/[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 579 by ph10, Wed Nov 24 17:39:25 2010 UTC revision 1219 by ph10, Sun Nov 11 18:04:37 2012 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2010 University of Cambridge             Copyright (c) 1997-2012 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. */  library. We do not need to select pcre16_printint.c specially, because the
59    COMPILE_PCREx macro will already be appropriately set. */
60    
61  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
62  #include "pcre_printint.src"  /* pcre_printint.c should not include any headers */
63    #define PCRE_INCLUDED
64    #include "pcre_printint.c"
65    #undef PCRE_INCLUDED
66  #endif  #endif
67    
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 73  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,
88        pcre_uint32 *, pcre_int32 *, pcre_uint32 *, pcre_int32 *, branch_chain *,
89        compile_data *, int *);
90    
91    
92    
93  /*************************************************  /*************************************************
94  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 88  so this number is very generous. Line 104  so this number is very generous.
104  The same workspace is used during the second, actual compile phase for  The same workspace is used during the second, actual compile phase for
105  remembering forward references to groups so that they can be filled in at the  remembering forward references to groups so that they can be filled in at the
106  end. Each entry in this list occupies LINK_SIZE bytes, so even when LINK_SIZE  end. Each entry in this list occupies LINK_SIZE bytes, so even when LINK_SIZE
107  is 4 there is plenty of room. */  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"
483    /* 55 */    /* 55 */
484    "repeating a DEFINE group is not allowed\0"    "repeating a DEFINE group is not allowed\0"  /** DEAD **/
485    "inconsistent NEWLINE options\0"    "inconsistent NEWLINE options\0"
486    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"
487    "a numbered reference must not be zero\0"    "a numbered reference must not be zero\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"
501      /* 70 */
502      "internal error: unknown opcode in find_fixedlength()\0"
503      "\\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    ;    ;
512    
513  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 427  For convenience, we use the same bit def Line 526  For convenience, we use the same bit def
526    
527  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
528    
529    /* Using a simple comparison for decimal numbers rather than a memory read
530    is much faster, and the resulting code is simpler (the compiler turns it
531    into a subtraction and unsigned comparison). */
532    
533    #define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9)
534    
535  #ifndef EBCDIC  #ifndef EBCDIC
536    
537  /* This is the "normal" case, for ASCII systems, and EBCDIC systems running in  /* This is the "normal" case, for ASCII systems, and EBCDIC systems running in
538  UTF-8 mode. */  UTF-8 mode. */
539    
540  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
541    {    {
542    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
543    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */
# Line 471  static const unsigned char digitab[] = Line 576  static const unsigned char digitab[] =
576    
577  /* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */  /* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */
578    
579  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
580    {    {
581    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
582    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */
# Line 506  static const unsigned char digitab[] = Line 611  static const unsigned char digitab[] =
611    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */
612    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */
613    
614  static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */  static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */
615    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */
616    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */
617    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */
# Line 542  static const unsigned char ebcdic_charta Line 647  static const unsigned char ebcdic_charta
647  #endif  #endif
648    
649    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,  
     int *, int *, branch_chain *, compile_data *, int *);  
   
   
650    
651  /*************************************************  /*************************************************
652  *            Find an error text                  *  *            Find an error text                  *
# Line 569  find_error_text(int n) Line 667  find_error_text(int n)
667  const char *s = error_texts;  const char *s = error_texts;
668  for (; n > 0; n--)  for (; n > 0; n--)
669    {    {
670    while (*s++ != 0) {};    while (*s++ != CHAR_NULL) {};
671    if (*s == 0) return "Error text not found (please report)";    if (*s == CHAR_NULL) return "Error text not found (please report)";
672    }    }
673  return s;  return s;
674  }  }
675    
676    
677  /*************************************************  /*************************************************
678    *           Expand the workspace                 *
679    *************************************************/
680    
681    /* This function is called during the second compiling phase, if the number of
682    forward references fills the existing workspace, which is originally a block on
683    the stack. A larger block is obtained from malloc() unless the ultimate limit
684    has been reached or the increase will be rather small.
685    
686    Argument: pointer to the compile data block
687    Returns:  0 if all went well, else an error number
688    */
689    
690    static int
691    expand_workspace(compile_data *cd)
692    {
693    pcre_uchar *newspace;
694    int newsize = cd->workspace_size * 2;
695    
696    if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;
697    if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX ||
698        newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)
699     return ERR72;
700    
701    newspace = (PUBL(malloc))(IN_UCHARS(newsize));
702    if (newspace == NULL) return ERR21;
703    memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar));
704    cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace);
705    if (cd->workspace_size > COMPILE_WORK_SIZE)
706      (PUBL(free))((void *)cd->start_workspace);
707    cd->start_workspace = newspace;
708    cd->workspace_size = newsize;
709    return 0;
710    }
711    
712    
713    
714    /*************************************************
715    *            Check for counted repeat            *
716    *************************************************/
717    
718    /* This function is called when a '{' is encountered in a place where it might
719    start a quantifier. It looks ahead to see if it really is a quantifier or not.
720    It is only a quantifier if it is one of the forms {ddd} {ddd,} or {ddd,ddd}
721    where the ddds are digits.
722    
723    Arguments:
724      p         pointer to the first char after '{'
725    
726    Returns:    TRUE or FALSE
727    */
728    
729    static BOOL
730    is_counted_repeat(const pcre_uchar *p)
731    {
732    if (!IS_DIGIT(*p)) return FALSE;
733    p++;
734    while (IS_DIGIT(*p)) p++;
735    if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
736    
737    if (*p++ != CHAR_COMMA) return FALSE;
738    if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
739    
740    if (!IS_DIGIT(*p)) return FALSE;
741    p++;
742    while (IS_DIGIT(*p)) p++;
743    
744    return (*p == CHAR_RIGHT_CURLY_BRACKET);
745    }
746    
747    
748    
749    /*************************************************
750  *            Handle escapes                      *  *            Handle escapes                      *
751  *************************************************/  *************************************************/
752    
753  /* This function is called when a \ has been encountered. It either returns a  /* This function is called when a \ has been encountered. It either returns a
754  positive value for a simple escape such as \n, or a negative value which  positive value for a simple escape such as \n, or 0 for a data character
755  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group n is returned as
756  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  negative n. When UTF-8 is enabled, a positive value greater than 255 may
757  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
758  ptr is pointing at the \. On exit, it is on the final character of the escape  On entry,ptr is pointing at the \. On exit, it is on the final character of the
759  sequence.  escape sequence.
760    
761  Arguments:  Arguments:
762    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
763      chptr          points to the data character
764    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
765    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
766    options        the options bits    options        the options bits
767    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
768    
769  Returns:         zero or positive => a data character  Returns:         zero => a data character
770                   negative => a special escape sequence                   positive => a special escape sequence
771                     negative => a back reference
772                   on error, errorcodeptr is set                   on error, errorcodeptr is set
773  */  */
774    
775  static int  static int
776  check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
777    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
778  {  {
779  BOOL utf8 = (options & PCRE_UTF8) != 0;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
780  const uschar *ptr = *ptrptr + 1;  BOOL utf = (options & PCRE_UTF8) != 0;
781  int c, i;  const pcre_uchar *ptr = *ptrptr + 1;
782    pcre_uint32 c;
783    int escape = 0;
784    int i;
785    
786  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
787  ptr--;                            /* Set pointer back to the last byte */  ptr--;                            /* Set pointer back to the last byte */
788    
789  /* If backslash is at the end of the pattern, it's an error. */  /* If backslash is at the end of the pattern, it's an error. */
790    
791  if (c == 0) *errorcodeptr = ERR1;  if (c == CHAR_NULL) *errorcodeptr = ERR1;
792    
793  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
794  in a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
795  Otherwise further processing may be required. */  Otherwise further processing may be required. */
796    
797  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
798  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */  /* Not alphanumeric */
799  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if (c < CHAR_0 || c > CHAR_z) {}
800    else if ((i = escapes[c - CHAR_0]) != 0) { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
801    
802  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
803  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */  /* Not alphanumeric */
804  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
805    else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
806  #endif  #endif
807    
808  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
809    
810  else  else
811    {    {
812    const uschar *oldptr;    const pcre_uchar *oldptr;
813    BOOL braced, negated;    BOOL braced, negated, overflow;
814      int s;
815    
816    switch (c)    switch (c)
817      {      {
# Line 642  else Line 820  else
820    
821      case CHAR_l:      case CHAR_l:
822      case CHAR_L:      case CHAR_L:
823        *errorcodeptr = ERR37;
824        break;
825    
826      case CHAR_u:      case CHAR_u:
827        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
828          {
829          /* In JavaScript, \u must be followed by four hexadecimal numbers.
830          Otherwise it is a lowercase u letter. */
831          if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
832            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
833            && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
834            && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
835            {
836            c = 0;
837            for (i = 0; i < 4; ++i)
838              {
839              register pcre_uint32 cc = *(++ptr);
840    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
841              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
842              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
843    #else           /* EBCDIC coding */
844              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
845              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
846    #endif
847              }
848    
849    #if defined COMPILE_PCRE8
850            if (c > (utf ? 0x10ffff : 0xff))
851    #elif defined COMPILE_PCRE16
852            if (c > (utf ? 0x10ffff : 0xffff))
853    #elif defined COMPILE_PCRE32
854            if (utf && c > 0x10ffff)
855    #endif
856              {
857              *errorcodeptr = ERR76;
858              }
859            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
860            }
861          }
862        else
863          *errorcodeptr = ERR37;
864        break;
865    
866      case CHAR_U:      case CHAR_U:
867      *errorcodeptr = ERR37;      /* In JavaScript, \U is an uppercase U letter. */
868        if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37;
869      break;      break;
870    
871      /* \g must be followed by one of a number of specific things:      /* In a character class, \g is just a literal "g". Outside a character
872        class, \g must be followed by one of a number of specific things:
873    
874      (1) A number, either plain or braced. If positive, it is an absolute      (1) A number, either plain or braced. If positive, it is an absolute
875      backreference. If negative, it is a relative backreference. This is a Perl      backreference. If negative, it is a relative backreference. This is a Perl
# Line 661  else Line 883  else
883      (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
884      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
885      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
886      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
887    
888      case CHAR_g:      case CHAR_g:
889        if (isclass) break;
890      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
891        {        {
892        c = -ESC_g;        escape = ESC_g;
893        break;        break;
894        }        }
895    
# Line 674  else Line 897  else
897    
898      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
899        {        {
900        const uschar *p;        const pcre_uchar *p;
901        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++)
902          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
903        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET)
904          {          {
905          c = -ESC_k;          escape = ESC_k;
906          break;          break;
907          }          }
908        braced = TRUE;        braced = TRUE;
# Line 694  else Line 917  else
917        }        }
918      else negated = FALSE;      else negated = FALSE;
919    
920      c = 0;      /* The integer range is limited by the machine's int representation. */
921      while ((digitab[ptr[1]] & ctype_digit) != 0)      s = 0;
922        c = c * 10 + *(++ptr) - CHAR_0;      overflow = FALSE;
923        while (IS_DIGIT(ptr[1]))
924      if (c < 0)   /* Integer overflow */        {
925          if (s > INT_MAX / 10 - 1) /* Integer overflow */
926            {
927            overflow = TRUE;
928            break;
929            }
930          s = s * 10 + (int)(*(++ptr) - CHAR_0);
931          }
932        if (overflow) /* Integer overflow */
933        {        {
934          while (IS_DIGIT(ptr[1]))
935            ptr++;
936        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
937        break;        break;
938        }        }
# Line 710  else Line 943  else
943        break;        break;
944        }        }
945    
946      if (c == 0)      if (s == 0)
947        {        {
948        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
949        break;        break;
# Line 718  else Line 951  else
951    
952      if (negated)      if (negated)
953        {        {
954        if (c > bracount)        if (s > bracount)
955          {          {
956          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
957          break;          break;
958          }          }
959        c = bracount - (c - 1);        s = bracount - (s - 1);
960        }        }
961    
962      c = -(ESC_REF + c);      escape = -s;
963      break;      break;
964    
965      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 747  else Line 980  else
980      if (!isclass)      if (!isclass)
981        {        {
982        oldptr = ptr;        oldptr = ptr;
983        c -= CHAR_0;        /* The integer range is limited by the machine's int representation. */
984        while ((digitab[ptr[1]] & ctype_digit) != 0)        s = (int)(c -CHAR_0);
985          c = c * 10 + *(++ptr) - CHAR_0;        overflow = FALSE;
986        if (c < 0)    /* Integer overflow */        while (IS_DIGIT(ptr[1]))
987            {
988            if (s > INT_MAX / 10 - 1) /* Integer overflow */
989              {
990              overflow = TRUE;
991              break;
992              }
993            s = s * 10 + (int)(*(++ptr) - CHAR_0);
994            }
995          if (overflow) /* Integer overflow */
996          {          {
997            while (IS_DIGIT(ptr[1]))
998              ptr++;
999          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1000          break;          break;
1001          }          }
1002        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1003          {          {
1004          c = -(ESC_REF + c);          escape = -s;
1005          break;          break;
1006          }          }
1007        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 777  else Line 1021  else
1021      /* \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
1022      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
1023      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
1024      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,
1025      than 3 octal digits. */      but no more than 3 octal digits. */
1026    
1027      case CHAR_0:      case CHAR_0:
1028      c -= CHAR_0;      c -= CHAR_0;
1029      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1030          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1031      if (!utf8 && c > 255) *errorcodeptr = ERR51;  #ifdef COMPILE_PCRE8
1032        if (!utf && c > 0xff) *errorcodeptr = ERR51;
1033    #endif
1034      break;      break;
1035    
1036      /* \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
1037      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.
1038      treated as a data character. */      If not, { is treated as a data character. */
1039    
1040      case CHAR_x:      case CHAR_x:
1041        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1042          {
1043          /* In JavaScript, \x must be followed by two hexadecimal numbers.
1044          Otherwise it is a lowercase x letter. */
1045          if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1046            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1047            {
1048            c = 0;
1049            for (i = 0; i < 2; ++i)
1050              {
1051              register pcre_uint32 cc = *(++ptr);
1052    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1053              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1054              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
1055    #else           /* EBCDIC coding */
1056              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
1057              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1058    #endif
1059              }
1060            }
1061          break;
1062          }
1063    
1064      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1065        {        {
1066        const uschar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
       int count = 0;  
1067    
1068        c = 0;        c = 0;
1069        while ((digitab[*pt] & ctype_xdigit) != 0)        overflow = FALSE;
1070          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1071          {          {
1072          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1073          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1074          count++;  
1075    #ifdef COMPILE_PCRE32
1076            if (c >= 0x10000000l) { overflow = TRUE; break; }
1077    #endif
1078    
1079  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1080          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
# Line 811  else Line 1083  else
1083          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 */
1084          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1085  #endif  #endif
1086    
1087    #if defined COMPILE_PCRE8
1088            if (c > (utf ? 0x10ffff : 0xff)) { overflow = TRUE; break; }
1089    #elif defined COMPILE_PCRE16
1090            if (c > (utf ? 0x10ffff : 0xffff)) { overflow = TRUE; break; }
1091    #elif defined COMPILE_PCRE32
1092            if (utf && c > 0x10ffff) { overflow = TRUE; break; }
1093    #endif
1094            }
1095    
1096          if (overflow)
1097            {
1098            while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1099            *errorcodeptr = ERR34;
1100          }          }
1101    
1102        if (*pt == CHAR_RIGHT_CURLY_BRACKET)        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
1103          {          {
1104          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1105          ptr = pt;          ptr = pt;
1106          break;          break;
1107          }          }
# Line 827  else Line 1113  else
1113      /* Read just a single-byte hex-defined char */      /* Read just a single-byte hex-defined char */
1114    
1115      c = 0;      c = 0;
1116      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1117        {        {
1118        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1119        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1120  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1121        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 848  else Line 1134  else
1134    
1135      case CHAR_c:      case CHAR_c:
1136      c = *(++ptr);      c = *(++ptr);
1137      if (c == 0)      if (c == CHAR_NULL)
1138        {        {
1139        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
1140        break;        break;
# Line 885  else Line 1171  else
1171    }    }
1172    
1173  /* Perl supports \N{name} for character names, as well as plain \N for "not  /* Perl supports \N{name} for character names, as well as plain \N for "not
1174  newline". PCRE does not support \N{name}. */  newline". PCRE does not support \N{name}. However, it does support
1175    quantification such as \N{2,3}. */
1176    
1177  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET)  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1178         !is_counted_repeat(ptr+2))
1179    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1180    
1181  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1182    
1183  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1184    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1185    
1186  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1187    
1188  *ptrptr = ptr;  *ptrptr = ptr;
1189  return c;  *chptr = c;
1190    return escape;
1191  }  }
1192    
   
