/[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 758 by ph10, Mon Nov 21 12:05:36 2011 UTC revision 1348 by ph10, Fri Jul 5 10:38:37 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) 1997-2013 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. */  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
109    /(?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
111    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()
113    kicks in at the same number of forward references in all cases. */
114    
115  #define COMPILE_WORK_SIZE (4096)  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)
116    #define COMPILE_WORK_SIZE_MAX (100*COMPILE_WORK_SIZE)
117    
118  /* The overrun tests check for a slightly smaller size so that they detect the  /* The overrun tests check for a slightly smaller size so that they detect the
119  overrun before it actually does run off the end of the data block. */  overrun before it actually does run off the end of the data block. */
120    
121  #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 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 231  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 266  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 365  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 388  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 399  static const char error_texts[] = Line 487  static const char error_texts[] =
487    "a numbered reference must not be zero\0"    "a numbered reference must not be zero\0"
488    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"
489    /* 60 */    /* 60 */
490    "(*VERB) not recognized\0"    "(*VERB) not recognized or malformed\0"
491    "number is too big\0"    "number is too big\0"
492    "subpattern name expected\0"    "subpattern name expected\0"
493    "digit expected after (?+\0"    "digit expected after (?+\0"
# Line 407  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"
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      "setting UTF is disabled by the application\0"
512    ;    ;
513    
514  /* 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 431  For convenience, we use the same bit def Line 527  For convenience, we use the same bit def
527    
528  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
529    
530    /* Using a simple comparison for decimal numbers rather than a memory read
531    is much faster, and the resulting code is simpler (the compiler turns it
532    into a subtraction and unsigned comparison). */
533    
534    #define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9)
535    
536  #ifndef EBCDIC  #ifndef EBCDIC
537    
538  /* 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
539  UTF-8 mode. */  UTF-8 mode. */
540    
541  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
542    {    {
543    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
544    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */
# Line 475  static const unsigned char digitab[] = Line 577  static const unsigned char digitab[] =
577    
578  /* 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. */
579    
580  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
581    {    {
582    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
583    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */
# Line 510  static const unsigned char digitab[] = Line 612  static const unsigned char digitab[] =
612    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */
613    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */
614    
615  static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */  static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */
616    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */
617    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */
618    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */
# Line 546  static const unsigned char ebcdic_charta Line 648  static const unsigned char ebcdic_charta
648  #endif  #endif
649    
650    
 /* 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 *);  
   
   
651    
652  /*************************************************  /*************************************************
653  *            Find an error text                  *  *            Find an error text                  *
# Line 573  find_error_text(int n) Line 668  find_error_text(int n)
668  const char *s = error_texts;  const char *s = error_texts;
669  for (; n > 0; n--)  for (; n > 0; n--)
670    {    {
671    while (*s++ != 0) {};    while (*s++ != CHAR_NULL) {};
672    if (*s == 0) return "Error text not found (please report)";    if (*s == CHAR_NULL) return "Error text not found (please report)";
673    }    }
674  return s;  return s;
675  }  }
676    
677    
678  /*************************************************  /*************************************************
679    *           Expand the workspace                 *
680    *************************************************/
681    
682    /* This function is called during the second compiling phase, if the number of
683    forward references fills the existing workspace, which is originally a block on
684    the stack. A larger block is obtained from malloc() unless the ultimate limit
685    has been reached or the increase will be rather small.
686    
687    Argument: pointer to the compile data block
688    Returns:  0 if all went well, else an error number
689    */
690    
691    static int
692    expand_workspace(compile_data *cd)
693    {
694    pcre_uchar *newspace;
695    int newsize = cd->workspace_size * 2;
696    
697    if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;
698    if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX ||
699        newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)
700     return ERR72;
701    
702    newspace = (PUBL(malloc))(IN_UCHARS(newsize));
703    if (newspace == NULL) return ERR21;
704    memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar));
705    cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace);
706    if (cd->workspace_size > COMPILE_WORK_SIZE)
707      (PUBL(free))((void *)cd->start_workspace);
708    cd->start_workspace = newspace;
709    cd->workspace_size = newsize;
710    return 0;
711    }
712    
713    
714    
715    /*************************************************
716  *            Check for counted repeat            *  *            Check for counted repeat            *
717  *************************************************/  *************************************************/
718    
# Line 596  Returns:    TRUE or FALSE Line 728  Returns:    TRUE or FALSE
728  */  */
729    
730  static BOOL  static BOOL
731  is_counted_repeat(const uschar *p)  is_counted_repeat(const pcre_uchar *p)
732  {  {
733  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
734  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
735    while (IS_DIGIT(*p)) p++;
736  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
737    
738  if (*p++ != CHAR_COMMA) return FALSE;  if (*p++ != CHAR_COMMA) return FALSE;
739  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
740    
741  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
742  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
743    while (IS_DIGIT(*p)) p++;
744    
745  return (*p == CHAR_RIGHT_CURLY_BRACKET);  return (*p == CHAR_RIGHT_CURLY_BRACKET);
746  }  }
# Line 618  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 752  return (*p == CHAR_RIGHT_CURLY_BRACKET);
752  *************************************************/  *************************************************/
753    
754  /* 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
755  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
756  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
757  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
758  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
759  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
760  sequence.  escape sequence.
761    
762  Arguments:  Arguments:
763    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
764      chptr          points to the data character
765    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
766    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
767    options        the options bits    options        the options bits
768    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
769    
770  Returns:         zero or positive => a data character  Returns:         zero => a data character
771                   negative => a special escape sequence                   positive => a special escape sequence
772                     negative => a back reference
773                   on error, errorcodeptr is set                   on error, errorcodeptr is set
774  */  */
775    
776  static int  static int
777  check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
778    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
779  {  {
780  BOOL utf8 = (options & PCRE_UTF8) != 0;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
781  const uschar *ptr = *ptrptr + 1;  BOOL utf = (options & PCRE_UTF8) != 0;
782  int c, i;  const pcre_uchar *ptr = *ptrptr + 1;
783    pcre_uint32 c;
784    int escape = 0;
785    int i;
786    
787  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
788  ptr--;                            /* Set pointer back to the last byte */  ptr--;                            /* Set pointer back to the last byte */
789    
790  /* 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. */
791    
792  if (c == 0) *errorcodeptr = ERR1;  if (c == CHAR_NULL) *errorcodeptr = ERR1;
793    
794  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
795  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.
796  Otherwise further processing may be required. */  Otherwise further processing may be required. */
797    
798  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
799  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */  /* Not alphanumeric */
800  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if (c < CHAR_0 || c > CHAR_z) {}
801    else if ((i = escapes[c - CHAR_0]) != 0)
802      { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
803    
804  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
805  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */  /* Not alphanumeric */
806  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
807    else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
808  #endif  #endif
809    
810  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
811    
812  else  else
813    {    {
814    const uschar *oldptr;    const pcre_uchar *oldptr;
815    BOOL braced, negated;    BOOL braced, negated, overflow;
816      int s;
817    
818    switch (c)    switch (c)
819      {      {
# Line 687  else Line 830  else
830        {        {
831        /* In JavaScript, \u must be followed by four hexadecimal numbers.        /* In JavaScript, \u must be followed by four hexadecimal numbers.
832        Otherwise it is a lowercase u letter. */        Otherwise it is a lowercase u letter. */
833        if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
834             && (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0)          && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
835            && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
836            && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
837          {          {
838          c = 0;          c = 0;
839          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
840            {            {
841            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
842  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
843            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
844            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 702  else Line 847  else
847            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
848  #endif  #endif
849            }            }
850    
851    #if defined COMPILE_PCRE8
852            if (c > (utf ? 0x10ffffU : 0xffU))
853    #elif defined COMPILE_PCRE16
854            if (c > (utf ? 0x10ffffU : 0xffffU))
855    #elif defined COMPILE_PCRE32
856            if (utf && c > 0x10ffffU)
857    #endif
858              {
859              *errorcodeptr = ERR76;
860              }
861            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
862          }          }
863        }        }
864      else      else
# Line 728  else Line 885  else
885      (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
886      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
887      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
888      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
889    
890      case CHAR_g:      case CHAR_g:
891      if (isclass) break;      if (isclass) break;
892      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
893        {        {
894        c = -ESC_g;        escape = ESC_g;
895        break;        break;
896        }        }
897    
# Line 742  else Line 899  else
899    
900      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
901        {        {
902        const uschar *p;        const pcre_uchar *p;
903        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++)
904          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
905        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET)
906          {          {
907          c = -ESC_k;          escape = ESC_k;
908          break;          break;
909          }          }
910        braced = TRUE;        braced = TRUE;
# Line 762  else Line 919  else
919        }        }
920      else negated = FALSE;      else negated = FALSE;
921    
922      c = 0;      /* The integer range is limited by the machine's int representation. */
923      while ((digitab[ptr[1]] & ctype_digit) != 0)      s = 0;
924        c = c * 10 + *(++ptr) - CHAR_0;      overflow = FALSE;
925        while (IS_DIGIT(ptr[1]))
926      if (c < 0)   /* Integer overflow */        {
927          if (s > INT_MAX / 10 - 1) /* Integer overflow */
928            {
929            overflow = TRUE;
930            break;
931            }
932          s = s * 10 + (int)(*(++ptr) - CHAR_0);
933          }
934        if (overflow) /* Integer overflow */
935        {        {
936          while (IS_DIGIT(ptr[1]))
937            ptr++;
938        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
939        break;        break;
940        }        }
# Line 778  else Line 945  else
945        break;        break;
946        }        }
947    
948      if (c == 0)      if (s == 0)
949        {        {
950        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
951        break;        break;
# Line 786  else Line 953  else
953    
954      if (negated)      if (negated)
955        {        {
956        if (c > bracount)        if (s > bracount)
957          {          {
958          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
959          break;          break;
960          }          }
961        c = bracount - (c - 1);        s = bracount - (s - 1);
962        }        }
963    
964      c = -(ESC_REF + c);      escape = -s;
965      break;      break;
966    
967      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 815  else Line 982  else
982      if (!isclass)      if (!isclass)
983        {        {
984        oldptr = ptr;        oldptr = ptr;
985        c -= CHAR_0;        /* The integer range is limited by the machine's int representation. */
986        while ((digitab[ptr[1]] & ctype_digit) != 0)        s = (int)(c -CHAR_0);
987          c = c * 10 + *(++ptr) - CHAR_0;        overflow = FALSE;
988        if (c < 0)    /* Integer overflow */        while (IS_DIGIT(ptr[1]))
989          {          {
990            if (s > INT_MAX / 10 - 1) /* Integer overflow */
991              {
992              overflow = TRUE;
993              break;
994              }
995            s = s * 10 + (int)(*(++ptr) - CHAR_0);
996            }
997          if (overflow) /* Integer overflow */
998            {
999            while (IS_DIGIT(ptr[1]))
1000              ptr++;
1001          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1002          break;          break;
1003          }          }
1004        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1005          {          {
1006          c = -(ESC_REF + c);          escape = -s;
1007          break;          break;
1008          }          }
1009        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 845  else Line 1023  else
1023      /* \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
1024      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
1025      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
1026      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,
1027      than 3 octal digits. */      but no more than 3 octal digits. */
1028    
1029      case CHAR_0:      case CHAR_0:
1030      c -= CHAR_0;      c -= CHAR_0;
1031      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1032          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1033      if (!utf8 && c > 255) *errorcodeptr = ERR51;  #ifdef COMPILE_PCRE8
1034        if (!utf && c > 0xff) *errorcodeptr = ERR51;
1035    #endif
1036      break;      break;
1037    
1038      /* \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
1039      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.
1040      treated as a data character. */      If not, { is treated as a data character. */
1041    
1042      case CHAR_x:      case CHAR_x:
1043      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1044        {        {
1045        /* In JavaScript, \x must be followed by two hexadecimal numbers.        /* In JavaScript, \x must be followed by two hexadecimal numbers.
1046        Otherwise it is a lowercase x letter. */        Otherwise it is a lowercase x letter. */
1047        if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0)        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1048            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1049          {          {
1050          c = 0;          c = 0;
1051          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1052            {            {
1053            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1054  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1055            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1056            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 884  else Line 1065  else
1065    
1066      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1067        {        {
1068        const uschar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
       int count = 0;  
1069    
1070        c = 0;        c = 0;
1071        while ((digitab[*pt] & ctype_xdigit) != 0)        overflow = FALSE;
1072          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1073          {          {
1074          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1075          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1076          count++;  
1077    #ifdef COMPILE_PCRE32
1078            if (c >= 0x10000000l) { overflow = TRUE; break; }
1079    #endif
1080    
1081  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1082          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
# Line 901  else Line 1085  else
1085          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 */
1086          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1087  #endif  #endif
1088    
1089    #if defined COMPILE_PCRE8
1090            if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }
1091    #elif defined COMPILE_PCRE16
1092            if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }
1093    #elif defined COMPILE_PCRE32
1094            if (utf && c > 0x10ffffU) { overflow = TRUE; break; }
1095    #endif
1096            }
1097    
1098          if (overflow)
1099            {
1100            while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1101            *errorcodeptr = ERR34;
1102          }          }
1103    
1104        if (*pt == CHAR_RIGHT_CURLY_BRACKET)        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
1105          {          {
1106          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1107          ptr = pt;          ptr = pt;
1108          break;          break;
1109          }          }
# Line 917  else Line 1115  else
1115      /* Read just a single-byte hex-defined char */      /* Read just a single-byte hex-defined char */
1116    
1117      c = 0;      c = 0;
1118      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1119        {        {
1120        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1121        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1122  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1123        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 938  else Line 1136  else
1136    
1137      case CHAR_c:      case CHAR_c:
1138      c = *(++ptr);      c = *(++ptr);
1139      if (c == 0)      if (c == CHAR_NULL)
1140        {        {
1141        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
1142        break;        break;
# Line 978  else Line 1176  else
1176  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1177  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1178    
1179  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1180       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1181    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1182    
1183  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1184    
1185  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1186    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1187    
1188  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1189    
1190  *ptrptr = ptr;  *ptrptr = ptr;
1191  return c;  *chptr = c;
1192    return escape;
1193  }  }
1194    
   
