/[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 638 by ph10, Mon Jul 25 09:41:19 2011 UTC revision 1133 by chpe, Thu Oct 18 18:35:30 2012 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2011 University of Cambridge             Copyright (c) 1997-2012 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. */  library. We do not need to select pcre16_printint.c specially, because the
59    COMPILE_PCREx macro will already be appropriately set. */
60    
61  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
62  #include "pcre_printint.src"  /* pcre_printint.c should not include any headers */
63    #define PCRE_INCLUDED
64    #include "pcre_printint.c"
65    #undef PCRE_INCLUDED
66  #endif  #endif
67    
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 73  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,
88        pcre_uint32 *, pcre_int32 *, pcre_uint32 *, pcre_int32 *, branch_chain *,
89        compile_data *, int *);
90    
91    
92    
93  /*************************************************  /*************************************************
94  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 88  so this number is very generous. Line 104  so this number is very generous.
104  The same workspace is used during the second, actual compile phase for  The same workspace is used during the second, actual compile phase for
105  remembering forward references to groups so that they can be filled in at the  remembering forward references to groups so that they can be filled in at the
106  end. Each entry in this list occupies LINK_SIZE bytes, so even when LINK_SIZE  end. Each entry in this list occupies LINK_SIZE bytes, so even when LINK_SIZE
107  is 4 there is plenty of room. */  is 4 there is plenty of room for most patterns. However, the memory can get
108    filled up by repetitions of forward references, for example patterns like
109    /(?1){0,1999}(b)/, and one user did hit the limit. The code has been changed so
110    that the workspace is expanded using malloc() in this situation. The value
111    below is therefore a minimum, and we put a maximum on it for safety. The
112    minimum is now also defined in terms of LINK_SIZE so that the use of malloc()
113    kicks in at the same number of forward references in all cases. */
114    
115  #define COMPILE_WORK_SIZE (4096)  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)
116    #define COMPILE_WORK_SIZE_MAX (100*COMPILE_WORK_SIZE)
117    
118  /* The overrun tests check for a slightly smaller size so that they detect the  /* The overrun tests check for a slightly smaller size so that they detect the
119  overrun before it actually does run off the end of the data block. */  overrun before it actually does run off the end of the data block. */
120    
121  #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100)  #define WORK_SIZE_SAFETY_MARGIN (100)
122    
123    /* Private flags added to firstchar and reqchar. */
124    
125    #define REQ_CASELESS    (1 << 0)        /* Indicates caselessness */
126    #define REQ_VARY        (1 << 1)        /* Reqchar followed non-literal item */
127    /* Negative values for the firstchar and reqchar flags */
128    #define REQ_UNSET       (-2)
129    #define REQ_NONE        (-1)
130    
131    /* Repeated character flags. */
132    
133    #define UTF_LENGTH     0x10000000l      /* The char contains its length. */
134    
135  /* Table for handling escaped characters in the range '0'-'z'. Positive returns  /* Table for handling escaped characters in the range '0'-'z'. Positive returns
136  are simple data values; negative values are for special things like \d and so  are simple data values; negative values are for special things like \d and so
# Line 231  static const char posix_names[] = Line 265  static const char posix_names[] =
265    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0
266    STRING_word0  STRING_xdigit;    STRING_word0  STRING_xdigit;
267    
268  static const uschar posix_name_lengths[] = {  static const pcre_uint8 posix_name_lengths[] = {
269    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };
270    
271  /* Table of class bit maps for each POSIX class. Each class is formed from a  /* Table of class bit maps for each POSIX class. Each class is formed from a
# Line 266  substitutes must be in the order of the Line 300  substitutes must be in the order of the
300  both positive and negative cases. NULL means no substitute. */  both positive and negative cases. NULL means no substitute. */
301    
302  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
303  static const uschar *substitutes[] = {  static const pcre_uchar string_PNd[]  = {
304    (uschar *)"\\P{Nd}",    /* \D */    CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
305    (uschar *)"\\p{Nd}",    /* \d */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
306    (uschar *)"\\P{Xsp}",   /* \S */       /* NOTE: Xsp is Perl space */  static const pcre_uchar string_pNd[]  = {
307    (uschar *)"\\p{Xsp}",   /* \s */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
308    (uschar *)"\\P{Xwd}",   /* \W */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
309    (uschar *)"\\p{Xwd}"    /* \w */  static const pcre_uchar string_PXsp[] = {
310      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
311      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
312    static const pcre_uchar string_pXsp[] = {
313      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
314      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
315    static const pcre_uchar string_PXwd[] = {
316      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
317      CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
318    static const pcre_uchar string_pXwd[] = {
319      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
320      CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
321    
322    static const pcre_uchar *substitutes[] = {
323      string_PNd,           /* \D */
324      string_pNd,           /* \d */
325      string_PXsp,          /* \S */       /* NOTE: Xsp is Perl space */
326      string_pXsp,          /* \s */
327      string_PXwd,          /* \W */
328      string_pXwd           /* \w */
329  };  };
330    
331  static const uschar *posix_substitutes[] = {  static const pcre_uchar string_pL[] =   {
332    (uschar *)"\\p{L}",     /* alpha */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
333    (uschar *)"\\p{Ll}",    /* lower */    CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
334    (uschar *)"\\p{Lu}",    /* upper */  static const pcre_uchar string_pLl[] =  {
335    (uschar *)"\\p{Xan}",   /* alnum */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
336    NULL,                   /* ascii */    CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
337    (uschar *)"\\h",        /* blank */  static const pcre_uchar string_pLu[] =  {
338    NULL,                   /* cntrl */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
339    (uschar *)"\\p{Nd}",    /* digit */    CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
340    NULL,                   /* graph */  static const pcre_uchar string_pXan[] = {
341    NULL,                   /* print */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
342    NULL,                   /* punct */    CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
343    (uschar *)"\\p{Xps}",   /* space */    /* NOTE: Xps is POSIX space */  static const pcre_uchar string_h[] =    {
344    (uschar *)"\\p{Xwd}",   /* word */    CHAR_BACKSLASH, CHAR_h, '\0' };
345    NULL,                   /* xdigit */  static const pcre_uchar string_pXps[] = {
346      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
347      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
348    static const pcre_uchar string_PL[] =   {
349      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
350      CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
351    static const pcre_uchar string_PLl[] =  {
352      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
353      CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
354    static const pcre_uchar string_PLu[] =  {
355      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
356      CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
357    static const pcre_uchar string_PXan[] = {
358      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
359      CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
360    static const pcre_uchar string_H[] =    {
361      CHAR_BACKSLASH, CHAR_H, '\0' };
362    static const pcre_uchar string_PXps[] = {
363      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
364      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
365    
366    static const pcre_uchar *posix_substitutes[] = {
367      string_pL,            /* alpha */
368      string_pLl,           /* lower */
369      string_pLu,           /* upper */
370      string_pXan,          /* alnum */
371      NULL,                 /* ascii */
372      string_h,             /* blank */
373      NULL,                 /* cntrl */
374      string_pNd,           /* digit */
375      NULL,                 /* graph */
376      NULL,                 /* print */
377      NULL,                 /* punct */
378      string_pXps,          /* space */    /* NOTE: Xps is POSIX space */
379      string_pXwd,          /* word */
380      NULL,                 /* xdigit */
381    /* Negated cases */    /* Negated cases */
382    (uschar *)"\\P{L}",     /* ^alpha */    string_PL,            /* ^alpha */
383    (uschar *)"\\P{Ll}",    /* ^lower */    string_PLl,           /* ^lower */
384    (uschar *)"\\P{Lu}",    /* ^upper */    string_PLu,           /* ^upper */
385    (uschar *)"\\P{Xan}",   /* ^alnum */    string_PXan,          /* ^alnum */
386    NULL,                   /* ^ascii */    NULL,                 /* ^ascii */
387    (uschar *)"\\H",        /* ^blank */    string_H,             /* ^blank */
388    NULL,                   /* ^cntrl */    NULL,                 /* ^cntrl */
389    (uschar *)"\\P{Nd}",    /* ^digit */    string_PNd,           /* ^digit */
390    NULL,                   /* ^graph */    NULL,                 /* ^graph */
391    NULL,                   /* ^print */    NULL,                 /* ^print */
392    NULL,                   /* ^punct */    NULL,                 /* ^punct */
393    (uschar *)"\\P{Xps}",   /* ^space */   /* NOTE: Xps is POSIX space */    string_PXps,          /* ^space */   /* NOTE: Xps is POSIX space */
394    (uschar *)"\\P{Xwd}",   /* ^word */    string_PXwd,          /* ^word */
395    NULL                    /* ^xdigit */    NULL                  /* ^xdigit */
396  };  };
397  #define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))  #define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))
398  #endif  #endif
399    
400  #define STRING(a)  # a  #define STRING(a)  # a
# Line 365  static const char error_texts[] = Line 453  static const char error_texts[] =
453    /* 30 */    /* 30 */
454    "unknown POSIX class name\0"    "unknown POSIX class name\0"
455    "POSIX collating elements are not supported\0"    "POSIX collating elements are not supported\0"
456    "this version of PCRE is not compiled with PCRE_UTF8 support\0"    "this version of PCRE is compiled without UTF support\0"
457    "spare error\0"  /** DEAD **/    "spare error\0"  /** DEAD **/
458    "character value in \\x{...} sequence is too large\0"    "character value in \\x{...} sequence is too large\0"
459    /* 35 */    /* 35 */
# Line 388  static const char error_texts[] = Line 476  static const char error_texts[] =
476    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
477    /* 50 */    /* 50 */
478    "repeated subpattern is too long\0"    /** DEAD **/    "repeated subpattern is too long\0"    /** DEAD **/
479    "octal value is greater than \\377 (not in UTF-8 mode)\0"    "octal value is greater than \\377 in 8-bit non-UTF-8 mode\0"
480    "internal error: overran compiling workspace\0"    "internal error: overran compiling workspace\0"
481    "internal error: previously-checked referenced subpattern not found\0"    "internal error: previously-checked referenced subpattern not found\0"
482    "DEFINE group contains more than one branch\0"    "DEFINE group contains more than one branch\0"
# Line 407  static const char error_texts[] = Line 495  static const char error_texts[] =
495    /* 65 */    /* 65 */
496    "different names for subpatterns of the same number are not allowed\0"    "different names for subpatterns of the same number are not allowed\0"
497    "(*MARK) must have an argument\0"    "(*MARK) must have an argument\0"
498    "this version of PCRE is not compiled with PCRE_UCP support\0"    "this version of PCRE is not compiled with Unicode property support\0"
499    "\\c must be followed by an ASCII character\0"    "\\c must be followed by an ASCII character\0"
500    "\\k is not followed by a braced, angle-bracketed, or quoted name\0"    "\\k is not followed by a braced, angle-bracketed, or quoted name\0"
501      /* 70 */
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 428  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 472  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 507  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 543  static const unsigned char ebcdic_charta Line 647  static const unsigned char ebcdic_charta
647  #endif  #endif
648    
649    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int *,  
     int *, branch_chain *, compile_data *, int *);  
   