   
1193  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1194  /*************************************************  /*************************************************
1195  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 916  escape sequence. Line 1203  escape sequence.
1203  Argument:  Argument:
1204    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1205    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
1206    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
1207      pdataptr       points to an unsigned int that is set to the detailed property value
1208    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1209    
1210  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
1211  */  */
1212    
1213  static int  static BOOL
1214  get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1215      unsigned int *pdataptr, int *errorcodeptr)
1216  {  {
1217  int c, i, bot, top;  pcre_uchar c;
1218  const uschar *ptr = *ptrptr;  int i, bot, top;
1219  char name[32];  const pcre_uchar *ptr = *ptrptr;
1220    pcre_uchar name[32];
1221    
1222  c = *(++ptr);  c = *(++ptr);
1223  if (c == 0) goto ERROR_RETURN;  if (c == CHAR_NULL) goto ERROR_RETURN;
1224    
1225  *negptr = FALSE;  *negptr = FALSE;
1226    
# Line 944  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1234  if (c == CHAR_LEFT_CURLY_BRACKET)
1234      *negptr = TRUE;      *negptr = TRUE;
1235      ptr++;      ptr++;
1236      }      }
1237    for (i = 0; i < (int)sizeof(name) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1238      {      {
1239      c = *(++ptr);      c = *(++ptr);
1240      if (c == 0) goto ERROR_RETURN;      if (c == CHAR_NULL) goto ERROR_RETURN;
1241      if (c == CHAR_RIGHT_CURLY_BRACKET) break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
1242      name[i] = c;      name[i] = c;
1243      }      }
# Line 968  else Line 1258  else
1258  /* Search for a recognized property name using binary chop */  /* Search for a recognized property name using binary chop */
1259    
1260  bot = 0;  bot = 0;
1261  top = _pcre_utt_size;  top = PRIV(utt_size);
1262    
1263  while (bot < top)  while (bot < top)
1264    {    {
1265      int r;
1266    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1267    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1268    if (c == 0)    if (r == 0)
1269      {      {
1270      *dptr = _pcre_utt[i].value;      *ptypeptr = PRIV(utt)[i].type;
1271      return _pcre_utt[i].type;      *pdataptr = PRIV(utt)[i].value;
1272        return TRUE;
1273      }      }
1274    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1275    }    }
1276    
1277  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1278  *ptrptr = ptr;  *ptrptr = ptr;
1279  return -1;  return FALSE;
1280    
1281  ERROR_RETURN:  ERROR_RETURN:
1282  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1283  *ptrptr = ptr;  *ptrptr = ptr;
1284  return -1;  return FALSE;
1285  }  }
1286  #endif  #endif
1287    
# Line 997  return -1; Line 1289  return -1;
1289    
1290    
1291  /*************************************************  /*************************************************
 *            Check for counted repeat            *  
 *************************************************/  
   
 /* This function is called when a '{' is encountered in a place where it might  
 start a quantifier. It looks ahead to see if it really is a quantifier or not.  
 It is only a quantifier if it is one of the forms {ddd} {ddd,} or {ddd,ddd}  
 where the ddds are digits.  
   
 Arguments:  
   p         pointer to the first char after '{'  
   
 Returns:    TRUE or FALSE  
 */  
   
 static BOOL  
 is_counted_repeat(const uschar *p)  
 {  
 if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  
 while ((digitab[*p] & ctype_digit) != 0) p++;  
 if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  
   
 if (*p++ != CHAR_COMMA) return FALSE;  
 if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  
   
 if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  
 while ((digitab[*p] & ctype_digit) != 0) p++;  
   
 return (*p == CHAR_RIGHT_CURLY_BRACKET);  
 }  
   
   
   
 /*************************************************  
1292  *         Read repeat counts                     *  *         Read repeat counts                     *
1293  *************************************************/  *************************************************/
1294    
# Line 1048  Returns:         pointer to '}' on succe Line 1307  Returns:         pointer to '}' on succe
1307                   current ptr on error, with errorcodeptr set non-zero                   current ptr on error, with errorcodeptr set non-zero
1308  */  */
1309    
1310  static const uschar *  static const pcre_uchar *
1311  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)
1312  {  {
1313  int min = 0;  int min = 0;
1314  int max = -1;  int max = -1;
# Line 1057  int max = -1; Line 1316  int max = -1;
1316  /* 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
1317  an integer overflow. */  an integer overflow. */
1318    
1319  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1320  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1321    {    {
1322    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1072  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1331  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1331    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1332      {      {
1333      max = 0;      max = 0;
1334      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1335      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1336        {        {
1337        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1127  Arguments: Line 1386  Arguments:
1386    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1387    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1388    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1389    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1390    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1391    
1392  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1393  */  */
1394    
1395  static int  static int
1396  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,
1397    BOOL xmode, BOOL utf8, int *count)    BOOL xmode, BOOL utf, int *count)
1398  {  {
1399  uschar *ptr = *ptrptr;  pcre_uchar *ptr = *ptrptr;
1400  int start_count = *count;  int start_count = *count;
1401  int hwm_count = start_count;  int hwm_count = start_count;
1402  BOOL dup_parens = FALSE;  BOOL dup_parens = FALSE;
# Line 1173  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1432  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1432    
1433    else if (ptr[2] == CHAR_NUMBER_SIGN)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1434      {      {
1435      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;      for (ptr += 3; *ptr != CHAR_NULL; ptr++)
1436          if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1437      goto FAIL_EXIT;      goto FAIL_EXIT;
1438      }      }
1439    
# Line 1186  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1446  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1446      ptr += 2;      ptr += 2;
1447      if (ptr[1] != CHAR_QUESTION_MARK)      if (ptr[1] != CHAR_QUESTION_MARK)
1448        {        {
1449        while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;        while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1450        if (*ptr != 0) ptr++;        if (*ptr != CHAR_NULL) ptr++;
1451        }        }
1452      }      }
1453    
# Line 1203  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1463  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1463      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1464          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1465        {        {
1466        int term;        pcre_uchar term;
1467        const uschar *thisname;        const pcre_uchar *thisname;
1468        *count += 1;        *count += 1;
1469        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1470        term = *ptr++;        term = *ptr++;
1471        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1472        thisname = ptr;        thisname = ptr;
1473        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1474        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1475            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1476          return *count;          return *count;
1477        term++;        term++;
1478        }        }
# Line 1230  for (; ptr < cd->end_pattern; ptr++) Line 1490  for (; ptr < cd->end_pattern; ptr++)
1490    
1491    if (*ptr == CHAR_BACKSLASH)    if (*ptr == CHAR_BACKSLASH)
1492      {      {
1493      if (*(++ptr) == 0) goto FAIL_EXIT;      if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1494      if (*ptr == CHAR_Q) for (;;)      if (*ptr == CHAR_Q) for (;;)
1495        {        {
1496        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};        while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1497        if (*ptr == 0) goto FAIL_EXIT;        if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1498        if (*(++ptr) == CHAR_E) break;        if (*(++ptr) == CHAR_E) break;
1499        }        }
1500      continue;      continue;
# Line 1255  for (; ptr < cd->end_pattern; ptr++) Line 1515  for (; ptr < cd->end_pattern; ptr++)
1515          {          {
1516          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1517            ptr+= 2;            ptr+= 2;
1518          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1519                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1520            ptr += 4;            ptr += 4;
1521          else          else
# Line 1278  for (; ptr < cd->end_pattern; ptr++) Line 1538  for (; ptr < cd->end_pattern; ptr++)
1538    
1539      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1540        {        {
1541        if (*ptr == 0) return -1;        if (*ptr == CHAR_NULL) return -1;
1542        if (*ptr == CHAR_BACKSLASH)        if (*ptr == CHAR_BACKSLASH)
1543          {          {
1544          if (*(++ptr) == 0) goto FAIL_EXIT;          if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1545          if (*ptr == CHAR_Q) for (;;)          if (*ptr == CHAR_Q) for (;;)
1546            {            {
1547            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};            while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1548            if (*ptr == 0) goto FAIL_EXIT;            if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1549            if (*(++ptr) == CHAR_E) break;            if (*(++ptr) == CHAR_E) break;
1550            }            }
1551          continue;          continue;
# Line 1299  for (; ptr < cd->end_pattern; ptr++) Line 1559  for (; ptr < cd->end_pattern; ptr++)
1559    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1560      {      {
1561      ptr++;      ptr++;
1562      while (*ptr != 0)      while (*ptr != CHAR_NULL)
1563        {        {
1564        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1565        ptr++;        ptr++;
1566  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1567        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1568  #endif  #endif
1569        }        }
1570      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1571      continue;      continue;
1572      }      }
1573    
# Line 1315  for (; ptr < cd->end_pattern; ptr++) Line 1575  for (; ptr < cd->end_pattern; ptr++)
1575    
1576    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1577      {      {
1578      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1579      if (rc > 0) return rc;      if (rc > 0) return rc;
1580      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1581      }      }
1582    
1583    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
# Line 1361  Arguments: Line 1621  Arguments:
1621    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1622    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1623    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1624    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1625    
1626  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1627  */  */
1628    
1629  static int  static int
1630  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,
1631    BOOL utf8)    BOOL utf)
1632  {  {
1633  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1634  int count = 0;  int count = 0;
1635  int rc;  int rc;
1636    
# Line 1381  matching closing parens. That is why we Line 1641  matching closing parens. That is why we
1641    
1642  for (;;)  for (;;)
1643    {    {
1644    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1645    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == CHAR_NULL) break;
1646    }    }
1647    
1648  return rc;  return rc;
# Line 1397  return rc; Line 1657  return rc;
1657    
1658  /* This is called by several functions that scan a compiled expression looking  /* This is called by several functions that scan a compiled expression looking
1659  for a fixed first character, or an anchoring op code etc. It skips over things  for a fixed first character, or an anchoring op code etc. It skips over things
1660  that do not influence this. For some calls, a change of option is important.  that do not influence this. For some calls, it makes sense to skip negative
1661  For some calls, it makes sense to skip negative forward and all backward  forward and all backward assertions, and also the \b assertion; for others it
1662  assertions, and also the \b assertion; for others it does not.  does not.
1663    
1664  Arguments:  Arguments:
1665    code         pointer to the start of the group    code         pointer to the start of the group
   options      pointer to external options  
   optbit       the option bit whose changing is significant, or  
                  zero if none are  
1666    skipassert   TRUE if certain assertions are to be skipped    skipassert   TRUE if certain assertions are to be skipped
1667    
1668  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1669  */  */
1670    
1671  static const uschar*  static const pcre_uchar*
1672  first_significant_code(const uschar *code, int *options, int optbit,  first_significant_code(const pcre_uchar *code, BOOL skipassert)
   BOOL skipassert)  
1673  {  {
1674  for (;;)  for (;;)
1675    {    {
1676    switch ((int)*code)    switch ((int)*code)
1677      {      {
     case OP_OPT:  
     if (optbit > 0 && ((int)code[1] & optbit) != (*options & optbit))  
       *options = (int)code[1];  
     code += 2;  
     break;  
   
