyte characters. In the positive  
     case, it can cause firstbyte to be set. Otherwise, there can be no first  
     char if this item is first, whatever repeat count may follow. In the case  
     of reqbyte, save the previous value for reinstating. */  
   
 #ifdef SUPPORT_UTF8  
     if (class_charcount == 1 && !class_utf8 &&  
       (!utf8 || !negate_class || class_lastchar < 128))  
 #else  
     if (class_charcount == 1)  
 #endif  
       {  
       zeroreqbyte = reqbyte;  
   
       /* The OP_NOT[I] opcodes work on one-byte characters only. */  
   
       if (negate_class)  
         {  
         if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;  
         zerofirstbyte = firstbyte;  
         *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;  
         *code++ = class_lastchar;  
         break;  
         }  
   
       /* For a single, positive character, get the value into mcbuffer, and  
       then we can handle this with the normal one-character code. */  
   
 #ifdef SUPPORT_UTF8  
       if (utf8 && class_lastchar > 127)  
         mclength = _pcre_ord2utf8(class_lastchar, mcbuffer);  
       else  
4588  #endif  #endif
         {  
         mcbuffer[0] = class_lastchar;  
         mclength = 1;  
         }  
       goto ONE_CHAR;  
       }       /* End of 1-char optimization */  
4589    
4590      /* The general case - not the one-char optimization. If this is the first      /* If this is the first thing in the branch, there can be no first char
4591      thing in the branch, there can be no first char setting, whatever the      setting, whatever the repeat count. Any reqchar setting must remain
4592      repeat count. Any reqbyte setting must remain unchanged after any kind of      unchanged after any kind of repeat. */
     repeat. */  
4593    
4594      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4595      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
4596      zeroreqbyte = reqbyte;      zeroreqchar = reqchar;
4597    
4598      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4599      extended class, with its own opcode, unless there was a negated special      extended class, with its own opcode, unless there was a negated special
# Line 4135  for (;; ptr++) Line 4603  for (;; ptr++)
4603      be listed) there are no characters < 256, we can omit the bitmap in the      be listed) there are no characters < 256, we can omit the bitmap in the
4604      actual compiled code. */      actual compiled code. */
4605    
4606  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4607      if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))      if (xclass && (!should_flip_negation || (options & PCRE_UCP) != 0))
4608    #elif !defined COMPILE_PCRE8
4609        if (xclass && !should_flip_negation)
4610    #endif
4611    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4612        {        {
4613        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_uchardata++ = XCL_END;    /* Marks the end of extra data */
4614        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
4615        code += LINK_SIZE;        code += LINK_SIZE;
4616        *code = negate_class? XCL_NOT : 0;        *code = negate_class? XCL_NOT:0;
4617    
4618        /* If the map is required, move up the extra data to make room for it;        /* If the map is required, move up the extra data to make room for it;
4619        otherwise just move the code pointer to the end of the extra data. */        otherwise just move the code pointer to the end of the extra data. */
4620    
4621        if (class_charcount > 0)        if (class_has_8bitchar > 0)
4622          {          {
4623          *code++ |= XCL_MAP;          *code++ |= XCL_MAP;
4624          memmove(code + 32, code, class_utf8data - code);          memmove(code + (32 / sizeof(pcre_uchar)), code,
4625              IN_UCHARS(class_uchardata - code));
4626          memcpy(code, classbits, 32);          memcpy(code, classbits, 32);
4627          code = class_utf8data + 32;          code = class_uchardata + (32 / sizeof(pcre_uchar));
4628          }          }
4629        else code = class_utf8data;        else code = class_uchardata;
4630    
4631        /* Now fill in the complete length of the item */        /* Now fill in the complete length of the item */
4632    
4633        PUT(previous, 1, code - previous);        PUT(previous, 1, (int)(code - previous));
4634        break;   /* End of class handling */        break;   /* End of class handling */
4635        }        }
4636  #endif  #endif
# Line 4169  for (;; ptr++) Line 4642  for (;; ptr++)
4642      negating it if necessary. */      negating it if necessary. */
4643    
4644      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
4645      if (negate_class)      if (lengthptr == NULL)    /* Save time in the pre-compile phase */
       {  
       if (lengthptr == NULL)    /* Save time in the pre-compile phase */  
         for (c = 0; c < 32; c++) code[c] = ~classbits[c];  
       }  
     else  
4646        {        {
4647          if (negate_class)
4648            for (c = 0; c < 32; c++) classbits[c] = ~classbits[c];
4649        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4650        }        }
4651      code += 32;      code += 32 / sizeof(pcre_uchar);
4652    
4653        END_CLASS:
4654      break;      break;
4655    
4656    
# Line 4215  for (;; ptr++) Line 4687  for (;; ptr++)
4687    
4688      if (repeat_min == 0)      if (repeat_min == 0)
4689        {        {
4690        firstbyte = zerofirstbyte;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4691        reqbyte = zeroreqbyte;        /* Ditto */        reqchar = zeroreqchar;        /* Ditto */
