/[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 574 by ph10, Sat Nov 20 17:47:27 2010 UTC revision 1060 by chpe, Tue Oct 16 15:53:57 2012 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2010 University of Cambridge             Copyright (c) 1997-2012 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. */  library. We do not need to select pcre16_printint.c specially, because the
59    COMPILE_PCREx macro will already be appropriately set. */
60    
61  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
62  #include "pcre_printint.src"  /* pcre_printint.c should not include any headers */
63    #define PCRE_INCLUDED
64    #include "pcre_printint.c"
65    #undef PCRE_INCLUDED
66  #endif  #endif
67    
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 73  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL,
88        int, int, int *, int *, branch_chain *, compile_data *, int *);
89    
90    
91    
92  /*************************************************  /*************************************************
93  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 88  so this number is very generous. Line 103  so this number is very generous.
103  The same workspace is used during the second, actual compile phase for  The same workspace is used during the second, actual compile phase for
104  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
105  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
106  is 4 there is plenty of room. */  is 4 there is plenty of room for most patterns. However, the memory can get
107    filled up by repetitions of forward references, for example patterns like
108    /(?1){0,1999}(b)/, and one user did hit the limit. The code has been changed so
109    that the workspace is expanded using malloc() in this situation. The value
110    below is therefore a minimum, and we put a maximum on it for safety. The
111    minimum is now also defined in terms of LINK_SIZE so that the use of malloc()
112    kicks in at the same number of forward references in all cases. */
113    
114  #define COMPILE_WORK_SIZE (4096)  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)
115    #define COMPILE_WORK_SIZE_MAX (100*COMPILE_WORK_SIZE)
116    
117  /* 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
118  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. */
119    
120  #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100)  #define WORK_SIZE_SAFETY_MARGIN (100)
121    
122    /* Private flags added to firstchar and reqchar. */
123    
124    #define REQ_CASELESS   0x10000000l      /* Indicates caselessness */
125    #define REQ_VARY       0x20000000l      /* Reqchar followed non-literal item */
126    #define REQ_MASK       (REQ_CASELESS | REQ_VARY)
127    
128    /* Repeated character flags. */
129    
130    #define UTF_LENGTH     0x10000000l      /* The char contains its length. */
131    
132  /* Table for handling escaped characters in the range '0'-'z'. Positive returns  /* Table for handling escaped characters in the range '0'-'z'. Positive returns
133  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 262  static const char posix_names[] =
262    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0
263    STRING_word0  STRING_xdigit;    STRING_word0  STRING_xdigit;
264    
265  static const uschar posix_name_lengths[] = {  static const pcre_uint8 posix_name_lengths[] = {
266    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 };
267    
268  /* 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 297  substitutes must be in the order of the
297  both positive and negative cases. NULL means no substitute. */  both positive and negative cases. NULL means no substitute. */
298    
299  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
300  static const uschar *substitutes[] = {  static const pcre_uchar string_PNd[]  = {
301    (uschar *)"\\P{Nd}",    /* \D */    CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
302    (uschar *)"\\p{Nd}",    /* \d */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
303    (uschar *)"\\P{Xsp}",   /* \S */       /* NOTE: Xsp is Perl space */  static const pcre_uchar string_pNd[]  = {
304    (uschar *)"\\p{Xsp}",   /* \s */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
305    (uschar *)"\\P{Xwd}",   /* \W */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
306    (uschar *)"\\p{Xwd}"    /* \w */  static const pcre_uchar string_PXsp[] = {
307      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
308      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
309    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_PXwd[] = {
313      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
314      CHAR_X, CHAR_w, CHAR_d, 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    
319    static const pcre_uchar *substitutes[] = {
320      string_PNd,           /* \D */
321      string_pNd,           /* \d */
322      string_PXsp,          /* \S */       /* NOTE: Xsp is Perl space */
323      string_pXsp,          /* \s */
324      string_PXwd,          /* \W */
325      string_pXwd           /* \w */
326  };  };
327    
328  static const uschar *posix_substitutes[] = {  static const pcre_uchar string_pL[] =   {
329    (uschar *)"\\p{L}",     /* alpha */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
330    (uschar *)"\\p{Ll}",    /* lower */    CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
331    (uschar *)"\\p{Lu}",    /* upper */  static const pcre_uchar string_pLl[] =  {
332    (uschar *)"\\p{Xan}",   /* alnum */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
333    NULL,                   /* ascii */    CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
334    (uschar *)"\\h",        /* blank */  static const pcre_uchar string_pLu[] =  {
335    NULL,                   /* cntrl */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
336    (uschar *)"\\p{Nd}",    /* digit */    CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
337    NULL,                   /* graph */  static const pcre_uchar string_pXan[] = {
338    NULL,                   /* print */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
339    NULL,                   /* punct */    CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
340    (uschar *)"\\p{Xps}",   /* space */    /* NOTE: Xps is POSIX space */  static const pcre_uchar string_h[] =    {
341    (uschar *)"\\p{Xwd}",   /* word */    CHAR_BACKSLASH, CHAR_h, '\0' };
342    NULL,                   /* xdigit */  static const pcre_uchar string_pXps[] = {
343      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
344      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
345    static const pcre_uchar string_PL[] =   {
346      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
347      CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
348    static const pcre_uchar string_PLl[] =  {
349      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
350      CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
351    static const pcre_uchar string_PLu[] =  {
352      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
353      CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
354    static const pcre_uchar string_PXan[] = {
355      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
356      CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
357    static const pcre_uchar string_H[] =    {
358      CHAR_BACKSLASH, CHAR_H, '\0' };
359    static const pcre_uchar string_PXps[] = {
360      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
361      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
362    
363    static const pcre_uchar *posix_substitutes[] = {
364      string_pL,            /* alpha */
365      string_pLl,           /* lower */
366      string_pLu,           /* upper */
367      string_pXan,          /* alnum */
368      NULL,                 /* ascii */
369      string_h,             /* blank */
370      NULL,                 /* cntrl */
371      string_pNd,           /* digit */
372      NULL,                 /* graph */
373      NULL,                 /* print */
374      NULL,                 /* punct */
375      string_pXps,          /* space */    /* NOTE: Xps is POSIX space */
376      string_pXwd,          /* word */
377      NULL,                 /* xdigit */
378    /* Negated cases */    /* Negated cases */
379    (uschar *)"\\P{L}",     /* ^alpha */    string_PL,            /* ^alpha */
380    (uschar *)"\\P{Ll}",    /* ^lower */    string_PLl,           /* ^lower */
381    (uschar *)"\\P{Lu}",    /* ^upper */    string_PLu,           /* ^upper */
382    (uschar *)"\\P{Xan}",   /* ^alnum */    string_PXan,          /* ^alnum */
383    NULL,                   /* ^ascii */    NULL,                 /* ^ascii */
384    (uschar *)"\\H",        /* ^blank */    string_H,             /* ^blank */
385    NULL,                   /* ^cntrl */    NULL,                 /* ^cntrl */
386    (uschar *)"\\P{Nd}",    /* ^digit */    string_PNd,           /* ^digit */
387    NULL,                   /* ^graph */    NULL,                 /* ^graph */
388    NULL,                   /* ^print */    NULL,                 /* ^print */
389    NULL,                   /* ^punct */    NULL,                 /* ^punct */
390    (uschar *)"\\P{Xps}",   /* ^space */   /* NOTE: Xps is POSIX space */    string_PXps,          /* ^space */   /* NOTE: Xps is POSIX space */
391    (uschar *)"\\P{Xwd}",   /* ^word */    string_PXwd,          /* ^word */
392    NULL                    /* ^xdigit */    NULL                  /* ^xdigit */
393  };  };
394  #define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))  #define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))
395  #endif  #endif
396    
397  #define STRING(a)  # a  #define STRING(a)  # a
# Line 365  static const char error_texts[] = Line 450  static const char error_texts[] =
450    /* 30 */    /* 30 */
451    "unknown POSIX class name\0"    "unknown POSIX class name\0"
452    "POSIX collating elements are not supported\0"    "POSIX collating elements are not supported\0"
453    "this version of PCRE is not compiled with PCRE_UTF8 support\0"    "this version of PCRE is compiled without UTF support\0"
454    "spare error\0"  /** DEAD **/    "spare error\0"  /** DEAD **/
455    "character value in \\x{...} sequence is too large\0"    "character value in \\x{...} sequence is too large\0"
456    /* 35 */    /* 35 */
# Line 388  static const char error_texts[] = Line 473  static const char error_texts[] =
473    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
474    /* 50 */    /* 50 */
475    "repeated subpattern is too long\0"    /** DEAD **/    "repeated subpattern is too long\0"    /** DEAD **/
476    "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"
477    "internal error: overran compiling workspace\0"    "internal error: overran compiling workspace\0"
478    "internal error: previously-checked referenced subpattern not found\0"    "internal error: previously-checked referenced subpattern not found\0"
479    "DEFINE group contains more than one branch\0"    "DEFINE group contains more than one branch\0"
480    /* 55 */    /* 55 */
481    "repeating a DEFINE group is not allowed\0"    "repeating a DEFINE group is not allowed\0"  /** DEAD **/
482    "inconsistent NEWLINE options\0"    "inconsistent NEWLINE options\0"
483    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"    "\\g is not followed by a braced, angle-bracketed, or quoted name/number or by a plain number\0"
484    "a numbered reference must not be zero\0"    "a numbered reference must not be zero\0"
# Line 407  static const char error_texts[] = Line 492  static const char error_texts[] =
492    /* 65 */    /* 65 */
493    "different names for subpatterns of the same number are not allowed\0"    "different names for subpatterns of the same number are not allowed\0"
494    "(*MARK) must have an argument\0"    "(*MARK) must have an argument\0"
495    "this version of PCRE is not compiled with PCRE_UCP support\0"    "this version of PCRE is not compiled with Unicode property support\0"
496    "\\c must be followed by an ASCII character\0"    "\\c must be followed by an ASCII character\0"
497      "\\k is not followed by a braced, angle-bracketed, or quoted name\0"
498      /* 70 */
499      "internal error: unknown opcode in find_fixedlength()\0"
500      "\\N is not supported in a class\0"
501      "too many forward references\0"
502      "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
503      "invalid UTF-16 string\0"
504      /* 75 */
505      "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
506      "character value in \\u.... sequence is too large\0"
507      "invalid UTF-32 string\0"
508    ;    ;
509    
510  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 427  For convenience, we use the same bit def Line 523  For convenience, we use the same bit def
523    
524  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
525    
526    /* Using a simple comparison for decimal numbers rather than a memory read
527    is much faster, and the resulting code is simpler (the compiler turns it
528    into a subtraction and unsigned comparison). */
529    
530    #define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9)
531    
532  #ifndef EBCDIC  #ifndef EBCDIC
533    
534  /* 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
535  UTF-8 mode. */  UTF-8 mode. */
536    
537  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
538    {    {
539    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
540    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */
# Line 471  static const unsigned char digitab[] = Line 573  static const unsigned char digitab[] =
573    
574  /* 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. */
575    
576  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
577    {    {
578    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
579    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */
# Line 506  static const unsigned char digitab[] = Line 608  static const unsigned char digitab[] =
608    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */
609    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */
610    
611  static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */  static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */
612    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */
613    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */
614    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */
# Line 542  static const unsigned char ebcdic_charta Line 644  static const unsigned char ebcdic_charta
644  #endif  #endif
645    
646    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, int, uschar **, const uschar **, int *, BOOL, BOOL, int,  
     int *, int *, branch_chain *, compile_data *, int *);  
   