1678      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1679      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1680      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1681      if (!skipassert) return code;      if (!skipassert) return code;
1682      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1683      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1684      break;      break;
1685    
1686      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1444  for (;;) Line 1694  for (;;)
1694      case OP_RREF:      case OP_RREF:
1695      case OP_NRREF:      case OP_NRREF:
1696      case OP_DEF:      case OP_DEF:
1697      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1698      break;      break;
1699    
1700      default:      default:
# Line 1474  and doing the check at the end; a flag s Line 1724  and doing the check at the end; a flag s
1724    
1725  Arguments:  Arguments:
1726    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1727    options  the compiling options    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1728    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1729    cd       the "compile data" structure    cd       the "compile data" structure
1730    
1731  Returns:   the fixed length,  Returns:   the fixed length,
1732               or -1 if there is no fixed length,               or -1 if there is no fixed length,
1733               or -2 if \C was encountered               or -2 if \C was encountered (in UTF-8 mode only)
1734               or -3 if an OP_RECURSE item was encountered and atend is FALSE               or -3 if an OP_RECURSE item was encountered and atend is FALSE
1735                 or -4 if an unknown opcode was encountered (internal error)
1736  */  */
1737    
1738  static int  static int
1739  find_fixedlength(uschar *code, int options, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1740  {  {
1741  int length = -1;  int length = -1;
1742    
1743  register int branchlength = 0;  register int branchlength = 0;
1744  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1745    
1746  /* 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
1747  branch, check the length against that of the other branches. */  branch, check the length against that of the other branches. */
# Line 1498  branch, check the length against that of Line 1749  branch, check the length against that of
1749  for (;;)  for (;;)
1750    {    {
1751    int d;    int d;
1752    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1753    register int op = *cc;    register pcre_uchar op = *cc;
1754    
1755    switch (op)    switch (op)
1756      {      {
1757        /* We only need to continue for OP_CBRA (normal capturing bracket) and
1758        OP_BRA (normal non-capturing bracket) because the other variants of these
1759        opcodes are all concerned with unlimited repeated groups, which of course
1760        are not of fixed length. */
1761    
1762      case OP_CBRA:      case OP_CBRA:
1763      case OP_BRA:      case OP_BRA:
1764      case OP_ONCE:      case OP_ONCE:
1765        case OP_ONCE_NC:
1766      case OP_COND:      case OP_COND:
1767      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), options, atend, cd);      d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd);
1768      if (d < 0) return d;      if (d < 0) return d;
1769      branchlength += d;      branchlength += d;
1770      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1771      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
1772      break;      break;
1773    
1774      /* Reached end of a branch; if it's a ket it is the end of a nested      /* Reached end of a branch; if it's a ket it is the end of a nested call.
1775      call. If it's ALT it is an alternation in a nested call. If it is      If it's ALT it is an alternation in a nested call. An ACCEPT is effectively
1776      END it's the end of the outer call. All can be handled by the same code. */      an ALT. If it is END it's the end of the outer call. All can be handled by
1777        the same code. Note that we must not include the OP_KETRxxx opcodes here,
1778        because they all imply an unlimited repeat. */
1779    
1780      case OP_ALT:      case OP_ALT:
1781      case OP_KET:      case OP_KET:
     case OP_KETRMAX:  
     case OP_KETRMIN:  
1782      case OP_END:      case OP_END:
1783        case OP_ACCEPT:
1784        case OP_ASSERT_ACCEPT:
1785      if (length < 0) length = branchlength;      if (length < 0) length = branchlength;
1786        else if (length != branchlength) return -1;        else if (length != branchlength) return -1;
1787      if (*cc != OP_ALT) return length;      if (*cc != OP_ALT) return length;
# Line 1535  for (;;) Line 1795  for (;;)
1795    
1796      case OP_RECURSE:      case OP_RECURSE:
1797      if (!atend) return -3;      if (!atend) return -3;
1798      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1799      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1800      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1801      d = find_fixedlength(cs + 2, options, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1802      if (d < 0) return d;      if (d < 0) return d;
1803      branchlength += d;      branchlength += d;
1804      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1551  for (;;) Line 1811  for (;;)
1811      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1812      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1813      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1814      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1815        break;
1816    
1817      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1818    
1819      case OP_REVERSE:      case OP_MARK:
1820        case OP_PRUNE_ARG:
1821        case OP_SKIP_ARG:
1822        case OP_THEN_ARG:
1823        cc += cc[1] + PRIV(OP_lengths)[*cc];
1824        break;
1825    
1826        case OP_CALLOUT:
1827        case OP_CIRC:
1828        case OP_CIRCM:
1829        case OP_CLOSE:
1830        case OP_COMMIT:
1831      case OP_CREF:      case OP_CREF:
     case OP_NCREF:  
     case OP_RREF:  
     case OP_NRREF:  
1832      case OP_DEF:      case OP_DEF:
1833      case OP_OPT:      case OP_DOLL:
1834      case OP_CALLOUT:      case OP_DOLLM:
     case OP_SOD:  
     case OP_SOM:  
     case OP_SET_SOM:  
1835      case OP_EOD:      case OP_EOD:
1836      case OP_EODN:      case OP_EODN:
1837      case OP_CIRC:      case OP_FAIL:
1838      case OP_DOLL:      case OP_NCREF:
1839        case OP_NRREF:
1840      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1841        case OP_PRUNE:
1842        case OP_REVERSE:
1843        case OP_RREF:
1844        case OP_SET_SOM:
1845        case OP_SKIP:
1846        case OP_SOD:
1847        case OP_SOM:
1848        case OP_THEN:
1849      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1850      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1851      break;      break;
1852    
1853      /* Handle literal characters */      /* Handle literal characters */
1854    
1855      case OP_CHAR:      case OP_CHAR:
1856      case OP_CHARNC:      case OP_CHARI:
1857      case OP_NOT:      case OP_NOT:
1858        case OP_NOTI:
1859      branchlength++;      branchlength++;
1860      cc += 2;      cc += 2;
1861  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1862      if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0)      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
       cc += _pcre_utf8_table4[cc[-1] & 0x3f];  
1863  #endif  #endif
1864      break;      break;
1865    
# Line 1592  for (;;) Line 1867  for (;;)
1867      need to skip over a multibyte character in UTF8 mode.  */      need to skip over a multibyte character in UTF8 mode.  */
1868    
1869      case OP_EXACT:      case OP_EXACT:
1870      branchlength += GET2(cc,1);      case OP_EXACTI:
1871      cc += 4;      case OP_NOTEXACT:
1872  #ifdef SUPPORT_UTF8      case OP_NOTEXACTI:
1873      if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0)      branchlength += (int)GET2(cc,1);
1874        cc += _pcre_utf8_table4[cc[-1] & 0x3f];      cc += 2 + IMM2_SIZE;
1875    #ifdef SUPPORT_UTF
1876        if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1877  #endif  #endif
1878      break;      break;
1879    
1880      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1881      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1882      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1883      cc += 4;        cc += 2;
1884        cc += 1 + IMM2_SIZE + 1;
1885      break;      break;
1886    
1887      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1613  for (;;) Line 1891  for (;;)
1891      cc += 2;      cc += 2;
1892      /* Fall through */      /* Fall through */
1893    
1894        case OP_HSPACE:
1895        case OP_VSPACE:
1896        case OP_NOT_HSPACE:
1897        case OP_NOT_VSPACE:
1898      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
1899      case OP_DIGIT:      case OP_DIGIT:
1900      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
# Line 1625  for (;;) Line 1907  for (;;)
1907      cc++;      cc++;
1908      break;      break;
1909    
1910      /* The single-byte matcher isn't allowed */      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;
1911        otherwise \C is coded as OP_ALLANY. */
1912    
1913      case OP_ANYBYTE:      case OP_ANYBYTE:
1914      return -2;      return -2;
1915    
1916      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1917    
 #ifdef SUPPORT_UTF8  
     case OP_XCLASS:  
     cc += GET(cc, 1) - 33;  
     /* Fall through */  
 #endif  
   
