/[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 773 by ph10, Wed Nov 30 18:10:27 2011 UTC revision 849 by ph10, Wed Jan 4 17:02:01 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    
# Line 88  so this number is very generous. Line 92  so this number is very generous.
92  The same workspace is used during the second, actual compile phase for  The same workspace is used during the second, actual compile phase for
93  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
94  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
95  is 4 there is plenty of room for most patterns. However, the memory can get  is 4 there is plenty of room for most patterns. However, the memory can get
96  filled up by repetitions of forward references, for example patterns like  filled up by repetitions of forward references, for example patterns like
97  /(?1){0,1999}(b)/, and one user did hit the limit. The code has been changed so  /(?1){0,1999}(b)/, and one user did hit the limit. The code has been changed so
98  that the workspace is expanded using malloc() in this situation. The value  that the workspace is expanded using malloc() in this situation. The value
99  below is therefore a minimum, and we put a maximum on it for safety. The  below is therefore a minimum, and we put a maximum on it for safety. The
100  minimum is now also defined in terms of LINK_SIZE so that the use of malloc()  minimum is now also defined in terms of LINK_SIZE so that the use of malloc()
101  kicks in at the same number of forward references in all cases. */  kicks in at the same number of forward references in all cases. */
102    
103  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)
# Line 104  overrun before it actually does run off Line 108  overrun before it actually does run off
108    
109  #define WORK_SIZE_SAFETY_MARGIN (100)  #define WORK_SIZE_SAFETY_MARGIN (100)
110    
111    /* Private flags added to firstchar and reqchar. */
112    
113    #define REQ_CASELESS   0x10000000l      /* Indicates caselessness */
114    #define REQ_VARY       0x20000000l      /* Reqchar followed non-literal item */
115    
116    /* Repeated character flags. */
117    
118    #define UTF_LENGTH     0x10000000l      /* The char contains its length. */
119    
120  /* Table for handling escaped characters in the range '0'-'z'. Positive returns  /* Table for handling escaped characters in the range '0'-'z'. Positive returns
121  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 238  static const char posix_names[] = Line 250  static const char posix_names[] =
250    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0
251    STRING_word0  STRING_xdigit;    STRING_word0  STRING_xdigit;
252    
253  static const uschar posix_name_lengths[] = {  static const pcre_uint8 posix_name_lengths[] = {
254    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 };
255    
256  /* 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 273  substitutes must be in the order of the Line 285  substitutes must be in the order of the
285  both positive and negative cases. NULL means no substitute. */  both positive and negative cases. NULL means no substitute. */
286    
287  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
288  static const uschar *substitutes[] = {  static const pcre_uchar string_PNd[]  = {
289    (uschar *)"\\P{Nd}",    /* \D */    CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
290    (uschar *)"\\p{Nd}",    /* \d */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
291    (uschar *)"\\P{Xsp}",   /* \S */       /* NOTE: Xsp is Perl space */  static const pcre_uchar string_pNd[]  = {
292    (uschar *)"\\p{Xsp}",   /* \s */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
293    (uschar *)"\\P{Xwd}",   /* \W */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
294    (uschar *)"\\p{Xwd}"    /* \w */  static const pcre_uchar string_PXsp[] = {
295      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
296      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
297    static const pcre_uchar string_pXsp[] = {
298      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
299      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
300    static const pcre_uchar string_PXwd[] = {
301      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
302      CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
303    static const pcre_uchar string_pXwd[] = {
304      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
305      CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
306    
307    static const pcre_uchar *substitutes[] = {
308      string_PNd,           /* \D */
309      string_pNd,           /* \d */
310      string_PXsp,          /* \S */       /* NOTE: Xsp is Perl space */
311      string_pXsp,          /* \s */
312      string_PXwd,          /* \W */
313      string_pXwd           /* \w */
314  };  };
315    
316  static const uschar *posix_substitutes[] = {  static const pcre_uchar string_pL[] =   {
317    (uschar *)"\\p{L}",     /* alpha */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
318    (uschar *)"\\p{Ll}",    /* lower */    CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
319    (uschar *)"\\p{Lu}",    /* upper */  static const pcre_uchar string_pLl[] =  {
320    (uschar *)"\\p{Xan}",   /* alnum */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
321    NULL,                   /* ascii */    CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
322    (uschar *)"\\h",        /* blank */  static const pcre_uchar string_pLu[] =  {
323    NULL,                   /* cntrl */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
324    (uschar *)"\\p{Nd}",    /* digit */    CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
325    NULL,                   /* graph */  static const pcre_uchar string_pXan[] = {
326    NULL,                   /* print */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
327    NULL,                   /* punct */    CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
328    (uschar *)"\\p{Xps}",   /* space */    /* NOTE: Xps is POSIX space */  static const pcre_uchar string_h[] =    {
329    (uschar *)"\\p{Xwd}",   /* word */    CHAR_BACKSLASH, CHAR_h, '\0' };
330    NULL,                   /* xdigit */  static const pcre_uchar string_pXps[] = {
331      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
332      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
333    static const pcre_uchar string_PL[] =   {
334      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
335      CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
336    static const pcre_uchar string_PLl[] =  {
337      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
338      CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
339    static const pcre_uchar string_PLu[] =  {
340      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
341      CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
342    static const pcre_uchar string_PXan[] = {
343      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
344      CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
345    static const pcre_uchar string_H[] =    {
346      CHAR_BACKSLASH, CHAR_H, '\0' };
347    static const pcre_uchar string_PXps[] = {
348      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
349      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
350    
351    static const pcre_uchar *posix_substitutes[] = {
352      string_pL,            /* alpha */
353      string_pLl,           /* lower */
354      string_pLu,           /* upper */
355      string_pXan,          /* alnum */
356      NULL,                 /* ascii */
357      string_h,             /* blank */
358      NULL,                 /* cntrl */
359      string_pNd,           /* digit */
360      NULL,                 /* graph */
361      NULL,                 /* print */
362      NULL,                 /* punct */
363      string_pXps,          /* space */    /* NOTE: Xps is POSIX space */
364      string_pXwd,          /* word */
365      NULL,                 /* xdigit */
366    /* Negated cases */    /* Negated cases */
367    (uschar *)"\\P{L}",     /* ^alpha */    string_PL,            /* ^alpha */
368    (uschar *)"\\P{Ll}",    /* ^lower */    string_PLl,           /* ^lower */
369    (uschar *)"\\P{Lu}",    /* ^upper */    string_PLu,           /* ^upper */
370    (uschar *)"\\P{Xan}",   /* ^alnum */    string_PXan,          /* ^alnum */
371    NULL,                   /* ^ascii */    NULL,                 /* ^ascii */
372    (uschar *)"\\H",        /* ^blank */    string_H,             /* ^blank */
373    NULL,                   /* ^cntrl */    NULL,                 /* ^cntrl */
374    (uschar *)"\\P{Nd}",    /* ^digit */    string_PNd,           /* ^digit */
375    NULL,                   /* ^graph */    NULL,                 /* ^graph */
376    NULL,                   /* ^print */    NULL,                 /* ^print */
377    NULL,                   /* ^punct */    NULL,                 /* ^punct */
378    (uschar *)"\\P{Xps}",   /* ^space */   /* NOTE: Xps is POSIX space */    string_PXps,          /* ^space */   /* NOTE: Xps is POSIX space */
379    (uschar *)"\\P{Xwd}",   /* ^word */    string_PXwd,          /* ^word */
380    NULL                    /* ^xdigit */    NULL                  /* ^xdigit */
381  };  };
382  #define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))  #define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))
383  #endif  #endif
384    
385  #define STRING(a)  # a  #define STRING(a)  # a
# Line 372  static const char error_texts[] = Line 438  static const char error_texts[] =
438    /* 30 */    /* 30 */
439    "unknown POSIX class name\0"    "unknown POSIX class name\0"
440    "POSIX collating elements are not supported\0"    "POSIX collating elements are not supported\0"
441    "this version of PCRE is not compiled with PCRE_UTF8 support\0"    "this version of PCRE is compiled without UTF support\0"
442    "spare error\0"  /** DEAD **/    "spare error\0"  /** DEAD **/
443    "character value in \\x{...} sequence is too large\0"    "character value in \\x{...} sequence is too large\0"
444    /* 35 */    /* 35 */
# Line 395  static const char error_texts[] = Line 461  static const char error_texts[] =
461    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
462    /* 50 */    /* 50 */
463    "repeated subpattern is too long\0"    /** DEAD **/    "repeated subpattern is too long\0"    /** DEAD **/
464    "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"
465    "internal error: overran compiling workspace\0"    "internal error: overran compiling workspace\0"
466    "internal error: previously-checked referenced subpattern not found\0"    "internal error: previously-checked referenced subpattern not found\0"
467    "DEFINE group contains more than one branch\0"    "DEFINE group contains more than one branch\0"
# Line 414  static const char error_texts[] = Line 480  static const char error_texts[] =
480    /* 65 */    /* 65 */
481    "different names for subpatterns of the same number are not allowed\0"    "different names for subpatterns of the same number are not allowed\0"
482    "(*MARK) must have an argument\0"    "(*MARK) must have an argument\0"
483    "this version of PCRE is not compiled with PCRE_UCP support\0"    "this version of PCRE is not compiled with Unicode property support\0"
484    "\\c must be followed by an ASCII character\0"    "\\c must be followed by an ASCII character\0"
485    "\\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"
486    /* 70 */    /* 70 */
487    "internal error: unknown opcode in find_fixedlength()\0"    "internal error: unknown opcode in find_fixedlength()\0"
488    "\\N is not supported in a class\0"    "\\N is not supported in a class\0"
489    "too many forward references\0"    "too many forward references\0"
490      "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
491      "invalid UTF-16 string\0"
492    ;    ;
493    
494  /* 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 439  For convenience, we use the same bit def Line 507  For convenience, we use the same bit def
507    
508  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
509    
510    /* Using a simple comparison for decimal numbers rather than a memory read
511    is much faster, and the resulting code is simpler (the compiler turns it
512    into a subtraction and unsigned comparison). */
513    
514    #define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9)
515    
516  #ifndef EBCDIC  #ifndef EBCDIC
517    
518  /* 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
519  UTF-8 mode. */  UTF-8 mode. */
520    
521  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
522    {    {
523    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
524    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */
# Line 483  static const unsigned char digitab[] = Line 557  static const unsigned char digitab[] =
557    
558  /* 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. */
559    
560  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
561    {    {
562    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
563    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */
# Line 518  static const unsigned char digitab[] = Line 592  static const unsigned char digitab[] =
592    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */
593    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */
594    
595  static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */  static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */
596    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */
597    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */
598    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */
# Line 557  static const unsigned char ebcdic_charta Line 631  static const unsigned char ebcdic_charta
631  /* Definition to allow mutual recursion */  /* Definition to allow mutual recursion */
632    
633  static BOOL  static BOOL
634    compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int,    compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,
635      int *, int *, branch_chain *, compile_data *, int *);      int *, int *, branch_chain *, compile_data *, int *);
636    
637    
# Line 592  return s; Line 666  return s;
666  *           Expand the workspace                 *  *           Expand the workspace                 *
667  *************************************************/  *************************************************/
668    
669  /* This function is called during the second compiling phase, if the number of  /* This function is called during the second compiling phase, if the number of
670  forward references fills the existing workspace, which is originally a block on  forward references fills the existing workspace, which is originally a block on
671  the stack. A larger block is obtained from malloc() unless the ultimate limit  the stack. A larger block is obtained from malloc() unless the ultimate limit
672  has been reached or the increase will be rather small.  has been reached or the increase will be rather small.
673    
674  Argument: pointer to the compile data block  Argument: pointer to the compile data block
# Line 604  Returns:  0 if all went well, else an er Line 678  Returns:  0 if all went well, else an er
678  static int  static int
679  expand_workspace(compile_data *cd)  expand_workspace(compile_data *cd)
680  {  {
681  uschar *newspace;  pcre_uchar *newspace;
682  int newsize = cd->workspace_size * 2;  int newsize = cd->workspace_size * 2;
683    
684  if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;  if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;
# Line 612  if (cd->workspace_size >= COMPILE_WORK_S Line 686  if (cd->workspace_size >= COMPILE_WORK_S
686      newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)      newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)
687   return ERR72;   return ERR72;
688    
689  newspace = (pcre_malloc)(newsize);  newspace = (PUBL(malloc))(IN_UCHARS(newsize));
690  if (newspace == NULL) return ERR21;  if (newspace == NULL) return ERR21;
691    memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar));
692  memcpy(newspace, cd->start_workspace, cd->workspace_size);  cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace);
693  cd->hwm = (uschar *)newspace + (cd->hwm - cd->start_workspace);  if (cd->workspace_size > COMPILE_WORK_SIZE)
694  if (cd->workspace_size > COMPILE_WORK_SIZE)    (PUBL(free))((void *)cd->start_workspace);
   (pcre_free)((void *)cd->start_workspace);  
695  cd->start_workspace = newspace;  cd->start_workspace = newspace;
696  cd->workspace_size = newsize;  cd->workspace_size = newsize;
697  return 0;  return 0;
# Line 642  Returns:    TRUE or FALSE Line 715  Returns:    TRUE or FALSE
715  */  */
716    
717  static BOOL  static BOOL
718  is_counted_repeat(const uschar *p)  is_counted_repeat(const pcre_uchar *p)
719  {  {
720  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
721  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
722    while (IS_DIGIT(*p)) p++;
723  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
724    
725  if (*p++ != CHAR_COMMA) return FALSE;  if (*p++ != CHAR_COMMA) return FALSE;
726  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
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    
732  return (*p == CHAR_RIGHT_CURLY_BRACKET);  return (*p == CHAR_RIGHT_CURLY_BRACKET);
733  }  }
# Line 684  Returns:         zero or positive => a d Line 759  Returns:         zero or positive => a d
759  */  */
760    
761  static int  static int
762  check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, int *errorcodeptr, int bracount,
763    int options, BOOL isclass)    int options, BOOL isclass)
764  {  {
765  BOOL utf8 = (options & PCRE_UTF8) != 0;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
766  const uschar *ptr = *ptrptr + 1;  BOOL utf = (options & PCRE_UTF8) != 0;
767  int c, i;  const pcre_uchar *ptr = *ptrptr + 1;
768    pcre_int32 c;
769    int i;
770    
771  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
772  ptr--;                            /* Set pointer back to the last byte */  ptr--;                            /* Set pointer back to the last byte */
# Line 703  in a table. A non-zero result is somethi Line 780  in a table. A non-zero result is somethi
780  Otherwise further processing may be required. */  Otherwise further processing may be required. */
781    
782  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
783  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */  /* Not alphanumeric */
784    else if (c < CHAR_0 || c > CHAR_z) {}
785  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if ((i = escapes[c - CHAR_0]) != 0) c = i;
786    
787  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
788  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */  /* Not alphanumeric */
789    else if (c < 'a' || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
790  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if ((i = escapes[c - 0x48]) != 0)  c = i;
791  #endif  #endif
792    
# Line 715  else if ((i = escapes[c - 0x48]) != 0) Line 794  else if ((i = escapes[c - 0x48]) != 0)
794    
795  else  else
796    {    {
797    const uschar *oldptr;    const pcre_uchar *oldptr;
798    BOOL braced, negated;    BOOL braced, negated;
799    
800    switch (c)    switch (c)
# Line 733  else Line 812  else
812        {        {
813        /* In JavaScript, \u must be followed by four hexadecimal numbers.        /* In JavaScript, \u must be followed by four hexadecimal numbers.
814        Otherwise it is a lowercase u letter. */        Otherwise it is a lowercase u letter. */
815        if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
816             && (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0)          && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
817            && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
818            && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
819          {          {
820          c = 0;          c = 0;
821          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
# Line 788  else Line 869  else
869    
870      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
871        {        {
872        const uschar *p;        const pcre_uchar *p;
873        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
874          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
875        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
876          {          {
877          c = -ESC_k;          c = -ESC_k;
# Line 808  else Line 889  else
889        }        }
890      else negated = FALSE;      else negated = FALSE;
891    
892        /* The integer range is limited by the machine's int representation. */
893      c = 0;      c = 0;
894      while ((digitab[ptr[1]] & ctype_digit) != 0)      while (IS_DIGIT(ptr[1]))
895          {
896          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */
897            {
898            c = -1;
899            break;
900            }
901        c = c * 10 + *(++ptr) - CHAR_0;        c = c * 10 + *(++ptr) - CHAR_0;
902          }
903      if (c < 0)   /* Integer overflow */      if (((unsigned int)c) > INT_MAX) /* Integer overflow */
904        {        {
905          while (IS_DIGIT(ptr[1]))
906            ptr++;
907        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
908        break;        break;
909        }        }
# Line 861  else Line 951  else
951      if (!isclass)      if (!isclass)
952        {        {
953        oldptr = ptr;        oldptr = ptr;
954          /* The integer range is limited by the machine's int representation. */
955        c -= CHAR_0;        c -= CHAR_0;
956        while ((digitab[ptr[1]] & ctype_digit) != 0)        while (IS_DIGIT(ptr[1]))
957            {
958            if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */
959              {
960              c = -1;
961              break;
962              }
963          c = c * 10 + *(++ptr) - CHAR_0;          c = c * 10 + *(++ptr) - CHAR_0;
964        if (c < 0)    /* Integer overflow */          }
965          if (((unsigned int)c) > INT_MAX) /* Integer overflow */
966          {          {
967            while (IS_DIGIT(ptr[1]))
968              ptr++;
969          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
970          break;          break;
971          }          }
# Line 891  else Line 991  else
991      /* \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
992      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
993      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
994      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,
995      than 3 octal digits. */      but no more than 3 octal digits. */
996    
997      case CHAR_0:      case CHAR_0:
998      c -= CHAR_0;      c -= CHAR_0;
999      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1000          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1001      if (!utf8 && c > 255) *errorcodeptr = ERR51;  #ifdef COMPILE_PCRE8
1002        if (!utf && c > 0xff) *errorcodeptr = ERR51;
1003    #endif
1004      break;      break;
1005    
1006      /* \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
1007      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.
