/[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 1062 by chpe, Tue Oct 16 15:54:05 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 n is returned as
754  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  negative n. When UTF-8 is enabled, a positive value greater than 255 may
755  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  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                     negative => a back reference
770                   on error, errorcodeptr is set                   on error, errorcodeptr is set
771  */  */
772    
773  static int  static int
774  check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
775    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
776  {  {
777  BOOL utf8 = (options & PCRE_UTF8) != 0;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
778  const uschar *ptr = *ptrptr + 1;  BOOL utf = (options & PCRE_UTF8) != 0;
779  int c, i;  const pcre_uchar *ptr = *ptrptr + 1;
780    pcre_uint32 c;
781    int escape = 0;
782    int i;
783    
784  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
785  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 793  in a table. A non-zero result is somethi
793  Otherwise further processing may be required. */  Otherwise further processing may be required. */
794    
795  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
796  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */  /* Not alphanumeric */
797  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if (c < CHAR_0 || c > CHAR_z) {}
798    else if ((i = escapes[c - CHAR_0]) != 0) { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
799    
800  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
801  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */  /* Not alphanumeric */
802  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
803    else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
804  #endif  #endif
805    
806  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
807    
808  else  else
809    {    {
810    const uschar *oldptr;    const pcre_uchar *oldptr;
811    BOOL braced, negated;    BOOL braced, negated, overflow;
812      int s;
813    
814    switch (c)    switch (c)
815      {      {
# Line 642  else Line 818  else
818    
819      case CHAR_l:      case CHAR_l:
820      case CHAR_L:      case CHAR_L:
821        *errorcodeptr = ERR37;
822        break;
823    
824      case CHAR_u:      case CHAR_u:
825        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
826          {
827          /* In JavaScript, \u must be followed by four hexadecimal numbers.
828          Otherwise it is a lowercase u letter. */
829          if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
830            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
831            && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
832            && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
833            {
834            c = 0;
835            for (i = 0; i < 4; ++i)
836              {
837              register pcre_uint32 cc = *(++ptr);
838    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
839              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
840              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
841    #else           /* EBCDIC coding */
842              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
843              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
844    #endif
845              }
846    
847    #if defined COMPILE_PCRE8
848            if (c > (utf ? 0x10ffff : 0xff))
849    #elif defined COMPILE_PCRE16
850            if (c > (utf ? 0x10ffff : 0xffff))
851    #elif defined COMPILE_PCRE32
852            if (utf && c > 0x10ffff)
853    #endif
854              {
855              *errorcodeptr = ERR76;
856              }
857            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
858            }
859          }
860        else
861          *errorcodeptr = ERR37;
862        break;
863    
864      case CHAR_U:      case CHAR_U:
865      *errorcodeptr = ERR37;      /* In JavaScript, \U is an uppercase U letter. */
866        if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37;
867      break;      break;
868    
869      /* \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
870        class, \g must be followed by one of a number of specific things:
871    
872      (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
873      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 881  else
881      (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
882      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
883      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
884      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
885    
886      case CHAR_g:      case CHAR_g:
887        if (isclass) break;
888      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
889        {        {
890        c = -ESC_g;        escape = ESC_g;
891        break;        break;
892        }        }
893    
# Line 674  else Line 895  else
895    
896      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
897        {        {
898        const uschar *p;        const pcre_uchar *p;
899        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
900          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
901        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
902          {          {
903          c = -ESC_k;          escape = ESC_k;
904          break;          break;
905          }          }
906        braced = TRUE;        braced = TRUE;
# Line 694  else Line 915  else
915        }        }
916      else negated = FALSE;      else negated = FALSE;
917    
918      c = 0;      /* The integer range is limited by the machine's int representation. */
919      while ((digitab[ptr[1]] & ctype_digit) != 0)      s = 0;
920        c = c * 10 + *(++ptr) - CHAR_0;      overflow = FALSE;
921        while (IS_DIGIT(ptr[1]))
     if (c < 0)   /* Integer overflow */  
922        {        {
923          if (s > INT_MAX / 10 - 1) /* Integer overflow */
924            {
925            overflow = TRUE;
926            break;
927            }
928          s = s * 10 + (int)(*(++ptr) - CHAR_0);
929          }
930        if (overflow) /* Integer overflow */
931          {
932          while (IS_DIGIT(ptr[1]))
933            ptr++;
934        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
935        break;        break;
936        }        }
# Line 710  else Line 941  else
941        break;        break;
942        }        }
943    
944      if (c == 0)      if (s == 0)
945        {        {
946        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
947        break;        break;
# Line 718  else Line 949  else
949    
950      if (negated)      if (negated)
951        {        {
952        if (c > bracount)        if (s > bracount)
953          {          {
954          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
955          break;          break;
956          }          }
957        c = bracount - (c - 1);        s = bracount - (s - 1);
958        }        }
959    
960      c = -(ESC_REF + c);      escape = -s;
961      break;      break;
962    
963      /* 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 978  else
978      if (!isclass)      if (!isclass)
979        {        {
980        oldptr = ptr;        oldptr = ptr;
981        c -= CHAR_0;        /* The integer range is limited by the machine's int representation. */
982        while ((digitab[ptr[1]] & ctype_digit) != 0)        s = (int)(c -CHAR_0);
983          c = c * 10 + *(++ptr) - CHAR_0;        overflow = FALSE;
984        if (c < 0)    /* Integer overflow */        while (IS_DIGIT(ptr[1]))
985          {          {
986            if (s > INT_MAX / 10 - 1) /* Integer overflow */
987              {
988              overflow = TRUE;
989              break;
990              }
991            s = s * 10 + (int)(*(++ptr) - CHAR_0);
992            }
993          if (overflow) /* Integer overflow */
994            {
995            while (IS_DIGIT(ptr[1]))
996              ptr++;
997          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
998          break;          break;
999          }          }
1000        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1001          {          {
1002          c = -(ESC_REF + c);          escape = -s;
1003          break;          break;
1004          }          }
1005        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 777  else Line 1019  else
1019      /* \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
1020      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
1021      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
1022      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,
1023      than 3 octal digits. */      but no more than 3 octal digits. */
1024    
1025      case CHAR_0:      case CHAR_0:
1026      c -= CHAR_0;      c -= CHAR_0;
1027      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1028          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1029      if (!utf8 && c > 255) *errorcodeptr = ERR51;  #ifdef COMPILE_PCRE8
1030        if (!utf && c > 0xff) *errorcodeptr = ERR51;
1031    #endif
1032      break;      break;
1033    
1034      /* \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
1035      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.
1036      treated as a data character. */      If not, { is treated as a data character. */
1037    
1038      case CHAR_x:      case CHAR_x:
1039        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1040          {
1041          /* In JavaScript, \x must be followed by two hexadecimal numbers.