1918      case OP_CLASS:      case OP_CLASS:
1919      case OP_NCLASS:      case OP_NCLASS:
1920      cc += 33;  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1921        case OP_XCLASS:
1922        /* The original code caused an unsigned overflow in 64 bit systems,
1923        so now we use a conditional statement. */
1924        if (op == OP_XCLASS)
1925          cc += GET(cc, 1);
1926        else
1927          cc += PRIV(OP_lengths)[OP_CLASS];
1928    #else
1929        cc += PRIV(OP_lengths)[OP_CLASS];
1930    #endif
1931    
1932      switch (*cc)      switch (*cc)
1933        {        {
1934          case OP_CRPLUS:
1935          case OP_CRMINPLUS:
1936        case OP_CRSTAR:        case OP_CRSTAR:
1937        case OP_CRMINSTAR:        case OP_CRMINSTAR:
1938        case OP_CRQUERY:        case OP_CRQUERY:
# Line 1652  for (;;) Line 1941  for (;;)
1941    
1942        case OP_CRRANGE:        case OP_CRRANGE:
1943        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1944        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1945        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1946        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1947        break;        break;
1948    
1949        default:        default:
# Line 1664  for (;;) Line 1953  for (;;)
1953    
1954      /* Anything else is variable length */      /* Anything else is variable length */
1955    
1956      default:      case OP_ANYNL:
1957        case OP_BRAMINZERO:
1958        case OP_BRAPOS:
1959        case OP_BRAPOSZERO:
1960        case OP_BRAZERO:
1961        case OP_CBRAPOS:
1962        case OP_EXTUNI:
1963        case OP_KETRMAX:
1964        case OP_KETRMIN:
1965        case OP_KETRPOS:
1966        case OP_MINPLUS:
1967        case OP_MINPLUSI:
1968        case OP_MINQUERY:
1969        case OP_MINQUERYI:
1970        case OP_MINSTAR:
1971        case OP_MINSTARI:
1972        case OP_MINUPTO:
1973        case OP_MINUPTOI:
1974        case OP_NOTMINPLUS:
1975        case OP_NOTMINPLUSI:
1976        case OP_NOTMINQUERY:
1977        case OP_NOTMINQUERYI:
1978        case OP_NOTMINSTAR:
1979        case OP_NOTMINSTARI:
1980        case OP_NOTMINUPTO:
1981        case OP_NOTMINUPTOI:
1982        case OP_NOTPLUS:
1983        case OP_NOTPLUSI:
1984        case OP_NOTPOSPLUS:
1985        case OP_NOTPOSPLUSI:
1986        case OP_NOTPOSQUERY:
1987        case OP_NOTPOSQUERYI:
1988        case OP_NOTPOSSTAR:
1989        case OP_NOTPOSSTARI:
1990        case OP_NOTPOSUPTO:
1991        case OP_NOTPOSUPTOI:
1992        case OP_NOTQUERY:
1993        case OP_NOTQUERYI:
1994        case OP_NOTSTAR:
1995        case OP_NOTSTARI:
1996        case OP_NOTUPTO:
1997        case OP_NOTUPTOI:
1998        case OP_PLUS:
1999        case OP_PLUSI:
2000        case OP_POSPLUS:
2001        case OP_POSPLUSI:
2002        case OP_POSQUERY:
2003        case OP_POSQUERYI:
2004        case OP_POSSTAR:
2005        case OP_POSSTARI:
2006        case OP_POSUPTO:
2007        case OP_POSUPTOI:
2008        case OP_QUERY:
2009        case OP_QUERYI:
2010        case OP_REF:
2011        case OP_REFI:
2012        case OP_SBRA:
2013        case OP_SBRAPOS:
2014        case OP_SCBRA:
2015        case OP_SCBRAPOS:
2016        case OP_SCOND:
2017        case OP_SKIPZERO:
2018        case OP_STAR:
2019        case OP_STARI:
2020        case OP_TYPEMINPLUS:
2021        case OP_TYPEMINQUERY:
2022        case OP_TYPEMINSTAR:
2023        case OP_TYPEMINUPTO:
2024        case OP_TYPEPLUS:
2025        case OP_TYPEPOSPLUS:
2026        case OP_TYPEPOSQUERY:
2027        case OP_TYPEPOSSTAR:
2028        case OP_TYPEPOSUPTO:
2029        case OP_TYPEQUERY:
2030        case OP_TYPESTAR:
2031        case OP_TYPEUPTO:
2032        case OP_UPTO:
2033        case OP_UPTOI:
2034      return -1;      return -1;
2035    
2036        /* Catch unrecognized opcodes so that when new ones are added they
2037        are not forgotten, as has happened in the past. */
2038    
2039        default:
2040        return -4;
2041      }      }
2042    }    }
2043  /* Control never gets here */  /* Control never gets here */
# Line 1686  length. Line 2058  length.
2058    
2059  Arguments:  Arguments:
2060    code        points to start of expression    code        points to start of expression
2061    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2062    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2063    
2064  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
2065  */  */
2066    
2067  const uschar *  const pcre_uchar *
2068  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2069  {  {
2070  for (;;)  for (;;)
2071    {    {
2072    register int c = *code;    register pcre_uchar c = *code;
2073    
2074    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2075    
2076    /* XCLASS is used for classes that cannot be represented just by a bit    /* XCLASS is used for classes that cannot be represented just by a bit
# Line 1710  for (;;) Line 2083  for (;;)
2083    
2084    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2085      {      {
2086      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2087      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2088      }      }
2089    
2090    /* Handle capturing bracket */    /* Handle capturing bracket */
2091    
2092    else if (c == OP_CBRA)    else if (c == OP_CBRA || c == OP_SCBRA ||
2093               c == OP_CBRAPOS || c == OP_SCBRAPOS)
2094      {      {
2095      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2096      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2097      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2098      }      }
2099    
2100    /* 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 1748  for (;;) Line 2122  for (;;)
2122        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2123        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2124        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2125        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2126            code += 2;
2127        break;        break;
2128    
2129        case OP_MARK:        case OP_MARK:
# Line 1758  for (;;) Line 2133  for (;;)
2133        break;        break;
2134    
2135        case OP_THEN_ARG:        case OP_THEN_ARG:
2136        code += code[1+LINK_SIZE];        code += code[1];
2137        break;        break;
2138        }        }
2139    
2140      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2141    
2142      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2143    
2144    /* 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
2145    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
2146    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2147    
2148  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2149      if (utf8) switch(c)      if (utf) switch(c)
2150        {        {
2151        case OP_CHAR:        case OP_CHAR:
2152        case OP_CHARNC:        case OP_CHARI:
2153        case OP_EXACT:        case OP_EXACT:
2154          case OP_EXACTI:
2155        case OP_UPTO:        case OP_UPTO:
2156          case OP_UPTOI:
2157        case OP_MINUPTO:        case OP_MINUPTO:
2158          case OP_MINUPTOI:
2159        case OP_POSUPTO:        case OP_POSUPTO:
2160          case OP_POSUPTOI:
2161        case OP_STAR:        case OP_STAR:
2162          case OP_STARI:
2163        case OP_MINSTAR:        case OP_MINSTAR:
2164          case OP_MINSTARI:
2165        case OP_POSSTAR:        case OP_POSSTAR:
2166          case OP_POSSTARI:
2167        case OP_PLUS:        case OP_PLUS:
2168          case OP_PLUSI:
2169        case OP_MINPLUS:        case OP_MINPLUS:
2170          case OP_MINPLUSI:
2171        case OP_POSPLUS:        case OP_POSPLUS:
2172          case OP_POSPLUSI:
2173        case OP_QUERY:        case OP_QUERY:
2174          case OP_QUERYI:
2175        case OP_MINQUERY:        case OP_MINQUERY:
2176          case OP_MINQUERYI:
2177        case OP_POSQUERY:        case OP_POSQUERY:
2178        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_POSQUERYI:
2179          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2180        break;        break;
2181        }        }
2182  #else  #else
2183      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2184  #endif  #endif
2185      }      }
2186    }    }
# Line 1809  instance of OP_RECURSE. Line 2197  instance of OP_RECURSE.
2197    
2198  Arguments:  Arguments:
2199    code        points to start of expression    code        points to start of expression
2200    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2201    
2202  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
2203  */  */
2204    
2205  static const uschar *  static const pcre_uchar *
2206  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2207  {  {
2208  for (;;)  for (;;)
2209    {    {
2210    register int c = *code;    register pcre_uchar c = *code;
2211    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2212    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2213    
# Line 1854  for (;;) Line 2242  for (;;)
2242        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2243        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2244        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2245        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2246            code += 2;
2247        break;        break;
2248    
2249        case OP_MARK:        case OP_MARK:
# Line 1864  for (;;) Line 2253  for (;;)
2253        break;        break;
2254    
2255        case OP_THEN_ARG:        case OP_THEN_ARG:
2256        code += code[1+LINK_SIZE];        code += code[1];
2257        break;        break;
2258        }        }
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_CHARNC:        case OP_CHARI:
2273          case OP_NOT:
2274          case OP_NOTI:
2275        case OP_EXACT:        case OP_EXACT:
2276          case OP_EXACTI:
2277          case OP_NOTEXACT:
2278          case OP_NOTEXACTI:
2279        case OP_UPTO:        case OP_UPTO:
2280          case OP_UPTOI:
2281          case OP_NOTUPTO:
2282          case OP_NOTUPTOI:
2283        case OP_MINUPTO:        case OP_MINUPTO:
2284          case OP_MINUPTOI:
2285          case OP_NOTMINUPTO:
2286          case OP_NOTMINUPTOI:
2287        case OP_POSUPTO:        case OP_POSUPTO:
2288          case OP_POSUPTOI:
2289          case OP_NOTPOSUPTO:
2290          case OP_NOTPOSUPTOI:
2291        case OP_STAR:        case OP_STAR:
2292          case OP_STARI:
2293          case OP_NOTSTAR:
2294          case OP_NOTSTARI:
2295        case OP_MINSTAR:        case OP_MINSTAR:
2296          case OP_MINSTARI:
2297          case OP_NOTMINSTAR:
2298          case OP_NOTMINSTARI:
2299        case OP_POSSTAR:        case OP_POSSTAR:
2300          case OP_POSSTARI:
2301          case OP_NOTPOSSTAR:
2302          case OP_NOTPOSSTARI:
2303        case OP_PLUS:        case OP_PLUS:
2304          case OP_PLUSI:
2305          case OP_NOTPLUS:
2306          case OP_NOTPLUSI:
2307        case OP_MINPLUS:        case OP_MINPLUS:
2308          case OP_MINPLUSI:
2309          case OP_NOTMINPLUS:
2310          case OP_NOTMINPLUSI:
2311        case OP_POSPLUS:        case OP_POSPLUS:
2312          case OP_POSPLUSI:
2313          case OP_NOTPOSPLUS:
2314          case OP_NOTPOSPLUSI:
2315        case OP_QUERY:        case OP_QUERY:
2316          case OP_QUERYI:
2317          case OP_NOTQUERY:
2318          case OP_NOTQUERYI:
2319        case OP_MINQUERY:        case OP_MINQUERY:
2320          case OP_MINQUERYI:
2321          case OP_NOTMINQUERY:
2322          case OP_NOTMINQUERYI:
2323        case OP_POSQUERY:        case OP_POSQUERY:
2324        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_POSQUERYI:
2325          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 1921  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    
2357  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2358  */  */
2359    
2360  static BOOL  static BOOL
2361  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2362    compile_data *cd)    BOOL utf, compile_data *cd)
2363  {  {
2364  register int c;  register pcre_uchar c;
2365  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2366       code < endcode;       code < endcode;
2367       code = first_significant_code(code + _pcre_OP_lengths[c], NULL, 0, TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2368    {    {
2369    const uschar *ccode;    const pcre_uchar *ccode;
2370    
2371    c = *code;    c = *code;
2372    
# Line 1950  for (code = first_significant_code(code Line 2380  for (code = first_significant_code(code
2380      continue;      continue;
2381      }      }
2382    
   /* Groups with zero repeats can of course be empty; skip them. */  
   
   if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO)  
     {  
     code += _pcre_OP_lengths[c];  
     do code += GET(code, 1); while (*code == OP_ALT);  
     c = *code;  
     continue;  
     }  
   