1008      treated as a data character. */      If not, { is treated as a data character. */
1009    
1010      case CHAR_x:      case CHAR_x:
1011      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1012        {        {
1013        /* In JavaScript, \x must be followed by two hexadecimal numbers.        /* In JavaScript, \x must be followed by two hexadecimal numbers.
1014        Otherwise it is a lowercase x letter. */        Otherwise it is a lowercase x letter. */
1015        if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0)        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1016            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1017          {          {
1018          c = 0;          c = 0;
1019          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
# Line 930  else Line 1033  else
1033    
1034      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1035        {        {
1036        const uschar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
       int count = 0;  
1037    
1038        c = 0;        c = 0;
1039        while ((digitab[*pt] & ctype_xdigit) != 0)        while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1040          {          {
1041          register int cc = *pt++;          register int cc = *pt++;
1042          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
         count++;  
1043    
1044  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1045          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
# Line 947  else Line 1048  else
1048          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 */
1049          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1050  #endif  #endif
1051    
1052    #ifdef COMPILE_PCRE8
1053            if (c > (utf ? 0x10ffff : 0xff)) { c = -1; break; }
1054    #else
1055    #ifdef COMPILE_PCRE16
1056            if (c > (utf ? 0x10ffff : 0xffff)) { c = -1; break; }
1057    #endif
1058    #endif
1059            }
1060    
1061          if (c < 0)
1062            {
1063            while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1064            *errorcodeptr = ERR34;
1065          }          }
1066    
1067        if (*pt == CHAR_RIGHT_CURLY_BRACKET)        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
1068          {          {
1069          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1070          ptr = pt;          ptr = pt;
1071          break;          break;
1072          }          }
# Line 963  else Line 1078  else
1078      /* Read just a single-byte hex-defined char */      /* Read just a single-byte hex-defined char */
1079    
1080      c = 0;      c = 0;
1081      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1082        {        {
1083        int cc;                                  /* Some compilers don't like */        int cc;                                  /* Some compilers don't like */
1084        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
# Line 1061  Returns:         type value from ucp_typ Line 1176  Returns:         type value from ucp_typ
1176  */  */
1177    
1178  static int  static int
1179  get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)
1180  {  {
1181  int c, i, bot, top;  int c, i, bot, top;
1182  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1183  char name[32];  pcre_uchar name[32];
1184    
1185  c = *(++ptr);  c = *(++ptr);
1186  if (c == 0) goto ERROR_RETURN;  if (c == 0) goto ERROR_RETURN;
# Line 1082  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1197  if (c == CHAR_LEFT_CURLY_BRACKET)
1197      *negptr = TRUE;      *negptr = TRUE;
1198      ptr++;      ptr++;
1199      }      }
1200    for (i = 0; i < (int)sizeof(name) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1201      {      {
1202      c = *(++ptr);      c = *(++ptr);
1203      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 1106  else Line 1221  else
1221  /* Search for a recognized property name using binary chop */  /* Search for a recognized property name using binary chop */
1222    
1223  bot = 0;  bot = 0;
1224  top = _pcre_utt_size;  top = PRIV(utt_size);
1225    
1226  while (bot < top)  while (bot < top)
1227    {    {
1228    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1229    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);    c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1230    if (c == 0)    if (c == 0)
1231      {      {
1232      *dptr = _pcre_utt[i].value;      *dptr = PRIV(utt)[i].value;
1233      return _pcre_utt[i].type;      return PRIV(utt)[i].type;
1234      }      }
1235    if (c > 0) bot = i + 1; else top = i;    if (c > 0) bot = i + 1; else top = i;
1236    }    }
# Line 1153  Returns:         pointer to '}' on succe Line 1268  Returns:         pointer to '}' on succe
1268                   current ptr on error, with errorcodeptr set non-zero                   current ptr on error, with errorcodeptr set non-zero
1269  */  */
1270    
1271  static const uschar *  static const pcre_uchar *
1272  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)
1273  {  {
1274  int min = 0;  int min = 0;
1275  int max = -1;  int max = -1;
# Line 1162  int max = -1; Line 1277  int max = -1;
1277  /* 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
1278  an integer overflow. */  an integer overflow. */
1279    
1280  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;
1281  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1282    {    {
1283    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1177  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1292  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1292    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1293      {      {
1294      max = 0;      max = 0;
1295      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;
1296      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1297        {        {
1298        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1232  Arguments: Line 1347  Arguments:
1347    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1348    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1349    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1350    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 mode
1351    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1352    
1353  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1354  */  */
1355    
1356  static int  static int
1357  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,
1358    BOOL xmode, BOOL utf8, int *count)    BOOL xmode, BOOL utf, int *count)
1359  {  {
1360  uschar *ptr = *ptrptr;  pcre_uchar *ptr = *ptrptr;
1361  int start_count = *count;  int start_count = *count;
1362  int hwm_count = start_count;  int hwm_count = start_count;
1363  BOOL dup_parens = FALSE;  BOOL dup_parens = FALSE;
# Line 1309  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1424  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1424          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1425        {        {
1426        int term;        int term;
1427        const uschar *thisname;        const pcre_uchar *thisname;
1428        *count += 1;        *count += 1;
1429        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1430        term = *ptr++;        term = *ptr++;
# Line 1317  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1432  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1432        thisname = ptr;        thisname = ptr;
1433        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1434        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == ptr - thisname &&
1435            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, lorn) == 0)
1436          return *count;          return *count;
1437        term++;        term++;
1438        }        }
# Line 1360  for (; ptr < cd->end_pattern; ptr++) Line 1475  for (; ptr < cd->end_pattern; ptr++)
1475          {          {
1476          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1477            ptr+= 2;            ptr+= 2;
1478          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1479                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1480            ptr += 4;            ptr += 4;
1481          else          else
# Line 1408  for (; ptr < cd->end_pattern; ptr++) Line 1523  for (; ptr < cd->end_pattern; ptr++)
1523        {        {
1524        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1525        ptr++;        ptr++;
1526  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1527        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1528  #endif  #endif
1529        }        }
1530      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
# Line 1420  for (; ptr < cd->end_pattern; ptr++) Line 1535  for (; ptr < cd->end_pattern; ptr++)
1535    
1536    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1537      {      {
1538      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1539      if (rc > 0) return rc;      if (rc > 0) return rc;
1540      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1541      }      }
# Line 1466  Arguments: Line 1581  Arguments:
1581    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1582    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1583    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1584    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 mode
1585    
1586  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1587  */  */
1588    
1589  static int  static int
1590  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,
1591    BOOL utf8)    BOOL utf)
1592  {  {
1593  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1594  int count = 0;  int count = 0;
1595  int rc;  int rc;
1596    
# Line 1486  matching closing parens. That is why we Line 1601  matching closing parens. That is why we
1601    
1602  for (;;)  for (;;)
1603    {    {
1604    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1605    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == 0) break;
1606    }    }
1607    
# Line 1513  Arguments: Line 1628  Arguments:
1628  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1629  */  */
1630    
1631  static const uschar*  static const pcre_uchar*
1632  first_significant_code(const uschar *code, BOOL skipassert)  first_significant_code(const pcre_uchar *code, BOOL skipassert)
1633  {  {
1634  for (;;)  for (;;)
1635    {    {
# Line 1525  for (;;) Line 1640  for (;;)
1640      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1641      if (!skipassert) return code;      if (!skipassert) return code;
1642      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1643      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1644      break;      break;
1645    
1646      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1539  for (;;) Line 1654  for (;;)
1654      case OP_RREF:      case OP_RREF:
1655      case OP_NRREF:      case OP_NRREF:
1656      case OP_DEF:      case OP_DEF:
1657      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1658      break;      break;
1659    
1660      default:      default:
# Line 1569  and doing the check at the end; a flag s Line 1684  and doing the check at the end; a flag s
1684    
1685  Arguments:  Arguments:
1686    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1687    utf8     TRUE in UTF-8 mode    utf      TRUE in UTF-8 / UTF-16 mode
1688    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1689    cd       the "compile data" structure    cd       the "compile data" structure
1690    
# Line 1581  Returns:   the fixed length, Line 1696  Returns:   the fixed length,
1696  */  */
1697    
1698  static int  static int
1699  find_fixedlength(uschar *code, BOOL utf8, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1700  {  {
1701  int length = -1;  int length = -1;
1702    
1703  register int branchlength = 0;  register int branchlength = 0;
1704  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1705    
1706  /* 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
1707  branch, check the length against that of the other branches. */  branch, check the length against that of the other branches. */
# Line 1594  branch, check the length against that of Line 1709  branch, check the length against that of
1709  for (;;)  for (;;)
1710    {    {
1711    int d;    int d;
1712    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1713    register int op = *cc;    register int op = *cc;
1714    
1715    switch (op)    switch (op)
1716      {      {
1717      /* We only need to continue for OP_CBRA (normal capturing bracket) and      /* We only need to continue for OP_CBRA (normal capturing bracket) and
# Line 1608  for (;;) Line 1724  for (;;)
1724      case OP_ONCE:      case OP_ONCE:
1725      case OP_ONCE_NC:      case OP_ONCE_NC:
1726      case OP_COND:      case OP_COND:
1727      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);
1728      if (d < 0) return d;      if (d < 0) return d;
1729      branchlength += d;      branchlength += d;
1730      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
# Line 1639  for (;;) Line 1755  for (;;)
1755    
1756      case OP_RECURSE:      case OP_RECURSE:
1757      if (!atend) return -3;      if (!atend) return -3;
1758      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1759      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1760      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1761      d = find_fixedlength(cs + 2, utf8, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1762      if (d < 0) return d;      if (d < 0) return d;
1763      branchlength += d;      branchlength += d;
1764      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1655  for (;;) Line 1771  for (;;)
1771      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1772      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1773      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1774      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1775        break;
1776    
1777      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1778    
# Line 1663  for (;;) Line 1780  for (;;)
1780      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
1781      case OP_SKIP_ARG:      case OP_SKIP_ARG:
1782      case OP_THEN_ARG:      case OP_THEN_ARG:
1783      cc += cc[1] + _pcre_OP_lengths[*cc];      cc += cc[1] + PRIV(OP_lengths)[*cc];
1784      break;      break;
1785    
1786      case OP_CALLOUT:      case OP_CALLOUT:
# Line 1690  for (;;) Line 1807  for (;;)
1807      case OP_SOM:      case OP_SOM:
1808      case OP_THEN:      case OP_THEN:
1809      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1810      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1811      break;      break;
1812    
1813      /* Handle literal characters */      /* Handle literal characters */
# Line 1701  for (;;) Line 1818  for (;;)
1818      case OP_NOTI:      case OP_NOTI:
1819      branchlength++;      branchlength++;
1820      cc += 2;      cc += 2;
1821  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1822      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1823  #endif  #endif
1824      break;      break;
1825    
# Line 1714  for (;;) Line 1831  for (;;)
1831      case OP_NOTEXACT:      case OP_NOTEXACT:
1832      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1833      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1834      cc += 4;      cc += 2 + IMM2_SIZE;
1835  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1836      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1837  #endif  #endif
1838      break;      break;
1839    
1840      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1841      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1842      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP) cc += 2;
1843      cc += 4;      cc += 1 + IMM2_SIZE + 1;
1844      break;      break;
1845    
1846      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1749  for (;;) Line 1866  for (;;)
1866      cc++;      cc++;
1867      break;      break;
1868    
1869      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;
1870      otherwise \C is coded as OP_ALLANY. */      otherwise \C is coded as OP_ALLANY. */
1871    
1872      case OP_ANYBYTE:      case OP_ANYBYTE:
# Line 1757  for (;;) Line 1874  for (;;)
1874    
1875      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1876    
1877  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || defined COMPILE_PCRE16
1878      case OP_XCLASS:      case OP_XCLASS:
1879      cc += GET(cc, 1) - 33;      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
1880      /* Fall through */      /* Fall through */
1881  #endif  #endif
1882    
1883      case OP_CLASS:      case OP_CLASS:
1884      case OP_NCLASS:      case OP_NCLASS:
1885      cc += 33;      cc += PRIV(OP_lengths)[OP_CLASS];
1886    
1887      switch (*cc)      switch (*cc)
1888        {        {
# Line 1779  for (;;) Line 1896  for (;;)
1896    
1897        case OP_CRRANGE:        case OP_CRRANGE:
1898        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1899        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1900        branchlength += GET2(cc,1);        branchlength += GET2(cc,1);
1901        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1902        break;        break;
1903    
1904        default:        default:
# Line 1896  length. Line 2013  length.
2013    
2014  Arguments:  Arguments:
2015    code        points to start of expression    code        points to start of expression
2016    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 mode
2017    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2018    
2019  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
2020  */  */
2021    
2022  const uschar *  const pcre_uchar *
2023  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2024  {  {
2025  for (;;)  for (;;)
2026    {    {
# Line 1921  for (;;) Line 2038  for (;;)
2038    
2039    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2040      {      {
2041      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2042      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2043      }      }
2044    
2045    /* Handle capturing bracket */    /* Handle capturing bracket */
# Line 1931  for (;;) Line 2048  for (;;)
2048             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2049      {      {
2050      int n = GET2(code, 1+LINK_SIZE);      int n = GET2(code, 1+LINK_SIZE);
2051      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2052      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2053      }      }
2054    
2055    /* 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 1960  for (;;) Line 2077  for (;;)
2077        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2078        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2079        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2080        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP
2081            || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;
2082        break;        break;
2083    
2084        case OP_MARK:        case OP_MARK:
# Line 1976  for (;;) Line 2094  for (;;)
2094    
2095      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2096    
2097      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2098    
2099    /* 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
2100    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
2101    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2102    
2103  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2104      if (utf8) switch(c)      if (utf) switch(c)
2105        {        {
2106        case OP_CHAR:        case OP_CHAR:
2107        case OP_CHARI:        case OP_CHARI:
# Line 2013  for (;;) Line 2131  for (;;)
2131        case OP_MINQUERYI:        case OP_MINQUERYI:
2132        case OP_POSQUERY:        case OP_POSQUERY:
2133        case OP_POSQUERYI:        case OP_POSQUERYI:
2134        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2135        break;        break;
2136        }        }
2137  #else  #else
2138      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2139  #endif  #endif
