/[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 1305 by ph10, Mon Apr 1 14:50:45 2013 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|32)_printint() function, which
57  also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. */  library. We do not need to select pcre16_printint.c specially, because the
59    COMPILE_PCREx macro will already be appropriately set. */
60    
61  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
62  #include "pcre_printint.src"  /* pcre_printint.c should not include any headers */
63    #define PCRE_INCLUDED
64    #include "pcre_printint.c"
65    #undef PCRE_INCLUDED
66  #endif  #endif
67    
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 73  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,
88        pcre_uint32 *, pcre_int32 *, pcre_uint32 *, pcre_int32 *, branch_chain *,
89        compile_data *, int *);
90    
91    
92    
93  /*************************************************  /*************************************************
94  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 88  so this number is very generous. Line 104  so this number is very generous.
104  The same workspace is used during the second, actual compile phase for  The same workspace is used during the second, actual compile phase for
105  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
106  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
107  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
108  filled up by repetitions of forward references, for example patterns like  filled up by repetitions of forward references, for example patterns like
109  /(?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
110  that the workspace is expanded using malloc() in this situation. The value  that the workspace is expanded using malloc() in this situation. The value
111  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
112  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()
113  kicks in at the same number of forward references in all cases. */  kicks in at the same number of forward references in all cases. */
114    
115  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)
# Line 104  overrun before it actually does run off Line 120  overrun before it actually does run off
120    
121  #define WORK_SIZE_SAFETY_MARGIN (100)  #define WORK_SIZE_SAFETY_MARGIN (100)
122    
123    /* Private flags added to firstchar and reqchar. */
124    
125    #define REQ_CASELESS    (1 << 0)        /* Indicates caselessness */
126    #define REQ_VARY        (1 << 1)        /* Reqchar followed non-literal item */
127    /* Negative values for the firstchar and reqchar flags */
128    #define REQ_UNSET       (-2)
129    #define REQ_NONE        (-1)
130    
131    /* Repeated character flags. */
132    
133    #define UTF_LENGTH     0x10000000l      /* The char contains its length. */
134    
135  /* Table for handling escaped characters in the range '0'-'z'. Positive returns  /* Table for handling escaped characters in the range '0'-'z'. Positive returns
136  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 265  static const char posix_names[] =
265    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0
266    STRING_word0  STRING_xdigit;    STRING_word0  STRING_xdigit;
267    
268  static const uschar posix_name_lengths[] = {  static const pcre_uint8 posix_name_lengths[] = {
269    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 };
270    
271  /* 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 300  substitutes must be in the order of the
300  both positive and negative cases. NULL means no substitute. */  both positive and negative cases. NULL means no substitute. */
301    
302  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
303  static const uschar *substitutes[] = {  static const pcre_uchar string_PNd[]  = {
304    (uschar *)"\\P{Nd}",    /* \D */    CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
305    (uschar *)"\\p{Nd}",    /* \d */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
306    (uschar *)"\\P{Xsp}",   /* \S */       /* NOTE: Xsp is Perl space */  static const pcre_uchar string_pNd[]  = {
307    (uschar *)"\\p{Xsp}",   /* \s */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
308    (uschar *)"\\P{Xwd}",   /* \W */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
309    (uschar *)"\\p{Xwd}"    /* \w */  static const pcre_uchar string_PXsp[] = {
310      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
311      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
312    static const pcre_uchar string_pXsp[] = {
313      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
314      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
315    static const pcre_uchar string_PXwd[] = {
316      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
317      CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
318    static const pcre_uchar string_pXwd[] = {
319      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
320      CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
321    
322    static const pcre_uchar *substitutes[] = {
323      string_PNd,           /* \D */
324      string_pNd,           /* \d */
325      string_PXsp,          /* \S */       /* NOTE: Xsp is Perl space */
326      string_pXsp,          /* \s */
327      string_PXwd,          /* \W */
328      string_pXwd           /* \w */
329  };  };
330    
331  static const uschar *posix_substitutes[] = {  static const pcre_uchar string_pL[] =   {
332    (uschar *)"\\p{L}",     /* alpha */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
333    (uschar *)"\\p{Ll}",    /* lower */    CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
334    (uschar *)"\\p{Lu}",    /* upper */  static const pcre_uchar string_pLl[] =  {
335    (uschar *)"\\p{Xan}",   /* alnum */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
336    NULL,                   /* ascii */    CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
337    (uschar *)"\\h",        /* blank */  static const pcre_uchar string_pLu[] =  {
338    NULL,                   /* cntrl */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
339    (uschar *)"\\p{Nd}",    /* digit */    CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
340    NULL,                   /* graph */  static const pcre_uchar string_pXan[] = {
341    NULL,                   /* print */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
342    NULL,                   /* punct */    CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
343    (uschar *)"\\p{Xps}",   /* space */    /* NOTE: Xps is POSIX space */  static const pcre_uchar string_h[] =    {
344    (uschar *)"\\p{Xwd}",   /* word */    CHAR_BACKSLASH, CHAR_h, '\0' };
345    NULL,                   /* xdigit */  static const pcre_uchar string_pXps[] = {
346      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
347      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
348    static const pcre_uchar string_PL[] =   {
349      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
350      CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
351    static const pcre_uchar string_PLl[] =  {
352      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
353      CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
354    static const pcre_uchar string_PLu[] =  {
355      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
356      CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
357    static const pcre_uchar string_PXan[] = {
358      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
359      CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
360    static const pcre_uchar string_H[] =    {
361      CHAR_BACKSLASH, CHAR_H, '\0' };
362    static const pcre_uchar string_PXps[] = {
363      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
364      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
365    
366    static const pcre_uchar *posix_substitutes[] = {
367      string_pL,            /* alpha */
368      string_pLl,           /* lower */
369      string_pLu,           /* upper */
370      string_pXan,          /* alnum */
371      NULL,                 /* ascii */
372      string_h,             /* blank */
373      NULL,                 /* cntrl */
374      string_pNd,           /* digit */
375      NULL,                 /* graph */
376      NULL,                 /* print */
377      NULL,                 /* punct */
378      string_pXps,          /* space */    /* NOTE: Xps is POSIX space */
379      string_pXwd,          /* word */
380      NULL,                 /* xdigit */
381    /* Negated cases */    /* Negated cases */
382    (uschar *)"\\P{L}",     /* ^alpha */    string_PL,            /* ^alpha */
383    (uschar *)"\\P{Ll}",    /* ^lower */    string_PLl,           /* ^lower */
384    (uschar *)"\\P{Lu}",    /* ^upper */    string_PLu,           /* ^upper */
385    (uschar *)"\\P{Xan}",   /* ^alnum */    string_PXan,          /* ^alnum */
386    NULL,                   /* ^ascii */    NULL,                 /* ^ascii */
387    (uschar *)"\\H",        /* ^blank */    string_H,             /* ^blank */
388    NULL,                   /* ^cntrl */    NULL,                 /* ^cntrl */
389    (uschar *)"\\P{Nd}",    /* ^digit */    string_PNd,           /* ^digit */
390    NULL,                   /* ^graph */    NULL,                 /* ^graph */
391    NULL,                   /* ^print */    NULL,                 /* ^print */
392    NULL,                   /* ^punct */    NULL,                 /* ^punct */
393    (uschar *)"\\P{Xps}",   /* ^space */   /* NOTE: Xps is POSIX space */    string_PXps,          /* ^space */   /* NOTE: Xps is POSIX space */
394    (uschar *)"\\P{Xwd}",   /* ^word */    string_PXwd,          /* ^word */
395    NULL                    /* ^xdigit */    NULL                  /* ^xdigit */
396  };  };
397  #define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))  #define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))
398  #endif  #endif
399    
400  #define STRING(a)  # a  #define STRING(a)  # a
# Line 372  static const char error_texts[] = Line 453  static const char error_texts[] =
453    /* 30 */    /* 30 */
454    "unknown POSIX class name\0"    "unknown POSIX class name\0"
455    "POSIX collating elements are not supported\0"    "POSIX collating elements are not supported\0"
456    "this version of PCRE is not compiled with PCRE_UTF8 support\0"    "this version of PCRE is compiled without UTF support\0"
457    "spare error\0"  /** DEAD **/    "spare error\0"  /** DEAD **/
458    "character value in \\x{...} sequence is too large\0"    "character value in \\x{...} sequence is too large\0"
459    /* 35 */    /* 35 */
# Line 395  static const char error_texts[] = Line 476  static const char error_texts[] =
476    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
477    /* 50 */    /* 50 */
478    "repeated subpattern is too long\0"    /** DEAD **/    "repeated subpattern is too long\0"    /** DEAD **/
479    "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"
480    "internal error: overran compiling workspace\0"    "internal error: overran compiling workspace\0"
481    "internal error: previously-checked referenced subpattern not found\0"    "internal error: previously-checked referenced subpattern not found\0"
482    "DEFINE group contains more than one branch\0"    "DEFINE group contains more than one branch\0"
# Line 414  static const char error_texts[] = Line 495  static const char error_texts[] =
495    /* 65 */    /* 65 */
496    "different names for subpatterns of the same number are not allowed\0"    "different names for subpatterns of the same number are not allowed\0"
497    "(*MARK) must have an argument\0"    "(*MARK) must have an argument\0"
498    "this version of PCRE is not compiled with PCRE_UCP support\0"    "this version of PCRE is not compiled with Unicode property support\0"
499    "\\c must be followed by an ASCII character\0"    "\\c must be followed by an ASCII character\0"
500    "\\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"
501    /* 70 */    /* 70 */
502    "internal error: unknown opcode in find_fixedlength()\0"    "internal error: unknown opcode in find_fixedlength()\0"
503    "\\N is not supported in a class\0"    "\\N is not supported in a class\0"
504    "too many forward references\0"    "too many forward references\0"
505      "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
506      "invalid UTF-16 string\0"
507      /* 75 */
508      "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
509      "character value in \\u.... sequence is too large\0"
510      "invalid UTF-32 string\0"
511    ;    ;
512    
513  /* 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 526  For convenience, we use the same bit def
526    
527  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
528    
529    /* Using a simple comparison for decimal numbers rather than a memory read
530    is much faster, and the resulting code is simpler (the compiler turns it
531    into a subtraction and unsigned comparison). */
532    
533    #define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9)
534    
535  #ifndef EBCDIC  #ifndef EBCDIC
536    
537  /* 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
538  UTF-8 mode. */  UTF-8 mode. */
539    
540  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
541    {    {
542    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
543    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 576  static const unsigned char digitab[] =
576    
577  /* 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. */
578    
579  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
580    {    {
581    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
582    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 611  static const unsigned char digitab[] =
611    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */
612    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */
613    
614  static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */  static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */
615    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */
616    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */
617    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */
# Line 554  static const unsigned char ebcdic_charta Line 647  static const unsigned char ebcdic_charta
647  #endif  #endif
648    
649    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int,  
     int *, int *, branch_chain *, compile_data *, int *);  
   