2383    /* For a recursion/subroutine call, if its end has been reached, which    /* For a recursion/subroutine call, if its end has been reached, which
2384    implies a subroutine call, we can scan it. */    implies a backward reference subroutine call, we can scan it. If it's a
2385      forward reference subroutine call, we can't. To detect forward reference
2386      we have to scan up the list that is kept in the workspace. This function is
2387      called only when doing the real compile, not during the pre-compile that
2388      measures the size of the compiled pattern. */
2389    
2390    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2391      {      {
2392      BOOL empty_branch = FALSE;      const pcre_uchar *scode;
2393      const uschar *scode = cd->start_code + GET(code, 1);      BOOL empty_branch;
2394    
2395        /* Test for forward reference */
2396    
2397        for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2398          if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2399    
2400        /* Not a forward reference, test for completed backward reference */
2401    
2402        empty_branch = FALSE;
2403        scode = cd->start_code + GET(code, 1);
2404      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2405    
2406        /* Completed backwards reference */
2407    
2408      do      do
2409        {        {
2410        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd))
2411          {          {
2412          empty_branch = TRUE;          empty_branch = TRUE;
2413          break;          break;
# Line 1978  for (code = first_significant_code(code Line 2415  for (code = first_significant_code(code
2415        scode += GET(scode, 1);        scode += GET(scode, 1);
2416        }        }
2417      while (*scode == OP_ALT);      while (*scode == OP_ALT);
2418    
2419      if (!empty_branch) return FALSE;  /* All branches are non-empty */      if (!empty_branch) return FALSE;  /* All branches are non-empty */
2420      continue;      continue;
2421      }      }
2422    
2423    /* For other groups, scan the branches. */    /* Groups with zero repeats can of course be empty; skip them. */
2424    
2425    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2426          c == OP_BRAPOSZERO)
2427      {      {
2428      BOOL empty_branch;      code += PRIV(OP_lengths)[c];
2429      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      do code += GET(code, 1); while (*code == OP_ALT);
2430        c = *code;
2431        continue;
2432        }
2433    
2434      /* If a conditional group has only one branch, there is a second, implied,    /* A nested group that is already marked as "could be empty" can just be
2435      empty branch, so just skip over the conditional, because it could be empty.    skipped. */
2436    
2437      if (c == OP_SBRA  || c == OP_SBRAPOS ||
2438          c == OP_SCBRA || c == OP_SCBRAPOS)
2439        {
2440        do code += GET(code, 1); while (*code == OP_ALT);
2441        c = *code;
2442        continue;
2443        }
2444    
2445      /* For other groups, scan the branches. */
2446    
2447      if (c == OP_BRA  || c == OP_BRAPOS ||
2448          c == OP_CBRA || c == OP_CBRAPOS ||
2449          c == OP_ONCE || c == OP_ONCE_NC ||
2450          c == OP_COND)
2451        {
2452        BOOL empty_branch;
2453        if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
2454    
2455        /* If a conditional group has only one branch, there is a second, implied,
2456        empty branch, so just skip over the conditional, because it could be empty.
2457      Otherwise, scan the individual branches of the group. */      Otherwise, scan the individual branches of the group. */
2458    
2459      if (c == OP_COND && code[GET(code, 1)] != OP_ALT)      if (c == OP_COND && code[GET(code, 1)] != OP_ALT)
# Line 2000  for (code = first_significant_code(code Line 2463  for (code = first_significant_code(code
2463        empty_branch = FALSE;        empty_branch = FALSE;
2464        do        do
2465          {          {
2466          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
2467            empty_branch = TRUE;            empty_branch = TRUE;
2468          code += GET(code, 1);          code += GET(code, 1);
2469          }          }
# Line 2018  for (code = first_significant_code(code Line 2481  for (code = first_significant_code(code
2481      {      {
2482      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2483      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2484      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2485      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"
2486      here. */      here. */
2487    
2488  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2489      case OP_XCLASS:      case OP_XCLASS:
2490      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2491      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2030  for (code = first_significant_code(code Line 2493  for (code = first_significant_code(code
2493    
2494      case OP_CLASS:      case OP_CLASS:
2495      case OP_NCLASS:      case OP_NCLASS:
2496      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2497    
2498  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2499      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2500  #endif  #endif
2501    
# Line 2071  for (code = first_significant_code(code Line 2534  for (code = first_significant_code(code
2534      case OP_ALLANY:      case OP_ALLANY:
2535      case OP_ANYBYTE:      case OP_ANYBYTE:
2536      case OP_CHAR:      case OP_CHAR:
2537      case OP_CHARNC:      case OP_CHARI:
2538      case OP_NOT:      case OP_NOT:
2539        case OP_NOTI:
2540      case OP_PLUS:      case OP_PLUS:
2541      case OP_MINPLUS:      case OP_MINPLUS:
2542      case OP_POSPLUS:      case OP_POSPLUS:
# Line 2104  for (code = first_significant_code(code Line 2568  for (code = first_significant_code(code
2568      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2569      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2570      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2571      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2572          code += 2;
2573      break;      break;
2574    
2575      /* End of branch */      /* End of branch */
# Line 2112  for (code = first_significant_code(code Line 2577  for (code = first_significant_code(code
2577      case OP_KET:      case OP_KET:
2578      case OP_KETRMAX:      case OP_KETRMAX:
2579      case OP_KETRMIN:      case OP_KETRMIN:
2580        case OP_KETRPOS:
2581      case OP_ALT:      case OP_ALT:
2582      return TRUE;      return TRUE;
2583    
2584      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2585      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2586    
2587  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2588      case OP_STAR:      case OP_STAR:
2589        case OP_STARI:
2590      case OP_MINSTAR:      case OP_MINSTAR:
2591        case OP_MINSTARI:
2592      case OP_POSSTAR:      case OP_POSSTAR:
2593        case OP_POSSTARI:
2594      case OP_QUERY:      case OP_QUERY:
2595        case OP_QUERYI:
2596      case OP_MINQUERY:      case OP_MINQUERY:
2597        case OP_MINQUERYI:
2598      case OP_POSQUERY:      case OP_POSQUERY:
2599      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      case OP_POSQUERYI:
2600        if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2601      break;      break;
2602    
2603      case OP_UPTO:      case OP_UPTO:
2604        case OP_UPTOI:
2605      case OP_MINUPTO:      case OP_MINUPTO:
2606        case OP_MINUPTOI:
2607      case OP_POSUPTO:      case OP_POSUPTO:
2608      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      case OP_POSUPTOI:
2609        if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);
2610      break;      break;
2611  #endif  #endif
2612    
# Line 2145  for (code = first_significant_code(code Line 2620  for (code = first_significant_code(code
2620      break;      break;
2621    
2622      case OP_THEN_ARG:      case OP_THEN_ARG:
2623      code += code[1+LINK_SIZE];      code += code[1];
2624      break;      break;
2625    
2626      /* None of the remaining opcodes are required to match a character. */      /* None of the remaining opcodes are required to match a character. */
# Line 2168  return TRUE; Line 2643  return TRUE;
2643  the current branch of the current pattern to see if it could match the empty  the current branch of the current pattern to see if it could match the empty
2644  string. If it could, we must look outwards for branches at other levels,  string. If it could, we must look outwards for branches at other levels,
2645  stopping when we pass beyond the bracket which is the subject of the recursion.  stopping when we pass beyond the bracket which is the subject of the recursion.
2646    This function is called only during the real compile, not during the
2647    pre-compile.
2648    
2649  Arguments:  Arguments:
2650    code        points to start of the recursion    code        points to start of the recursion
2651    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2652    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2653    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2654    cd          pointers to tables etc    cd          pointers to tables etc
2655    
2656  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2657  */  */
2658    
2659  static BOOL  static BOOL
2660  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2661    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2662  {  {
2663  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2664    {    {
2665    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
2666      return FALSE;      return FALSE;
2667    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2668    }    }
# Line 2218  where Perl recognizes it as the POSIX cl Line 2695  where Perl recognizes it as the POSIX cl
2695  "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,  "l\ower". This is a lesser evil that not diagnosing bad classes when Perl does,
2696  I think.  I think.
2697    
2698    A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.
2699    It seems that the appearance of a nested POSIX class supersedes an apparent
2700    external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or
2701    a digit.
2702    
2703    In Perl, unescaped square brackets may also appear as part of class names. For
2704    example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for
2705    [:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not
2706    seem right at all. PCRE does not allow closing square brackets in POSIX class
2707    names.
2708    
2709  Arguments:  Arguments:
2710    ptr      pointer to the initial [    ptr      pointer to the initial [
2711    endptr   where to return the end pointer    endptr   where to return the end pointer
# Line 2226  Returns:   TRUE or FALSE Line 2714  Returns:   TRUE or FALSE
2714  */  */
2715    
2716  static BOOL  static BOOL
2717  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2718  {  {
2719  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2720  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2721  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != CHAR_NULL; ptr++)
2722    {    {
2723    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2724        ptr++;
2725      else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
2726      else
2727      {      {
     if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;  
2728      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2729        {        {
2730        *endptr = ptr;        *endptr = ptr;
2731        return TRUE;        return TRUE;
2732        }        }
2733        if (*ptr == CHAR_LEFT_SQUARE_BRACKET &&
2734             (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2735              ptr[1] == CHAR_EQUALS_SIGN) &&
2736            check_posix_syntax(ptr, endptr))
2737          return FALSE;
2738      }      }
2739    }    }
2740  return FALSE;  return FALSE;
# Line 2263  Returns:     a value representing the na Line 2758  Returns:     a value representing the na
2758  */  */
2759    
2760  static int  static int
2761  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2762  {  {
2763  const char *pn = posix_names;  const char *pn = posix_names;
2764  register int yield = 0;  register int yield = 0;
2765  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2766    {    {
2767    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2768      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2769    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2770    yield++;    yield++;
2771    }    }
# Line 2302  value in the reference (which is a group Line 2797  value in the reference (which is a group
2797  Arguments:  Arguments:
2798    group      points to the start of the group    group      points to the start of the group
2799    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2800    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2801    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2802    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
2803    
# Line 2310  Returns:     nothing Line 2805  Returns:     nothing
2805  */  */
2806    
2807  static void  static void
2808  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2809    uschar *save_hwm)    pcre_uchar *save_hwm)
2810  {  {
2811  uschar *ptr = group;  pcre_uchar *ptr = group;
2812    
2813  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2814    {    {
2815    int offset;    int offset;
2816    uschar *hc;    pcre_uchar *hc;
2817    
2818    /* 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
2819    reference. */    reference. */
2820    
2821    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2822      {      {
2823      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2824      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2825        {        {
2826        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2338  while ((ptr = (uschar *)find_recurse(ptr Line 2833  while ((ptr = (uschar *)find_recurse(ptr
2833    
2834    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2835      {      {
2836      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2837      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2838      }      }
2839    
# Line 2363  Arguments: Line 2858  Arguments:
2858  Returns:         new code pointer  Returns:         new code pointer
2859  */  */
2860    
2861  static uschar *  static pcre_uchar *
2862  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2863  {  {
2864  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2865  *code++ = 255;  *code++ = 255;
2866  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2867  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2868  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2869  }  }
2870    
2871    
# Line 2392  Returns:             nothing Line 2887  Returns:             nothing
2887  */  */
2888    
2889  static void  static void
2890  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2891  {  {
2892  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2893  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2406  PUT(previous_callout, 2 + LINK_SIZE, len Line 2901  PUT(previous_callout, 2 + LINK_SIZE, len
2901  *************************************************/  *************************************************/
2902    
2903  /* 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
2904  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
2905  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
2906  start address.  start address. A character with multiple other cases is returned on its own
2907    with a special return value.
2908    
2909  Arguments:  Arguments:
2910    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2416  Arguments: Line 2912  Arguments:
2912    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2913    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2914    
2915  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2916                   0 when a range is returned
2917                  >0 the CASESET offset for char with multiple other cases
2918                    in this case, ocptr contains the original
2919  */  */
2920    
2921  static BOOL  static int
2922  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2923    unsigned int *odptr)    pcre_uint32 *odptr)
2924  {  {
2925  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2926    unsigned int co;
2927    
2928    /* Find the first character that has an other case. If it has multiple other
2929    cases, return its case offset value. */
2930    
2931  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2932    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2933      if ((co = UCD_CASESET(c)) != 0)
2934        {
2935        *ocptr = c++;   /* Character that has the set */
2936        *cptr = c;      /* Rest of input range */
2937        return (int)co;
2938        }
2939      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2940      }
2941    
2942  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2943    
2944  *ocptr = othercase;  *ocptr = othercase;
2945  next = othercase + 1;  next = othercase + 1;
# Line 2439  for (++c; c <= d; c++) Line 2950  for (++c; c <= d; c++)
2950    next++;    next++;
2951    }    }
2952    
2953  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2954  *cptr = c;  *cptr = c;             /* Rest of input range */
2955    return 0;
 return TRUE;  
2956  }  }
2957    
2958    
# Line 2464  Returns:       TRUE if auto-possessifyin Line 2974  Returns:       TRUE if auto-possessifyin
2974  */  */
2975    
2976  static BOOL  static BOOL
2977  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)
2978  {  {
2979    #ifdef SUPPORT_UCP
2980    const pcre_uint32 *p;
2981    #endif
2982    
2983  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2984    
2985  switch(ptype)  switch(ptype)
2986    {    {
2987    case PT_LAMP:    case PT_LAMP:
# Line 2475  switch(ptype) Line 2990  switch(ptype)
2990            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
2991    
2992    case PT_GC:    case PT_GC:
2993    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2994    
2995    case PT_PC:    case PT_PC:
2996    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2486  switch(ptype) Line 3001  switch(ptype)
3001    /* These are specials */    /* These are specials */
3002    
3003    case PT_ALNUM:    case PT_ALNUM:
3004    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3005            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
3006    
3007    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
3008    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3009            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
3010            == negated;            == negated;
3011    
3012    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
3013    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3014            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
3015            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
3016            == negated;            == negated;
3017    
3018    case PT_WORD:    case PT_WORD:
3019    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3020            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3021            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3022    
3023    #ifdef SUPPORT_UCP
3024      case PT_CLIST:
3025      p = PRIV(ucd_caseless_sets) + prop->caseset;
3026      for (;;)
3027        {
3028        if (c < *p) return !negated;
3029        if (c == *p++) return negated;
3030        }
3031      break;  /* Control never reaches here */
3032    #endif
3033    }    }
3034    
3035  return FALSE;  return FALSE;
3036  }  }
3037  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2521  sense to automatically possessify the re Line 3048  sense to automatically possessify the re
3048    
3049  Arguments:  Arguments:
3050    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3051    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3052    ptr           next character in pattern    ptr           next character in pattern
3053    options       options bits    options       options bits
3054    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2530  Returns:        TRUE if possessifying is Line 3057  Returns:        TRUE if possessifying is
3057  */  */
3058    
3059  static BOOL  static BOOL
3060  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3061    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3062  {  {
3063  int c, next;  pcre_uint32 c = NOTACHAR;
3064  int op_code = *previous++;  pcre_uint32 next;
3065    int escape;
3066    pcre_uchar op_code = *previous++;
3067    
3068  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3069    
# Line 2542  if ((options & PCRE_EXTENDED) != 0) Line 3071  if ((options & PCRE_EXTENDED) != 0)
3071    {    {
3072    for (;;)    for (;;)
3073      {      {
3074      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3075      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3076        {        {
3077        ptr++;        ptr++;
3078        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3079          {          {
3080          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3081          ptr++;          ptr++;
3082  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3083          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3084  #endif  #endif
3085          }          }
3086        }        }
# Line 2565  value is a character, a negative value i Line 3094  value is a character, a negative value i
3094  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3095    {    {
3096    int temperrorcode = 0;    int temperrorcode = 0;
3097    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3098    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3099    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3100    }    }
3101    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
3102    {    {
3103  #ifdef SUPPORT_UTF8    escape = 0;
3104    if (utf8) { GETCHARINC(next, ptr); } else  #ifdef SUPPORT_UTF
3105      if (utf) { GETCHARINC(next, ptr); } else
3106  #endif  #endif
3107    next = *ptr++;    next = *ptr++;
3108    }    }
   