2140      }      }
2141    }    }
# Line 2034  instance of OP_RECURSE. Line 2152  instance of OP_RECURSE.
2152    
2153  Arguments:  Arguments:
2154    code        points to start of expression    code        points to start of expression
2155    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 mode
2156    
2157  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
2158  */  */
2159    
2160  static const uschar *  static const pcre_uchar *
2161  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2162  {  {
2163  for (;;)  for (;;)
2164    {    {
# Line 2079  for (;;) Line 2197  for (;;)
2197        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2198        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2199        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2200        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP
2201            || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;
2202        break;        break;
2203    
2204        case OP_MARK:        case OP_MARK:
# Line 2095  for (;;) Line 2214  for (;;)
2214    
2215      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2216    
2217      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2218    
2219      /* 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
2220      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
2221      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2222    
2223  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2224      if (utf8) switch(c)      if (utf) switch(c)
2225        {        {
2226        case OP_CHAR:        case OP_CHAR:
2227        case OP_CHARI:        case OP_CHARI:
# Line 2132  for (;;) Line 2251  for (;;)
2251        case OP_MINQUERYI:        case OP_MINQUERYI:
2252        case OP_POSQUERY:        case OP_POSQUERY:
2253        case OP_POSQUERYI:        case OP_POSQUERYI:
2254        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2255        break;        break;
2256        }        }
2257  #else  #else
2258      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2259  #endif  #endif
2260      }      }
2261    }    }
# Line 2159  bracket whose current branch will alread Line 2278  bracket whose current branch will alread
2278  Arguments:  Arguments:
2279    code        points to start of search    code        points to start of search
2280    endcode     points to where to stop    endcode     points to where to stop
2281    utf8        TRUE if in UTF8 mode    utf         TRUE if in UTF-8 / UTF-16 mode
2282    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2283    
2284  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2285  */  */
2286    
2287  static BOOL  static BOOL
2288  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2289    compile_data *cd)    BOOL utf, compile_data *cd)
2290  {  {
2291  register int c;  register int c;
2292  for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2293       code < endcode;       code < endcode;
2294       code = first_significant_code(code + _pcre_OP_lengths[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2295    {    {
2296    const uschar *ccode;    const pcre_uchar *ccode;
2297    
2298    c = *code;    c = *code;
2299    
# Line 2197  for (code = first_significant_code(code Line 2316  for (code = first_significant_code(code
2316    
2317    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2318      {      {
2319      const uschar *scode;      const pcre_uchar *scode;
2320      BOOL empty_branch;      BOOL empty_branch;
2321    
2322      /* Test for forward reference */      /* Test for forward reference */
# Line 2215  for (code = first_significant_code(code Line 2334  for (code = first_significant_code(code
2334    
2335      do      do
2336        {        {
2337        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd))
2338          {          {
2339          empty_branch = TRUE;          empty_branch = TRUE;
2340          break;          break;
# Line 2233  for (code = first_significant_code(code Line 2352  for (code = first_significant_code(code
2352    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2353        c == OP_BRAPOSZERO)        c == OP_BRAPOSZERO)
2354      {      {
2355      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2356      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
2357      c = *code;      c = *code;
2358      continue;      continue;
# Line 2271  for (code = first_significant_code(code Line 2390  for (code = first_significant_code(code
2390        empty_branch = FALSE;        empty_branch = FALSE;
2391        do        do
2392          {          {
2393          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
2394            empty_branch = TRUE;            empty_branch = TRUE;
2395          code += GET(code, 1);          code += GET(code, 1);
2396          }          }
# Line 2289  for (code = first_significant_code(code Line 2408  for (code = first_significant_code(code
2408      {      {
2409      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2410      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2411      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2412      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"
2413      here. */      here. */
2414    
2415  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2416      case OP_XCLASS:      case OP_XCLASS:
2417      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2418      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2301  for (code = first_significant_code(code Line 2420  for (code = first_significant_code(code
2420    
2421      case OP_CLASS:      case OP_CLASS:
2422      case OP_NCLASS:      case OP_NCLASS:
2423      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2424    
2425  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2426      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2427  #endif  #endif
2428    
# Line 2376  for (code = first_significant_code(code Line 2495  for (code = first_significant_code(code
2495      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2496      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2497      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2498      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP
2499          || code[1 + IMM2_SIZE] == OP_NOTPROP) code += 2;
2500      break;      break;
2501    
2502      /* End of branch */      /* End of branch */
# Line 2391  for (code = first_significant_code(code Line 2511  for (code = first_significant_code(code
2511      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2512      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2513    
2514  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2515      case OP_STAR:      case OP_STAR:
2516      case OP_STARI:      case OP_STARI:
2517      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2404  for (code = first_significant_code(code Line 2524  for (code = first_significant_code(code
2524      case OP_MINQUERYI:      case OP_MINQUERYI:
2525      case OP_POSQUERY:      case OP_POSQUERY:
2526      case OP_POSQUERYI:      case OP_POSQUERYI:
2527      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2528      break;      break;
2529    
2530      case OP_UPTO:      case OP_UPTO:
# Line 2413  for (code = first_significant_code(code Line 2533  for (code = first_significant_code(code
2533      case OP_MINUPTOI:      case OP_MINUPTOI:
2534      case OP_POSUPTO:      case OP_POSUPTO:
2535      case OP_POSUPTOI:      case OP_POSUPTOI:
2536      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]);
2537      break;      break;
2538  #endif  #endif
2539    
# Line 2457  Arguments: Line 2577  Arguments:
2577    code        points to start of the recursion    code        points to start of the recursion
2578    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2579    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2580    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 mode
2581    cd          pointers to tables etc    cd          pointers to tables etc
2582    
2583  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2584  */  */
2585    
2586  static BOOL  static BOOL
2587  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2588    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2589  {  {
2590  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2591    {    {
2592    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
2593      return FALSE;      return FALSE;
2594    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2595    }    }
# Line 2521  Returns:   TRUE or FALSE Line 2641  Returns:   TRUE or FALSE
2641  */  */
2642    
2643  static BOOL  static BOOL
2644  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2645  {  {
2646  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
2647  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
# Line 2565  Returns:     a value representing the na Line 2685  Returns:     a value representing the na
2685  */  */
2686    
2687  static int  static int
2688  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2689  {  {
2690  const char *pn = posix_names;  const char *pn = posix_names;
2691  register int yield = 0;  register int yield = 0;
2692  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2693    {    {
2694    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2695      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;
2696    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2697    yield++;    yield++;
2698    }    }
# Line 2604  value in the reference (which is a group Line 2724  value in the reference (which is a group
2724  Arguments:  Arguments:
2725    group      points to the start of the group    group      points to the start of the group
2726    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2727    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 mode
2728    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2729    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
2730    
# Line 2612  Returns:     nothing Line 2732  Returns:     nothing
2732  */  */
2733    
2734  static void  static void
2735  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2736    uschar *save_hwm)    pcre_uchar *save_hwm)
2737  {  {
2738  uschar *ptr = group;  pcre_uchar *ptr = group;
2739    
2740  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2741    {    {
2742    int offset;    int offset;
2743    uschar *hc;    pcre_uchar *hc;
2744    
2745    /* 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
2746    reference. */    reference. */
# Line 2665  Arguments: Line 2785  Arguments:
2785  Returns:         new code pointer  Returns:         new code pointer
2786  */  */
2787    
2788  static uschar *  static pcre_uchar *
2789  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2790  {  {
2791  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2792  *code++ = 255;  *code++ = 255;
2793  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2794  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2795  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2796  }  }
2797    
2798    
# Line 2694  Returns:             nothing Line 2814  Returns:             nothing
2814  */  */
2815    
2816  static void  static void
2817  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2818  {  {
2819  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2820  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2777  switch(ptype) Line 2897  switch(ptype)
2897            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
2898    
2899    case PT_GC:    case PT_GC:
2900    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2901    
2902    case PT_PC:    case PT_PC:
2903    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2788  switch(ptype) Line 2908  switch(ptype)
2908    /* These are specials */    /* These are specials */
2909    
2910    case PT_ALNUM:    case PT_ALNUM:
2911    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2912            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
2913    
2914    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
2915    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2916            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2917            == negated;            == negated;
2918    
2919    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
2920    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2921            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
2922            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
2923            == negated;            == negated;
2924    
2925    case PT_WORD:    case PT_WORD:
2926    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2927            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
2928            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
2929    }    }
2930  return FALSE;  return FALSE;
# Line 2823  sense to automatically possessify the re Line 2943  sense to automatically possessify the re
2943    
2944  Arguments:  Arguments:
2945    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
2946    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 mode
2947    ptr           next character in pattern    ptr           next character in pattern
2948    options       options bits    options       options bits
2949    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2832  Returns:        TRUE if possessifying is Line 2952  Returns:        TRUE if possessifying is
2952  */  */
2953    
2954  static BOOL  static BOOL
2955  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
2956    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
2957  {  {
2958  int c, next;  pcre_int32 c, next;
2959  int op_code = *previous++;  int op_code = *previous++;
2960    
2961  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2844  if ((options & PCRE_EXTENDED) != 0) Line 2964  if ((options & PCRE_EXTENDED) != 0)
2964    {    {
2965    for (;;)    for (;;)
2966      {      {
2967      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
2968      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
2969        {        {
2970        ptr++;        ptr++;
# Line 2852  if ((options & PCRE_EXTENDED) != 0) Line 2972  if ((options & PCRE_EXTENDED) != 0)
2972          {          {
2973          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
2974          ptr++;          ptr++;
2975  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2976          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
2977  #endif  #endif
2978          }          }
2979        }        }
# Line 2871  if (*ptr == CHAR_BACKSLASH) Line 2991  if (*ptr == CHAR_BACKSLASH)
2991    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
2992    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
2993    }    }
2994    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
2995    {    {
2996  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
2997    if (utf8) { GETCHARINC(next, ptr); } else    if (utf) { GETCHARINC(next, ptr); } else
2998  #endif  #endif
2999    next = *ptr++;    next = *ptr++;
3000    }    }
   
3001  else return FALSE;  else return FALSE;
3002    
3003  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2888  if ((options & PCRE_EXTENDED) != 0) Line 3006  if ((options & PCRE_EXTENDED) != 0)
3006    {    {
3007    for (;;)    for (;;)
3008      {      {
3009      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3010      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3011        {        {
3012        ptr++;        ptr++;
# Line 2896  if ((options & PCRE_EXTENDED) != 0) Line 3014  if ((options & PCRE_EXTENDED) != 0)
3014          {          {
3015          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3016          ptr++;          ptr++;
3017  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3018          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3019  #endif  #endif
3020          }          }
3021        }        }
# Line 2908  if ((options & PCRE_EXTENDED) != 0) Line 3026  if ((options & PCRE_EXTENDED) != 0)
3026  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3027    
3028  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3029    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)
3030      return FALSE;      return FALSE;
3031    
3032  /* Now compare the next item with the previous opcode. First, handle cases when  /* Now compare the next item with the previous opcode. First, handle cases when
# Line 2917  the next item is a character. */ Line 3035  the next item is a character. */
3035  if (next >= 0) switch(op_code)  if (next >= 0) switch(op_code)
3036    {    {
3037    case OP_CHAR:    case OP_CHAR:
3038  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3039    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3040  #else  #else
3041    c = *previous;    c = *previous;
# Line 2929  if (next >= 0) switch(op_code) Line 3047  if (next >= 0) switch(op_code)
3047    high-valued characters. */    high-valued characters. */
3048    
3049    case OP_CHARI:    case OP_CHARI:
3050  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3051    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3052  #else  #else
3053    c = *previous;    c = *previous;
3054  #endif  #endif
3055    if (c == next) return FALSE;    if (c == next) return FALSE;
3056  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3057    if (utf8)    if (utf)
3058      {      {
3059      unsigned int othercase;      unsigned int othercase;
3060      if (next < 128) othercase = cd->fcc[next]; else      if (next < 128) othercase = cd->fcc[next]; else
# Line 2948  if (next >= 0) switch(op_code) Line 3066  if (next >= 0) switch(op_code)
3066      return (unsigned int)c != othercase;      return (unsigned int)c != othercase;
3067      }      }
3068    else    else
3069  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3070    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */    return (c != TABLE_GET(next, cd->fcc, next));  /* Non-UTF-8 mode */
3071    
3072    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These
3073    opcodes are not used for multi-byte characters, because they are coded using    opcodes are not used for multi-byte characters, because they are coded using
# Line 2960  if (next >= 0) switch(op_code) Line 3078  if (next >= 0) switch(op_code)
3078    
3079    case OP_NOTI:    case OP_NOTI:
3080    if ((c = *previous) == next) return TRUE;    if ((c = *previous) == next) return TRUE;
3081  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3082    if (utf8)    if (utf)
3083      {      {
3084      unsigned int othercase;      unsigned int othercase;
3085      if (next < 128) othercase = cd->fcc[next]; else      if (next < 128) othercase = cd->fcc[next]; else
# Line 2973  if (next >= 0) switch(op_code) Line 3091  if (next >= 0) switch(op_code)
3091      return (unsigned int)c == othercase;      return (unsigned int)c == othercase;
3092      }      }
3093    else    else
3094  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3095    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */    return (c == TABLE_GET(next, cd->fcc, next));  /* Non-UTF-8 mode */
3096    
3097    /* 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.
3098    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. */
# Line 3065  switch(op_code) Line 3183  switch(op_code)
3183    {    {
3184    case OP_CHAR:    case OP_CHAR:
3185    case OP_CHARI:    case OP_CHARI:
3186  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3187    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3188  #else  #else
3189    c = *previous;    c = *previous;
# Line 3170  switch(op_code) Line 3288  switch(op_code)
3288        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. */
3289    
3290        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3291          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)
3292            return FALSE;            return FALSE;
3293    
3294        /* Do the property check. */        /* Do the property check. */
# Line 3248  Arguments: Line 3366  Arguments:
3366    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3367    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3368    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3369    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)
3370    reqbyteptr     set to the last literal character required, else < 0    reqcharptr     set to the last literal character required, else < 0
3371    bcptr          points to current branch chain    bcptr          points to current branch chain
3372    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3373    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3261  Returns:         TRUE on success Line 3379  Returns:         TRUE on success
3379  */  */
3380    
3381  static BOOL  static BOOL
3382  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3383    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,
3384    int cond_depth, compile_data *cd, int *lengthptr)    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,
3385      compile_data *cd, int *lengthptr)
3386  {  {
3387  int repeat_type, op_type;  int repeat_type, op_type;
3388  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3389  int bravalue = 0;  int bravalue = 0;
3390  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3391  int firstbyte, reqbyte;  pcre_int32 firstchar, reqchar;
3392  int zeroreqbyte, zerofirstbyte;  pcre_int32 zeroreqchar, zerofirstchar;
3393  int req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3394  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3395  int after_manual_callout = 0;  int after_manual_callout = 0;
3396  int length_prevgroup = 0;  int length_prevgroup = 0;
3397  register int c;  register int c;
3398  register uschar *code = *codeptr;  register pcre_uchar *code = *codeptr;
3399  uschar *last_code = code;  pcre_uchar *last_code = code;
3400  uschar *orig_code = code;  pcre_uchar *orig_code = code;
3401  uschar *tempcode;  pcre_uchar *tempcode;
3402  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3403  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3404  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3405  const uschar *tempptr;  const pcre_uchar *tempptr;
3406  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3407  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3408  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3409  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3410  uschar classbits[32];  pcre_uint8 classbits[32];
3411    
3412  /* 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
3413  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
3414  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3415    
3416  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3417  BOOL class_utf8;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
3418  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3419  uschar *class_utf8data;  pcre_uchar utf_chars[6];
 uschar *class_utf8data_base;  
 uschar utf8_char[6];  
3420  #else  #else
3421  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3422    #endif
3423    
3424    /* Helper variables for OP_XCLASS opcode (for characters > 255). */
3425    
3426    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3427    BOOL xclass;
3428    pcre_uchar *class_uchardata;
3429    pcre_uchar *class_uchardata_base;
3430  #endif  #endif
3431    
3432  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3315  greedy_non_default = greedy_default ^ 1; Line 3440  greedy_non_default = greedy_default ^ 1;
3440    
3441  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3442  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
3443  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
3444  find one.  find one.
3445    
3446  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
3447  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
3448  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3449  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3450    
3451  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;
3452    
3453  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3454  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
3455  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
3456  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3457    value. This is used only for ASCII characters. */
3458    
3459  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3460    
3461  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3462    
# Line 3342  for (;; ptr++) Line 3468  for (;; ptr++)
3468    BOOL is_quantifier;    BOOL is_quantifier;
3469    BOOL is_recurse;    BOOL is_recurse;
3470    BOOL reset_bracount;    BOOL reset_bracount;
3471    int class_charcount;    int class_has_8bitchar;
3472    int class_lastchar;    int class_single_char;
3473    int newoptions;    int newoptions;
3474    int recno;    int recno;
3475    int refsign;    int refsign;
3476    int skipbytes;    int skipbytes;
3477    int subreqbyte;    int subreqchar;
3478    int subfirstbyte;    int subfirstchar;
3479    int terminator;    int terminator;
3480    int mclength;    int mclength;
3481    int tempbracount;    int tempbracount;
3482    uschar mcbuffer[8];    pcre_uchar mcbuffer[8];
3483    
3484    /* Get next byte in the pattern */    /* Get next character in the pattern */
3485    
3486    c = *ptr;    c = *ptr;
3487    
# Line 3377  for (;; ptr++) Line 3503  for (;; ptr++)
3503  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3504      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3505  #endif  #endif
3506      if (code > cd->start_workspace + cd->workspace_size -      if (code > cd->start_workspace + cd->workspace_size -
3507          WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */          WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3508        {        {
3509        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
# Line 3401  for (;; ptr++) Line 3527  for (;; ptr++)
3527        }        }
3528    
3529      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3530      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, (int)(code - last_code),      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3531        c));        (int)(code - last_code), c, c));
3532    
3533      /* 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
3534      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 3412  for (;; ptr++) Line 3538  for (;; ptr++)
3538        {        {
3539        if (previous > orig_code)        if (previous > orig_code)
3540          {          {
3541          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3542          code -= previous - orig_code;          code -= previous - orig_code;
3543          previous = orig_code;          previous = orig_code;
3544          }          }
# Line 3428  for (;; ptr++) Line 3554  for (;; ptr++)
3554    /* 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
3555    reference list. */    reference list. */
3556    
3557    else if (cd->hwm > cd->start_workspace + cd->workspace_size -    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3558             WORK_SIZE_SAFETY_MARGIN)             WORK_SIZE_SAFETY_MARGIN)
3559      {      {
3560      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
# Line 3481  for (;; ptr++) Line 3607  for (;; ptr++)
3607    
3608    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
3609      {      {
3610      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
3611      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3612        {        {
3613        ptr++;        ptr++;
# Line 3489  for (;; ptr++) Line 3615  for (;; ptr++)
3615          {          {
3616          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3617          ptr++;          ptr++;
3618  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3619          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3620  #endif  #endif
3621          }          }
3622        if (*ptr != 0) continue;        if (*ptr != 0) continue;
# Line 3514  for (;; ptr++) Line 3640  for (;; ptr++)
3640      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
3641      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3642      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3643      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
3644      *reqbyteptr = reqbyte;      *reqcharptr = reqchar;
3645      *codeptr = code;      *codeptr = code;
3646      *ptrptr = ptr;      *ptrptr = ptr;
3647      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3539  for (;; ptr++) Line 3665  for (;; ptr++)
3665      previous = NULL;      previous = NULL;
3666      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3667        {        {
3668        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3669        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3670        }        }
3671      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3551  for (;; ptr++) Line 3677  for (;; ptr++)
3677      break;      break;
3678    
3679      /* 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
3680      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
3681    
3682      case CHAR_DOT:      case CHAR_DOT:
3683      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3684      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
3685      zeroreqbyte = reqbyte;      zeroreqchar = reqchar;
3686      previous = code;      previous = code;
3687      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
3688      break;      break;
# Line 3611  for (;; ptr++) Line 3737  for (;; ptr++)
3737          {          {
3738          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
3739            ptr++;            ptr++;
3740          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)  
3741            ptr += 3;            ptr += 3;
3742          else          else
3743            break;            break;
# Line 3631  for (;; ptr++) Line 3756  for (;; ptr++)
3756          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
3757        {        {
3758        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
3759        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3760        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
3761        break;        break;
3762        }        }
3763    
# Line 3642  for (;; ptr++) Line 3767  for (;; ptr++)
3767    
3768      should_flip_negation = FALSE;      should_flip_negation = FALSE;
3769    
3770      /* 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.