   
650    
651  /*************************************************  /*************************************************
652  *            Find an error text                  *  *            Find an error text                  *
# Line 581  find_error_text(int n) Line 667  find_error_text(int n)
667  const char *s = error_texts;  const char *s = error_texts;
668  for (; n > 0; n--)  for (; n > 0; n--)
669    {    {
670    while (*s++ != 0) {};    while (*s++ != CHAR_NULL) {};
671    if (*s == 0) return "Error text not found (please report)";    if (*s == CHAR_NULL) return "Error text not found (please report)";
672    }    }
673  return s;  return s;
674  }  }
# Line 592  return s; Line 678  return s;
678  *           Expand the workspace                 *  *           Expand the workspace                 *
679  *************************************************/  *************************************************/
680    
681  /* 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
682  forward references fills the existing workspace, which is originally a block on  forward references fills the existing workspace, which is originally a block on
683  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
684  has been reached or the increase will be rather small.  has been reached or the increase will be rather small.
685    
686  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 690  Returns:  0 if all went well, else an er
690  static int  static int
691  expand_workspace(compile_data *cd)  expand_workspace(compile_data *cd)
692  {  {
693  uschar *newspace;  pcre_uchar *newspace;
694  int newsize = cd->workspace_size * 2;  int newsize = cd->workspace_size * 2;
695    
696  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 698  if (cd->workspace_size >= COMPILE_WORK_S
698      newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)      newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)
699   return ERR72;   return ERR72;
700    
701  newspace = (pcre_malloc)(newsize);  newspace = (PUBL(malloc))(IN_UCHARS(newsize));
702  if (newspace == NULL) return ERR21;  if (newspace == NULL) return ERR21;
703    memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar));
704  memcpy(newspace, cd->start_workspace, cd->workspace_size);  cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace);
705  cd->hwm = (uschar *)newspace + (cd->hwm - cd->start_workspace);  if (cd->workspace_size > COMPILE_WORK_SIZE)
706  if (cd->workspace_size > COMPILE_WORK_SIZE)    (PUBL(free))((void *)cd->start_workspace);
   (pcre_free)((void *)cd->start_workspace);  
707  cd->start_workspace = newspace;  cd->start_workspace = newspace;
708  cd->workspace_size = newsize;  cd->workspace_size = newsize;
709  return 0;  return 0;
# Line 642  Returns:    TRUE or FALSE Line 727  Returns:    TRUE or FALSE
727  */  */
728    
729  static BOOL  static BOOL
730  is_counted_repeat(const uschar *p)  is_counted_repeat(const pcre_uchar *p)
731  {  {
732  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
733  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
734    while (IS_DIGIT(*p)) p++;
735  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
736    
737  if (*p++ != CHAR_COMMA) return FALSE;  if (*p++ != CHAR_COMMA) return FALSE;
738  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
739    
740  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
741  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
742    while (IS_DIGIT(*p)) p++;
743    
744  return (*p == CHAR_RIGHT_CURLY_BRACKET);  return (*p == CHAR_RIGHT_CURLY_BRACKET);
745  }  }
# Line 664  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 751  return (*p == CHAR_RIGHT_CURLY_BRACKET);
751  *************************************************/  *************************************************/
752    
753  /* This function is called when a \ has been encountered. It either returns a  /* This function is called when a \ has been encountered. It either returns a
754  positive value for a simple escape such as \n, or a negative value which  positive value for a simple escape such as \n, or 0 for a data character
755  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group n is returned as
756  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  negative n. When UTF-8 is enabled, a positive value greater than 255 may
757  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
758  ptr is pointing at the \. On exit, it is on the final character of the escape  On entry,ptr is pointing at the \. On exit, it is on the final character of the
759  sequence.  escape sequence.
760    
761  Arguments:  Arguments:
762    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
763      chptr          points to the data character
764    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
765    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
766    options        the options bits    options        the options bits
767    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
768    
769  Returns:         zero or positive => a data character  Returns:         zero => a data character
770                   negative => a special escape sequence                   positive => a special escape sequence
771                     negative => a back reference
772                   on error, errorcodeptr is set                   on error, errorcodeptr is set
773  */  */
774    
775  static int  static int
776  check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
777    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
778  {  {
779  BOOL utf8 = (options & PCRE_UTF8) != 0;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
780  const uschar *ptr = *ptrptr + 1;  BOOL utf = (options & PCRE_UTF8) != 0;
781  int c, i;  const pcre_uchar *ptr = *ptrptr + 1;
782    pcre_uint32 c;
783    int escape = 0;
784    int i;
785    
786  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
787  ptr--;                            /* Set pointer back to the last byte */  ptr--;                            /* Set pointer back to the last byte */
788    
789  /* If backslash is at the end of the pattern, it's an error. */  /* If backslash is at the end of the pattern, it's an error. */
790    
791  if (c == 0) *errorcodeptr = ERR1;  if (c == CHAR_NULL) *errorcodeptr = ERR1;
792    
793  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
794  in a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
795  Otherwise further processing may be required. */  Otherwise further processing may be required. */
796    
797  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
798  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */  /* Not alphanumeric */
799  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if (c < CHAR_0 || c > CHAR_z) {}
800    else if ((i = escapes[c - CHAR_0]) != 0)
801      { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
802    
803  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
804  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */  /* Not alphanumeric */
805  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
806    else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
807  #endif  #endif
808    
809  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
810    
811  else  else
812    {    {
813    const uschar *oldptr;    const pcre_uchar *oldptr;
814    BOOL braced, negated;    BOOL braced, negated, overflow;
815      int s;
816    
817    switch (c)    switch (c)
818      {      {
# Line 733  else Line 829  else
829        {        {
830        /* In JavaScript, \u must be followed by four hexadecimal numbers.        /* In JavaScript, \u must be followed by four hexadecimal numbers.
831        Otherwise it is a lowercase u letter. */        Otherwise it is a lowercase u letter. */
832        if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
833             && (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0)          && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
834            && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
835            && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
836          {          {
837          c = 0;          c = 0;
838          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
839            {            {
840            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
841  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
842            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
843            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 748  else Line 846  else
846            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
847  #endif  #endif
848            }            }
849    
850    #if defined COMPILE_PCRE8
851            if (c > (utf ? 0x10ffffU : 0xffU))
852    #elif defined COMPILE_PCRE16
853            if (c > (utf ? 0x10ffffU : 0xffffU))
854    #elif defined COMPILE_PCRE32
855            if (utf && c > 0x10ffffU)
856    #endif
857              {
858              *errorcodeptr = ERR76;
859              }
860            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
861          }          }
862        }        }
863      else      else
# Line 774  else Line 884  else
884      (3) For Oniguruma compatibility we also support \g followed by a name or a      (3) For Oniguruma compatibility we also support \g followed by a name or a
885      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
886      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
887      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
888    
889      case CHAR_g:      case CHAR_g:
890      if (isclass) break;      if (isclass) break;
891      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
892        {        {
893        c = -ESC_g;        escape = ESC_g;
894        break;        break;
895        }        }
896    
# Line 788  else Line 898  else
898    
899      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
900        {        {
901        const uschar *p;        const pcre_uchar *p;
902        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
903          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
904        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET)
905          {          {
906          c = -ESC_k;          escape = ESC_k;
907          break;          break;
908          }          }
909        braced = TRUE;        braced = TRUE;
# Line 808  else Line 918  else
918        }        }
919      else negated = FALSE;      else negated = FALSE;
920    
921      c = 0;      /* The integer range is limited by the machine's int representation. */
922      while ((digitab[ptr[1]] & ctype_digit) != 0)      s = 0;
923        c = c * 10 + *(++ptr) - CHAR_0;      overflow = FALSE;
924        while (IS_DIGIT(ptr[1]))
     if (c < 0)   /* Integer overflow */  
925        {        {
926          if (s > INT_MAX / 10 - 1) /* Integer overflow */
927            {
928            overflow = TRUE;
929            break;
930            }
931          s = s * 10 + (int)(*(++ptr) - CHAR_0);
932          }
933        if (overflow) /* Integer overflow */
934          {
935          while (IS_DIGIT(ptr[1]))
936            ptr++;
937        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
938        break;        break;
939        }        }
# Line 824  else Line 944  else
944        break;        break;
945        }        }
946    
947      if (c == 0)      if (s == 0)
948        {        {
949        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
950        break;        break;
# Line 832  else Line 952  else
952    
953      if (negated)      if (negated)
954        {        {
955        if (c > bracount)        if (s > bracount)
956          {          {
957          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
958          break;          break;
959          }          }
960        c = bracount - (c - 1);        s = bracount - (s - 1);
961        }        }
962    
963      c = -(ESC_REF + c);      escape = -s;
964      break;      break;
965    
966      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 861  else Line 981  else
981      if (!isclass)      if (!isclass)
982        {        {
983        oldptr = ptr;        oldptr = ptr;
984        c -= CHAR_0;        /* The integer range is limited by the machine's int representation. */
985        while ((digitab[ptr[1]] & ctype_digit) != 0)        s = (int)(c -CHAR_0);
986          c = c * 10 + *(++ptr) - CHAR_0;        overflow = FALSE;
987        if (c < 0)    /* Integer overflow */        while (IS_DIGIT(ptr[1]))
988          {          {
989            if (s > INT_MAX / 10 - 1) /* Integer overflow */
990              {
991              overflow = TRUE;
992              break;
993              }
994            s = s * 10 + (int)(*(++ptr) - CHAR_0);
995            }
996          if (overflow) /* Integer overflow */
997            {
998            while (IS_DIGIT(ptr[1]))
999              ptr++;
1000          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1001          break;          break;
1002          }          }
1003        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1004          {          {
1005          c = -(ESC_REF + c);          escape = -s;
1006          break;          break;
1007          }          }
1008        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 891  else Line 1022  else
1022      /* \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
1023      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
1024      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
1025      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,
1026      than 3 octal digits. */      but no more than 3 octal digits. */
1027    
1028      case CHAR_0:      case CHAR_0:
1029      c -= CHAR_0;      c -= CHAR_0;
1030      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1031          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1032      if (!utf8 && c > 255) *errorcodeptr = ERR51;  #ifdef COMPILE_PCRE8
1033        if (!utf && c > 0xff) *errorcodeptr = ERR51;
1034    #endif
1035      break;      break;
1036    
1037      /* \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
1038      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.
1039      treated as a data character. */      If not, { is treated as a data character. */
1040    
1041      case CHAR_x:      case CHAR_x:
1042      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1043        {        {
1044        /* In JavaScript, \x must be followed by two hexadecimal numbers.        /* In JavaScript, \x must be followed by two hexadecimal numbers.
1045        Otherwise it is a lowercase x letter. */        Otherwise it is a lowercase x letter. */
1046        if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0)        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1047            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1048          {          {
1049          c = 0;          c = 0;
1050          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1051            {            {
1052            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1053  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1054            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1055            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 930  else Line 1064  else
1064    
1065      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1066        {        {
1067        const uschar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
       int count = 0;  
1068    
1069        c = 0;        c = 0;
1070        while ((digitab[*pt] & ctype_xdigit) != 0)        overflow = FALSE;
1071          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1072          {          {
1073          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1074          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1075          count++;  
1076    #ifdef COMPILE_PCRE32
1077            if (c >= 0x10000000l) { overflow = TRUE; break; }
1078    #endif
1079    
1080  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1081          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
# Line 947  else Line 1084  else
1084          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 */
1085          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1086  #endif  #endif
1087    
1088    #if defined COMPILE_PCRE8
1089            if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }
1090    #elif defined COMPILE_PCRE16
1091            if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }
1092    #elif defined COMPILE_PCRE32
1093            if (utf && c > 0x10ffffU) { overflow = TRUE; break; }
1094    #endif
1095            }
1096    
1097          if (overflow)
1098            {
1099            while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1100            *errorcodeptr = ERR34;
1101          }          }
1102    
1103        if (*pt == CHAR_RIGHT_CURLY_BRACKET)        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
1104          {          {
1105          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1106          ptr = pt;          ptr = pt;
1107          break;          break;
1108          }          }
# Line 963  else Line 1114  else
1114      /* Read just a single-byte hex-defined char */      /* Read just a single-byte hex-defined char */
1115    
1116      c = 0;      c = 0;
1117      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1118        {        {
1119        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1120        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1121  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1122        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 984  else Line 1135  else
1135    
1136      case CHAR_c:      case CHAR_c:
1137      c = *(++ptr);      c = *(++ptr);
1138      if (c == 0)      if (c == CHAR_NULL)
1139        {        {
1140        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
1141        break;        break;
# Line 1024  else Line 1175  else
1175  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1176  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1177    
1178  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1179       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1180    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1181    
1182  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1183    
1184  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1185    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1186    
1187  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1188    
1189  *ptrptr = ptr;  *ptrptr = ptr;
1190  return c;  *chptr = c;
1191    return escape;
1192  }  }
1193    
   