3109  else return FALSE;  else return FALSE;
3110    
3111  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2586  if ((options & PCRE_EXTENDED) != 0) Line 3114  if ((options & PCRE_EXTENDED) != 0)
3114    {    {
3115    for (;;)    for (;;)
3116      {      {
3117      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3118      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3119        {        {
3120        ptr++;        ptr++;
3121        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3122          {          {
3123          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3124          ptr++;          ptr++;
3125  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3126          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3127  #endif  #endif
3128          }          }
3129        }        }
# Line 2606  if ((options & PCRE_EXTENDED) != 0) Line 3134  if ((options & PCRE_EXTENDED) != 0)
3134  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3135    
3136  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3137    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)
3138      return FALSE;      return FALSE;
3139    
3140  /* 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. */  
3141    
3142  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3143        op_code == OP_NOT || op_code == OP_NOTI)
3144    {    {
3145    case OP_CHAR:  #ifdef SUPPORT_UTF
 #ifdef SUPPORT_UTF8  
3146    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3147  #else  #else
3148    c = *previous;    c = *previous;
3149  #endif  #endif
3150    return c != next;    }
3151    
3152    /* For CHARNC (caseless character) we must check the other case. If we have  /* Now compare the next item with the previous opcode. First, handle cases when
3153    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3154    
3155    case OP_CHARNC:  if (escape == 0)
3156  #ifdef SUPPORT_UTF8    {
3157    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3158  #else    case, which maps to the special PT_CLIST property. Check this first. */
3159    c = *previous;  
3160    #ifdef SUPPORT_UCP
3161      if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3162        {
3163        unsigned int ocs = UCD_CASESET(next);
3164        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3165        }
3166  #endif  #endif
3167    if (c == next) return FALSE;  
3168  #ifdef SUPPORT_UTF8    switch(op_code)
   if (utf8)  
3169      {      {
3170      unsigned int othercase;      case OP_CHAR:
3171      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3172    
3173        /* For CHARI (caseless character) we must check the other case. If we have
3174        Unicode property support, we can use it to test the other case of
3175        high-valued characters. We know that next can have only one other case,
3176        because multi-other-case characters are dealt with above. */
3177    
3178        case OP_CHARI:
3179        if (c == next) return FALSE;
3180    #ifdef SUPPORT_UTF
3181        if (utf)
3182          {
3183          pcre_uint32 othercase;
3184          if (next < 128) othercase = cd->fcc[next]; else
3185  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3186      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3187  #else  #else
3188      othercase = NOTACHAR;        othercase = NOTACHAR;
3189  #endif  #endif
3190      return (unsigned int)c != othercase;        return c != othercase;
3191      }        }
3192    else      else
3193  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3194    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3195    
3196    /* For OP_NOT, its data is always a single-byte character. */      case OP_NOT:
3197        return c == next;
3198    case OP_NOT:  
3199    if ((c = *previous) == next) return TRUE;      case OP_NOTI:
3200    if ((options & PCRE_CASELESS) == 0) return FALSE;      if (c == next) return TRUE;
3201  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3202    if (utf8)      if (utf)
3203      {        {
3204      unsigned int othercase;        pcre_uint32 othercase;
3205      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3206  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3207      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE(next);
3208  #else  #else
3209      othercase = NOTACHAR;        othercase = NOTACHAR;
3210  #endif  #endif
3211      return (unsigned int)c == othercase;        return c == othercase;
3212      }        }
3213    else      else
3214  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3215    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3216    
3217    /* 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.
3218    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. */
3219    
3220    case OP_DIGIT:      case OP_DIGIT:
3221    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3222    
3223    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3224    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3225    
3226    case OP_WHITESPACE:      case OP_WHITESPACE:
3227    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3228    
3229    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3230    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3231    
3232    case OP_WORDCHAR:      case OP_WORDCHAR:
3233    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3234    
3235    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3236    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3237    
3238    case OP_HSPACE:      case OP_HSPACE:
3239    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3240    switch(next)      switch(next)
3241      {        {
3242      case 0x09:        HSPACE_CASES:
3243      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;  
     }  
3244    
3245    case OP_ANYNL:        default:
3246    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3247    case OP_NOT_VSPACE:        }
3248    switch(next)  
3249      {      case OP_ANYNL:
3250      case 0x0a:      case OP_VSPACE:
3251      case 0x0b:      case OP_NOT_VSPACE:
3252      case 0x0c:      switch(next)
3253      case 0x0d:        {
3254      case 0x85:        VSPACE_CASES:
3255      case 0x2028:        return op_code == OP_NOT_VSPACE;
3256      case 0x2029:  
3257      return op_code == OP_NOT_VSPACE;        default:
3258      default:        return op_code != OP_NOT_VSPACE;
3259      return op_code != OP_NOT_VSPACE;        }
     }  
3260    
3261  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3262    case OP_PROP:      case OP_PROP:
3263    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3264    
3265    case OP_NOTPROP:      case OP_NOTPROP:
3266    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3267  #endif  #endif
3268    
3269    default:      default:
3270    return FALSE;      return FALSE;
3271        }
3272    }    }
3273    
   