647    
648    
649  /*************************************************  /*************************************************
# Line 577  return s; Line 673  return s;
673    
674    
675  /*************************************************  /*************************************************
676    *           Expand the workspace                 *
677    *************************************************/
678    
679    /* This function is called during the second compiling phase, if the number of
680    forward references fills the existing workspace, which is originally a block on
681    the stack. A larger block is obtained from malloc() unless the ultimate limit
682    has been reached or the increase will be rather small.
683    
684    Argument: pointer to the compile data block
685    Returns:  0 if all went well, else an error number
686    */
687    
688    static int
689    expand_workspace(compile_data *cd)
690    {
691    pcre_uchar *newspace;
692    int newsize = cd->workspace_size * 2;
693    
694    if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;
695    if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX ||
696        newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)
697     return ERR72;
698    
699    newspace = (PUBL(malloc))(IN_UCHARS(newsize));
700    if (newspace == NULL) return ERR21;
701    memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar));
702    cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace);
703    if (cd->workspace_size > COMPILE_WORK_SIZE)
704      (PUBL(free))((void *)cd->start_workspace);
705    cd->start_workspace = newspace;
706    cd->workspace_size = newsize;
707    return 0;
708    }
709    
710    
711    
712    /*************************************************
713    *            Check for counted repeat            *
714    *************************************************/
715    
716    /* This function is called when a '{' is encountered in a place where it might
717    start a quantifier. It looks ahead to see if it really is a quantifier or not.
718    It is only a quantifier if it is one of the forms {ddd} {ddd,} or {ddd,ddd}
719    where the ddds are digits.
720    
721    Arguments:
722      p         pointer to the first char after '{'
723    
724    Returns:    TRUE or FALSE
725    */
726    
727    static BOOL
728    is_counted_repeat(const pcre_uchar *p)
729    {
730    if (!IS_DIGIT(*p)) return FALSE;
731    p++;
732    while (IS_DIGIT(*p)) p++;
733    if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
734    
735    if (*p++ != CHAR_COMMA) return FALSE;
736    if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
737    
738    if (!IS_DIGIT(*p)) return FALSE;
739    p++;
740    while (IS_DIGIT(*p)) p++;
741    
742    return (*p == CHAR_RIGHT_CURLY_BRACKET);
743    }
744    
745    
746    
747    /*************************************************
748  *            Handle escapes                      *  *            Handle escapes                      *
749  *************************************************/  *************************************************/
750    
751  /* 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
752  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
753  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group
754  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  n is returned as ESC_REF + n; ESC_REF is the highest ESC_xxx macro. When
755  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  UTF-8 is enabled, a positive value greater than 255 may be returned in chptr.
756  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
757  sequence.  escape sequence.
758    
759  Arguments:  Arguments:
760    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
761      chptr          points to the data character
762    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
763    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
764    options        the options bits    options        the options bits
765    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
766    
767  Returns:         zero or positive => a data character  Returns:         zero => a data character
768                   negative => a special escape sequence                   positive => a special escape sequence
769                   on error, errorcodeptr is set                   on error, errorcodeptr is set
770  */  */
771    
772  static int  static int
773  check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, int *chptr, int *errorcodeptr,
774    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
775  {  {
776  BOOL utf8 = (options & PCRE_UTF8) != 0;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
777  const uschar *ptr = *ptrptr + 1;  BOOL utf = (options & PCRE_UTF8) != 0;
778  int c, i;  const pcre_uchar *ptr = *ptrptr + 1;
779    pcre_uint32 c;
780    int escape = 0;
781    int i;
782    
783  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
784  ptr--;                            /* Set pointer back to the last byte */  ptr--;                            /* Set pointer back to the last byte */
# Line 620  in a table. A non-zero result is somethi Line 792  in a table. A non-zero result is somethi
792  Otherwise further processing may be required. */  Otherwise further processing may be required. */
793    
794  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
795  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */  /* Not alphanumeric */
796  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if (c < CHAR_0 || c > CHAR_z) {}
797    else if ((i = escapes[c - CHAR_0]) != 0) { if (i > 0) c = i; else escape = -i; }
798    
799  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
800  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */  /* Not alphanumeric */
801  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
802    else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = i; else escape = -i; }
803  #endif  #endif
804    
805  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
806    
807  else  else
808    {    {
809    const uschar *oldptr;    const pcre_uchar *oldptr;
810    BOOL braced, negated;    BOOL braced, negated;
811    
812    switch (c)    switch (c)
# Line 642  else Line 816  else
816    
817      case CHAR_l:      case CHAR_l:
818      case CHAR_L:      case CHAR_L:
819        *errorcodeptr = ERR37;
820        break;
821    
822      case CHAR_u:      case CHAR_u:
823        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
824          {
825          /* In JavaScript, \u must be followed by four hexadecimal numbers.
826          Otherwise it is a lowercase u letter. */
827          if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
828            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
829            && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
830            && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
831            {
832            c = 0;
833            for (i = 0; i < 4; ++i)
834              {
835              register pcre_uint32 cc = *(++ptr);
836    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
837              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
838              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
839    #else           /* EBCDIC coding */
840              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
841              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
842    #endif
843              }
844    
845    #if defined COMPILE_PCRE8
846            if (c > (utf ? 0x10ffff : 0xff))
847    #elif defined COMPILE_PCRE16
848            if (c > (utf ? 0x10ffff : 0xffff))
849    #elif defined COMPILE_PCRE32
850            if (utf && c > 0x10ffff)
851    #endif
852              {
853              *errorcodeptr = ERR76;
854              }
855            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
856            }
857          }
858        else
859          *errorcodeptr = ERR37;
860        break;
861    
862      case CHAR_U:      case CHAR_U:
863      *errorcodeptr = ERR37;      /* In JavaScript, \U is an uppercase U letter. */
864        if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37;
865      break;      break;
866    
867      /* \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
868        class, \g must be followed by one of a number of specific things:
869    
870      (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
871      backreference. If negative, it is a relative backreference. This is a Perl      backreference. If negative, it is a relative backreference. This is a Perl
# Line 661  else Line 879  else
879      (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
880      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
881      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
882      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
883    
884      case CHAR_g:      case CHAR_g:
885        if (isclass) break;
886      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
887        {        {
888        c = -ESC_g;        escape = ESC_g;
889        break;        break;
890        }        }
891    
# Line 674  else Line 893  else
893    
894      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
895        {        {
896        const uschar *p;        const pcre_uchar *p;
897        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
898          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
899        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
900          {          {
901          c = -ESC_k;          escape = ESC_k;
902          break;          break;
903          }          }
904        braced = TRUE;        braced = TRUE;
# Line 694  else Line 913  else
913        }        }
914      else negated = FALSE;      else negated = FALSE;
915    
916        /* The integer range is limited by the machine's int representation. */
917      c = 0;      c = 0;
918      while ((digitab[ptr[1]] & ctype_digit) != 0)      while (IS_DIGIT(ptr[1]))
919          {
920          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */
921            {
922            c = -1;
923            break;
924            }
925        c = c * 10 + *(++ptr) - CHAR_0;        c = c * 10 + *(++ptr) - CHAR_0;
926          }
927      if (c < 0)   /* Integer overflow */      if (((unsigned int)c) > INT_MAX) /* Integer overflow */
928        {        {
929          while (IS_DIGIT(ptr[1]))
930            ptr++;
931        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
932        break;        break;
933        }        }
# Line 726  else Line 954  else
954        c = bracount - (c - 1);        c = bracount - (c - 1);
955        }        }
956    
957      c = -(ESC_REF + c);      escape = ESC_REF + c;
958      break;      break;
959    
960      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 747  else Line 975  else
975      if (!isclass)      if (!isclass)
976        {        {
977        oldptr = ptr;        oldptr = ptr;
978          /* The integer range is limited by the machine's int representation. */
979        c -= CHAR_0;        c -= CHAR_0;
980        while ((digitab[ptr[1]] & ctype_digit) != 0)        while (IS_DIGIT(ptr[1]))
981            {
982            if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */
983              {
984              c = -1;
985              break;
986              }
987          c = c * 10 + *(++ptr) - CHAR_0;          c = c * 10 + *(++ptr) - CHAR_0;
988        if (c < 0)    /* Integer overflow */          }
989          if (((unsigned int)c) > INT_MAX) /* Integer overflow */
990          {          {
991            while (IS_DIGIT(ptr[1]))
992              ptr++;
993          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
994          break;          break;
995          }          }
996        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
997          {          {
998          c = -(ESC_REF + c);          escape = ESC_REF + c;
999          break;          break;
1000          }          }
1001        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 777  else Line 1015  else
1015      /* \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
1016      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
1017      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
1018      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,
1019      than 3 octal digits. */      but no more than 3 octal digits. */
1020    
1021      case CHAR_0:      case CHAR_0:
1022      c -= CHAR_0;      c -= CHAR_0;
1023      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1024          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1025      if (!utf8 && c > 255) *errorcodeptr = ERR51;  #ifdef COMPILE_PCRE8
1026        if (!utf && c > 0xff) *errorcodeptr = ERR51;
1027    #endif
1028      break;      break;
1029    
1030      /* \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
1031      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.
1032      treated as a data character. */      If not, { is treated as a data character. */
1033    
1034      case CHAR_x:      case CHAR_x:
1035        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1036          {
1037          /* In JavaScript, \x must be followed by two hexadecimal numbers.
1038          Otherwise it is a lowercase x letter. */
1039          if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1040            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1041            {
1042            c = 0;
1043            for (i = 0; i < 2; ++i)
1044              {
1045              register pcre_uint32 cc = *(++ptr);
1046    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1047              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1048              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
1049    #else           /* EBCDIC coding */
1050              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
1051              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1052    #endif
1053              }
1054            }
1055          break;
1056          }
1057    
1058      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1059        {        {
1060        const uschar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
1061        int count = 0;        BOOL overflow;
1062    
1063        c = 0;        c = 0;
1064        while ((digitab[*pt] & ctype_xdigit) != 0)        overflow = FALSE;
1065          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1066          {          {
1067          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1068          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1069          count++;  
1070    #ifdef COMPILE_PCRE32
1071            if (c >= 0x10000000l) { overflow = TRUE; break; }
1072    #endif
1073    
1074  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1075          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
# Line 811  else Line 1078  else
1078          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 */
1079          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1080  #endif  #endif
1081    
1082    #if defined COMPILE_PCRE8
1083            if (c > (utf ? 0x10ffff : 0xff)) { overflow = TRUE; break; }
1084    #elif defined COMPILE_PCRE16
1085            if (c > (utf ? 0x10ffff : 0xffff)) { overflow = TRUE; break; }
1086    #elif defined COMPILE_PCRE32
1087            if (utf && c > 0x10ffff) { overflow = TRUE; break; }
1088    #endif
1089            }
1090    
1091          if (overflow)
1092            {
1093            while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1094            *errorcodeptr = ERR34;
1095          }          }
1096    
1097        if (*pt == CHAR_RIGHT_CURLY_BRACKET)        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
1098          {          {
1099          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1100          ptr = pt;          ptr = pt;
1101          break;          break;
1102          }          }
# Line 827  else Line 1108  else
1108      /* Read just a single-byte hex-defined char */      /* Read just a single-byte hex-defined char */
1109    
1110      c = 0;      c = 0;
1111      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1112        {        {
1113        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1114        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1115  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1116        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 857  else Line 1138  else
1138      if (c > 127)  /* Excludes all non-ASCII in either mode */      if (c > 127)  /* Excludes all non-ASCII in either mode */
1139        {        {
1140        *errorcodeptr = ERR68;        *errorcodeptr = ERR68;
1141        break;        break;
1142        }        }
1143      if (c >= CHAR_a && c <= CHAR_z) c -= 32;      if (c >= CHAR_a && c <= CHAR_z) c -= 32;
1144      c ^= 0x40;      c ^= 0x40;
1145  #else             /* EBCDIC coding */  #else             /* EBCDIC coding */
# Line 885  else Line 1166  else
1166    }    }
1167    
1168  /* 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
1169  newline". PCRE does not support \N{name}. */  newline". PCRE does not support \N{name}. However, it does support
1170    quantification such as \N{2,3}. */
1171    
1172  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET)  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1173         !is_counted_repeat(ptr+2))
1174    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1175    
1176  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1177    
1178  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1179    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1180    
1181  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1182    
1183  *ptrptr = ptr;  *ptrptr = ptr;
1184  return c;  *chptr = c;
1185    return escape;
1186  }  }
1187    
   
   
1188  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1189  /*************************************************  /*************************************************
1190  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 923  Returns:         type value from ucp_typ Line 1205  Returns:         type value from ucp_typ
1205  */  */
1206    
1207  static int  static int
1208  get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)
1209  {  {
1210  int c, i, bot, top;  int c, i, bot, top;
1211  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1212  char name[32];  pcre_uchar name[32];
1213    
1214  c = *(++ptr);  c = *(++ptr);
1215  if (c == 0) goto ERROR_RETURN;  if (c == 0) goto ERROR_RETURN;
# Line 944  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1226  if (c == CHAR_LEFT_CURLY_BRACKET)
1226      *negptr = TRUE;      *negptr = TRUE;
1227      ptr++;      ptr++;
1228      }      }
1229    for (i = 0; i < (int)sizeof(name) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1230      {      {
1231      c = *(++ptr);      c = *(++ptr);
1232      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 968  else Line 1250  else
1250  /* Search for a recognized property name using binary chop */  /* Search for a recognized property name using binary chop */
1251    
1252  bot = 0;  bot = 0;
1253  top = _pcre_utt_size;  top = PRIV(utt_size);
1254    
1255  while (bot < top)  while (bot < top)
1256    {    {
1257    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1258    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);    c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1259    if (c == 0)    if (c == 0)
1260      {      {
1261      *dptr = _pcre_utt[i].value;      *dptr = PRIV(utt)[i].value;
1262      return _pcre_utt[i].type;      return PRIV(utt)[i].type;
1263      }      }
1264    if (c > 0) bot = i + 1; else top = i;    if (c > 0) bot = i + 1; else top = i;
1265    }    }
# Line 997  return -1; Line 1279  return -1;
1279    
1280    
1281  /*************************************************  /*************************************************
 *            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);  
 }  
   
   
   