   
1195  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1196  /*************************************************  /*************************************************
1197  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1008  escape sequence. Line 1205  escape sequence.
1205  Argument:  Argument:
1206    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1207    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
1208    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
1209      pdataptr       points to an unsigned int that is set to the detailed property value
1210    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1211    
1212  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
1213  */  */
1214    
1215  static int  static BOOL
1216  get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1217      unsigned int *pdataptr, int *errorcodeptr)
1218  {  {
1219  int c, i, bot, top;  pcre_uchar c;
1220  const uschar *ptr = *ptrptr;  int i, bot, top;
1221  char name[32];  const pcre_uchar *ptr = *ptrptr;
1222    pcre_uchar name[32];
1223    
1224  c = *(++ptr);  c = *(++ptr);
1225  if (c == 0) goto ERROR_RETURN;  if (c == CHAR_NULL) goto ERROR_RETURN;
1226    
1227  *negptr = FALSE;  *negptr = FALSE;
1228    
# Line 1036  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1236  if (c == CHAR_LEFT_CURLY_BRACKET)
1236      *negptr = TRUE;      *negptr = TRUE;
1237      ptr++;      ptr++;
1238      }      }
1239    for (i = 0; i < (int)sizeof(name) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1240      {      {
1241      c = *(++ptr);      c = *(++ptr);
1242      if (c == 0) goto ERROR_RETURN;      if (c == CHAR_NULL) goto ERROR_RETURN;
1243      if (c == CHAR_RIGHT_CURLY_BRACKET) break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
1244      name[i] = c;      name[i] = c;
1245      }      }
# Line 1060  else Line 1260  else
1260  /* Search for a recognized property name using binary chop */  /* Search for a recognized property name using binary chop */
1261    
1262  bot = 0;  bot = 0;
1263  top = _pcre_utt_size;  top = PRIV(utt_size);
1264    
1265  while (bot < top)  while (bot < top)
1266    {    {
1267      int r;
1268    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1269    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1270    if (c == 0)    if (r == 0)
1271      {      {
1272      *dptr = _pcre_utt[i].value;      *ptypeptr = PRIV(utt)[i].type;
1273      return _pcre_utt[i].type;      *pdataptr = PRIV(utt)[i].value;
1274        return TRUE;
1275      }      }
1276    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1277    }    }
1278    
1279  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1280  *ptrptr = ptr;  *ptrptr = ptr;
1281  return -1;  return FALSE;
1282    
1283  ERROR_RETURN:  ERROR_RETURN:
1284  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1285  *ptrptr = ptr;  *ptrptr = ptr;
1286  return -1;  return FALSE;
1287  }  }
1288  #endif  #endif
1289    
# Line 1107  Returns:         pointer to '}' on succe Line 1309  Returns:         pointer to '}' on succe
1309                   current ptr on error, with errorcodeptr set non-zero                   current ptr on error, with errorcodeptr set non-zero
1310  */  */
1311    
1312  static const uschar *  static const pcre_uchar *
1313  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)
1314  {  {
1315  int min = 0;  int min = 0;
1316  int max = -1;  int max = -1;
# Line 1116  int max = -1; Line 1318  int max = -1;
1318  /* 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
1319  an integer overflow. */  an integer overflow. */
1320    
1321  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1322  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1323    {    {
1324    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1131  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1333  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1333    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1334      {      {
1335      max = 0;      max = 0;
1336      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1337      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1338        {        {
1339        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1186  Arguments: Line 1388  Arguments:
1388    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1389    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1390    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1391    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1392    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1393    
1394  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1395  */  */
1396    
1397  static int  static int
1398  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,
1399    BOOL xmode, BOOL utf8, int *count)    BOOL xmode, BOOL utf, int *count)
1400  {  {
1401  uschar *ptr = *ptrptr;  pcre_uchar *ptr = *ptrptr;
1402  int start_count = *count;  int start_count = *count;
1403  int hwm_count = start_count;  int hwm_count = start_count;
1404  BOOL dup_parens = FALSE;  BOOL dup_parens = FALSE;
# Line 1208  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1410  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1410    {    {
1411    /* Handle specials such as (*SKIP) or (*UTF8) etc. */    /* Handle specials such as (*SKIP) or (*UTF8) etc. */
1412    
1413    if (ptr[1] == CHAR_ASTERISK) ptr += 2;    if (ptr[1] == CHAR_ASTERISK)
1414        {
1415        ptr += 2;
1416        while (ptr < cd->end_pattern && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1417        }
1418    
1419    /* Handle a normal, unnamed capturing parenthesis. */    /* Handle a normal, unnamed capturing parenthesis. */
1420    
# Line 1232  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1438  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1438    
1439    else if (ptr[2] == CHAR_NUMBER_SIGN)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1440      {      {
1441      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;      for (ptr += 3; *ptr != CHAR_NULL; ptr++)
1442          if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1443      goto FAIL_EXIT;      goto FAIL_EXIT;
1444      }      }
1445    
# Line 1245  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1452  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1452      ptr += 2;      ptr += 2;
1453      if (ptr[1] != CHAR_QUESTION_MARK)      if (ptr[1] != CHAR_QUESTION_MARK)
1454        {        {
1455        while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;        while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1456        if (*ptr != 0) ptr++;        if (*ptr != CHAR_NULL) ptr++;
1457        }        }
1458      }      }
1459    
# Line 1262  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1469  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1469      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1470          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1471        {        {
1472        int term;        pcre_uchar term;
1473        const uschar *thisname;        const pcre_uchar *thisname;
1474        *count += 1;        *count += 1;
1475        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1476        term = *ptr++;        term = *ptr++;
1477        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1478        thisname = ptr;        thisname = ptr;
1479        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1480        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1481            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1482          return *count;          return *count;
1483        term++;        term++;
1484        }        }
# Line 1289  for (; ptr < cd->end_pattern; ptr++) Line 1496  for (; ptr < cd->end_pattern; ptr++)
1496    
1497    if (*ptr == CHAR_BACKSLASH)    if (*ptr == CHAR_BACKSLASH)
1498      {      {
1499      if (*(++ptr) == 0) goto FAIL_EXIT;      if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1500      if (*ptr == CHAR_Q) for (;;)      if (*ptr == CHAR_Q) for (;;)
1501        {        {
1502        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};        while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1503        if (*ptr == 0) goto FAIL_EXIT;        if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1504        if (*(++ptr) == CHAR_E) break;        if (*(++ptr) == CHAR_E) break;
1505        }        }
1506      continue;      continue;
# Line 1314  for (; ptr < cd->end_pattern; ptr++) Line 1521  for (; ptr < cd->end_pattern; ptr++)
1521          {          {
1522          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1523            ptr+= 2;            ptr+= 2;
1524          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1525                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1526            ptr += 4;            ptr += 4;
1527          else          else
# Line 1337  for (; ptr < cd->end_pattern; ptr++) Line 1544  for (; ptr < cd->end_pattern; ptr++)
1544    
1545      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1546        {        {
1547        if (*ptr == 0) return -1;        if (*ptr == CHAR_NULL) return -1;
1548        if (*ptr == CHAR_BACKSLASH)        if (*ptr == CHAR_BACKSLASH)
1549          {          {
1550          if (*(++ptr) == 0) goto FAIL_EXIT;          if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1551          if (*ptr == CHAR_Q) for (;;)          if (*ptr == CHAR_Q) for (;;)
1552            {            {
1553            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};            while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1554            if (*ptr == 0) goto FAIL_EXIT;            if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1555            if (*(++ptr) == CHAR_E) break;            if (*(++ptr) == CHAR_E) break;
1556            }            }
1557          continue;          continue;
# Line 1358  for (; ptr < cd->end_pattern; ptr++) Line 1565  for (; ptr < cd->end_pattern; ptr++)
1565    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1566      {      {
1567      ptr++;      ptr++;
1568      while (*ptr != 0)      while (*ptr != CHAR_NULL)
1569        {        {
1570        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1571        ptr++;        ptr++;
1572  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1573        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1574  #endif  #endif
1575        }        }
1576      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1577      continue;      continue;
1578      }      }
1579    
# Line 1374  for (; ptr < cd->end_pattern; ptr++) Line 1581  for (; ptr < cd->end_pattern; ptr++)
1581    
1582    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1583      {      {
1584      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1585      if (rc > 0) return rc;      if (rc > 0) return rc;
1586      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1587      }      }
1588    
1589    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
# Line 1420  Arguments: Line 1627  Arguments:
1627    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1628    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1629    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1630    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1631    
1632  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1633  */  */
1634    
1635  static int  static int
1636  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,
1637    BOOL utf8)    BOOL utf)
1638  {  {
1639  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1640  int count = 0;  int count = 0;
1641  int rc;  int rc;
1642    
# Line 1440  matching closing parens. That is why we Line 1647  matching closing parens. That is why we
1647    
1648  for (;;)  for (;;)
1649    {    {
1650    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1651    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == CHAR_NULL) break;
1652    }    }
1653    
1654  return rc;  return rc;
# Line 1467  Arguments: Line 1674  Arguments:
1674  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1675  */  */
1676    
1677  static const uschar*  static const pcre_uchar*
1678  first_significant_code(const uschar *code, BOOL skipassert)  first_significant_code(const pcre_uchar *code, BOOL skipassert)
1679  {  {
1680  for (;;)  for (;;)
1681    {    {
# Line 1479  for (;;) Line 1686  for (;;)
1686      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1687      if (!skipassert) return code;      if (!skipassert) return code;
1688      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1689      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1690      break;      break;
1691    
1692      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1493  for (;;) Line 1700  for (;;)
1700      case OP_RREF:      case OP_RREF:
1701      case OP_NRREF:      case OP_NRREF:
1702      case OP_DEF:      case OP_DEF:
1703      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1704      break;      break;
1705    
1706      default:      default:
# Line 1523  and doing the check at the end; a flag s Line 1730  and doing the check at the end; a flag s
1730    
1731  Arguments:  Arguments:
1732    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1733    utf8     TRUE in UTF-8 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1734    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1735    cd       the "compile data" structure    cd       the "compile data" structure
1736    
# Line 1535  Returns:   the fixed length, Line 1742  Returns:   the fixed length,
1742  */  */
1743    
1744  static int  static int
1745  find_fixedlength(uschar *code, BOOL utf8, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1746  {  {
1747  int length = -1;  int length = -1;
1748    
1749  register int branchlength = 0;  register int branchlength = 0;
1750  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1751    
1752  /* 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
1753  branch, check the length against that of the other branches. */  branch, check the length against that of the other branches. */
# Line 1548  branch, check the length against that of Line 1755  branch, check the length against that of
1755  for (;;)  for (;;)
1756    {    {
1757    int d;    int d;
1758    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1759    register int op = *cc;    register pcre_uchar op = *cc;
1760    
1761    switch (op)    switch (op)
1762      {      {
1763      /* 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 1562  for (;;) Line 1770  for (;;)
1770      case OP_ONCE:      case OP_ONCE:
1771      case OP_ONCE_NC:      case OP_ONCE_NC:
1772      case OP_COND:      case OP_COND:
1773      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);
1774      if (d < 0) return d;      if (d < 0) return d;
1775      branchlength += d;      branchlength += d;
1776      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
# Line 1593  for (;;) Line 1801  for (;;)
1801    
1802      case OP_RECURSE:      case OP_RECURSE:
1803      if (!atend) return -3;      if (!atend) return -3;
1804      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1805      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1806      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1807      d = find_fixedlength(cs + 2, utf8, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1808      if (d < 0) return d;      if (d < 0) return d;
1809      branchlength += d;      branchlength += d;
1810      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1609  for (;;) Line 1817  for (;;)
1817      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1818      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1819      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1820      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1821        break;
1822    
1823      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1824    
# Line 1617  for (;;) Line 1826  for (;;)
1826      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
1827      case OP_SKIP_ARG:      case OP_SKIP_ARG:
1828      case OP_THEN_ARG:      case OP_THEN_ARG:
1829      cc += cc[1] + _pcre_OP_lengths[*cc];      cc += cc[1] + PRIV(OP_lengths)[*cc];
1830      break;      break;
1831    
1832      case OP_CALLOUT:      case OP_CALLOUT:
# Line 1644  for (;;) Line 1853  for (;;)
1853      case OP_SOM:      case OP_SOM:
1854      case OP_THEN:      case OP_THEN:
1855      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1856      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1857      break;      break;
1858    
1859      /* Handle literal characters */      /* Handle literal characters */
# Line 1655  for (;;) Line 1864  for (;;)
1864      case OP_NOTI:      case OP_NOTI:
1865      branchlength++;      branchlength++;
1866      cc += 2;      cc += 2;
1867  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1868      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1869  #endif  #endif
1870      break;      break;
1871    
# Line 1667  for (;;) Line 1876  for (;;)
1876      case OP_EXACTI:      case OP_EXACTI:
1877      case OP_NOTEXACT:      case OP_NOTEXACT:
1878      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1879      branchlength += GET2(cc,1);      branchlength += (int)GET2(cc,1);
1880      cc += 4;      cc += 2 + IMM2_SIZE;
1881  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1882      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1883  #endif  #endif
1884      break;      break;
1885    
1886      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1887      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1888      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1889      cc += 4;        cc += 2;
1890        cc += 1 + IMM2_SIZE + 1;
1891      break;      break;
1892    
1893      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1703  for (;;) Line 1913  for (;;)
1913      cc++;      cc++;
1914      break;      break;
1915    
1916      /* 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;
1917      otherwise \C is coded as OP_ALLANY. */      otherwise \C is coded as OP_ALLANY. */
1918    
1919      case OP_ANYBYTE:      case OP_ANYBYTE:
# Line 1711  for (;;) Line 1921  for (;;)
1921    
1922      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1923    
 #ifdef SUPPORT_UTF8  
     case OP_XCLASS:  
     cc += GET(cc, 1) - 33;  
     /* Fall through */  
 #endif  
   