3274  /* 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
3275  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
3276  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 2758  replaced by OP_PROP codes when PCRE_UCP Line 3280  replaced by OP_PROP codes when PCRE_UCP
3280  switch(op_code)  switch(op_code)
3281    {    {
3282    case OP_CHAR:    case OP_CHAR:
3283    case OP_CHARNC:    case OP_CHARI:
3284  #ifdef SUPPORT_UTF8    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3285      {      {
3286      case ESC_d:      case ESC_d:
3287      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3288    
3289      case ESC_D:      case ESC_D:
3290      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3291    
3292      case ESC_s:      case ESC_s:
3293      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3294    
3295      case ESC_S:      case ESC_S:
3296      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3297    
3298      case ESC_w:      case ESC_w:
3299      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3300    
3301      case ESC_W:      case ESC_W:
3302      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3303    
3304      case ESC_h:      case ESC_h:
3305      case ESC_H:      case ESC_H:
3306      switch(c)      switch(c)
3307        {        {
3308        case 0x09:        HSPACE_CASES:
3309        case 0x20:        return escape != ESC_h;
3310        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;  
3311        default:        default:
3312        return -next == ESC_h;        return escape == ESC_h;
3313        }        }
3314    
3315      case ESC_v:      case ESC_v:
3316      case ESC_V:      case ESC_V:
3317      switch(c)      switch(c)
3318        {        {
3319        case 0x0a:        VSPACE_CASES:
3320        case 0x0b:        return escape != ESC_v;
3321        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3322        default:        default:
3323        return -next == ESC_v;        return escape == ESC_v;
3324        }        }
3325    
3326      /* 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
3327      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3328      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3329    
3330  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3331      case ESC_du:      case ESC_du:
# Line 2841  switch(op_code) Line 3336  switch(op_code)
3336      case ESC_SU:      case ESC_SU:
3337        {        {
3338        int temperrorcode = 0;        int temperrorcode = 0;
3339        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3340        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3341        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3342        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3343        }        }
# Line 2851  switch(op_code) Line 3346  switch(op_code)
3346      case ESC_p:      case ESC_p:
3347      case ESC_P:      case ESC_P:
3348        {        {
3349        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3350          int errorcodeptr;
3351        BOOL negated;        BOOL negated;
3352    
3353        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3354        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3355        if (ptype < 0) return FALSE;          return FALSE;
3356        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3357    
3358        /* 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 2864  switch(op_code) Line 3360  switch(op_code)
3360        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. */
3361    
3362        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3363          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)
3364            return FALSE;            return FALSE;
3365    
3366        /* Do the property check. */        /* Do the property check. */
3367    
3368        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3369        }        }
3370  #endif  #endif
3371    
# Line 2884  switch(op_code) Line 3380  switch(op_code)
3380    these op-codes are never generated.) */    these op-codes are never generated.) */
3381    
3382    case OP_DIGIT:    case OP_DIGIT:
3383    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3384           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3385    
3386    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3387    return next == -ESC_d;    return escape == ESC_d;
3388    
3389    case OP_WHITESPACE:    case OP_WHITESPACE:
3390    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3391    
3392    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3393    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3394    
3395    case OP_HSPACE:    case OP_HSPACE:
3396    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3397           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3398    
3399    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3400    return next == -ESC_h;    return escape == ESC_h;
3401    
3402    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3403    case OP_ANYNL:    case OP_ANYNL:
3404    case OP_VSPACE:    case OP_VSPACE:
3405    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3406    
3407    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3408    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3409    
3410    case OP_WORDCHAR:    case OP_WORDCHAR:
3411    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3412           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3413    
3414    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3415    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3416    
3417    default:    default:
3418    return FALSE;    return FALSE;
# Line 2928  switch(op_code) Line 3424  switch(op_code)
3424    
3425    
3426  /*************************************************  /*************************************************
3427    *        Add a character or range to a class     *
3428    *************************************************/
3429    
3430    /* This function packages up the logic of adding a character or range of
3431    characters to a class. The character values in the arguments will be within the
3432    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3433    mutually recursive with the function immediately below.
3434    
3435    Arguments:
3436      classbits     the bit map for characters < 256
3437      uchardptr     points to the pointer for extra data
3438      options       the options word
3439      cd            contains pointers to tables etc.
3440      start         start of range character
3441      end           end of range character
3442    
3443    Returns:        the number of < 256 characters added
3444                    the pointer to extra data is updated
3445    */
3446    
3447    static int
3448    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3449      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3450    {
3451    pcre_uint32 c;
3452    int n8 = 0;
3453    
3454    /* If caseless matching is required, scan the range and process alternate
3455    cases. In Unicode, there are 8-bit characters that have alternate cases that
3456    are greater than 255 and vice-versa. Sometimes we can just extend the original
3457    range. */
3458    
3459    if ((options & PCRE_CASELESS) != 0)
3460      {
3461    #ifdef SUPPORT_UCP
3462      if ((options & PCRE_UTF8) != 0)
3463        {
3464        int rc;
3465        pcre_uint32 oc, od;
3466    
3467        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3468        c = start;
3469    
3470        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3471          {
3472          /* Handle a single character that has more than one other case. */
3473    
3474          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3475            PRIV(ucd_caseless_sets) + rc, oc);
3476    
3477          /* Do nothing if the other case range is within the original range. */
3478    
3479          else if (oc >= start && od <= end) continue;
3480    
3481          /* Extend the original range if there is overlap, noting that if oc < c, we
3482          can't have od > end because a subrange is always shorter than the basic
3483          range. Otherwise, use a recursive call to add the additional range. */
3484    
3485          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3486          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3487          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3488          }
3489        }
3490      else
3491    #endif  /* SUPPORT_UCP */
3492    
3493      /* Not UTF-mode, or no UCP */
3494    
3495      for (c = start; c <= end && c < 256; c++)
3496        {
3497        SETBIT(classbits, cd->fcc[c]);
3498        n8++;
3499        }
3500      }
3501    
3502    /* Now handle the original range. Adjust the final value according to the bit
3503    length - this means that the same lists of (e.g.) horizontal spaces can be used
3504    in all cases. */
3505    
3506    #if defined COMPILE_PCRE8
3507    #ifdef SUPPORT_UTF
3508      if ((options & PCRE_UTF8) == 0)
3509    #endif
3510      if (end > 0xff) end = 0xff;
3511    
3512    #elif defined COMPILE_PCRE16
3513    #ifdef SUPPORT_UTF
3514      if ((options & PCRE_UTF16) == 0)
3515    #endif
3516      if (end > 0xffff) end = 0xffff;
3517    
3518    #endif /* COMPILE_PCRE[8|16] */
3519    
3520    /* If all characters are less than 256, use the bit map. Otherwise use extra
3521    data. */
3522    
3523    if (end < 0x100)
3524      {
3525      for (c = start; c <= end; c++)
3526        {
3527        n8++;
3528        SETBIT(classbits, c);
3529        }
3530      }
3531    
3532    else
3533      {
3534      pcre_uchar *uchardata = *uchardptr;
3535    
3536    #ifdef SUPPORT_UTF
3537      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3538        {
3539        if (start < end)
3540          {
3541          *uchardata++ = XCL_RANGE;
3542          uchardata += PRIV(ord2utf)(start, uchardata);
3543          uchardata += PRIV(ord2utf)(end, uchardata);
3544          }
3545        else if (start == end)
3546          {
3547          *uchardata++ = XCL_SINGLE;
3548          uchardata += PRIV(ord2utf)(start, uchardata);
3549          }
3550        }
3551      else
3552    #endif  /* SUPPORT_UTF */
3553    
3554      /* Without UTF support, character values are constrained by the bit length,
3555      and can only be > 256 for 16-bit and 32-bit libraries. */
3556    
3557    #ifdef COMPILE_PCRE8
3558        {}
3559    #else
3560      if (start < end)
3561        {
3562        *uchardata++ = XCL_RANGE;
3563        *uchardata++ = start;
3564        *uchardata++ = end;
3565        }
3566      else if (start == end)
3567        {
3568        *uchardata++ = XCL_SINGLE;
3569        *uchardata++ = start;
3570        }
3571    #endif
3572    
3573      *uchardptr = uchardata;   /* Updata extra data pointer */
3574      }
3575    
3576    return n8;    /* Number of 8-bit characters */
3577    }
3578    
3579    
3580    
3581    
3582    /*************************************************
3583    *        Add a list of characters to a class     *
3584    *************************************************/
3585    
3586    /* This function is used for adding a list of case-equivalent characters to a
3587    class, and also for adding a list of horizontal or vertical whitespace. If the
3588    list is in order (which it should be), ranges of characters are detected and
3589    handled appropriately. This function is mutually recursive with the function
3590    above.
3591    
3592    Arguments:
3593      classbits     the bit map for characters < 256
3594      uchardptr     points to the pointer for extra data
3595      options       the options word
3596      cd            contains pointers to tables etc.
3597      p             points to row of 32-bit values, terminated by NOTACHAR
3598      except        character to omit; this is used when adding lists of
3599                      case-equivalent characters to avoid including the one we
3600                      already know about
3601    
3602    Returns:        the number of < 256 characters added
3603                    the pointer to extra data is updated
3604    */
3605    
3606    static int
3607    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3608      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3609    {
3610    int n8 = 0;
3611    while (p[0] < NOTACHAR)
3612      {
3613      int n = 0;
3614      if (p[0] != except)
3615        {
3616        while(p[n+1] == p[0] + n + 1) n++;
3617        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3618        }
3619      p += n + 1;
3620      }
3621    return n8;
3622    }
3623    
3624    
3625    
3626    /*************************************************
3627    *    Add characters not in a list to a class     *
3628    *************************************************/
3629    
3630    /* This function is used for adding the complement of a list of horizontal or
3631    vertical whitespace to a class. The list must be in order.
3632    
3633    Arguments:
3634      classbits     the bit map for characters < 256
3635      uchardptr     points to the pointer for extra data
3636      options       the options word
3637      cd            contains pointers to tables etc.
3638      p             points to row of 32-bit values, terminated by NOTACHAR
3639    
3640    Returns:        the number of < 256 characters added
3641                    the pointer to extra data is updated
3642    */
3643    
3644    static int
3645    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3646      int options, compile_data *cd, const pcre_uint32 *p)
3647    {
3648    BOOL utf = (options & PCRE_UTF8) != 0;
3649    int n8 = 0;
3650    if (p[0] > 0)
3651      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3652    while (p[0] < NOTACHAR)
3653      {
3654      while (p[1] == p[0] + 1) p++;
3655      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3656        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3657      p++;
3658      }
3659    return n8;
3660    }
3661    
3662    
3663    
3664    /*************************************************
3665  *           Compile one branch                   *  *           Compile one branch                   *
3666  *************************************************/  *************************************************/
3667    
# Line 2942  Arguments: Line 3676  Arguments:
3676    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3677    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3678    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3679    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3680    reqbyteptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3681      reqcharptr     place to put the last required character
3682      reqcharflagsptr place to put the last required character flags, or a negative number
3683    bcptr          points to current branch chain    bcptr          points to current branch chain
3684      cond_depth     conditional nesting depth
3685    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
3686    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
3687                   points to length accumulator during pre-compile phase                   points to length accumulator during pre-compile phase
# Line 2954  Returns:         TRUE on success Line 3691  Returns:         TRUE on success
3691  */  */
3692    
3693  static BOOL  static BOOL
3694  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3695    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3696      pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3697      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3698      branch_chain *bcptr, int cond_depth,
3699    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3700  {  {
3701  int repeat_type, op_type;  int repeat_type, op_type;
3702  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3703  int bravalue = 0;  int bravalue = 0;
3704  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3705  int firstbyte, reqbyte;  pcre_uint32 firstchar, reqchar;
3706  int zeroreqbyte, zerofirstbyte;  pcre_int32 firstcharflags, reqcharflags;
3707  int req_caseopt, reqvary, tempreqvary;  pcre_uint32 zeroreqchar, zerofirstchar;
3708  int options = *optionsptr;  pcre_int32 zeroreqcharflags, zerofirstcharflags;
3709    pcre_int32 req_caseopt, reqvary, tempreqvary;
3710    int options = *optionsptr;               /* May change dynamically */
3711  int after_manual_callout = 0;  int after_manual_callout = 0;
3712  int length_prevgroup = 0;  int length_prevgroup = 0;
3713  register int c;  register pcre_uint32 c;
3714  register uschar *code = *codeptr;  int escape;
3715  uschar *last_code = code;  register pcre_uchar *code = *codeptr;
3716  uschar *orig_code = code;  pcre_uchar *last_code = code;
3717  uschar *tempcode;  pcre_uchar *orig_code = code;
3718    pcre_uchar *tempcode;
3719  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3720  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3721  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3722  const uschar *tempptr;  const pcre_uchar *tempptr;
3723  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3724  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3725  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3726  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3727  uschar classbits[32];  pcre_uint8 classbits[32];
3728    
3729  #ifdef SUPPORT_UTF8  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we
3730  BOOL class_utf8;  must not do this for other options (e.g. PCRE_EXTENDED) because they may change
3731  BOOL utf8 = (options & PCRE_UTF8) != 0;  dynamically as we process the pattern. */
3732  uschar *class_utf8data;  
3733  uschar *class_utf8data_base;  #ifdef SUPPORT_UTF
3734  uschar utf8_char[6];  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3735    BOOL utf = (options & PCRE_UTF8) != 0;
3736    #ifndef COMPILE_PCRE32
3737    pcre_uchar utf_chars[6];
3738    #endif
3739  #else  #else
3740  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3741  uschar *utf8_char = NULL;  #endif
3742    
3743    /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3744    class_uchardata always so that it can be passed to add_to_class() always,
3745    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3746    alternative calls for the different cases. */
3747    
3748    pcre_uchar *class_uchardata;
3749    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3750    BOOL xclass;
3751    pcre_uchar *class_uchardata_base;
3752  #endif  #endif
3753    
3754  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3005  greedy_non_default = greedy_default ^ 1; Line 3762  greedy_non_default = greedy_default ^ 1;
3762    
3763  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3764  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
3765  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
3766  find one.  find one.
3767    
3768  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
3769  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
3770  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3771  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3772    
3773  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3774    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3775    
3776  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3777  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
3778  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
3779  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3780    value. This is used only for ASCII characters. */
3781    
3782  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3783    
3784  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3785    
# Line 3032  for (;; ptr++) Line 3791  for (;; ptr++)
3791    BOOL is_quantifier;    BOOL is_quantifier;
3792    BOOL is_recurse;    BOOL is_recurse;
3793    BOOL reset_bracount;    BOOL reset_bracount;
3794    int class_charcount;    int class_has_8bitchar;
3795    int class_lastchar;    int class_one_char;
3796    int newoptions;    int newoptions;
3797    int recno;    int recno;
3798    int refsign;    int refsign;
3799    int skipbytes;    int skipbytes;
3800    int subreqbyte;    pcre_uint32 subreqchar, subfirstchar;
3801    int subfirstbyte;    pcre_int32 subreqcharflags, subfirstcharflags;
3802    int terminator;    int terminator;
3803    int mclength;    unsigned int mclength;
3804    uschar mcbuffer[8];    unsigned int tempbracount;
3805      pcre_uint32 ec;
3806      pcre_uchar mcbuffer[8];
3807    
3808    /* Get next byte in the pattern */    /* Get next character in the pattern */
3809    
3810    c = *ptr;    c = *ptr;
3811    
3812    /* 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
3813    string. Nesting only happens one level deep. */    string. Nesting only happens one level deep. */
3814    
3815    if (c == 0 && nestptr != NULL)    if (c == CHAR_NULL && nestptr != NULL)
3816      {      {
3817      ptr = nestptr;      ptr = nestptr;
3818      nestptr = NULL;      nestptr = NULL;
# Line 3066  for (;; ptr++) Line 3827  for (;; ptr++)
3827  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3828      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3829  #endif  #endif
3830      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */      if (code > cd->start_workspace + cd->workspace_size -
3831            WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3832        {        {
3833        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3834        goto FAILED;        goto FAILED;
# Line 3089  for (;; ptr++) Line 3851  for (;; ptr++)
3851        }        }
3852    
3853      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3854      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3855          (int)(code - last_code), c, c));
3856    
3857      /* 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
3858      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 3099  for (;; ptr++) Line 3862  for (;; ptr++)
3862        {        {
3863        if (previous > orig_code)        if (previous > orig_code)
3864          {          {
3865          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3866          code -= previous - orig_code;          code -= previous - orig_code;
3867          previous = orig_code;          previous = orig_code;
3868          }          }
# Line 3115  for (;; ptr++) Line 3878  for (;; ptr++)
3878    /* 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
3879    reference list. */    reference list. */
3880    
3881    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3882               WORK_SIZE_SAFETY_MARGIN)
3883      {      {
3884      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3885      goto FAILED;      goto FAILED;
# Line 3123  for (;; ptr++) Line 3887  for (;; ptr++)
3887    
3888    /* 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 */
3889    
3890    if (inescq && c != 0)    if (inescq && c != CHAR_NULL)
3891      {      {
3892      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3893        {        {
# Line 3163  for (;; ptr++) Line 3927  for (;; ptr++)
3927      previous_callout = NULL;      previous_callout = NULL;
3928      }      }
3929    
3930    /* In extended mode, skip white space and comments */    /* In extended mode, skip white space and comments. */
3931    
3932    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
3933      {      {
3934      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
3935      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3936        {        {
3937        ptr++;        ptr++;
3938        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3939          {          {
3940          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3941          ptr++;          ptr++;
3942  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3943          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3944  #endif  #endif
3945          }          }
3946        if (*ptr != 0) continue;        if (*ptr != CHAR_NULL) continue;
3947    
3948        /* Else fall through to handle end of string */        /* Else fall through to handle end of string */
3949        c = 0;        c = 0;
# Line 3200  for (;; ptr++) Line 3964  for (;; ptr++)
3964      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
3965      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3966      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3967      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
3968      *reqbyteptr = reqbyte;      *firstcharflagsptr = firstcharflags;
3969        *reqcharptr = reqchar;
3970        *reqcharflagsptr = reqcharflags;
3971      *codeptr = code;      *codeptr = code;
3972      *ptrptr = ptr;      *ptrptr = ptr;
3973      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3222  for (;; ptr++) Line 3988  for (;; ptr++)
3988      the setting of any following char as a first character. */      the setting of any following char as a first character. */
3989    
3990      case CHAR_CIRCUMFLEX_ACCENT:      case CHAR_CIRCUMFLEX_ACCENT:
3991        previous = NULL;
3992      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3993        {        {
3994        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
3995          *code++ = OP_CIRCM;
3996        }        }
3997      previous = NULL;      else *code++ = OP_CIRC;
     *code++ = OP_CIRC;  
3998      break;      break;
3999    
4000      case CHAR_DOLLAR_SIGN:      case CHAR_DOLLAR_SIGN:
4001      previous = NULL;      previous = NULL;
4002      *code++ = OP_DOLL;      *code++ = ((options & PCRE_MULTILINE) != 0)? OP_DOLLM : OP_DOLL;
4003      break;      break;
4004    
4005      /* 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
4006      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4007    
4008      case CHAR_DOT:      case CHAR_DOT:
4009      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4010      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
4011      zeroreqbyte = reqbyte;      zerofirstcharflags = firstcharflags;
4012        zeroreqchar = reqchar;
4013        zeroreqcharflags = reqcharflags;
4014      previous = code;      previous = code;
4015      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4016      break;      break;
# Line 3296  for (;; ptr++) Line 4065  for (;; ptr++)
4065          {          {
4066          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
4067            ptr++;            ptr++;
4068          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)  
4069            ptr += 3;            ptr += 3;
4070          else          else
4071            break;            break;
# Line 3316  for (;; ptr++) Line 4084  for (;; ptr++)
4084          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4085        {        {
4086        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4087        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4088        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
4089          zerofirstcharflags = firstcharflags;
4090        break;        break;
4091        }        }
4092    
# Line 3327  for (;; ptr++) Line 4096  for (;; ptr++)
4096    
4097      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4098    
4099      /* 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:
4100      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 <
4101      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
4102        character. */
4103    
4104      class_charcount = 0;      class_has_8bitchar = 0;
4105      class_lastchar = -1;      class_one_char = 0;
4106    
4107      /* 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
4108      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
4109      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
4110      */      map. */
4111    
4112      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4113    
4114  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4115      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;
4116      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4117      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4118  #endif  #endif
4119    
4120      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
4121      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
4122      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
4123    
4124      if (c != 0) do      if (c != CHAR_NULL) do
4125        {        {
4126        const uschar *oldptr;        const pcre_uchar *oldptr;
4127    
4128  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4129        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
4130          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4131          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
4132          }          }
4133    #endif
4134    
4135        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4136          /* In the pre-compile phase, accumulate the length of any extra
4137        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
4138        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4139        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4140          however. */
4141        if (lengthptr != NULL)  
4142          {        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4143          *lengthptr += class_utf8data - class_utf8data_base;          {
4144          class_utf8data = class_utf8data_base;          xclass = TRUE;
4145            *lengthptr += class_uchardata - class_uchardata_base;
4146            class_uchardata = class_uchardata_base;
4147          }          }
   