 /*************************************************  
1282  *         Read repeat counts                     *  *         Read repeat counts                     *
1283  *************************************************/  *************************************************/
1284    
# Line 1048  Returns:         pointer to '}' on succe Line 1297  Returns:         pointer to '}' on succe
1297                   current ptr on error, with errorcodeptr set non-zero                   current ptr on error, with errorcodeptr set non-zero
1298  */  */
1299    
1300  static const uschar *  static const pcre_uchar *
1301  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)
1302  {  {
1303  int min = 0;  int min = 0;
1304  int max = -1;  int max = -1;
# Line 1057  int max = -1; Line 1306  int max = -1;
1306  /* 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
1307  an integer overflow. */  an integer overflow. */
1308    
1309  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;
1310  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1311    {    {
1312    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1072  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1321  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1321    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1322      {      {
1323      max = 0;      max = 0;
1324      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;
1325      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1326        {        {
1327        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1105  top-level call starts at the beginning o Line 1354  top-level call starts at the beginning o
1354  start at a parenthesis. It scans along a pattern's text looking for capturing  start at a parenthesis. It scans along a pattern's text looking for capturing
1355  subpatterns, and counting them. If it finds a named pattern that matches the  subpatterns, and counting them. If it finds a named pattern that matches the
1356  name it is given, it returns its number. Alternatively, if the name is NULL, it  name it is given, it returns its number. Alternatively, if the name is NULL, it
1357  returns when it reaches a given numbered subpattern. We know that if (?P< is  returns when it reaches a given numbered subpattern. Recursion is used to keep
1358  encountered, the name will be terminated by '>' because that is checked in the  track of subpatterns that reset the capturing group numbers - the (?| feature.
1359  first pass. Recursion is used to keep track of subpatterns that reset the  
1360  capturing group numbers - the (?| feature.  This function was originally called only from the second pass, in which we know
1361    that if (?< or (?' or (?P< is encountered, the name will be correctly
1362    terminated because that is checked in the first pass. There is now one call to
1363    this function in the first pass, to check for a recursive back reference by
1364    name (so that we can make the whole group atomic). In this case, we need check
1365    only up to the current position in the pattern, and that is still OK because
1366    and previous occurrences will have been checked. To make this work, the test
1367    for "end of pattern" is a check against cd->end_pattern in the main loop,
1368    instead of looking for a binary zero. This means that the special first-pass
1369    call can adjust cd->end_pattern temporarily. (Checks for binary zero while
1370    processing items within the loop are OK, because afterwards the main loop will
1371    terminate.)
1372    
1373  Arguments:  Arguments:
1374    ptrptr       address of the current character pointer (updated)    ptrptr       address of the current character pointer (updated)
# Line 1116  Arguments: Line 1376  Arguments:
1376    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1377    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1378    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1379    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1380    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1381    
1382  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1383  */  */
1384    
1385  static int  static int
1386  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,
1387    BOOL xmode, BOOL utf8, int *count)    BOOL xmode, BOOL utf, int *count)
1388  {  {
1389  uschar *ptr = *ptrptr;  pcre_uchar *ptr = *ptrptr;
1390  int start_count = *count;  int start_count = *count;
1391  int hwm_count = start_count;  int hwm_count = start_count;
1392  BOOL dup_parens = FALSE;  BOOL dup_parens = FALSE;
# Line 1193  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1453  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1453          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1454        {        {
1455        int term;        int term;
1456        const uschar *thisname;        const pcre_uchar *thisname;
1457        *count += 1;        *count += 1;
1458        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1459        term = *ptr++;        term = *ptr++;
# Line 1201  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1461  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1461        thisname = ptr;        thisname = ptr;
1462        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1463        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == ptr - thisname &&
1464            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, lorn) == 0)
1465          return *count;          return *count;
1466        term++;        term++;
1467        }        }
# Line 1209  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1469  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1469    }    }
1470    
1471  /* Past any initial parenthesis handling, scan for parentheses or vertical  /* Past any initial parenthesis handling, scan for parentheses or vertical
1472  bars. */  bars. Stop if we get to cd->end_pattern. Note that this is important for the
1473    first-pass call when this value is temporarily adjusted to stop at the current
1474    position. So DO NOT change this to a test for binary zero. */
1475    
1476  for (; *ptr != 0; ptr++)  for (; ptr < cd->end_pattern; ptr++)
1477    {    {
1478    /* Skip over backslashed characters and also entire \Q...\E */    /* Skip over backslashed characters and also entire \Q...\E */
1479    
# Line 1242  for (; *ptr != 0; ptr++) Line 1504  for (; *ptr != 0; ptr++)
1504          {          {
1505          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1506            ptr+= 2;            ptr+= 2;
1507          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1508                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1509            ptr += 4;            ptr += 4;
1510          else          else
# Line 1285  for (; *ptr != 0; ptr++) Line 1547  for (; *ptr != 0; ptr++)
1547    
1548    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1549      {      {
1550      ptr++;      ptr++;
1551      while (*ptr != 0)      while (*ptr != 0)
1552        {        {
1553        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1554        ptr++;        ptr++;
1555  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1556        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1557  #endif  #endif
1558        }        }
1559      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
# Line 1302  for (; *ptr != 0; ptr++) Line 1564  for (; *ptr != 0; ptr++)
1564    
1565    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1566      {      {
1567      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1568      if (rc > 0) return rc;      if (rc > 0) return rc;
1569      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1570      }      }
# Line 1348  Arguments: Line 1610  Arguments:
1610    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1611    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1612    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1613    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1614    
1615  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1616  */  */
1617    
1618  static int  static int
1619  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,
1620    BOOL utf8)    BOOL utf)
1621  {  {
1622  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1623  int count = 0;  int count = 0;
1624  int rc;  int rc;
1625    
# Line 1368  matching closing parens. That is why we Line 1630  matching closing parens. That is why we
1630    
1631  for (;;)  for (;;)
1632    {    {
1633    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1634    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == 0) break;
1635    }    }
1636    
# Line 1384  return rc; Line 1646  return rc;
1646    
1647  /* This is called by several functions that scan a compiled expression looking  /* This is called by several functions that scan a compiled expression looking
1648  for a fixed first character, or an anchoring op code etc. It skips over things  for a fixed first character, or an anchoring op code etc. It skips over things
1649  that do not influence this. For some calls, a change of option is important.  that do not influence this. For some calls, it makes sense to skip negative
1650  For some calls, it makes sense to skip negative forward and all backward  forward and all backward assertions, and also the \b assertion; for others it
1651  assertions, and also the \b assertion; for others it does not.  does not.
1652    
1653  Arguments:  Arguments:
1654    code         pointer to the start of the group    code         pointer to the start of the group
   options      pointer to external options  
   optbit       the option bit whose changing is significant, or  
                  zero if none are  
1655    skipassert   TRUE if certain assertions are to be skipped    skipassert   TRUE if certain assertions are to be skipped
1656    
1657  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1658  */  */
1659    
1660  static const uschar*  static const pcre_uchar*
1661  first_significant_code(const uschar *code, int *options, int optbit,  first_significant_code(const pcre_uchar *code, BOOL skipassert)
   BOOL skipassert)  
1662  {  {
1663  for (;;)  for (;;)
1664    {    {
1665    switch ((int)*code)    switch ((int)*code)
1666      {      {
     case OP_OPT:  
     if (optbit > 0 && ((int)code[1] & optbit) != (*options & optbit))  
       *options = (int)code[1];  
     code += 2;  
     break;  
   