   
1194  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1195  /*************************************************  /*************************************************
1196  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1054  escape sequence. Line 1204  escape sequence.
1204  Argument:  Argument:
1205    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1206    negptr         points to a boolean that is set TRUE for negation else FALSE    negptr         points to a boolean that is set TRUE for negation else FALSE
1207    dptr           points to an int that is set to the detailed property value    ptypeptr       points to an unsigned int that is set to the type value
1208      pdataptr       points to an unsigned int that is set to the detailed property value
1209    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1210    
1211  Returns:         type value from ucp_type_table, or -1 for an invalid type  Returns:         TRUE if the type value was found, or FALSE for an invalid type
1212  */  */
1213    
1214  static int  static BOOL
1215  get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1216      unsigned int *pdataptr, int *errorcodeptr)
1217  {  {
1218  int c, i, bot, top;  pcre_uchar c;
1219  const uschar *ptr = *ptrptr;  int i, bot, top;
1220  char name[32];  const pcre_uchar *ptr = *ptrptr;
1221    pcre_uchar name[32];
1222    
1223  c = *(++ptr);  c = *(++ptr);
1224  if (c == 0) goto ERROR_RETURN;  if (c == CHAR_NULL) goto ERROR_RETURN;
1225    
1226  *negptr = FALSE;  *negptr = FALSE;
1227    
# Line 1082  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1235  if (c == CHAR_LEFT_CURLY_BRACKET)
1235      *negptr = TRUE;      *negptr = TRUE;
1236      ptr++;      ptr++;
1237      }      }
1238    for (i = 0; i < (int)sizeof(name) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1239      {      {
1240      c = *(++ptr);      c = *(++ptr);
1241      if (c == 0) goto ERROR_RETURN;      if (c == CHAR_NULL) goto ERROR_RETURN;
1242      if (c == CHAR_RIGHT_CURLY_BRACKET) break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
1243      name[i] = c;      name[i] = c;
1244      }      }
# Line 1106  else Line 1259  else
1259  /* Search for a recognized property name using binary chop */  /* Search for a recognized property name using binary chop */
1260    
1261  bot = 0;  bot = 0;
1262  top = _pcre_utt_size;  top = PRIV(utt_size);
1263    
1264  while (bot < top)  while (bot < top)
1265    {    {
1266      int r;
1267    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1268    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1269    if (c == 0)    if (r == 0)
1270      {      {
1271      *dptr = _pcre_utt[i].value;      *ptypeptr = PRIV(utt)[i].type;
1272      return _pcre_utt[i].type;      *pdataptr = PRIV(utt)[i].value;
1273        return TRUE;
1274      }      }
1275    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1276    }    }
1277    
1278  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1279  *ptrptr = ptr;  *ptrptr = ptr;
1280  return -1;  return FALSE;
1281    
1282  ERROR_RETURN:  ERROR_RETURN:
1283  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1284  *ptrptr = ptr;  *ptrptr = ptr;
1285  return -1;  return FALSE;
1286  }  }
1287  #endif  #endif
1288    
# Line 1153  Returns:         pointer to '}' on succe Line 1308  Returns:         pointer to '}' on succe
1308                   current ptr on error, with errorcodeptr set non-zero                   current ptr on error, with errorcodeptr set non-zero
1309  */  */
1310    
1311  static const uschar *  static const pcre_uchar *
1312  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)
1313  {  {
1314  int min = 0;  int min = 0;
1315  int max = -1;  int max = -1;
# Line 1162  int max = -1; Line 1317  int max = -1;
1317  /* 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
1318  an integer overflow. */  an integer overflow. */
1319    
1320  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1321  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1322    {    {
1323    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1177  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1332  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1332    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1333      {      {
1334      max = 0;      max = 0;
1335      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1336      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1337        {        {
1338        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1232  Arguments: Line 1387  Arguments:
1387    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1388    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1389    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1390    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1391    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1392    
1393  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1394  */  */
1395    
1396  static int  static int
1397  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,
1398    BOOL xmode, BOOL utf8, int *count)    BOOL xmode, BOOL utf, int *count)
1399  {  {
1400  uschar *ptr = *ptrptr;  pcre_uchar *ptr = *ptrptr;
1401  int start_count = *count;  int start_count = *count;
1402  int hwm_count = start_count;  int hwm_count = start_count;
1403  BOOL dup_parens = FALSE;  BOOL dup_parens = FALSE;
# Line 1254  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1409  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1409    {    {
1410    /* Handle specials such as (*SKIP) or (*UTF8) etc. */    /* Handle specials such as (*SKIP) or (*UTF8) etc. */
1411    
1412    if (ptr[1] == CHAR_ASTERISK) ptr += 2;    if (ptr[1] == CHAR_ASTERISK)
1413        {
1414        ptr += 2;
1415        while (ptr < cd->end_pattern && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1416        }
1417    
1418    /* Handle a normal, unnamed capturing parenthesis. */    /* Handle a normal, unnamed capturing parenthesis. */
1419    
# Line 1278  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1437  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1437    
1438    else if (ptr[2] == CHAR_NUMBER_SIGN)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1439      {      {
1440      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;      for (ptr += 3; *ptr != CHAR_NULL; ptr++)
1441          if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1442      goto FAIL_EXIT;      goto FAIL_EXIT;
1443      }      }
1444    
# Line 1291  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1451  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1451      ptr += 2;      ptr += 2;
1452      if (ptr[1] != CHAR_QUESTION_MARK)      if (ptr[1] != CHAR_QUESTION_MARK)
1453        {        {
1454        while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;        while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1455        if (*ptr != 0) ptr++;        if (*ptr != CHAR_NULL) ptr++;
1456        }        }
1457      }      }
1458    
# Line 1308  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1468  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1468      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1469          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1470        {        {
1471        int term;        pcre_uchar term;
1472        const uschar *thisname;        const pcre_uchar *thisname;
1473        *count += 1;        *count += 1;
1474        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1475        term = *ptr++;        term = *ptr++;
1476        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1477        thisname = ptr;        thisname = ptr;
1478        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1479        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1480            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1481          return *count;          return *count;
1482        term++;        term++;
1483        }        }
# Line 1335  for (; ptr < cd->end_pattern; ptr++) Line 1495  for (; ptr < cd->end_pattern; ptr++)
1495    
1496    if (*ptr == CHAR_BACKSLASH)    if (*ptr == CHAR_BACKSLASH)
1497      {      {
1498      if (*(++ptr) == 0) goto FAIL_EXIT;      if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1499      if (*ptr == CHAR_Q) for (;;)      if (*ptr == CHAR_Q) for (;;)
1500        {        {
1501        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};        while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1502        if (*ptr == 0) goto FAIL_EXIT;        if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1503        if (*(++ptr) == CHAR_E) break;        if (*(++ptr) == CHAR_E) break;
1504        }        }
1505      continue;      continue;
# Line 1360  for (; ptr < cd->end_pattern; ptr++) Line 1520  for (; ptr < cd->end_pattern; ptr++)
1520          {          {
1521          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1522            ptr+= 2;            ptr+= 2;
1523          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1524                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1525            ptr += 4;            ptr += 4;
1526          else          else
# Line 1383  for (; ptr < cd->end_pattern; ptr++) Line 1543  for (; ptr < cd->end_pattern; ptr++)
1543    
1544      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1545        {        {
1546        if (*ptr == 0) return -1;        if (*ptr == CHAR_NULL) return -1;
1547        if (*ptr == CHAR_BACKSLASH)        if (*ptr == CHAR_BACKSLASH)
1548          {          {
1549          if (*(++ptr) == 0) goto FAIL_EXIT;          if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1550          if (*ptr == CHAR_Q) for (;;)          if (*ptr == CHAR_Q) for (;;)
1551            {            {
1552            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};            while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1553            if (*ptr == 0) goto FAIL_EXIT;            if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1554            if (*(++ptr) == CHAR_E) break;            if (*(++ptr) == CHAR_E) break;
1555            }            }
1556          continue;          continue;
# Line 1404  for (; ptr < cd->end_pattern; ptr++) Line 1564  for (; ptr < cd->end_pattern; ptr++)
1564    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1565      {      {
1566      ptr++;      ptr++;
1567      while (*ptr != 0)      while (*ptr != CHAR_NULL)
1568        {        {
1569        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1570        ptr++;        ptr++;
1571  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1572        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1573  #endif  #endif
1574        }        }
1575      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1576      continue;      continue;
1577      }      }
1578    
# Line 1420  for (; ptr < cd->end_pattern; ptr++) Line 1580  for (; ptr < cd->end_pattern; ptr++)
1580    
1581    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1582      {      {
1583      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1584      if (rc > 0) return rc;      if (rc > 0) return rc;
1585      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1586      }      }
1587    
1588    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
# Line 1466  Arguments: Line 1626  Arguments:
1626    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1627    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1628    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1629    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1630    
1631  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1632  */  */
1633    
1634  static int  static int
1635  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,
1636    BOOL utf8)    BOOL utf)
1637  {  {
1638  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1639  int count = 0;  int count = 0;
1640  int rc;  int rc;
1641    
# Line 1486  matching closing parens. That is why we Line 1646  matching closing parens. That is why we
1646    
1647  for (;;)  for (;;)
1648    {    {
1649    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1650    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == CHAR_NULL) break;
1651    }    }
1652    
1653  return rc;  return rc;
# Line 1513  Arguments: Line 1673  Arguments:
1673  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1674  */  */
1675    
1676  static const uschar*  static const pcre_uchar*
1677  first_significant_code(const uschar *code, BOOL skipassert)  first_significant_code(const pcre_uchar *code, BOOL skipassert)
1678  {  {
1679  for (;;)  for (;;)
1680    {    {
# Line 1525  for (;;) Line 1685  for (;;)
1685      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1686      if (!skipassert) return code;      if (!skipassert) return code;
1687      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1688      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1689      break;      break;
1690    
1691      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1539  for (;;) Line 1699  for (;;)
1699      case OP_RREF:      case OP_RREF:
1700      case OP_NRREF:      case OP_NRREF:
1701      case OP_DEF:      case OP_DEF:
1702      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1703      break;      break;
1704    
1705      default:      default:
# Line 1569  and doing the check at the end; a flag s Line 1729  and doing the check at the end; a flag s
1729    
1730  Arguments:  Arguments:
1731    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1732    utf8     TRUE in UTF-8 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1733    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1734    cd       the "compile data" structure    cd       the "compile data" structure
1735    
# Line 1581  Returns:   the fixed length, Line 1741  Returns:   the fixed length,
1741  */  */
1742    
1743  static int  static int
1744  find_fixedlength(uschar *code, BOOL utf8, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1745  {  {
1746  int length = -1;  int length = -1;
1747    
1748  register int branchlength = 0;  register int branchlength = 0;
1749  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1750    
1751  /* 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
1752  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 1754  branch, check the length against that of
1754  for (;;)  for (;;)
1755    {    {
1756    int d;    int d;
1757    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1758    register int op = *cc;    register pcre_uchar op = *cc;
1759    
1760    switch (op)    switch (op)
1761      {      {
1762      /* 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 1769  for (;;)
1769      case OP_ONCE:      case OP_ONCE:
1770      case OP_ONCE_NC:      case OP_ONCE_NC:
1771      case OP_COND:      case OP_COND:
1772      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);
1773      if (d < 0) return d;      if (d < 0) return d;
1774      branchlength += d;      branchlength += d;
1775      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
# Line 1639  for (;;) Line 1800  for (;;)
1800    
1801      case OP_RECURSE:      case OP_RECURSE:
1802      if (!atend) return -3;      if (!atend) return -3;
1803      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1804      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1805      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1806      d = find_fixedlength(cs + 2, utf8, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1807      if (d < 0) return d;      if (d < 0) return d;
1808      branchlength += d;      branchlength += d;
1809      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1655  for (;;) Line 1816  for (;;)
1816      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1817      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1818      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1819      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1820        break;
1821    
1822      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1823    
# Line 1663  for (;;) Line 1825  for (;;)
1825      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
1826      case OP_SKIP_ARG:      case OP_SKIP_ARG:
1827      case OP_THEN_ARG:      case OP_THEN_ARG:
1828      cc += cc[1] + _pcre_OP_lengths[*cc];      cc += cc[1] + PRIV(OP_lengths)[*cc];
1829      break;      break;
1830    
1831      case OP_CALLOUT:      case OP_CALLOUT:
# Line 1690  for (;;) Line 1852  for (;;)
1852      case OP_SOM:      case OP_SOM:
1853      case OP_THEN:      case OP_THEN:
1854      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1855      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1856      break;      break;
1857    
1858      /* Handle literal characters */      /* Handle literal characters */
# Line 1701  for (;;) Line 1863  for (;;)
1863      case OP_NOTI:      case OP_NOTI:
1864      branchlength++;      branchlength++;
1865      cc += 2;      cc += 2;
1866  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1867      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1868  #endif  #endif
1869      break;      break;
1870    
# Line 1713  for (;;) Line 1875  for (;;)
1875      case OP_EXACTI:      case OP_EXACTI:
1876      case OP_NOTEXACT:      case OP_NOTEXACT:
1877      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1878      branchlength += GET2(cc,1);      branchlength += (int)GET2(cc,1);
1879      cc += 4;      cc += 2 + IMM2_SIZE;
1880  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1881      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1882  #endif  #endif
1883      break;      break;
1884    
1885      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1886      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1887      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1888      cc += 4;        cc += 2;
1889        cc += 1 + IMM2_SIZE + 1;
1890      break;      break;
1891    
1892      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1749  for (;;) Line 1912  for (;;)
1912      cc++;      cc++;
1913      break;      break;
1914    
1915      /* 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;
1916      otherwise \C is coded as OP_ALLANY. */      otherwise \C is coded as OP_ALLANY. */
1917    
1918      case OP_ANYBYTE:      case OP_ANYBYTE:
# Line 1757  for (;;) Line 1920  for (;;)
1920    
1921      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1922    
 #ifdef SUPPORT_UTF8  
     case OP_XCLASS:  
     cc += GET(cc, 1) - 33;  
     /* Fall through */  
 #endif  
   