1924      case OP_CLASS:      case OP_CLASS:
1925      case OP_NCLASS:      case OP_NCLASS:
1926      cc += 33;  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1927        case OP_XCLASS:
1928        /* The original code caused an unsigned overflow in 64 bit systems,
1929        so now we use a conditional statement. */
1930        if (op == OP_XCLASS)
1931          cc += GET(cc, 1);
1932        else
1933          cc += PRIV(OP_lengths)[OP_CLASS];
1934    #else
1935        cc += PRIV(OP_lengths)[OP_CLASS];
1936    #endif
1937    
1938      switch (*cc)      switch (*cc)
1939        {        {
# Line 1733  for (;;) Line 1947  for (;;)
1947    
1948        case OP_CRRANGE:        case OP_CRRANGE:
1949        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1950        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1951        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1952        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1953        break;        break;
1954    
1955        default:        default:
# Line 1850  length. Line 2064  length.
2064    
2065  Arguments:  Arguments:
2066    code        points to start of expression    code        points to start of expression
2067    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2068    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2069    
2070  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
2071  */  */
2072    
2073  const uschar *  const pcre_uchar *
2074  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2075  {  {
2076  for (;;)  for (;;)
2077    {    {
2078    register int c = *code;    register pcre_uchar c = *code;
2079    
2080    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2081    
# Line 1875  for (;;) Line 2089  for (;;)
2089    
2090    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2091      {      {
2092      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2093      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2094      }      }
2095    
2096    /* Handle capturing bracket */    /* Handle capturing bracket */
# Line 1884  for (;;) Line 2098  for (;;)
2098    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2099             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2100      {      {
2101      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2102      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2103      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2104      }      }
2105    
2106    /* 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 1914  for (;;) Line 2128  for (;;)
2128        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2129        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2130        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2131        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2132            code += 2;
2133        break;        break;
2134    
2135        case OP_MARK:        case OP_MARK:
2136        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2137        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2138        case OP_THEN_ARG:        case OP_THEN_ARG:
2139        code += code[1];        code += code[1];
2140        break;        break;
# Line 1930  for (;;) Line 2142  for (;;)
2142    
2143      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2144    
2145      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2146    
2147    /* 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
2148    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
2149    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2150    
2151  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2152      if (utf8) switch(c)      if (utf) switch(c)
2153        {        {
2154        case OP_CHAR:        case OP_CHAR:
2155        case OP_CHARI:        case OP_CHARI:
# Line 1967  for (;;) Line 2179  for (;;)
2179        case OP_MINQUERYI:        case OP_MINQUERYI:
2180        case OP_POSQUERY:        case OP_POSQUERY:
2181        case OP_POSQUERYI:        case OP_POSQUERYI:
2182        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2183        break;        break;
2184        }        }
2185  #else  #else
2186      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2187  #endif  #endif
2188      }      }
2189    }    }
# Line 1988  instance of OP_RECURSE. Line 2200  instance of OP_RECURSE.
2200    
2201  Arguments:  Arguments:
2202    code        points to start of expression    code        points to start of expression
2203    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2204    
2205  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
2206  */  */
2207    
2208  static const uschar *  static const pcre_uchar *
2209  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2210  {  {
2211  for (;;)  for (;;)
2212    {    {
2213    register int c = *code;    register pcre_uchar c = *code;
2214    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2215    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2216    
# Line 2033  for (;;) Line 2245  for (;;)
2245        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2246        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2247        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2248        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2249            code += 2;
2250        break;        break;
2251    
2252        case OP_MARK:        case OP_MARK:
2253        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2254        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2255        case OP_THEN_ARG:        case OP_THEN_ARG:
2256        code += code[1];        code += code[1];
2257        break;        break;
# Line 2049  for (;;) Line 2259  for (;;)
2259    
2260      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2261    
2262      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2263    
2264      /* 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
2265      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
2266      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2267    
2268  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2269      if (utf8) switch(c)      if (utf) switch(c)
2270        {        {
2271        case OP_CHAR:        case OP_CHAR:
2272        case OP_CHARI:        case OP_CHARI:
2273          case OP_NOT:
2274          case OP_NOTI:
2275        case OP_EXACT:        case OP_EXACT:
2276        case OP_EXACTI:        case OP_EXACTI:
2277          case OP_NOTEXACT:
2278          case OP_NOTEXACTI:
2279        case OP_UPTO:        case OP_UPTO:
2280        case OP_UPTOI:        case OP_UPTOI:
2281          case OP_NOTUPTO:
2282          case OP_NOTUPTOI:
2283        case OP_MINUPTO:        case OP_MINUPTO:
2284        case OP_MINUPTOI:        case OP_MINUPTOI:
2285          case OP_NOTMINUPTO:
2286          case OP_NOTMINUPTOI:
2287        case OP_POSUPTO:        case OP_POSUPTO:
2288        case OP_POSUPTOI:        case OP_POSUPTOI:
2289          case OP_NOTPOSUPTO:
2290          case OP_NOTPOSUPTOI:
2291        case OP_STAR:        case OP_STAR:
2292        case OP_STARI:        case OP_STARI:
2293          case OP_NOTSTAR:
2294          case OP_NOTSTARI:
2295        case OP_MINSTAR:        case OP_MINSTAR:
2296        case OP_MINSTARI:        case OP_MINSTARI:
2297          case OP_NOTMINSTAR:
2298          case OP_NOTMINSTARI:
2299        case OP_POSSTAR:        case OP_POSSTAR:
2300        case OP_POSSTARI:        case OP_POSSTARI:
2301          case OP_NOTPOSSTAR:
2302          case OP_NOTPOSSTARI:
2303        case OP_PLUS:        case OP_PLUS:
2304        case OP_PLUSI:        case OP_PLUSI:
2305          case OP_NOTPLUS:
2306          case OP_NOTPLUSI:
2307        case OP_MINPLUS:        case OP_MINPLUS:
2308        case OP_MINPLUSI:        case OP_MINPLUSI:
2309          case OP_NOTMINPLUS:
2310          case OP_NOTMINPLUSI:
2311        case OP_POSPLUS:        case OP_POSPLUS:
2312        case OP_POSPLUSI:        case OP_POSPLUSI:
2313          case OP_NOTPOSPLUS:
2314          case OP_NOTPOSPLUSI:
2315        case OP_QUERY:        case OP_QUERY:
2316        case OP_QUERYI:        case OP_QUERYI:
2317          case OP_NOTQUERY:
2318          case OP_NOTQUERYI:
2319        case OP_MINQUERY:        case OP_MINQUERY:
2320        case OP_MINQUERYI:        case OP_MINQUERYI:
2321          case OP_NOTMINQUERY:
2322          case OP_NOTMINQUERYI:
2323        case OP_POSQUERY:        case OP_POSQUERY:
2324        case OP_POSQUERYI:        case OP_POSQUERYI:
2325        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_NOTPOSQUERY:
2326          case OP_NOTPOSQUERYI:
2327          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2328        break;        break;
2329        }        }
2330  #else  #else
2331      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2332  #endif  #endif
2333      }      }
2334    }    }
# Line 2113  bracket whose current branch will alread Line 2351  bracket whose current branch will alread
2351  Arguments:  Arguments:
2352    code        points to start of search    code        points to start of search
2353    endcode     points to where to stop    endcode     points to where to stop
2354    utf8        TRUE if in UTF8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2355    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2356      recurses    chain of recurse_check to catch mutual recursion
2357    
2358  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2359  */  */
2360    
2361    typedef struct recurse_check {
2362      struct recurse_check *prev;
2363      const pcre_uchar *group;
2364    } recurse_check;
2365    
2366  static BOOL  static BOOL
2367  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2368    compile_data *cd)    BOOL utf, compile_data *cd, recurse_check *recurses)
2369  {  {
2370  register int c;  register pcre_uchar c;
2371  for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE);  recurse_check this_recurse;
2372    
2373    for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2374       code < endcode;       code < endcode;
2375       code = first_significant_code(code + _pcre_OP_lengths[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2376    {    {
2377    const uschar *ccode;    const pcre_uchar *ccode;
2378    
2379    c = *code;    c = *code;
2380    
2381    /* Skip over forward assertions; the other assertions are skipped by    /* Skip over forward assertions; the other assertions are skipped by
2382    first_significant_code() with a TRUE final argument. */    first_significant_code() with a TRUE final argument. */
2383    
# Line 2151  for (code = first_significant_code(code Line 2397  for (code = first_significant_code(code
2397    
2398    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2399      {      {
2400      const uschar *scode;      const pcre_uchar *scode = cd->start_code + GET(code, 1);
2401      BOOL empty_branch;      BOOL empty_branch;
2402    
2403      /* Test for forward reference */      /* Test for forward reference or uncompleted reference. This is disabled
2404        when called to scan a completed pattern by setting cd->start_workspace to
2405      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      NULL. */
2406        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;  
2407        if (cd->start_workspace != NULL)
2408          {
2409          const pcre_uchar *tcode;
2410          for (tcode = cd->start_workspace; tcode < cd->hwm; tcode += LINK_SIZE)
2411            if ((int)GET(tcode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2412          if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2413          }
2414    
2415        /* If we are scanning a completed pattern, there are no forward references
2416        and all groups are complete. We need to detect whether this is a recursive
2417        call, as otherwise there will be an infinite loop. If it is a recursion,
2418        just skip over it. Simple recursions are easily detected. For mutual
2419        recursions we keep a chain on the stack. */
2420    
2421        else
2422          {
2423          recurse_check *r = recurses;
2424          const pcre_uchar *endgroup = scode;
2425    
2426          do endgroup += GET(endgroup, 1); while (*endgroup == OP_ALT);
2427          if (code >= scode && code <= endgroup) continue;  /* Simple recursion */
2428    
2429          for (r = recurses; r != NULL; r = r->prev)
2430            if (r->group == scode) break;
2431          if (r != NULL) continue;   /* Mutual recursion */
2432          }
2433    
2434      /* Not a forward reference, test for completed backward reference */      /* Completed reference; scan the referenced group, remembering it on the
2435        stack chain to detect mutual recursions. */
2436    
2437      empty_branch = FALSE;      empty_branch = FALSE;
2438      scode = cd->start_code + GET(code, 1);      this_recurse.prev = recurses;
2439      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */      this_recurse.group = scode;
2440    
     /* Completed backwards reference */  
   