1667      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1668      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1669      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1670      if (!skipassert) return code;      if (!skipassert) return code;
1671      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1672      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1673      break;      break;
1674    
1675      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1431  for (;;) Line 1683  for (;;)
1683      case OP_RREF:      case OP_RREF:
1684      case OP_NRREF:      case OP_NRREF:
1685      case OP_DEF:      case OP_DEF:
1686      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1687      break;      break;
1688    
1689      default:      default:
# Line 1461  and doing the check at the end; a flag s Line 1713  and doing the check at the end; a flag s
1713    
1714  Arguments:  Arguments:
1715    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1716    options  the compiling options    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1717    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1718    cd       the "compile data" structure    cd       the "compile data" structure
1719    
1720  Returns:   the fixed length,  Returns:   the fixed length,
1721               or -1 if there is no fixed length,               or -1 if there is no fixed length,
1722               or -2 if \C was encountered               or -2 if \C was encountered (in UTF-8 mode only)
1723               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
1724                 or -4 if an unknown opcode was encountered (internal error)
1725  */  */
1726    
1727  static int  static int
1728  find_fixedlength(uschar *code, int options, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1729  {  {
1730  int length = -1;  int length = -1;
1731    
1732  register int branchlength = 0;  register int branchlength = 0;
1733  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1734    
1735  /* 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
1736  branch, check the length against that of the other branches. */  branch, check the length against that of the other branches. */
# Line 1485  branch, check the length against that of Line 1738  branch, check the length against that of
1738  for (;;)  for (;;)
1739    {    {
1740    int d;    int d;
1741    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1742    register int op = *cc;    register int op = *cc;
1743    
1744    switch (op)    switch (op)
1745      {      {
1746        /* We only need to continue for OP_CBRA (normal capturing bracket) and
1747        OP_BRA (normal non-capturing bracket) because the other variants of these
1748        opcodes are all concerned with unlimited repeated groups, which of course
1749        are not of fixed length. */
1750    
1751      case OP_CBRA:      case OP_CBRA:
1752      case OP_BRA:      case OP_BRA:
1753      case OP_ONCE:      case OP_ONCE:
1754        case OP_ONCE_NC:
1755      case OP_COND:      case OP_COND:
1756      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), options, atend, cd);      d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd);
1757      if (d < 0) return d;      if (d < 0) return d;
1758      branchlength += d;      branchlength += d;
1759      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1760      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
1761      break;      break;
1762    
1763      /* 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.
1764      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
1765      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
1766        the same code. Note that we must not include the OP_KETRxxx opcodes here,
1767        because they all imply an unlimited repeat. */
1768    
1769      case OP_ALT:      case OP_ALT:
1770      case OP_KET:      case OP_KET:
     case OP_KETRMAX:  
     case OP_KETRMIN:  
1771      case OP_END:      case OP_END:
1772        case OP_ACCEPT:
1773        case OP_ASSERT_ACCEPT:
1774      if (length < 0) length = branchlength;      if (length < 0) length = branchlength;
1775        else if (length != branchlength) return -1;        else if (length != branchlength) return -1;
1776      if (*cc != OP_ALT) return length;      if (*cc != OP_ALT) return length;
# Line 1522  for (;;) Line 1784  for (;;)
1784    
1785      case OP_RECURSE:      case OP_RECURSE:
1786      if (!atend) return -3;      if (!atend) return -3;
1787      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1788      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1789      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1790      d = find_fixedlength(cs + 2, options, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1791      if (d < 0) return d;      if (d < 0) return d;
1792      branchlength += d;      branchlength += d;
1793      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1538  for (;;) Line 1800  for (;;)
1800      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1801      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1802      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1803      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1804        break;
1805    
1806      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1807    
1808      case OP_REVERSE:      case OP_MARK:
1809        case OP_PRUNE_ARG:
1810        case OP_SKIP_ARG:
1811        case OP_THEN_ARG:
1812        cc += cc[1] + PRIV(OP_lengths)[*cc];
1813        break;
1814    
1815        case OP_CALLOUT:
1816        case OP_CIRC:
1817        case OP_CIRCM:
1818        case OP_CLOSE:
1819        case OP_COMMIT:
1820      case OP_CREF:      case OP_CREF:
     case OP_NCREF:  
     case OP_RREF:  
     case OP_NRREF:  
1821      case OP_DEF:      case OP_DEF:
1822      case OP_OPT:      case OP_DOLL:
1823      case OP_CALLOUT:      case OP_DOLLM:
     case OP_SOD:  
     case OP_SOM:  
     case OP_SET_SOM:  
1824      case OP_EOD:      case OP_EOD:
1825      case OP_EODN:      case OP_EODN:
1826      case OP_CIRC:      case OP_FAIL:
1827      case OP_DOLL:      case OP_NCREF:
1828        case OP_NRREF:
1829      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1830        case OP_PRUNE:
1831        case OP_REVERSE:
1832        case OP_RREF:
1833        case OP_SET_SOM:
1834        case OP_SKIP:
1835        case OP_SOD:
1836        case OP_SOM:
1837        case OP_THEN:
1838      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1839      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1840      break;      break;
1841    
1842      /* Handle literal characters */      /* Handle literal characters */
1843    
1844      case OP_CHAR:      case OP_CHAR:
1845      case OP_CHARNC:      case OP_CHARI:
1846      case OP_NOT:      case OP_NOT:
1847        case OP_NOTI:
1848      branchlength++;      branchlength++;
1849      cc += 2;      cc += 2;
1850  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1851      if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0)      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
       cc += _pcre_utf8_table4[cc[-1] & 0x3f];  
1852  #endif  #endif
1853      break;      break;
1854    
# Line 1579  for (;;) Line 1856  for (;;)
1856      need to skip over a multibyte character in UTF8 mode.  */      need to skip over a multibyte character in UTF8 mode.  */
1857    
1858      case OP_EXACT:      case OP_EXACT:
1859        case OP_EXACTI:
1860        case OP_NOTEXACT:
1861        case OP_NOTEXACTI:
1862      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1863      cc += 4;      cc += 2 + IMM2_SIZE;
1864  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1865      if ((options & PCRE_UTF8) != 0 && cc[-1] >= 0xc0)      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
       cc += _pcre_utf8_table4[cc[-1] & 0x3f];  
1866  #endif  #endif
1867      break;      break;
1868    
1869      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1870      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1871      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1872      cc += 4;        cc += 2;
1873        cc += 1 + IMM2_SIZE + 1;
1874      break;      break;
1875    
1876      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1600  for (;;) Line 1880  for (;;)
1880      cc += 2;      cc += 2;
1881      /* Fall through */      /* Fall through */
1882    
1883        case OP_HSPACE:
1884        case OP_VSPACE:
1885        case OP_NOT_HSPACE:
1886        case OP_NOT_VSPACE:
1887      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
1888      case OP_DIGIT:      case OP_DIGIT:
1889      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
# Line 1612  for (;;) Line 1896  for (;;)
1896      cc++;      cc++;
1897      break;      break;
1898    
1899      /* The single-byte matcher isn't allowed */      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;
1900        otherwise \C is coded as OP_ALLANY. */
1901    
1902      case OP_ANYBYTE:      case OP_ANYBYTE:
1903      return -2;      return -2;
1904    
1905      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1906    
1907  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1908      case OP_XCLASS:      case OP_XCLASS:
1909      cc += GET(cc, 1) - 33;      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
1910      /* Fall through */      /* Fall through */
1911  #endif  #endif
1912    
1913      case OP_CLASS:      case OP_CLASS:
1914      case OP_NCLASS:      case OP_NCLASS:
1915      cc += 33;      cc += PRIV(OP_lengths)[OP_CLASS];
1916    
1917      switch (*cc)      switch (*cc)
1918        {        {
1919          case OP_CRPLUS:
1920          case OP_CRMINPLUS:
1921        case OP_CRSTAR:        case OP_CRSTAR:
1922        case OP_CRMINSTAR:        case OP_CRMINSTAR:
1923        case OP_CRQUERY:        case OP_CRQUERY:
# Line 1639  for (;;) Line 1926  for (;;)
1926    
1927        case OP_CRRANGE:        case OP_CRRANGE:
1928        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1929        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1930        branchlength += GET2(cc,1);        branchlength += GET2(cc,1);
1931        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1932        break;        break;
1933    
1934        default:        default:
# Line 1651  for (;;) Line 1938  for (;;)
1938    
1939      /* Anything else is variable length */      /* Anything else is variable length */
1940    
1941      default:      case OP_ANYNL:
1942        case OP_BRAMINZERO:
1943        case OP_BRAPOS:
1944        case OP_BRAPOSZERO:
1945        case OP_BRAZERO:
1946        case OP_CBRAPOS:
1947        case OP_EXTUNI:
1948        case OP_KETRMAX:
1949        case OP_KETRMIN:
1950        case OP_KETRPOS:
1951        case OP_MINPLUS:
1952        case OP_MINPLUSI:
1953        case OP_MINQUERY:
1954        case OP_MINQUERYI:
1955        case OP_MINSTAR:
1956        case OP_MINSTARI:
1957        case OP_MINUPTO:
1958        case OP_MINUPTOI:
1959        case OP_NOTMINPLUS:
1960        case OP_NOTMINPLUSI:
1961        case OP_NOTMINQUERY:
1962        case OP_NOTMINQUERYI:
1963        case OP_NOTMINSTAR:
1964        case OP_NOTMINSTARI:
1965        case OP_NOTMINUPTO:
1966        case OP_NOTMINUPTOI:
1967        case OP_NOTPLUS:
1968        case OP_NOTPLUSI:
1969        case OP_NOTPOSPLUS:
1970        case OP_NOTPOSPLUSI:
1971        case OP_NOTPOSQUERY:
1972        case OP_NOTPOSQUERYI:
1973        case OP_NOTPOSSTAR:
1974        case OP_NOTPOSSTARI:
1975        case OP_NOTPOSUPTO:
1976        case OP_NOTPOSUPTOI:
1977        case OP_NOTQUERY:
1978        case OP_NOTQUERYI:
1979        case OP_NOTSTAR:
1980        case OP_NOTSTARI:
1981        case OP_NOTUPTO:
1982        case OP_NOTUPTOI:
1983        case OP_PLUS:
1984        case OP_PLUSI:
1985        case OP_POSPLUS:
1986        case OP_POSPLUSI:
1987        case OP_POSQUERY:
1988        case OP_POSQUERYI:
1989        case OP_POSSTAR:
1990        case OP_POSSTARI:
1991        case OP_POSUPTO:
1992        case OP_POSUPTOI:
1993        case OP_QUERY:
1994        case OP_QUERYI:
1995        case OP_REF:
1996        case OP_REFI:
1997        case OP_SBRA:
1998        case OP_SBRAPOS:
1999        case OP_SCBRA:
2000        case OP_SCBRAPOS:
2001        case OP_SCOND:
2002        case OP_SKIPZERO:
2003        case OP_STAR:
2004        case OP_STARI:
2005        case OP_TYPEMINPLUS:
2006        case OP_TYPEMINQUERY:
2007        case OP_TYPEMINSTAR:
2008        case OP_TYPEMINUPTO:
2009        case OP_TYPEPLUS:
2010        case OP_TYPEPOSPLUS:
2011        case OP_TYPEPOSQUERY:
2012        case OP_TYPEPOSSTAR:
2013        case OP_TYPEPOSUPTO:
2014        case OP_TYPEQUERY:
2015        case OP_TYPESTAR:
2016        case OP_TYPEUPTO:
2017        case OP_UPTO:
2018        case OP_UPTOI:
2019      return -1;      return -1;
2020    
2021        /* Catch unrecognized opcodes so that when new ones are added they
2022        are not forgotten, as has happened in the past. */
2023    
2024        default:
2025        return -4;
2026      }      }
2027    }    }
2028  /* Control never gets here */  /* Control never gets here */
# Line 1673  length. Line 2043  length.
2043    
2044  Arguments:  Arguments:
2045    code        points to start of expression    code        points to start of expression
2046    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2047    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2048    
2049  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
2050  */  */
2051    
2052  const uschar *  const pcre_uchar *
2053  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2054  {  {
2055  for (;;)  for (;;)
2056    {    {
2057    register int c = *code;    register int c = *code;
2058    
2059    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2060    
2061    /* XCLASS is used for classes that cannot be represented just by a bit    /* XCLASS is used for classes that cannot be represented just by a bit
# Line 1697  for (;;) Line 2068  for (;;)
2068    
2069    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2070      {      {
2071      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2072      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2073      }      }
2074    
2075    /* Handle capturing bracket */    /* Handle capturing bracket */
2076    
2077    else if (c == OP_CBRA)    else if (c == OP_CBRA || c == OP_SCBRA ||
2078               c == OP_CBRAPOS || c == OP_SCBRAPOS)
2079      {      {
2080      int n = GET2(code, 1+LINK_SIZE);      int n = GET2(code, 1+LINK_SIZE);
2081      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2082      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2083      }      }
2084    
2085    /* 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 1735  for (;;) Line 2107  for (;;)
2107        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2108        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2109        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2110        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2111            code += 2;
2112        break;        break;
2113    
2114        case OP_MARK:        case OP_MARK:
# Line 1745  for (;;) Line 2118  for (;;)
2118        break;        break;
2119    
2120        case OP_THEN_ARG:        case OP_THEN_ARG:
2121        code += code[1+LINK_SIZE];        code += code[1];
2122        break;        break;
2123        }        }
2124    
2125      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2126    
2127      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2128    
2129    /* 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
2130    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
2131    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2132    
2133  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2134      if (utf8) switch(c)      if (utf) switch(c)
2135        {        {
2136        case OP_CHAR:        case OP_CHAR:
2137        case OP_CHARNC:        case OP_CHARI:
2138        case OP_EXACT:        case OP_EXACT:
2139          case OP_EXACTI:
2140        case OP_UPTO:        case OP_UPTO:
2141          case OP_UPTOI:
2142        case OP_MINUPTO:        case OP_MINUPTO:
2143          case OP_MINUPTOI:
2144        case OP_POSUPTO:        case OP_POSUPTO:
2145          case OP_POSUPTOI:
2146        case OP_STAR:        case OP_STAR:
2147          case OP_STARI:
2148        case OP_MINSTAR:        case OP_MINSTAR:
2149          case OP_MINSTARI:
2150        case OP_POSSTAR:        case OP_POSSTAR:
2151          case OP_POSSTARI:
2152        case OP_PLUS:        case OP_PLUS:
2153          case OP_PLUSI:
2154        case OP_MINPLUS:        case OP_MINPLUS:
2155          case OP_MINPLUSI:
2156        case OP_POSPLUS:        case OP_POSPLUS:
2157          case OP_POSPLUSI:
2158        case OP_QUERY:        case OP_QUERY:
2159          case OP_QUERYI:
2160        case OP_MINQUERY:        case OP_MINQUERY:
2161          case OP_MINQUERYI:
2162        case OP_POSQUERY:        case OP_POSQUERY:
2163        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_POSQUERYI:
2164          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2165        break;        break;
2166        }        }
2167  #else  #else
2168      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2169  #endif  #endif
2170      }      }
2171    }    }
# Line 1796  instance of OP_RECURSE. Line 2182  instance of OP_RECURSE.
2182    
2183  Arguments:  Arguments:
2184    code        points to start of expression    code        points to start of expression
2185    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2186    
2187  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
2188  */  */
2189    
2190  static const uschar *  static const pcre_uchar *
2191  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2192  {  {
2193  for (;;)  for (;;)
2194    {    {
# Line 1841  for (;;) Line 2227  for (;;)
2227        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2228        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2229        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2230        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2231            code += 2;
2232        break;        break;
2233    
2234        case OP_MARK:        case OP_MARK:
# Line 1851  for (;;) Line 2238  for (;;)
2238        break;        break;
2239    
2240        case OP_THEN_ARG:        case OP_THEN_ARG:
2241        code += code[1+LINK_SIZE];        code += code[1];
2242        break;        break;
2243        }        }
2244    
2245      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2246    
2247      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2248    
2249      /* 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
2250      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
2251      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2252    
2253  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2254      if (utf8) switch(c)      if (utf) switch(c)
2255        {        {
2256        case OP_CHAR:        case OP_CHAR:
2257        case OP_CHARNC:        case OP_CHARI:
2258          case OP_NOT:
2259          case OP_NOTI:
2260        case OP_EXACT:        case OP_EXACT:
2261          case OP_EXACTI:
2262          case OP_NOTEXACT:
2263          case OP_NOTEXACTI:
2264        case OP_UPTO:        case OP_UPTO:
2265          case OP_UPTOI:
2266          case OP_NOTUPTO:
2267          case OP_NOTUPTOI:
2268        case OP_MINUPTO:        case OP_MINUPTO:
2269          case OP_MINUPTOI:
2270          case OP_NOTMINUPTO:
2271          case OP_NOTMINUPTOI:
2272        case OP_POSUPTO:        case OP_POSUPTO:
2273          case OP_POSUPTOI:
2274          case OP_NOTPOSUPTO:
2275          case OP_NOTPOSUPTOI:
2276        case OP_STAR:        case OP_STAR:
2277          case OP_STARI:
2278          case OP_NOTSTAR:
2279          case OP_NOTSTARI:
2280        case OP_MINSTAR:        case OP_MINSTAR:
2281          case OP_MINSTARI:
2282          case OP_NOTMINSTAR:
2283          case OP_NOTMINSTARI:
2284        case OP_POSSTAR:        case OP_POSSTAR:
2285          case OP_POSSTARI:
2286          case OP_NOTPOSSTAR:
2287          case OP_NOTPOSSTARI:
2288        case OP_PLUS:        case OP_PLUS:
2289          case OP_PLUSI:
2290          case OP_NOTPLUS:
2291          case OP_NOTPLUSI:
2292        case OP_MINPLUS:        case OP_MINPLUS:
2293          case OP_MINPLUSI:
2294          case OP_NOTMINPLUS:
2295          case OP_NOTMINPLUSI:
2296        case OP_POSPLUS:        case OP_POSPLUS:
2297          case OP_POSPLUSI:
2298          case OP_NOTPOSPLUS:
2299          case OP_NOTPOSPLUSI:
2300        case OP_QUERY:        case OP_QUERY:
2301          case OP_QUERYI:
2302          case OP_NOTQUERY:
2303          case OP_NOTQUERYI:
2304        case OP_MINQUERY:        case OP_MINQUERY:
2305          case OP_MINQUERYI:
2306          case OP_NOTMINQUERY:
2307          case OP_NOTMINQUERYI:
2308        case OP_POSQUERY:        case OP_POSQUERY:
2309        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_POSQUERYI:
2310          case OP_NOTPOSQUERY:
2311          case OP_NOTPOSQUERYI:
2312          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2313        break;        break;
2314        }        }
2315  #else  #else
2316      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2317  #endif  #endif
2318      }      }
2319    }    }
# Line 1908  bracket whose current branch will alread Line 2336  bracket whose current branch will alread
2336  Arguments:  Arguments:
2337    code        points to start of search    code        points to start of search
2338    endcode     points to where to stop    endcode     points to where to stop
2339    utf8        TRUE if in UTF8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2340    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2341    
2342  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2343  */  */
2344    
2345  static BOOL  static BOOL
2346  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2347    compile_data *cd)    BOOL utf, compile_data *cd)
2348  {  {
2349  register int c;  register int c;
2350  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2351       code < endcode;       code < endcode;
2352       code = first_significant_code(code + _pcre_OP_lengths[c], NULL, 0, TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2353    {    {
2354    const uschar *ccode;    const pcre_uchar *ccode;
2355    
2356    c = *code;    c = *code;
2357    
# Line 1937  for (code = first_significant_code(code Line 2365  for (code = first_significant_code(code
2365      continue;      continue;
2366      }      }
2367    
   /* Groups with zero repeats can of course be empty; skip them. */  
   
   if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO)  
     {  
     code += _pcre_OP_lengths[c];  
     do code += GET(code, 1); while (*code == OP_ALT);  
     c = *code;  
     continue;  
     }  
   