4692        }        }
4693    
4694      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4249  for (;; ptr++) Line 4721  for (;; ptr++)
4721        ptr++;        ptr++;
4722        }        }
4723      else repeat_type = greedy_default;      else repeat_type = greedy_default;
4724    
4725      /* If previous was a recursion call, wrap it in atomic brackets so that      /* If previous was a recursion call, wrap it in atomic brackets so that
4726      previous becomes the atomic group. All recursions were so wrapped in the      previous becomes the atomic group. All recursions were so wrapped in the
4727      past, but it no longer happens for non-repeated recursions. In fact, the      past, but it no longer happens for non-repeated recursions. In fact, the
4728      repeated ones could be re-implemented independently so as not to need this,      repeated ones could be re-implemented independently so as not to need this,
4729      but for the moment we rely on the code for repeating groups. */      but for the moment we rely on the code for repeating groups. */
4730    
4731      if (*previous == OP_RECURSE)      if (*previous == OP_RECURSE)
4732        {        {
4733        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);        memmove(previous + 1 + LINK_SIZE, previous, IN_UCHARS(1 + LINK_SIZE));
4734        *previous = OP_ONCE;        *previous = OP_ONCE;
4735        PUT(previous, 1, 2 + 2*LINK_SIZE);        PUT(previous, 1, 2 + 2*LINK_SIZE);
4736        previous[2 + 2*LINK_SIZE] = OP_KET;        previous[2 + 2*LINK_SIZE] = OP_KET;
4737        PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE);        PUT(previous, 3 + 2*LINK_SIZE, 2 + 2*LINK_SIZE);
4738        code += 2 + 2 * LINK_SIZE;        code += 2 + 2 * LINK_SIZE;
4739        length_prevgroup = 3 + 3*LINK_SIZE;        length_prevgroup = 3 + 3*LINK_SIZE;
4740    
4741        /* When actually compiling, we need to check whether this was a forward        /* When actually compiling, we need to check whether this was a forward
4742        reference, and if so, adjust the offset. */        reference, and if so, adjust the offset. */
4743    
4744        if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE)        if (lengthptr == NULL && cd->hwm >= cd->start_workspace + LINK_SIZE)
4745          {          {
4746          int offset = GET(cd->hwm, -LINK_SIZE);          int offset = GET(cd->hwm, -LINK_SIZE);
4747          if (offset == previous + 1 - cd->start_code)          if (offset == previous + 1 - cd->start_code)
4748            PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE);            PUT(cd->hwm, -LINK_SIZE, offset + 1 + LINK_SIZE);
4749          }          }
4750        }        }
4751    
4752      /* Now handle repetition for the different types of item. */      /* Now handle repetition for the different types of item. */
4753    
4754      /* If previous was a character match, abolish the item and generate a      /* If previous was a character or negated character match, abolish the item
4755      repeat item instead. If a char item has a minumum of more than one, ensure      and generate a repeat item instead. If a char item has a minimum of more
4756      that it is set in reqbyte - it might not be if a sequence such as x{3} is      than one, ensure that it is set in reqchar - it might not be if a sequence
4757      the first thing in a branch because the x will have gone into firstbyte      such as x{3} is the first thing in a branch because the x will have gone
4758      instead.  */      into firstchar instead.  */
4759    
4760      if (*previous == OP_CHAR || *previous == OP_CHARI)      if (*previous == OP_CHAR || *previous == OP_CHARI
4761            || *previous == OP_NOT || *previous == OP_NOTI)
4762        {        {
4763        op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;        switch (*previous)
4764            {
4765            default: /* Make compiler happy. */
4766            case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;
4767            case OP_CHARI: op_type = OP_STARI - OP_STAR; break;
4768            case OP_NOT:   op_type = OP_NOTSTAR - OP_STAR; break;
4769            case OP_NOTI:  op_type = OP_NOTSTARI - OP_STAR; break;
4770            }
4771    
4772        /* Deal with UTF-8 characters that take up more than one byte. It's        /* Deal with UTF characters that take up more than one character. It's
4773        easier to write this out separately than try to macrify it. Use c to        easier to write this out separately than try to macrify it. Use c to
4774        hold the length of the character in bytes, plus 0x80 to flag that it's a        hold the length of the character in bytes, plus UTF_LENGTH to flag that
4775        length rather than a small character. */        it's a length rather than a small character. */
4776    
4777  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4778        if (utf8 && (code[-1] & 0x80) != 0)        if (utf && NOT_FIRSTCHAR(code[-1]))
4779          {          {
4780          uschar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
4781          while((*lastchar & 0xc0) == 0x80) lastchar--;          BACKCHAR(lastchar);
4782          c = code - lastchar;            /* Length of UTF-8 character */          c = (int)(code - lastchar);     /* Length of UTF-8 character */
4783          memcpy(utf8_char, lastchar, c); /* Save the char */          memcpy(utf_chars, lastchar, IN_UCHARS(c)); /* Save the char */
4784          c |= 0x80;                      /* Flag c as a length */          c |= UTF_LENGTH;                /* Flag c as a length */
4785          }          }
4786        else        else
4787  #endif  #endif /* SUPPORT_UTF */
   