2441      do      do
2442        {        {
2443        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd, &this_recurse))
2444          {          {
2445          empty_branch = TRUE;          empty_branch = TRUE;
2446          break;          break;
# Line 2187  for (code = first_significant_code(code Line 2458  for (code = first_significant_code(code
2458    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2459        c == OP_BRAPOSZERO)        c == OP_BRAPOSZERO)
2460      {      {
2461      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2462      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
2463      c = *code;      c = *code;
2464      continue;      continue;
# Line 2225  for (code = first_significant_code(code Line 2496  for (code = first_significant_code(code
2496        empty_branch = FALSE;        empty_branch = FALSE;
2497        do        do
2498          {          {
2499          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd, NULL))
2500            empty_branch = TRUE;            empty_branch = TRUE;
2501          code += GET(code, 1);          code += GET(code, 1);
2502          }          }
# Line 2243  for (code = first_significant_code(code Line 2514  for (code = first_significant_code(code
2514      {      {
2515      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2516      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2517      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2518      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"
2519      here. */      here. */
2520    
2521  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2522      case OP_XCLASS:      case OP_XCLASS:
2523      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2524      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2255  for (code = first_significant_code(code Line 2526  for (code = first_significant_code(code
2526    
2527      case OP_CLASS:      case OP_CLASS:
2528      case OP_NCLASS:      case OP_NCLASS:
2529      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2530    
2531  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2532      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2533  #endif  #endif
2534    
# Line 2283  for (code = first_significant_code(code Line 2554  for (code = first_significant_code(code
2554    
2555      /* Opcodes that must match a character */      /* Opcodes that must match a character */
2556    
2557        case OP_ANY:
2558        case OP_ALLANY:
2559        case OP_ANYBYTE:
2560    
2561      case OP_PROP:      case OP_PROP:
2562      case OP_NOTPROP:      case OP_NOTPROP:
2563        case OP_ANYNL:
2564    
2565        case OP_NOT_HSPACE:
2566        case OP_HSPACE:
2567        case OP_NOT_VSPACE:
2568        case OP_VSPACE:
2569      case OP_EXTUNI:      case OP_EXTUNI:
2570    
2571      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
2572      case OP_DIGIT:      case OP_DIGIT:
2573      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
2574      case OP_WHITESPACE:      case OP_WHITESPACE:
2575      case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
2576      case OP_WORDCHAR:      case OP_WORDCHAR:
2577      case OP_ANY:  
     case OP_ALLANY:  
     case OP_ANYBYTE:  
2578      case OP_CHAR:      case OP_CHAR:
2579      case OP_CHARI:      case OP_CHARI:
2580      case OP_NOT:      case OP_NOT:
2581      case OP_NOTI:      case OP_NOTI:
2582    
2583      case OP_PLUS:      case OP_PLUS:
2584        case OP_PLUSI:
2585      case OP_MINPLUS:      case OP_MINPLUS:
2586      case OP_POSPLUS:      case OP_MINPLUSI:
2587      case OP_EXACT:  
2588      case OP_NOTPLUS:      case OP_NOTPLUS:
2589        case OP_NOTPLUSI:
2590      case OP_NOTMINPLUS:      case OP_NOTMINPLUS:
2591        case OP_NOTMINPLUSI:
2592    
2593        case OP_POSPLUS:
2594        case OP_POSPLUSI:
2595      case OP_NOTPOSPLUS:      case OP_NOTPOSPLUS:
2596        case OP_NOTPOSPLUSI:
2597    
2598        case OP_EXACT:
2599        case OP_EXACTI:
2600      case OP_NOTEXACT:      case OP_NOTEXACT:
2601        case OP_NOTEXACTI:
2602    
2603      case OP_TYPEPLUS:      case OP_TYPEPLUS:
2604      case OP_TYPEMINPLUS:      case OP_TYPEMINPLUS:
2605      case OP_TYPEPOSPLUS:      case OP_TYPEPOSPLUS:
2606      case OP_TYPEEXACT:      case OP_TYPEEXACT:
2607    
2608      return FALSE;      return FALSE;
2609    
2610      /* These are going to continue, as they may be empty, but we have to      /* These are going to continue, as they may be empty, but we have to
# Line 2330  for (code = first_significant_code(code Line 2624  for (code = first_significant_code(code
2624      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2625      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2626      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2627      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2628          code += 2;
2629      break;      break;
2630    
2631      /* End of branch */      /* End of branch */
# Line 2343  for (code = first_significant_code(code Line 2638  for (code = first_significant_code(code
2638      return TRUE;      return TRUE;
2639    
2640      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2641      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO and their caseless and negative versions may be
2642        followed by a multibyte character. */
2643    
2644  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2645      case OP_STAR:      case OP_STAR:
2646      case OP_STARI:      case OP_STARI:
2647        case OP_NOTSTAR:
2648        case OP_NOTSTARI:
2649    
2650      case OP_MINSTAR:      case OP_MINSTAR:
2651      case OP_MINSTARI:      case OP_MINSTARI:
2652        case OP_NOTMINSTAR:
2653        case OP_NOTMINSTARI:
2654    
2655      case OP_POSSTAR:      case OP_POSSTAR:
2656      case OP_POSSTARI:      case OP_POSSTARI:
2657        case OP_NOTPOSSTAR:
2658        case OP_NOTPOSSTARI:
2659    
2660      case OP_QUERY:      case OP_QUERY:
2661      case OP_QUERYI:      case OP_QUERYI:
2662        case OP_NOTQUERY:
2663        case OP_NOTQUERYI:
2664    
2665      case OP_MINQUERY:      case OP_MINQUERY:
2666      case OP_MINQUERYI:      case OP_MINQUERYI:
2667        case OP_NOTMINQUERY:
2668        case OP_NOTMINQUERYI:
2669    
2670      case OP_POSQUERY:      case OP_POSQUERY:
2671      case OP_POSQUERYI:      case OP_POSQUERYI:
2672      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      case OP_NOTPOSQUERY:
2673        case OP_NOTPOSQUERYI:
2674    
2675        if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2676      break;      break;
2677    
2678      case OP_UPTO:      case OP_UPTO:
2679      case OP_UPTOI:      case OP_UPTOI:
2680        case OP_NOTUPTO:
2681        case OP_NOTUPTOI:
2682    
2683      case OP_MINUPTO:      case OP_MINUPTO:
2684      case OP_MINUPTOI:      case OP_MINUPTOI:
2685        case OP_NOTMINUPTO:
2686        case OP_NOTMINUPTOI:
2687    
2688      case OP_POSUPTO:      case OP_POSUPTO:
2689      case OP_POSUPTOI:      case OP_POSUPTOI:
2690      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      case OP_NOTPOSUPTO:
2691        case OP_NOTPOSUPTOI:
2692    
2693        if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);
2694      break;      break;
2695  #endif  #endif
2696    
# Line 2377  for (code = first_significant_code(code Line 2700  for (code = first_significant_code(code
2700      case OP_MARK:      case OP_MARK:
2701      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
2702      case OP_SKIP_ARG:      case OP_SKIP_ARG:
     code += code[1];  
     break;  
   