1923      case OP_CLASS:      case OP_CLASS:
1924      case OP_NCLASS:      case OP_NCLASS:
1925      cc += 33;  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1926        case OP_XCLASS:
1927        /* The original code caused an unsigned overflow in 64 bit systems,
1928        so now we use a conditional statement. */
1929        if (op == OP_XCLASS)
1930          cc += GET(cc, 1);
1931        else
1932          cc += PRIV(OP_lengths)[OP_CLASS];
1933    #else
1934        cc += PRIV(OP_lengths)[OP_CLASS];
1935    #endif
1936    
1937      switch (*cc)      switch (*cc)
1938        {        {
# Line 1779  for (;;) Line 1946  for (;;)
1946    
1947        case OP_CRRANGE:        case OP_CRRANGE:
1948        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1949        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1950        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1951        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1952        break;        break;
1953    
1954        default:        default:
# Line 1896  length. Line 2063  length.
2063    
2064  Arguments:  Arguments:
2065    code        points to start of expression    code        points to start of expression
2066    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2067    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2068    
2069  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
2070  */  */
2071    
2072  const uschar *  const pcre_uchar *
2073  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2074  {  {
2075  for (;;)  for (;;)
2076    {    {
2077    register int c = *code;    register pcre_uchar c = *code;
2078    
2079    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2080    
# Line 1921  for (;;) Line 2088  for (;;)
2088    
2089    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2090      {      {
2091      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2092      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2093      }      }
2094    
2095    /* Handle capturing bracket */    /* Handle capturing bracket */
# Line 1930  for (;;) Line 2097  for (;;)
2097    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2098             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2099      {      {
2100      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2101      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2102      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2103      }      }
2104    
2105    /* 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 2127  for (;;)
2127        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2128        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2129        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2130        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2131            code += 2;
2132        break;        break;
2133    
2134        case OP_MARK:        case OP_MARK:
2135        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2136        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2137        case OP_THEN_ARG:        case OP_THEN_ARG:
2138        code += code[1];        code += code[1];
2139        break;        break;
# Line 1976  for (;;) Line 2141  for (;;)
2141    
2142      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2143    
2144      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2145    
2146    /* 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
2147    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
2148    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2149    
2150  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2151      if (utf8) switch(c)      if (utf) switch(c)
2152        {        {
2153        case OP_CHAR:        case OP_CHAR:
2154        case OP_CHARI:        case OP_CHARI:
# Line 2013  for (;;) Line 2178  for (;;)
2178        case OP_MINQUERYI:        case OP_MINQUERYI:
2179        case OP_POSQUERY:        case OP_POSQUERY:
2180        case OP_POSQUERYI:        case OP_POSQUERYI:
2181        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2182        break;        break;
2183        }        }
2184  #else  #else
2185      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2186  #endif  #endif
2187      }      }
2188    }    }
# Line 2034  instance of OP_RECURSE. Line 2199  instance of OP_RECURSE.
2199    
2200  Arguments:  Arguments:
2201    code        points to start of expression    code        points to start of expression
2202    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2203    
2204  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
2205  */  */
2206    
2207  static const uschar *  static const pcre_uchar *
2208  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2209  {  {
2210  for (;;)  for (;;)
2211    {    {
2212    register int c = *code;    register pcre_uchar c = *code;
2213    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2214    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2215    
# Line 2079  for (;;) Line 2244  for (;;)
2244        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2245        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2246        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2247        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2248            code += 2;
2249        break;        break;
2250    
2251        case OP_MARK:        case OP_MARK:
2252        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2253        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2254        case OP_THEN_ARG:        case OP_THEN_ARG:
2255        code += code[1];        code += code[1];
2256        break;        break;
# Line 2095  for (;;) Line 2258  for (;;)
2258    
2259      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2260    
2261      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2262    
2263      /* 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
2264      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
2265      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2266    
2267  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2268      if (utf8) switch(c)      if (utf) switch(c)
2269        {        {
2270        case OP_CHAR:        case OP_CHAR:
2271        case OP_CHARI:        case OP_CHARI:
2272          case OP_NOT:
2273          case OP_NOTI:
2274        case OP_EXACT:        case OP_EXACT:
2275        case OP_EXACTI:        case OP_EXACTI:
2276          case OP_NOTEXACT:
2277          case OP_NOTEXACTI:
2278        case OP_UPTO:        case OP_UPTO:
2279        case OP_UPTOI:        case OP_UPTOI:
2280          case OP_NOTUPTO:
2281          case OP_NOTUPTOI:
2282        case OP_MINUPTO:        case OP_MINUPTO:
2283        case OP_MINUPTOI:        case OP_MINUPTOI:
2284          case OP_NOTMINUPTO:
2285          case OP_NOTMINUPTOI:
2286        case OP_POSUPTO:        case OP_POSUPTO:
2287        case OP_POSUPTOI:        case OP_POSUPTOI:
2288          case OP_NOTPOSUPTO:
2289          case OP_NOTPOSUPTOI:
2290        case OP_STAR:        case OP_STAR:
2291        case OP_STARI:        case OP_STARI:
2292          case OP_NOTSTAR:
2293          case OP_NOTSTARI:
2294        case OP_MINSTAR:        case OP_MINSTAR:
2295        case OP_MINSTARI:        case OP_MINSTARI:
2296          case OP_NOTMINSTAR:
2297          case OP_NOTMINSTARI:
2298        case OP_POSSTAR:        case OP_POSSTAR:
2299        case OP_POSSTARI:        case OP_POSSTARI:
2300          case OP_NOTPOSSTAR:
2301          case OP_NOTPOSSTARI:
2302        case OP_PLUS:        case OP_PLUS:
2303        case OP_PLUSI:        case OP_PLUSI:
2304          case OP_NOTPLUS:
2305          case OP_NOTPLUSI:
2306        case OP_MINPLUS:        case OP_MINPLUS:
2307        case OP_MINPLUSI:        case OP_MINPLUSI:
2308          case OP_NOTMINPLUS:
2309          case OP_NOTMINPLUSI:
2310        case OP_POSPLUS:        case OP_POSPLUS:
2311        case OP_POSPLUSI:        case OP_POSPLUSI:
2312          case OP_NOTPOSPLUS:
2313          case OP_NOTPOSPLUSI:
2314        case OP_QUERY:        case OP_QUERY:
2315        case OP_QUERYI:        case OP_QUERYI:
2316          case OP_NOTQUERY:
2317          case OP_NOTQUERYI:
2318        case OP_MINQUERY:        case OP_MINQUERY:
2319        case OP_MINQUERYI:        case OP_MINQUERYI:
2320          case OP_NOTMINQUERY:
2321          case OP_NOTMINQUERYI:
2322        case OP_POSQUERY:        case OP_POSQUERY:
2323        case OP_POSQUERYI:        case OP_POSQUERYI:
2324        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_NOTPOSQUERY:
2325          case OP_NOTPOSQUERYI:
2326          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2327        break;        break;
2328        }        }
2329  #else  #else
2330      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2331  #endif  #endif
2332      }      }
2333    }    }
# Line 2159  bracket whose current branch will alread Line 2350  bracket whose current branch will alread
2350  Arguments:  Arguments:
2351    code        points to start of search    code        points to start of search
2352    endcode     points to where to stop    endcode     points to where to stop
2353    utf8        TRUE if in UTF8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2354    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2355    
2356  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2357  */  */
2358    
2359  static BOOL  static BOOL
2360  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2361    compile_data *cd)    BOOL utf, compile_data *cd)
2362  {  {
2363  register int c;  register pcre_uchar c;
2364  for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2365       code < endcode;       code < endcode;
2366       code = first_significant_code(code + _pcre_OP_lengths[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2367    {    {
2368    const uschar *ccode;    const pcre_uchar *ccode;
2369    
2370    c = *code;    c = *code;
2371    
# Line 2197  for (code = first_significant_code(code Line 2388  for (code = first_significant_code(code
2388    
2389    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2390      {      {
2391      const uschar *scode;      const pcre_uchar *scode;
2392      BOOL empty_branch;      BOOL empty_branch;
2393    
2394      /* Test for forward reference */      /* Test for forward reference */
2395    
2396      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2397        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2398    
2399      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2400    
# Line 2215  for (code = first_significant_code(code Line 2406  for (code = first_significant_code(code
2406    
2407      do      do
2408        {        {
2409        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd))
2410          {          {
2411          empty_branch = TRUE;          empty_branch = TRUE;
2412          break;          break;
# Line 2233  for (code = first_significant_code(code Line 2424  for (code = first_significant_code(code
2424    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2425        c == OP_BRAPOSZERO)        c == OP_BRAPOSZERO)
2426      {      {
2427      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2428      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
2429      c = *code;      c = *code;
2430      continue;      continue;
# Line 2271  for (code = first_significant_code(code Line 2462  for (code = first_significant_code(code
2462        empty_branch = FALSE;        empty_branch = FALSE;
2463        do        do
2464          {          {
2465          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
2466            empty_branch = TRUE;            empty_branch = TRUE;
2467          code += GET(code, 1);          code += GET(code, 1);
2468          }          }
# Line 2289  for (code = first_significant_code(code Line 2480  for (code = first_significant_code(code
2480      {      {
2481      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2482      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2483      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2484      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"
2485      here. */      here. */
2486    
2487  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2488      case OP_XCLASS:      case OP_XCLASS:
2489      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2490      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2301  for (code = first_significant_code(code Line 2492  for (code = first_significant_code(code
2492    
2493      case OP_CLASS:      case OP_CLASS:
2494      case OP_NCLASS:      case OP_NCLASS:
2495      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2496    
2497  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2498      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2499  #endif  #endif
2500    
# Line 2376  for (code = first_significant_code(code Line 2567  for (code = first_significant_code(code
2567      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2568      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2569      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2570      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2571          code += 2;
2572      break;      break;
2573    
2574      /* End of branch */      /* End of branch */
# Line 2391  for (code = first_significant_code(code Line 2583  for (code = first_significant_code(code
2583      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2584      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2585    
2586  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2587      case OP_STAR:      case OP_STAR:
2588      case OP_STARI:      case OP_STARI:
2589      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2404  for (code = first_significant_code(code Line 2596  for (code = first_significant_code(code
2596      case OP_MINQUERYI:      case OP_MINQUERYI:
2597      case OP_POSQUERY:      case OP_POSQUERY:
2598      case OP_POSQUERYI:      case OP_POSQUERYI:
2599      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2600      break;      break;
2601    
2602      case OP_UPTO:      case OP_UPTO:
# Line 2413  for (code = first_significant_code(code Line 2605  for (code = first_significant_code(code
2605      case OP_MINUPTOI:      case OP_MINUPTOI:
2606      case OP_POSUPTO:      case OP_POSUPTO:
2607      case OP_POSUPTOI:      case OP_POSUPTOI:
2608      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]);
2609      break;      break;
2610  #endif  #endif
2611    
# Line 2423  for (code = first_significant_code(code Line 2615  for (code = first_significant_code(code
2615      case OP_MARK:      case OP_MARK:
2616      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
2617      case OP_SKIP_ARG:      case OP_SKIP_ARG:
     code += code[1];  
     break;  
   
2618      case OP_THEN_ARG:      case OP_THEN_ARG:
2619      code += code[1];      code += code[1];
2620      break;      break;
# Line 2457  Arguments: Line 2646  Arguments:
2646    code        points to start of the recursion    code        points to start of the recursion
2647    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2648    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2649    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2650    cd          pointers to tables etc    cd          pointers to tables etc
2651    
2652  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2653  */  */
2654    
2655  static BOOL  static BOOL
2656  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2657    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2658  {  {
2659  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2660    {    {
2661    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
2662      return FALSE;      return FALSE;
2663    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2664    }    }
# Line 2521  Returns:   TRUE or FALSE Line 2710  Returns:   TRUE or FALSE
2710  */  */
2711    
2712  static BOOL  static BOOL
2713  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2714  {  {
2715  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2716  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2717  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != CHAR_NULL; ptr++)
2718    {    {
2719    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2720      ptr++;      ptr++;
# Line 2565  Returns:     a value representing the na Line 2754  Returns:     a value representing the na
2754  */  */
2755    
2756  static int  static int
2757  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2758  {  {
2759  const char *pn = posix_names;  const char *pn = posix_names;
2760  register int yield = 0;  register int yield = 0;
2761  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2762    {    {
2763    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2764      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2765    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2766    yield++;    yield++;
2767    }    }
# Line 2604  value in the reference (which is a group Line 2793  value in the reference (which is a group
2793  Arguments:  Arguments:
2794    group      points to the start of the group    group      points to the start of the group
2795    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2796    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2797    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2798    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
2799    
# Line 2612  Returns:     nothing Line 2801  Returns:     nothing
2801  */  */
2802    
2803  static void  static void
2804  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2805    uschar *save_hwm)    pcre_uchar *save_hwm)
2806  {  {
2807  uschar *ptr = group;  pcre_uchar *ptr = group;
2808    
2809  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2810    {    {
2811    int offset;    int offset;
2812    uschar *hc;    pcre_uchar *hc;
2813    
2814    /* 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
2815    reference. */    reference. */
2816    
2817    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2818      {      {
2819      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2820      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2821        {        {
2822        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2640  while ((ptr = (uschar *)find_recurse(ptr Line 2829  while ((ptr = (uschar *)find_recurse(ptr
2829    
2830    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2831      {      {
2832      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2833      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2834      }      }
2835    
# Line 2665  Arguments: Line 2854  Arguments:
2854  Returns:         new code pointer  Returns:         new code pointer
2855  */  */
2856    
2857  static uschar *  static pcre_uchar *
2858  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2859  {  {
2860  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2861  *code++ = 255;  *code++ = 255;
2862  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2863  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2864  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2865  }  }
2866    
2867    
# Line 2694  Returns:             nothing Line 2883  Returns:             nothing
2883  */  */
2884    
2885  static void  static void
2886  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2887  {  {
2888  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2889  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2708  PUT(previous_callout, 2 + LINK_SIZE, len Line 2897  PUT(previous_callout, 2 + LINK_SIZE, len
2897  *************************************************/  *************************************************/
2898    
2899  /* This function is passed the start and end of a class range, in UTF-8 mode  /* This function is passed the start and end of a class range, in UTF-8 mode
2900  with UCP support. It searches up the characters, looking for internal ranges of  with UCP support. It searches up the characters, looking for ranges of
2901  characters in the "other" case. Each call returns the next one, updating the  characters in the "other" case. Each call returns the next one, updating the
2902  start address.  start address. A character with multiple other cases is returned on its own
2903    with a special return value.
2904    
2905  Arguments:  Arguments:
2906    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2718  Arguments: Line 2908  Arguments:
2908    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2909    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2910    
2911  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2912                   0 when a range is returned
2913                  >0 the CASESET offset for char with multiple other cases
2914                    in this case, ocptr contains the original
2915  */  */
2916    
2917  static BOOL  static int
2918  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2919    unsigned int *odptr)    pcre_uint32 *odptr)
2920  {  {
2921  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2922    unsigned int co;
2923    
2924    /* Find the first character that has an other case. If it has multiple other
2925    cases, return its case offset value. */
2926    
2927  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2928    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2929      if ((co = UCD_CASESET(c)) != 0)
2930        {
2931        *ocptr = c++;   /* Character that has the set */
2932        *cptr = c;      /* Rest of input range */
2933        return (int)co;
2934        }
2935      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2936      }
2937    
2938  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2939    
2940  *ocptr = othercase;  *ocptr = othercase;
2941  next = othercase + 1;  next = othercase + 1;
# Line 2741  for (++c; c <= d; c++) Line 2946  for (++c; c <= d; c++)
2946    next++;    next++;
2947    }    }
2948    
2949  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2950  *cptr = c;  *cptr = c;             /* Rest of input range */
2951    return 0;
 return TRUE;  
2952  }  }
2953    
2954    
# Line 2766  Returns:       TRUE if auto-possessifyin Line 2970  Returns:       TRUE if auto-possessifyin
2970  */  */
2971    
2972  static BOOL  static BOOL
2973  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(pcre_uint32 c, unsigned int ptype, unsigned int pdata, BOOL negated)
2974  {  {
2975    #ifdef SUPPORT_UCP
2976    const pcre_uint32 *p;
2977    #endif
2978    
2979  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2980    
2981  switch(ptype)  switch(ptype)
2982    {    {
2983    case PT_LAMP:    case PT_LAMP:
# Line 2777  switch(ptype) Line 2986  switch(ptype)
2986            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
2987    
2988    case PT_GC:    case PT_GC:
2989    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2990    
2991    case PT_PC:    case PT_PC:
2992    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2788  switch(ptype) Line 2997  switch(ptype)
2997    /* These are specials */    /* These are specials */
2998    
2999    case PT_ALNUM:    case PT_ALNUM:
3000    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3001            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
3002    
3003    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
3004    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3005            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
3006            == negated;            == negated;
3007    
3008    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
3009    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3010            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
3011            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
3012            == negated;            == negated;
3013    
3014    case PT_WORD:    case PT_WORD:
3015    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3016            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3017            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3018    
3019    #ifdef SUPPORT_UCP
3020      case PT_CLIST:
3021      p = PRIV(ucd_caseless_sets) + prop->caseset;
3022      for (;;)
3023        {
3024        if (c < *p) return !negated;
3025        if (c == *p++) return negated;
3026        }
3027      break;  /* Control never reaches here */
3028    #endif
3029    }    }
3030    
3031  return FALSE;  return FALSE;
3032  }  }
3033  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2823  sense to automatically possessify the re Line 3044  sense to automatically possessify the re
3044    
3045  Arguments:  Arguments:
3046    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3047    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3048    ptr           next character in pattern    ptr           next character in pattern
3049    options       options bits    options       options bits
3050    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2832  Returns:        TRUE if possessifying is Line 3053  Returns:        TRUE if possessifying is
3053  */  */
3054    
3055  static BOOL  static BOOL
3056  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3057    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3058  {  {
3059  int c, next;  pcre_uint32 c = NOTACHAR;
3060  int op_code = *previous++;  pcre_uint32 next;
3061    int escape;
3062    pcre_uchar op_code = *previous++;
3063    
3064  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3065    
# Line 2844  if ((options & PCRE_EXTENDED) != 0) Line 3067  if ((options & PCRE_EXTENDED) != 0)
3067    {    {
3068    for (;;)    for (;;)
3069      {      {
3070      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3071      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3072        {        {
3073        ptr++;        ptr++;
3074        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3075          {          {
3076          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3077          ptr++;          ptr++;
3078  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3079          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3080  #endif  #endif
3081          }          }
3082        }        }
# Line 2867  value is a character, a negative value i Line 3090  value is a character, a negative value i
3090  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3091    {    {
3092    int temperrorcode = 0;    int temperrorcode = 0;
3093    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options,
3094        FALSE);
3095    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3096    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3097    }    }
3098    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
3099    {    {
3100  #ifdef SUPPORT_UTF8    escape = 0;
3101    if (utf8) { GETCHARINC(next, ptr); } else  #ifdef SUPPORT_UTF
3102      if (utf) { GETCHARINC(next, ptr); } else
3103  #endif  #endif
3104    next = *ptr++;    next = *ptr++;
3105    }    }
   