       /* Handle the case of a single byte - either with no UTF8 support, or  
       with UTF-8 disabled, or for a UTF-8 character < 128. */  
4788    
4789          /* Handle the case of a single charater - either with no UTF support, or
4790          with UTF disabled, or for a single character UTF character. */
4791          {          {
4792          c = code[-1];          c = code[-1];
4793          if (repeat_min > 1) reqbyte = c | req_caseopt | cd->req_varyopt;          if (*previous <= OP_CHARI && repeat_min > 1)
4794              reqchar = c | req_caseopt | cd->req_varyopt;
4795          }          }
4796    
4797        /* If the repetition is unlimited, it pays to see if the next thing on        /* If the repetition is unlimited, it pays to see if the next thing on
# Line 4321  for (;; ptr++) Line 4801  for (;; ptr++)
4801    
4802        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4803            repeat_max < 0 &&            repeat_max < 0 &&
4804            check_auto_possessive(previous, utf8, ptr + 1, options, cd))            check_auto_possessive(previous, utf, ptr + 1, options, cd))
4805          {          {
4806          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4807          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 4330  for (;; ptr++) Line 4810  for (;; ptr++)
4810        goto OUTPUT_SINGLE_REPEAT;   /* Code shared with single character types */        goto OUTPUT_SINGLE_REPEAT;   /* Code shared with single character types */
4811        }        }
4812    
     /* If previous was a single negated character ([^a] or similar), we use  
     one of the special opcodes, replacing it. The code is shared with single-  
     character repeats by setting opt_type to add a suitable offset into  
     repeat_type. We can also test for auto-possessification. OP_NOT and OP_NOTI  
     are currently used only for single-byte chars. */  
   
     else if (*previous == OP_NOT || *previous == OP_NOTI)  
       {  
       op_type = ((*previous == OP_NOT)? OP_NOTSTAR : OP_NOTSTARI) - OP_STAR;  
       c = previous[1];  
       if (!possessive_quantifier &&  
           repeat_max < 0 &&  
           check_auto_possessive(previous, utf8, ptr + 1, options, cd))  
         {  
         repeat_type = 0;    /* Force greedy */  
         possessive_quantifier = TRUE;  
         }  
       goto OUTPUT_SINGLE_REPEAT;  
       }  
   