1042          Otherwise it is a lowercase x letter. */
1043          if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1044            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1045            {
1046            c = 0;
1047            for (i = 0; i < 2; ++i)
1048              {
1049              register pcre_uint32 cc = *(++ptr);
1050    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1051              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1052              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
1053    #else           /* EBCDIC coding */
1054              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
1055              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1056    #endif
1057              }
1058            }
1059          break;
1060          }
1061    
1062      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1063        {        {
1064        const uschar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
       int count = 0;  
1065    
1066        c = 0;        c = 0;
1067        while ((digitab[*pt] & ctype_xdigit) != 0)        overflow = FALSE;
1068          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1069          {          {
1070          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1071          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1072          count++;  
1073    #ifdef COMPILE_PCRE32
1074            if (c >= 0x10000000l) { overflow = TRUE; break; }
1075    #endif
1076    
1077  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1078          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
# Line 811  else Line 1081  else
1081          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 */
1082          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1083  #endif  #endif
1084    
1085    #if defined COMPILE_PCRE8
1086            if (c > (utf ? 0x10ffff : 0xff)) { overflow = TRUE; break; }
1087    #elif defined COMPILE_PCRE16
1088            if (c > (utf ? 0x10ffff : 0xffff)) { overflow = TRUE; break; }
1089    #elif defined COMPILE_PCRE32
1090            if (utf && c > 0x10ffff) { overflow = TRUE; break; }
1091    #endif
1092            }
1093    
1094          if (overflow)
1095            {
1096            while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1097            *errorcodeptr = ERR34;
1098          }          }
1099    
1100        if (*pt == CHAR_RIGHT_CURLY_BRACKET)        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
1101          {          {
1102          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1103          ptr = pt;          ptr = pt;
1104          break;          break;
1105          }          }
# Line 827  else Line 1111  else
1111      /* Read just a single-byte hex-defined char */      /* Read just a single-byte hex-defined char */
1112    
1113      c = 0;      c = 0;
1114      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1115        {        {
1116        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1117        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1118  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1119        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 857  else Line 1141  else
1141      if (c > 127)  /* Excludes all non-ASCII in either mode */      if (c > 127)  /* Excludes all non-ASCII in either mode */
1142        {        {
1143        *errorcodeptr = ERR68;        *errorcodeptr = ERR68;
1144        break;        break;
1145        }        }
1146      if (c >= CHAR_a && c <= CHAR_z) c -= 32;      if (c >= CHAR_a && c <= CHAR_z) c -= 32;
1147      c ^= 0x40;      c ^= 0x40;
1148  #else             /* EBCDIC coding */  #else             /* EBCDIC coding */
# Line 885  else Line 1169  else
1169    }    }
1170    
1171  /* 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
1172  newline". PCRE does not support \N{name}. */  newline". PCRE does not support \N{name}. However, it does support
1173    quantification such as \N{2,3}. */
1174    
1175  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET)  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1176         !is_counted_repeat(ptr+2))
1177    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1178    
1179  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1180    
1181  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1182    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1183    
1184  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1185    
1186  *ptrptr = ptr;  *ptrptr = ptr;
1187  return c;  *chptr = c;
1188    return escape;
1189  }  }
1190    
   
   
1191  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1192  /*************************************************  /*************************************************
1193  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 923  Returns:         type value from ucp_typ Line 1208  Returns:         type value from ucp_typ
1208  */  */
1209    
1210  static int  static int
1211  get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)
1212  {  {
1213  int c, i, bot, top;  int c, i, bot, top;
1214  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1215  char name[32];  pcre_uchar name[32];
1216    
1217  c = *(++ptr);  c = *(++ptr);
1218  if (c == 0) goto ERROR_RETURN;  if (c == 0) goto ERROR_RETURN;
# Line 944  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1229  if (c == CHAR_LEFT_CURLY_BRACKET)
1229      *negptr = TRUE;      *negptr = TRUE;
1230      ptr++;      ptr++;
1231      }      }
1232    for (i = 0; i < (int)sizeof(name) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1233      {      {
1234      c = *(++ptr);      c = *(++ptr);
1235      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 968  else Line 1253  else
1253  /* Search for a recognized property name using binary chop */  /* Search for a recognized property name using binary chop */
1254    
1255  bot = 0;  bot = 0;
1256  top = _pcre_utt_size;  top = PRIV(utt_size);
1257    
1258  while (bot < top)  while (bot < top)
1259    {    {
1260    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1261    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);    c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1262    if (c == 0)    if (c == 0)
1263      {      {
1264      *dptr = _pcre_utt[i].value;      *dptr = PRIV(utt)[i].value;
1265      return _pcre_utt[i].type;      return PRIV(utt)[i].type;
1266      }      }
1267    if (c > 0) bot = i + 1; else top = i;    if (c > 0) bot = i + 1; else top = i;
1268    }    }
# Line 997  return -1; Line 1282  return -1;
1282    
1283    
1284  /*************************************************  /*************************************************
 *            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);  
 }  
   
   
   
 /*************************************************  
1285  *         Read repeat counts                     *  *         Read repeat counts                     *
1286  *************************************************/  *************************************************/
1287    
# Line 1048  Returns:         pointer to '}' on succe Line 1300  Returns:         pointer to '}' on succe
1300                   current ptr on error, with errorcodeptr set non-zero                   current ptr on error, with errorcodeptr set non-zero
1301  */  */
1302    
1303  static const uschar *  static const pcre_uchar *
1304  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)
1305  {  {
1306  int min = 0;  int min = 0;
1307  int max = -1;  int max = -1;
# Line 1057  int max = -1; Line 1309  int max = -1;
1309  /* 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
1310  an integer overflow. */  an integer overflow. */
1311    
1312  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;
1313  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1314    {    {
1315    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1072  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1324  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1324    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1325      {      {
1326      max = 0;      max = 0;
1327      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;
1328      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1329        {        {
1330        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1105  top-level call starts at the beginning o Line 1357  top-level call starts at the beginning o
1357  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
1358  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
1359  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
1360  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
1361  encountered, the name will be terminated by '>' because that is checked in the  track of subpatterns that reset the capturing group numbers - the (?| feature.
1362  first pass. Recursion is used to keep track of subpatterns that reset the  
1363  capturing group numbers - the (?| feature.  This function was originally called only from the second pass, in which we know
1364    that if (?< or (?' or (?P< is encountered, the name will be correctly
1365    terminated because that is checked in the first pass. There is now one call to
1366    this function in the first pass, to check for a recursive back reference by
1367    name (so that we can make the whole group atomic). In this case, we need check
1368    only up to the current position in the pattern, and that is still OK because
1369    and previous occurrences will have been checked. To make this work, the test
1370    for "end of pattern" is a check against cd->end_pattern in the main loop,
1371    instead of looking for a binary zero. This means that the special first-pass
1372    call can adjust cd->end_pattern temporarily. (Checks for binary zero while
1373    processing items within the loop are OK, because afterwards the main loop will
1374    terminate.)