3106  else return FALSE;  else return FALSE;
3107    
3108  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2888  if ((options & PCRE_EXTENDED) != 0) Line 3111  if ((options & PCRE_EXTENDED) != 0)
3111    {    {
3112    for (;;)    for (;;)
3113      {      {
3114      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3115      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3116        {        {
3117        ptr++;        ptr++;
3118        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3119          {          {
3120          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3121          ptr++;          ptr++;
3122  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3123          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3124  #endif  #endif
3125          }          }
3126        }        }
# Line 2908  if ((options & PCRE_EXTENDED) != 0) Line 3131  if ((options & PCRE_EXTENDED) != 0)
3131  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3132    
3133  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3134    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)
3135      return FALSE;      return FALSE;
3136    
3137  /* Now compare the next item with the previous opcode. First, handle cases when  /* If the previous item is a character, get its value. */
 the next item is a character. */  
3138    
3139  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3140        op_code == OP_NOT || op_code == OP_NOTI)
3141    {    {
3142    case OP_CHAR:  #ifdef SUPPORT_UTF
 #ifdef SUPPORT_UTF8  
3143    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3144  #else  #else
3145    c = *previous;    c = *previous;
3146  #endif  #endif
3147    return c != next;    }
3148    
3149    /* For CHARI (caseless character) we must check the other case. If we have  /* Now compare the next item with the previous opcode. First, handle cases when
3150    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3151    
3152    case OP_CHARI:  if (escape == 0)
3153  #ifdef SUPPORT_UTF8    {
3154    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3155  #else    case, which maps to the special PT_CLIST property. Check this first. */
3156    c = *previous;  
3157    #ifdef SUPPORT_UCP
3158      if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3159        {
3160        unsigned int ocs = UCD_CASESET(next);
3161        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3162        }
3163  #endif  #endif
3164    if (c == next) return FALSE;  
3165  #ifdef SUPPORT_UTF8    switch(op_code)
   if (utf8)  
3166      {      {
3167      unsigned int othercase;      case OP_CHAR:
3168      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3169    
3170        /* For CHARI (caseless character) we must check the other case. If we have
3171        Unicode property support, we can use it to test the other case of
3172        high-valued characters. We know that next can have only one other case,
3173        because multi-other-case characters are dealt with above. */
3174    
3175        case OP_CHARI:
3176        if (c == next) return FALSE;
3177    #ifdef SUPPORT_UTF
3178        if (utf)
3179          {
3180          pcre_uint32 othercase;
3181          if (next < 128) othercase = cd->fcc[next]; else
3182  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3183      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3184  #else  #else
3185      othercase = NOTACHAR;        othercase = NOTACHAR;
3186  #endif  #endif
3187      return (unsigned int)c != othercase;        return c != othercase;
3188      }        }
3189    else      else
3190  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3191    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3192    
3193    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These      case OP_NOT:
3194    opcodes are not used for multi-byte characters, because they are coded using      return c == next;
3195    an XCLASS instead. */  
3196        case OP_NOTI:
3197    case OP_NOT:      if (c == next) return TRUE;
3198    return (c = *previous) == next;  #ifdef SUPPORT_UTF
3199        if (utf)
3200    case OP_NOTI:        {
3201    if ((c = *previous) == next) return TRUE;        pcre_uint32 othercase;
3202  #ifdef SUPPORT_UTF8        if (next < 128) othercase = cd->fcc[next]; else
   if (utf8)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3203  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3204      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE(next);
3205  #else  #else
3206      othercase = NOTACHAR;        othercase = NOTACHAR;
3207  #endif  #endif
3208      return (unsigned int)c == othercase;        return c == othercase;
3209      }        }
3210    else      else
3211  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3212    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3213    
3214    /* 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.
3215    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. */
3216    
3217    case OP_DIGIT:      case OP_DIGIT:
3218    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3219    
3220    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3221    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3222    
3223    case OP_WHITESPACE:      case OP_WHITESPACE:
3224    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3225    
3226    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3227    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3228    
3229    case OP_WORDCHAR:      case OP_WORDCHAR:
3230    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3231    
3232    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3233    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3234    
3235    case OP_HSPACE:      case OP_HSPACE:
3236    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3237    switch(next)      switch(next)
3238      {        {
3239      case 0x09:        HSPACE_CASES:
3240      case 0x20:        return op_code == OP_NOT_HSPACE;
     case 0xa0:  
     case 0x1680:  
     case 0x180e:  
     case 0x2000:  
     case 0x2001:  
     case 0x2002:  
     case 0x2003:  
     case 0x2004:  
     case 0x2005:  
     case 0x2006:  
     case 0x2007:  
     case 0x2008:  
     case 0x2009:  
     case 0x200A:  
     case 0x202f:  
     case 0x205f:  
     case 0x3000:  
     return op_code == OP_NOT_HSPACE;  
     default:  
     return op_code != OP_NOT_HSPACE;  
     }  
3241    
3242    case OP_ANYNL:        default:
3243    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3244    case OP_NOT_VSPACE:        }
3245    switch(next)  
3246      {      case OP_ANYNL:
3247      case 0x0a:      case OP_VSPACE:
3248      case 0x0b:      case OP_NOT_VSPACE:
3249      case 0x0c:      switch(next)
3250      case 0x0d:        {
3251      case 0x85:        VSPACE_CASES:
3252      case 0x2028:        return op_code == OP_NOT_VSPACE;
3253      case 0x2029:  
3254      return op_code == OP_NOT_VSPACE;        default:
3255      default:        return op_code != OP_NOT_VSPACE;
3256      return op_code != OP_NOT_VSPACE;        }
     }  
3257    
3258  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3259    case OP_PROP:      case OP_PROP:
3260    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3261    
3262    case OP_NOTPROP:      case OP_NOTPROP:
3263    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3264  #endif  #endif
3265    
3266    default:      default:
3267    return FALSE;      return FALSE;
3268        }
3269    }    }
3270    
   