2368    /* For a recursion/subroutine call, if its end has been reached, which    /* For a recursion/subroutine call, if its end has been reached, which
2369    implies a subroutine call, we can scan it. */    implies a backward reference subroutine call, we can scan it. If it's a
2370      forward reference subroutine call, we can't. To detect forward reference
2371      we have to scan up the list that is kept in the workspace. This function is
2372      called only when doing the real compile, not during the pre-compile that
2373      measures the size of the compiled pattern. */
2374    
2375    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2376      {      {
2377      BOOL empty_branch = FALSE;      const pcre_uchar *scode;
2378      const uschar *scode = cd->start_code + GET(code, 1);      BOOL empty_branch;
2379    
2380        /* Test for forward reference */
2381    
2382        for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2383          if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;
2384    
2385        /* Not a forward reference, test for completed backward reference */
2386    
2387        empty_branch = FALSE;
2388        scode = cd->start_code + GET(code, 1);
2389      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2390    
2391        /* Completed backwards reference */
2392    
2393      do      do
2394        {        {
2395        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd))
2396          {          {
2397          empty_branch = TRUE;          empty_branch = TRUE;
2398          break;          break;
# Line 1965  for (code = first_significant_code(code Line 2400  for (code = first_significant_code(code
2400        scode += GET(scode, 1);        scode += GET(scode, 1);
2401        }        }
2402      while (*scode == OP_ALT);      while (*scode == OP_ALT);
2403    
2404      if (!empty_branch) return FALSE;  /* All branches are non-empty */      if (!empty_branch) return FALSE;  /* All branches are non-empty */
2405      continue;      continue;
2406      }      }
2407    
2408      /* Groups with zero repeats can of course be empty; skip them. */
2409    
2410      if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2411          c == OP_BRAPOSZERO)
2412        {
2413        code += PRIV(OP_lengths)[c];
2414        do code += GET(code, 1); while (*code == OP_ALT);
2415        c = *code;
2416        continue;
2417        }
2418    
2419      /* A nested group that is already marked as "could be empty" can just be
2420      skipped. */
2421    
2422      if (c == OP_SBRA  || c == OP_SBRAPOS ||
2423          c == OP_SCBRA || c == OP_SCBRAPOS)
2424        {
2425        do code += GET(code, 1); while (*code == OP_ALT);
2426        c = *code;
2427        continue;
2428        }
2429    
2430    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
2431    
2432    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)    if (c == OP_BRA  || c == OP_BRAPOS ||
2433          c == OP_CBRA || c == OP_CBRAPOS ||
2434          c == OP_ONCE || c == OP_ONCE_NC ||
2435          c == OP_COND)
2436      {      {
2437      BOOL empty_branch;      BOOL empty_branch;
2438      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */      if (GET(code, 1) == 0) return TRUE;    /* Hit unclosed bracket */
# Line 1987  for (code = first_significant_code(code Line 2448  for (code = first_significant_code(code
2448        empty_branch = FALSE;        empty_branch = FALSE;
2449        do        do
2450          {          {
2451          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
2452            empty_branch = TRUE;            empty_branch = TRUE;
2453          code += GET(code, 1);          code += GET(code, 1);
2454          }          }
# Line 2005  for (code = first_significant_code(code Line 2466  for (code = first_significant_code(code
2466      {      {
2467      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2468      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2469      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2470      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"
2471      here. */      here. */
2472    
2473  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2474      case OP_XCLASS:      case OP_XCLASS:
2475      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2476      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2017  for (code = first_significant_code(code Line 2478  for (code = first_significant_code(code
2478    
2479      case OP_CLASS:      case OP_CLASS:
2480      case OP_NCLASS:      case OP_NCLASS:
2481      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2482    
2483  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2484      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2485  #endif  #endif
2486    
# Line 2058  for (code = first_significant_code(code Line 2519  for (code = first_significant_code(code
2519      case OP_ALLANY:      case OP_ALLANY:
2520      case OP_ANYBYTE:      case OP_ANYBYTE:
2521      case OP_CHAR:      case OP_CHAR:
2522      case OP_CHARNC:      case OP_CHARI:
2523      case OP_NOT:      case OP_NOT:
2524        case OP_NOTI:
2525      case OP_PLUS:      case OP_PLUS:
2526      case OP_MINPLUS:      case OP_MINPLUS:
2527      case OP_POSPLUS:      case OP_POSPLUS:
# Line 2091  for (code = first_significant_code(code Line 2553  for (code = first_significant_code(code
2553      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2554      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2555      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2556      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2557          code += 2;
2558      break;      break;
2559    
2560      /* End of branch */      /* End of branch */
# Line 2099  for (code = first_significant_code(code Line 2562  for (code = first_significant_code(code
2562      case OP_KET:      case OP_KET:
2563      case OP_KETRMAX:      case OP_KETRMAX:
2564      case OP_KETRMIN:      case OP_KETRMIN:
2565        case OP_KETRPOS:
2566      case OP_ALT:      case OP_ALT:
2567      return TRUE;      return TRUE;
2568    
2569      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2570      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2571    
2572  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2573      case OP_STAR:      case OP_STAR:
2574        case OP_STARI:
2575      case OP_MINSTAR:      case OP_MINSTAR:
2576        case OP_MINSTARI:
2577      case OP_POSSTAR:      case OP_POSSTAR:
2578        case OP_POSSTARI:
2579      case OP_QUERY:      case OP_QUERY:
2580        case OP_QUERYI:
2581      case OP_MINQUERY:      case OP_MINQUERY:
2582        case OP_MINQUERYI:
2583      case OP_POSQUERY:      case OP_POSQUERY:
2584      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      case OP_POSQUERYI:
2585        if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2586      break;      break;
2587    
2588      case OP_UPTO:      case OP_UPTO:
2589        case OP_UPTOI:
2590      case OP_MINUPTO:      case OP_MINUPTO:
2591        case OP_MINUPTOI:
2592      case OP_POSUPTO:      case OP_POSUPTO:
2593      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      case OP_POSUPTOI:
2594        if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);
2595      break;      break;
2596  #endif  #endif
2597    
# Line 2132  for (code = first_significant_code(code Line 2605  for (code = first_significant_code(code
2605      break;      break;
2606    
2607      case OP_THEN_ARG:      case OP_THEN_ARG:
2608      code += code[1+LINK_SIZE];      code += code[1];
2609      break;      break;
2610    
2611      /* None of the remaining opcodes are required to match a character. */      /* None of the remaining opcodes are required to match a character. */
# Line 2155  return TRUE; Line 2628  return TRUE;
2628  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
2629  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,
2630  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.
2631    This function is called only during the real compile, not during the
2632    pre-compile.
2633    
2634  Arguments:  Arguments:
2635    code        points to start of the recursion    code        points to start of the recursion
2636    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2637    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2638    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2639    cd          pointers to tables etc    cd          pointers to tables etc
2640    
2641  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2642  */  */
2643    
2644  static BOOL  static BOOL
2645  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2646    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2647  {  {
2648  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2649    {    {
2650    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
2651      return FALSE;      return FALSE;
2652    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2653    }    }
# Line 2205  where Perl recognizes it as the POSIX cl Line 2680  where Perl recognizes it as the POSIX cl
2680  "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,
2681  I think.  I think.
2682    
2683    A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.
2684    It seems that the appearance of a nested POSIX class supersedes an apparent
2685    external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or
2686    a digit.
2687    
2688    In Perl, unescaped square brackets may also appear as part of class names. For
2689    example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for
2690    [:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not
2691    seem right at all. PCRE does not allow closing square brackets in POSIX class
2692    names.
2693    
2694  Arguments:  Arguments:
2695    ptr      pointer to the initial [    ptr      pointer to the initial [
2696    endptr   where to return the end pointer    endptr   where to return the end pointer
# Line 2213  Returns:   TRUE or FALSE Line 2699  Returns:   TRUE or FALSE
2699  */  */
2700    
2701  static BOOL  static BOOL
2702  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2703  {  {
2704  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
2705  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2706  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != 0; ptr++)
2707    {    {
2708    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2709        ptr++;
2710      else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
2711      else
2712      {      {
     if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;  
2713      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2714        {        {
2715        *endptr = ptr;        *endptr = ptr;
2716        return TRUE;        return TRUE;
2717        }        }
2718        if (*ptr == CHAR_LEFT_SQUARE_BRACKET &&
2719             (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2720              ptr[1] == CHAR_EQUALS_SIGN) &&
2721            check_posix_syntax(ptr, endptr))
2722          return FALSE;
2723      }      }
2724    }    }
2725  return FALSE;  return FALSE;
# Line 2250  Returns:     a value representing the na Line 2743  Returns:     a value representing the na
2743  */  */
2744    
2745  static int  static int
2746  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2747  {  {
2748  const char *pn = posix_names;  const char *pn = posix_names;
2749  register int yield = 0;  register int yield = 0;
2750  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2751    {    {
2752    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2753      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;
2754    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2755    yield++;    yield++;
2756    }    }
# Line 2289  value in the reference (which is a group Line 2782  value in the reference (which is a group
2782  Arguments:  Arguments:
2783    group      points to the start of the group    group      points to the start of the group
2784    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2785    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2786    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2787    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
2788    
# Line 2297  Returns:     nothing Line 2790  Returns:     nothing
2790  */  */
2791    
2792  static void  static void
2793  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2794    uschar *save_hwm)    pcre_uchar *save_hwm)
2795  {  {
2796  uschar *ptr = group;  pcre_uchar *ptr = group;
2797    
2798  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2799    {    {
2800    int offset;    int offset;
2801    uschar *hc;    pcre_uchar *hc;
2802    
2803    /* 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
2804    reference. */    reference. */
# Line 2350  Arguments: Line 2843  Arguments:
2843  Returns:         new code pointer  Returns:         new code pointer
2844  */  */
2845    
2846  static uschar *  static pcre_uchar *
2847  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2848  {  {
2849  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2850  *code++ = 255;  *code++ = 255;
2851  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2852  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2853  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2854  }  }
2855    
2856    
# Line 2379  Returns:             nothing Line 2872  Returns:             nothing
2872  */  */
2873    
2874  static void  static void
2875  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2876  {  {
2877  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2878  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2393  PUT(previous_callout, 2 + LINK_SIZE, len Line 2886  PUT(previous_callout, 2 + LINK_SIZE, len
2886  *************************************************/  *************************************************/
2887    
2888  /* 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
2889  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
2890  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
2891  start address.  start address. A character with multiple other cases is returned on its own
2892    with a special return value.
2893    
2894  Arguments:  Arguments:
2895    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2403  Arguments: Line 2897  Arguments:
2897    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2898    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2899    
2900  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2901                   0 when a range is returned
2902                  >0 the CASESET offset for char with multiple other cases
2903                    in this case, ocptr contains the original
2904  */  */
2905    
2906  static BOOL  static int
2907  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,
2908    unsigned int *odptr)    unsigned int *odptr)
2909  {  {
2910  unsigned int c, othercase, next;  unsigned int c, othercase, next;
2911    int co;
2912    
2913    /* Find the first character that has an other case. If it has multiple other
2914    cases, return its case offset value. */
2915    
2916  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2917    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2918      if ((co = UCD_CASESET(c)) != 0)
2919        {
2920        *ocptr = c++;   /* Character that has the set */
2921        *cptr = c;      /* Rest of input range */
2922        return co;
2923        }
2924      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2925      }
2926    
2927  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2928    
2929  *ocptr = othercase;  *ocptr = othercase;
2930  next = othercase + 1;  next = othercase + 1;
# Line 2426  for (++c; c <= d; c++) Line 2935  for (++c; c <= d; c++)
2935    next++;    next++;
2936    }    }
2937    
2938  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2939  *cptr = c;  *cptr = c;             /* Rest of input range */
2940    return 0;
 return TRUE;  
2941  }  }
2942    
2943    
# Line 2453  Returns:       TRUE if auto-possessifyin Line 2961  Returns:       TRUE if auto-possessifyin
2961  static BOOL  static BOOL
2962  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(int c, int ptype, int pdata, BOOL negated)
2963  {  {
2964    #ifdef SUPPORT_UCP
2965    const pcre_uint32 *p;
2966    #endif
2967    
2968  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2969    
2970  switch(ptype)  switch(ptype)
2971    {    {
2972    case PT_LAMP:    case PT_LAMP:
# Line 2462  switch(ptype) Line 2975  switch(ptype)
2975            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
2976    
2977    case PT_GC:    case PT_GC:
2978    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2979    
2980    case PT_PC:    case PT_PC:
2981    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2473  switch(ptype) Line 2986  switch(ptype)
2986    /* These are specials */    /* These are specials */
2987    
2988    case PT_ALNUM:    case PT_ALNUM:
2989    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2990            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
2991    
2992    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
2993    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2994            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2995            == negated;            == negated;
2996    
2997    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
2998    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2999            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
3000            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
3001            == negated;            == negated;
3002    
3003    case PT_WORD:    case PT_WORD:
3004    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3005            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3006            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3007    
3008    #ifdef SUPPORT_UCP
3009      case PT_CLIST:
3010      p = PRIV(ucd_caseless_sets) + prop->caseset;
3011      for (;;)
3012        {
3013        if ((unsigned int)c < *p) return !negated;
3014        if ((unsigned int)c == *p++) return negated;
3015        }
3016      break;  /* Control never reaches here */
3017    #endif
3018    }    }
3019    
3020  return FALSE;  return FALSE;
3021  }  }
3022  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2508  sense to automatically possessify the re Line 3033  sense to automatically possessify the re
3033    
3034  Arguments:  Arguments:
3035    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3036    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3037    ptr           next character in pattern    ptr           next character in pattern
3038    options       options bits    options       options bits
3039    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2517  Returns:        TRUE if possessifying is Line 3042  Returns:        TRUE if possessifying is
3042  */  */
3043    
3044  static BOOL  static BOOL
3045  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3046    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3047  {  {
3048  int c, next;  pcre_int32 c = NOTACHAR; // FIXMEchpe pcre_uint32
3049    pcre_int32 next;
3050    int escape;
3051  int op_code = *previous++;  int op_code = *previous++;
3052    
3053  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2529  if ((options & PCRE_EXTENDED) != 0) Line 3056  if ((options & PCRE_EXTENDED) != 0)
3056    {    {
3057    for (;;)    for (;;)
3058      {      {
3059      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3060      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3061        {        {
3062        ptr++;        ptr++;
3063        while (*ptr != 0)        while (*ptr != 0)
3064          {          {
3065          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3066          ptr++;          ptr++;
3067  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3068          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3069  #endif  #endif
3070          }          }
3071        }        }
# Line 2552  value is a character, a negative value i Line 3079  value is a character, a negative value i
3079  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3080    {    {
3081    int temperrorcode = 0;    int temperrorcode = 0;
3082    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3083    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3084    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3085    }    }
3086    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
3087    {    {
3088  #ifdef SUPPORT_UTF8    escape = 0;
3089    if (utf8) { GETCHARINC(next, ptr); } else  #ifdef SUPPORT_UTF
3090      if (utf) { GETCHARINC(next, ptr); } else
3091  #endif  #endif
3092    next = *ptr++;    next = *ptr++;
3093    }    }
   