4148  #endif  #endif
4149    
4150        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3399  for (;; ptr++) Line 4172  for (;; ptr++)
4172          {          {
4173          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
4174          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
4175          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
4176          uschar pbits[32];          pcre_uint8 pbits[32];
4177    
4178          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
4179            {            {
# Line 3427  for (;; ptr++) Line 4200  for (;; ptr++)
4200          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4201          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4202    
4203          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4204            posix_class = 0;            posix_class = 0;
4205    
4206          /* When PCRE_UCP is set, some of the POSIX classes are converted to          /* When PCRE_UCP is set, some of the POSIX classes are converted to
# Line 3455  for (;; ptr++) Line 4228  for (;; ptr++)
4228          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
4229    
4230          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
4231            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
4232    
4233          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
4234    
# Line 3470  for (;; ptr++) Line 4243  for (;; ptr++)
4243              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4244            }            }
4245    
4246          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4247          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4248    
4249          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3486  for (;; ptr++) Line 4259  for (;; ptr++)
4259            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4260    
4261          ptr = tempptr + 1;          ptr = tempptr + 1;
4262          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 character. */
4263            class_has_8bitchar = 1;
4264            /* Every class contains at least two characters. */
4265            class_one_char = 2;
4266          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4267          }          }
4268    
4269        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
4270        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
4271        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
4272        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
4273        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_one_char bigger
4274        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
4275          as literal characters (by default), or are faulted if
4276        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4277    
4278        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4279          {          {
4280          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4281    
4282          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4283    
4284          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4285          else if (-c == ESC_Q)            /* Handle start of quoted string */            c = ec;
4286            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4287            else if (escape == ESC_N)            /* \N is not supported in a class */
4288              {
4289              *errorcodeptr = ERR71;
4290              goto FAILED;
4291              }
4292            else if (escape == ESC_Q)            /* Handle start of quoted string */
4293            {            {
4294            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4295              {              {
# Line 3513  for (;; ptr++) Line 4298  for (;; ptr++)
4298            else inescq = TRUE;            else inescq = TRUE;
4299            continue;            continue;
4300            }            }
4301          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4302    
4303          if (c < 0)          else
4304            {            {
4305            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4306            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
4307              class_has_8bitchar++;
4308              /* Every class contains at least two characters. */
4309              class_one_char += 2;
4310    
4311            switch (-c)            switch (escape)
4312              {              {
4313  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4314              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3530  for (;; ptr++) Line 4318  for (;; ptr++)
4318              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4319              case ESC_SU:              case ESC_SU:
4320              nestptr = ptr;              nestptr = ptr;
4321              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4322              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4323              continue;              continue;
4324  #endif  #endif
4325              case ESC_d:              case ESC_d:
# Line 3554  for (;; ptr++) Line 4342  for (;; ptr++)
4342    
4343              /* 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
4344              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4345              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4346                EBCDIC, so we lazily just adjust the appropriate bit. */
4347    
4348              case ESC_s:              case ESC_s:
4349              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 3567  for (;; ptr++) Line 4356  for (;; ptr++)
4356              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4357              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4358              continue;              continue;
4359    
4360                /* The rest apply in both UCP and non-UCP cases. */
4361    
4362              case ESC_h:              case ESC_h:
4363              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4364              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  
4365              continue;              continue;
4366    
4367              case ESC_H:              case ESC_H:
4368              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4369                {                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  
4370              continue;              continue;
4371    
4372              case ESC_v:              case ESC_v:
4373              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4374              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  
4375              continue;              continue;
4376    
4377              case ESC_V:              case ESC_V:
4378              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4379                {                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  
4380              continue;              continue;
4381    
4382  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3689  for (;; ptr++) Line 4384  for (;; ptr++)
4384              case ESC_P:              case ESC_P:
4385                {                {
4386                BOOL negated;                BOOL negated;
4387                int pdata;                unsigned int ptype = 0, pdata = 0;
4388                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4389                if (ptype < 0) goto FAILED;                  goto FAILED;
4390                class_utf8 = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_utf8data++ = ((-c == ESC_p) != negated)?  
4391                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4392                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4393                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4394                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4395                continue;                continue;
4396                }                }
4397  #endif  #endif
# Line 3711  for (;; ptr++) Line 4405  for (;; ptr++)
4405                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4406                goto FAILED;                goto FAILED;
4407                }                }
4408              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4409              c = *ptr;              /* Get the final character and fall through */              class_one_char -= 2;     /* Undo the speculative increase. */
4410                c = *ptr;                /* Get the final character and fall through */
4411              break;              break;
4412              }              }
4413            }            }
4414    
4415          /* 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).
4416          greater than 256 in UTF-8 mode. */          This may be greater than 256. */
4417    
4418            escape = 0;
4419    
4420          }   /* End of backslash handling */          }   /* End of backslash handling */
4421    
4422        /* 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
4423        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
4424        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4425        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4426    
4427        CHECK_RANGE:        CHECK_RANGE:
4428        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3733  for (;; ptr++) Line 4430  for (;; ptr++)
4430          inescq = FALSE;          inescq = FALSE;
4431          ptr += 2;          ptr += 2;
4432          }          }
   
4433        oldptr = ptr;        oldptr = ptr;
4434    
4435        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4436    
4437        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4438    
# Line 3744  for (;; ptr++) Line 4440  for (;; ptr++)
4440    
4441        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4442          {          {
4443          int d;          pcre_uint32 d;
4444          ptr += 2;          ptr += 2;
4445          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4446    
# Line 3759  for (;; ptr++) Line 4455  for (;; ptr++)
4455            inescq = TRUE;            inescq = TRUE;
4456            break;            break;
4457            }            }
4458    
4459            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4460            back the pointer and jump to handle the character that preceded it. */
4461    
4462          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == CHAR_NULL || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4463            {            {
4464            ptr = oldptr;            ptr = oldptr;
4465            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4466            }            }
4467    
4468            /* Otherwise, we have a potential range; pick up the next character */
4469    
4470  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4471          if (utf8)          if (utf)
4472            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4473            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4474            }            }
# Line 3781  for (;; ptr++) Line 4482  for (;; ptr++)
4482    
4483          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4484            {            {
4485            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4486              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4487            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4488    
4489            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4490    
4491            if (d < 0)            if (descape != 0)
4492              {              {
4493              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4494                {                {
4495                ptr = oldptr;                ptr = oldptr;
4496                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4497                }                }
4498              }              }
4499            }            }
4500    
4501          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4502          one-character ranges */          one-character ranges. */
4503    
4504          if (d < c)          if (d < c)
4505            {            {
4506            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4507            goto FAILED;            goto FAILED;
4508            }            }
4509            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4510    
4511          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4512            cannot be done anymore. Any value greater than 1 indicates that there
4513            is more than one character. */
4514    
4515            class_one_char = 2;
4516    
4517          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4518    
4519          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4520    
4521          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless          class_has_8bitchar +=
4522          matching, we have to use an XCLASS with extra data items. Caseless            add_to_class(classbits, &class_uchardata, options, cd, c, d);
4523          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 */  
   
           d = 127;  
   
 #endif  /* SUPPORT_UCP */  
           }  
 #endif  /* SUPPORT_UTF8 */  
   
         /* We use the bit map for all cases when not in UTF-8 mode; else  
         ranges that lie entirely within 0-127 when there is UCP support; else  
         for partial ranges without UCP support. */  
   
         class_charcount += d - c + 1;  
         class_lastchar = d;  
   
         /* We can save a bit of time by skipping this in the pre-compile. */  
   
         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));  
             }  
           }  
   
4524          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4525          }          }
4526    
4527        /* 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
4528        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
4529        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4530          increase it if it is already 2 or more ... just in case there's a class
4531          with a zillion characters in it. */
4532    
4533          CLASS_SINGLE_CHARACTER:
4534          if (class_one_char < 2) class_one_char++;
4535    
4536          /* If class_one_char is 1, we have the first single character in the
4537          class, and there have been no prior ranges, or XCLASS items generated by
4538          escapes. If this is the final character in the class, we can optimize by
4539          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4540          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4541          to be set. Otherwise, there can be no first char if this item is first,
4542          whatever repeat count may follow. In the case of reqchar, save the
4543          previous value for reinstating. */
4544    
4545        LONE_SINGLE_CHARACTER:        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
   
       /* Handle a character that cannot go in the bit map */  
   
 #ifdef SUPPORT_UTF8  
       if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
4546          {          {
4547          class_utf8 = TRUE;          ptr++;
4548          *class_utf8data++ = XCL_SINGLE;          zeroreqchar = reqchar;
4549          class_utf8data += _pcre_ord2utf8(c, class_utf8data);          zeroreqcharflags = reqcharflags;
4550    
4551  #ifdef SUPPORT_UCP          if (negate_class)
         if ((options & PCRE_CASELESS) != 0)  
4552            {            {
4553            unsigned int othercase;  #ifdef SUPPORT_UCP
4554            if ((othercase = UCD_OTHERCASE(c)) != c)            int d;
4555    #endif
4556              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4557              zerofirstchar = firstchar;
4558              zerofirstcharflags = firstcharflags;
4559    
4560              /* For caseless UTF-8 mode when UCP support is available, check
4561              whether this character has more than one other case. If so, generate
4562              a special OP_NOTPROP item instead of OP_NOTI. */
4563    
4564    #ifdef SUPPORT_UCP
4565              if (utf && (options & PCRE_CASELESS) != 0 &&
4566                  (d = UCD_CASESET(c)) != 0)
4567              {              {
4568              *class_utf8data++ = XCL_SINGLE;              *code++ = OP_NOTPROP;
4569              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);              *code++ = PT_CLIST;
4570                *code++ = d;
4571              }              }
4572              else
4573    #endif
4574              /* Char has only one other case, or UCP not available */
4575    
4576                {
4577                *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4578    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4579                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4580                  code += PRIV(ord2utf)(c, code);
4581                else
4582    #endif
4583                  *code++ = c;
4584                }
4585    
4586              /* We are finished with this character class */
4587    
4588              goto END_CLASS;
4589            }            }
 #endif  /* SUPPORT_UCP */  
4590    
4591          }          /* For a single, positive character, get the value into mcbuffer, and
4592        else          then we can handle this with the normal one-character code. */
 #endif  /* SUPPORT_UTF8 */  
4593    
4594        /* Handle a single-byte character */  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4595          {          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4596          classbits[c/8] |= (1 << (c&7));            mclength = PRIV(ord2utf)(c, mcbuffer);
4597          if ((options & PCRE_CASELESS) != 0)          else
4598    #endif
4599            {            {
4600            c = cd->fcc[c];   /* flip case */            mcbuffer[0] = c;
4601            classbits[c/8] |= (1 << (c&7));            mclength = 1;
4602            }            }
4603          class_charcount++;          goto ONE_CHAR;
4604          class_lastchar = c;          }       /* End of 1-char optimization */
4605          }  
4606        }        /* There is more than one character in the class, or an XCLASS item
4607          has been generated. Add this character to the class. */
4608