   
650    
651  /*************************************************  /*************************************************
652  *            Find an error text                  *  *            Find an error text                  *
# Line 578  return s; Line 675  return s;
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 */
# Line 621  in a table. A non-zero result is somethi Line 795  in a table. A non-zero result is somethi
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 643  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 662  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 675  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 != 0 && *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 != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
904          {          {
905          c = -ESC_k;          escape = ESC_k;
906          break;          break;
907          }          }
908        braced = TRUE;        braced = TRUE;
# Line 695  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 711  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 719  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 748  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 778  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 812  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 828  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 886  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 917  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 == 0) goto ERROR_RETURN;
# Line 945  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 == 0) goto ERROR_RETURN;
# Line 969  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 998  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 1049  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 1058  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 1073  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 1128  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 1204  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1462  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1462      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1463          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1464        {        {
1465        int term;        pcre_uchar term;
1466        const uschar *thisname;        const pcre_uchar *thisname;
1467        *count += 1;        *count += 1;
1468        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1469        term = *ptr++;        term = *ptr++;
1470        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1471        thisname = ptr;        thisname = ptr;
1472        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1473        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1474            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1475          return *count;          return *count;
1476        term++;        term++;
1477        }        }
# Line 1256  for (; ptr < cd->end_pattern; ptr++) Line 1514  for (; ptr < cd->end_pattern; ptr++)
1514          {          {
1515          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1516            ptr+= 2;            ptr+= 2;
1517          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1518                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1519            ptr += 4;            ptr += 4;
1520          else          else
# Line 1304  for (; ptr < cd->end_pattern; ptr++) Line 1562  for (; ptr < cd->end_pattern; ptr++)
1562        {        {
1563        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1564        ptr++;        ptr++;
1565  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1566        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1567  #endif  #endif
1568        }        }
1569      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
# Line 1316  for (; ptr < cd->end_pattern; ptr++) Line 1574  for (; ptr < cd->end_pattern; ptr++)
1574    
1575    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1576      {      {
1577      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1578      if (rc > 0) return rc;      if (rc > 0) return rc;
1579      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1580      }      }
# Line 1362  Arguments: Line 1620  Arguments:
1620    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1621    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1622    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1623    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1624    
1625  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1626  */  */
1627    
1628  static int  static int
1629  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,
1630    BOOL utf8)    BOOL utf)
1631  {  {
1632  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1633  int count = 0;  int count = 0;
1634  int rc;  int rc;
1635    
# Line 1382  matching closing parens. That is why we Line 1640  matching closing parens. That is why we
1640    
1641  for (;;)  for (;;)
1642    {    {
1643    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1644    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == 0) break;
1645    }    }
1646    
# Line 1409  Arguments: Line 1667  Arguments:
1667  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1668  */  */
1669    
1670  static const uschar*  static const pcre_uchar*
1671  first_significant_code(const uschar *code, BOOL skipassert)  first_significant_code(const pcre_uchar *code, BOOL skipassert)
1672  {  {
1673  for (;;)  for (;;)
1674    {    {
# Line 1421  for (;;) Line 1679  for (;;)
1679      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1680      if (!skipassert) return code;      if (!skipassert) return code;
1681      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1682      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1683      break;      break;
1684    
1685      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1435  for (;;) Line 1693  for (;;)
1693      case OP_RREF:      case OP_RREF:
1694      case OP_NRREF:      case OP_NRREF:
1695      case OP_DEF:      case OP_DEF:
1696      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1697      break;      break;
1698    
1699      default:      default:
# Line 1465  and doing the check at the end; a flag s Line 1723  and doing the check at the end; a flag s
1723    
1724  Arguments:  Arguments:
1725    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1726    utf8     TRUE in UTF-8 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1727    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1728    cd       the "compile data" structure    cd       the "compile data" structure
1729    
1730  Returns:   the fixed length,  Returns:   the fixed length,
1731               or -1 if there is no fixed length,               or -1 if there is no fixed length,
1732               or -2 if \C was encountered               or -2 if \C was encountered (in UTF-8 mode only)
1733               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
1734                 or -4 if an unknown opcode was encountered (internal error)
1735  */  */
1736    
1737  static int  static int
1738  find_fixedlength(uschar *code, BOOL utf8, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1739  {  {
1740  int length = -1;  int length = -1;
1741    
1742  register int branchlength = 0;  register int branchlength = 0;
1743  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1744    
1745  /* 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
1746  branch, check the length against that of the other branches. */  branch, check the length against that of the other branches. */
# Line 1489  branch, check the length against that of Line 1748  branch, check the length against that of
1748  for (;;)  for (;;)
1749    {    {
1750    int d;    int d;
1751    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1752    register int op = *cc;    register pcre_uchar op = *cc;
1753    
1754    switch (op)    switch (op)
1755      {      {
1756      /* We only need to continue for OP_CBRA (normal capturing bracket) and      /* We only need to continue for OP_CBRA (normal capturing bracket) and
1757      OP_BRA (normal non-capturing bracket) because the other variants of these      OP_BRA (normal non-capturing bracket) because the other variants of these
1758      opcodes are all concerned with unlimited repeated groups, which of course      opcodes are all concerned with unlimited repeated groups, which of course
1759      are not of fixed length. They will cause a -1 response from the default      are not of fixed length. */
     case of this switch. */  
1760    
1761      case OP_CBRA:      case OP_CBRA:
1762      case OP_BRA:      case OP_BRA:
1763      case OP_ONCE:      case OP_ONCE:
1764        case OP_ONCE_NC:
1765      case OP_COND:      case OP_COND:
1766      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd);      d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd);
1767      if (d < 0) return d;      if (d < 0) return d;
1768      branchlength += d;      branchlength += d;
1769      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1770      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
1771      break;      break;
1772    
1773      /* 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.
1774      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
1775      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
1776      Note that we must not include the OP_KETRxxx opcodes here, because they      the same code. Note that we must not include the OP_KETRxxx opcodes here,
1777      all imply an unlimited repeat. */      because they all imply an unlimited repeat. */
1778    
1779      case OP_ALT:      case OP_ALT:
1780      case OP_KET:      case OP_KET:
1781      case OP_END:      case OP_END:
1782        case OP_ACCEPT:
1783        case OP_ASSERT_ACCEPT:
1784      if (length < 0) length = branchlength;      if (length < 0) length = branchlength;
1785        else if (length != branchlength) return -1;        else if (length != branchlength) return -1;
1786      if (*cc != OP_ALT) return length;      if (*cc != OP_ALT) return length;
# Line 1532  for (;;) Line 1794  for (;;)
1794    
1795      case OP_RECURSE:      case OP_RECURSE:
1796      if (!atend) return -3;      if (!atend) return -3;
1797      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1798      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1799      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1800      d = find_fixedlength(cs + 2, utf8, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1801      if (d < 0) return d;      if (d < 0) return d;
1802      branchlength += d;      branchlength += d;
1803      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1548  for (;;) Line 1810  for (;;)
1810      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1811      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1812      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1813      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1814        break;
1815    
1816      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1817    
1818      case OP_REVERSE:      case OP_MARK:
1819      case OP_CREF:      case OP_PRUNE_ARG:
1820      case OP_NCREF:      case OP_SKIP_ARG:
1821      case OP_RREF:      case OP_THEN_ARG:
1822      case OP_NRREF:      cc += cc[1] + PRIV(OP_lengths)[*cc];
1823      case OP_DEF:      break;
1824    
1825      case OP_CALLOUT:      case OP_CALLOUT:
     case OP_SOD:  
     case OP_SOM:  
     case OP_SET_SOM:  
     case OP_EOD:  
     case OP_EODN:  
1826      case OP_CIRC:      case OP_CIRC:
1827      case OP_CIRCM:      case OP_CIRCM:
1828        case OP_CLOSE:
1829        case OP_COMMIT:
1830        case OP_CREF:
1831        case OP_DEF:
1832      case OP_DOLL:      case OP_DOLL:
1833      case OP_DOLLM:      case OP_DOLLM:
1834        case OP_EOD:
1835        case OP_EODN:
1836        case OP_FAIL:
1837        case OP_NCREF:
1838        case OP_NRREF:
1839      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1840        case OP_PRUNE:
1841        case OP_REVERSE:
1842        case OP_RREF:
1843        case OP_SET_SOM:
1844        case OP_SKIP:
1845        case OP_SOD:
1846        case OP_SOM:
1847        case OP_THEN:
1848      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1849      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1850      break;      break;
1851    
1852      /* Handle literal characters */      /* Handle literal characters */
# Line 1581  for (;;) Line 1857  for (;;)
1857      case OP_NOTI:      case OP_NOTI:
1858      branchlength++;      branchlength++;
1859      cc += 2;      cc += 2;
1860  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1861      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1862  #endif  #endif
1863      break;      break;
1864    
# Line 1590  for (;;) Line 1866  for (;;)
1866      need to skip over a multibyte character in UTF8 mode.  */      need to skip over a multibyte character in UTF8 mode.  */
1867    
1868      case OP_EXACT:      case OP_EXACT:
1869      branchlength += GET2(cc,1);      case OP_EXACTI:
1870      cc += 4;      case OP_NOTEXACT:
1871  #ifdef SUPPORT_UTF8      case OP_NOTEXACTI:
1872      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      branchlength += (int)GET2(cc,1);
1873        cc += 2 + IMM2_SIZE;
1874    #ifdef SUPPORT_UTF
1875        if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1876  #endif  #endif
1877      break;      break;
1878    
1879      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1880      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1881      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1882      cc += 4;        cc += 2;
1883        cc += 1 + IMM2_SIZE + 1;
1884      break;      break;
1885    
1886      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1610  for (;;) Line 1890  for (;;)
1890      cc += 2;      cc += 2;
1891      /* Fall through */      /* Fall through */
1892    
1893        case OP_HSPACE:
1894        case OP_VSPACE:
1895        case OP_NOT_HSPACE:
1896        case OP_NOT_VSPACE:
1897      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
1898      case OP_DIGIT:      case OP_DIGIT:
1899      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
# Line 1622  for (;;) Line 1906  for (;;)
1906      cc++;      cc++;
1907      break;      break;
1908    
1909      /* The single-byte matcher isn't allowed */      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;
1910        otherwise \C is coded as OP_ALLANY. */
1911    
1912      case OP_ANYBYTE:      case OP_ANYBYTE:
1913      return -2;      return -2;
1914    
1915      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1916    
1917  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1918      case OP_XCLASS:      case OP_XCLASS:
1919      cc += GET(cc, 1) - 33;      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
1920      /* Fall through */      /* Fall through */
1921  #endif  #endif
1922    
1923      case OP_CLASS:      case OP_CLASS:
1924      case OP_NCLASS:      case OP_NCLASS:
1925      cc += 33;      cc += PRIV(OP_lengths)[OP_CLASS];
1926    
1927      switch (*cc)      switch (*cc)
1928        {        {
1929          case OP_CRPLUS:
1930          case OP_CRMINPLUS:
1931        case OP_CRSTAR:        case OP_CRSTAR:
1932        case OP_CRMINSTAR:        case OP_CRMINSTAR:
1933        case OP_CRQUERY:        case OP_CRQUERY:
# Line 1649  for (;;) Line 1936  for (;;)
1936    
1937        case OP_CRRANGE:        case OP_CRRANGE:
1938        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1939        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1940        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1941        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1942        break;        break;
1943    
1944        default:        default:
# Line 1661  for (;;) Line 1948  for (;;)
1948    
1949      /* Anything else is variable length */      /* Anything else is variable length */
1950    
1951      default:      case OP_ANYNL:
1952        case OP_BRAMINZERO:
1953        case OP_BRAPOS:
1954        case OP_BRAPOSZERO:
1955        case OP_BRAZERO:
1956        case OP_CBRAPOS:
1957        case OP_EXTUNI:
1958        case OP_KETRMAX:
1959        case OP_KETRMIN:
1960        case OP_KETRPOS:
1961        case OP_MINPLUS:
1962        case OP_MINPLUSI:
1963        case OP_MINQUERY:
1964        case OP_MINQUERYI:
1965        case OP_MINSTAR:
1966        case OP_MINSTARI:
1967        case OP_MINUPTO:
1968        case OP_MINUPTOI:
1969        case OP_NOTMINPLUS:
1970        case OP_NOTMINPLUSI:
1971        case OP_NOTMINQUERY:
1972        case OP_NOTMINQUERYI:
1973        case OP_NOTMINSTAR:
1974        case OP_NOTMINSTARI:
1975        case OP_NOTMINUPTO:
1976        case OP_NOTMINUPTOI:
1977        case OP_NOTPLUS:
1978        case OP_NOTPLUSI:
1979        case OP_NOTPOSPLUS:
1980        case OP_NOTPOSPLUSI:
1981        case OP_NOTPOSQUERY:
1982        case OP_NOTPOSQUERYI:
1983        case OP_NOTPOSSTAR:
1984        case OP_NOTPOSSTARI:
1985        case OP_NOTPOSUPTO:
1986        case OP_NOTPOSUPTOI:
1987        case OP_NOTQUERY:
1988        case OP_NOTQUERYI:
1989        case OP_NOTSTAR:
1990        case OP_NOTSTARI:
1991        case OP_NOTUPTO:
1992        case OP_NOTUPTOI:
1993        case OP_PLUS:
1994        case OP_PLUSI:
1995        case OP_POSPLUS:
1996        case OP_POSPLUSI:
1997        case OP_POSQUERY:
1998        case OP_POSQUERYI:
1999        case OP_POSSTAR:
2000        case OP_POSSTARI:
2001        case OP_POSUPTO:
2002        case OP_POSUPTOI:
2003        case OP_QUERY:
2004        case OP_QUERYI:
2005        case OP_REF:
2006        case OP_REFI:
2007        case OP_SBRA:
2008        case OP_SBRAPOS:
2009        case OP_SCBRA:
2010        case OP_SCBRAPOS:
2011        case OP_SCOND:
2012        case OP_SKIPZERO:
2013        case OP_STAR:
2014        case OP_STARI:
2015        case OP_TYPEMINPLUS:
2016        case OP_TYPEMINQUERY:
2017        case OP_TYPEMINSTAR:
2018        case OP_TYPEMINUPTO:
2019        case OP_TYPEPLUS:
2020        case OP_TYPEPOSPLUS:
2021        case OP_TYPEPOSQUERY:
2022        case OP_TYPEPOSSTAR:
2023        case OP_TYPEPOSUPTO:
2024        case OP_TYPEQUERY:
2025        case OP_TYPESTAR:
2026        case OP_TYPEUPTO:
2027        case OP_UPTO:
2028        case OP_UPTOI:
2029      return -1;      return -1;
2030    
2031        /* Catch unrecognized opcodes so that when new ones are added they
2032        are not forgotten, as has happened in the past. */
2033    
2034        default:
2035        return -4;
2036      }      }
2037    }    }
2038  /* Control never gets here */  /* Control never gets here */
# Line 1683  length. Line 2053  length.
2053    
2054  Arguments:  Arguments:
2055    code        points to start of expression    code        points to start of expression
2056    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2057    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2058    
2059  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
2060  */  */
2061    
2062  const uschar *  const pcre_uchar *
2063  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2064  {  {
2065  for (;;)  for (;;)
2066    {    {
2067    register int c = *code;    register pcre_uchar c = *code;
2068    
2069    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2070    
# Line 1708  for (;;) Line 2078  for (;;)
2078    
2079    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2080      {      {
2081      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2082      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2083      }      }
2084    
2085    /* Handle capturing bracket */    /* Handle capturing bracket */
# Line 1717  for (;;) Line 2087  for (;;)
2087    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2088             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2089      {      {
2090      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2091      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2092      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2093      }      }
2094    
2095    /* 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 1747  for (;;) Line 2117  for (;;)
2117        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2118        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2119        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2120        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2121            code += 2;
2122        break;        break;
2123    
2124        case OP_MARK:        case OP_MARK:
# Line 1757  for (;;) Line 2128  for (;;)
2128        break;        break;
2129    
2130        case OP_THEN_ARG:        case OP_THEN_ARG:
2131        code += code[1+LINK_SIZE];        code += code[1];
2132        break;        break;
2133        }        }
2134    
2135      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2136    
2137      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2138    
2139    /* 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
2140    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
2141    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2142    
2143  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2144      if (utf8) switch(c)      if (utf) switch(c)
2145        {        {
2146        case OP_CHAR:        case OP_CHAR:
2147        case OP_CHARI:        case OP_CHARI:
# Line 1800  for (;;) Line 2171  for (;;)
2171        case OP_MINQUERYI:        case OP_MINQUERYI:
2172        case OP_POSQUERY:        case OP_POSQUERY:
2173        case OP_POSQUERYI:        case OP_POSQUERYI:
2174        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2175        break;        break;
2176        }        }
2177  #else  #else
2178      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2179  #endif  #endif
2180      }      }
2181    }    }
# Line 1821  instance of OP_RECURSE. Line 2192  instance of OP_RECURSE.
2192    
2193  Arguments:  Arguments:
2194    code        points to start of expression    code        points to start of expression
2195    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2196    
2197  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
2198  */  */
2199    
2200  static const uschar *  static const pcre_uchar *
2201  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2202  {  {
2203  for (;;)  for (;;)
2204    {    {
2205    register int c = *code;    register pcre_uchar c = *code;
2206    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2207    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2208    
# Line 1866  for (;;) Line 2237  for (;;)
2237        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2238        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2239        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2240        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2241            code += 2;
2242        break;        break;
2243    
2244        case OP_MARK:        case OP_MARK:
# Line 1876  for (;;) Line 2248  for (;;)
2248        break;        break;
2249    
2250        case OP_THEN_ARG:        case OP_THEN_ARG:
2251        code += code[1+LINK_SIZE];        code += code[1];
2252        break;        break;
2253        }        }
2254    
2255      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2256    
2257      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2258    
2259      /* 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
2260      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
2261      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2262    
2263  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2264      if (utf8) switch(c)      if (utf) switch(c)
2265        {        {
2266        case OP_CHAR:        case OP_CHAR:
2267        case OP_CHARI:        case OP_CHARI:
2268          case OP_NOT:
2269          case OP_NOTI:
2270        case OP_EXACT:        case OP_EXACT:
2271        case OP_EXACTI:        case OP_EXACTI:
2272          case OP_NOTEXACT:
2273          case OP_NOTEXACTI:
2274        case OP_UPTO:        case OP_UPTO:
2275        case OP_UPTOI:        case OP_UPTOI:
2276          case OP_NOTUPTO:
2277          case OP_NOTUPTOI:
2278        case OP_MINUPTO:        case OP_MINUPTO:
2279        case OP_MINUPTOI:        case OP_MINUPTOI:
2280          case OP_NOTMINUPTO:
2281          case OP_NOTMINUPTOI:
2282        case OP_POSUPTO:        case OP_POSUPTO:
2283        case OP_POSUPTOI:        case OP_POSUPTOI:
2284          case OP_NOTPOSUPTO:
2285          case OP_NOTPOSUPTOI:
2286        case OP_STAR:        case OP_STAR:
2287        case OP_STARI:        case OP_STARI:
2288          case OP_NOTSTAR:
2289          case OP_NOTSTARI:
2290        case OP_MINSTAR:        case OP_MINSTAR:
2291        case OP_MINSTARI:        case OP_MINSTARI:
2292          case OP_NOTMINSTAR:
2293          case OP_NOTMINSTARI:
2294        case OP_POSSTAR:        case OP_POSSTAR:
2295        case OP_POSSTARI:        case OP_POSSTARI:
2296          case OP_NOTPOSSTAR:
2297          case OP_NOTPOSSTARI:
2298        case OP_PLUS:        case OP_PLUS:
2299        case OP_PLUSI:        case OP_PLUSI:
2300          case OP_NOTPLUS:
2301          case OP_NOTPLUSI:
2302        case OP_MINPLUS:        case OP_MINPLUS:
2303        case OP_MINPLUSI:        case OP_MINPLUSI:
2304          case OP_NOTMINPLUS:
2305          case OP_NOTMINPLUSI:
2306        case OP_POSPLUS:        case OP_POSPLUS:
2307        case OP_POSPLUSI:        case OP_POSPLUSI:
2308          case OP_NOTPOSPLUS:
2309          case OP_NOTPOSPLUSI:
2310        case OP_QUERY:        case OP_QUERY:
2311        case OP_QUERYI:        case OP_QUERYI:
2312          case OP_NOTQUERY:
2313          case OP_NOTQUERYI:
2314        case OP_MINQUERY:        case OP_MINQUERY:
2315        case OP_MINQUERYI:        case OP_MINQUERYI:
2316          case OP_NOTMINQUERY:
2317          case OP_NOTMINQUERYI:
2318        case OP_POSQUERY:        case OP_POSQUERY:
2319        case OP_POSQUERYI:        case OP_POSQUERYI:
2320        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_NOTPOSQUERY:
2321          case OP_NOTPOSQUERYI:
2322          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2323        break;        break;
2324        }        }
2325  #else  #else
2326      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2327  #endif  #endif
2328      }      }
2329    }    }
# Line 1946  bracket whose current branch will alread Line 2346  bracket whose current branch will alread
2346  Arguments:  Arguments:
2347    code        points to start of search    code        points to start of search
2348    endcode     points to where to stop    endcode     points to where to stop
2349    utf8        TRUE if in UTF8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2350    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2351    
2352  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2353  */  */
2354    
2355  static BOOL  static BOOL
2356  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2357    compile_data *cd)    BOOL utf, compile_data *cd)
2358  {  {
2359  register int c;  register pcre_uchar c;
2360  for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2361       code < endcode;       code < endcode;
2362       code = first_significant_code(code + _pcre_OP_lengths[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2363    {    {
2364    const uschar *ccode;    const pcre_uchar *ccode;
2365    
2366    c = *code;    c = *code;
2367    
# Line 1978  for (code = first_significant_code(code Line 2378  for (code = first_significant_code(code
2378    /* For a recursion/subroutine call, if its end has been reached, which    /* For a recursion/subroutine call, if its end has been reached, which
2379    implies a backward reference subroutine call, we can scan it. If it's a    implies a backward reference subroutine call, we can scan it. If it's a
2380    forward reference subroutine call, we can't. To detect forward reference    forward reference subroutine call, we can't. To detect forward reference
2381    we have to scan up the list that is kept in the workspace. This function is    we have to scan up the list that is kept in the workspace. This function is
2382    called only when doing the real compile, not during the pre-compile that    called only when doing the real compile, not during the pre-compile that
2383    measures the size of the compiled pattern. */    measures the size of the compiled pattern. */
2384    
2385    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2386      {      {
2387      const uschar *scode;      const pcre_uchar *scode;
2388      BOOL empty_branch;      BOOL empty_branch;
2389    
2390      /* Test for forward reference */      /* Test for forward reference */
2391    
2392      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2393        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2394    
2395      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2396    
2397      empty_branch = FALSE;      empty_branch = FALSE;
2398      scode = cd->start_code + GET(code, 1);      scode = cd->start_code + GET(code, 1);
2399      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2400    
2401      /* Completed backwards reference */      /* Completed backwards reference */
2402    
2403      do      do
2404        {        {
2405        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd))
2406          {          {
2407          empty_branch = TRUE;          empty_branch = TRUE;
2408          break;          break;
# Line 2010  for (code = first_significant_code(code Line 2410  for (code = first_significant_code(code
2410        scode += GET(scode, 1);        scode += GET(scode, 1);
2411        }        }
2412      while (*scode == OP_ALT);      while (*scode == OP_ALT);
2413    
2414      if (!empty_branch) return FALSE;  /* All branches are non-empty */      if (!empty_branch) return FALSE;  /* All branches are non-empty */
2415      continue;      continue;
2416      }      }
# Line 2020  for (code = first_significant_code(code Line 2420  for (code = first_significant_code(code
2420    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2421        c == OP_BRAPOSZERO)        c == OP_BRAPOSZERO)
2422      {      {
2423      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2424      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
2425      c = *code;      c = *code;
2426      continue;      continue;
# Line 2041  for (code = first_significant_code(code Line 2441  for (code = first_significant_code(code
2441    
2442    if (c == OP_BRA  || c == OP_BRAPOS ||    if (c == OP_BRA  || c == OP_BRAPOS ||
2443        c == OP_CBRA || c == OP_CBRAPOS ||        c == OP_CBRA || c == OP_CBRAPOS ||
2444        c == OP_ONCE || c == OP_COND)        c == OP_ONCE || c == OP_ONCE_NC ||
2445          c == OP_COND)
2446      {      {
2447      BOOL empty_branch;      BOOL empty_branch;
2448      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 2057  for (code = first_significant_code(code Line 2458  for (code = first_significant_code(code
2458        empty_branch = FALSE;        empty_branch = FALSE;
2459        do        do
2460          {          {
2461          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
2462            empty_branch = TRUE;            empty_branch = TRUE;
2463          code += GET(code, 1);          code += GET(code, 1);
2464          }          }
# Line 2075  for (code = first_significant_code(code Line 2476  for (code = first_significant_code(code
2476      {      {
2477      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2478      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2479      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2480      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"
2481      here. */      here. */
2482    
2483  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2484      case OP_XCLASS:      case OP_XCLASS:
2485      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2486      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2087  for (code = first_significant_code(code Line 2488  for (code = first_significant_code(code
2488    
2489      case OP_CLASS:      case OP_CLASS:
2490      case OP_NCLASS:      case OP_NCLASS:
2491      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2492    
2493  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2494      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2495  #endif  #endif
2496    
# Line 2162  for (code = first_significant_code(code Line 2563  for (code = first_significant_code(code
2563      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2564      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2565      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2566      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2567          code += 2;
2568      break;      break;
2569    
2570      /* End of branch */      /* End of branch */
# Line 2177  for (code = first_significant_code(code Line 2579  for (code = first_significant_code(code
2579      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2580      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2581    
2582  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2583      case OP_STAR:      case OP_STAR:
2584      case OP_STARI:      case OP_STARI:
2585      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2190  for (code = first_significant_code(code Line 2592  for (code = first_significant_code(code
2592      case OP_MINQUERYI:      case OP_MINQUERYI:
2593      case OP_POSQUERY:      case OP_POSQUERY:
2594      case OP_POSQUERYI:      case OP_POSQUERYI:
2595      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2596      break;      break;
2597    
2598      case OP_UPTO:      case OP_UPTO:
# Line 2199  for (code = first_significant_code(code Line 2601  for (code = first_significant_code(code
2601      case OP_MINUPTOI:      case OP_MINUPTOI:
2602      case OP_POSUPTO:      case OP_POSUPTO:
2603      case OP_POSUPTOI:      case OP_POSUPTOI:
2604      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);
2605      break;      break;
2606  #endif  #endif
2607    
# Line 2213  for (code = first_significant_code(code Line 2615  for (code = first_significant_code(code
2615      break;      break;
2616    
2617      case OP_THEN_ARG:      case OP_THEN_ARG:
2618      code += code[1+LINK_SIZE];      code += code[1];
2619      break;      break;
2620    
2621      /* None of the remaining opcodes are required to match a character. */      /* None of the remaining opcodes are required to match a character. */
# Line 2236  return TRUE; Line 2638  return TRUE;
2638  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
2639  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,
2640  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.
2641  This function is called only during the real compile, not during the  This function is called only during the real compile, not during the
2642  pre-compile.  pre-compile.
2643    
2644  Arguments:  Arguments:
2645    code        points to start of the recursion    code        points to start of the recursion
2646    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2647    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2648    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2649    cd          pointers to tables etc    cd          pointers to tables etc
2650    
2651  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2652  */  */
2653    
2654  static BOOL  static BOOL
2655  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2656    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2657  {  {
2658  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2659    {    {
2660    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
2661      return FALSE;      return FALSE;
2662    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2663    }    }
# Line 2288  where Perl recognizes it as the POSIX cl Line 2690  where Perl recognizes it as the POSIX cl
2690  "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,
2691  I think.  I think.
2692    
2693    A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.
2694    It seems that the appearance of a nested POSIX class supersedes an apparent
2695    external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or
2696    a digit.
2697    
2698    In Perl, unescaped square brackets may also appear as part of class names. For
2699    example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for
2700    [:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not
2701    seem right at all. PCRE does not allow closing square brackets in POSIX class
2702    names.
2703    
2704  Arguments:  Arguments:
2705    ptr      pointer to the initial [    ptr      pointer to the initial [
2706    endptr   where to return the end pointer    endptr   where to return the end pointer
# Line 2296  Returns:   TRUE or FALSE Line 2709  Returns:   TRUE or FALSE
2709  */  */
2710    
2711  static BOOL  static BOOL
2712  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2713  {  {
2714  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2715  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2716  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != 0; ptr++)
2717    {    {
2718    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2719        ptr++;
2720      else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
2721      else
2722      {      {
     if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;  
2723      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2724        {        {
2725        *endptr = ptr;        *endptr = ptr;
2726        return TRUE;        return TRUE;
2727        }        }
2728        if (*ptr == CHAR_LEFT_SQUARE_BRACKET &&
2729             (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2730              ptr[1] == CHAR_EQUALS_SIGN) &&
2731            check_posix_syntax(ptr, endptr))
2732          return FALSE;
2733      }      }
2734    }    }
2735  return FALSE;  return FALSE;
# Line 2333  Returns:     a value representing the na Line 2753  Returns:     a value representing the na
2753  */  */
2754    
2755  static int  static int
2756  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2757  {  {
2758  const char *pn = posix_names;  const char *pn = posix_names;
2759  register int yield = 0;  register int yield = 0;
2760  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2761    {    {
2762    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2763      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2764    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2765    yield++;    yield++;
2766    }    }
# Line 2372  value in the reference (which is a group Line 2792  value in the reference (which is a group
2792  Arguments:  Arguments:
2793    group      points to the start of the group    group      points to the start of the group
2794    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2795    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2796    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2797    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
2798    
# Line 2380  Returns:     nothing Line 2800  Returns:     nothing
2800  */  */
2801    
2802  static void  static void
2803  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2804    uschar *save_hwm)    pcre_uchar *save_hwm)
2805  {  {
2806  uschar *ptr = group;  pcre_uchar *ptr = group;
2807    
2808  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2809    {    {
2810    int offset;    int offset;
2811    uschar *hc;    pcre_uchar *hc;
2812    
2813    /* 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
2814    reference. */    reference. */
2815    
2816    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2817      {      {
2818      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2819      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2820        {        {
2821        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2408  while ((ptr = (uschar *)find_recurse(ptr Line 2828  while ((ptr = (uschar *)find_recurse(ptr
2828    
2829    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2830      {      {
2831      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2832      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2833      }      }
2834    
# Line 2433  Arguments: Line 2853  Arguments:
2853  Returns:         new code pointer  Returns:         new code pointer
2854  */  */
2855    
2856  static uschar *  static pcre_uchar *
2857  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2858  {  {
2859  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2860  *code++ = 255;  *code++ = 255;
2861  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2862  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2863  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2864  }  }
2865    
2866    
# Line 2462  Returns:             nothing Line 2882  Returns:             nothing
2882  */  */
2883    
2884  static void  static void
2885  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2886  {  {
2887  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2888  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2476  PUT(previous_callout, 2 + LINK_SIZE, len Line 2896  PUT(previous_callout, 2 + LINK_SIZE, len
2896  *************************************************/  *************************************************/
2897    
2898  /* 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
2899  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
2900  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
2901  start address.  start address. A character with multiple other cases is returned on its own
2902    with a special return value.
2903    
2904  Arguments:  Arguments:
2905    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2486  Arguments: Line 2907  Arguments:
2907    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2908    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2909    
2910  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2911                   0 when a range is returned
2912                  >0 the CASESET offset for char with multiple other cases
2913                    in this case, ocptr contains the original
2914  */  */
2915    
2916  static BOOL  static int
2917  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2918    unsigned int *odptr)    pcre_uint32 *odptr)
2919  {  {
2920  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2921    unsigned int co;
2922    
2923    /* Find the first character that has an other case. If it has multiple other
2924    cases, return its case offset value. */
2925    
2926  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2927    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2928      if ((co = UCD_CASESET(c)) != 0)
2929        {
2930        *ocptr = c++;   /* Character that has the set */
2931        *cptr = c;      /* Rest of input range */
2932        return (int)co;
2933        }
2934      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2935      }
2936    
2937  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2938    
2939  *ocptr = othercase;  *ocptr = othercase;
2940  next = othercase + 1;  next = othercase + 1;
# Line 2509  for (++c; c <= d; c++) Line 2945  for (++c; c <= d; c++)
2945    next++;    next++;
2946    }    }
2947    
2948  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2949  *cptr = c;  *cptr = c;             /* Rest of input range */
2950    return 0;
 return TRUE;  
2951  }  }
2952    
2953    
# Line 2534  Returns:       TRUE if auto-possessifyin Line 2969  Returns:       TRUE if auto-possessifyin
2969  */  */
2970    
2971  static BOOL  static BOOL
2972  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)
2973  {  {
2974    #ifdef SUPPORT_UCP
2975    const pcre_uint32 *p;
2976    #endif
2977    
2978  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2979    
2980  switch(ptype)  switch(ptype)
2981    {    {
2982    case PT_LAMP:    case PT_LAMP:
# Line 2545  switch(ptype) Line 2985  switch(ptype)
2985            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
2986    
2987    case PT_GC:    case PT_GC:
2988    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2989    
2990    case PT_PC:    case PT_PC:
2991    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2556  switch(ptype) Line 2996  switch(ptype)
2996    /* These are specials */    /* These are specials */
2997    
2998    case PT_ALNUM:    case PT_ALNUM:
2999    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3000            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
3001    
3002    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
3003    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3004            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
3005            == negated;            == negated;
3006    
3007    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX 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_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
3010            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
3011            == negated;            == negated;
3012    
3013    case PT_WORD:    case PT_WORD:
3014    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3015            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3016            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3017    
3018    #ifdef SUPPORT_UCP
3019      case PT_CLIST:
3020      p = PRIV(ucd_caseless_sets) + prop->caseset;
3021      for (;;)
3022        {
3023        if ((unsigned int)c < *p) return !negated;
3024        if ((unsigned int)c == *p++) return negated;
3025        }
3026      break;  /* Control never reaches here */
3027    #endif
3028    }    }
3029    
3030  return FALSE;  return FALSE;
3031  }  }
3032  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2591  sense to automatically possessify the re Line 3043  sense to automatically possessify the re
3043    
3044  Arguments:  Arguments:
3045    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3046    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3047    ptr           next character in pattern    ptr           next character in pattern
3048    options       options bits    options       options bits
3049    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2600  Returns:        TRUE if possessifying is Line 3052  Returns:        TRUE if possessifying is
3052  */  */
3053    
3054  static BOOL  static BOOL
3055  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3056    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3057  {  {
3058  int c, next;  pcre_uint32 c = NOTACHAR;
3059  int op_code = *previous++;  pcre_uint32 next;
3060    int escape;
3061    pcre_uchar op_code = *previous++;
3062    
3063  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3064    
# Line 2612  if ((options & PCRE_EXTENDED) != 0) Line 3066  if ((options & PCRE_EXTENDED) != 0)
3066    {    {
3067    for (;;)    for (;;)
3068      {      {
3069      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3070      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3071        {        {
3072        ptr++;        ptr++;
# Line 2620  if ((options & PCRE_EXTENDED) != 0) Line 3074  if ((options & PCRE_EXTENDED) != 0)
3074          {          {
3075          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3076          ptr++;          ptr++;
3077  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3078          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3079  #endif  #endif
3080          }          }
3081        }        }
# Line 2635  value is a character, a negative value i Line 3089  value is a character, a negative value i
3089  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3090    {    {
3091    int temperrorcode = 0;    int temperrorcode = 0;
3092    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3093    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3094    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3095    }    }
3096    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
3097    {    {
3098  #ifdef SUPPORT_UTF8    escape = 0;
3099    if (utf8) { GETCHARINC(next, ptr); } else  #ifdef SUPPORT_UTF
3100      if (utf) { GETCHARINC(next, ptr); } else
3101  #endif  #endif
3102    next = *ptr++;    next = *ptr++;
3103    }    }
   