1375    
1376  Arguments:  Arguments:
1377    ptrptr       address of the current character pointer (updated)    ptrptr       address of the current character pointer (updated)
# Line 1116  Arguments: Line 1379  Arguments:
1379    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1380    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1381    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1382    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1383    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1384    
1385  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1386  */  */
1387    
1388  static int  static int
1389  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,
1390    BOOL xmode, BOOL utf8, int *count)    BOOL xmode, BOOL utf, int *count)
1391  {  {
1392  uschar *ptr = *ptrptr;  pcre_uchar *ptr = *ptrptr;
1393  int start_count = *count;  int start_count = *count;
1394  int hwm_count = start_count;  int hwm_count = start_count;
1395  BOOL dup_parens = FALSE;  BOOL dup_parens = FALSE;
# Line 1193  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1456  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1456          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1457        {        {
1458        int term;        int term;
1459        const uschar *thisname;        const pcre_uchar *thisname;
1460        *count += 1;        *count += 1;
1461        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1462        term = *ptr++;        term = *ptr++;
# Line 1201  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1464  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1464        thisname = ptr;        thisname = ptr;
1465        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1466        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == ptr - thisname &&
1467            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, lorn) == 0)
1468          return *count;          return *count;
1469        term++;        term++;
1470        }        }
# Line 1209  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1472  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1472    }    }
1473    
1474  /* Past any initial parenthesis handling, scan for parentheses or vertical  /* Past any initial parenthesis handling, scan for parentheses or vertical
1475  bars. */  bars. Stop if we get to cd->end_pattern. Note that this is important for the
1476    first-pass call when this value is temporarily adjusted to stop at the current
1477    position. So DO NOT change this to a test for binary zero. */
1478    
1479  for (; *ptr != 0; ptr++)  for (; ptr < cd->end_pattern; ptr++)
1480    {    {
1481    /* Skip over backslashed characters and also entire \Q...\E */    /* Skip over backslashed characters and also entire \Q...\E */
1482    
# Line 1242  for (; *ptr != 0; ptr++) Line 1507  for (; *ptr != 0; ptr++)
1507          {          {
1508          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1509            ptr+= 2;            ptr+= 2;
1510          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1511                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1512            ptr += 4;            ptr += 4;
1513          else          else
# Line 1285  for (; *ptr != 0; ptr++) Line 1550  for (; *ptr != 0; ptr++)
1550    
1551    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1552      {      {
1553      ptr++;      ptr++;
1554      while (*ptr != 0)      while (*ptr != 0)
1555        {        {
1556        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1557        ptr++;        ptr++;
1558  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1559        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1560  #endif  #endif
1561        }        }
1562      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
# Line 1302  for (; *ptr != 0; ptr++) Line 1567  for (; *ptr != 0; ptr++)
1567    
1568    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1569      {      {
1570      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1571      if (rc > 0) return rc;      if (rc > 0) return rc;
1572      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1573      }      }
# Line 1348  Arguments: Line 1613  Arguments:
1613    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1614    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1615    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1616    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1617    
1618  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1619  */  */
1620    
1621  static int  static int
1622  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,
1623    BOOL utf8)    BOOL utf)
1624  {  {
1625  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1626  int count = 0;  int count = 0;
1627  int rc;  int rc;
1628    
# Line 1368  matching closing parens. That is why we Line 1633  matching closing parens. That is why we
1633    
1634  for (;;)  for (;;)
1635    {    {
1636    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1637    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == 0) break;
1638    }    }
1639    
# Line 1384  return rc; Line 1649  return rc;
1649    
1650  /* This is called by several functions that scan a compiled expression looking  /* This is called by several functions that scan a compiled expression looking
1651  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
1652  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
1653  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
1654  assertions, and also the \b assertion; for others it does not.  does not.
1655    
1656  Arguments:  Arguments:
1657    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  
1658    skipassert   TRUE if certain assertions are to be skipped    skipassert   TRUE if certain assertions are to be skipped
1659    
1660  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1661  */  */
1662    
1663  static const uschar*  static const pcre_uchar*
1664  first_significant_code(const uschar *code, int *options, int optbit,  first_significant_code(const pcre_uchar *code, BOOL skipassert)
   BOOL skipassert)  
1665  {  {
1666  for (;;)  for (;;)
1667    {    {
1668    switch ((int)*code)    switch ((int)*code)
1669      {      {
     case OP_OPT:  
     if (optbit > 0 && ((int)code[1] & optbit) != (*options & optbit))  
       *options = (int)code[1];  
     code += 2;  
     break;  
   
1670      case OP_ASSERT_NOT:      case OP_ASSERT_NOT:
1671      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1672      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1673      if (!skipassert) return code;      if (!skipassert) return code;
1674      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1675      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1676      break;      break;
1677    
1678      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1431  for (;;) Line 1686  for (;;)
1686      case OP_RREF:      case OP_RREF:
1687      case OP_NRREF:      case OP_NRREF:
1688      case OP_DEF:      case OP_DEF:
1689      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1690      break;      break;
1691    
1692      default:      default:
# Line 1461  and doing the check at the end; a flag s Line 1716  and doing the check at the end; a flag s
1716    
1717  Arguments:  Arguments:
1718    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1719    options  the compiling options    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1720    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1721    cd       the "compile data" structure    cd       the "compile data" structure
1722    
1723  Returns:   the fixed length,  Returns:   the fixed length,
1724               or -1 if there is no fixed length,               or -1 if there is no fixed length,
1725               or -2 if \C was encountered               or -2 if \C was encountered (in UTF-8 mode only)
1726               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
1727                 or -4 if an unknown opcode was encountered (internal error)
1728  */  */
1729    
1730  static int  static int
1731  find_fixedlength(uschar *code, int options, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1732  {  {
1733  int length = -1;  int length = -1;
1734    
1735  register int branchlength = 0;  register int branchlength = 0;
1736  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1737    
1738  /* 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
1739  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 1741  branch, check the length against that of
1741  for (;;)  for (;;)
1742    {    {
1743    int d;    int d;
1744    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1745    register int op = *cc;    register int op = *cc;
1746    
1747    switch (op)    switch (op)
1748      {      {
1749        /* We only need to continue for OP_CBRA (normal capturing bracket) and
1750        OP_BRA (normal non-capturing bracket) because the other variants of these
1751        opcodes are all concerned with unlimited repeated groups, which of course
1752        are not of fixed length. */
1753    
1754      case OP_CBRA:      case OP_CBRA:
1755      case OP_BRA:      case OP_BRA:
1756      case OP_ONCE:      case OP_ONCE:
1757        case OP_ONCE_NC:
1758      case OP_COND:      case OP_COND:
1759      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);
1760      if (d < 0) return d;      if (d < 0) return d;
1761      branchlength += d;      branchlength += d;
1762      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1763      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
1764      break;      break;
1765    
1766      /* 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.