3771      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
3772      valued UTF-8 characters, we don't yet do any optimization. */      < 256 character. class_single_char will be 1 if the class contains only
3773        a single character. */
3774    
3775      class_charcount = 0;      class_has_8bitchar = 0;
3776      class_lastchar = -1;      class_single_char = 0;
3777    
3778      /* 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
3779      temporary bit of memory, in case the class contains only 1 character (less      temporary bit of memory, in case the class contains only 1 character (less
3780      than 256), because in that case the compiled code doesn't use the bit map.      than 256), because in that case the compiled code doesn't use the bit map.
3781      */      */
3782    
3783      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
3784    
3785  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3786      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;                           /* No chars >= 256 */
3787      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For UTF-8 items */
3788      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* For resetting in pass 1 */
3789  #endif  #endif
3790    
3791      /* 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 3668  for (;; ptr++) Line 3794  for (;; ptr++)
3794    
3795      if (c != 0) do      if (c != 0) do
3796        {        {
3797        const uschar *oldptr;        const pcre_uchar *oldptr;
3798    
3799  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3800        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
3801          {                           /* Braces are required because the */          {                           /* Braces are required because the */
3802          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
3803          }          }
3804    #endif
3805    
3806        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3807          /* In the pre-compile phase, accumulate the length of any extra
3808        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
3809        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
3810        (which is on the stack). */        (which is on the stack). */
3811    
3812        if (lengthptr != NULL)        if (lengthptr != NULL)
3813          {          {
3814          *lengthptr += class_utf8data - class_utf8data_base;          *lengthptr += class_uchardata - class_uchardata_base;
3815          class_utf8data = class_utf8data_base;          class_uchardata = class_uchardata_base;
3816          }          }
   
3817  #endif  #endif
3818    
3819        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3714  for (;; ptr++) Line 3841  for (;; ptr++)
3841          {          {
3842          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
3843          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
3844          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
3845          uschar pbits[32];          pcre_uint8 pbits[32];
3846    
3847          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
3848            {            {
# Line 3770  for (;; ptr++) Line 3897  for (;; ptr++)
3897          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
3898    
3899          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
3900            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
3901    
3902          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
3903    
# Line 3801  for (;; ptr++) Line 3928  for (;; ptr++)
3928            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
3929    
3930          ptr = tempptr + 1;          ptr = tempptr + 1;
3931          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 characters. */
3932            class_has_8bitchar = 1;
3933            /* Every class contains at least two characters. */
3934            class_single_char = 2;
3935          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
3936          }          }
3937    
3938        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
3939        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
3940        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
3941        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
3942        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_single_char bigger
3943        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
3944          as literal characters (by default), or are faulted if
3945        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
3946    
3947        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
# Line 3822  for (;; ptr++) Line 3953  for (;; ptr++)
3953          else if (-c == ESC_N)            /* \N is not supported in a class */          else if (-c == ESC_N)            /* \N is not supported in a class */
3954            {            {
3955            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
3956            goto FAILED;            goto FAILED;
3957            }            }
3958          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (-c == ESC_Q)            /* Handle start of quoted string */
3959            {            {
3960            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3837  for (;; ptr++) Line 3968  for (;; ptr++)
3968    
3969          if (c < 0)          if (c < 0)
3970            {            {
3971            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
3972            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
3973              class_has_8bitchar++;
3974              /* Every class contains at least two characters. */
3975              class_single_char += 2;
3976    
3977            switch (-c)            switch (-c)
3978              {              {
# Line 3851  for (;; ptr++) Line 3985  for (;; ptr++)
3985              case ESC_SU:              case ESC_SU:
3986              nestptr = ptr;              nestptr = ptr;
3987              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */
3988              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
3989              continue;              continue;
3990  #endif  #endif
3991              case ESC_d:              case ESC_d:
# Line 3892  for (;; ptr++) Line 4026  for (;; ptr++)
4026              SETBIT(classbits, 0x09); /* VT */              SETBIT(classbits, 0x09); /* VT */
4027              SETBIT(classbits, 0x20); /* SPACE */              SETBIT(classbits, 0x20); /* SPACE */
4028              SETBIT(classbits, 0xa0); /* NSBP */              SETBIT(classbits, 0xa0); /* NSBP */
4029  #ifdef SUPPORT_UTF8  #ifndef COMPILE_PCRE8
4030              if (utf8)              xclass = TRUE;
4031                *class_uchardata++ = XCL_SINGLE;
4032                *class_uchardata++ = 0x1680;
4033                *class_uchardata++ = XCL_SINGLE;
4034                *class_uchardata++ = 0x180e;
4035                *class_uchardata++ = XCL_RANGE;
4036                *class_uchardata++ = 0x2000;
4037                *class_uchardata++ = 0x200a;
4038                *class_uchardata++ = XCL_SINGLE;
4039                *class_uchardata++ = 0x202f;
4040                *class_uchardata++ = XCL_SINGLE;
4041                *class_uchardata++ = 0x205f;
4042                *class_uchardata++ = XCL_SINGLE;
4043                *class_uchardata++ = 0x3000;
4044    #elif defined SUPPORT_UTF
4045                if (utf)
4046                {                {
4047                class_utf8 = TRUE;                xclass = TRUE;
4048                *class_utf8data++ = XCL_SINGLE;                *class_uchardata++ = XCL_SINGLE;
4049                class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x1680, class_uchardata);
4050                *class_utf8data++ = XCL_SINGLE;                *class_uchardata++ = XCL_SINGLE;
4051                class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x180e, class_uchardata);
4052                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4053                class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x2000, class_uchardata);
4054                class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x200a, class_uchardata);
4055                *class_utf8data++ = XCL_SINGLE;                *class_uchardata++ = XCL_SINGLE;
4056                class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x202f, class_uchardata);
4057                *class_utf8data++ = XCL_SINGLE;                *class_uchardata++ = XCL_SINGLE;
4058                class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x205f, class_uchardata);
4059                *class_utf8data++ = XCL_SINGLE;                *class_uchardata++ = XCL_SINGLE;
4060                class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x3000, class_uchardata);
4061                }                }
4062  #endif  #endif
4063              continue;              continue;
# Line 3926  for (;; ptr++) Line 4075  for (;; ptr++)
4075                  }                  }
4076                classbits[c] |= x;                classbits[c] |= x;
4077                }                }
4078    #ifndef COMPILE_PCRE8
4079  #ifdef SUPPORT_UTF8              xclass = TRUE;
4080              if (utf8)              *class_uchardata++ = XCL_RANGE;
4081                *class_uchardata++ = 0x0100;
4082                *class_uchardata++ = 0x167f;
4083                *class_uchardata++ = XCL_RANGE;
4084                *class_uchardata++ = 0x1681;
4085                *class_uchardata++ = 0x180d;
4086                *class_uchardata++ = XCL_RANGE;
4087                *class_uchardata++ = 0x180f;
4088                *class_uchardata++ = 0x1fff;
4089                *class_uchardata++ = XCL_RANGE;
4090                *class_uchardata++ = 0x200b;
4091                *class_uchardata++ = 0x202e;
4092                *class_uchardata++ = XCL_RANGE;
4093                *class_uchardata++ = 0x2030;
4094                *class_uchardata++ = 0x205e;
4095                *class_uchardata++ = XCL_RANGE;
4096                *class_uchardata++ = 0x2060;
4097                *class_uchardata++ = 0x2fff;
4098                *class_uchardata++ = XCL_RANGE;
4099                *class_uchardata++ = 0x3001;
4100    #ifdef SUPPORT_UTF
4101                if (utf)
4102                  class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);
4103                else
4104    #endif
4105                  *class_uchardata++ = 0xffff;
4106    #elif defined SUPPORT_UTF
4107                if (utf)
4108                {                {
4109                class_utf8 = TRUE;                xclass = TRUE;
4110                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4111                class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata);
4112                class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x167f, class_uchardata);
4113                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4114                class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x1681, class_uchardata);
4115                class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x180d, class_uchardata);
4116                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4117                class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x180f, class_uchardata);
4118                class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x1fff, class_uchardata);
4119                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4120                class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x200b, class_uchardata);
4121                class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x202e, class_uchardata);
4122                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4123                class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x2030, class_uchardata);
4124                class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x205e, class_uchardata);
4125                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4126                class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x2060, class_uchardata);
4127                class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x2fff, class_uchardata);
4128                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4129                class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x3001, class_uchardata);
4130                class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);
4131                }                }
4132  #endif  #endif
4133              continue;              continue;
# Line 3962  for (;; ptr++) Line 4138  for (;; ptr++)
4138              SETBIT(classbits, 0x0c); /* FF */              SETBIT(classbits, 0x0c); /* FF */
4139              SETBIT(classbits, 0x0d); /* CR */              SETBIT(classbits, 0x0d); /* CR */
4140              SETBIT(classbits, 0x85); /* NEL */              SETBIT(classbits, 0x85); /* NEL */
4141  #ifdef SUPPORT_UTF8  #ifndef COMPILE_PCRE8
4142              if (utf8)              xclass = TRUE;
4143                *class_uchardata++ = XCL_RANGE;
4144                *class_uchardata++ = 0x2028;
4145                *class_uchardata++ = 0x2029;
4146    #elif defined SUPPORT_UTF
4147                if (utf)
4148                {                {
4149                class_utf8 = TRUE;                xclass = TRUE;
4150                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4151                class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x2028, class_uchardata);
4152                class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x2029, class_uchardata);
4153                }                }
4154  #endif  #endif
4155              continue;              continue;
# Line 3990  for (;; ptr++) Line 4171  for (;; ptr++)
4171                classbits[c] |= x;                classbits[c] |= x;
4172                }                }
4173    
4174  #ifdef SUPPORT_UTF8  #ifndef COMPILE_PCRE8
4175              if (utf8)              xclass = TRUE;
4176                *class_uchardata++ = XCL_RANGE;
4177                *class_uchardata++ = 0x0100;
4178                *class_uchardata++ = 0x2027;
4179                *class_uchardata++ = XCL_RANGE;
4180                *class_uchardata++ = 0x202a;
4181    #ifdef SUPPORT_UTF
4182                if (utf)
4183                  class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);
4184                else
4185    #endif
4186                  *class_uchardata++ = 0xffff;
4187    #elif defined SUPPORT_UTF
4188                if (utf)
4189                {                {
4190                class_utf8 = TRUE;                xclass = TRUE;
4191                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4192                class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x0100, class_uchardata);
4193                class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x2027, class_uchardata);
4194                *class_utf8data++ = XCL_RANGE;                *class_uchardata++ = XCL_RANGE;
4195                class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x202a, class_uchardata);
4196                class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);                class_uchardata += PRIV(ord2utf)(0x10ffff, class_uchardata);
4197                }                }
4198  #endif  #endif
4199              continue;              continue;
# Line 4012  for (;; ptr++) Line 4206  for (;; ptr++)
4206                int pdata;                int pdata;
4207                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
4208                if (ptype < 0) goto FAILED;                if (ptype < 0) goto FAILED;
4209                class_utf8 = TRUE;                xclass = TRUE;
4210                *class_utf8data++ = ((-c == ESC_p) != negated)?                *class_uchardata++ = ((-c == ESC_p) != negated)?
4211                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4212                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4213                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4214                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4215                continue;                continue;
4216                }                }
4217  #endif  #endif
# Line 4031  for (;; ptr++) Line 4225  for (;; ptr++)
4225                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4226                goto FAILED;                goto FAILED;
4227                }                }
4228              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4229              c = *ptr;              /* Get the final character and fall through */              class_single_char -= 2;  /* Undo the speculative increase. */
4230                c = *ptr;                /* Get the final character and fall through */
4231              break;              break;
4232              }              }
4233            }            }
4234    
4235          /* Fall through if we have a single character (c >= 0). This may be          /* Fall through if we have a single character (c >= 0). This may be
4236          greater than 256 in UTF-8 mode. */          greater than 256. */
4237    
4238          }   /* End of backslash handling */          }   /* End of backslash handling */
4239    
# Line 4086  for (;; ptr++) Line 4281  for (;; ptr++)
4281            goto LONE_SINGLE_CHARACTER;            goto LONE_SINGLE_CHARACTER;
4282            }            }
4283    
4284  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4285          if (utf8)          if (utf)
4286            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4287            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4288            }            }
# Line 4131  for (;; ptr++) Line 4326  for (;; ptr++)
4326    
4327          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4328    
4329            /* Since we found a character range, single character optimizations
4330            cannot be done anymore. */
4331            class_single_char = 2;
4332    
4333          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless
4334          matching, we have to use an XCLASS with extra data items. Caseless          matching, we have to use an XCLASS with extra data items. Caseless
4335          matching for characters > 127 is available only if UCP support is          matching for characters > 127 is available only if UCP support is
4336          available. */          available. */
4337    
4338  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)
4339          if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))          if ((d > 255) || (utf && ((options & PCRE_CASELESS) != 0 && d > 127)))
4340    #elif defined  SUPPORT_UTF
4341            if (utf && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))
4342    #elif !(defined COMPILE_PCRE8)
4343            if (d > 255)
4344    #endif
4345    #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
4346            {            {
4347            class_utf8 = TRUE;            xclass = TRUE;
4348    
4349            /* With UCP support, we can find the other case equivalents of            /* With UCP support, we can find the other case equivalents of
4350            the relevant characters. There may be several ranges. Optimize how            the relevant characters. There may be several ranges. Optimize how
4351            they fit with the basic range. */            they fit with the basic range. */
4352    
4353  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4354    #ifndef COMPILE_PCRE8
4355              if (utf && (options & PCRE_CASELESS) != 0)
4356    #else
4357            if ((options & PCRE_CASELESS) != 0)            if ((options & PCRE_CASELESS) != 0)
4358    #endif
4359              {              {
4360              unsigned int occ, ocd;              unsigned int occ, ocd;
4361              unsigned int cc = c;              unsigned int cc = c;
# Line 4172  for (;; ptr++) Line 4381  for (;; ptr++)
4381    
4382                if (occ == ocd)                if (occ == ocd)
4383                  {                  {
4384                  *class_utf8data++ = XCL_SINGLE;                  *class_uchardata++ = XCL_SINGLE;
4385                  }                  }
4386                else                else
4387                  {                  {
4388                  *class_utf8data++ = XCL_RANGE;                  *class_uchardata++ = XCL_RANGE;
4389                  class_utf8data += _pcre_ord2utf8(occ, class_utf8data);                  class_uchardata += PRIV(ord2utf)(occ, class_uchardata);
4390                  }                  }
4391                class_utf8data += _pcre_ord2utf8(ocd, class_utf8data);                class_uchardata += PRIV(ord2utf)(ocd, class_uchardata);
4392                }                }
4393              }              }
4394  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 4187  for (;; ptr++) Line 4396  for (;; ptr++)
4396            /* Now record the original range, possibly modified for UCP caseless            /* Now record the original range, possibly modified for UCP caseless
4397            overlapping ranges. */            overlapping ranges. */
4398    
4399            *class_utf8data++ = XCL_RANGE;            *class_uchardata++ = XCL_RANGE;
4400            class_utf8data += _pcre_ord2utf8(c, class_utf8data);  #ifdef SUPPORT_UTF
4401            class_utf8data += _pcre_ord2utf8(d, class_utf8data);  #ifndef COMPILE_PCRE8
4402              if (utf)
4403                {
4404                class_uchardata += PRIV(ord2utf)(c, class_uchardata);
4405                class_uchardata += PRIV(ord2utf)(d, class_uchardata);
4406                }
4407              else
4408                {
4409                *class_uchardata++ = c;
4410                *class_uchardata++ = d;
4411                }
4412    #else
4413              class_uchardata += PRIV(ord2utf)(c, class_uchardata);
4414              class_uchardata += PRIV(ord2utf)(d, class_uchardata);
4415    #endif
4416    #else /* SUPPORT_UTF */
4417              *class_uchardata++ = c;
4418              *class_uchardata++ = d;
4419    #endif /* SUPPORT_UTF */
4420    
4421            /* With UCP support, we are done. Without UCP support, there is no            /* With UCP support, we are done. Without UCP support, there is no
4422            caseless matching for UTF-8 characters > 127; we can use the bit map            caseless matching for UTF characters > 127; we can use the bit map
4423            for the smaller ones. */            for the smaller ones. As for 16 bit characters without UTF, we
4424              can still use  */
4425    
4426  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4427            continue;    /* With next character in the class */  #ifndef COMPILE_PCRE8
4428  #else            if (utf)
4429            if ((options & PCRE_CASELESS) == 0 || c > 127) continue;  #endif
4430                continue;    /* With next character in the class */
4431    #endif  /* SUPPORT_UCP */
4432    
4433    #if defined SUPPORT_UTF && !defined(SUPPORT_UCP) && !(defined COMPILE_PCRE8)
4434              if (utf)
4435                {
4436                if ((options & PCRE_CASELESS) == 0 || c > 127) continue;
4437                /* Adjust upper limit and fall through to set up the map */
4438                d = 127;
4439                }
4440              else
4441                {
4442                if (c > 255) continue;
4443                /* Adjust upper limit and fall through to set up the map */
4444                d = 255;
4445                }
4446    #elif defined SUPPORT_UTF && !defined(SUPPORT_UCP)
4447              if ((options & PCRE_CASELESS) == 0 || c > 127) continue;
4448            /* Adjust upper limit and fall through to set up the map */            /* Adjust upper limit and fall through to set up the map */
   
4449            d = 127;            d = 127;
4450    #else
4451  #endif  /* SUPPORT_UCP */            if (c > 255) continue;
4452              /* Adjust upper limit and fall through to set up the map */
4453              d = 255;
4454    #endif  /* SUPPORT_UTF && !SUPPORT_UCP && !COMPILE_PCRE8 */
4455            }            }
4456  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF || !COMPILE_PCRE8 */
4457    
4458          /* We use the bit map for all cases when not in UTF-8 mode; else          /* We use the bit map for 8 bit mode, or when the characters fall
4459          ranges that lie entirely within 0-127 when there is UCP support; else          partially or entirely to [0-255] ([0-127] for UCP) ranges. */
         for partial ranges without UCP support. */  
4460    
4461          class_charcount += d - c + 1;          class_has_8bitchar = 1;
         class_lastchar = d;  
4462    
4463          /* We can save a bit of time by skipping this in the pre-compile. */          /* We can save a bit of time by skipping this in the pre-compile. */
4464    
# Line 4222  for (;; ptr++) Line 4467  for (;; ptr++)
4467            classbits[c/8] |= (1 << (c&7));            classbits[c/8] |= (1 << (c&7));
4468            if ((options & PCRE_CASELESS) != 0)            if ((options & PCRE_CASELESS) != 0)
4469              {              {
4470              int uc = cd->fcc[c];           /* flip case */              int uc = cd->fcc[c]; /* flip case */
4471              classbits[uc/8] |= (1 << (uc&7));              classbits[uc/8] |= (1 << (uc&7));
4472              }              }
4473            }            }
# Line 4236  for (;; ptr++) Line 4481  for (;; ptr++)
4481    
4482        LONE_SINGLE_CHARACTER:        LONE_SINGLE_CHARACTER:
4483    
4484        /* Handle a character that cannot go in the bit map */        /* Only the value of 1 matters for class_single_char. */
4485          if (class_single_char < 2) class_single_char++;
4486    
4487  #ifdef SUPPORT_UTF8        /* If class_charcount is 1, we saw precisely one character. As long as
4488        if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))        there were no negated characters >= 128 and there was no use of \p or \P,
4489          in other words, no use of any XCLASS features, we can optimize.