3104  else return FALSE;  else return FALSE;
3105    
3106  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2656  if ((options & PCRE_EXTENDED) != 0) Line 3109  if ((options & PCRE_EXTENDED) != 0)
3109    {    {
3110    for (;;)    for (;;)
3111      {      {
3112      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3113      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3114        {        {
3115        ptr++;        ptr++;
# Line 2664  if ((options & PCRE_EXTENDED) != 0) Line 3117  if ((options & PCRE_EXTENDED) != 0)
3117          {          {
3118          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3119          ptr++;          ptr++;
3120  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3121          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3122  #endif  #endif
3123          }          }
3124        }        }
# Line 2676  if ((options & PCRE_EXTENDED) != 0) Line 3129  if ((options & PCRE_EXTENDED) != 0)
3129  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3130    
3131  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3132    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)
3133      return FALSE;      return FALSE;
3134    
3135  /* 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. */  
3136    
3137  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3138        op_code == OP_NOT || op_code == OP_NOTI)
3139    {    {
3140    case OP_CHAR:  #ifdef SUPPORT_UTF
 #ifdef SUPPORT_UTF8  
3141    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3142  #else  #else
3143    c = *previous;    c = *previous;
3144  #endif  #endif
3145    return c != next;    }
3146    
3147    /* For CHARI (caseless character) we must check the other case. If we have  /* Now compare the next item with the previous opcode. First, handle cases when
3148    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3149    
3150    case OP_CHARI:  if (escape == 0)
3151  #ifdef SUPPORT_UTF8    {
3152    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3153  #else    case, which maps to the special PT_CLIST property. Check this first. */
3154    c = *previous;  
3155    #ifdef SUPPORT_UCP
3156      if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3157        {
3158        unsigned int ocs = UCD_CASESET(next);
3159        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3160        }
3161  #endif  #endif
3162    if (c == next) return FALSE;  
3163  #ifdef SUPPORT_UTF8    switch(op_code)
   if (utf8)  
3164      {      {
3165      unsigned int othercase;      case OP_CHAR:
3166      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3167    
3168        /* For CHARI (caseless character) we must check the other case. If we have
3169        Unicode property support, we can use it to test the other case of
3170        high-valued characters. We know that next can have only one other case,
3171        because multi-other-case characters are dealt with above. */
3172    
3173        case OP_CHARI:
3174        if (c == next) return FALSE;
3175    #ifdef SUPPORT_UTF
3176        if (utf)
3177          {
3178          pcre_uint32 othercase;
3179          if (next < 128) othercase = cd->fcc[next]; else
3180  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3181      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3182  #else  #else
3183      othercase = NOTACHAR;        othercase = NOTACHAR;
3184  #endif  #endif
3185      return (unsigned int)c != othercase;        return c != othercase;
3186      }        }
3187    else      else
3188  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3189    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3190    
3191    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These      case OP_NOT:
3192    opcodes are not used for multi-byte characters, because they are coded using      return c == next;
3193    an XCLASS instead. */  
3194        case OP_NOTI:
3195    case OP_NOT:      if (c == next) return TRUE;
3196    return (c = *previous) == next;  #ifdef SUPPORT_UTF
3197        if (utf)
3198    case OP_NOTI:        {
3199    if ((c = *previous) == next) return TRUE;        pcre_uint32 othercase;
3200  #ifdef SUPPORT_UTF8        if (next < 128) othercase = cd->fcc[next]; else
   if (utf8)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3201  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3202      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE(next);
3203  #else  #else
3204      othercase = NOTACHAR;        othercase = NOTACHAR;
3205  #endif  #endif
3206      return (unsigned int)c == othercase;        return c == othercase;
3207      }        }
3208    else      else
3209  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3210    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3211    
3212    /* 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.
3213    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. */
3214    
3215    case OP_DIGIT:      case OP_DIGIT:
3216    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3217    
3218    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3219    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3220    
3221    case OP_WHITESPACE:      case OP_WHITESPACE:
3222    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3223    
3224    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3225    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3226    
3227    case OP_WORDCHAR:      case OP_WORDCHAR:
3228    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3229    
3230    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3231    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3232    
3233    case OP_HSPACE:      case OP_HSPACE:
3234    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3235    switch(next)      switch(next)
3236      {        {
3237      case 0x09:        HSPACE_CASES:
3238      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;  
     }  
3239    
3240    case OP_ANYNL:        default:
3241    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3242    case OP_NOT_VSPACE:        }
3243    switch(next)  
3244      {      case OP_ANYNL:
3245      case 0x0a:      case OP_VSPACE:
3246      case 0x0b:      case OP_NOT_VSPACE:
3247      case 0x0c:      switch(next)
3248      case 0x0d:        {
3249      case 0x85:        VSPACE_CASES:
3250      case 0x2028:        return op_code == OP_NOT_VSPACE;
3251      case 0x2029:  
3252      return op_code == OP_NOT_VSPACE;        default:
3253      default:        return op_code != OP_NOT_VSPACE;
3254      return op_code != OP_NOT_VSPACE;        }
     }  
3255    
3256  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3257    case OP_PROP:      case OP_PROP:
3258    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3259    
3260    case OP_NOTPROP:      case OP_NOTPROP:
3261    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3262  #endif  #endif
3263    
3264    default:      default:
3265    return FALSE;      return FALSE;
3266        }
3267    }    }
3268    
   