2703      case OP_THEN_ARG:      case OP_THEN_ARG:
2704      code += code[1];      code += code[1];
2705      break;      break;
# Line 2411  Arguments: Line 2731  Arguments:
2731    code        points to start of the recursion    code        points to start of the recursion
2732    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2733    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2734    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2735    cd          pointers to tables etc    cd          pointers to tables etc
2736    
2737  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2738  */  */
2739    
2740  static BOOL  static BOOL
2741  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2742    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2743  {  {
2744  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2745    {    {
2746    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd, NULL))
2747      return FALSE;      return FALSE;
2748    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2749    }    }
# Line 2475  Returns:   TRUE or FALSE Line 2795  Returns:   TRUE or FALSE
2795  */  */
2796    
2797  static BOOL  static BOOL
2798  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2799  {  {
2800  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2801  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2802  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != CHAR_NULL; ptr++)
2803    {    {
2804    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2805      ptr++;      ptr++;
# Line 2519  Returns:     a value representing the na Line 2839  Returns:     a value representing the na
2839  */  */
2840    
2841  static int  static int
2842  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2843  {  {
2844  const char *pn = posix_names;  const char *pn = posix_names;
2845  register int yield = 0;  register int yield = 0;
2846  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2847    {    {
2848    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2849      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2850    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2851    yield++;    yield++;
2852    }    }
# Line 2558  value in the reference (which is a group Line 2878  value in the reference (which is a group
2878  Arguments:  Arguments:
2879    group      points to the start of the group    group      points to the start of the group
2880    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2881    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2882    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2883    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
2884    
# Line 2566  Returns:     nothing Line 2886  Returns:     nothing
2886  */  */
2887    
2888  static void  static void
2889  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2890    uschar *save_hwm)    pcre_uchar *save_hwm)
2891  {  {
2892  uschar *ptr = group;  pcre_uchar *ptr = group;
2893    
2894  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2895    {    {
2896    int offset;    int offset;
2897    uschar *hc;    pcre_uchar *hc;
2898    
2899    /* 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
2900    reference. */    reference. */
2901    
2902    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2903      {      {
2904      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2905      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2906        {        {
2907        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2594  while ((ptr = (uschar *)find_recurse(ptr Line 2914  while ((ptr = (uschar *)find_recurse(ptr
2914    
2915    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2916      {      {
2917      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2918      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2919      }      }
2920    
# Line 2619  Arguments: Line 2939  Arguments:
2939  Returns:         new code pointer  Returns:         new code pointer
2940  */  */
2941    
2942  static uschar *  static pcre_uchar *
2943  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2944  {  {
2945  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2946  *code++ = 255;  *code++ = 255;
2947  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2948  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2949  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2950  }  }
2951    
2952    
# Line 2648  Returns:             nothing Line 2968  Returns:             nothing
2968  */  */
2969    
2970  static void  static void
2971  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2972  {  {
2973  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2974  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2662  PUT(previous_callout, 2 + LINK_SIZE, len Line 2982  PUT(previous_callout, 2 + LINK_SIZE, len
2982  *************************************************/  *************************************************/
2983    
2984  /* 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
2985  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
2986  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
2987  start address.  start address. A character with multiple other cases is returned on its own
2988    with a special return value.
2989    
2990  Arguments:  Arguments:
2991    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2672  Arguments: Line 2993  Arguments:
2993    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2994    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2995    
2996  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2997                   0 when a range is returned
2998                  >0 the CASESET offset for char with multiple other cases
2999                    in this case, ocptr contains the original
3000  */  */
3001    
3002  static BOOL  static int
3003  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
3004    unsigned int *odptr)    pcre_uint32 *odptr)
3005  {  {
3006  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
3007    unsigned int co;
3008    
3009    /* Find the first character that has an other case. If it has multiple other
3010    cases, return its case offset value. */
3011    
3012  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
3013    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
3014      if ((co = UCD_CASESET(c)) != 0)
3015        {
3016        *ocptr = c++;   /* Character that has the set */
3017        *cptr = c;      /* Rest of input range */
3018        return (int)co;
3019        }
3020      if ((othercase = UCD_OTHERCASE(c)) != c) break;
3021      }
3022    
3023  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
3024    
3025  *ocptr = othercase;  *ocptr = othercase;
3026  next = othercase + 1;  next = othercase + 1;
# Line 2695  for (++c; c <= d; c++) Line 3031  for (++c; c <= d; c++)
3031    next++;    next++;
3032    }    }
3033    
3034  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
3035  *cptr = c;  *cptr = c;             /* Rest of input range */
3036    return 0;
 return TRUE;  
3037  }  }
3038    
3039    
# Line 2720  Returns:       TRUE if auto-possessifyin Line 3055  Returns:       TRUE if auto-possessifyin
3055  */  */
3056    
3057  static BOOL  static BOOL
3058  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)
3059  {  {
3060    #ifdef SUPPORT_UCP
3061    const pcre_uint32 *p;
3062    #endif
3063    
3064  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
3065    
3066  switch(ptype)  switch(ptype)
3067    {    {
3068    case PT_LAMP:    case PT_LAMP:
# Line 2731  switch(ptype) Line 3071  switch(ptype)
3071            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
3072    
3073    case PT_GC:    case PT_GC:
3074    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
3075    
3076    case PT_PC:    case PT_PC:
3077    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2742  switch(ptype) Line 3082  switch(ptype)
3082    /* These are specials */    /* These are specials */
3083    
3084    case PT_ALNUM:    case PT_ALNUM:
3085    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3086            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
3087    
3088    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
3089    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3090            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
3091            == negated;            == negated;
3092    
3093    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
3094    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3095            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
3096            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
3097            == negated;            == negated;
3098    
3099    case PT_WORD:    case PT_WORD:
3100    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3101            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3102            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3103    
3104    #ifdef SUPPORT_UCP
3105      case PT_CLIST:
3106      p = PRIV(ucd_caseless_sets) + prop->caseset;
3107      for (;;)
3108        {
3109        if (c < *p) return !negated;
3110        if (c == *p++) return negated;
3111        }
3112      break;  /* Control never reaches here */
3113    #endif
3114    }    }
3115    
3116  return FALSE;  return FALSE;
3117  }  }
3118  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2777  sense to automatically possessify the re Line 3129  sense to automatically possessify the re
3129    
3130  Arguments:  Arguments:
3131    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3132    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3133    ptr           next character in pattern    ptr           next character in pattern
3134    options       options bits    options       options bits
3135    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2786  Returns:        TRUE if possessifying is Line 3138  Returns:        TRUE if possessifying is
3138  */  */
3139    
3140  static BOOL  static BOOL
3141  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3142    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3143  {  {
3144  int c, next;  pcre_uint32 c = NOTACHAR;
3145  int op_code = *previous++;  pcre_uint32 next;
3146    int escape;
3147    pcre_uchar op_code = *previous++;
3148    
3149  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3150    
# Line 2798  if ((options & PCRE_EXTENDED) != 0) Line 3152  if ((options & PCRE_EXTENDED) != 0)
3152    {    {
3153    for (;;)    for (;;)
3154      {      {
3155      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3156      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3157        {        {
3158        ptr++;        ptr++;
3159        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3160          {          {
3161          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3162          ptr++;          ptr++;
3163  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3164          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3165  #endif  #endif
3166          }          }
3167        }        }
# Line 2821  value is a character, a negative value i Line 3175  value is a character, a negative value i
3175  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3176    {    {
3177    int temperrorcode = 0;    int temperrorcode = 0;
3178    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options,
3179        FALSE);
3180    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3181    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3182    }    }
3183    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
3184    {    {
3185  #ifdef SUPPORT_UTF8    escape = 0;
3186    if (utf8) { GETCHARINC(next, ptr); } else  #ifdef SUPPORT_UTF
3187      if (utf) { GETCHARINC(next, ptr); } else
3188  #endif  #endif
3189    next = *ptr++;    next = *ptr++;
3190    }    }
   
3191  else return FALSE;  else return FALSE;
3192    
3193  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2842  if ((options & PCRE_EXTENDED) != 0) Line 3196  if ((options & PCRE_EXTENDED) != 0)
3196    {    {
3197    for (;;)    for (;;)
3198      {      {
3199      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3200      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3201        {        {
3202        ptr++;        ptr++;
3203        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3204          {          {
3205          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3206          ptr++;          ptr++;
3207  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3208          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3209  #endif  #endif
3210          }          }
3211        }        }
# Line 2862  if ((options & PCRE_EXTENDED) != 0) Line 3216  if ((options & PCRE_EXTENDED) != 0)
3216  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3217    
3218  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3219    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)
3220      return FALSE;      return FALSE;
3221    
3222  /* 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. */  
3223    
3224  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3225        op_code == OP_NOT || op_code == OP_NOTI)
3226    {    {
3227    case OP_CHAR:  #ifdef SUPPORT_UTF
 #ifdef SUPPORT_UTF8  
3228    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3229  #else  #else
3230    c = *previous;    c = *previous;
3231  #endif  #endif
3232    return c != next;    }
3233    
3234    /* Now compare the next item with the previous opcode. First, handle cases when
3235    the next item is a character. */
3236    
3237    /* For CHARI (caseless character) we must check the other case. If we have  if (escape == 0)
3238    Unicode property support, we can use it to test the other case of    {
3239    high-valued characters. */    /* For a caseless UTF match, the next character may have more than one other
3240      case, which maps to the special PT_CLIST property. Check this first. */
3241    
3242    case OP_CHARI:  #ifdef SUPPORT_UCP
3243  #ifdef SUPPORT_UTF8    if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3244    GETCHARTEST(c, previous);      {
3245  #else      unsigned int ocs = UCD_CASESET(next);
3246    c = *previous;      if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3247        }
3248  #endif  #endif
3249    if (c == next) return FALSE;  
3250  #ifdef SUPPORT_UTF8    switch(op_code)
   if (utf8)  
3251      {      {
3252      unsigned int othercase;      case OP_CHAR:
3253      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3254    
3255        /* For CHARI (caseless character) we must check the other case. If we have
3256        Unicode property support, we can use it to test the other case of
3257        high-valued characters. We know that next can have only one other case,
3258        because multi-other-case characters are dealt with above. */
3259    
3260        case OP_CHARI:
3261        if (c == next) return FALSE;
3262    #ifdef SUPPORT_UTF
3263        if (utf)
3264          {
3265          pcre_uint32 othercase;
3266          if (next < 128) othercase = cd->fcc[next]; else
3267  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3268      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3269  #else  #else
3270      othercase = NOTACHAR;        othercase = NOTACHAR;
3271  #endif  #endif
3272      return (unsigned int)c != othercase;        return c != othercase;
3273      }        }
3274    else      else
3275  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3276    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3277    
3278    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These      case OP_NOT:
3279    opcodes are not used for multi-byte characters, because they are coded using      return c == next;
3280    an XCLASS instead. */  
3281        case OP_NOTI:
3282    case OP_NOT:      if (c == next) return TRUE;
3283    return (c = *previous) == next;  #ifdef SUPPORT_UTF
3284        if (utf)
3285    case OP_NOTI:        {
3286    if ((c = *previous) == next) return TRUE;        pcre_uint32 othercase;
3287  #ifdef SUPPORT_UTF8        if (next < 128) othercase = cd->fcc[next]; else
   if (utf8)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3288  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3289      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE(next);
3290  #else  #else
3291      othercase = NOTACHAR;        othercase = NOTACHAR;
3292  #endif  #endif
3293      return (unsigned int)c == othercase;        return c == othercase;
3294      }        }
3295    else      else
3296  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3297    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3298    
3299    /* 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.
3300    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. */
3301    
3302    case OP_DIGIT:      case OP_DIGIT:
3303    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3304    
3305    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3306    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3307    
3308    case OP_WHITESPACE:      case OP_WHITESPACE:
3309    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3310    
3311    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3312    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3313    
3314    case OP_WORDCHAR:      case OP_WORDCHAR:
3315    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3316    
3317    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3318    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3319    
3320    case OP_HSPACE:      case OP_HSPACE:
3321    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3322    switch(next)      switch(next)
3323      {        {
3324      case 0x09:        HSPACE_CASES:
3325      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;  
     }  
3326    
3327    case OP_ANYNL:        default:
3328    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3329    case OP_NOT_VSPACE:        }
3330    switch(next)  
3331      {      case OP_ANYNL:
3332      case 0x0a:      case OP_VSPACE:
3333      case 0x0b:      case OP_NOT_VSPACE:
3334      case 0x0c:      switch(next)
3335      case 0x0d:        {
3336      case 0x85:        VSPACE_CASES:
3337      case 0x2028:        return op_code == OP_NOT_VSPACE;
3338      case 0x2029:  
3339      return op_code == OP_NOT_VSPACE;        default:
3340      default:        return op_code != OP_NOT_VSPACE;
3341      return op_code != OP_NOT_VSPACE;        }
     }  
3342    
3343  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3344    case OP_PROP:      case OP_PROP:
3345    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3346    
3347    case OP_NOTPROP:      case OP_NOTPROP:
3348    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3349  #endif  #endif
3350    
3351    default:      default:
3352    return FALSE;      return FALSE;
3353        }
3354    }    }
3355    
   