4490    
4491          In UTF-8 mode, we can optimize the negative case only if there were no
4492          characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
4493          operate on single-bytes characters only. This is an historical hangover.
4494          Maybe one day we can tidy these opcodes to handle multi-byte characters.
4495    
4496          The optimization throws away the bit map. We turn the item into a
4497          1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.
4498          Note that OP_NOT[I] does not support multibyte characters. In the positive
4499          case, it can cause firstchar to be set. Otherwise, there can be no first
4500          char if this item is first, whatever repeat count may follow. In the case
4501          of reqchar, save the previous value for reinstating. */
4502    
4503    #ifdef SUPPORT_UTF
4504          if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET
4505            && (!utf || !negate_class || c < (MAX_VALUE_FOR_SINGLE_CHAR + 1)))
4506    #else
4507          if (class_single_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
4508    #endif
4509          {          {
4510          class_utf8 = TRUE;          ptr++;
4511          *class_utf8data++ = XCL_SINGLE;          zeroreqchar = reqchar;
4512          class_utf8data += _pcre_ord2utf8(c, class_utf8data);  
4513            /* The OP_NOT[I] opcodes work on single characters only. */
4514    
4515            if (negate_class)
4516              {
4517              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4518              zerofirstchar = firstchar;
4519              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4520              *code++ = c;
4521              goto NOT_CHAR;
4522              }
4523    
4524            /* For a single, positive character, get the value into mcbuffer, and
4525            then we can handle this with the normal one-character code. */
4526    
4527    #ifdef SUPPORT_UTF
4528            if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4529              mclength = PRIV(ord2utf)(c, mcbuffer);
4530            else
4531    #endif
4532              {
4533              mcbuffer[0] = c;
4534              mclength = 1;
4535              }
4536            goto ONE_CHAR;
4537            }       /* End of 1-char optimization */
4538    
4539          /* Handle a character that cannot go in the bit map. */
4540    
4541    #if defined SUPPORT_UTF && !(defined COMPILE_PCRE8)
4542          if ((c > 255) || (utf && ((options & PCRE_CASELESS) != 0 && c > 127)))
4543    #elif defined SUPPORT_UTF
4544          if (utf && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))
4545    #elif !(defined COMPILE_PCRE8)
4546          if (c > 255)
4547    #endif
4548    
4549    #if defined SUPPORT_UTF || !(defined COMPILE_PCRE8)
4550            {
4551            xclass = TRUE;
4552            *class_uchardata++ = XCL_SINGLE;
4553    #ifdef SUPPORT_UTF
4554    #ifndef COMPILE_PCRE8
4555            /* In non 8 bit mode, we can get here even if we are not in UTF mode. */
4556            if (!utf)
4557              *class_uchardata++ = c;
4558            else
4559    #endif
4560              class_uchardata += PRIV(ord2utf)(c, class_uchardata);
4561    #else /* SUPPORT_UTF */
4562            *class_uchardata++ = c;
4563    #endif /* SUPPORT_UTF */
4564    
4565  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4566    #ifdef COMPILE_PCRE8
4567          if ((options & PCRE_CASELESS) != 0)          if ((options & PCRE_CASELESS) != 0)
4568    #else
4569            /* In non 8 bit mode, we can get here even if we are not in UTF mode. */
4570            if (utf && (options & PCRE_CASELESS) != 0)
4571    #endif
4572            {            {
4573            unsigned int othercase;            unsigned int othercase;
4574            if ((othercase = UCD_OTHERCASE(c)) != c)            if ((othercase = UCD_OTHERCASE(c)) != c)
4575              {              {
4576              *class_utf8data++ = XCL_SINGLE;              *class_uchardata++ = XCL_SINGLE;
4577              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);              class_uchardata += PRIV(ord2utf)(othercase, class_uchardata);
4578              }              }
4579            }            }
4580  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
4581    
4582          }          }
4583        else        else
4584  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF || COMPILE_PCRE16 */
4585    
4586        /* Handle a single-byte character */        /* Handle a single-byte character */
4587          {          {
4588            class_has_8bitchar = 1;
4589          classbits[c/8] |= (1 << (c&7));          classbits[c/8] |= (1 << (c&7));
4590          if ((options & PCRE_CASELESS) != 0)          if ((options & PCRE_CASELESS) != 0)
4591            {            {
4592            c = cd->fcc[c];   /* flip case */            c = cd->fcc[c]; /* flip case */
4593            classbits[c/8] |= (1 << (c&7));            classbits[c/8] |= (1 << (c&7));
4594            }            }
         class_charcount++;  
         class_lastchar = c;  
4595          }          }
4596        }        }
4597    
# Line 4291  for (;; ptr++) Line 4612  for (;; ptr++)
4612        goto FAILED;        goto FAILED;
4613        }        }
4614    
4615      /* If class_charcount is 1, we saw precisely one character whose value is      /* If this is the first thing in the branch, there can be no first char
4616      less than 256. As long as there were no characters >= 128 and there was no      setting, whatever the repeat count. Any reqchar setting must remain
4617      use of \p or \P, in other words, no use of any XCLASS features, we can      unchanged after any kind of repeat. */
4618      optimize.  
4619        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4620      In UTF-8 mode, we can optimize the negative case only if there were no      zerofirstchar = firstchar;
4621      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      zeroreqchar = reqchar;
     operate on single-bytes characters only. This is an historical hangover.  
     Maybe one day we can tidy these opcodes to handle multi-byte characters.  
   
     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  
 #endif  
         {  
         mcbuffer[0] = class_lastchar;  
         mclength = 1;  
         }  
       goto ONE_CHAR;  
       }       /* End of 1-char optimization */  
   
     /* The general case - not the one-char optimization. If this is the first  
     thing in the branch, there can be no first char setting, whatever the  
     repeat count. Any reqbyte setting must remain unchanged after any kind of  
     repeat. */  
   
     if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;  
     zerofirstbyte = firstbyte;  
     zeroreqbyte = reqbyte;  
4622    
4623      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4624      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 4360  for (;; ptr++) Line 4628  for (;; ptr++)
4628      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
4629      actual compiled code. */      actual compiled code. */
4630    
4631  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4632      if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))      if (xclass && (!should_flip_negation || (options & PCRE_UCP) != 0))
4633    #elif !defined COMPILE_PCRE8
4634        if (xclass && !should_flip_negation)
4635    #endif
4636    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4637        {        {
4638        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_uchardata++ = XCL_END;    /* Marks the end of extra data */
4639        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
4640        code += LINK_SIZE;        code += LINK_SIZE;
4641        *code = negate_class? XCL_NOT : 0;        *code = negate_class? XCL_NOT:0;
4642    
4643        /* 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;
4644        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. */
4645    
4646        if (class_charcount > 0)        if (class_has_8bitchar > 0)
4647          {          {
4648          *code++ |= XCL_MAP;          *code++ |= XCL_MAP;
4649          memmove(code + 32, code, class_utf8data - code);          memmove(code + (32 / sizeof(pcre_uchar)), code,
4650              IN_UCHARS(class_uchardata - code));
4651          memcpy(code, classbits, 32);          memcpy(code, classbits, 32);
4652          code = class_utf8data + 32;          code = class_uchardata + (32 / sizeof(pcre_uchar));
4653          }          }
4654        else code = class_utf8data;        else code = class_uchardata;
4655    
4656        /* Now fill in the complete length of the item */        /* Now fill in the complete length of the item */
4657    
4658        PUT(previous, 1, code - previous);        PUT(previous, 1, (int)(code - previous));
4659        break;   /* End of class handling */        break;   /* End of class handling */
4660        }        }
4661  #endif  #endif
# Line 4394  for (;; ptr++) Line 4667  for (;; ptr++)
4667      negating it if necessary. */      negating it if necessary. */
4668    
4669      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
4670      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  
4671        {        {
4672          if (negate_class)
4673            for (c = 0; c < 32; c++) classbits[c] = ~classbits[c];
4674        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4675        }        }
4676      code += 32;      code += 32 / sizeof(pcre_uchar);
4677        NOT_CHAR:
4678      break;      break;
4679    
4680    
# Line 4440  for (;; ptr++) Line 4711  for (;; ptr++)
4711    
4712      if (repeat_min == 0)      if (repeat_min == 0)
4713        {        {
4714        firstbyte = zerofirstbyte;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4715        reqbyte = zeroreqbyte;        /* Ditto */        reqchar = zeroreqchar;        /* Ditto */
4716        }        }
4717    
4718      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4480  for (;; ptr++) Line 4751  for (;; ptr++)
4751      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
4752      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,
4753      but for the moment we rely on the code for repeating groups. */      but for the moment we rely on the code for repeating groups. */
4754    
4755      if (*previous == OP_RECURSE)      if (*previous == OP_RECURSE)
4756        {        {
4757        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);        memmove(previous + 1 + LINK_SIZE, previous, IN_UCHARS(1 + LINK_SIZE));
4758        *previous = OP_ONCE;        *previous = OP_ONCE;
4759        PUT(previous, 1, 2 + 2*LINK_SIZE);        PUT(previous, 1, 2 + 2*LINK_SIZE);
4760        previous[2 + 2*LINK_SIZE] = OP_KET;        previous[2 + 2*LINK_SIZE] = OP_KET;
# Line 4506  for (;; ptr++) Line 4777  for (;; ptr++)
4777    
4778      /* If previous was a character match, abolish the item and generate a      /* If previous was a character match, abolish the item and generate a
4779      repeat item instead. If a char item has a minumum of more than one, ensure      repeat item instead. If a char item has a minumum of more than one, ensure
4780      that it is set in reqbyte - it might not be if a sequence such as x{3} is      that it is set in reqchar - it might not be if a sequence such as x{3} is
4781      the first thing in a branch because the x will have gone into firstbyte      the first thing in a branch because the x will have gone into firstchar
4782      instead.  */      instead.  */
4783    
4784      if (*previous == OP_CHAR || *previous == OP_CHARI)      if (*previous == OP_CHAR || *previous == OP_CHARI)
4785        {        {
4786        op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;        op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;
4787    
4788        /* 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
4789        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
4790        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
4791        length rather than a small character. */        it's a length rather than a small character. */
4792    
4793  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4794        if (utf8 && (code[-1] & 0x80) != 0)        if (utf && NOT_FIRSTCHAR(code[-1]))
4795          {          {
4796          uschar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
4797          while((*lastchar & 0xc0) == 0x80) lastchar--;          BACKCHAR(lastchar);
4798          c = code - lastchar;            /* Length of UTF-8 character */          c = (int)(code - lastchar);     /* Length of UTF-8 character */
4799          memcpy(utf8_char, lastchar, c); /* Save the char */          memcpy(utf_chars, lastchar, IN_UCHARS(c)); /* Save the char */
4800          c |= 0x80;                      /* Flag c as a length */          c |= UTF_LENGTH;                /* Flag c as a length */
4801          }          }
4802        else        else
4803  #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. */  
4804    
4805          /* Handle the case of a single charater - either with no UTF support, or
4806          with UTF disabled, or for a single character UTF character. */
4807          {          {
4808          c = code[-1];          c = code[-1];
4809          if (repeat_min > 1) reqbyte = c | req_caseopt | cd->req_varyopt;          if (repeat_min > 1) reqchar = c | req_caseopt | cd->req_varyopt;
4810          }          }
4811    
4812        /* 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 4546  for (;; ptr++) Line 4816  for (;; ptr++)
4816    
4817        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4818            repeat_max < 0 &&            repeat_max < 0 &&
4819            check_auto_possessive(previous, utf8, ptr + 1, options, cd))            check_auto_possessive(previous, utf, ptr + 1, options, cd))
4820          {          {
4821          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4822          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 4567  for (;; ptr++) Line 4837  for (;; ptr++)
4837        c = previous[1];        c = previous[1];
4838        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4839            repeat_max < 0 &&            repeat_max < 0 &&
4840            check_auto_possessive(previous, utf8, ptr + 1, options, cd))            check_auto_possessive(previous, utf, ptr + 1, options, cd))
4841          {          {
4842          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4843          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 4584  for (;; ptr++) Line 4854  for (;; ptr++)
4854    
4855      else if (*previous < OP_EODN)      else if (*previous < OP_EODN)
4856        {        {
4857        uschar *oldcode;        pcre_uchar *oldcode;
4858        int prop_type, prop_value;        int prop_type, prop_value;
4859        op_type = OP_TYPESTAR - OP_STAR;  /* Use type opcodes */        op_type = OP_TYPESTAR - OP_STAR;  /* Use type opcodes */
4860        c = *previous;        c = *previous;
4861    
4862        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4863            repeat_max < 0 &&            repeat_max < 0 &&
4864            check_auto_possessive(previous, utf8, ptr + 1, options, cd))            check_auto_possessive(previous, utf, ptr + 1, options, cd))
4865          {          {
4866          repeat_type = 0;    /* Force greedy */          repeat_type = 0;    /* Force greedy */
4867          possessive_quantifier = TRUE;          possessive_quantifier = TRUE;
# Line 4671  for (;; ptr++) Line 4941  for (;; ptr++)
4941          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
4942          Unicode property match, there are two extra bytes that define the          Unicode property match, there are two extra bytes that define the
4943          required property. In UTF-8 mode, long characters have their length in          required property. In UTF-8 mode, long characters have their length in
4944          c, with the 0x80 bit as a flag. */          c, with the UTF_LENGTH bit as a flag. */
4945    
4946          if (repeat_max < 0)          if (repeat_max < 0)
4947            {            {
4948  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4949            if (utf8 && c >= 128)            if (utf && (c & UTF_LENGTH) != 0)
4950              {              {
4951              memcpy(code, utf8_char, c & 7);              memcpy(code, utf_chars, IN_UCHARS(c & 7));
4952              code += c & 7;              code += c & 7;
4953              }              }
4954            else            else
# Line 4700  for (;; ptr++) Line 4970  for (;; ptr++)
4970    
4971          else if (repeat_max != repeat_min)          else if (repeat_max != repeat_min)
4972            {            {
4973  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4974            if (utf8 && c >= 128)            if (utf && (c & UTF_LENGTH) != 0)
4975              {              {
4976              memcpy(code, utf8_char, c & 7);              memcpy(code, utf_chars, IN_UCHARS(c & 7));
4977              code += c & 7;              code += c & 7;
4978              }              }
4979            else            else
# Line 4730  for (;; ptr++) Line 5000  for (;; ptr++)
5000    
5001        /* The character or character type itself comes last in all cases. */        /* The character or character type itself comes last in all cases. */
5002    
5003  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
5004        if (utf8 && c >= 128)        if (utf && (c & UTF_LENGTH) != 0)
5005          {          {
5006          memcpy(code, utf8_char, c & 7);          memcpy(code, utf_chars, IN_UCHARS(c & 7));
5007          code += c & 7;          code += c & 7;
5008          }          }
5009        else        else
# Line 4757  for (;; ptr++) Line 5027  for (;; ptr++)
5027    
5028      else if (*previous == OP_CLASS ||      else if (*previous == OP_CLASS ||
5029               *previous == OP_NCLASS ||               *previous == OP_NCLASS ||
5030  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
5031               *previous == OP_XCLASS ||               *previous == OP_XCLASS ||
5032  #endif  #endif
5033               *previous == OP_REF ||               *previous == OP_REF ||
# Line 4806  for (;; ptr++) Line 5076  for (;; ptr++)
5076        {        {
5077        register int i;        register int i;
5078        int len = (int)(code - previous);        int len = (int)(code - previous);
5079        uschar *bralink = NULL;        pcre_uchar *bralink = NULL;
5080        uschar *brazeroptr = NULL;        pcre_uchar *brazeroptr = NULL;
5081    
5082        /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so        /* Repeating a DEFINE group is pointless, but Perl allows the syntax, so
5083        we just ignore the repeat. */        we just ignore the repeat. */
# Line 4860  for (;; ptr++) Line 5130  for (;; ptr++)
5130          if (repeat_max <= 1)    /* Covers 0, 1, and unlimited */          if (repeat_max <= 1)    /* Covers 0, 1, and unlimited */
5131            {            {
5132            *code = OP_END;            *code = OP_END;
5133            adjust_recurse(previous, 1, utf8, cd, save_hwm);            adjust_recurse(previous, 1, utf, cd, save_hwm);
5134            memmove(previous+1, previous, len);            memmove(previous + 1, previous, IN_UCHARS(len));
5135            code++;            code++;
5136            if (repeat_max == 0)            if (repeat_max == 0)
5137              {              {
# Line 4884  for (;; ptr++) Line 5154  for (;; ptr++)
5154            {            {
5155            int offset;            int offset;
5156            *code = OP_END;            *code = OP_END;
5157            adjust_recurse(previous, 2 + LINK_SIZE, utf8, cd, save_hwm);            adjust_recurse(previous, 2 + LINK_SIZE, utf, cd, save_hwm);