4813      /* If previous was a character type match (\d or similar), abolish it and      /* If previous was a character type match (\d or similar), abolish it and
4814      create a suitable repeat item. The code is shared with single-character      create a suitable repeat item. The code is shared with single-character
4815      repeats by setting op_type to add a suitable offset into repeat_type. Note      repeats by setting op_type to add a suitable offset into repeat_type. Note
# Line 4359  for (;; ptr++) Line 4819  for (;; ptr++)
4819    
4820      else if (*previous < OP_EODN)      else if (*previous < OP_EODN)
4821        {        {
4822        uschar *oldcode;        pcre_uchar *oldcode;
4823        int prop_type, prop_value;        int prop_type, prop_value;
4824        op_type = OP_TYPESTAR - OP_STAR;  /* Use type opcodes */        op_type = OP_TYPESTAR - OP_STAR;  /* Use type opcodes */
4825        c = *previous;        c = *previous;
4826    
4827        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4828            repeat_max < 0 &&            repeat_max < 0 &&
4829            check_auto_possessive(previous, utf8, ptr + 1, options, cd))            check_auto_possessive(previous, utf, ptr + 1, options, cd))
4830          {          {
4831          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4832          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 4446  for (;; ptr++) Line 4906  for (;; ptr++)
4906          we have to insert the character for the previous code. For a repeated          we have to insert the character for the previous code. For a repeated
4907          Unicode property match, there are two extra bytes that define the          Unicode property match, there are two extra bytes that define the
4908          required property. In UTF-8 mode, long characters have their length in          required property. In UTF-8 mode, long characters have their length in
4909          c, with the 0x80 bit as a flag. */          c, with the UTF_LENGTH bit as a flag. */
4910    
4911          if (repeat_max < 0)          if (repeat_max < 0)
4912            {            {
4913  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4914            if (utf8 && c >= 128)            if (utf && (c & UTF_LENGTH) != 0)
4915              {              {
4916              memcpy(code, utf8_char, c & 7);              memcpy(code, utf_chars, IN_UCHARS(c & 7));
4917              code += c & 7;              code += c & 7;
4918              }              }
4919            else            else
# Line 4475  for (;; ptr++) Line 4935  for (;; ptr++)
4935    
4936          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
4937            {            {
4938  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4939            if (utf8 && c >= 128)            if (utf && (c & UTF_LENGTH) != 0)
4940              {              {
4941              memcpy(code, utf8_char, c & 7);              memcpy(code, utf_chars, IN_UCHARS(c & 7));
4942              code += c & 7;              code += c & 7;
4943              }              }
4944            else            else
# Line 4505  for (;; ptr++) Line 4965  for (;; ptr++)
4965    
4966        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
4967    
4968  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4969        if (utf8 && c >= 128)        if (utf && (c & UTF_LENGTH) != 0)
4970          {          {
4971          memcpy(code, utf8_char, c & 7);          memcpy(code, utf_chars, IN_UCHARS(c & 7));
4972          code += c & 7;          code += c & 7;
4973          }          }
4974        else        else
# Line 4532  for (;; ptr++) Line 4992  for (;; ptr++)
4992    
4993      else if (*previous == OP_CLASS ||      else if (*previous == OP_CLASS ||
4994               *previous == OP_NCLASS ||               *previous == OP_NCLASS ||
4995  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4996               *previous == OP_XCLASS ||               *previous == OP_XCLASS ||
4997  #endif  #endif
4998               *previous == OP_REF ||               *previous == OP_REF ||
# Line 4574  for (;; ptr++) Line 5034  for (;; ptr++)
5034      opcodes such as BRA and CBRA, as this is the place where they get converted      opcodes such as BRA and CBRA, as this is the place where they get converted
5035      into the more special varieties such as BRAPOS and SBRA. A test for >=      into the more special varieties such as BRAPOS and SBRA. A test for >=
5036      OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK,      OP_ASSERT and <= OP_COND includes ASSERT, ASSERT_NOT, ASSERTBACK,
5037      ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow      ASSERTBACK_NOT, ONCE, BRA, CBRA, and COND. Originally, PCRE did not allow
5038      repetition of assertions, but now it does, for Perl compatibility. */      repetition of assertions, but now it does, for Perl compatibility. */
5039    
5040      else if (*previous >= OP_ASSERT && *previous <= OP_COND)      else if (*previous >= OP_ASSERT && *previous <= OP_COND)
5041        {        {
5042        register int i;        register int i;
5043        int len = (int)(code - previous);        int len = (int)(code - previous);
5044        uschar *bralink = NULL;        pcre_uchar *bralink = NULL;
5045        uschar *brazeroptr = NULL;        pcre_uchar *brazeroptr = NULL;
5046    
5047        /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so        /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so
5048        we just ignore the repeat. */        we just ignore the repeat. */
5049    
5050        if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF)        if (*previous == OP_COND && previous[LINK_SIZE+1] == OP_DEF)
5051          goto END_REPEAT;          goto END_REPEAT;
5052    
5053        /* There is no sense in actually repeating assertions. The only potential        /* There is no sense in actually repeating assertions. The only potential
5054        use of repetition is in cases when the assertion is optional. Therefore,        use of repetition is in cases when the assertion is optional. Therefore,
5055        if the minimum is greater than zero, just ignore the repeat. If the        if the minimum is greater than zero, just ignore the repeat. If the
5056        maximum is not not zero or one, set it to 1. */        maximum is not not zero or one, set it to 1. */
5057    
5058        if (*previous < OP_ONCE)    /* Assertion */        if (*previous < OP_ONCE)    /* Assertion */
5059          {          {
5060          if (repeat_min > 0) goto END_REPEAT;          if (repeat_min > 0) goto END_REPEAT;
5061          if (repeat_max < 0 || repeat_max > 1) repeat_max = 1;          if (repeat_max < 0 || repeat_max > 1) repeat_max = 1;
5062          }          }
5063    
5064        /* The case of a zero minimum is special because of the need to stick        /* The case of a zero minimum is special because of the need to stick
5065        OP_BRAZERO in front of it, and because the group appears once in the        OP_BRAZERO in front of it, and because the group appears once in the
# Line 4635  for (;; ptr++) Line 5095  for (;; ptr++)
5095          if (repeat_max <= 1)    /* Covers 0, 1, and unlimited */          if (repeat_max <= 1)    /* Covers 0, 1, and unlimited */
5096            {            {