3269  /* 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
3270  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
3271  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 2833  switch(op_code) Line 3276  switch(op_code)
3276    {    {
3277    case OP_CHAR:    case OP_CHAR:
3278    case OP_CHARI:    case OP_CHARI:
3279  #ifdef SUPPORT_UTF8    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3280      {      {
3281      case ESC_d:      case ESC_d:
3282      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3283    
3284      case ESC_D:      case ESC_D:
3285      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3286    
3287      case ESC_s:      case ESC_s:
3288      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3289    
3290      case ESC_S:      case ESC_S:
3291      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3292    
3293      case ESC_w:      case ESC_w:
3294      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3295    
3296      case ESC_W:      case ESC_W:
3297      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3298    
3299      case ESC_h:      case ESC_h:
3300      case ESC_H:      case ESC_H:
3301      switch(c)      switch(c)
3302        {        {
3303        case 0x09:        HSPACE_CASES:
3304        case 0x20:        return escape != ESC_h;
3305        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;  
3306        default:        default:
3307        return -next == ESC_h;        return escape == ESC_h;
3308        }        }
3309    
3310      case ESC_v:      case ESC_v:
3311      case ESC_V:      case ESC_V:
3312      switch(c)      switch(c)
3313        {        {
3314        case 0x0a:        VSPACE_CASES:
3315        case 0x0b:        return escape != ESC_v;
3316        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3317        default:        default:
3318        return -next == ESC_v;        return escape == ESC_v;
3319        }        }
3320    
3321      /* 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
3322      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3323      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3324    
3325  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3326      case ESC_du:      case ESC_du:
# Line 2915  switch(op_code) Line 3331  switch(op_code)
3331      case ESC_SU:      case ESC_SU:
3332        {        {
3333        int temperrorcode = 0;        int temperrorcode = 0;
3334        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3335        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3336        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3337        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3338        }        }
# Line 2925  switch(op_code) Line 3341  switch(op_code)
3341      case ESC_p:      case ESC_p:
3342      case ESC_P:      case ESC_P:
3343        {        {
3344        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3345          int errorcodeptr;
3346        BOOL negated;        BOOL negated;
3347    
3348        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3349        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3350        if (ptype < 0) return FALSE;          return FALSE;
3351        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3352    
3353        /* 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 2938  switch(op_code) Line 3355  switch(op_code)
3355        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. */
3356    
3357        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3358          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)
3359            return FALSE;            return FALSE;
3360    
3361        /* Do the property check. */        /* Do the property check. */
3362    
3363        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3364        }        }
3365  #endif  #endif
3366    
# Line 2958  switch(op_code) Line 3375  switch(op_code)
3375    these op-codes are never generated.) */    these op-codes are never generated.) */
3376    
3377    case OP_DIGIT:    case OP_DIGIT:
3378    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3379           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3380    
3381    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3382    return next == -ESC_d;    return escape == ESC_d;
3383    
3384    case OP_WHITESPACE:    case OP_WHITESPACE:
3385    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3386    
3387    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3388    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3389    
3390    case OP_HSPACE:    case OP_HSPACE:
3391    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3392           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3393    
3394    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3395    return next == -ESC_h;    return escape == ESC_h;
3396    
3397    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3398    case OP_ANYNL:    case OP_ANYNL:
3399    case OP_VSPACE:    case OP_VSPACE:
3400    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3401    
3402    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3403    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3404    
3405    case OP_WORDCHAR:    case OP_WORDCHAR:
3406    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3407           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3408    
3409    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3410    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3411    
3412    default:    default:
3413    return FALSE;    return FALSE;
# Line 3002  switch(op_code) Line 3419  switch(op_code)
3419    
3420    
3421  /*************************************************  /*************************************************
3422    *        Add a character or range to a class     *
3423    *************************************************/
3424    
3425    /* This function packages up the logic of adding a character or range of
3426    characters to a class. The character values in the arguments will be within the
3427    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3428    mutually recursive with the function immediately below.
3429    
3430    Arguments:
3431      classbits     the bit map for characters < 256
3432      uchardptr     points to the pointer for extra data
3433      options       the options word
3434      cd            contains pointers to tables etc.
3435      start         start of range character
3436      end           end of range character
3437    
3438    Returns:        the number of < 256 characters added
3439                    the pointer to extra data is updated
3440    */
3441    
3442    static int
3443    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3444      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3445    {
3446    pcre_uint32 c;
3447    int n8 = 0;
3448    
3449    /* If caseless matching is required, scan the range and process alternate
3450    cases. In Unicode, there are 8-bit characters that have alternate cases that
3451    are greater than 255 and vice-versa. Sometimes we can just extend the original
3452    range. */
3453    
3454    if ((options & PCRE_CASELESS) != 0)
3455      {
3456    #ifdef SUPPORT_UCP
3457      if ((options & PCRE_UTF8) != 0)
3458        {
3459        int rc;
3460        pcre_uint32 oc, od;
3461    
3462        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3463        c = start;
3464    
3465        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3466          {
3467          /* Handle a single character that has more than one other case. */
3468    
3469          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3470            PRIV(ucd_caseless_sets) + rc, oc);
3471    
3472          /* Do nothing if the other case range is within the original range. */
3473    
3474          else if (oc >= start && od <= end) continue;
3475    
3476          /* Extend the original range if there is overlap, noting that if oc < c, we
3477          can't have od > end because a subrange is always shorter than the basic
3478          range. Otherwise, use a recursive call to add the additional range. */
3479    
3480          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3481          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3482          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3483          }
3484        }
3485      else
3486    #endif  /* SUPPORT_UCP */
3487    
3488      /* Not UTF-mode, or no UCP */
3489    
3490      for (c = start; c <= end && c < 256; c++)
3491        {
3492        SETBIT(classbits, cd->fcc[c]);
3493        n8++;
3494        }
3495      }
3496    
3497    /* Now handle the original range. Adjust the final value according to the bit
3498    length - this means that the same lists of (e.g.) horizontal spaces can be used
3499    in all cases. */
3500    
3501    #if defined COMPILE_PCRE8
3502    #ifdef SUPPORT_UTF
3503      if ((options & PCRE_UTF8) == 0)
3504    #endif
3505      if (end > 0xff) end = 0xff;
3506    
3507    #elif defined COMPILE_PCRE16
3508    #ifdef SUPPORT_UTF
3509      if ((options & PCRE_UTF16) == 0)
3510    #endif
3511      if (end > 0xffff) end = 0xffff;
3512    
3513    #endif /* COMPILE_PCRE[8|16] */
3514    
3515    /* If all characters are less than 256, use the bit map. Otherwise use extra
3516    data. */
3517    
3518    if (end < 0x100)
3519      {
3520      for (c = start; c <= end; c++)
3521        {
3522        n8++;
3523        SETBIT(classbits, c);
3524        }
3525      }
3526    
3527    else
3528      {
3529      pcre_uchar *uchardata = *uchardptr;
3530    
3531    #ifdef SUPPORT_UTF
3532      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3533        {
3534        if (start < end)
3535          {
3536          *uchardata++ = XCL_RANGE;
3537          uchardata += PRIV(ord2utf)(start, uchardata);
3538          uchardata += PRIV(ord2utf)(end, uchardata);
3539          }
3540        else if (start == end)
3541          {
3542          *uchardata++ = XCL_SINGLE;
3543          uchardata += PRIV(ord2utf)(start, uchardata);
3544          }
3545        }
3546      else
3547    #endif  /* SUPPORT_UTF */
3548    
3549      /* Without UTF support, character values are constrained by the bit length,
3550      and can only be > 256 for 16-bit and 32-bit libraries. */
3551    
3552    #ifdef COMPILE_PCRE8
3553        {}
3554    #else
3555      if (start < end)
3556        {
3557        *uchardata++ = XCL_RANGE;
3558        *uchardata++ = start;
3559        *uchardata++ = end;
3560        }
3561      else if (start == end)
3562        {
3563        *uchardata++ = XCL_SINGLE;
3564        *uchardata++ = start;
3565        }
3566    #endif
3567    
3568      *uchardptr = uchardata;   /* Updata extra data pointer */
3569      }
3570    
3571    return n8;    /* Number of 8-bit characters */
3572    }
3573    
3574    
3575    
3576    
3577    /*************************************************
3578    *        Add a list of characters to a class     *
3579    *************************************************/
3580    
3581    /* This function is used for adding a list of case-equivalent characters to a
3582    class, and also for adding a list of horizontal or vertical whitespace. If the
3583    list is in order (which it should be), ranges of characters are detected and
3584    handled appropriately. This function is mutually recursive with the function
3585    above.
3586    
3587    Arguments:
3588      classbits     the bit map for characters < 256
3589      uchardptr     points to the pointer for extra data
3590      options       the options word
3591      cd            contains pointers to tables etc.
3592      p             points to row of 32-bit values, terminated by NOTACHAR
3593      except        character to omit; this is used when adding lists of
3594                      case-equivalent characters to avoid including the one we
3595                      already know about
3596    
3597    Returns:        the number of < 256 characters added
3598                    the pointer to extra data is updated
3599    */
3600    
3601    static int
3602    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3603      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3604    {
3605    int n8 = 0;
3606    while (p[0] < NOTACHAR)
3607      {
3608      int n = 0;
3609      if (p[0] != except)
3610        {
3611        while(p[n+1] == p[0] + n + 1) n++;
3612        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3613        }
3614      p += n + 1;
3615      }
3616    return n8;
3617    }
3618    
3619    
3620    
3621    /*************************************************
3622    *    Add characters not in a list to a class     *
3623    *************************************************/
3624    
3625    /* This function is used for adding the complement of a list of horizontal or
3626    vertical whitespace to a class. The list must be in order.
3627    
3628    Arguments:
3629      classbits     the bit map for characters < 256
3630      uchardptr     points to the pointer for extra data
3631      options       the options word
3632      cd            contains pointers to tables etc.
3633      p             points to row of 32-bit values, terminated by NOTACHAR
3634    
3635    Returns:        the number of < 256 characters added
3636                    the pointer to extra data is updated
3637    */
3638    
3639    static int
3640    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3641      int options, compile_data *cd, const pcre_uint32 *p)
3642    {
3643    BOOL utf = (options & PCRE_UTF8) != 0;
3644    int n8 = 0;
3645    if (p[0] > 0)
3646      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3647    while (p[0] < NOTACHAR)
3648      {
3649      while (p[1] == p[0] + 1) p++;
3650      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3651        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3652      p++;
3653      }
3654    return n8;
3655    }
3656    
3657    
3658    
3659    /*************************************************
3660  *           Compile one branch                   *  *           Compile one branch                   *
3661  *************************************************/  *************************************************/
3662    
# Line 3016  Arguments: Line 3671  Arguments:
3671    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3672    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3673    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3674    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3675    reqbyteptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3676      reqcharptr     place to put the last required character
3677      reqcharflagsptr place to put the last required character flags, or a negative number
3678    bcptr          points to current branch chain    bcptr          points to current branch chain
3679      cond_depth     conditional nesting depth
3680    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
3681    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
3682                   points to length accumulator during pre-compile phase                   points to length accumulator during pre-compile phase
# Line 3028  Returns:         TRUE on success Line 3686  Returns:         TRUE on success
3686  */  */
3687    
3688  static BOOL  static BOOL
3689  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3690    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3691      pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3692      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3693      branch_chain *bcptr, int cond_depth,
3694    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3695  {  {
3696  int repeat_type, op_type;  int repeat_type, op_type;
3697  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3698  int bravalue = 0;  int bravalue = 0;
3699  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3700  int firstbyte, reqbyte;  pcre_uint32 firstchar, reqchar;
3701  int zeroreqbyte, zerofirstbyte;  pcre_int32 firstcharflags, reqcharflags;
3702  int req_caseopt, reqvary, tempreqvary;  pcre_uint32 zeroreqchar, zerofirstchar;
3703    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3704    pcre_int32 req_caseopt, reqvary, tempreqvary;
3705  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3706  int after_manual_callout = 0;  int after_manual_callout = 0;
3707  int length_prevgroup = 0;  int length_prevgroup = 0;
3708  register int c;  register pcre_uint32 c;
3709  register uschar *code = *codeptr;  int escape;
3710  uschar *last_code = code;  register pcre_uchar *code = *codeptr;
3711  uschar *orig_code = code;  pcre_uchar *last_code = code;
3712  uschar *tempcode;  pcre_uchar *orig_code = code;
3713    pcre_uchar *tempcode;
3714  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3715  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3716  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3717  const uschar *tempptr;  const pcre_uchar *tempptr;
3718  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3719  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3720  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3721  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3722  uschar classbits[32];  pcre_uint8 classbits[32];
3723    
3724  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we
3725  must not do this for other options (e.g. PCRE_EXTENDED) because they may change  must not do this for other options (e.g. PCRE_EXTENDED) because they may change
3726  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3727    
3728  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3729  BOOL class_utf8;  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3730  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3731  uschar *class_utf8data;  #ifndef COMPILE_PCRE32
3732  uschar *class_utf8data_base;  pcre_uchar utf_chars[6];
3733  uschar utf8_char[6];  #endif
3734  #else  #else
3735  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3736  uschar *utf8_char = NULL;  #endif
3737    
3738    /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3739    class_uchardata always so that it can be passed to add_to_class() always,
3740    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3741    alternative calls for the different cases. */
3742    
3743    pcre_uchar *class_uchardata;
3744    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3745    BOOL xclass;
3746    pcre_uchar *class_uchardata_base;
3747  #endif  #endif
3748    
3749  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3083  greedy_non_default = greedy_default ^ 1; Line 3757  greedy_non_default = greedy_default ^ 1;
3757    
3758  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3759  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
3760  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
3761  find one.  find one.
3762    
3763  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
3764  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
3765  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3766  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3767    
3768  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3769    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3770    
3771  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3772  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
3773  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
3774  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3775    value. This is used only for ASCII characters. */
3776    
3777  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3778    
3779  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3780    
# Line 3110  for (;; ptr++) Line 3786  for (;; ptr++)
3786    BOOL is_quantifier;    BOOL is_quantifier;
3787    BOOL is_recurse;    BOOL is_recurse;
3788    BOOL reset_bracount;    BOOL reset_bracount;
3789    int class_charcount;    int class_has_8bitchar;
3790    int class_lastchar;    int class_one_char;
3791    int newoptions;    int newoptions;
3792    int recno;    int recno;
3793    int refsign;    int refsign;
3794    int skipbytes;    int skipbytes;
3795    int subreqbyte;    pcre_uint32 subreqchar, subfirstchar;
3796    int subfirstbyte;    pcre_int32 subreqcharflags, subfirstcharflags;
3797    int terminator;    int terminator;
3798    int mclength;    unsigned int mclength;
3799    uschar mcbuffer[8];    int tempbracount;
3800      pcre_uint32 ec;
3801      pcre_uchar mcbuffer[8];
3802    
3803    /* Get next byte in the pattern */    /* Get next character in the pattern */
3804    
3805    c = *ptr;    c = *ptr;
3806    
# Line 3144  for (;; ptr++) Line 3822  for (;; ptr++)
3822  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3823      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3824  #endif  #endif
3825      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */      if (code > cd->start_workspace + cd->workspace_size -
3826            WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3827        {        {
3828        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3829        goto FAILED;        goto FAILED;
# Line 3167  for (;; ptr++) Line 3846  for (;; ptr++)
3846        }        }
3847    
3848      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3849      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3850          (int)(code - last_code), c, c));
3851    
3852      /* 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
3853      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 3177  for (;; ptr++) Line 3857  for (;; ptr++)
3857        {        {
3858        if (previous > orig_code)        if (previous > orig_code)
3859          {          {
3860          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3861          code -= previous - orig_code;          code -= previous - orig_code;
3862          previous = orig_code;          previous = orig_code;
3863          }          }
# Line 3193  for (;; ptr++) Line 3873  for (;; ptr++)
3873    /* 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
3874    reference list. */    reference list. */
3875    
3876    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3877               WORK_SIZE_SAFETY_MARGIN)
3878      {      {
3879      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3880      goto FAILED;      goto FAILED;
# Line 3245  for (;; ptr++) Line 3926  for (;; ptr++)
3926    
3927    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
3928      {      {
3929      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
3930      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3931        {        {
3932        ptr++;        ptr++;
# Line 3253  for (;; ptr++) Line 3934  for (;; ptr++)
3934          {          {
3935          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3936          ptr++;          ptr++;
3937  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3938          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3939  #endif  #endif
3940          }          }
3941        if (*ptr != 0) continue;        if (*ptr != 0) continue;
# Line 3278  for (;; ptr++) Line 3959  for (;; ptr++)
3959      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
3960      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3961      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3962      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
3963      *reqbyteptr = reqbyte;      *firstcharflagsptr = firstcharflags;
3964        *reqcharptr = reqchar;
3965        *reqcharflagsptr = reqcharflags;
3966      *codeptr = code;      *codeptr = code;
3967      *ptrptr = ptr;      *ptrptr = ptr;
3968      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3303  for (;; ptr++) Line 3986  for (;; ptr++)
3986      previous = NULL;      previous = NULL;
3987      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3988        {        {
3989        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
3990        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3991        }        }
3992      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3315  for (;; ptr++) Line 3998  for (;; ptr++)
3998      break;      break;
3999    
4000      /* 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
4001      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4002    
4003      case CHAR_DOT:      case CHAR_DOT:
4004      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4005      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
4006      zeroreqbyte = reqbyte;      zerofirstcharflags = firstcharflags;
4007        zeroreqchar = reqchar;
4008        zeroreqcharflags = reqcharflags;
4009      previous = code;      previous = code;
4010      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4011      break;      break;
# Line 3375  for (;; ptr++) Line 4060  for (;; ptr++)
4060          {          {
4061          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
4062            ptr++;            ptr++;
4063          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)  
4064            ptr += 3;            ptr += 3;
4065          else          else
4066            break;            break;
# Line 3395  for (;; ptr++) Line 4079  for (;; ptr++)
4079          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4080        {        {
4081        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4082        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4083        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
4084          zerofirstcharflags = firstcharflags;
4085        break;        break;
4086        }        }
4087    
# Line 3406  for (;; ptr++) Line 4091  for (;; ptr++)
4091    
4092      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4093    
4094      /* 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:
4095      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 <
4096      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
4097        character. */
4098    
4099      class_charcount = 0;      class_has_8bitchar = 0;
4100      class_lastchar = -1;      class_one_char = 0;
4101    
4102      /* 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
4103      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
4104      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
4105      */      map. */
4106    
4107      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4108    
4109  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4110      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;
4111      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4112      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4113  #endif  #endif
4114    
4115      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
# Line 3432  for (;; ptr++) Line 4118  for (;; ptr++)
4118    
4119      if (c != 0) do      if (c != 0) do
4120        {        {
4121        const uschar *oldptr;        const pcre_uchar *oldptr;
4122    
4123  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4124        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
4125          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4126          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
4127          }          }
4128    #endif
4129    
4130        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4131          /* In the pre-compile phase, accumulate the length of any extra
4132        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
4133        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4134        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4135          however. */
4136        if (lengthptr != NULL)  
4137          {        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4138          *lengthptr += class_utf8data - class_utf8data_base;          {
4139          class_utf8data = class_utf8data_base;          xclass = TRUE;
4140            *lengthptr += class_uchardata - class_uchardata_base;
4141            class_uchardata = class_uchardata_base;
4142          }          }
   