3356  /* 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
3357  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
3358  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 3019  switch(op_code) Line 3363  switch(op_code)
3363    {    {
3364    case OP_CHAR:    case OP_CHAR:
3365    case OP_CHARI:    case OP_CHARI:
3366  #ifdef SUPPORT_UTF8    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3367      {      {
3368      case ESC_d:      case ESC_d:
3369      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3370    
3371      case ESC_D:      case ESC_D:
3372      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3373    
3374      case ESC_s:      case ESC_s:
3375      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3376    
3377      case ESC_S:      case ESC_S:
3378      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3379    
3380      case ESC_w:      case ESC_w:
3381      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3382    
3383      case ESC_W:      case ESC_W:
3384      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3385    
3386      case ESC_h:      case ESC_h:
3387      case ESC_H:      case ESC_H:
3388      switch(c)      switch(c)
3389        {        {
3390        case 0x09:        HSPACE_CASES:
3391        case 0x20:        return escape != ESC_h;
3392        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;  
3393        default:        default:
3394        return -next == ESC_h;        return escape == ESC_h;
3395        }        }
3396    
3397      case ESC_v:      case ESC_v:
3398      case ESC_V:      case ESC_V:
3399      switch(c)      switch(c)
3400        {        {
3401        case 0x0a:        VSPACE_CASES:
3402        case 0x0b:        return escape != ESC_v;
3403        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3404        default:        default:
3405        return -next == ESC_v;        return escape == ESC_v;
3406        }        }
3407    
3408      /* 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
3409      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3410      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3411    
3412  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3413      case ESC_du:      case ESC_du:
# Line 3101  switch(op_code) Line 3418  switch(op_code)
3418      case ESC_SU:      case ESC_SU:
3419        {        {
3420        int temperrorcode = 0;        int temperrorcode = 0;
3421        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3422        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3423        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3424        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3425        }        }
# Line 3111  switch(op_code) Line 3428  switch(op_code)
3428      case ESC_p:      case ESC_p:
3429      case ESC_P:      case ESC_P:
3430        {        {
3431        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3432          int errorcodeptr;
3433        BOOL negated;        BOOL negated;
3434    
3435        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3436        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3437        if (ptype < 0) return FALSE;          return FALSE;
3438        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3439    
3440        /* 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 3124  switch(op_code) Line 3442  switch(op_code)
3442        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. */
3443    
3444        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3445          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)
3446            return FALSE;            return FALSE;
3447    
3448        /* Do the property check. */        /* Do the property check. */
3449    
3450        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3451        }        }
3452  #endif  #endif
3453    
# Line 3144  switch(op_code) Line 3462  switch(op_code)
3462    these op-codes are never generated.) */    these op-codes are never generated.) */
3463    
3464    case OP_DIGIT:    case OP_DIGIT:
3465    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3466           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3467    
3468    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3469    return next == -ESC_d;    return escape == ESC_d;
3470    
3471    case OP_WHITESPACE:    case OP_WHITESPACE:
3472    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3473    
3474    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3475    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3476    
3477    case OP_HSPACE:    case OP_HSPACE:
3478    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3479           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3480    
3481    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3482    return next == -ESC_h;    return escape == ESC_h;
3483    
3484    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3485    case OP_ANYNL:    case OP_ANYNL:
3486    case OP_VSPACE:    case OP_VSPACE:
3487    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3488    
3489    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3490    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3491    
3492    case OP_WORDCHAR:    case OP_WORDCHAR:
3493    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3494           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3495    
3496    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3497    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3498    
3499    default:    default:
3500    return FALSE;    return FALSE;
# Line 3188  switch(op_code) Line 3506  switch(op_code)
3506    
3507    
3508  /*************************************************  /*************************************************
3509    *        Add a character or range to a class     *
3510    *************************************************/
3511    
3512    /* This function packages up the logic of adding a character or range of
3513    characters to a class. The character values in the arguments will be within the
3514    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3515    mutually recursive with the function immediately below.
3516    
3517    Arguments:
3518      classbits     the bit map for characters < 256
3519      uchardptr     points to the pointer for extra data
3520      options       the options word
3521      cd            contains pointers to tables etc.
3522      start         start of range character
3523      end           end of range character
3524    
3525    Returns:        the number of < 256 characters added
3526                    the pointer to extra data is updated
3527    */
3528    
3529    static int
3530    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3531      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3532    {
3533    pcre_uint32 c;
3534    int n8 = 0;
3535    
3536    /* If caseless matching is required, scan the range and process alternate
3537    cases. In Unicode, there are 8-bit characters that have alternate cases that
3538    are greater than 255 and vice-versa. Sometimes we can just extend the original
3539    range. */
3540    
3541    if ((options & PCRE_CASELESS) != 0)
3542      {
3543    #ifdef SUPPORT_UCP
3544      if ((options & PCRE_UTF8) != 0)
3545        {
3546        int rc;
3547        pcre_uint32 oc, od;
3548    
3549        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3550        c = start;
3551    
3552        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3553          {
3554          /* Handle a single character that has more than one other case. */
3555    
3556          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3557            PRIV(ucd_caseless_sets) + rc, oc);
3558    
3559          /* Do nothing if the other case range is within the original range. */
3560    
3561          else if (oc >= start && od <= end) continue;
3562    
3563          /* Extend the original range if there is overlap, noting that if oc < c, we
3564          can't have od > end because a subrange is always shorter than the basic
3565          range. Otherwise, use a recursive call to add the additional range. */
3566    
3567          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3568          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3569          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3570          }
3571        }
3572      else
3573    #endif  /* SUPPORT_UCP */
3574    
3575      /* Not UTF-mode, or no UCP */
3576    
3577      for (c = start; c <= end && c < 256; c++)
3578        {
3579        SETBIT(classbits, cd->fcc[c]);
3580        n8++;
3581        }
3582      }
3583    
3584    /* Now handle the original range. Adjust the final value according to the bit
3585    length - this means that the same lists of (e.g.) horizontal spaces can be used
3586    in all cases. */
3587    
3588    #if defined COMPILE_PCRE8
3589    #ifdef SUPPORT_UTF
3590      if ((options & PCRE_UTF8) == 0)
3591    #endif
3592      if (end > 0xff) end = 0xff;
3593    
3594    #elif defined COMPILE_PCRE16
3595    #ifdef SUPPORT_UTF
3596      if ((options & PCRE_UTF16) == 0)
3597    #endif
3598      if (end > 0xffff) end = 0xffff;
3599    
3600    #endif /* COMPILE_PCRE[8|16] */
3601    
3602    /* If all characters are less than 256, use the bit map. Otherwise use extra
3603    data. */
3604    
3605    if (end < 0x100)
3606      {
3607      for (c = start; c <= end; c++)
3608        {
3609        n8++;
3610        SETBIT(classbits, c);
3611        }
3612      }
3613    
3614    else
3615      {
3616      pcre_uchar *uchardata = *uchardptr;
3617    
3618    #ifdef SUPPORT_UTF
3619      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3620        {
3621        if (start < end)
3622          {
3623          *uchardata++ = XCL_RANGE;
3624          uchardata += PRIV(ord2utf)(start, uchardata);
3625          uchardata += PRIV(ord2utf)(end, uchardata);
3626          }
3627        else if (start == end)
3628          {
3629          *uchardata++ = XCL_SINGLE;
3630          uchardata += PRIV(ord2utf)(start, uchardata);
3631          }
3632        }
3633      else
3634    #endif  /* SUPPORT_UTF */
3635    
3636      /* Without UTF support, character values are constrained by the bit length,
3637      and can only be > 256 for 16-bit and 32-bit libraries. */
3638    
3639    #ifdef COMPILE_PCRE8
3640        {}
3641    #else
3642      if (start < end)
3643        {
3644        *uchardata++ = XCL_RANGE;
3645        *uchardata++ = start;
3646        *uchardata++ = end;
3647        }
3648      else if (start == end)
3649        {
3650        *uchardata++ = XCL_SINGLE;
3651        *uchardata++ = start;
3652        }
3653    #endif
3654    
3655      *uchardptr = uchardata;   /* Updata extra data pointer */
3656      }
3657    
3658    return n8;    /* Number of 8-bit characters */
3659    }
3660    
3661    
3662    
3663    
3664    /*************************************************
3665    *        Add a list of characters to a class     *
3666    *************************************************/
3667    
3668    /* This function is used for adding a list of case-equivalent characters to a
3669    class, and also for adding a list of horizontal or vertical whitespace. If the
3670    list is in order (which it should be), ranges of characters are detected and
3671    handled appropriately. This function is mutually recursive with the function
3672    above.
3673    
3674    Arguments:
3675      classbits     the bit map for characters < 256
3676      uchardptr     points to the pointer for extra data
3677      options       the options word
3678      cd            contains pointers to tables etc.
3679      p             points to row of 32-bit values, terminated by NOTACHAR
3680      except        character to omit; this is used when adding lists of
3681                      case-equivalent characters to avoid including the one we
3682                      already know about
3683    
3684    Returns:        the number of < 256 characters added
3685                    the pointer to extra data is updated
3686    */
3687    
3688    static int
3689    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3690      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3691    {
3692    int n8 = 0;
3693    while (p[0] < NOTACHAR)
3694      {
3695      int n = 0;
3696      if (p[0] != except)
3697        {
3698        while(p[n+1] == p[0] + n + 1) n++;
3699        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3700        }
3701      p += n + 1;
3702      }
3703    return n8;
3704    }
3705    
3706    
3707    
3708    /*************************************************
3709    *    Add characters not in a list to a class     *
3710    *************************************************/
3711    
3712    /* This function is used for adding the complement of a list of horizontal or
3713    vertical whitespace to a class. The list must be in order.
3714    
3715    Arguments:
3716      classbits     the bit map for characters < 256
3717      uchardptr     points to the pointer for extra data
3718      options       the options word
3719      cd            contains pointers to tables etc.
3720      p             points to row of 32-bit values, terminated by NOTACHAR
3721    
3722    Returns:        the number of < 256 characters added
3723                    the pointer to extra data is updated
3724    */
3725    
3726    static int
3727    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3728      int options, compile_data *cd, const pcre_uint32 *p)
3729    {
3730    BOOL utf = (options & PCRE_UTF8) != 0;
3731    int n8 = 0;
3732    if (p[0] > 0)
3733      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3734    while (p[0] < NOTACHAR)
3735      {
3736      while (p[1] == p[0] + 1) p++;
3737      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3738        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3739      p++;
3740      }
3741    return n8;
3742    }
3743    
3744    
3745    
3746    /*************************************************
3747  *           Compile one branch                   *  *           Compile one branch                   *
3748  *************************************************/  *************************************************/
3749    
# Line 3202  Arguments: Line 3758  Arguments:
3758    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3759    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3760    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3761    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3762    reqbyteptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3763      reqcharptr     place to put the last required character
3764      reqcharflagsptr place to put the last required character flags, or a negative number
3765    bcptr          points to current branch chain    bcptr          points to current branch chain
3766    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3767    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3215  Returns:         TRUE on success Line 3773  Returns:         TRUE on success
3773  */  */
3774    
3775  static BOOL  static BOOL
3776  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3777    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3778    int cond_depth, compile_data *cd, int *lengthptr)    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3779      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3780      branch_chain *bcptr, int cond_depth,
3781      compile_data *cd, int *lengthptr)
3782  {  {
3783  int repeat_type, op_type;  int repeat_type, op_type;
3784  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3785  int bravalue = 0;  int bravalue = 0;
3786  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3787  int firstbyte, reqbyte;  pcre_uint32 firstchar, reqchar;
3788  int zeroreqbyte, zerofirstbyte;  pcre_int32 firstcharflags, reqcharflags;
3789  int req_caseopt, reqvary, tempreqvary;  pcre_uint32 zeroreqchar, zerofirstchar;
3790    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3791    pcre_int32 req_caseopt, reqvary, tempreqvary;
3792  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3793  int after_manual_callout = 0;  int after_manual_callout = 0;
3794  int length_prevgroup = 0;  int length_prevgroup = 0;
3795  register int c;  register pcre_uint32 c;
3796  register uschar *code = *codeptr;  int escape;
3797  uschar *last_code = code;  register pcre_uchar *code = *codeptr;
3798  uschar *orig_code = code;  pcre_uchar *last_code = code;
3799  uschar *tempcode;  pcre_uchar *orig_code = code;
3800    pcre_uchar *tempcode;
3801  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3802  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3803  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3804  const uschar *tempptr;  const pcre_uchar *tempptr;
3805  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3806  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3807  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3808  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3809  uschar classbits[32];  pcre_uint8 classbits[32];
3810    
3811  /* 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
3812  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
3813  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3814    
3815  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3816  BOOL class_utf8;  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3817  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3818  uschar *class_utf8data;  #ifndef COMPILE_PCRE32
3819  uschar *class_utf8data_base;  pcre_uchar utf_chars[6];
3820  uschar utf8_char[6];  #endif
3821  #else  #else
3822  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3823    #endif
3824    
3825    /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3826    class_uchardata always so that it can be passed to add_to_class() always,
3827    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3828    alternative calls for the different cases. */
3829    
3830    pcre_uchar *class_uchardata;
3831    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3832    BOOL xclass;
3833    pcre_uchar *class_uchardata_base;
3834  #endif  #endif
3835    
3836  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3269  greedy_non_default = greedy_default ^ 1; Line 3844  greedy_non_default = greedy_default ^ 1;
3844    
3845  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3846  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
3847  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
3848  find one.  find one.
3849    
3850  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
3851  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
3852  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3853  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3854    
3855  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3856    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3857    
3858  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3859  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
3860  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
3861  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3862    value. This is used only for ASCII characters. */
3863    
3864  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3865    
3866  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3867    
# Line 3296  for (;; ptr++) Line 3873  for (;; ptr++)
3873    BOOL is_quantifier;    BOOL is_quantifier;
3874    BOOL is_recurse;    BOOL is_recurse;
3875    BOOL reset_bracount;    BOOL reset_bracount;
3876    int class_charcount;    int class_has_8bitchar;
3877    int class_lastchar;    int class_one_char;
3878    int newoptions;    int newoptions;
3879    int recno;    int recno;
3880    int refsign;    int refsign;
3881    int skipbytes;    int skipbytes;
3882    int subreqbyte;    pcre_uint32 subreqchar, subfirstchar;
3883    int subfirstbyte;    pcre_int32 subreqcharflags, subfirstcharflags;
3884    int terminator;    int terminator;
3885    int mclength;    unsigned int mclength;
3886    int tempbracount;    unsigned int tempbracount;
3887    uschar mcbuffer[8];    pcre_uint32 ec;
3888      pcre_uchar mcbuffer[8];
3889    
3890    /* Get next byte in the pattern */    /* Get next character in the pattern */
3891    
3892    c = *ptr;    c = *ptr;
3893    
3894    /* 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
3895    string. Nesting only happens one level deep. */    string. Nesting only happens one level deep. */
3896    
3897    if (c == 0 && nestptr != NULL)    if (c == CHAR_NULL && nestptr != NULL)
3898      {      {
3899      ptr = nestptr;      ptr = nestptr;
3900      nestptr = NULL;      nestptr = NULL;
# Line 3331  for (;; ptr++) Line 3909  for (;; ptr++)
3909  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3910      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3911  #endif  #endif
3912      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */      if (code > cd->start_workspace + cd->workspace_size -
3913            WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3914        {        {
3915        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3916        goto FAILED;        goto FAILED;
# Line 3354  for (;; ptr++) Line 3933  for (;; ptr++)
3933        }        }
3934    
3935      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3936      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, (int)(code - last_code),      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3937        c));        (int)(code - last_code), c, c));
3938    
3939      /* 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
3940      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 3365  for (;; ptr++) Line 3944  for (;; ptr++)
3944        {        {
3945        if (previous > orig_code)        if (previous > orig_code)
3946          {          {
3947          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3948          code -= previous - orig_code;          code -= previous - orig_code;
3949          previous = orig_code;          previous = orig_code;
3950          }          }
# Line 3381  for (;; ptr++) Line 3960  for (;; ptr++)
3960    /* 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
3961    reference list. */    reference list. */
3962    
3963    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3964               WORK_SIZE_SAFETY_MARGIN)
3965      {      {
3966      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3967      goto FAILED;      goto FAILED;
# Line 3389  for (;; ptr++) Line 3969  for (;; ptr++)
3969    
3970    /* 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 */
3971    
3972    if (inescq && c != 0)    if (inescq && c != CHAR_NULL)
3973      {      {
3974      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3975        {        {
# Line 3433  for (;; ptr++) Line 4013  for (;; ptr++)
4013    
4014    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
4015      {      {
4016      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
4017      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
4018        {        {
4019        ptr++;        ptr++;
4020        while (*ptr != 0)        while (*ptr != CHAR_NULL)
4021          {          {
4022          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
4023          ptr++;          ptr++;
4024  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4025          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
4026  #endif  #endif
4027          }          }
4028        if (*ptr != 0) continue;        if (*ptr != CHAR_NULL) continue;
4029    
4030        /* Else fall through to handle end of string */        /* Else fall through to handle end of string */
4031        c = 0;        c = 0;
# Line 3466  for (;; ptr++) Line 4046  for (;; ptr++)
4046      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
4047      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
4048      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
4049      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
4050      *reqbyteptr = reqbyte;      *firstcharflagsptr = firstcharflags;
4051        *reqcharptr = reqchar;
4052        *reqcharflagsptr = reqcharflags;
4053      *codeptr = code;      *codeptr = code;
4054      *ptrptr = ptr;      *ptrptr = ptr;
4055      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3491  for (;; ptr++) Line 4073  for (;; ptr++)
4073      previous = NULL;      previous = NULL;
4074      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
4075        {        {
4076        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4077        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
4078        }        }
4079      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3503  for (;; ptr++) Line 4085  for (;; ptr++)
4085      break;      break;
4086    
4087      /* 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
4088      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4089    
4090      case CHAR_DOT:      case CHAR_DOT:
4091      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4092      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
4093      zeroreqbyte = reqbyte;      zerofirstcharflags = firstcharflags;
4094        zeroreqchar = reqchar;
4095        zeroreqcharflags = reqcharflags;
4096      previous = code;      previous = code;
4097      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4098      break;      break;
# Line 3563  for (;; ptr++) Line 4147  for (;; ptr++)
4147          {          {
4148          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
4149            ptr++;            ptr++;
4150          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)  
4151            ptr += 3;            ptr += 3;
4152          else          else
4153            break;            break;
# Line 3583  for (;; ptr++) Line 4166  for (;; ptr++)
4166          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4167        {        {
4168        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4169        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4170        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
4171          zerofirstcharflags = firstcharflags;
4172        break;        break;
4173        }        }
4174    
# Line 3594  for (;; ptr++) Line 4178  for (;; ptr++)
4178    
4179      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4180    
4181      /* 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:
4182      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 <
4183      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
4184        character. */
4185    
4186      class_charcount = 0;      class_has_8bitchar = 0;
4187      class_lastchar = -1;      class_one_char = 0;
4188    
4189      /* 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
4190      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
4191      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
4192      */      map. */
4193    
4194      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4195    
4196  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4197      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;
4198      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4199      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4200  #endif  #endif
4201    
4202      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
4203      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
4204      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
4205    
4206      if (c != 0) do      if (c != CHAR_NULL) do
4207        {        {
4208        const uschar *oldptr;        const pcre_uchar *oldptr;
4209    
4210  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4211        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
4212          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4213          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
4214          }          }
4215    #endif
4216    
4217        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4218          /* In the pre-compile phase, accumulate the length of any extra
4219        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
4220        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4221        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4222          however. */
4223        if (lengthptr != NULL)  
4224          {        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4225          *lengthptr += class_utf8data - class_utf8data_base;          {
4226          class_utf8data = class_utf8data_base;          xclass = TRUE;
4227            *lengthptr += class_uchardata - class_uchardata_base;
4228            class_uchardata = class_uchardata_base;
4229          }          }
   