3271  /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP  /* Handle the case when the next item is \d, \s, etc. Note that when PCRE_UCP
3272  is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are  is set, \d turns into ESC_du rather than ESC_d, etc., so ESC_d etc. are
3273  generated only when PCRE_UCP is *not* set, that is, when only ASCII  generated only when PCRE_UCP is *not* set, that is, when only ASCII
# Line 3065  switch(op_code) Line 3278  switch(op_code)
3278    {    {
3279    case OP_CHAR:    case OP_CHAR:
3280    case OP_CHARI:    case OP_CHARI:
3281  #ifdef SUPPORT_UTF8    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3282      {      {
3283      case ESC_d:      case ESC_d:
3284      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3285    
3286      case ESC_D:      case ESC_D:
3287      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3288    
3289      case ESC_s:      case ESC_s:
3290      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3291    
3292      case ESC_S:      case ESC_S:
3293      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3294    
3295      case ESC_w:      case ESC_w:
3296      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3297    
3298      case ESC_W:      case ESC_W:
3299      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3300    
3301      case ESC_h:      case ESC_h:
3302      case ESC_H:      case ESC_H:
3303      switch(c)      switch(c)
3304        {        {
3305        case 0x09:        HSPACE_CASES:
3306        case 0x20:        return escape != ESC_h;
3307        case 0xa0:  
       case 0x1680:  
       case 0x180e:  
       case 0x2000:  
       case 0x2001:  
       case 0x2002:  
       case 0x2003:  
       case 0x2004:  
       case 0x2005:  
       case 0x2006:  
       case 0x2007:  
       case 0x2008:  
       case 0x2009:  
       case 0x200A:  
       case 0x202f:  
       case 0x205f:  
       case 0x3000:  
       return -next != ESC_h;  
3308        default:        default:
3309        return -next == ESC_h;        return escape == ESC_h;
3310        }        }
3311    
3312      case ESC_v:      case ESC_v:
3313      case ESC_V:      case ESC_V:
3314      switch(c)      switch(c)
3315        {        {
3316        case 0x0a:        VSPACE_CASES:
3317        case 0x0b:        return escape != ESC_v;
3318        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3319        default:        default:
3320        return -next == ESC_v;        return escape == ESC_v;
3321        }        }
3322    
3323      /* When PCRE_UCP is set, these values get generated for \d etc. Find      /* When PCRE_UCP is set, these values get generated for \d etc. Find
3324      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3325      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3326    
3327  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3328      case ESC_du:      case ESC_du:
# Line 3147  switch(op_code) Line 3333  switch(op_code)
3333      case ESC_SU:      case ESC_SU:
3334        {        {
3335        int temperrorcode = 0;        int temperrorcode = 0;
3336        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3337        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3338        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3339        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3340        }        }
# Line 3157  switch(op_code) Line 3343  switch(op_code)
3343      case ESC_p:      case ESC_p:
3344      case ESC_P:      case ESC_P:
3345        {        {
3346        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3347          int errorcodeptr;
3348        BOOL negated;        BOOL negated;
3349    
3350        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3351        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3352        if (ptype < 0) return FALSE;          return FALSE;
3353        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3354    
3355        /* If the property item is optional, we have to give up. (When generated        /* If the property item is optional, we have to give up. (When generated
# Line 3170  switch(op_code) Line 3357  switch(op_code)
3357        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. */
3358    
3359        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3360          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)
3361            return FALSE;            return FALSE;
3362    
3363        /* Do the property check. */        /* Do the property check. */
3364    
3365        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3366        }        }
3367  #endif  #endif
3368    
# Line 3190  switch(op_code) Line 3377  switch(op_code)
3377    these op-codes are never generated.) */    these op-codes are never generated.) */
3378    
3379    case OP_DIGIT:    case OP_DIGIT:
3380    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3381           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3382    
3383    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3384    return next == -ESC_d;    return escape == ESC_d;
3385    
3386    case OP_WHITESPACE:    case OP_WHITESPACE:
3387    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3388    
3389    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3390    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3391    
3392    case OP_HSPACE:    case OP_HSPACE:
3393    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3394           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3395    
3396    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3397    return next == -ESC_h;    return escape == ESC_h;
3398    
3399    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3400    case OP_ANYNL:    case OP_ANYNL:
3401    case OP_VSPACE:    case OP_VSPACE:
3402    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3403    
3404    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3405    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3406    
3407    case OP_WORDCHAR:    case OP_WORDCHAR:
3408    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3409           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3410    
3411    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3412    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3413    
3414    default:    default:
3415    return FALSE;    return FALSE;
# Line 3234  switch(op_code) Line 3421  switch(op_code)
3421    
3422    
3423  /*************************************************  /*************************************************
3424    *        Add a character or range to a class     *
3425    *************************************************/
3426    
3427    /* This function packages up the logic of adding a character or range of
3428    characters to a class. The character values in the arguments will be within the
3429    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3430    mutually recursive with the function immediately below.
3431    
3432    Arguments:
3433      classbits     the bit map for characters < 256
3434      uchardptr     points to the pointer for extra data
3435      options       the options word
3436      cd            contains pointers to tables etc.
3437      start         start of range character
3438      end           end of range character
3439    
3440    Returns:        the number of < 256 characters added
3441                    the pointer to extra data is updated
3442    */
3443    
3444    static int
3445    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3446      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3447    {
3448    pcre_uint32 c;
3449    int n8 = 0;
3450    
3451    /* If caseless matching is required, scan the range and process alternate
3452    cases. In Unicode, there are 8-bit characters that have alternate cases that
3453    are greater than 255 and vice-versa. Sometimes we can just extend the original
3454    range. */
3455    
3456    if ((options & PCRE_CASELESS) != 0)
3457      {
3458    #ifdef SUPPORT_UCP
3459      if ((options & PCRE_UTF8) != 0)
3460        {
3461        int rc;
3462        pcre_uint32 oc, od;
3463    
3464        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3465        c = start;
3466    
3467        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3468          {
3469          /* Handle a single character that has more than one other case. */
3470    
3471          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3472            PRIV(ucd_caseless_sets) + rc, oc);
3473    
3474          /* Do nothing if the other case range is within the original range. */
3475    
3476          else if (oc >= start && od <= end) continue;
3477    
3478          /* Extend the original range if there is overlap, noting that if oc < c, we
3479          can't have od > end because a subrange is always shorter than the basic
3480          range. Otherwise, use a recursive call to add the additional range. */
3481    
3482          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3483          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3484          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3485          }
3486        }
3487      else
3488    #endif  /* SUPPORT_UCP */
3489    
3490      /* Not UTF-mode, or no UCP */
3491    
3492      for (c = start; c <= end && c < 256; c++)
3493        {
3494        SETBIT(classbits, cd->fcc[c]);
3495        n8++;
3496        }
3497      }
3498    
3499    /* Now handle the original range. Adjust the final value according to the bit
3500    length - this means that the same lists of (e.g.) horizontal spaces can be used
3501    in all cases. */
3502    
3503    #if defined COMPILE_PCRE8
3504    #ifdef SUPPORT_UTF
3505      if ((options & PCRE_UTF8) == 0)
3506    #endif
3507      if (end > 0xff) end = 0xff;
3508    
3509    #elif defined COMPILE_PCRE16
3510    #ifdef SUPPORT_UTF
3511      if ((options & PCRE_UTF16) == 0)
3512    #endif
3513      if (end > 0xffff) end = 0xffff;
3514    
3515    #endif /* COMPILE_PCRE[8|16] */
3516    
3517    /* If all characters are less than 256, use the bit map. Otherwise use extra
3518    data. */
3519    
3520    if (end < 0x100)
3521      {
3522      for (c = start; c <= end; c++)
3523        {
3524        n8++;
3525        SETBIT(classbits, c);
3526        }
3527      }
3528    
3529    else
3530      {
3531      pcre_uchar *uchardata = *uchardptr;
3532    
3533    #ifdef SUPPORT_UTF
3534      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3535        {
3536        if (start < end)
3537          {
3538          *uchardata++ = XCL_RANGE;
3539          uchardata += PRIV(ord2utf)(start, uchardata);
3540          uchardata += PRIV(ord2utf)(end, uchardata);
3541          }
3542        else if (start == end)
3543          {
3544          *uchardata++ = XCL_SINGLE;
3545          uchardata += PRIV(ord2utf)(start, uchardata);
3546          }
3547        }
3548      else
3549    #endif  /* SUPPORT_UTF */
3550    
3551      /* Without UTF support, character values are constrained by the bit length,
3552      and can only be > 256 for 16-bit and 32-bit libraries. */
3553    
3554    #ifdef COMPILE_PCRE8
3555        {}
3556    #else
3557      if (start < end)
3558        {
3559        *uchardata++ = XCL_RANGE;
3560        *uchardata++ = start;
3561        *uchardata++ = end;
3562        }
3563      else if (start == end)
3564        {
3565        *uchardata++ = XCL_SINGLE;
3566        *uchardata++ = start;
3567        }
3568    #endif
3569    
3570      *uchardptr = uchardata;   /* Updata extra data pointer */
3571      }
3572    
3573    return n8;    /* Number of 8-bit characters */
3574    }
3575    
3576    
3577    
3578    
3579    /*************************************************
3580    *        Add a list of characters to a class     *
3581    *************************************************/
3582    
3583    /* This function is used for adding a list of case-equivalent characters to a
3584    class, and also for adding a list of horizontal or vertical whitespace. If the
3585    list is in order (which it should be), ranges of characters are detected and
3586    handled appropriately. This function is mutually recursive with the function
3587    above.
3588    
3589    Arguments:
3590      classbits     the bit map for characters < 256
3591      uchardptr     points to the pointer for extra data
3592      options       the options word
3593      cd            contains pointers to tables etc.
3594      p             points to row of 32-bit values, terminated by NOTACHAR
3595      except        character to omit; this is used when adding lists of
3596                      case-equivalent characters to avoid including the one we
3597                      already know about
3598    
3599    Returns:        the number of < 256 characters added
3600                    the pointer to extra data is updated
3601    */
3602    
3603    static int
3604    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3605      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3606    {
3607    int n8 = 0;
3608    while (p[0] < NOTACHAR)
3609      {
3610      int n = 0;
3611      if (p[0] != except)
3612        {
3613        while(p[n+1] == p[0] + n + 1) n++;
3614        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3615        }
3616      p += n + 1;
3617      }
3618    return n8;
3619    }
3620    
3621    
3622    
3623    /*************************************************
3624    *    Add characters not in a list to a class     *
3625    *************************************************/
3626    
3627    /* This function is used for adding the complement of a list of horizontal or
3628    vertical whitespace to a class. The list must be in order.
3629    
3630    Arguments:
3631      classbits     the bit map for characters < 256
3632      uchardptr     points to the pointer for extra data
3633      options       the options word
3634      cd            contains pointers to tables etc.
3635      p             points to row of 32-bit values, terminated by NOTACHAR
3636    
3637    Returns:        the number of < 256 characters added
3638                    the pointer to extra data is updated
3639    */
3640    
3641    static int
3642    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3643      int options, compile_data *cd, const pcre_uint32 *p)
3644    {
3645    BOOL utf = (options & PCRE_UTF8) != 0;
3646    int n8 = 0;
3647    if (p[0] > 0)
3648      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3649    while (p[0] < NOTACHAR)
3650      {
3651      while (p[1] == p[0] + 1) p++;
3652      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3653        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3654      p++;
3655      }
3656    return n8;
3657    }
3658    
3659    
3660    
3661    /*************************************************
3662  *           Compile one branch                   *  *           Compile one branch                   *
3663  *************************************************/  *************************************************/
3664    
# Line 3248  Arguments: Line 3673  Arguments:
3673    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3674    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3675    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3676    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3677    reqbyteptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3678      reqcharptr     place to put the last required character
3679      reqcharflagsptr place to put the last required character flags, or a negative number
3680    bcptr          points to current branch chain    bcptr          points to current branch chain
3681    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3682    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3261  Returns:         TRUE on success Line 3688  Returns:         TRUE on success
3688  */  */
3689    
3690  static BOOL  static BOOL
3691  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3692    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3693    int cond_depth, compile_data *cd, int *lengthptr)    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3694      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3695      branch_chain *bcptr, int cond_depth,
3696      compile_data *cd, int *lengthptr)
3697  {  {
3698  int repeat_type, op_type;  int repeat_type, op_type;
3699  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3700  int bravalue = 0;  int bravalue = 0;
3701  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3702  int firstbyte, reqbyte;  pcre_uint32 firstchar, reqchar;
3703  int zeroreqbyte, zerofirstbyte;  pcre_int32 firstcharflags, reqcharflags;
3704  int req_caseopt, reqvary, tempreqvary;  pcre_uint32 zeroreqchar, zerofirstchar;
3705    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3706    pcre_int32 req_caseopt, reqvary, tempreqvary;
3707  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3708  int after_manual_callout = 0;  int after_manual_callout = 0;
3709  int length_prevgroup = 0;  int length_prevgroup = 0;
3710  register int c;  register pcre_uint32 c;
3711  register uschar *code = *codeptr;  int escape;
3712  uschar *last_code = code;  register pcre_uchar *code = *codeptr;
3713  uschar *orig_code = code;  pcre_uchar *last_code = code;
3714  uschar *tempcode;  pcre_uchar *orig_code = code;
3715    pcre_uchar *tempcode;
3716  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3717  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3718  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3719  const uschar *tempptr;  const pcre_uchar *tempptr;
3720  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3721  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3722  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3723  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3724  uschar classbits[32];  pcre_uint8 classbits[32];
3725    
3726  /* 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
3727  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
3728  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3729    
3730  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3731  BOOL class_utf8;  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3732  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3733  uschar *class_utf8data;  #ifndef COMPILE_PCRE32
3734  uschar *class_utf8data_base;  pcre_uchar utf_chars[6];
3735  uschar utf8_char[6];  #endif
3736  #else  #else
3737  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3738    #endif
3739    
3740    /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3741    class_uchardata always so that it can be passed to add_to_class() always,
3742    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3743    alternative calls for the different cases. */
3744    
3745    pcre_uchar *class_uchardata;
3746    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3747    BOOL xclass;
3748    pcre_uchar *class_uchardata_base;
3749  #endif  #endif
3750    
3751  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3315  greedy_non_default = greedy_default ^ 1; Line 3759  greedy_non_default = greedy_default ^ 1;
3759    
3760  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3761  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
3762  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
3763  find one.  find one.
3764    
3765  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
3766  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
3767  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3768  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3769    
3770  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3771    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3772    
3773  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3774  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
3775  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
3776  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3777    value. This is used only for ASCII characters. */
3778    
3779  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3780    
3781  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3782    
# Line 3342  for (;; ptr++) Line 3788  for (;; ptr++)
3788    BOOL is_quantifier;    BOOL is_quantifier;
3789    BOOL is_recurse;    BOOL is_recurse;
3790    BOOL reset_bracount;    BOOL reset_bracount;
3791    int class_charcount;    int class_has_8bitchar;
3792    int class_lastchar;    int class_one_char;
3793    int newoptions;    int newoptions;
3794    int recno;    int recno;
3795    int refsign;    int refsign;
3796    int skipbytes;    int skipbytes;
3797    int subreqbyte;    pcre_uint32 subreqchar, subfirstchar;
3798    int subfirstbyte;    pcre_int32 subreqcharflags, subfirstcharflags;
3799    int terminator;    int terminator;
3800    int mclength;    unsigned int mclength;
3801    int tempbracount;    unsigned int tempbracount;
3802    uschar mcbuffer[8];    pcre_uint32 ec;
3803      pcre_uchar mcbuffer[8];
3804    
3805    /* Get next byte in the pattern */    /* Get next character in the pattern */
3806    
3807    c = *ptr;    c = *ptr;
3808    
3809    /* If we are at the end of a nested substitution, revert to the outer level    /* If we are at the end of a nested substitution, revert to the outer level
3810    string. Nesting only happens one level deep. */    string. Nesting only happens one level deep. */
3811    
3812    if (c == 0 && nestptr != NULL)    if (c == CHAR_NULL && nestptr != NULL)
3813      {      {
3814      ptr = nestptr;      ptr = nestptr;
3815      nestptr = NULL;      nestptr = NULL;
# Line 3377  for (;; ptr++) Line 3824  for (;; ptr++)
3824  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3825      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3826  #endif  #endif
3827      if (code > cd->start_workspace + cd->workspace_size -      if (code > cd->start_workspace + cd->workspace_size -
3828          WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */          WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3829        {        {
3830        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
# Line 3401  for (;; ptr++) Line 3848  for (;; ptr++)
3848        }        }
3849    
3850      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3851      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, (int)(code - last_code),      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3852        c));        (int)(code - last_code), c, c));
3853    
3854      /* 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
3855      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 3859  for (;; ptr++)
3859        {        {
3860        if (previous > orig_code)        if (previous > orig_code)
3861          {          {
3862          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3863          code -= previous - orig_code;          code -= previous - orig_code;
3864          previous = orig_code;          previous = orig_code;
3865          }          }
# Line 3428  for (;; ptr++) Line 3875  for (;; ptr++)
3875    /* 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
3876    reference list. */    reference list. */
3877    
3878    else if (cd->hwm > cd->start_workspace + cd->workspace_size -    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3879             WORK_SIZE_SAFETY_MARGIN)             WORK_SIZE_SAFETY_MARGIN)
3880      {      {
3881      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
# Line 3437  for (;; ptr++) Line 3884  for (;; ptr++)
3884    
3885    /* If in \Q...\E, check for the end; if not, we have a literal */    /* If in \Q...\E, check for the end; if not, we have a literal */
3886    
3887    if (inescq && c != 0)    if (inescq && c != CHAR_NULL)
3888      {      {
3889      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3890        {        {
# Line 3481  for (;; ptr++) Line 3928  for (;; ptr++)
3928    
3929    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
3930      {      {
3931      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
3932      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3933        {        {
3934        ptr++;        ptr++;
3935        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3936          {          {
3937          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3938          ptr++;          ptr++;
3939  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3940          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3941  #endif  #endif
3942          }          }
3943        if (*ptr != 0) continue;        if (*ptr != CHAR_NULL) continue;
3944    
3945        /* Else fall through to handle end of string */        /* Else fall through to handle end of string */
3946        c = 0;        c = 0;
# Line 3514  for (;; ptr++) Line 3961  for (;; ptr++)
3961      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
3962      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3963      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3964      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
3965      *reqbyteptr = reqbyte;      *firstcharflagsptr = firstcharflags;
3966        *reqcharptr = reqchar;
3967        *reqcharflagsptr = reqcharflags;
3968      *codeptr = code;      *codeptr = code;
3969      *ptrptr = ptr;      *ptrptr = ptr;
3970      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3539  for (;; ptr++) Line 3988  for (;; ptr++)
3988      previous = NULL;      previous = NULL;
3989      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3990        {        {
3991        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
3992        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3993        }        }
3994      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3551  for (;; ptr++) Line 4000  for (;; ptr++)
4000      break;      break;
4001    
4002      /* 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
4003      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4004    
4005      case CHAR_DOT:      case CHAR_DOT:
4006      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4007      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
4008      zeroreqbyte = reqbyte;      zerofirstcharflags = firstcharflags;
4009        zeroreqchar = reqchar;
4010        zeroreqcharflags = reqcharflags;
4011      previous = code;      previous = code;
4012      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4013      break;      break;
# Line 3611  for (;; ptr++) Line 4062  for (;; ptr++)
4062          {          {
4063          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
4064            ptr++;            ptr++;
4065          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)  
4066            ptr += 3;            ptr += 3;
4067          else          else
4068            break;            break;
# Line 3631  for (;; ptr++) Line 4081  for (;; ptr++)
4081          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4082        {        {
4083        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4084        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4085        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
4086          zerofirstcharflags = firstcharflags;
4087        break;        break;
4088        }        }
4089    
# Line 3642  for (;; ptr++) Line 4093  for (;; ptr++)
4093    
4094      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4095    
4096      /* 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:
4097      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 <
4098      valued UTF-8 characters, we don't yet do any optimization. */      256 character; class_one_char will be 1 if the class contains just one
4099        character. */
4100    
4101      class_charcount = 0;      class_has_8bitchar = 0;
4102      class_lastchar = -1;      class_one_char = 0;
4103    
4104      /* 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
4105      temporary bit of memory, in case the class contains only 1 character (less      temporary bit of memory, in case the class contains fewer than two
4106      than 256), because in that case the compiled code doesn't use the bit map.      8-bit characters because in that case the compiled code doesn't use the bit
4107      */      map. */
4108    
4109      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4110    
4111  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4112      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;
4113      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4114      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4115  #endif  #endif
4116    
4117      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
4118      means that an initial ] is taken as a data character. At the start of the      means that an initial ] is taken as a data character. At the start of the
4119      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
4120    
4121      if (c != 0) do      if (c != CHAR_NULL) do
4122        {        {
4123        const uschar *oldptr;        const pcre_uchar *oldptr;
4124    
4125  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4126        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
4127          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4128          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
4129          }          }
4130    #endif
4131    
4132        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4133          /* In the pre-compile phase, accumulate the length of any extra
4134        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
4135        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4136        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4137          however. */
4138        if (lengthptr != NULL)  
4139          {        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4140          *lengthptr += class_utf8data - class_utf8data_base;          {
4141          class_utf8data = class_utf8data_base;          xclass = TRUE;
4142            *lengthptr += class_uchardata - class_uchardata_base;
4143            class_uchardata = class_uchardata_base;
4144          }          }
   