5158            memmove(previous + 2 + LINK_SIZE, previous, len);            memmove(previous + 2 + LINK_SIZE, previous, IN_UCHARS(len));
5159            code += 2 + LINK_SIZE;            code += 2 + LINK_SIZE;
5160            *previous++ = OP_BRAZERO + repeat_type;            *previous++ = OP_BRAZERO + repeat_type;
5161            *previous++ = OP_BRA;            *previous++ = OP_BRA;
# Line 4932  for (;; ptr++) Line 5202  for (;; ptr++)
5202              }              }
5203    
5204            /* This is compiling for real. If there is a set first byte for            /* This is compiling for real. If there is a set first byte for
5205            the group, and we have not yet set a "required byte", set it. Make            the group, and we have not yet set a "required byte", set it. Make
5206            sure there is enough workspace for copying forward references before            sure there is enough workspace for copying forward references before
5207            doing the copy. */            doing the copy. */
5208    
5209            else            else
5210              {              {
5211              if (groupsetfirstbyte && reqbyte < 0) reqbyte = firstbyte;              if (groupsetfirstchar && reqchar < 0) reqchar = firstchar;
5212    
5213              for (i = 1; i < repeat_min; i++)              for (i = 1; i < repeat_min; i++)
5214                {                {
5215                uschar *hc;                pcre_uchar *hc;
5216                uschar *this_hwm = cd->hwm;                pcre_uchar *this_hwm = cd->hwm;
5217                memcpy(code, previous, len);                memcpy(code, previous, IN_UCHARS(len));
5218    
5219                while (cd->hwm > cd->start_workspace + cd->workspace_size -                while (cd->hwm > cd->start_workspace + cd->workspace_size -
5220                       WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))                       WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))
5221                  {                  {
# Line 4953  for (;; ptr++) Line 5223  for (;; ptr++)
5223                  int this_offset = this_hwm - cd->start_workspace;                  int this_offset = this_hwm - cd->start_workspace;
5224                  *errorcodeptr = expand_workspace(cd);                  *errorcodeptr = expand_workspace(cd);
5225                  if (*errorcodeptr != 0) goto FAILED;                  if (*errorcodeptr != 0) goto FAILED;
5226                  save_hwm = (uschar *)cd->start_workspace + save_offset;                  save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;
5227                  this_hwm = (uschar *)cd->start_workspace + this_offset;                  this_hwm = (pcre_uchar *)cd->start_workspace + this_offset;
5228                  }                  }
5229    
5230                for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE)                for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE)
5231                  {                  {
5232                  PUT(cd->hwm, 0, GET(hc, 0) + len);                  PUT(cd->hwm, 0, GET(hc, 0) + len);
# Line 4986  for (;; ptr++) Line 5256  for (;; ptr++)
5256          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some          add 2 + 2*LINKSIZE to allow for the nesting that occurs. Do some
5257          paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is          paranoid checks to avoid integer overflow. The INT64_OR_DOUBLE type is
5258          a 64-bit integer type when available, otherwise double. */          a 64-bit integer type when available, otherwise double. */
5259    
5260          if (lengthptr != NULL && repeat_max > 0)          if (lengthptr != NULL && repeat_max > 0)
5261            {            {
5262            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -            int delta = repeat_max * (length_prevgroup + 1 + 2 + 2*LINK_SIZE) -
# Line 5006  for (;; ptr++) Line 5276  for (;; ptr++)
5276    
5277          else for (i = repeat_max - 1; i >= 0; i--)          else for (i = repeat_max - 1; i >= 0; i--)
5278            {            {
5279            uschar *hc;            pcre_uchar *hc;
5280            uschar *this_hwm = cd->hwm;            pcre_uchar *this_hwm = cd->hwm;
5281    
5282            *code++ = OP_BRAZERO + repeat_type;            *code++ = OP_BRAZERO + repeat_type;
5283    
# Line 5023  for (;; ptr++) Line 5293  for (;; ptr++)
5293              PUTINC(code, 0, offset);              PUTINC(code, 0, offset);
5294              }              }
5295    
5296            memcpy(code, previous, len);            memcpy(code, previous, IN_UCHARS(len));
5297    
5298            /* Ensure there is enough workspace for forward references before            /* Ensure there is enough workspace for forward references before
5299            copying them. */            copying them. */
5300    
5301            while (cd->hwm > cd->start_workspace + cd->workspace_size -            while (cd->hwm > cd->start_workspace + cd->workspace_size -
5302                   WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))                   WORK_SIZE_SAFETY_MARGIN - (this_hwm - save_hwm))
5303              {              {
# Line 5035  for (;; ptr++) Line 5305  for (;; ptr++)
5305              int this_offset = this_hwm - cd->start_workspace;              int this_offset = this_hwm - cd->start_workspace;
5306              *errorcodeptr = expand_workspace(cd);              *errorcodeptr = expand_workspace(cd);
5307              if (*errorcodeptr != 0) goto FAILED;              if (*errorcodeptr != 0) goto FAILED;
5308              save_hwm = (uschar *)cd->start_workspace + save_offset;              save_hwm = (pcre_uchar *)cd->start_workspace + save_offset;
5309              this_hwm = (uschar *)cd->start_workspace + this_offset;              this_hwm = (pcre_uchar *)cd->start_workspace + this_offset;
5310              }              }
5311    
5312            for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE)            for (hc = save_hwm; hc < this_hwm; hc += LINK_SIZE)
5313              {              {
5314              PUT(cd->hwm, 0, GET(hc, 0) + len + ((i != 0)? 2+LINK_SIZE : 1));              PUT(cd->hwm, 0, GET(hc, 0) + len + ((i != 0)? 2+LINK_SIZE : 1));
# Line 5055  for (;; ptr++) Line 5325  for (;; ptr++)
5325            {            {
5326            int oldlinkoffset;            int oldlinkoffset;
5327            int offset = (int)(code - bralink + 1);            int offset = (int)(code - bralink + 1);
5328            uschar *bra = code - offset;            pcre_uchar *bra = code - offset;
5329            oldlinkoffset = GET(bra, 1);            oldlinkoffset = GET(bra, 1);
5330            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;            bralink = (oldlinkoffset == 0)? NULL : bralink - oldlinkoffset;
5331            *code++ = OP_KET;            *code++ = OP_KET;
# Line 5069  for (;; ptr++) Line 5339  for (;; ptr++)
5339        ONCE brackets can be converted into non-capturing brackets, as the        ONCE brackets can be converted into non-capturing brackets, as the
5340        behaviour of (?:xx)++ is the same as (?>xx)++ and this saves having to        behaviour of (?:xx)++ is the same as (?>xx)++ and this saves having to
5341        deal with possessive ONCEs specially.        deal with possessive ONCEs specially.
5342    
5343        Otherwise, when we are doing the actual compile phase, check to see        Otherwise, when we are doing the actual compile phase, check to see
5344        whether this group is one that could match an empty string. If so,        whether this group is one that could match an empty string. If so,
5345        convert the initial operator to the S form (e.g. OP_BRA -> OP_SBRA) so        convert the initial operator to the S form (e.g. OP_BRA -> OP_SBRA) so
5346        that runtime checking can be done. [This check is also applied to ONCE        that runtime checking can be done. [This check is also applied to ONCE
5347        groups at runtime, but in a different way.]        groups at runtime, but in a different way.]
5348    
5349        Then, if the quantifier was possessive and the bracket is not a        Then, if the quantifier was possessive and the bracket is not a
5350        conditional, we convert the BRA code to the POS form, and the KET code to        conditional, we convert the BRA code to the POS form, and the KET code to
5351        KETRPOS. (It turns out to be convenient at runtime to detect this kind of        KETRPOS. (It turns out to be convenient at runtime to detect this kind of
5352        subpattern at both the start and at the end.) The use of special opcodes        subpattern at both the start and at the end.) The use of special opcodes
5353        makes it possible to reduce greatly the stack usage in pcre_exec(). If        makes it possible to reduce greatly the stack usage in pcre_exec(). If
5354        the group is preceded by OP_BRAZERO, convert this to OP_BRAPOSZERO.        the group is preceded by OP_BRAZERO, convert this to OP_BRAPOSZERO.
5355    
5356        Then, if the minimum number of matches is 1 or 0, cancel the possessive        Then, if the minimum number of matches is 1 or 0, cancel the possessive
5357        flag so that the default action below, of wrapping everything inside        flag so that the default action below, of wrapping everything inside
5358        atomic brackets, does not happen. When the minimum is greater than 1,        atomic brackets, does not happen. When the minimum is greater than 1,
5359        there will be earlier copies of the group, and so we still have to wrap        there will be earlier copies of the group, and so we still have to wrap
5360        the whole thing. */        the whole thing. */
5361    
5362        else        else
5363          {          {
5364          uschar *ketcode = code - 1 - LINK_SIZE;          pcre_uchar *ketcode = code - 1 - LINK_SIZE;
5365          uschar *bracode = ketcode - GET(ketcode, 1);          pcre_uchar *bracode = ketcode - GET(ketcode, 1);
5366    
5367          /* Convert possessive ONCE brackets to non-capturing */          /* Convert possessive ONCE brackets to non-capturing */
5368    
5369          if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) &&          if ((*bracode == OP_ONCE || *bracode == OP_ONCE_NC) &&
5370              possessive_quantifier) *bracode = OP_BRA;              possessive_quantifier) *bracode = OP_BRA;
5371    
5372          /* For non-possessive ONCE brackets, all we need to do is to          /* For non-possessive ONCE brackets, all we need to do is to
5373          set the KET. */          set the KET. */
5374    
5375          if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC)          if (*bracode == OP_ONCE || *bracode == OP_ONCE_NC)
5376            *ketcode = OP_KETRMAX + repeat_type;            *ketcode = OP_KETRMAX + repeat_type;
5377    
5378          /* Handle non-ONCE brackets and possessive ONCEs (which have been          /* Handle non-ONCE brackets and possessive ONCEs (which have been
5379          converted to non-capturing above). */          converted to non-capturing above). */
5380    
5381          else          else
5382            {            {
5383            /* In the compile phase, check for empty string matching. */            /* In the compile phase, check for empty string matching. */
5384    
5385            if (lengthptr == NULL)            if (lengthptr == NULL)
5386              {              {
5387              uschar *scode = bracode;              pcre_uchar *scode = bracode;
5388              do              do
5389                {                {
5390                if (could_be_empty_branch(scode, ketcode, utf8, cd))                if (could_be_empty_branch(scode, ketcode, utf, cd))
5391                  {                  {
5392                  *bracode += OP_SBRA - OP_BRA;                  *bracode += OP_SBRA - OP_BRA;
5393                  break;                  break;
# Line 5126  for (;; ptr++) Line 5396  for (;; ptr++)
5396                }                }
5397              while (*scode == OP_ALT);              while (*scode == OP_ALT);
5398              }              }
5399    
5400            /* Handle possessive quantifiers. */            /* Handle possessive quantifiers. */
5401    
5402            if (possessive_quantifier)            if (possessive_quantifier)
# Line 5135  for (;; ptr++) Line 5405  for (;; ptr++)
5405              repeated non-capturing bracket, because we have not invented POS              repeated non-capturing bracket, because we have not invented POS
5406              versions of the COND opcodes. Because we are moving code along, we              versions of the COND opcodes. Because we are moving code along, we
5407              must ensure that any pending recursive references are updated. */              must ensure that any pending recursive references are updated. */
5408    
5409              if (*bracode == OP_COND || *bracode == OP_SCOND)              if (*bracode == OP_COND || *bracode == OP_SCOND)
5410                {                {
5411                int nlen = (int)(code - bracode);                int nlen = (int)(code - bracode);
5412                *code = OP_END;                *code = OP_END;
5413                adjust_recurse(bracode, 1 + LINK_SIZE, utf8, cd, save_hwm);                adjust_recurse(bracode, 1 + LINK_SIZE, utf, cd, save_hwm);
5414                memmove(bracode + 1+LINK_SIZE, bracode, nlen);                memmove(bracode + 1 + LINK_SIZE, bracode, IN_UCHARS(nlen));
5415                code += 1 + LINK_SIZE;                code += 1 + LINK_SIZE;
5416                nlen += 1 + LINK_SIZE;                nlen += 1 + LINK_SIZE;
5417                *bracode = OP_BRAPOS;                *bracode = OP_BRAPOS;
5418                *code++ = OP_KETRPOS;                *code++ = OP_KETRPOS;
5419                PUTINC(code, 0, nlen);                PUTINC(code, 0, nlen);
5420                PUT(bracode, 1, nlen);                PUT(bracode, 1, nlen);
5421                }                }
5422    
5423              /* For non-COND brackets, we modify the BRA code and use KETRPOS. */              /* For non-COND brackets, we modify the BRA code and use KETRPOS. */
5424    
5425              else              else
5426                {                {
5427                *bracode += 1;              /* Switch to xxxPOS opcodes */                *bracode += 1;              /* Switch to xxxPOS opcodes */
5428                *ketcode = OP_KETRPOS;                *ketcode = OP_KETRPOS;
5429                }                }
5430    
5431              /* If the minimum is zero, mark it as possessive, then unset the              /* If the minimum is zero, mark it as possessive, then unset the
5432              possessive flag when the minimum is 0 or 1. */              possessive flag when the minimum is 0 or 1. */
5433    
5434              if (brazeroptr != NULL) *brazeroptr = OP_BRAPOSZERO;              if (brazeroptr != NULL) *brazeroptr = OP_BRAPOSZERO;
5435              if (repeat_min < 2) possessive_quantifier = FALSE;              if (repeat_min < 2) possessive_quantifier = FALSE;
5436              }              }
5437    
5438            /* Non-possessive quantifier */            /* Non-possessive quantifier */
5439    
5440            else *ketcode = OP_KETRMAX + repeat_type;            else *ketcode = OP_KETRMAX + repeat_type;
5441            }            }
5442          }          }
# Line 5210  for (;; ptr++) Line 5480  for (;; ptr++)
5480        int len;        int len;
5481    
5482        if (*tempcode == OP_TYPEEXACT)        if (*tempcode == OP_TYPEEXACT)
5483          tempcode += _pcre_OP_lengths[*tempcode] +          tempcode += PRIV(OP_lengths)[*tempcode] +
5484            ((tempcode[3] == OP_PROP || tempcode[3] == OP_NOTPROP)? 2 : 0);            ((tempcode[1 + IMM2_SIZE] == OP_PROP
5485              || tempcode[1 + IMM2_SIZE] == OP_NOTPROP)? 2 : 0);
5486    
5487        else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)        else if (*tempcode == OP_EXACT || *tempcode == OP_NOTEXACT)
5488          {          {
5489          tempcode += _pcre_OP_lengths[*tempcode];          tempcode += PRIV(OP_lengths)[*tempcode];
5490  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
5491          if (utf8 && tempcode[-1] >= 0xc0)          if (utf && HAS_EXTRALEN(tempcode[-1]))
5492            tempcode += _pcre_utf8_table4[tempcode[-1] & 0x3f];            tempcode += GET_EXTRALEN(tempcode[-1]);
5493  #endif  #endif
5494          }          }
5495    
# Line 5255  for (;; ptr++) Line 5526  for (;; ptr++)
5526    
5527          default:          default:
5528          *code = OP_END;          *code = OP_END;
5529          adjust_recurse(tempcode, 1 + LINK_SIZE, utf8, cd, save_hwm);          adjust_recurse(tempcode, 1 + LINK_SIZE, utf, cd, save_hwm);
5530          memmove(tempcode + 1+LINK_SIZE, tempcode, len);          memmove(tempcode + 1 + LINK_SIZE, tempcode, IN_UCHARS(len));
5531          code += 1 + LINK_SIZE;          code += 1 + LINK_SIZE;
5532          len += 1 + LINK_SIZE;          len += 1 + LINK_SIZE;
5533          tempcode[0] = OP_ONCE;          tempcode[0] = OP_ONCE;
# Line 5268  for (;; ptr++) Line 5539  for (;; ptr++)
5539        }        }
5540    
5541      /* In all case we no longer have a previous item. We also set the      /* In all case we no longer have a previous item. We also set the
5542      "follows varying string" flag for subsequently encountered reqbytes if      "follows varying string" flag for subsequently encountered reqchars if
5543      it isn't already set and we have just passed a varying length item. */      it isn't already set and we have just passed a varying length item. */
5544    
5545      END_REPEAT:      END_REPEAT:
# Line 5291  for (;; ptr++) Line 5562  for (;; ptr++)
5562    
5563      /* First deal with various "verbs" that can be introduced by '*'. */      /* First deal with various "verbs" that can be introduced by '*'. */
5564    
5565      if (*(++ptr) == CHAR_ASTERISK &&      ptr++;
5566           ((cd->ctypes[ptr[1]] & ctype_letter) != 0 || ptr[1] == ':'))      if (ptr[0] == CHAR_ASTERISK && (ptr[1] == ':'
5567             || (MAX_255(ptr[1]) && ((cd->ctypes[ptr[1]] & ctype_letter) != 0))))
5568        {        {
5569        int i, namelen;        int i, namelen;
5570        int arglen = 0;        int arglen = 0;
5571        const char *vn = verbnames;        const char *vn = verbnames;
5572        const uschar *name = ptr + 1;        const pcre_uchar *name = ptr + 1;
5573        const uschar *arg = NULL;        const pcre_uchar *arg = NULL;
5574        previous = NULL;        previous = NULL;
5575        while ((cd->ctypes[*++ptr] & ctype_letter) != 0) {};        ptr++;
5576          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_letter) != 0) ptr++;
5577        namelen = (int)(ptr - name);        namelen = (int)(ptr - name);
5578    
5579        /* It appears that Perl allows any characters whatsoever, other than        /* It appears that Perl allows any characters whatsoever, other than
# Line 5325  for (;; ptr++) Line 5598  for (;; ptr++)
5598        for (i = 0; i < verbcount; i++)        for (i = 0; i < verbcount; i++)
5599          {          {
5600          if (namelen == verbs[i].len &&          if (namelen == verbs[i].len &&
5601              strncmp((char *)name, vn, namelen) == 0)              STRNCMP_UC_C8(name, vn, namelen) == 0)
5602            {            {
5603            /* Check for open captures before ACCEPT and convert it to            /* Check for open captures before ACCEPT and convert it to
5604            ASSERT_ACCEPT if in an assertion. */            ASSERT_ACCEPT if in an assertion. */
# Line 5346  for (;; ptr++) Line 5619  for (;; ptr++)
5619                }                }
5620              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;              *code++ = (cd->assert_depth > 0)? OP_ASSERT_ACCEPT : OP_ACCEPT;