4143  #endif  #endif
4144    
4145        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3478  for (;; ptr++) Line 4167  for (;; ptr++)
4167          {          {
4168          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
4169          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
4170          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
4171          uschar pbits[32];          pcre_uint8 pbits[32];
4172    
4173          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
4174            {            {
# Line 3506  for (;; ptr++) Line 4195  for (;; ptr++)
4195          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4196          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4197    
4198          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4199            posix_class = 0;            posix_class = 0;
4200    
4201          /* 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 3534  for (;; ptr++) Line 4223  for (;; ptr++)
4223          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
4224    
4225          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
4226            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
4227    
4228          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
4229    
# Line 3549  for (;; ptr++) Line 4238  for (;; ptr++)
4238              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4239            }            }
4240    
4241          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4242          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4243    
4244          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3565  for (;; ptr++) Line 4254  for (;; ptr++)
4254            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4255    
4256          ptr = tempptr + 1;          ptr = tempptr + 1;
4257          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 character. */
4258            class_has_8bitchar = 1;
4259            /* Every class contains at least two characters. */
4260            class_one_char = 2;
4261          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4262          }          }
4263    
4264        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
4265        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
4266        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
4267        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
4268        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_one_char bigger
4269        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
4270          as literal characters (by default), or are faulted if
4271        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4272    
4273        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4274          {          {
4275          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4276    
4277          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4278    
4279          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4280          else if (-c == ESC_Q)            /* Handle start of quoted string */            c = ec;
4281            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4282            else if (escape == ESC_N)            /* \N is not supported in a class */
4283              {
4284              *errorcodeptr = ERR71;
4285              goto FAILED;
4286              }
4287            else if (escape == ESC_Q)            /* Handle start of quoted string */
4288            {            {
4289            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4290              {              {
# Line 3592  for (;; ptr++) Line 4293  for (;; ptr++)
4293            else inescq = TRUE;            else inescq = TRUE;
4294            continue;            continue;
4295            }            }
4296          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4297    
4298          if (c < 0)          else
4299            {            {
4300            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4301            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
4302              class_has_8bitchar++;
4303              /* Every class contains at least two characters. */
4304              class_one_char += 2;
4305    
4306            switch (-c)            switch (escape)
4307              {              {
4308  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4309              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3609  for (;; ptr++) Line 4313  for (;; ptr++)
4313              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4314              case ESC_SU:              case ESC_SU:
4315              nestptr = ptr;              nestptr = ptr;
4316              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4317              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4318              continue;              continue;
4319  #endif  #endif
4320              case ESC_d:              case ESC_d:
# Line 3633  for (;; ptr++) Line 4337  for (;; ptr++)
4337    
4338              /* Perl 5.004 onwards omits VT from \s, but we must preserve it              /* Perl 5.004 onwards omits VT from \s, but we must preserve it
4339              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4340              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4341                EBCDIC, so we lazily just adjust the appropriate bit. */
4342    
4343              case ESC_s:              case ESC_s:
4344              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 3646  for (;; ptr++) Line 4351  for (;; ptr++)
4351              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4352              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4353              continue;              continue;
4354    
4355                /* The rest apply in both UCP and non-UCP cases. */
4356    
4357              case ESC_h:              case ESC_h:
4358              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4359              SETBIT(classbits, 0x20); /* SPACE */                PRIV(hspace_list), NOTACHAR);
             SETBIT(classbits, 0xa0); /* NSBP */  
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x1680, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x180e, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2000, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x200A, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x202f, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x205f, class_utf8data);  
               *class_utf8data++ = XCL_SINGLE;  
               class_utf8data += _pcre_ord2utf8(0x3000, class_utf8data);  
               }  
 #endif  
4360              continue;              continue;
4361    
4362              case ESC_H:              case ESC_H:
4363              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4364                {                cd, PRIV(hspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x09/8: x ^= 1 << (0x09%8); break;  
                 case 0x20/8: x ^= 1 << (0x20%8); break;  
                 case 0xa0/8: x ^= 1 << (0xa0%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
   