1767      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
1768      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
1769        the same code. Note that we must not include the OP_KETRxxx opcodes here,
1770        because they all imply an unlimited repeat. */
1771    
1772      case OP_ALT:      case OP_ALT:
1773      case OP_KET:      case OP_KET:
     case OP_KETRMAX:  
     case OP_KETRMIN:  
1774      case OP_END:      case OP_END:
1775        case OP_ACCEPT:
1776        case OP_ASSERT_ACCEPT:
1777      if (length < 0) length = branchlength;      if (length < 0) length = branchlength;
1778        else if (length != branchlength) return -1;        else if (length != branchlength) return -1;
1779      if (*cc != OP_ALT) return length;      if (*cc != OP_ALT) return length;
# Line 1522  for (;;) Line 1787  for (;;)
1787    
1788      case OP_RECURSE:      case OP_RECURSE:
1789      if (!atend) return -3;      if (!atend) return -3;
1790      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1791      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1792      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1793      d = find_fixedlength(cs + 2, options, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1794      if (d < 0) return d;      if (d < 0) return d;
1795      branchlength += d;      branchlength += d;
1796      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1538  for (;;) Line 1803  for (;;)
1803      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1804      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1805      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1806      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1807        break;
1808    
1809      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1810    
1811      case OP_REVERSE:      case OP_MARK:
1812        case OP_PRUNE_ARG:
1813        case OP_SKIP_ARG:
1814        case OP_THEN_ARG:
1815        cc += cc[1] + PRIV(OP_lengths)[*cc];
1816        break;
1817    
1818        case OP_CALLOUT:
1819        case OP_CIRC:
1820        case OP_CIRCM:
1821        case OP_CLOSE:
1822        case OP_COMMIT:
1823      case OP_CREF:      case OP_CREF:
     case OP_NCREF:  
     case OP_RREF:  
     case OP_NRREF:  
1824      case OP_DEF:      case OP_DEF:
1825      case OP_OPT:      case OP_DOLL:
1826      case OP_CALLOUT:      case OP_DOLLM:
     case OP_SOD:  
     case OP_SOM:  
     case OP_SET_SOM:  
1827      case OP_EOD:      case OP_EOD:
1828      case OP_EODN:      case OP_EODN:
1829      case OP_CIRC:      case OP_FAIL:
1830      case OP_DOLL:      case OP_NCREF:
1831        case OP_NRREF:
1832      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1833        case OP_PRUNE:
1834        case OP_REVERSE:
1835        case OP_RREF:
1836        case OP_SET_SOM:
1837        case OP_SKIP:
1838        case OP_SOD:
1839        case OP_SOM:
1840        case OP_THEN:
1841      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1842      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1843      break;      break;
1844    
1845      /* Handle literal characters */      /* Handle literal characters */
1846    
1847      case OP_CHAR:      case OP_CHAR:
1848      case OP_CHARNC:      case OP_CHARI:
1849      case OP_NOT:      case OP_NOT:
1850        case OP_NOTI:
1851      branchlength++;      branchlength++;
1852      cc += 2;      cc += 2;
1853  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1854      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];  
1855  #endif  #endif
1856      break;      break;
1857    
# Line 1579  for (;;) Line 1859  for (;;)
1859      need to skip over a multibyte character in UTF8 mode.  */      need to skip over a multibyte character in UTF8 mode.  */
1860    
1861      case OP_EXACT:      case OP_EXACT:
1862        case OP_EXACTI:
1863        case OP_NOTEXACT:
1864        case OP_NOTEXACTI:
1865      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1866      cc += 4;      cc += 2 + IMM2_SIZE;
1867  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1868      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];  
1869  #endif  #endif
1870      break;      break;
1871    
1872      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1873      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1874      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1875      cc += 4;        cc += 2;
1876        cc += 1 + IMM2_SIZE + 1;
1877      break;      break;
1878    
1879      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1600  for (;;) Line 1883  for (;;)
1883      cc += 2;      cc += 2;
1884      /* Fall through */      /* Fall through */
1885    
1886        case OP_HSPACE:
1887        case OP_VSPACE:
1888        case OP_NOT_HSPACE:
1889        case OP_NOT_VSPACE:
1890      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
1891      case OP_DIGIT:      case OP_DIGIT:
1892      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
# Line 1612  for (;;) Line 1899  for (;;)
1899      cc++;      cc++;
1900      break;      break;
1901    
1902      /* The single-byte matcher isn't allowed */      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;
1903        otherwise \C is coded as OP_ALLANY. */
1904    
1905      case OP_ANYBYTE:      case OP_ANYBYTE:
1906      return -2;      return -2;
1907    
1908      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1909    
1910  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1911      case OP_XCLASS:      case OP_XCLASS:
1912      cc += GET(cc, 1) - 33;      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
1913      /* Fall through */      /* Fall through */
1914  #endif  #endif
1915    
1916      case OP_CLASS:      case OP_CLASS:
1917      case OP_NCLASS:      case OP_NCLASS:
1918      cc += 33;      cc += PRIV(OP_lengths)[OP_CLASS];
1919    
1920      switch (*cc)      switch (*cc)
1921        {        {
1922          case OP_CRPLUS:
1923          case OP_CRMINPLUS:
1924        case OP_CRSTAR:        case OP_CRSTAR:
1925        case OP_CRMINSTAR:        case OP_CRMINSTAR:
1926        case OP_CRQUERY:        case OP_CRQUERY:
# Line 1639  for (;;) Line 1929  for (;;)
1929    
1930        case OP_CRRANGE:        case OP_CRRANGE:
1931        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1932        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1933        branchlength += GET2(cc,1);        branchlength += GET2(cc,1);
1934        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1935        break;        break;
1936    
1937        default:        default:
# Line 1651  for (;;) Line 1941  for (;;)
1941    
1942      /* Anything else is variable length */      /* Anything else is variable length */
1943    
1944      default:      case OP_ANYNL:
1945        case OP_BRAMINZERO:
1946        case OP_BRAPOS:
1947        case OP_BRAPOSZERO:
1948        case OP_BRAZERO:
1949        case OP_CBRAPOS:
1950        case OP_EXTUNI:
1951        case OP_KETRMAX:
1952        case OP_KETRMIN:
1953        case OP_KETRPOS:
1954        case OP_MINPLUS:
1955        case OP_MINPLUSI:
1956        case OP_MINQUERY:
1957        case OP_MINQUERYI:
1958        case OP_MINSTAR:
1959        case OP_MINSTARI:
1960        case OP_MINUPTO:
1961        case OP_MINUPTOI:
1962        case OP_NOTMINPLUS:
1963        case OP_NOTMINPLUSI:
1964        case OP_NOTMINQUERY:
1965        case OP_NOTMINQUERYI:
1966        case OP_NOTMINSTAR:
1967        case OP_NOTMINSTARI:
1968        case OP_NOTMINUPTO:
1969        case OP_NOTMINUPTOI:
1970        case OP_NOTPLUS:
1971        case OP_NOTPLUSI:
1972        case OP_NOTPOSPLUS:
1973        case OP_NOTPOSPLUSI:
1974        case OP_NOTPOSQUERY:
1975        case OP_NOTPOSQUERYI:
1976        case OP_NOTPOSSTAR:
1977        case OP_NOTPOSSTARI:
1978        case OP_NOTPOSUPTO:
1979        case OP_NOTPOSUPTOI:
1980        case OP_NOTQUERY:
1981        case OP_NOTQUERYI:
1982        case OP_NOTSTAR:
1983        case OP_NOTSTARI:
1984        case OP_NOTUPTO:
1985        case OP_NOTUPTOI:
1986        case OP_PLUS:
1987        case OP_PLUSI:
1988        case OP_POSPLUS:
1989        case OP_POSPLUSI:
1990        case OP_POSQUERY:
1991        case OP_POSQUERYI:
1992        case OP_POSSTAR:
1993        case OP_POSSTARI:
1994        case OP_POSUPTO:
1995        case OP_POSUPTOI:
1996        case OP_QUERY:
1997        case OP_QUERYI:
1998        case OP_REF:
1999        case OP_REFI:
2000        case OP_SBRA:
2001        case OP_SBRAPOS:
2002        case OP_SCBRA:
2003        case OP_SCBRAPOS:
2004        case OP_SCOND:
2005        case OP_SKIPZERO:
2006        case OP_STAR:
2007        case OP_STARI:
2008        case OP_TYPEMINPLUS:
2009        case OP_TYPEMINQUERY:
2010        case OP_TYPEMINSTAR:
2011        case OP_TYPEMINUPTO:
2012        case OP_TYPEPLUS:
2013        case OP_TYPEPOSPLUS:
2014        case OP_TYPEPOSQUERY:
2015        case OP_TYPEPOSSTAR:
2016        case OP_TYPEPOSUPTO:
2017        case OP_TYPEQUERY:
2018        case OP_TYPESTAR:
2019        case OP_TYPEUPTO:
2020        case OP_UPTO:
2021        case OP_UPTOI:
2022      return -1;      return -1;
2023    
2024        /* Catch unrecognized opcodes so that when new ones are added they
2025        are not forgotten, as has happened in the past. */
2026    
2027        default:
2028        return -4;