5621    
5622              /* Do not set firstbyte after *ACCEPT */              /* Do not set firstchar after *ACCEPT */
5623              if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
5624              }              }
5625    
5626            /* Handle other cases with/without an argument */            /* Handle other cases with/without an argument */
# Line 5373  for (;; ptr++) Line 5646  for (;; ptr++)
5646              *code = verbs[i].op_arg;              *code = verbs[i].op_arg;
5647              if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;              if (*code++ == OP_THEN_ARG) cd->external_flags |= PCRE_HASTHEN;
5648              *code++ = arglen;              *code++ = arglen;
5649              memcpy(code, arg, arglen);              memcpy(code, arg, IN_UCHARS(arglen));
5650              code += arglen;              code += arglen;
5651              *code++ = 0;              *code++ = 0;
5652              }              }
# Line 5396  for (;; ptr++) Line 5669  for (;; ptr++)
5669        {        {
5670        int i, set, unset, namelen;        int i, set, unset, namelen;
5671        int *optset;        int *optset;
5672        const uschar *name;        const pcre_uchar *name;
5673        uschar *slot;        pcre_uchar *slot;
5674    
5675        switch (*(++ptr))        switch (*(++ptr))
5676          {          {
# Line 5450  for (;; ptr++) Line 5723  for (;; ptr++)
5723            break;            break;
5724    
5725          /* Most other conditions use OP_CREF (a couple change to OP_RREF          /* Most other conditions use OP_CREF (a couple change to OP_RREF
5726          below), and all need to skip 3 bytes at the start of the group. */          below), and all need to skip 1+IMM2_SIZE bytes at the start of the group. */
5727    
5728          code[1+LINK_SIZE] = OP_CREF;          code[1+LINK_SIZE] = OP_CREF;
5729          skipbytes = 3;          skipbytes = 1+IMM2_SIZE;
5730          refsign = -1;          refsign = -1;
5731    
5732          /* Check for a test for recursion in a named group. */          /* Check for a test for recursion in a named group. */
# Line 5486  for (;; ptr++) Line 5759  for (;; ptr++)
5759    
5760          /* We now expect to read a name; any thing else is an error */          /* We now expect to read a name; any thing else is an error */
5761    
5762          if ((cd->ctypes[ptr[1]] & ctype_word) == 0)          if (!MAX_255(ptr[1]) || (cd->ctypes[ptr[1]] & ctype_word) == 0)
5763            {            {
5764            ptr += 1;  /* To get the right offset */            ptr += 1;  /* To get the right offset */
5765            *errorcodeptr = ERR28;            *errorcodeptr = ERR28;
# Line 5497  for (;; ptr++) Line 5770  for (;; ptr++)
5770    
5771          recno = 0;          recno = 0;
5772          name = ++ptr;          name = ++ptr;
5773          while ((cd->ctypes[*ptr] & ctype_word) != 0)          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0)
5774            {            {
5775            if (recno >= 0)            if (recno >= 0)
5776              recno = ((digitab[*ptr] & ctype_digit) != 0)?              recno = (IS_DIGIT(*ptr))? recno * 10 + *ptr - CHAR_0 : -1;
               recno * 10 + *ptr - CHAR_0 : -1;  
5777            ptr++;            ptr++;
5778            }            }
5779          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
# Line 5549  for (;; ptr++) Line 5821  for (;; ptr++)
5821          slot = cd->name_table;          slot = cd->name_table;
5822          for (i = 0; i < cd->names_found; i++)          for (i = 0; i < cd->names_found; i++)
5823            {            {
5824            if (strncmp((char *)name, (char *)slot+2, namelen) == 0) break;            if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) == 0) break;
5825            slot += cd->name_entry_size;            slot += cd->name_entry_size;
5826            }            }
5827    
# Line 5565  for (;; ptr++) Line 5837  for (;; ptr++)
5837          /* Search the pattern for a forward reference */          /* Search the pattern for a forward reference */
5838    
5839          else if ((i = find_parens(cd, name, namelen,          else if ((i = find_parens(cd, name, namelen,
5840                          (options & PCRE_EXTENDED) != 0, utf8)) > 0)                          (options & PCRE_EXTENDED) != 0, utf)) > 0)
5841            {            {
5842            PUT2(code, 2+LINK_SIZE, i);            PUT2(code, 2+LINK_SIZE, i);
5843            code[1+LINK_SIZE]++;            code[1+LINK_SIZE]++;
# Line 5591  for (;; ptr++) Line 5863  for (;; ptr++)
5863            recno = 0;            recno = 0;
5864            for (i = 1; i < namelen; i++)            for (i = 1; i < namelen; i++)
5865              {              {
5866              if ((digitab[name[i]] & ctype_digit) == 0)              if (!IS_DIGIT(name[i]))
5867                {                {
5868                *errorcodeptr = ERR15;                *errorcodeptr = ERR15;
5869                goto FAILED;                goto FAILED;
# Line 5606  for (;; ptr++) Line 5878  for (;; ptr++)
5878          /* Similarly, check for the (?(DEFINE) "condition", which is always          /* Similarly, check for the (?(DEFINE) "condition", which is always
5879          false. */          false. */
5880    
5881          else if (namelen == 6 && strncmp((char *)name, STRING_DEFINE, 6) == 0)          else if (namelen == 6 && STRNCMP_UC_C8(name, STRING_DEFINE, 6) == 0)
5882            {            {
5883            code[1+LINK_SIZE] = OP_DEF;            code[1+LINK_SIZE] = OP_DEF;
5884            skipbytes = 1;            skipbytes = 1;
# Line 5669  for (;; ptr++) Line 5941  for (;; ptr++)
5941            break;            break;
5942    
5943            default:                /* Could be name define, else bad */            default:                /* Could be name define, else bad */
5944            if ((cd->ctypes[ptr[1]] & ctype_word) != 0) goto DEFINE_NAME;            if (MAX_255(ptr[1]) && (cd->ctypes[ptr[1]] & ctype_word) != 0)
5945                goto DEFINE_NAME;
5946            ptr++;                  /* Correct offset for error */            ptr++;                  /* Correct offset for error */
5947            *errorcodeptr = ERR24;            *errorcodeptr = ERR24;
5948            goto FAILED;            goto FAILED;
# Line 5691  for (;; ptr++) Line 5964  for (;; ptr++)
5964          *code++ = OP_CALLOUT;          *code++ = OP_CALLOUT;
5965            {            {
5966            int n = 0;            int n = 0;
5967            while ((digitab[*(++ptr)] & ctype_digit) != 0)            ptr++;
5968              n = n * 10 + *ptr - CHAR_0;            while(IS_DIGIT(*ptr))
5969                n = n * 10 + *ptr++ - CHAR_0;
5970            if (*ptr != CHAR_RIGHT_PARENTHESIS)            if (*ptr != CHAR_RIGHT_PARENTHESIS)
5971              {              {
5972              *errorcodeptr = ERR39;              *errorcodeptr = ERR39;
# Line 5737  for (;; ptr++) Line 6011  for (;; ptr++)
6011              CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;              CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
6012            name = ++ptr;            name = ++ptr;
6013    
6014            while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;            while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
6015            namelen = (int)(ptr - name);            namelen = (int)(ptr - name);
6016    
6017            /* In the pre-compile phase, just do a syntax check. */            /* In the pre-compile phase, just do a syntax check. */
# Line 5754  for (;; ptr++) Line 6028  for (;; ptr++)
6028                *errorcodeptr = ERR49;                *errorcodeptr = ERR49;
6029                goto FAILED;                goto FAILED;
6030                }                }
6031              if (namelen + 3 > cd->name_entry_size)              if (namelen + IMM2_SIZE + 1 > cd->name_entry_size)
6032                {                {
6033                cd->name_entry_size = namelen + 3;                cd->name_entry_size = namelen + IMM2_SIZE + 1;
6034                if (namelen > MAX_NAME_SIZE)                if (namelen > MAX_NAME_SIZE)
6035                  {                  {
6036                  *errorcodeptr = ERR48;                  *errorcodeptr = ERR48;
# Line 5785  for (;; ptr++) Line 6059  for (;; ptr++)
6059    
6060              for (i = 0; i < cd->names_found; i++)              for (i = 0; i < cd->names_found; i++)
6061                {                {
6062                int crc = memcmp(name, slot+2, namelen);                int crc = memcmp(name, slot+IMM2_SIZE, IN_UCHARS(namelen));
6063                if (crc == 0)                if (crc == 0)
6064                  {                  {
6065                  if (slot[2+namelen] == 0)                  if (slot[IMM2_SIZE+namelen] == 0)
6066                    {                    {
6067                    if (GET2(slot, 0) != cd->bracount + 1 &&                    if (GET2(slot, 0) != cd->bracount + 1 &&
6068                        (options & PCRE_DUPNAMES) == 0)                        (options & PCRE_DUPNAMES) == 0)
# Line 5809  for (;; ptr++) Line 6083  for (;; ptr++)
6083                if (crc < 0)                if (crc < 0)
6084                  {                  {
6085                  memmove(slot + cd->name_entry_size, slot,                  memmove(slot + cd->name_entry_size, slot,
6086                    (cd->names_found - i) * cd->name_entry_size);                    IN_UCHARS((cd->names_found - i) * cd->name_entry_size));
6087                  break;                  break;
6088                  }                  }
6089    
# Line 5823  for (;; ptr++) Line 6097  for (;; ptr++)
6097    
6098              if (!dupname)              if (!dupname)
6099                {                {
6100                uschar *cslot = cd->name_table;                pcre_uchar *cslot = cd->name_table;
6101                for (i = 0; i < cd->names_found; i++)                for (i = 0; i < cd->names_found; i++)
6102                  {                  {
6103                  if (cslot != slot)                  if (cslot != slot)
# Line 5840  for (;; ptr++) Line 6114  for (;; ptr++)
6114                }                }
6115    
6116              PUT2(slot, 0, cd->bracount + 1);              PUT2(slot, 0, cd->bracount + 1);
6117              memcpy(slot + 2, name, namelen);              memcpy(slot + IMM2_SIZE, name, IN_UCHARS(namelen));
6118              slot[2+namelen] = 0;              slot[IMM2_SIZE + namelen] = 0;
6119              }              }
6120            }            }
6121    
# Line 5867  for (;; ptr++) Line 6141  for (;; ptr++)
6141    
6142          NAMED_REF_OR_RECURSE:          NAMED_REF_OR_RECURSE:
6143          name = ++ptr;          name = ++ptr;
6144          while ((cd->ctypes[*ptr] & ctype_word) != 0) ptr++;          while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_word) != 0) ptr++;
6145          namelen = (int)(ptr - name);          namelen = (int)(ptr - name);
6146    
6147          /* In the pre-compile phase, do a syntax check. We used to just set          /* In the pre-compile phase, do a syntax check. We used to just set
# Line 5879  for (;; ptr++) Line 6153  for (;; ptr++)
6153    
6154          if (lengthptr != NULL)          if (lengthptr != NULL)
6155            {            {
6156            const uschar *temp;            const pcre_uchar *temp;
6157    
6158            if (namelen == 0)            if (namelen == 0)
6159              {              {
# Line 5909  for (;; ptr++) Line 6183  for (;; ptr++)
6183            temp = cd->end_pattern;            temp = cd->end_pattern;
6184            cd->end_pattern = ptr;            cd->end_pattern = ptr;
6185            recno = find_parens(cd, name, namelen,            recno = find_parens(cd, name, namelen,
6186              (options & PCRE_EXTENDED) != 0, utf8);              (options & PCRE_EXTENDED) != 0, utf);
6187            cd->end_pattern = temp;            cd->end_pattern = temp;
6188            if (recno < 0) recno = 0;    /* Forward ref; set dummy number */            if (recno < 0) recno = 0;    /* Forward ref; set dummy number */
6189            }            }
# Line 5924  for (;; ptr++) Line 6198  for (;; ptr++)
6198            slot = cd->name_table;            slot = cd->name_table;
6199            for (i = 0; i < cd->names_found; i++)            for (i = 0; i < cd->names_found; i++)
6200              {              {
6201              if (strncmp((char *)name, (char *)slot+2, namelen) == 0 &&              if (STRNCMP_UC_UC(name, slot+IMM2_SIZE, namelen) == 0 &&
6202                  slot[2+namelen] == 0)                  slot[IMM2_SIZE+namelen] == 0)
6203                break;                break;
6204              slot += cd->name_entry_size;              slot += cd->name_entry_size;
6205              }              }
# Line 5936  for (;; ptr++) Line 6210  for (;; ptr++)
6210              }              }
6211            else if ((recno =                /* Forward back reference */            else if ((recno =                /* Forward back reference */
6212                      find_parens(cd, name, namelen,                      find_parens(cd, name, namelen,
6213                        (options & PCRE_EXTENDED) != 0, utf8)) <= 0)                        (options & PCRE_EXTENDED) != 0, utf)) <= 0)
6214              {              {
6215              *errorcodeptr = ERR15;              *errorcodeptr = ERR15;
6216              goto FAILED;              goto FAILED;
# Line 5961  for (;; ptr++) Line 6235  for (;; ptr++)
6235          case CHAR_0: case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4:          case CHAR_0: case CHAR_1: case CHAR_2: case CHAR_3: case CHAR_4:
6236          case CHAR_5: case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:          case CHAR_5: case CHAR_6: case CHAR_7: case CHAR_8: case CHAR_9:
6237            {            {
6238            const uschar *called;            const pcre_uchar *called;
6239            terminator = CHAR_RIGHT_PARENTHESIS;            terminator = CHAR_RIGHT_PARENTHESIS;
6240    
6241            /* Come here from the \g<...> and \g'...' code (Oniguruma            /* Come here from the \g<...> and \g'...' code (Oniguruma
# Line 5975  for (;; ptr++) Line 6249  for (;; ptr++)
6249            if ((refsign = *ptr) == CHAR_PLUS)            if ((refsign = *ptr) == CHAR_PLUS)
6250              {              {
6251              ptr++;              ptr++;
6252              if ((digitab[*ptr] & ctype_digit) == 0)              if (!IS_DIGIT(*ptr))
6253                {                {
6254                *errorcodeptr = ERR63;                *errorcodeptr = ERR63;
6255                goto FAILED;                goto FAILED;
# Line 5983  for (;; ptr++) Line 6257  for (;; ptr++)
6257              }              }
6258            else if (refsign == CHAR_MINUS)            else if (refsign == CHAR_MINUS)
6259              {              {
6260              if ((digitab[ptr[1]] & ctype_digit) == 0)              if (!IS_DIGIT(ptr[1]))
6261                goto OTHER_CHAR_AFTER_QUERY;                goto OTHER_CHAR_AFTER_QUERY;
6262              ptr++;              ptr++;
6263              }              }
6264    
6265            recno = 0;            recno = 0;
6266            while((digitab[*ptr] & ctype_digit) != 0)            while(IS_DIGIT(*ptr))
6267              recno = recno * 10 + *ptr++ - CHAR_0;              recno = recno * 10 + *ptr++ - CHAR_0;
6268    
6269            if (*ptr != terminator)            if (*ptr != terminator)
# Line 6040  for (;; ptr++) Line 6314  for (;; ptr++)
6314              {              {
6315              *code = OP_END;              *code = OP_END;
6316              if (recno != 0)              if (recno != 0)
6317                called = _pcre_find_bracket(cd->start_code, utf8, recno);                called = PRIV(find_bracket)(cd->start_code, utf, recno);
6318    
6319              /* Forward reference */              /* Forward reference */
6320    
6321              if (called == NULL)              if (called == NULL)
6322                {                {
6323                if (find_parens(cd, NULL, recno,                if (find_parens(cd, NULL, recno,
6324                      (options & PCRE_EXTENDED) != 0, utf8) < 0)                      (options & PCRE_EXTENDED) != 0, utf) < 0)
6325                  {                  {
6326                  *errorcodeptr = ERR15;                  *errorcodeptr = ERR15;
6327                  goto FAILED;                  goto FAILED;
# Line 6056  for (;; ptr++) Line 6330  for (;; ptr++)
6330                /* Fudge the value of "called" so that when it is inserted as an                /* Fudge the value of "called" so that when it is inserted as an
6331                offset below, what it actually inserted is the reference number                offset below, what it actually inserted is the reference number
6332                of the group. Then remember the forward reference. */                of the group. Then remember the forward reference. */
6333    
6334                called = cd->start_code + recno;                called = cd->start_code + recno;
6335                if (cd->hwm >= cd->start_workspace + cd->workspace_size -                if (cd->hwm >= cd->start_workspace + cd->workspace_size -
6336                    WORK_SIZE_SAFETY_MARGIN)                    WORK_SIZE_SAFETY_MARGIN)
6337                  {                  {
6338                  *errorcodeptr = expand_workspace(cd);                  *errorcodeptr = expand_workspace(cd);
6339                  if (*errorcodeptr != 0) goto FAILED;                  if (*errorcodeptr != 0) goto FAILED;
6340                  }                  }
6341                PUTINC(cd->hwm, 0, (int)(code + 1 - cd->start_code));                PUTINC(cd->hwm, 0, (int)(code + 1 - cd->start_code));
6342                }                }
6343    
# Line 6077  for (;; ptr++) Line 6351  for (;; ptr++)
6351              conditional subpatterns will be picked up then. */              conditional subpatterns will be picked up then. */
6352    
6353              else if (GET(called, 1) == 0 && cond_depth <= 0 &&              else if (GET(called, 1) == 0 && cond_depth <= 0 &&
6354                       could_be_empty(called, code, bcptr, utf8, cd))                       could_be_empty(called, code, bcptr, utf, cd))
6355                {                {
6356                *errorcodeptr = ERR40;                *errorcodeptr = ERR40;
6357                goto FAILED;                goto FAILED;
# Line 6085  for (;; ptr++) Line 6359  for (;; ptr++)
6359              }              }
6360    
6361            /* Insert the recursion/subroutine item. It does not have a set first            /* Insert the recursion/subroutine item. It does not have a set first
6362            byte (relevant if it is repeated, because it will then be wrapped            character (relevant if it is repeated, because it will then be
6363            with ONCE brackets). */            wrapped with ONCE brackets). */
6364    
6365            *code = OP_RECURSE;            *code = OP_RECURSE;
6366            PUT(code, 1, (int)(called - cd->start_code));            PUT(code, 1, (int)(called - cd->start_code));
6367            code += 1 + LINK_SIZE;            code += 1 + LINK_SIZE;
6368            groupsetfirstbyte = FALSE;            groupsetfirstchar = FALSE;
6369            }            }
6370    
6371          /* Can't determine a first byte now */          /* Can't determine a first byte now */
6372    
6373          if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
6374          continue;          continue;
6375    
6376    
# Line 6153  for (;; ptr++) Line 6427  for (;; ptr++)
6427          both phases.          both phases.