4230  #endif  #endif
4231    
4232        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3666  for (;; ptr++) Line 4254  for (;; ptr++)
4254          {          {
4255          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
4256          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
4257          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
4258          uschar pbits[32];          pcre_uint8 pbits[32];
4259    
4260          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
4261            {            {
# Line 3722  for (;; ptr++) Line 4310  for (;; ptr++)
4310          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
4311    
4312          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
4313            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
4314    
4315          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
4316    
# Line 3737  for (;; ptr++) Line 4325  for (;; ptr++)
4325              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4326            }            }
4327    
4328          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4329          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4330    
4331          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3753  for (;; ptr++) Line 4341  for (;; ptr++)
4341            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4342    
4343          ptr = tempptr + 1;          ptr = tempptr + 1;
4344          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 character. */
4345            class_has_8bitchar = 1;
4346            /* Every class contains at least two characters. */
4347            class_one_char = 2;
4348          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4349          }          }
4350    
4351        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
4352        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
4353        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
4354        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
4355        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_one_char bigger
4356        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
4357          as literal characters (by default), or are faulted if
4358        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4359    
4360        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4361          {          {
4362          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options,
4363              TRUE);
4364          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4365            if (escape == 0) c = ec;
4366          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 */
4367          else if (-c == ESC_N)            /* \N is not supported in a class */          else if (escape == ESC_N)          /* \N is not supported in a class */
4368            {            {
4369            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4370            goto FAILED;            goto FAILED;
4371            }            }
4372          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4373            {            {
4374            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4375              {              {
# Line 3785  for (;; ptr++) Line 4378  for (;; ptr++)
4378            else inescq = TRUE;            else inescq = TRUE;
4379            continue;            continue;
4380            }            }
4381          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4382    
4383          if (c < 0)          else
4384            {            {
4385            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4386            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
4387              class_has_8bitchar++;
4388              /* Every class contains at least two characters. */
4389              class_one_char += 2;
4390    
4391            switch (-c)            switch (escape)
4392              {              {
4393  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4394              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3802  for (;; ptr++) Line 4398  for (;; ptr++)
4398              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4399              case ESC_SU:              case ESC_SU:
4400              nestptr = ptr;              nestptr = ptr;
4401              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4402              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4403              continue;              continue;
4404  #endif  #endif
4405              case ESC_d:              case ESC_d:
# Line 3826  for (;; ptr++) Line 4422  for (;; ptr++)
4422    
4423              /* 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
4424              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4425              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4426                EBCDIC, so we lazily just adjust the appropriate bit. */
4427    
4428              case ESC_s:              case ESC_s:
4429              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 3840  for (;; ptr++) Line 4437  for (;; ptr++)
4437              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4438              continue;              continue;
4439    
4440                /* The rest apply in both UCP and non-UCP cases. */
4441    
4442              case ESC_h:              case ESC_h:
4443              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4444              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  
4445              continue;              continue;
4446    
4447              case ESC_H:              case ESC_H:
4448              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4449                {                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  
4450              continue;              continue;
4451    
4452              case ESC_v:              case ESC_v:
4453              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4454              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  
4455              continue;              continue;
4456    
4457              case ESC_V:              case ESC_V:
4458              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4459                {                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  
4460              continue;              continue;
4461    
4462  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3961  for (;; ptr++) Line 4464  for (;; ptr++)
4464              case ESC_P:              case ESC_P:
4465                {                {
4466                BOOL negated;                BOOL negated;
4467                int pdata;                unsigned int ptype = 0, pdata = 0;
4468                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4469                if (ptype < 0) goto FAILED;                  goto FAILED;
4470                class_utf8 = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_utf8data++ = ((-c == ESC_p) != negated)?  
4471                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4472                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4473                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4474                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4475                continue;                continue;
4476                }                }
4477  #endif  #endif
# Line 3983  for (;; ptr++) Line 4485  for (;; ptr++)
4485                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4486                goto FAILED;                goto FAILED;
4487                }                }
4488              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4489              c = *ptr;              /* Get the final character and fall through */              class_one_char -= 2;     /* Undo the speculative increase. */
4490                c = *ptr;                /* Get the final character and fall through */
4491              break;              break;
4492              }              }
4493            }            }
4494    
4495          /* 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).
4496          greater than 256 in UTF-8 mode. */          This may be greater than 256. */
4497    
4498            escape = 0;
4499    
4500          }   /* End of backslash handling */          }   /* End of backslash handling */
4501    
4502        /* 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
4503        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
4504        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4505        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4506    
4507        CHECK_RANGE:        CHECK_RANGE:
4508        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4005  for (;; ptr++) Line 4510  for (;; ptr++)
4510          inescq = FALSE;          inescq = FALSE;
4511          ptr += 2;          ptr += 2;
4512          }          }
   