3094  else return FALSE;  else return FALSE;
3095    
3096  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2573  if ((options & PCRE_EXTENDED) != 0) Line 3099  if ((options & PCRE_EXTENDED) != 0)
3099    {    {
3100    for (;;)    for (;;)
3101      {      {
3102      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3103      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3104        {        {
3105        ptr++;        ptr++;
3106        while (*ptr != 0)        while (*ptr != 0)
3107          {          {
3108          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3109          ptr++;          ptr++;
3110  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3111          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3112  #endif  #endif
3113          }          }
3114        }        }
# Line 2593  if ((options & PCRE_EXTENDED) != 0) Line 3119  if ((options & PCRE_EXTENDED) != 0)
3119  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3120    
3121  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3122    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)
3123      return FALSE;      return FALSE;
3124    
3125  /* 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. */  
3126    
3127  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3128        op_code == OP_NOT || op_code == OP_NOTI)
3129      //if (escape == 0) switch(op_code)
3130    {    {
3131    case OP_CHAR:  #ifdef SUPPORT_UTF
 #ifdef SUPPORT_UTF8  
3132    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3133  #else  #else
3134    c = *previous;    c = *previous;
3135  #endif  #endif
3136    return c != next;    }
3137    
3138    /* For CHARNC (caseless character) we must check the other case. If we have  /* Now compare the next item with the previous opcode. First, handle cases when
3139    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3140    
3141    case OP_CHARNC:  if (escape == 0)
3142  #ifdef SUPPORT_UTF8    {
3143    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3144  #else    case, which maps to the special PT_CLIST property. Check this first. */
3145    c = *previous;  
3146    #ifdef SUPPORT_UCP
3147      if (utf && (unsigned int)c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3148        {
3149        int ocs = UCD_CASESET(next);
3150        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3151        }
3152  #endif  #endif
3153    if (c == next) return FALSE;  
3154  #ifdef SUPPORT_UTF8    switch(op_code)
   if (utf8)  
3155      {      {
3156      unsigned int othercase;      case OP_CHAR:
3157      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3158    
3159        /* For CHARI (caseless character) we must check the other case. If we have
3160        Unicode property support, we can use it to test the other case of
3161        high-valued characters. We know that next can have only one other case,
3162        because multi-other-case characters are dealt with above. */
3163    
3164        case OP_CHARI:
3165        if (c == next) return FALSE;
3166    #ifdef SUPPORT_UTF
3167        if (utf)
3168          {
3169          unsigned int othercase;
3170          if (next < 128) othercase = cd->fcc[next]; else
3171  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3172      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE((unsigned int)next);
3173  #else  #else
3174      othercase = NOTACHAR;        othercase = NOTACHAR;
3175  #endif  #endif
3176      return (unsigned int)c != othercase;        return (unsigned int)c != othercase;
3177      }        }
3178    else      else
3179  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3180    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */      return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3181    
3182    /* For OP_NOT, its data is always a single-byte character. */      case OP_NOT:
3183        return c == next;
3184    case OP_NOT:  
3185    if ((c = *previous) == next) return TRUE;      case OP_NOTI:
3186    if ((options & PCRE_CASELESS) == 0) return FALSE;      if (c == next) return TRUE;
3187  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3188    if (utf8)      if (utf)
3189      {        {
3190      unsigned int othercase;        unsigned int othercase;
3191      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3192  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3193      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE((unsigned int)next);
3194  #else  #else
3195      othercase = NOTACHAR;        othercase = NOTACHAR;
3196  #endif  #endif
3197      return (unsigned int)c == othercase;        return (unsigned int)c == othercase;
3198      }        }
3199    else      else
3200  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3201    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */      return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3202    
3203    /* 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.
3204    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. */
3205    
3206    case OP_DIGIT:      case OP_DIGIT:
3207    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3208    
3209    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3210    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3211    
3212    case OP_WHITESPACE:      case OP_WHITESPACE:
3213    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3214    
3215    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3216    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3217    
3218    case OP_WORDCHAR:      case OP_WORDCHAR:
3219    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3220    
3221    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3222    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3223    
3224    case OP_HSPACE:      case OP_HSPACE:
3225    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3226    switch(next)      switch(next)
3227      {        {
3228      case 0x09:        HSPACE_CASES:
3229      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;  
     }  
3230    
3231    case OP_ANYNL:        default:
3232    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3233    case OP_NOT_VSPACE:        }
3234    switch(next)  
3235      {      case OP_ANYNL:
3236      case 0x0a:      case OP_VSPACE:
3237      case 0x0b:      case OP_NOT_VSPACE:
3238      case 0x0c:      switch(next)
3239      case 0x0d:        {
3240      case 0x85:        VSPACE_CASES:
3241      case 0x2028:        return op_code == OP_NOT_VSPACE;
3242      case 0x2029:  
3243      return op_code == OP_NOT_VSPACE;        default:
3244      default:        return op_code != OP_NOT_VSPACE;
3245      return op_code != OP_NOT_VSPACE;        }
     }  
3246    
3247  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3248    case OP_PROP:      case OP_PROP:
3249    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3250    
3251    case OP_NOTPROP:      case OP_NOTPROP:
3252    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3253  #endif  #endif
3254    
3255    default:      default:
3256    return FALSE;      return FALSE;
3257        }
3258    }    }
3259    
   