2029      }      }
2030    }    }
2031  /* Control never gets here */  /* Control never gets here */
# Line 1673  length. Line 2046  length.
2046    
2047  Arguments:  Arguments:
2048    code        points to start of expression    code        points to start of expression
2049    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2050    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2051    
2052  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
2053  */  */
2054    
2055  const uschar *  const pcre_uchar *
2056  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2057  {  {
2058  for (;;)  for (;;)
2059    {    {
2060    register int c = *code;    register int c = *code;
2061    
2062    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2063    
2064    /* 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 2071  for (;;)
2071    
2072    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2073      {      {
2074      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2075      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2076      }      }
2077    
2078    /* Handle capturing bracket */    /* Handle capturing bracket */
2079    
2080    else if (c == OP_CBRA)    else if (c == OP_CBRA || c == OP_SCBRA ||
2081               c == OP_CBRAPOS || c == OP_SCBRAPOS)
2082      {      {
2083      int n = GET2(code, 1+LINK_SIZE);      int n = GET2(code, 1+LINK_SIZE);
2084      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2085      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2086      }      }
2087    
2088    /* 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 2110  for (;;)
2110        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2111        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2112        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2113        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2114            code += 2;
2115        break;        break;
2116    
2117        case OP_MARK:        case OP_MARK:
# Line 1745  for (;;) Line 2121  for (;;)
2121        break;        break;
2122    
2123        case OP_THEN_ARG:        case OP_THEN_ARG:
2124        code += code[1+LINK_SIZE];        code += code[1];
2125        break;        break;
2126        }        }
2127    
2128      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2129    
2130      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2131    
2132    /* 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
2133    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
2134    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2135    
2136  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2137      if (utf8) switch(c)      if (utf) switch(c)
2138        {        {
2139        case OP_CHAR:        case OP_CHAR:
2140        case OP_CHARNC:        case OP_CHARI:
2141        case OP_EXACT:        case OP_EXACT:
2142          case OP_EXACTI:
2143        case OP_UPTO:        case OP_UPTO:
2144          case OP_UPTOI:
2145        case OP_MINUPTO:        case OP_MINUPTO:
2146          case OP_MINUPTOI:
2147        case OP_POSUPTO:        case OP_POSUPTO:
2148          case OP_POSUPTOI:
2149        case OP_STAR:        case OP_STAR:
2150          case OP_STARI:
2151        case OP_MINSTAR:        case OP_MINSTAR:
2152          case OP_MINSTARI:
2153        case OP_POSSTAR:        case OP_POSSTAR:
2154          case OP_POSSTARI:
2155        case OP_PLUS:        case OP_PLUS:
2156          case OP_PLUSI:
2157        case OP_MINPLUS:        case OP_MINPLUS:
2158          case OP_MINPLUSI:
2159        case OP_POSPLUS:        case OP_POSPLUS:
2160          case OP_POSPLUSI:
2161        case OP_QUERY:        case OP_QUERY:
2162          case OP_QUERYI:
2163        case OP_MINQUERY:        case OP_MINQUERY:
2164          case OP_MINQUERYI:
2165        case OP_POSQUERY:        case OP_POSQUERY:
2166        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_POSQUERYI:
2167          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2168        break;        break;
2169        }        }
2170  #else  #else
2171      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2172  #endif  #endif
2173      }      }
2174    }    }
# Line 1796  instance of OP_RECURSE. Line 2185  instance of OP_RECURSE.
2185    
2186  Arguments:  Arguments:
2187    code        points to start of expression    code        points to start of expression
2188    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2189    
2190  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
2191  */  */
2192    
2193  static const uschar *  static const pcre_uchar *
2194  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2195  {  {
2196  for (;;)  for (;;)
2197    {    {
# Line 1841  for (;;) Line 2230  for (;;)
2230        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2231        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2232        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2233        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2234            code += 2;
2235        break;        break;
2236    
2237        case OP_MARK:        case OP_MARK:
# Line 1851  for (;;) Line 2241  for (;;)
2241        break;        break;
2242    
2243        case OP_THEN_ARG:        case OP_THEN_ARG:
2244        code += code[1+LINK_SIZE];        code += code[1];
2245        break;        break;
2246        }        }
2247    
2248      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2249    
2250      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2251    
2252      /* 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
2253      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
2254      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2255    
2256  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2257      if (utf8) switch(c)      if (utf) switch(c)
2258        {        {
2259        case OP_CHAR:        case OP_CHAR:
2260        case OP_CHARNC:        case OP_CHARI:
2261          case OP_NOT:
2262          case OP_NOTI:
2263        case OP_EXACT:        case OP_EXACT:
2264          case OP_EXACTI:
2265          case OP_NOTEXACT:
2266          case OP_NOTEXACTI:
2267        case OP_UPTO:        case OP_UPTO:
2268          case OP_UPTOI:
2269          case OP_NOTUPTO:
2270          case OP_NOTUPTOI:
2271        case OP_MINUPTO:        case OP_MINUPTO:
2272          case OP_MINUPTOI:
2273          case OP_NOTMINUPTO:
2274          case OP_NOTMINUPTOI:
2275        case OP_POSUPTO:        case OP_POSUPTO:
2276          case OP_POSUPTOI:
2277          case OP_NOTPOSUPTO:
2278          case OP_NOTPOSUPTOI:
2279        case OP_STAR:        case OP_STAR:
2280          case OP_STARI:
2281          case OP_NOTSTAR:
2282          case OP_NOTSTARI:
2283        case OP_MINSTAR:        case OP_MINSTAR:
2284          case OP_MINSTARI:
2285          case OP_NOTMINSTAR:
2286          case OP_NOTMINSTARI:
2287        case OP_POSSTAR:        case OP_POSSTAR:
2288          case OP_POSSTARI:
2289          case OP_NOTPOSSTAR:
2290          case OP_NOTPOSSTARI:
2291        case OP_PLUS:        case OP_PLUS:
2292          case OP_PLUSI:
2293          case OP_NOTPLUS:
2294          case OP_NOTPLUSI:
2295        case OP_MINPLUS:        case OP_MINPLUS:
2296          case OP_MINPLUSI:
2297          case OP_NOTMINPLUS:
2298          case OP_NOTMINPLUSI:
2299        case OP_POSPLUS:        case OP_POSPLUS:
2300          case OP_POSPLUSI:
2301          case OP_NOTPOSPLUS:
2302          case OP_NOTPOSPLUSI:
2303        case OP_QUERY:        case OP_QUERY:
2304          case OP_QUERYI:
2305          case OP_NOTQUERY:
2306          case OP_NOTQUERYI:
2307        case OP_MINQUERY:        case OP_MINQUERY:
2308          case OP_MINQUERYI:
2309          case OP_NOTMINQUERY:
2310          case OP_NOTMINQUERYI:
2311        case OP_POSQUERY:        case OP_POSQUERY:
2312        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_POSQUERYI:
2313          case OP_NOTPOSQUERY:
2314          case OP_NOTPOSQUERYI:
2315          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2316        break;        break;
2317        }        }
2318  #else  #else
2319      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2320  #endif  #endif
2321      }      }
2322    }    }
# Line 1908  bracket whose current branch will alread Line 2339  bracket whose current branch will alread
2339  Arguments:  Arguments:
2340    code        points to start of search    code        points to start of search
2341    endcode     points to where to stop    endcode     points to where to stop
2342    utf8        TRUE if in UTF8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2343    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2344    
2345  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2346  */  */