4513        oldptr = ptr;        oldptr = ptr;
4514    
4515        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4516    
4517        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4518    
# Line 4016  for (;; ptr++) Line 4520  for (;; ptr++)
4520    
4521        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4522          {          {
4523          int d;          pcre_uint32 d;
4524          ptr += 2;          ptr += 2;
4525          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4526    
# Line 4032  for (;; ptr++) Line 4536  for (;; ptr++)
4536            break;            break;
4537            }            }
4538    
4539          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          /* Minus (hyphen) at the end of a class is treated as a literal, so put
4540            back the pointer and jump to handle the character that preceded it. */
4541    
4542            if (*ptr == CHAR_NULL || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4543            {            {
4544            ptr = oldptr;            ptr = oldptr;
4545            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4546            }            }
4547    
4548  #ifdef SUPPORT_UTF8          /* Otherwise, we have a potential range; pick up the next character */
4549          if (utf8)  
4550    #ifdef SUPPORT_UTF
4551            if (utf)
4552            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4553            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4554            }            }
# Line 4053  for (;; ptr++) Line 4562  for (;; ptr++)
4562    
4563          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4564            {            {
4565            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4566              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4567            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4568    
4569            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4570    
4571            if (d < 0)            if (descape != 0)
4572              {              {
4573              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4574                {                {
4575                ptr = oldptr;                ptr = oldptr;
4576                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4577                }                }
4578              }              }
4579            }            }
4580    
4581          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4582          one-character ranges */          one-character ranges. */
4583    
4584          if (d < c)          if (d < c)
4585            {            {
4586            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4587            goto FAILED;            goto FAILED;
4588            }            }
4589            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4590    
4591          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4592            cannot be done anymore. Any value greater than 1 indicates that there
4593          /* Remember \r or \n */          is more than one character. */
   
         if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;  
   
         /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless  
         matching, we have to use an XCLASS with extra data items. Caseless  
         matching for characters > 127 is available only if UCP support is  
         available. */  
   
 #ifdef SUPPORT_UTF8  
         if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))  
           {  
           class_utf8 = TRUE;  
   
           /* With UCP support, we can find the other case equivalents of  
           the relevant characters. There may be several ranges. Optimize how  
           they fit with the basic range. */  
   
 #ifdef SUPPORT_UCP  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             unsigned int occ, ocd;  
             unsigned int cc = c;  
             unsigned int origd = d;  
             while (get_othercase_range(&cc, origd, &occ, &ocd))  
               {  
               if (occ >= (unsigned int)c &&  
                   ocd <= (unsigned int)d)  
                 continue;                          /* Skip embedded ranges */  
   
               if (occ < (unsigned int)c  &&  
                   ocd >= (unsigned int)c - 1)      /* Extend the basic range */  
                 {                                  /* if there is overlap,   */  
                 c = occ;                           /* noting that if occ < c */  
                 continue;                          /* we can't have ocd > d  */  
                 }                                  /* because a subrange is  */  
               if (ocd > (unsigned int)d &&  
                   occ <= (unsigned int)d + 1)      /* always shorter than    */  
                 {                                  /* the basic range.       */  
                 d = ocd;  
                 continue;  
                 }  
   
               if (occ == ocd)  
                 {  
                 *class_utf8data++ = XCL_SINGLE;  
                 }  
               else  
                 {  
                 *class_utf8data++ = XCL_RANGE;  
                 class_utf8data += _pcre_ord2utf8(occ, class_utf8data);  
                 }  
               class_utf8data += _pcre_ord2utf8(ocd, class_utf8data);  
               }  
             }  
 #endif  /* SUPPORT_UCP */  
   
           /* Now record the original range, possibly modified for UCP caseless  
           overlapping ranges. */  
   
           *class_utf8data++ = XCL_RANGE;  
           class_utf8data += _pcre_ord2utf8(c, class_utf8data);  
           class_utf8data += _pcre_ord2utf8(d, class_utf8data);  
   
           /* With UCP support, we are done. Without UCP support, there is no  
           caseless matching for UTF-8 characters > 127; we can use the bit map  
           for the smaller ones. */  
   
 #ifdef SUPPORT_UCP  
           continue;    /* With next character in the class */  
 #else  
           if ((options & PCRE_CASELESS) == 0 || c > 127) continue;  
   
           /* Adjust upper limit and fall through to set up the map */  
4594    
4595            d = 127;          class_one_char = 2;
4596    
4597  #endif  /* SUPPORT_UCP */          /* Remember an explicit \r or \n, and add the range to the class. */
           }  
 #endif  /* SUPPORT_UTF8 */  
4598    
4599          /* We use the bit map for all cases when not in UTF-8 mode; else          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
         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;  
4600    
4601          /* We can save a bit of time by skipping this in the pre-compile. */          class_has_8bitchar +=
4602              add_to_class(classbits, &class_uchardata, options, cd, c, d);
         if (lengthptr == NULL) for (; c <= d; c++)  
           {  
           classbits[c/8] |= (1 << (c&7));  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             int uc = cd->fcc[c];           /* flip case */  
             classbits[uc/8] |= (1 << (uc&7));  
             }  
           }  
4603    
4604          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4605          }          }
4606    
4607        /* Handle a lone single character - we can get here for a normal        /* Handle a single character - we can get here for a normal non-escape
4608        non-escape char, or after \ that introduces a single character or for an        char, or after \ that introduces a single character or for an apparent
4609        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4610          increase it if it is already 2 or more ... just in case there's a class
4611        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4612    
4613        /* Handle a character that cannot go in the bit map */        CLASS_SINGLE_CHARACTER:
4614          if (class_one_char < 2) class_one_char++;
4615    
4616          /* If class_one_char is 1, we have the first single character in the
4617          class, and there have been no prior ranges, or XCLASS items generated by
4618          escapes. If this is the final character in the class, we can optimize by
4619          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4620          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4621          to be set. Otherwise, there can be no first char if this item is first,
4622          whatever repeat count may follow. In the case of reqchar, save the
4623          previous value for reinstating. */
4624    
4625  #ifdef SUPPORT_UTF8        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
       if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
4626          {          {
4627          class_utf8 = TRUE;          ptr++;
4628          *class_utf8data++ = XCL_SINGLE;          zeroreqchar = reqchar;
4629          class_utf8data += _pcre_ord2utf8(c, class_utf8data);          zeroreqcharflags = reqcharflags;
4630    
4631  #ifdef SUPPORT_UCP          if (negate_class)
         if ((options & PCRE_CASELESS) != 0)  
4632            {            {
4633            unsigned int othercase;  #ifdef SUPPORT_UCP
4634            if ((othercase = UCD_OTHERCASE(c)) != c)            int d;
4635    #endif
4636              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4637              zerofirstchar = firstchar;
4638              zerofirstcharflags = firstcharflags;
4639    
4640              /* For caseless UTF-8 mode when UCP support is available, check
4641              whether this character has more than one other case. If so, generate
4642              a special OP_NOTPROP item instead of OP_NOTI. */
4643    
4644    #ifdef SUPPORT_UCP
4645              if (utf && (options & PCRE_CASELESS) != 0 &&
4646                  (d = UCD_CASESET(c)) != 0)
4647                {
4648                *code++ = OP_NOTPROP;
4649                *code++ = PT_CLIST;
4650                *code++ = d;
4651                }
4652              else
4653    #endif
4654              /* Char has only one other case, or UCP not available */
4655    
4656              {              {
4657              *class_utf8data++ = XCL_SINGLE;              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4658              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4659                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4660                  code += PRIV(ord2utf)(c, code);
4661                else
4662    #endif
4663                  *code++ = c;
4664              }              }
4665    
4666              /* We are finished with this character class */
4667    
4668              goto END_CLASS;
4669            }            }
 #endif  /* SUPPORT_UCP */  
4670    
4671          }          /* For a single, positive character, get the value into mcbuffer, and
4672        else          then we can handle this with the normal one-character code. */
 #endif  /* SUPPORT_UTF8 */  
4673    
4674        /* Handle a single-byte character */  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4675          {          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4676          classbits[c/8] |= (1 << (c&7));            mclength = PRIV(ord2utf)(c, mcbuffer);
4677          if ((options & PCRE_CASELESS) != 0)          else
4678    #endif
4679            {            {
4680            c = cd->fcc[c];   /* flip case */            mcbuffer[0] = c;
4681            classbits[c/8] |= (1 << (c&7));            mclength = 1;
4682            }            }
4683          class_charcount++;          goto ONE_CHAR;
4684          class_lastchar = c;          }       /* End of 1-char optimization */
4685          }  
4686          /* There is more than one character in the class, or an XCLASS item
4687          has been generated. Add this character to the class. */
4688    
4689          class_has_8bitchar +=
4690            add_to_class(classbits, &class_uchardata, options, cd, c, c);
4691        }        }
4692    
4693      /* 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.
4694      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
4695      string. */      string. */
4696    
4697      while (((c = *(++ptr)) != 0 ||      while (((c = *(++ptr)) != CHAR_NULL ||
4698             (nestptr != NULL &&             (nestptr != NULL &&
4699               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != CHAR_NULL)) &&
4700             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
4701    
4702      /* Check for missing terminating ']' */      /* Check for missing terminating ']' */
4703    
4704      if (c == 0)      if (c == CHAR_NULL)
4705        {        {
4706        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
4707        goto FAILED;        goto FAILED;
4708        }        }
4709    
4710      /* 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
4711      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
4712      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
4713      optimize.      only if the very last character in the class needs XCLASS will it contain
4714        anything at this point. For this reason, xclass gets set TRUE above when
4715      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
4716      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      instead of just doing a test on class_uchardata below. */
4717      operate on single-bytes characters only. This is an historical hangover.  
4718      Maybe one day we can tidy these opcodes to handle multi-byte characters.  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4719        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4720      The optimization throws away the bit map. We turn the item into a  #endif
4721      1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
4722      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
4723      case, it can cause firstbyte to be set. Otherwise, there can be no first      setting, whatever the repeat count. Any reqchar setting must remain
4724      char if this item is first, whatever repeat count may follow. In the case      unchanged after any kind of repeat. */
4725      of reqbyte, save the previous value for reinstating. */  
4726        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4727  #ifdef SUPPORT_UTF8      zerofirstchar = firstchar;
4728      if (class_charcount == 1 && !class_utf8 &&      zerofirstcharflags = firstcharflags;
4729        (!utf8 || !negate_class || class_lastchar < 128))      zeroreqchar = reqchar;
4730  #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;  
&nb