3260  /* 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
3261  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
3262  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 2745  replaced by OP_PROP codes when PCRE_UCP Line 3266  replaced by OP_PROP codes when PCRE_UCP
3266  switch(op_code)  switch(op_code)
3267    {    {
3268    case OP_CHAR:    case OP_CHAR:
3269    case OP_CHARNC:    case OP_CHARI:
3270  #ifdef SUPPORT_UTF8    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3271      {      {
3272      case ESC_d:      case ESC_d:
3273      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3274    
3275      case ESC_D:      case ESC_D:
3276      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3277    
3278      case ESC_s:      case ESC_s:
3279      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3280    
3281      case ESC_S:      case ESC_S:
3282      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3283    
3284      case ESC_w:      case ESC_w:
3285      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3286    
3287      case ESC_W:      case ESC_W:
3288      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3289    
3290      case ESC_h:      case ESC_h:
3291      case ESC_H:      case ESC_H:
3292      switch(c)      switch(c)
3293        {        {
3294        case 0x09:        HSPACE_CASES:
3295        case 0x20:        return escape != ESC_h;
3296        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;  
3297        default:        default:
3298        return -next == ESC_h;        return escape == ESC_h;
3299        }        }
3300    
3301      case ESC_v:      case ESC_v:
3302      case ESC_V:      case ESC_V:
3303      switch(c)      switch(c)
3304        {        {
3305        case 0x0a:        VSPACE_CASES:
3306        case 0x0b:        return escape != ESC_v;
3307        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3308        default:        default:
3309        return -next == ESC_v;        return escape == ESC_v;
3310        }        }
3311    
3312      /* 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
3313      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3314      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3315    
3316  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3317      case ESC_du:      case ESC_du:
# Line 2828  switch(op_code) Line 3322  switch(op_code)
3322      case ESC_SU:      case ESC_SU:
3323        {        {
3324        int temperrorcode = 0;        int temperrorcode = 0;
3325        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3326        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3327        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3328        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3329        }        }
# Line 2851  switch(op_code) Line 3345  switch(op_code)
3345        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. */
3346    
3347        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3348          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)
3349            return FALSE;            return FALSE;
3350    
3351        /* Do the property check. */        /* Do the property check. */
3352    
3353        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3354        }        }
3355  #endif  #endif
3356    
# Line 2871  switch(op_code) Line 3365  switch(op_code)
3365    these op-codes are never generated.) */    these op-codes are never generated.) */
3366    
3367    case OP_DIGIT:    case OP_DIGIT:
3368    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3369           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3370    
3371    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3372    return next == -ESC_d;    return escape == ESC_d;
3373    
3374    case OP_WHITESPACE:    case OP_WHITESPACE:
3375    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3376    
3377    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3378    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3379    
3380    case OP_HSPACE:    case OP_HSPACE:
3381    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3382           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3383    
3384    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3385    return next == -ESC_h;    return escape == ESC_h;
3386    
3387    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3388    case OP_ANYNL:    case OP_ANYNL:
3389    case OP_VSPACE:    case OP_VSPACE:
3390    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3391    
3392    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3393    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3394    
3395    case OP_WORDCHAR:    case OP_WORDCHAR:
3396    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3397           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3398    
3399    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3400    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3401    
3402    default:    default:
3403    return FALSE;    return FALSE;
# Line 2915  switch(op_code) Line 3409  switch(op_code)
3409    
3410    
3411  /*************************************************  /*************************************************
3412    *        Add a character or range to a class     *
3413    *************************************************/
3414    
3415    /* This function packages up the logic of adding a character or range of
3416    characters to a class. The character values in the arguments will be within the
3417    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3418    mutually recursive with the function immediately below.
3419    
3420    Arguments:
3421      classbits     the bit map for characters < 256
3422      uchardptr     points to the pointer for extra data
3423      options       the options word
3424      cd            contains pointers to tables etc.
3425      start         start of range character
3426      end           end of range character
3427    
3428    Returns:        the number of < 256 characters added
3429                    the pointer to extra data is updated
3430    */
3431    
3432    static int
3433    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3434      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3435    {
3436    pcre_uint32 c;
3437    int n8 = 0;
3438    
3439    /* If caseless matching is required, scan the range and process alternate
3440    cases. In Unicode, there are 8-bit characters that have alternate cases that
3441    are greater than 255 and vice-versa. Sometimes we can just extend the original
3442    range. */
3443    
3444    if ((options & PCRE_CASELESS) != 0)
3445      {
3446    #ifdef SUPPORT_UCP
3447      if ((options & PCRE_UTF8) != 0)
3448        {
3449        int rc;
3450        pcre_uint32 oc, od;
3451    
3452        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3453        c = start;
3454    
3455        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3456          {
3457          /* Handle a single character that has more than one other case. */
3458    
3459          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3460            PRIV(ucd_caseless_sets) + rc, oc);
3461    
3462          /* Do nothing if the other case range is within the original range. */
3463    
3464          else if (oc >= start && od <= end) continue;
3465    
3466          /* Extend the original range if there is overlap, noting that if oc < c, we
3467          can't have od > end because a subrange is always shorter than the basic
3468          range. Otherwise, use a recursive call to add the additional range. */
3469    
3470          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3471          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3472          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3473          }
3474        }
3475      else
3476    #endif  /* SUPPORT_UCP */
3477    
3478      /* Not UTF-mode, or no UCP */
3479    
3480      for (c = start; c <= end && c < 256; c++)
3481        {
3482        SETBIT(classbits, cd->fcc[c]);
3483        n8++;
3484        }
3485      }
3486    
3487    /* Now handle the original range. Adjust the final value according to the bit
3488    length - this means that the same lists of (e.g.) horizontal spaces can be used
3489    in all cases. */
3490    
3491    #if defined COMPILE_PCRE8
3492    #ifdef SUPPORT_UTF
3493      if ((options & PCRE_UTF8) == 0)
3494    #endif
3495      if (end > 0xff) end = 0xff;
3496    
3497    #elif defined COMPILE_PCRE16
3498    #ifdef SUPPORT_UTF
3499      if ((options & PCRE_UTF16) == 0)
3500    #endif
3501      if (end > 0xffff) end = 0xffff;
3502    
3503    #endif /* COMPILE_PCRE[8|16] */
3504    
3505    /* If all characters are less than 256, use the bit map. Otherwise use extra
3506    data. */
3507    
3508    if (end < 0x100)
3509      {
3510      for (c = start; c <= end; c++)
3511        {
3512        n8++;
3513        SETBIT(classbits, c);
3514        }
3515      }
3516    
3517    else
3518      {
3519      pcre_uchar *uchardata = *uchardptr;
3520    
3521    #ifdef SUPPORT_UTF
3522      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3523        {
3524        if (start < end)
3525          {
3526          *uchardata++ = XCL_RANGE;
3527          uchardata += PRIV(ord2utf)(start, uchardata);
3528          uchardata += PRIV(ord2utf)(end, uchardata);
3529          }
3530        else if (start == end)
3531          {
3532          *uchardata++ = XCL_SINGLE;
3533          uchardata += PRIV(ord2utf)(start, uchardata);
3534          }
3535        }
3536      else
3537    #endif  /* SUPPORT_UTF */
3538    
3539      /* Without UTF support, character values are constrained by the bit length,
3540      and can only be > 256 for 16-bit and 32-bit libraries. */
3541    
3542    #ifdef COMPILE_PCRE8
3543        {}
3544    #else
3545      if (start < end)
3546        {
3547        *uchardata++ = XCL_RANGE;
3548        *uchardata++ = start;
3549        *uchardata++ = end;
3550        }
3551      else if (start == end)
3552        {
3553        *uchardata++ = XCL_SINGLE;
3554        *uchardata++ = start;
3555        }
3556    #endif
3557    
3558      *uchardptr = uchardata;   /* Updata extra data pointer */
3559      }
3560    
3561    return n8;    /* Number of 8-bit characters */
3562    }
3563    
3564    
3565    
3566    
3567    /*************************************************
3568    *        Add a list of characters to a class     *
3569    *************************************************/
3570    
3571    /* This function is used for adding a list of case-equivalent characters to a
3572    class, and also for adding a list of horizontal or vertical whitespace. If the
3573    list is in order (which it should be), ranges of characters are detected and
3574    handled appropriately. This function is mutually recursive with the function
3575    above.
3576    
3577    Arguments:
3578      classbits     the bit map for characters < 256
3579      uchardptr     points to the pointer for extra data
3580      options       the options word
3581      cd            contains pointers to tables etc.
3582      p             points to row of 32-bit values, terminated by NOTACHAR
3583      except        character to omit; this is used when adding lists of
3584                      case-equivalent characters to avoid including the one we
3585                      already know about
3586    
3587    Returns:        the number of < 256 characters added
3588                    the pointer to extra data is updated
3589    */
3590    
3591    static int
3592    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3593      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3594    {
3595    int n8 = 0;
3596    while (p[0] < NOTACHAR)
3597      {
3598      int n = 0;
3599      if (p[0] != except)
3600        {
3601        while(p[n+1] == p[0] + n + 1) n++;
3602        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3603        }
3604      p += n + 1;
3605      }
3606    return n8;
3607    }
3608    
3609    
3610    
3611    /*************************************************
3612    *    Add characters not in a list to a class     *
3613    *************************************************/
3614    
3615    /* This function is used for adding the complement of a list of horizontal or
3616    vertical whitespace to a class. The list must be in order.
3617    
3618    Arguments:
3619      classbits     the bit map for characters < 256
3620      uchardptr     points to the pointer for extra data
3621      options       the options word
3622      cd            contains pointers to tables etc.
3623      p             points to row of 32-bit values, terminated by NOTACHAR
3624    
3625    Returns:        the number of < 256 characters added
3626                    the pointer to extra data is updated
3627    */
3628    
3629    static int
3630    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3631      int options, compile_data *cd, const pcre_uint32 *p)
3632    {
3633    BOOL utf = (options & PCRE_UTF8) != 0;
3634    int n8 = 0;
3635    if (p[0] > 0)
3636      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3637    while (p[0] < NOTACHAR)
3638      {
3639      while (p[1] == p[0] + 1) p++;
3640      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3641        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3642      p++;
3643      }
3644    return n8;
3645    }
3646    
3647    
3648    
3649    /*************************************************
3650  *           Compile one branch                   *  *           Compile one branch                   *
3651  *************************************************/  *************************************************/
3652    
# Line 2929  Arguments: Line 3661  Arguments:
3661    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3662    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3663    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3664    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)
3665    reqbyteptr     set to the last literal character required, else < 0    reqcharptr     set to the last literal character required, else < 0
3666    bcptr          points to current branch chain    bcptr          points to current branch chain
3667      cond_depth     conditional nesting depth
3668    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
3669    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
3670                   points to length accumulator during pre-compile phase                   points to length accumulator during pre-compile phase
# Line 2941  Returns:         TRUE on success Line 3674  Returns:         TRUE on success
3674  */  */
3675    
3676  static BOOL  static BOOL
3677  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3678    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,
3679      pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,
3680    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3681  {  {
3682  int repeat_type, op_type;  int repeat_type, op_type;
3683  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3684  int bravalue = 0;  int bravalue = 0;
3685  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3686  int firstbyte, reqbyte;  pcre_int32 firstchar, reqchar;
3687  int zeroreqbyte, zerofirstbyte;  pcre_int32 zeroreqchar, zerofirstchar;
3688  int req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3689  int options = *optionsptr;  int options = *optionsptr;               /* May change dynamically */
3690  int after_manual_callout = 0;  int after_manual_callout = 0;
3691  int length_prevgroup = 0;  int length_prevgroup = 0;
3692  register int c;  register int c;
3693  register uschar *code = *codeptr;  int escape;
3694  uschar *last_code = code;  register pcre_uchar *code = *codeptr;
3695  uschar *orig_code = code;  pcre_uchar *last_code = code;
3696  uschar *tempcode;  pcre_uchar *orig_code = code;
3697    pcre_uchar *tempcode;
3698  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3699  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3700  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3701  const uschar *tempptr;  const pcre_uchar *tempptr;
3702  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3703  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3704  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3705  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3706  uschar classbits[32];  pcre_uint8 classbits[32];
3707    
3708  #ifdef SUPPORT_UTF8  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we
3709  BOOL class_utf8;  must not do this for other options (e.g. PCRE_EXTENDED) because they may change
3710  BOOL utf8 = (options & PCRE_UTF8) != 0;  dynamically as we process the pattern. */
3711  uschar *class_utf8data;  
3712  uschar *class_utf8data_base;  #ifdef SUPPORT_UTF
3713  uschar utf8_char[6];  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3714    BOOL utf = (options & PCRE_UTF8) != 0;
3715    pcre_uchar utf_chars[6];
3716  #else  #else
3717  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3718  uschar *utf8_char = NULL;  #endif
3719    
3720    /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3721    class_uchardata always so that it can be passed to add_to_class() always,
3722    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3723    alternative calls for the different cases. */
3724    
3725    pcre_uchar *class_uchardata;
3726    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3727    BOOL xclass;
3728    pcre_uchar *class_uchardata_base;
3729  #endif  #endif
3730    
3731  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 2992  greedy_non_default = greedy_default ^ 1; Line 3739  greedy_non_default = greedy_default ^ 1;
3739    
3740  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3741  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
3742  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
3743  find one.  find one.
3744    
3745  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
3746  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
3747  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3748  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3749    
3750  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;
3751    
3752  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3753  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
3754  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
3755  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3756    value. This is used only for ASCII characters. */
3757    
3758  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3759    
3760  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3761    
# Line 3019  for (;; ptr++) Line 3767  for (;; ptr++)
3767    BOOL is_quantifier;    BOOL is_quantifier;
3768    BOOL is_recurse;    BOOL is_recurse;
3769    BOOL reset_bracount;    BOOL reset_bracount;
3770    int class_charcount;    int class_has_8bitchar;
3771    int class_lastchar;    int class_one_char;
3772    int newoptions;    int newoptions;
3773    int recno;    int recno;
3774    int refsign;    int refsign;
3775    int skipbytes;    int skipbytes;
3776    int subreqbyte;    int subreqchar;
3777    int subfirstbyte;    int subfirstchar;
3778    int terminator;    int terminator;
3779    int mclength;    int mclength;
3780    uschar mcbuffer[8];    int tempbracount;
3781      int ec; // FIXMEchpe pcre_uint32
3782      pcre_uchar mcbuffer[8];
3783    
3784    /* Get next byte in the pattern */    /* Get next character in the pattern */
3785    
3786    c = *ptr;    c = *ptr;
3787    
# Line 3053  for (;; ptr++) Line 3803  for (;; ptr++)
3803  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3804      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3805  #endif  #endif
3806      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */      if (code > cd->start_workspace + cd->workspace_size -
3807            WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3808        {        {
3809        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3810        goto FAILED;        goto FAILED;
# Line 3076  for (;; ptr++) Line 3827  for (;; ptr++)
3827        }        }
3828    
3829      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3830      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3831          (int)(code - last_code), c, c));
3832    
3833      /* 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
3834      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 3086  for (;; ptr++) Line 3838  for (;; ptr++)
3838        {        {
3839        if (previous > orig_code)        if (previous > orig_code)
3840          {          {
3841          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3842          code -= previous - orig_code;          code -= previous - orig_code;
3843          previous = orig_code;          previous = orig_code;
3844          }          }
# Line 3102  for (;; ptr++) Line 3854  for (;; ptr++)
3854    /* 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
3855    reference list. */    reference list. */
3856    
3857    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3858               WORK_SIZE_SAFETY_MARGIN)
3859      {      {
3860      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3861      goto FAILED;      goto FAILED;
# Line 3150  for (;; ptr++) Line 3903  for (;; ptr++)
3903      previous_callout = NULL;      previous_callout = NULL;
3904      }      }
3905    
3906    /* In extended mode, skip white space and comments */    /* In extended mode, skip white space and comments. */
3907    
3908    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
3909      {      {
3910      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
3911      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3912        {        {
3913        ptr++;        ptr++;
3914        while (*ptr != 0)        while (*ptr != 0)
3915          {          {
3916          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3917          ptr++;          ptr++;
3918  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3919          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3920  #endif  #endif
3921          }          }
3922        if (*ptr != 0) continue;        if (*ptr != 0) continue;
# Line 3187  for (;; ptr++) Line 3940  for (;; ptr++)
3940      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
3941      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3942      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3943      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
3944      *reqbyteptr = reqbyte;      *reqcharptr = reqchar;
3945      *codeptr = code;      *codeptr = code;
3946      *ptrptr = ptr;      *ptrptr = ptr;
3947      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3209  for (;; ptr++) Line 3962  for (;; ptr++)
3962      the setting of any following char as a first character. */      the setting of any following char as a first character. */
3963    
3964      case CHAR_CIRCUMFLEX_ACCENT:      case CHAR_CIRCUMFLEX_ACCENT:
3965        previous = NULL;
3966      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3967        {        {
3968        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3969          *code++ = OP_CIRCM;
3970        }        }
3971      previous = NULL;      else *code++ = OP_CIRC;
     *code++ = OP_CIRC;  
3972      break;      break;
3973    
3974      case CHAR_DOLLAR_SIGN:      case CHAR_DOLLAR_SIGN:
3975      previous = NULL;      previous = NULL;
3976      *code++ = OP_DOLL;      *code++ = ((options & PCRE_MULTILINE) != 0)? OP_DOLLM : OP_DOLL;
3977      break;      break;
3978    
3979      /* 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
3980      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
3981    
3982      case CHAR_DOT:      case CHAR_DOT:
3983      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3984      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
3985      zeroreqbyte = reqbyte;      zeroreqchar = reqchar;
3986      previous = code;      previous = code;
3987      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
3988      break;      break;
# Line 3283  for (;; ptr++) Line 4037  for (;; ptr++)
4037          {          {
4038          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
4039            ptr++;            ptr++;
4040          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)  
4041            ptr += 3;            ptr += 3;
4042          else          else
4043            break;            break;
# Line 3303  for (;; ptr++) Line 4056  for (;; ptr++)
4056          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4057        {        {
4058        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4059        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4060        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
4061        break;        break;
4062        }        }
4063    
# Line 3314  for (;; ptr++) Line 4067  for (;; ptr++)
4067    
4068      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4069    
4070      /* 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:
4071      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 <
4072      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
4073        character. */
4074    
4075      class_charcount = 0;      class_has_8bitchar = 0;
4076      class_lastchar = -1;      class_one_char = 0;
4077    
4078      /* 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
4079      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
4080      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
4081      */      map. */
4082    
4083      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4084    
4085  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4086      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;
4087      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4088      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4089  #endif  #endif
4090    
4091      /* 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 3340  for (;; ptr++) Line 4094  for (;; ptr++)
4094    
4095      if (c != 0) do      if (c != 0) do
4096        {        {
4097        const uschar *oldptr;        const pcre_uchar *oldptr;
4098    
4099  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4100        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
4101          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4102          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
4103          }          }
4104    #endif
4105    
4106        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4107          /* In the pre-compile phase, accumulate the length of any extra
4108        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
4109        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4110        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4111          however. */
4112        if (lengthptr != NULL)  
4113          {        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4114          *lengthptr += class_utf8data - class_utf8data_base;          {
4115          class_utf8data = class_utf8data_base;          xclass = TRUE;
4116            *lengthptr += class_uchardata - class_uchardata_base;
4117            class_uchardata = class_uchardata_base;
4118          }          }
   