2347    
2348  static BOOL  static BOOL
2349  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2350    compile_data *cd)    BOOL utf, compile_data *cd)
2351  {  {
2352  register int c;  register int c;
2353  for (code = first_significant_code(code + _pcre_OP_lengths[*code], NULL, 0, TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2354       code < endcode;       code < endcode;
2355       code = first_significant_code(code + _pcre_OP_lengths[c], NULL, 0, TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2356    {    {
2357    const uschar *ccode;    const pcre_uchar *ccode;
2358    
2359    c = *code;    c = *code;
2360    
# Line 1937  for (code = first_significant_code(code Line 2368  for (code = first_significant_code(code
2368      continue;      continue;
2369      }      }
2370    
   /* 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;  
     }  
   
2371    /* For a recursion/subroutine call, if its end has been reached, which    /* For a recursion/subroutine call, if its end has been reached, which
2372    implies a subroutine call, we can scan it. */    implies a backward reference subroutine call, we can scan it. If it's a
2373      forward reference subroutine call, we can't. To detect forward reference
2374      we have to scan up the list that is kept in the workspace. This function is
2375      called only when doing the real compile, not during the pre-compile that
2376      measures the size of the compiled pattern. */
2377    
2378    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2379      {      {
2380      BOOL empty_branch = FALSE;      const pcre_uchar *scode;
2381      const uschar *scode = cd->start_code + GET(code, 1);      BOOL empty_branch;
2382    
2383        /* Test for forward reference */
2384    
2385        for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2386          if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;
2387    
2388        /* Not a forward reference, test for completed backward reference */
2389    
2390        empty_branch = FALSE;
2391        scode = cd->start_code + GET(code, 1);
2392      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */      if (GET(scode, 1) == 0) return TRUE;    /* Unclosed */
2393    
2394        /* Completed backwards reference */
2395    
2396      do      do
2397        {        {
2398        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd))
2399          {          {
2400          empty_branch = TRUE;          empty_branch = TRUE;
2401          break;          break;
# Line 1965  for (code = first_significant_code(code Line 2403  for (code = first_significant_code(code
2403        scode += GET(scode, 1);        scode += GET(scode, 1);
2404        }        }
2405      while (*scode == OP_ALT);      while (*scode == OP_ALT);
2406    
2407      if (!empty_branch) return FALSE;  /* All branches are non-empty */      if (!empty_branch) return FALSE;  /* All branches are non-empty */
2408      continue;      continue;
2409      }      }
2410    
2411      /* Groups with zero repeats can of course be empty; skip them. */
2412    
2413      if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2414          c == OP_BRAPOSZERO)
2415        {
2416        code += PRIV(OP_lengths)[c];
2417        do code += GET(code, 1); while (*code == OP_ALT);
2418        c = *code;
2419        continue;
2420        }
2421    
2422      /* A nested group that is already marked as "could be empty" can just be
2423      skipped. */
2424    
2425      if (c == OP_SBRA  || c == OP_SBRAPOS ||
2426          c == OP_SCBRA || c == OP_SCBRAPOS)
2427        {
2428        do code += GET(code, 1); while (*code == OP_ALT);
2429        c = *code;
2430        continue;
2431        }
2432    
2433    /* For other groups, scan the branches. */    /* For other groups, scan the branches. */
2434    
2435    if (c == OP_BRA || c == OP_CBRA || c == OP_ONCE || c == OP_COND)    if (c == OP_BRA  || c == OP_BRAPOS ||
2436          c == OP_CBRA || c == OP_CBRAPOS ||
2437          c == OP_ONCE || c == OP_ONCE_NC ||
2438          c == OP_COND)
2439      {      {
2440      BOOL empty_branch;      BOOL empty_branch;
2441      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 2451  for (code = first_significant_code(code
2451        empty_branch = FALSE;        empty_branch = FALSE;
2452        do        do
2453          {          {
2454          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
2455            empty_branch = TRUE;            empty_branch = TRUE;
2456          code += GET(code, 1);          code += GET(code, 1);
2457          }          }
# Line 2005  for (code = first_significant_code(code Line 2469  for (code = first_significant_code(code
2469      {      {
2470      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2471      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2472      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2473      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"
2474      here. */      here. */
2475    
2476  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2477      case OP_XCLASS:      case OP_XCLASS:
2478      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2479      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2017  for (code = first_significant_code(code Line 2481  for (code = first_significant_code(code
2481    
2482      case OP_CLASS:      case OP_CLASS:
2483      case OP_NCLASS:      case OP_NCLASS:
2484      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2485    
2486  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2487      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2488  #endif  #endif
2489    
# Line 2058  for (code = first_significant_code(code Line 2522  for (code = first_significant_code(code
2522      case OP_ALLANY:      case OP_ALLANY:
2523      case OP_ANYBYTE:      case OP_ANYBYTE:
2524      case OP_CHAR:      case OP_CHAR:
2525      case OP_CHARNC:      case OP_CHARI:
2526      case OP_NOT:      case OP_NOT:
2527        case OP_NOTI:
2528      case OP_PLUS:      case OP_PLUS:
2529      case OP_MINPLUS:      case OP_MINPLUS:
2530      case OP_POSPLUS:      case OP_POSPLUS:
# Line 2091  for (code = first_significant_code(code Line 2556  for (code = first_significant_code(code
2556      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2557      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2558      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2559      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2560          code += 2;
2561      break;      break;
2562    
2563      /* End of branch */      /* End of branch */
# Line 2099  for (code = first_significant_code(code Line 2565  for (code = first_significant_code(code
2565      case OP_KET:      case OP_KET:
2566      case OP_KETRMAX:      case OP_KETRMAX:
2567      case OP_KETRMIN:      case OP_KETRMIN:
2568        case OP_KETRPOS:
2569      case OP_ALT:      case OP_ALT:
2570      return TRUE;      return TRUE;
2571    
2572      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2573      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2574    
2575  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2576      case OP_STAR:      case OP_STAR:
2577        case OP_STARI:
2578      case OP_MINSTAR:      case OP_MINSTAR:
2579        case OP_MINSTARI:
2580      case OP_POSSTAR:      case OP_POSSTAR:
2581        case OP_POSSTARI:
2582      case OP_QUERY:      case OP_QUERY:
2583        case OP_QUERYI:
2584      case OP_MINQUERY:      case OP_MINQUERY:
2585        case OP_MINQUERYI:
2586      case OP_POSQUERY:      case OP_POSQUERY:
2587      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      case OP_POSQUERYI:
2588        if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2589      break;      break;
2590    
2591      case OP_UPTO:      case OP_UPTO:
2592        case OP_UPTOI:
2593      case OP_MINUPTO:      case OP_MINUPTO:
2594        case OP_MINUPTOI:
2595      case OP_POSUPTO:      case OP_POSUPTO:
2596      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      case OP_POSUPTOI:
2597        if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);
2598      break;      break;
2599  #endif  #endif
2600    
# Line 2132  for (code = first_significant_code(code Line 2608  for (code = first_significant_code(code
2608      break;      break;
2609    
2610      case OP_THEN_ARG:      case OP_THEN_ARG:
2611      code += code[1+LINK_SIZE];      code += code[1];
2612      break;      break;
2613    
2614      /* 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 2631  return TRUE;
2631  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
2632  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,
2633  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.