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x167f, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x1681, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x180d, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x180f, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x1fff, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x200B, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x202e, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2030, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x205e, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2060, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x2fff, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x3001, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);  
               }  
 #endif  
4365              continue;              continue;
4366    
4367              case ESC_v:              case ESC_v:
4368              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4369              SETBIT(classbits, 0x0b); /* VT */                PRIV(vspace_list), NOTACHAR);
             SETBIT(classbits, 0x0c); /* FF */  
             SETBIT(classbits, 0x0d); /* CR */  
             SETBIT(classbits, 0x85); /* NEL */  
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2028, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);  
               }  
 #endif  
4370              continue;              continue;
4371    
4372              case ESC_V:              case ESC_V:
4373              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4374                {                cd, PRIV(vspace_list));
               int x = 0xff;  
               switch (c)  
                 {  
                 case 0x0a/8: x ^= 1 << (0x0a%8);  
                              x ^= 1 << (0x0b%8);  
                              x ^= 1 << (0x0c%8);  
                              x ^= 1 << (0x0d%8);  
                              break;  
                 case 0x85/8: x ^= 1 << (0x85%8); break;  
                 default: break;  
                 }  
               classbits[c] |= x;  
               }  
   
 #ifdef SUPPORT_UTF8  
             if (utf8)  
               {  
               class_utf8 = TRUE;  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x0100, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x2027, class_utf8data);  
               *class_utf8data++ = XCL_RANGE;  
               class_utf8data += _pcre_ord2utf8(0x2029, class_utf8data);  
               class_utf8data += _pcre_ord2utf8(0x7fffffff, class_utf8data);  
               }  
 #endif  
4375              continue;              continue;
4376    
4377  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3768  for (;; ptr++) Line 4379  for (;; ptr++)
4379              case ESC_P:              case ESC_P:
4380                {                {
4381                BOOL negated;                BOOL negated;
4382                int pdata;                unsigned int ptype = 0, pdata = 0;
4383                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4384                if (ptype < 0) goto FAILED;                  goto FAILED;
4385                class_utf8 = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_utf8data++ = ((-c == ESC_p) != negated)?  
4386                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4387                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4388                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4389                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4390                continue;                continue;
4391                }                }
4392  #endif  #endif
# Line 3790  for (;; ptr++) Line 4400  for (;; ptr++)
4400                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4401                goto FAILED;                goto FAILED;
4402                }                }
4403              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4404              c = *ptr;              /* Get the final character and fall through */              class_one_char -= 2;     /* Undo the speculative increase. */
4405                c = *ptr;                /* Get the final character and fall through */
4406              break;              break;
4407              }              }
4408            }            }
4409    
4410          /* Fall through if we have a single character (c >= 0). This may be          /* Fall through if the escape just defined a single character (c >= 0).
4411          greater than 256 in UTF-8 mode. */          This may be greater than 256. */
4412    
4413            escape = 0;
4414    
4415          }   /* End of backslash handling */          }   /* End of backslash handling */
4416    
4417        /* A single character may be followed by '-' to form a range. However,        /* A character may be followed by '-' to form a range. However, Perl does
4418        Perl does not permit ']' to be the end of the range. A '-' character        not permit ']' to be the end of the range. A '-' character at the end is
4419        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4420        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4421    
4422        CHECK_RANGE:        CHECK_RANGE:
4423        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3812  for (;; ptr++) Line 4425  for (;; ptr++)
4425          inescq = FALSE;          inescq = FALSE;
4426          ptr += 2;          ptr += 2;
4427          }          }
   