6428    
6429          If we are not at the pattern start, reset the greedy defaults and the          If we are not at the pattern start, reset the greedy defaults and the
6430          case value for firstbyte and reqbyte. */          case value for firstchar and reqchar. */
6431    
6432          if (*ptr == CHAR_RIGHT_PARENTHESIS)          if (*ptr == CHAR_RIGHT_PARENTHESIS)
6433            {            {
# Line 6166  for (;; ptr++) Line 6440  for (;; ptr++)
6440              {              {
6441              greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);              greedy_default = ((newoptions & PCRE_UNGREEDY) != 0);
6442              greedy_non_default = greedy_default ^ 1;              greedy_non_default = greedy_default ^ 1;
6443              req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;              req_caseopt = ((newoptions & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
6444              }              }
6445    
6446            /* Change options at this level, and pass them back for use            /* Change options at this level, and pass them back for use
# Line 6203  for (;; ptr++) Line 6477  for (;; ptr++)
6477        NUMBERED_GROUP:        NUMBERED_GROUP:
6478        cd->bracount += 1;        cd->bracount += 1;
6479        PUT2(code, 1+LINK_SIZE, cd->bracount);        PUT2(code, 1+LINK_SIZE, cd->bracount);
6480        skipbytes = 2;        skipbytes = IMM2_SIZE;
6481        }        }
6482    
6483      /* Process nested bracketed regex. Assertions used not to be repeatable,      /* Process nested bracketed regex. Assertions used not to be repeatable,
# Line 6229  for (;; ptr++) Line 6503  for (;; ptr++)
6503           skipbytes,                       /* Skip over bracket number */           skipbytes,                       /* Skip over bracket number */
6504           cond_depth +           cond_depth +
6505             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */             ((bravalue == OP_COND)?1:0),   /* Depth of condition subpatterns */
6506           &subfirstbyte,                   /* For possible first char */           &subfirstchar,                   /* For possible first char */
6507           &subreqbyte,                     /* For possible last char */           &subreqchar,                     /* For possible last char */
6508           bcptr,                           /* Current branch chain */           bcptr,                           /* Current branch chain */
6509           cd,                              /* Tables block */           cd,                              /* Tables block */
6510           (lengthptr == NULL)? NULL :      /* Actual compile phase */           (lengthptr == NULL)? NULL :      /* Actual compile phase */
# Line 6258  for (;; ptr++) Line 6532  for (;; ptr++)
6532    
6533      if (bravalue == OP_COND && lengthptr == NULL)      if (bravalue == OP_COND && lengthptr == NULL)
6534        {        {
6535        uschar *tc = code;        pcre_uchar *tc = code;
6536        int condcount = 0;        int condcount = 0;
6537    
6538        do {        do {
# Line 6281  for (;; ptr++) Line 6555  for (;; ptr++)
6555          }          }
6556    
6557        /* A "normal" conditional group. If there is just one branch, we must not        /* A "normal" conditional group. If there is just one branch, we must not
6558        make use of its firstbyte or reqbyte, because this is equivalent to an        make use of its firstchar or reqchar, because this is equivalent to an
6559        empty second branch. */        empty second branch. */
6560    
6561        else        else
# Line 6291  for (;; ptr++) Line 6565  for (;; ptr++)
6565            *errorcodeptr = ERR27;            *errorcodeptr = ERR27;
6566            goto FAILED;            goto FAILED;
6567            }            }
6568          if (condcount == 1) subfirstbyte = subreqbyte = REQ_NONE;          if (condcount == 1) subfirstchar = subreqchar = REQ_NONE;
6569          }          }
6570        }        }
6571    
# Line 6335  for (;; ptr++) Line 6609  for (;; ptr++)
6609      /* Handle updating of the required and first characters for other types of      /* Handle updating of the required and first characters for other types of
6610      group. Update for normal brackets of all kinds, and conditions with two      group. Update for normal brackets of all kinds, and conditions with two
6611      branches (see code above). If the bracket is followed by a quantifier with      branches (see code above). If the bracket is followed by a quantifier with
6612      zero repeat, we have to back off. Hence the definition of zeroreqbyte and      zero repeat, we have to back off. Hence the definition of zeroreqchar and
6613      zerofirstbyte outside the main loop so that they can be accessed for the      zerofirstchar outside the main loop so that they can be accessed for the
6614      back off. */      back off. */
6615    
6616      zeroreqbyte = reqbyte;      zeroreqchar = reqchar;
6617      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
6618      groupsetfirstbyte = FALSE;      groupsetfirstchar = FALSE;
6619    
6620      if (bravalue >= OP_ONCE)      if (bravalue >= OP_ONCE)
6621        {        {
6622        /* If we have not yet set a firstbyte in this branch, take it from the        /* If we have not yet set a firstchar in this branch, take it from the
6623        subpattern, remembering that it was set here so that a repeat of more        subpattern, remembering that it was set here so that a repeat of more
6624        than one can replicate it as reqbyte if necessary. If the subpattern has        than one can replicate it as reqchar if necessary. If the subpattern has
6625        no firstbyte, set "none" for the whole branch. In both cases, a zero        no firstchar, set "none" for the whole branch. In both cases, a zero
6626        repeat forces firstbyte to "none". */        repeat forces firstchar to "none". */
6627    
6628        if (firstbyte == REQ_UNSET)        if (firstchar == REQ_UNSET)
6629          {          {
6630          if (subfirstbyte >= 0)          if (subfirstchar >= 0)
6631            {            {
6632            firstbyte = subfirstbyte;            firstchar = subfirstchar;
6633            groupsetfirstbyte = TRUE;            groupsetfirstchar = TRUE;
6634            }            }
6635          else firstbyte = REQ_NONE;          else firstchar = REQ_NONE;
6636          zerofirstbyte = REQ_NONE;          zerofirstchar = REQ_NONE;
6637          }          }
6638    
6639        /* If firstbyte was previously set, convert the subpattern's firstbyte        /* If firstchar was previously set, convert the subpattern's firstchar
6640        into reqbyte if there wasn't one, using the vary flag that was in        into reqchar if there wasn't one, using the vary flag that was in
6641        existence beforehand. */        existence beforehand. */
6642    
6643        else if (subfirstbyte >= 0 && subreqbyte < 0)        else if (subfirstchar >= 0 && subreqchar < 0)
6644          subreqbyte = subfirstbyte | tempreqvary;          subreqchar = subfirstchar | tempreqvary;
6645    
6646        /* If the subpattern set a required byte (or set a first byte that isn't        /* If the subpattern set a required byte (or set a first byte that isn't
6647        really the first byte - see above), set it. */        really the first byte - see above), set it. */
6648    
6649        if (subreqbyte >= 0) reqbyte = subreqbyte;        if (subreqchar >= 0) reqchar = subreqchar;
6650        }        }
6651    
6652      /* For a forward assertion, we take the reqbyte, if set. This can be      /* For a forward assertion, we take the reqchar, if set. This can be
6653      helpful if the pattern that follows the assertion doesn't set a different      helpful if the pattern that follows the assertion doesn't set a different
6654      char. For example, it's useful for /(?=abcde).+/. We can't set firstbyte      char. For example, it's useful for /(?=abcde).+/. We can't set firstchar
6655      for an assertion, however because it leads to incorrect effect for patterns      for an assertion, however because it leads to incorrect effect for patterns
6656      such as /(?=a)a.+/ when the "real" "a" would then become a reqbyte instead      such as /(?=a)a.+/ when the "real" "a" would then become a reqchar instead
6657      of a firstbyte. This is overcome by a scan at the end if there's no      of a firstchar. This is overcome by a scan at the end if there's no
6658      firstbyte, looking for an asserted first char. */      firstchar, looking for an asserted first char. */
6659    
6660      else if (bravalue == OP_ASSERT && subreqbyte >= 0) reqbyte = subreqbyte;      else if (bravalue == OP_ASSERT && subreqchar >= 0) reqchar = subreqchar;
6661      break;     /* End of processing '(' */      break;     /* End of processing '(' */
6662    
6663    
# Line 6416  for (;; ptr++) Line 6690  for (;; ptr++)
6690        /* For metasequences that actually match a character, we disable the        /* For metasequences that actually match a character, we disable the
6691        setting of a first character if it hasn't already been set. */        setting of a first character if it hasn't already been set. */
6692    
6693        if (firstbyte == REQ_UNSET && -c > ESC_b && -c < ESC_Z)        if (firstchar == REQ_UNSET && -c > ESC_b && -c < ESC_Z)
6694          firstbyte = REQ_NONE;          firstchar = REQ_NONE;
6695    
6696        /* Set values to reset to if this is followed by a zero repeat. */        /* Set values to reset to if this is followed by a zero repeat. */
6697    
6698        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
6699        zeroreqbyte = reqbyte;        zeroreqchar = reqchar;
6700    
6701        /* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'        /* \g<name> or \g'name' is a subroutine call by name and \g<n> or \g'n'
6702        is a subroutine call by number (Oniguruma syntax). In fact, the value        is a subroutine call by number (Oniguruma syntax). In fact, the value
# Line 6433  for (;; ptr++) Line 6707  for (;; ptr++)
6707    
6708        if (-c == ESC_g)        if (-c == ESC_g)
6709          {          {
6710          const uschar *p;          const pcre_uchar *p;
6711          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */          save_hwm = cd->hwm;   /* Normally this is set when '(' is read */
6712          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?          terminator = (*(++ptr) == CHAR_LESS_THAN_SIGN)?
6713            CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;            CHAR_GREATER_THAN_SIGN : CHAR_APOSTROPHE;
# Line 6450  for (;; ptr++) Line 6724  for (;; ptr++)
6724    
6725          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)          if (ptr[1] != CHAR_PLUS && ptr[1] != CHAR_MINUS)
6726            {            {
6727            BOOL isnumber = TRUE;            BOOL is_a_number = TRUE;
6728            for (p = ptr + 1; *p != 0 && *p != terminator; p++)            for (p = ptr + 1; *p != 0 && *p != terminator; p++)
6729              {              {
6730              if ((cd->ctypes[*p] & ctype_digit) == 0) isnumber = FALSE;              if (!MAX_255(*p)) { is_a_number = FALSE; break; }
6731                if ((cd->ctypes[*p] & ctype_digit) == 0) is_a_number = FALSE;
6732              if ((cd->ctypes[*p] & ctype_word) == 0) break;              if ((cd->ctypes[*p] & ctype_word) == 0) break;
6733              }              }
6734            if (*p != terminator)            if (*p != terminator)
# Line 6461  for (;; ptr++) Line 6736  for (;; ptr++)
6736              *errorcodeptr = ERR57;              *errorcodeptr = ERR57;
6737              break;              break;
6738              }              }
6739            if (isnumber)            if (is_a_number)
6740              {              {
6741              ptr++;              ptr++;
6742              goto HANDLE_NUMERICAL_RECURSION;              goto HANDLE_NUMERICAL_RECURSION;
# Line 6473  for (;; ptr++) Line 6748  for (;; ptr++)
6748          /* Test a signed number in angle brackets or quotes. */          /* Test a signed number in angle brackets or quotes. */
6749    
6750          p = ptr + 2;          p = ptr + 2;
6751          while ((digitab[*p] & ctype_digit) != 0) p++;          while (IS_DIGIT(*p)) p++;
6752          if (*p != terminator)          if (*p != terminator)
6753            {            {
6754            *errorcodeptr = ERR57;            *errorcodeptr = ERR57;
# Line 6501  for (;; ptr++) Line 6776  for (;; ptr++)
6776          goto NAMED_REF_OR_RECURSE;          goto NAMED_REF_OR_RECURSE;
6777          }          }
6778    
6779        /* Back references are handled specially; must disable firstbyte if        /* Back references are handled specially; must disable firstchar if
6780        not set to cope with cases like (?=(\w+))\1: which would otherwise set        not set to cope with cases like (?=(\w+))\1: which would otherwise set
6781        ':' later. */        ':' later. */
6782    
# Line 6511  for (;; ptr++) Line 6786  for (;; ptr++)
6786          recno = -c - ESC_REF;          recno = -c - ESC_REF;
6787    
6788          HANDLE_REFERENCE:    /* Come here from named backref handling */          HANDLE_REFERENCE:    /* Come here from named backref handling */
6789          if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;          if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
6790          previous = code;          previous = code;
6791          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;          *code++ = ((options & PCRE_CASELESS) != 0)? OP_REFI : OP_REF;
6792          PUT2INC(code, 0, recno);          PUT2INC(code, 0, recno);
# Line 6575  for (;; ptr++) Line 6850  for (;; ptr++)
6850  #endif  #endif
6851          /* In non-UTF-8 mode, we turn \C into OP_ALLANY instead of OP_ANYBYTE          /* In non-UTF-8 mode, we turn \C into OP_ALLANY instead of OP_ANYBYTE
6852          so that it works in DFA mode and in lookbehinds. */          so that it works in DFA mode and in lookbehinds. */
6853    
6854            {            {
6855            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;            previous = (-c > ESC_b && -c < ESC_Z)? code : NULL;
6856            *code++ = (!utf8 && c == -ESC_C)? OP_ALLANY : -c;            *code++ = (!utf && c == -ESC_C)? OP_ALLANY : -c;
6857            }            }
6858          }          }
6859        continue;        continue;
# Line 6588  for (;; ptr++) Line 6863  for (;; ptr++)
6863