2634    This function is called only during the real compile, not during the
2635    pre-compile.
2636    
2637  Arguments:  Arguments:
2638    code        points to start of the recursion    code        points to start of the recursion
2639    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2640    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2641    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2642    cd          pointers to tables etc    cd          pointers to tables etc
2643    
2644  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2645  */  */
2646    
2647  static BOOL  static BOOL
2648  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2649    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2650  {  {
2651  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2652    {    {
2653    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
2654      return FALSE;      return FALSE;
2655    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2656    }    }
# Line 2205  where Perl recognizes it as the POSIX cl Line 2683  where Perl recognizes it as the POSIX cl
2683  "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,
2684  I think.  I think.
2685    
2686    A user pointed out that PCRE was rejecting [:a[:digit:]] whereas Perl was not.
2687    It seems that the appearance of a nested POSIX class supersedes an apparent
2688    external class. For example, [:a[:digit:]b:] matches "a", "b", ":", or
2689    a digit.
2690    
2691    In Perl, unescaped square brackets may also appear as part of class names. For
2692    example, [:a[:abc]b:] gives unknown POSIX class "[:abc]b:]". However, for
2693    [:a[:abc]b][b:] it gives unknown POSIX class "[:abc]b][b:]", which does not
2694    seem right at all. PCRE does not allow closing square brackets in POSIX class
2695    names.
2696    
2697  Arguments:  Arguments:
2698    ptr      pointer to the initial [    ptr      pointer to the initial [
2699    endptr   where to return the end pointer    endptr   where to return the end pointer
# Line 2213  Returns:   TRUE or FALSE Line 2702  Returns:   TRUE or FALSE
2702  */  */
2703    
2704  static BOOL  static BOOL
2705  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2706  {  {
2707  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
2708  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2709  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != 0; ptr++)
2710    {    {
2711    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET) ptr++; else    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2712        ptr++;
2713      else if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;
2714      else
2715      {      {
     if (*ptr == CHAR_RIGHT_SQUARE_BRACKET) return FALSE;  
2716      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)      if (*ptr == terminator && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2717        {        {
2718        *endptr = ptr;        *endptr = ptr;
2719        return TRUE;        return TRUE;
2720        }        }
2721        if (*ptr == CHAR_LEFT_SQUARE_BRACKET &&
2722             (ptr[1] == CHAR_COLON || ptr[1] == CHAR_DOT ||
2723              ptr[1] == CHAR_EQUALS_SIGN) &&
2724            check_posix_syntax(ptr, endptr))
2725          return FALSE;
2726      }      }
2727    }    }
2728  return FALSE;  return FALSE;
# Line 2250  Returns:     a value representing the na Line 2746  Returns:     a value representing the na
2746  */  */
2747    
2748  static int  static int
2749  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2750  {  {
2751  const char *pn = posix_names;  const char *pn = posix_names;
2752  register int yield = 0;  register int yield = 0;
2753  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2754    {    {
2755    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2756      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;
2757    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2758    yield++;    yield++;
2759    }    }
# Line 2289  value in the reference (which is a group Line 2785  value in the reference (which is a group
2785  Arguments:  Arguments:
2786    group      points to the start of the group    group      points to the start of the group
2787    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2788    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2789    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2790    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
2791    
# Line 2297  Returns:     nothing Line 2793  Returns:     nothing
2793  */  */
2794    
2795  static void  static void
2796  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2797    uschar *save_hwm)    pcre_uchar *save_hwm)
2798  {  {
2799  uschar *ptr = group;  pcre_uchar *ptr = group;
2800    
2801  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2802    {    {
2803    int offset;    int offset;
2804    uschar *hc;    pcre_uchar *hc;
2805    
2806    /* 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
2807    reference. */    reference. */
# Line 2350  Arguments: Line 2846  Arguments:
2846  Returns:         new code pointer  Returns:         new code pointer
2847  */  */
2848    
2849  static uschar *  static pcre_uchar *
2850  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2851  {  {
2852  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2853  *code++ = 255;  *code++ = 255;
2854  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2855  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2856  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2857  }  }
2858    
2859    
# Line 2379  Returns:             nothing Line 2875  Returns:             nothing
2875  */  */
2876    
2877  static void  static void
2878  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2879  {  {
2880  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2881  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2393  PUT(previous_callout, 2 + LINK_SIZE, len Line 2889  PUT(previous_callout, 2 + LINK_SIZE, len
2889  *************************************************/  *************************************************/
2890    
2891  /* 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
2892  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
2893  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
2894  start address.  start address. A character with multiple other cases is returned on its own
2895    with a special return value.