4145  #endif  #endif
4146    
4147        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3714  for (;; ptr++) Line 4169  for (;; ptr++)
4169          {          {
4170          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
4171          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
4172          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
4173          uschar pbits[32];          pcre_uint8 pbits[32];
4174    
4175          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
4176            {            {
# Line 3770  for (;; ptr++) Line 4225  for (;; ptr++)
4225          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
4226    
4227          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
4228            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
4229    
4230          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
4231    
# Line 3785  for (;; ptr++) Line 4240  for (;; ptr++)
4240              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4241            }            }
4242    
4243          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4244          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4245    
4246          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3801  for (;; ptr++) Line 4256  for (;; ptr++)
4256            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4257    
4258          ptr = tempptr + 1;          ptr = tempptr + 1;
4259          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 character. */
4260            class_has_8bitchar = 1;
4261            /* Every class contains at least two characters. */
4262            class_one_char = 2;
4263          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4264          }          }
4265    
4266        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
4267        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
4268        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
4269        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
4270        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_one_char bigger
4271        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
4272          as literal characters (by default), or are faulted if
4273        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4274    
4275        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4276          {          {
4277          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options,
4278              TRUE);
4279          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4280            if (escape == 0) c = ec;
4281          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4282          else if (-c == ESC_N)            /* \N is not supported in a class */          else if (escape == ESC_N)          /* \N is not supported in a class */
4283            {            {
4284            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4285            goto FAILED;            goto FAILED;
4286            }            }
4287          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4288            {            {
4289            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4290              {              {
# Line 3833  for (;; ptr++) Line 4293  for (;; ptr++)
4293            else inescq = TRUE;            else inescq = TRUE;
4294            continue;            continue;
4295            }            }
4296          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4297    
4298          if (c < 0)          else
4299            {            {
4300            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4301            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
4302              class_has_8bitchar++;
4303              /* Every class contains at least two characters. */
4304              class_one_char += 2;
4305    
4306            switch (-c)            switch (escape)
4307              {              {
4308  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4309              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3850  for (;; ptr++) Line 4313  for (;; ptr++)
4313              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4314              case ESC_SU:              case ESC_SU:
4315              nestptr = ptr;              nestptr = ptr;
4316              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4317              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4318              continue;              continue;
4319  #endif  #endif
4320              case ESC_d:              case ESC_d:
# Line 3874  for (;; ptr++) Line 4337  for (;; ptr++)
4337    
4338              /* Perl 5.004 onwards omits VT from \s, but we must preserve it              /* Perl 5.004 onwards omits VT from \s, but we must preserve it
4339              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4340              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4341                EBCDIC, so we lazily just adjust the appropriate bit. */
4342    
4343              case ESC_s:              case ESC_s:
4344              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 3888  for (;; ptr++) Line 4352  for (;; ptr++)
4352              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4353              continue;              continue;
4354    
4355                /* The rest apply in both UCP and non-UCP cases. */
4356    
4357              case ESC_h:              case ESC_h:
4358              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4359              SETBIT(classbits, 0x20); /* SPACE */                PRIV(hspace_list), NOTACHAR);
             SETBIT(classbits, 0xa0); /* NSBP */  
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);  
               }  
 #endif  
4360              continue;              continue;
4361    
4362              case ESC_H:              case ESC_H:
4363              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4364                {                cd, PRIV(hspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x09/8: x ^= 1 << (0x09%8); break;  
                 case 0x20/8: x ^= 1 << (0x20%8); break;  
                 case 0xa0/8: x ^= 1 << (0xa0%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
   
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);  
               }  
 #endif  
4365              continue;              continue;
4366    
4367              case ESC_v:              case ESC_v:
4368              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4369              SETBIT(classbits, 0x0b); /* VT */                PRIV(vspace_list), NOTACHAR);
             SETBIT(classbits, 0x0c); /* FF */  
             SETBIT(classbits, 0x0d); /* CR */  
             SETBIT(classbits, 0x85); /* NEL */  
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);  
               }  
 #endif  
4370              continue;              continue;
4371    
4372              case ESC_V:              case ESC_V:
4373              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4374                {                cd, PRIV(vspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x0a/8: x ^= 1 << (0x0a%8);  
                              x ^= 1 << (0x0b%8);  
                              x ^= 1 << (0x0c%8);  
                              x ^= 1 << (0x0d%8);  
                              break;  
                 case 0x85/8: x ^= 1 << (0x85%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
   
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);  
               }  
 #endif  
4375              continue;              continue;
4376    
4377  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4009  for (;; ptr++) Line 4379  for (;; ptr++)
4379              case ESC_P:              case ESC_P:
4380                {                {
4381                BOOL negated;                BOOL negated;
4382                int pdata;                unsigned int ptype = 0, pdata = 0;
4383                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4384                if (ptype < 0) goto FAILED;                  goto FAILED;
4385                class_utf8 = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_utf8data++ = ((-c == ESC_p) != negated)?  
4386                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4387                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4388                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4389                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4390                continue;                continue;
4391                }                }
4392  #endif  #endif
# Line 4031  for (;; ptr++) Line 4400  for (;; ptr++)
4400                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4401                goto FAILED;                goto FAILED;
4402                }                }
4403              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4404              c = *ptr;              /* Get the final character and fall through */              class_one_char -= 2;     /* Undo the speculative increase. */
4405                c = *ptr;                /* Get the final character and fall through */
4406              break;              break;
4407              }              }
4408            }            }
4409    
4410          /* Fall through if we have a single character (c >= 0). This may be          /* Fall through if the escape just defined a single character (c >= 0).
4411          greater than 256 in UTF-8 mode. */          This may be greater than 256. */
4412    
4413            escape = 0;
4414    
4415          }   /* End of backslash handling */          }   /* End of backslash handling */
4416    
4417        /* A single character may be followed by '-' to form a range. However,        /* A character may be followed by '-' to form a range. However, Perl does
4418        Perl does not permit ']' to be the end of the range. A '-' character        not permit ']' to be the end of the range. A '-' character at the end is
4419        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4420        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4421    
4422        CHECK_RANGE:        CHECK_RANGE:
4423        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4053  for (;; ptr++) Line 4425  for (;; ptr++)
4425          inescq = FALSE;          inescq = FALSE;
4426          ptr += 2;          ptr += 2;
4427          }          }
   
4428        oldptr = ptr;        oldptr = ptr;
4429    
4430        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4431    
4432        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4433    
# Line 4064  for (;; ptr++) Line 4435  for (;; ptr++)
4435    
4436        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4437          {          {
4438          int d;          pcre_uint32 d;
4439          ptr += 2;          ptr += 2;
4440          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4441    
# Line 4080  for (;; ptr++) Line 4451  for (;; ptr++)
4451            break;            break;
4452            }            }
4453    
4454          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          /* Minus (hyphen) at the end of a class is treated as a literal, so put
4455            back the pointer and jump to handle the character that preceded it. */
4456    
4457            if (*ptr == CHAR_NULL || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4458            {            {
4459            ptr = oldptr;            ptr = oldptr;
4460            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4461            }            }
4462    
4463  #ifdef SUPPORT_UTF8          /* Otherwise, we have a potential range; pick up the next character */
4464          if (utf8)  
4465    #ifdef SUPPORT_UTF
4466            if (utf)
4467            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4468            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4469            }            }
# Line 4101  for (;; ptr++) Line 4477  for (;; ptr++)
4477    
4478          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4479            {            {
4480            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4481              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4482            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4483    
4484            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4485    
4486            if (d < 0)            if (descape != 0)
4487              {              {
4488              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4489                {                {
4490                ptr = oldptr;                ptr = oldptr;
4491                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4492                }                }
4493              }              }
4494            }            }
4495    
4496          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4497          one-character ranges */          one-character ranges. */
4498    
4499          if (d < c)          if (d < c)
4500            {            {
4501            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4502            goto FAILED;            goto FAILED;
4503            }            }
4504            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4505    
4506          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4507            cannot be done anymore. Any value greater than 1 indicates that there
4508            is more than one character. */
4509    
4510          /* Remember \r or \n */          class_one_char = 2;
4511    
4512          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          /* Remember an explicit \r or \n, and add the range to the class. */
4513    
4514          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
         matching, we have to use an XCLASS with extra data items. Caseless  
         matching for characters > 127 is available only if UCP support is  
         available. */  
   
 #ifdef SUPPORT_UTF8  
         if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))  
           {  
           class_utf8 = TRUE;  
   
           /* With UCP support, we can find the other case equivalents of  
           the relevant characters. There may be several ranges. Optimize how  
           they fit with the basic range. */  
4515    
4516  #ifdef SUPPORT_UCP          class_has_8bitchar +=
4517            if ((options & PCRE_CASELESS) != 0)            add_to_class(classbits, &class_uchardata, options, cd, c, d);
             {  
             unsigned int occ, ocd;  
             unsigned int cc = c;  
             unsigned int origd = d;  
             while (get_othercase_range(&cc, origd, &occ, &ocd))  
               {  
               if (occ >= (unsigned int)c &&  
                   ocd <= (unsigned int)d)  
                 continue;                          /* Skip embedded ranges */  
   
               if (occ < (unsigned int)c  &&  
                   ocd >= (unsigned int)c - 1)      /* Extend the basic range */  
                 {                                  /* if there is overlap,   */  
                 c = occ;                           /* noting that if occ < c */  
                 continue;                          /* we can't have ocd > d  */  
                 }                                  /* because a subrange is  */  
               if (ocd > (unsigned int)d &&  
                   occ <= (unsigned int)d + 1)      /* always shorter than    */  
                 {                                  /* the basic range.       */  
                 d = ocd;  
                 continue;  
                 }  
4518    
4519                if (occ == ocd)          continue;   /* Go get the next char in the class */
4520                  {          }
                 *class_utf8data++ = XCL_SINGLE;  
                 }  
               else  
                 {  
                 *class_utf8data++ = XCL_RANGE;  
                 class_utf8data += _pcre_ord2utf8(occ, class_utf8data);  
                 }  
               class_utf8data += _pcre_ord2utf8(ocd, class_utf8data);  
               }  
             }  
 #endif  /* SUPPORT_UCP */  
4521    
4522            /* Now record the original range, possibly modified for UCP caseless        /* Handle a single character - we can get here for a normal non-escape
4523            overlapping ranges. */        char, or after \ that introduces a single character or for an apparent
4524          range that isn't. Only the value 1 matters for class_one_char, so don't
4525          increase it if it is already 2 or more ... just in case there's a class
4526          with a zillion characters in it. */
4527    
4528          CLASS_SINGLE_CHARACTER:
4529          if (class_one_char < 2) class_one_char++;
4530    
4531          /* If class_one_char is 1, we have the first single character in the
4532          class, and there have been no prior ranges, or XCLASS items generated by
4533          escapes. If this is the final character in the class, we can optimize by
4534          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4535          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4536          to be set. Otherwise, there can be no first char if this item is first,
4537          whatever repeat count may follow. In the case of reqchar, save the
4538          previous value for reinstating. */
4539    
4540            *class_utf8data++ = XCL_RANGE;        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
4541            class_utf8data += _pcre_ord2utf8(c, class_utf8data);          {
4542            class_utf8data += _pcre_ord2utf8(d, class_utf8data);          ptr++;
4543            zeroreqchar = reqchar;
4544            /* With UCP support, we are done. Without UCP support, there is no          zeroreqcharflags = reqcharflags;
           caseless matching for UTF-8 characters > 127; we can use the bit map  
           for the smaller ones. */  
4545    
4546            if (negate_class)
4547              {
4548  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4549            continue;    /* With next character in the class */            int d;
4550  #else  #endif
4551            if ((options & PCRE_CASELESS) == 0 || c > 127) continue;            if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4552              zerofirstchar = firstchar;
4553            /* Adjust upper limit and fall through to set up the map */            zerofirstcharflags = firstcharflags;
4554    
4555            d = 127;            /* For caseless UTF-8 mode when UCP support is available, check
4556              whether this character has more than one other case. If so, generate
4557  #endif  /* SUPPORT_UCP */            a special OP_NOTPROP item instead of OP_NOTI. */
           }  
 #endif  /* SUPPORT_UTF8 */  
   