4428        oldptr = ptr;        oldptr = ptr;
4429    
4430        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4431    
4432        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4433    
# Line 3823  for (;; ptr++) Line 4435  for (;; ptr++)
4435    
4436        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4437          {          {
4438          int d;          pcre_uint32 d;
4439          ptr += 2;          ptr += 2;
4440          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4441    
# Line 3838  for (;; ptr++) Line 4450  for (;; ptr++)
4450            inescq = TRUE;            inescq = TRUE;
4451            break;            break;
4452            }            }
4453    
4454            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4455            back the pointer and jump to handle the character that preceded it. */
4456    
4457          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4458            {            {
4459            ptr = oldptr;            ptr = oldptr;
4460            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4461            }            }
4462    
4463            /* Otherwise, we have a potential range; pick up the next character */
4464    
4465  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4466          if (utf8)          if (utf)
4467            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4468            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4469            }            }
# Line 3860  for (;; ptr++) Line 4477  for (;; ptr++)
4477    
4478          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4479            {            {
4480            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4481              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4482            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4483    
4484            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4485    
4486            if (d < 0)            if (descape != 0)
4487              {              {
4488              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4489                {                {
4490                ptr = oldptr;                ptr = oldptr;
4491                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4492                }                }
4493              }              }
4494            }            }
4495    
4496          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4497          one-character ranges */          one-character ranges. */
4498    
4499          if (d < c)          if (d < c)
4500            {            {
4501            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4502            goto FAILED;            goto FAILED;
4503            }            }
4504            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4505    
4506          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4507            cannot be done anymore. Any value greater than 1 indicates that there
4508          /* Remember \r or \n */          is more than one character. */
4509    
4510          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          class_one_char = 2;
   
         /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless  
         matching, we have to use an XCLASS with extra data items. Caseless  
         matching for characters > 127 is available only if UCP support is  
         available. */  
   
 #ifdef SUPPORT_UTF8  
         if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))  
           {  
           class_utf8 = TRUE;  
   
           /* With UCP support, we can find the other case equivalents of  
           the relevant characters. There may be several ranges. Optimize how  
           they fit with the basic range. */  
   
 #ifdef SUPPORT_UCP  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             unsigned int occ, ocd;  
             unsigned int cc = c;  
             unsigned int origd = d;  
             while (get_othercase_range(&cc, origd, &occ, &ocd))  
               {  
               if (occ >= (unsigned int)c &&  
                   ocd <= (unsigned int)d)  
                 continue;                          /* Skip embedded ranges */  
   
               if (occ < (unsigned int)c  &&  
                   ocd >= (unsigned int)c - 1)      /* Extend the basic range */  
                 {                                  /* if there is overlap,   */  
                 c = occ;                           /* noting that if occ < c */  
                 continue;                          /* we can't have ocd > d  */  
                 }                                  /* because a subrange is  */  
               if (ocd > (unsigned int)d &&  
                   occ <= (unsigned int)d + 1)      /* always shorter than    */  
                 {                                  /* the basic range.       */  
                 d = ocd;  
                 continue;  
                 }  
   
               if (occ == ocd)  
                 {  
                 *class_utf8data++ = XCL_SINGLE;  
                 }  
               else  
                 {  
                 *class_utf8data++ = XCL_RANGE;  
                 class_utf8data += _pcre_ord2utf8(occ, class_utf8data);  
                 }  
               class_utf8data += _pcre_ord2utf8(ocd, class_utf8data);  
               }  
             }  
 #endif  /* SUPPORT_UCP */  
   
           /* Now record the original range, possibly modified for UCP caseless  
           overlapping ranges. */  
   
           *class_utf8data++ = XCL_RANGE;  
           class_utf8data += _pcre_ord2utf8(c, class_utf8data);  
           class_utf8data += _pcre_ord2utf8(d, class_utf8data);  
   
           /* With UCP support, we are done. Without UCP support, there is no  
           caseless matching for UTF-8 characters > 127; we can use the bit map  
           for the smaller ones. */  
   
 #ifdef SUPPORT_UCP  
           continue;    /* With next character in the class */  
 #else  
           if ((options & PCRE_CASELESS) == 0 || c > 127) continue;  
   
           /* Adjust upper limit and fall through to set up the map */  
   
           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. */  
4511    
4512          if (lengthptr == NULL) for (; c <= d; c++)          /* Remember an explicit \r or \n, and add the range to the class. */
           {  
           classbits[c/8] |= (1 << (c&7));  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             int uc = cd->fcc[c];           /* flip case */  
             classbits[uc/8] |= (1 << (uc&7));  
             }  
           }  
4513    
4514            if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4515    
4516            class_has_8bitchar +=
4517              add_to_class(classbits, &class_uchardata, options, cd, c, d);
4518    
4519          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4520          }          }
4521    
4522        /* 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
4523        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
4524        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4525          increase it if it is already 2 or more ... just in case there's a class
4526        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4527    
4528        /* Handle a character that cannot go in the bit map */        CLASS_SINGLE_CHARACTER:
4529          if (class_one_char < 2) class_one_char++;
4530    
4531          /* If class_one_char is 1, we have the first single character in the
4532          class, and there have been no prior ranges, or XCLASS items generated by
4533          escapes. If this is the final character in the class, we can optimize by
4534          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4535          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4536          to be set. Otherwise, there can be no first char if this item is first,
4537          whatever repeat count may follow. In the case of reqchar, save the
4538          previous value for reinstating. */
4539    
4540  #ifdef SUPPORT_UTF8        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
       if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
4541          {          {
4542          class_utf8 = TRUE;          ptr++;
4543          *class_utf8data++ = XCL_SINGLE;          zeroreqchar = reqchar;
4544          class_utf8data += _pcre_ord2utf8(c, class_utf8data);          zeroreqcharflags = reqcharflags;
4545    
4546  #ifdef SUPPORT_UCP          if (negate_class)
         if ((options & PCRE_CASELESS) != 0)  
4547            {            {
4548            unsigned int othercase;  #ifdef SUPPORT_UCP
4549            if ((othercase = UCD_OTHERCASE(c)) != c)            int d;
4550    #endif
4551              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4552              zerofirstchar = firstchar;
4553              zerofirstcharflags = firstcharflags;
4554    
4555              /* For caseless UTF-8 mode when UCP support is available, check
4556              whether this character has more than one other case. If so, generate
4557              a special OP_NOTPROP item instead of OP_NOTI. */
4558    
4559    #ifdef SUPPORT_UCP
4560              if (utf && (options & PCRE_CASELESS) != 0 &&
4561                  (d = UCD_CASESET(c)) != 0)
4562              {              {
4563              *class_utf8data++ = XCL_SINGLE;              *code++ = OP_NOTPROP;
4564              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);              *code++ = PT_CLIST;
4565                *code++ = d;
4566              }              }
4567              else
4568    #endif
4569              /* Char has only one other case, or UCP not available */
4570    
4571                {
4572                *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4573    #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4574                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4575                  code += PRIV(ord2utf)(c, code);
4576                else
4577    #endif
4578                  *code++ = c;
4579                }
4580    
4581              /* We are finished with this character class */
4582    
4583              goto END_CLASS;
4584            }            }
 #endif  /* SUPPORT_UCP */  
4585    
4586          }          /* For a single, positive character, get the value into mcbuffer, and
4587        else          then we can handle this with the normal one-character code. */
 #endif  /* SUPPORT_UTF8 */  
4588    
4589        /* Handle a single-byte character */  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4590          {          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4591          classbits[c/8] |= (1 << (c&7));            mclength = PRIV(ord2utf)(c, mcbuffer);
4592          if ((options & PCRE_CASELESS) != 0)          else
4593    #endif
4594            {            {
4595            c = cd->fcc[c];   /* flip case */            mcbuffer[0] = c;
4596            classbits[c/8] |= (1 << (c&7));            mclength = 1;
4597            }            }
4598          class_charcount++;          goto ONE_CHAR;
4599          class_lastchar = c;          }       /* End of 1-char optimization */
4600          }  
4601          /* There is more than one character in the class, or an XCLASS item
4602          has been generated. Add this character to the class. */
4603    
4604          class_has_8bitchar +=
4605            add_to_class(classbits, &class_uchardata, options, cd, c, c);
4606        }        }
4607    
4608      /* Loop until ']' reached. This "while" is the end of the "do" far above.      /* Loop until ']' reached. This "while" is the end of the "do" far above.
# Line 4050  for (;; ptr++) Line 4622  for (;; ptr++)
4622        goto FAILED;        goto FAILED;
4623        }        }
4624    
4625      /* If class_charcount is 1, we saw precisely one character whose value is      /* We will need an XCLASS if data has been placed in class_uchardata. In
4626      less than 256. As long as there were no characters >= 128 and there was no      the second phase this is a sufficient test. However, in the pre-compile
4627      use of \p or \P, in other words, no use of any XCLASS features, we can      phase, class_uchardata gets emptied to prevent workspace overflow, so it
4628      optimize.      only if the very last character in the class needs XCLASS will it contain
4629        anything at this point. For this reason, xclass gets set TRUE above when
4630      In UTF-8 mode, we can optimize the negative case only if there were no      uchar_classdata is emptied, and that's why this code is the way it is here
4631      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      instead of just doing a test on class_uchardata below. */
4632      operate on single-bytes characters only. This is an historical hangover.  
4633      Maybe one day we can tidy these opcodes to handle multi-byte characters.  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4634        if (class_uchardata > class_uchardata_base) xclass = TRUE;
     The optimization throws away the bit map. We turn the item into a  
     1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
     Note that OP_NOT[I] does not support multibyte characters. In the positive  
     case, it can cause firstbyte to be set. Otherwise, there can be no first  
     char if this item is first, whatever repeat count may follow. In the case  
     of reqbyte, save the previous value for reinstating. */  
   
 #ifdef SUPPORT_UTF8  
     if (class_charcount == 1 && !class_utf8 &&  
       (!utf8 || !negate_class || class_lastchar < 128))  
 #else  
     if (class_charcount == 1)  
 #endif  
       {  
       zeroreqbyte = reqbyte;  
   
       /* The OP_NOT[I] opcodes work on one-byte characters only. */  
   
       if (negate_class)  
         {  
         if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;  
         zerofirstbyte = firstbyte;  
         *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;  
         *code++ = class_lastchar;  
         break;  
         }  
   
       /* For a single, positive character, get the value into mcbuffer, and  
       then we can handle this with the normal one-character code. */  
   
 #ifdef SUPPORT_UTF8  
       if (utf8 && class_lastchar > 127)  
         mclength = _pcre_ord2utf8(class_lastchar, mcbuffer);