4119  #endif  #endif
4120    
4121        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3386  for (;; ptr++) Line 4143  for (;; ptr++)
4143          {          {
4144          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
4145          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
4146          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
4147          uschar pbits[32];          pcre_uint8 pbits[32];
4148    
4149          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
4150            {            {
# Line 3414  for (;; ptr++) Line 4171  for (;; ptr++)
4171          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4172          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4173    
4174          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4175            posix_class = 0;            posix_class = 0;
4176    
4177          /* 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 3442  for (;; ptr++) Line 4199  for (;; ptr++)
4199          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
4200    
4201          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
4202            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
4203    
4204          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
4205    
# Line 3457  for (;; ptr++) Line 4214  for (;; ptr++)
4214              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4215            }            }
4216    
4217          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4218          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4219    
4220          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3473  for (;; ptr++) Line 4230  for (;; ptr++)
4230            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4231    
4232          ptr = tempptr + 1;          ptr = tempptr + 1;
4233          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 character. */
4234            class_has_8bitchar = 1;
4235            /* Every class contains at least two characters. */
4236            class_one_char = 2;
4237          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4238          }          }
4239    
4240        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
4241        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
4242        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
4243        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
4244        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_one_char bigger
4245        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
4246          as literal characters (by default), or are faulted if
4247        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4248    
4249        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4250          {          {
4251          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4252    
4253          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4254    
4255          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4256          else if (-c == ESC_Q)            /* Handle start of quoted string */            c = ec;
4257            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4258            else if (escape == ESC_N)            /* \N is not supported in a class */
4259              {
4260              *errorcodeptr = ERR71;
4261              goto FAILED;
4262              }
4263            else if (escape == ESC_Q)            /* Handle start of quoted string */
4264            {            {
4265            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4266              {              {
# Line 3500  for (;; ptr++) Line 4269  for (;; ptr++)
4269            else inescq = TRUE;            else inescq = TRUE;
4270            continue;            continue;
4271            }            }
4272          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4273    
4274          if (c < 0)          else
4275            {            {
4276            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4277            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
4278              class_has_8bitchar++;
4279              /* Every class contains at least two characters. */
4280              class_one_char += 2;
4281    
4282            switch (-c)            switch (escape)
4283              {              {
4284  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4285              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3517  for (;; ptr++) Line 4289  for (;; ptr++)
4289              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4290              case ESC_SU:              case ESC_SU:
4291              nestptr = ptr;              nestptr = ptr;
4292              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4293              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4294              continue;              continue;
4295  #endif  #endif
4296              case ESC_d:              case ESC_d:
# Line 3540  for (;; ptr++) Line 4312  for (;; ptr++)
4312              continue;              continue;
4313    
4314              /* 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
4315              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4316              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4317                EBCDIC, so we lazily just adjust the appropriate bit. */
4318    
4319              case ESC_s:              case ESC_s:
4320              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
4321              classbits[1] |= cbits[cbit_space+1] & ~0x08;              classbits[1] |= cbits[cbit_space+1] & ~0x08;
4322              for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];              for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];
4323              continue;              continue;
4324    
# Line 3554  for (;; ptr++) Line 4327  for (;; ptr++)
4327              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4328              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4329              continue;              continue;
4330    
4331                /* The rest apply in both UCP and non-UCP cases. */
4332    
4333              case ESC_h:              case ESC_h:
4334              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4335              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  
4336              continue;              continue;
4337    
4338              case ESC_H:              case ESC_H:
4339              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4340                {                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  
4341              continue;              continue;
4342    
4343              case ESC_v:              case ESC_v:
4344              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4345              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  
4346              continue;              continue;
4347    
4348              case ESC_V:              case ESC_V:
4349              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4350                {                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  
4351              continue;              continue;
4352    
4353  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3679  for (;; ptr++) Line 4358  for (;; ptr++)
4358                int pdata;                int pdata;
4359                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
4360                if (ptype < 0) goto FAILED;                if (ptype < 0) goto FAILED;
4361                class_utf8 = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_utf8data++ = ((-c == ESC_p) != negated)?  
4362                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4363                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4364                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4365                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4366                continue;                continue;
4367                }                }
4368  #endif  #endif
# Line 3698  for (;; ptr++) Line 4376  for (;; ptr++)
4376                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4377                goto FAILED;                goto FAILED;
4378                }                }
4379              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4380              c = *ptr;              /* Get the final character and fall through */              class_one_char -= 2;     /* Undo the speculative increase. */
4381                c = *ptr;                /* Get the final character and fall through */
4382              break;              break;
4383              }              }
4384            }            }
4385    
4386          /* 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).
4387          greater than 256 in UTF-8 mode. */          This may be greater than 256. */
4388    
4389            escape = 0;
4390    
4391          }   /* End of backslash handling */          }   /* End of backslash handling */
4392    
4393        /* 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
4394        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
4395        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4396        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4397    
4398        CHECK_RANGE:        CHECK_RANGE:
4399        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3720  for (;; ptr++) Line 4401  for (;; ptr++)
4401          inescq = FALSE;          inescq = FALSE;
4402          ptr += 2;          ptr += 2;
4403          }          }
   
4404        oldptr = ptr;        oldptr = ptr;
4405    
4406        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4407    
4408        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4409    
# Line 3746  for (;; ptr++) Line 4426  for (;; ptr++)
4426            inescq = TRUE;            inescq = TRUE;
4427            break;            break;
4428            }            }
4429    
4430            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4431            back the pointer and jump to handle the character that preceded it. */
4432    
4433          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4434            {            {
4435            ptr = oldptr;            ptr = oldptr;
4436            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4437            }            }
4438    
4439            /* Otherwise, we have a potential range; pick up the next character */
4440    
4441  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4442          if (utf8)          if (utf)
4443            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4444            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4445            }            }
# Line 3768  for (;; ptr++) Line 4453  for (;; ptr++)
4453    
4454          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4455            {            {
4456            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4457              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4458            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4459    
4460            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4461    
4462            if (d < 0)            if (descape > 0)
4463              {              {
4464              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4465                {                {
4466                ptr = oldptr;                ptr = oldptr;
4467                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
               }  
             }  
           }  
   
         /* Check that the two values are in the correct order. Optimize  
         one-character ranges */  
   
         if (d < c)  
           {  
           *errorcodeptr = ERR8;  
           goto FAILED;  
           }  
   
         if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */  
   
         /* Remember \r or \n */  
   
         if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;  
   
         /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless  
         matching, we have to use an XCLASS with extra data items. Caseless  
         matching for characters > 127 is available only if UCP support is  
         available. */  
   
 #ifdef SUPPORT_UTF8  
         if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))  
           {  
           class_utf8 = TRUE;  
   
           /* With UCP support, we can find the other case equivalents of  
           the relevant characters. There may be several ranges. Optimize how  
           they fit with the basic range. */  
   
 #ifdef SUPPORT_UCP  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             unsigned int occ, ocd;  
             unsigned int cc = c;  
             unsigned int origd = d;  
             while (get_othercase_range(&cc, origd, &occ, &ocd))  
               {  
               if (occ >= (unsigned int)c &&  
                   ocd <= (unsigned int)d)  
                 continue;                          /* Skip embedded ranges */  
   
               if (occ < (unsigned int)c  &&  
                   ocd >= (unsigned int)c - 1)      /* Extend the basic range */  
                 {                                  /* if there is overlap,   */  
                 c = occ;                           /* noting that if occ < c */  
                 continue;                          /* we can't have ocd > d  */  
                 }                                  /* because a subrange is  */  
               if (ocd > (unsigned int)d &&  
                   occ <= (unsigned int)d + 1)      /* always shorter than    */  
                 {                                  /* the basic range.       */  
                 d = ocd;  
                 continue;  
                 }  
   
               if (occ == ocd)  
                 {  
                 *class_utf8data++ = XCL_SINGLE;  
                 }  
               else  
                 {  
                 *class_utf8data++ = XCL_RANGE;  
                 class_utf8data += _pcre_ord2utf8(occ, class_utf8data);  
                 }  
               class_utf8data += _pcre_ord2utf8(ocd, class_utf8data);  
4468                }                }
4469              }              }
4470  #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;  
4471    
4472          /* We can save a bit of time by skipping this in the pre-compile. */          /* Check that the two values are in the correct order. Optimize
4473            one-character ranges. */
4474    
4475          if (lengthptr == NULL) for (; c <= d; c++)          if (d < c)
4476            {            {
4477            classbits[c/8] |= (1 << (c&7));            *errorcodeptr = ERR8;
4478            if ((options & PCRE_CASELESS) != 0)            goto FAILED;
             {  
             int uc = cd->fcc[c];           /* flip case */  
             classbits[uc/8] |= (1 << (uc&7));  
             }  
4479            }            }
4480            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4481    
4482          continue;   /* Go get the next char in the class */          /* We have found a character range, so single character optimizations
4483          }          cannot be done anymore. Any value greater than 1 indicates that there
4484            is more than one character. */
4485    
4486            class_one_char = 2;
4487    
4488        /* Handle a lone single character - we can get here for a normal          /* Remember an explicit \r or \n, and add the range to the class. */
       non-escape char, or after \ that introduces a single character or for an  
       apparent range that isn't. */  
4489    
4490        LONE_SINGLE_CHARACTER:          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4491    
4492            class_has_8bitchar +=
4493              add_to_class(classbits, &class_uchardata, options, cd, c, d);
4494    
4495            continue;   /* Go get the next char in the class */
4496            }
4497    
4498        /* Handle a character that cannot go in the bit map */        /* Handle a single character - we can get here for a normal non-escape
4499          char, or after \ that introduces a single character or for an apparent
4500          range that isn't. Only the value 1 matters for class_one_char, so don't
4501          increase it if it is already 2 or more ... just in case there's a class
4502          with a zillion characters in it. */
4503    
4504          CLASS_SINGLE_CHARACTER:
4505          if (class_one_char < 2) class_one_char++;
4506    
4507          /* If class_one_char is 1, we have the first single character in the
4508          class, and there have been no prior ranges, or XCLASS items generated by
4509          escapes. If this is the final character in the class, we can optimize by
4510          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4511          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4512          to be set. Otherwise, there can be no first char if this item is first,
4513          whatever repeat count may follow. In the case of reqchar, save the
4514          previous value for reinstating. */
4515    
4516  #ifdef SUPPORT_UTF8        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
       if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
4517          {          {
4518          class_utf8 = TRUE;          ptr++;
4519          *class_utf8data++ = XCL_SINGLE;          zeroreqchar = reqchar;
         class_utf8data += _pcre_ord2utf8(c, class_utf8data);  
4520    
4521  #ifdef SUPPORT_UCP          if (negate_class)
         if ((options & PCRE_CASELESS) != 0)  
4522            {            {
4523            unsigned int othercase;  #ifdef SUPPORT_UCP
4524            if ((othercase = UCD_OTHERCASE(c)) != c)              // FIXMEchpe pcreuint32?
4525              int d;
4526    #endif
4527              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4528              zerofirstchar = firstchar;
4529    
4530              /* For caseless UTF-8 mode when UCP support is available, check
4531              whether this character has more than one other case. If so, generate
4532              a special OP_NOTPROP item instead of OP_NOTI. */
4533    
4534    #ifdef SUPPORT_UCP
4535              if (utf && (options & PCRE_CASELESS) != 0 &&
4536                  (d = UCD_CASESET(c)) != 0)
4537                {
4538                *code++ = OP_NOTPROP;
4539                *code++ = PT_CLIST;
4540                *code++ = d;
4541                }
4542              else
4543    #endif
4544              /* Char has only one other case, or UCP not available */
4545    
4546              {              {
4547              *class_utf8data++ = XCL_SINGLE;              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4548              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4549                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4550                  code += PRIV(ord2utf)(c, code);
4551                else
4552    #endif
4553                  *code++ = c;
4554              }              }
4555    
4556              /* We are finished with this character class */
4557    
4558              goto END_CLASS;
4559            }            }
 #endif  /* SUPPORT_UCP */  
4560    
4561          }          /* For a single, positive character, get the value into mcbuffer, and
4562        else          then we can handle this with the normal one-character code. */
 #endif  /* SUPPORT_UTF8 */  
4563    
4564        /* Handle a single-byte character */  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4565          {          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4566          classbits[c/8] |= (1 << (c&7));            mclength = PRIV(ord2utf)(c, mcbuffer);
4567          if ((options & PCRE_CASELESS) != 0)          else
4568    #endif
4569            {            {
4570            c = cd->fcc[c];   /* flip case */            mcbuffer[0] = c;
4571            classbits[c/8] |= (1 << (c&7));            mclength = 1;
4572            }       &nbs