         /* We use the bit map for all cases when not in UTF-8 mode; else  
         ranges that lie entirely within 0-127 when there is UCP support; else  
         for partial ranges without UCP support. */  
   
         class_charcount += d - c + 1;  
         class_lastchar = d;  
   
         /* We can save a bit of time by skipping this in the pre-compile. */  
4558    
4559          if (lengthptr == NULL) for (; c <= d; c++)  #ifdef SUPPORT_UCP
4560            {            if (utf && (options & PCRE_CASELESS) != 0 &&
4561            classbits[c/8] |= (1 << (c&7));                (d = UCD_CASESET(c)) != 0)
           if ((options & PCRE_CASELESS) != 0)  
4562              {              {
4563              int uc = cd->fcc[c];           /* flip case */              *code++ = OP_NOTPROP;
4564              classbits[uc/8] |= (1 << (uc&7));              *code++ = PT_CLIST;
4565                *code++ = d;
4566              }              }
4567            }            else
4568    #endif
4569          continue;   /* Go get the next char in the class */            /* Char has only one other case, or UCP not available */
         }  
4570    
4571        /* Handle a lone single character - we can get here for a normal              {
4572        non-escape char, or after \ that introduces a single character or for an              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4573        apparent range that isn't. */  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4574                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4575                  code += PRIV(ord2utf)(c, code);
4576                else
4577    #endif
4578                  *code++ = c;
4579                }
4580    
4581        LONE_SINGLE_CHARACTER:            /* We are finished with this character class */
4582    
4583        /* Handle a character that cannot go in the bit map */            goto END_CLASS;
4584              }
4585    
4586  #ifdef SUPPORT_UTF8          /* For a single, positive character, get the value into mcbuffer, and
4587        if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))          then we can handle this with the normal one-character code. */
         {  
         class_utf8 = TRUE;  
         *class_utf8data++ = XCL_SINGLE;  
         class_utf8data += _pcre_ord2utf8(c, class_utf8data);  
4588    
4589  #ifdef SUPPORT_UCP  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4590          if ((options & PCRE_CASELESS) != 0)          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4591              mclength = PRIV(ord2utf)(c, mcbuffer);
4592            else
4593    #endif
4594            {            {
4595            unsigned int othercase;            mcbuffer[0] = c;
4596            if ((othercase = UCD_OTHERCASE(c)) != c)            mclength = 1;
             {  
             *class_utf8data++ = XCL_SINGLE;  
             class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);  
             }  
4597            }            }
4598  #endif  /* SUPPORT_UCP */          goto ONE_CHAR;
4599            }       /* End of 1-char optimization */
4600    
4601          }        /* There is more than one character in the class, or an XCLASS item
4602        else        has been generated. Add this character to the class. */
 #endif  /* SUPPORT_UTF8 */  
4603    
4604        /* Handle a single-byte character */        class_has_8bitchar +=
4605          {          add_to_class(classbits, &class_uchardata, options, cd, c, c);
         classbits[c/8] |= (1 << (c&7));  
         if ((options & PCRE_CASELESS) != 0)  
           {  
           c = cd->fcc[c];   /* flip case */  
           classbits[c/8] |= (1 << (c&7));  
           }  
         class_charcount++;  
         class_lastchar = c;  
         }  
4606        }        }
4607    
4608      /* Loop until ']' reached. This "while" is the end of the "do" far above.      /* Loop until ']' reached. This "while" is the end of the "do" far above.
4609      If we are at the end of an internal nested string, revert to the outer      If we are at the end of an internal nested string, revert to the outer
4610      string. */      string. */
4611    
4612      while (((c = *(++ptr)) != 0 ||      while (((c = *(++ptr)) != CHAR_NULL ||
4613             (nestptr != NULL &&             (nestptr != NULL &&
4614               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != CHAR_NULL)) &&
4615             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
4616    
4617      /* Check for missing terminating ']' */      /* Check for missing terminating ']' */
4618    
4619      if (c == 0)      if (c == CHAR_NULL)
4620        {        {
4621        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
4622        goto FAILED;        goto FAILED;
4623        }        }
4624    
4625      /* If class_charcount is 1, we saw precisely one character whose value is      /* We will need an XCLASS if data has been placed in class_uchardata. In
4626      less than 256. As long as there were no characters >= 128 and there was no      the second phase this is a sufficient test. However, in the pre-compile
4627      use of \p or \P, in other words, no use of any XCLASS features, we can      phase, class_uchardata gets emptied to prevent workspace overflow, so it
4628      optimize.      only if the very last character in the class needs XCLASS will it contain
4629        anything at this point. For this reason, xclass gets set TRUE above when
4630      In UTF-8 mode, we can optimize the negative case only if there were no      uchar_classdata is emptied, and that's why this code is the way it is here
4631      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      instead of just doing a test on class_uchardata below. */
4632      operate on single-bytes characters only. This is an historical hangover.  
4633      Maybe one day we can tidy these opcodes to handle multi-byte characters.  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4634        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4635      The optimization throws away the bit map. We turn the item into a  #endif
4636      1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
4637      Note that OP_NOT[I] does not support multibyte characters. In the positive      /* If this is the first thing in the branch, there can be no first char
4638      case, it can cause firstbyte to be set. Otherwise, there can be no first      setting, whatever the repeat count. Any reqchar setting must remain
4639      char if this item is first, whatever repeat count may follow. In the case      unchanged after any kind of repeat. */
4640      of reqbyte, save the previous value for reinstating. */  
4641        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4642  #ifdef SUPPORT_UTF8      zerofirstchar = firstchar;
4643      if (class_charcount == 1 && !class_utf8 &&      zerofirstcharflags = firstcharflags;
4644        (!utf8 || !negate_class || class_lastchar < 128))      zeroreqchar = reqchar;
4645  #else      zeroreqcharflags = reqcharflags;
     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;  
4646    
4647      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4648      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 4652  for (;; ptr++)
4652      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
4653      actual compiled code. */      actual compiled code. */
4654    
4655  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4656      if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))      if (xclass && (!should_flip_negation || (options & PCRE_UCP) != 0))
4657    #elif !defined COMPILE_PCRE8
4658        if (xclass && !should_flip_negation)
4659    #endif
4660    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4661        {        {
4662        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_uchardata++ = XCL_END;    /* Marks the end of extra data */
4663        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
4664        code += LINK_SIZE;        code += LINK_SIZE;
4665        *code = negate_class? XCL_NOT : 0;        *code = negate_class? XCL_NOT:0;
4666    
4667        /* 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;
4668        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. */
4669    
4670        if (class_charcount > 0)        if (class_has_8bitchar > 0)
4671          {          {
4672          *code++ |= XCL_MAP;          *code++ |= XCL_MAP;
4673          memmove(code + 32, code, class_utf8data - code);          memmove(code + (32 / sizeof(pcre_uchar)), code,
4674              IN_UCHARS(class_uchardata - code));
4675          memcpy(code, classbits, 32);          memcpy(code, classbits, 32);
4676          code = class_utf8data + 32;          code = class_uchardata + (32 / sizeof(pcre_uchar));
4677          }          }
4678        else code = class_utf8data;        else code = class_uchardata;
4679    
4680        /* Now fill in the complete length of the item */        /* Now fill in the complete length of the item */
4681    
4682        PUT(previous, 1, code - previous);        PUT(previous, 1, (int)(code - previous));
4683        break;   /* End of class handling */        break;   /* End of class handling */
4684        }        }
4685  #endif  #endif
# Line 4394  for (;; ptr++) Line 4691  for (;; ptr++)
4691      negating it if necessary. */      negating it if necessary. */
4692    
4693      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
4694      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  
4695        {        {
4696          if (negate_class)
4697            for (c = 0; c < 32; c++) classbits[c] = ~classbits[c];
4698        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4699        }        }
4700      code += 32;      code += 32 / sizeof(pcre_uchar);
4701    
4702        END_CLASS:
4703      break;      break;
4704    
4705    
# Line 4440  for (;; ptr++) Line 4736  for (;; ptr++)
4736    
4737      if (repeat_min == 0)      if (repeat_min == 0)
4738        {        {
4739        firstbyte = zerofirstbyte;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4740        reqbyte = zeroreqbyte;        /* Ditto */        firstcharflags = zerofirstcharflags;
4741          reqchar = zeroreqchar;        /* Ditto */
4742          reqcharflags = zeroreqcharflags;
4743        }        }
4744    
4745      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4480  for (;; ptr++) Line 4778  for (;; ptr++)
4778      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
4779      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,
4780      but for the moment we rely on the code for repeating groups. */      but for the moment we rely on the code for repeating groups. */
4781    
4782      if (*previous == OP_RECURSE)      if (*previous == OP_RECURSE)
4783        {        {
4784        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);        memmove(previous + 1 + LINK_SIZE, previous, IN_UCHARS(1 + LINK_SIZE));
4785        *previous = OP_ONCE;        *previous = OP_ONCE;
4786        PUT(previous, 1, 2 + 2*LINK_SIZE);        PUT(previous, 1, 2 + 2*LINK_SIZE);
4787        previous[2 + 2*LINK_SIZE] = OP_KET;        previous[2 + 2*LINK_SIZE] = OP_KET;
# Line 4504  for (;; ptr++) Line 4802  for (;; ptr++)
4802    
4803      /* Now handle repetition for the different types of item. */      /* Now handle repetition for the different types of item. */
4804    
4805      /* If previous was a character match, abolish the item and generate a      /* If previous was a character or negated character match, abolish the item
4806      repeat item instead. If a char item has a minumum of more than one, ensure      and generate a repeat item instead. If a char item has a minimum of more
4807      that it is set in reqbyte - it might not be if a sequence such as x{3} is      than one, ensure that it is set in reqchar - it might not be if a sequence
4808      the first thing in a branch because the x will have gone into firstbyte      such as x{3} is the first thing in a branch because the x will have gone
4809      instead.  */      into firstchar instead.  */
4810    
4811      if (*previous == OP_CHAR || *previous == OP_CHARI)      if (*previous == OP_CHAR || *previous == OP_CHARI
4812            || *previous == OP_NOT || *previous == OP_NOTI)
4813        {        {
4814        op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;        switch (*previous)
4815            {
4816            default: /* Make compiler happy. */
4817            case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;
4818            case OP_CHARI: op_type = OP_STARI - OP_STAR; break;
4819            case OP_NOT:   op_type = OP_NOTSTAR - OP_STAR; break;
4820            case OP_NOTI:  op_type = OP_NOTSTARI - OP_STAR; break;
4821            }
4822    
4823        /* 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
4824        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
4825        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
4826        length rather than a small character. */        it's a length rather than a small character. */
4827    
4828  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4829        if (utf8 && (code[-1] & 0x80) != 0)        if (utf && NOT_FIRSTCHAR(code[-1]))
4830          {          {
4831          uschar *lastchar = code - 1;          pcre_uchar *lastchar = code - 1;
4832          while((*lastchar & 0xc0) == 0x80) lastchar--;          BACKCHAR(lastchar);
4833          c = code - lastchar;            /* Length of UTF-8 character */          c = (int)(code - lastchar);     /* Length of UTF-8 character */
4834          memcpy(utf8_char, lastchar, c); /* Save the char */          memcpy(utf_chars, lastchar, IN_UCHARS(c)); /* Save the char */
4835          c |= 0x80;                      /* Flag c as a length */          c |= UTF_LENGTH;                /* Flag c as a length */
4836          }          }
4837        else        else
4838  #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. */  
4839    
4840          /* Handle the case of a single charater - either with no UTF support, or
4841          with UTF disabled, or for a single character UTF character. */
4842          {          {
4843          c = code[-1];          c = code[-1];
4844          if (repeat_min > 1) reqbyte = c | req_caseopt | cd->req_varyopt;          if (*previous <= OP_CHARI && repeat_min > 1)
4845              {
4846              reqchar = c;
4847              reqcharflags = req_caseopt | cd->req_varyopt;
4848              }
4849          }          }
4850    
4851        /* 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 4855  for (;; ptr++)
4855    
4856        if (!possessive_quantifier &&        if (!possessive_quantifier &&
4857            repeat_max < 0 &&            repeat_max < 0 &&
4858            check_auto_possessive(previous, utf8, ptr + 1, options, cd))            check_auto_possessive(previous, utf, ptr + 1, options, cd))
4859          {          {