2896    
2897  Arguments:  Arguments:
2898    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2403  Arguments: Line 2900  Arguments:
2900    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2901    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2902    
2903  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2904                   0 when a range is returned
2905                  >0 the CASESET offset for char with multiple other cases
2906                    in this case, ocptr contains the original
2907  */  */
2908    
2909  static BOOL  static int
2910  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,
2911    unsigned int *odptr)    unsigned int *odptr)
2912  {  {
2913  unsigned int c, othercase, next;  unsigned int c, othercase, next;
2914    int co;
2915    
2916    /* Find the first character that has an other case. If it has multiple other
2917    cases, return its case offset value. */
2918    
2919  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2920    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2921      if ((co = UCD_CASESET(c)) != 0)
2922        {
2923        *ocptr = c++;   /* Character that has the set */
2924        *cptr = c;      /* Rest of input range */
2925        return co;
2926        }
2927      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2928      }
2929    
2930  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2931    
2932  *ocptr = othercase;  *ocptr = othercase;
2933  next = othercase + 1;  next = othercase + 1;
# Line 2426  for (++c; c <= d; c++) Line 2938  for (++c; c <= d; c++)
2938    next++;    next++;
2939    }    }
2940    
2941  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2942  *cptr = c;  *cptr = c;             /* Rest of input range */
2943    return 0;
 return TRUE;  
2944  }  }
2945    
2946    
# Line 2453  Returns:       TRUE if auto-possessifyin Line 2964  Returns:       TRUE if auto-possessifyin
2964  static BOOL  static BOOL
2965  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(int c, int ptype, int pdata, BOOL negated)
2966  {  {
2967    #ifdef SUPPORT_UCP
2968    const pcre_uint32 *p;
2969    #endif
2970    
2971  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2972    
2973  switch(ptype)  switch(ptype)
2974    {    {
2975    case PT_LAMP:    case PT_LAMP:
# Line 2462  switch(ptype) Line 2978  switch(ptype)
2978            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
2979    
2980    case PT_GC:    case PT_GC:
2981    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2982    
2983    case PT_PC:    case PT_PC:
2984    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2473  switch(ptype) Line 2989  switch(ptype)
2989    /* These are specials */    /* These are specials */
2990    
2991    case PT_ALNUM:    case PT_ALNUM:
2992    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2993            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
2994    
2995    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
2996    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2997            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2998            == negated;            == negated;
2999    
3000    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
3001    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3002            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
3003            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
3004            == negated;            == negated;
3005    
3006    case PT_WORD:    case PT_WORD:
3007    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3008            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3009            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3010    
3011    #ifdef SUPPORT_UCP
3012      case PT_CLIST:
3013      p = PRIV(ucd_caseless_sets) + prop->caseset;
3014      for (;;)
3015        {
3016        if ((unsigned int)c < *p) return !negated;
3017        if ((unsigned int)c == *p++) return negated;
3018        }
3019      break;  /* Control never reaches here */
3020    #endif
3021    }    }
3022    
3023  return FALSE;  return FALSE;
3024  }  }
3025  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2508  sense to automatically possessify the re Line 3036  sense to automatically possessify the re
3036    
3037  Arguments:  Arguments:
3038    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3039    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3040    ptr           next character in pattern    ptr           next character in pattern
3041    options       options bits    options       options bits
3042    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2517  Returns:        TRUE if possessifying is Line 3045  Returns:        TRUE if possessifying is
3045  */  */
3046    
3047  static BOOL  static BOOL
3048  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3049    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3050  {  {
3051  int c, next;  pcre_int32 c = NOTACHAR; // FIXMEchpe pcre_uint32
3052    pcre_int32 next;
3053    int escape;
3054  int op_code = *previous++;  int op_code = *previous++;
3055    
3056  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2529  if ((options & PCRE_EXTENDED) != 0) Line 3059  if ((options & PCRE_EXTENDED) != 0)
3059    {    {
3060    for (;;)    for (;;)
3061      {      {
3062      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3063      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3064        {        {
3065        ptr++;        ptr++;
3066        while (*ptr != 0)        while (*ptr != 0)
3067          {          {
3068          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3069          ptr++;          ptr++;
3070  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3071          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3072  #endif  #endif
3073          }          }
3074        }        }
# Line 2552  value is a character, a negative value i Line 3082  value is a character, a negative value i
3082  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3083    {    {
3084    int temperrorcode = 0;    int temperrorcode = 0;
3085    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3086    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3087    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3088    }    }
3089    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
3090    {    {
3091  #ifdef SUPPORT_UTF8    escape = 0;
3092    if (utf8) { GETCHARINC(next, ptr); } else  #ifdef SUPPORT_UTF
3093      if (utf) { GETCHARINC(next, ptr); } else
3094  #endif  #endif
3095    next = *ptr++;    next = *ptr++;
3096    }    }
   
3097  else return FALSE;  else return FALSE;
3098    
3099  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2573  if ((options & PCRE_EXTENDED) != 0) Line 3102  if ((options & PCRE_EXTENDED) != 0)
3102    {    {
3103    for (;;)    for (;;)
3104      {      {
3105      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3106      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3107        {        {
3108        ptr++;        ptr++;
3109        while (*ptr != 0)        while (*ptr != 0)
3110          {          {
3111          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3112          ptr++;          ptr++;
3113  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3114          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3115  #endif  #endif
3116          }          }
3117        }        }
# Line 2593  if ((options & PCRE_EXTENDED) != 0) Line 3122  if ((options & PCRE_EXTENDED) != 0)
3122  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3123    
3124  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3125    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)
3126      return FALSE;      return FALSE;
3127    
3128  /* 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. */  
3129    
3130  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3131        op_code == OP_NOT || op_code == OP_NOTI)
3132      //if (escape == 0) switch(op_code)
3133    {    {
3134    case OP_CHAR:  #ifdef SUPPORT_UTF
 #ifdef SUPPORT_UTF8  
3135    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3136  #else  #else
3137    c = *previous;    c = *previous;
3138  #endif  #endif
3139    return c != next;    }
3140    
3141    /* 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
3142    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3143    
3144    case OP_CHARNC:  if (escape == 0)
3145  #ifdef SUPPORT_UTF8    {
3146    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3147  #else    case, which maps to the special PT_CLIST property. Check this first. */
3148    c = *previous;  
3149    #ifdef SUPPORT_UCP
3150      if (utf && (unsigned int)c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3151        {
3152        int ocs = UCD_CASESET(next);
3153        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3154        }
3155  #endif  #endif
3156    if (c == next) return FALSE;  
3157  #ifdef SUPPORT_UTF8    switch(op_code)
   if (utf8)  
3158      {      {
3159      unsigned int othercase;      case OP_CHAR:
3160      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3161    
3162        /* For CHARI (caseless character) we must check the other case. If we have
3163        Unicode property support, we can use it to test the other case of
3164        high-valued characters. We know that next can have only one other case,
3165        because multi-other-case characters are dealt with above. */
3166    
3167        case OP_CHARI:
3168        if (c == next) return FALSE;
3169    #ifdef SUPPORT_UTF
3170        if (utf)
3171          {
3172          unsigned int othercase;
3173          if (next < 128) othercase = cd->fcc[next]; else
3174  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3175      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE((unsigned int)next);
3176  #else  #else
3177      othercase = NOTACHAR;        othercase = NOTACHAR;
3178  #endif  #endif
3179      return (unsigned int)c != othercase;        return (unsigned int)c != othercase;
3180      }        }
3181    else      else
3182  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3183    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */      return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3184    
3185    /* For OP_NOT, its data is always a single-byte character. */      case OP_NOT:
3186        return c == next;
3187    case OP_NOT:  
3188    if ((c = *previous) == next) return TRUE;      case OP_NOTI:
3189    if ((options & PCRE_CASELESS) == 0) return FALSE;      if (c == next) return TRUE;
3190  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3191    if (utf8)      if (utf)
3192      {        {
3193      unsigned int othercase;        unsigned int othercase;
3194      if (next < 128) othercase = cd->fcc[next]; else        if (next < 128) othercase = cd->fcc[next]; else
3195  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3196      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE((unsigned int)next);
3197  #else  #else
3198      othercase = NOTACHAR;        othercase = NOTACHAR;
3199  #endif  #endif
3200      return (unsigned int)c == othercase;        return (unsigned int)c == othercase;
3201      }        }
3202    else      else
3203  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3204    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */      return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3205    
3206    /* 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.
3207    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. */
3208    
3209    case OP_DIGIT:      case OP_DIGIT:
3210    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3211    
3212    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3213    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3214    
3215    case OP_WHITESPACE:      case OP_WHITESPACE:
3216    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3217    
3218    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3219    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3220    
3221    case OP_WORDCHAR:      case OP_WORDCHAR:
3222    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3223    
3224    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3225    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3226    
3227    case OP_HSPACE:      case OP_HSPACE:
3228    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3229    switch(next)      switch(next)
3230      {        {
3231      case 0x09:        HSPACE_CASES:
3232      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;  
     }  
3233    
3234    case OP_ANYNL:        default:
3235    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3236    case OP_NOT_VSPACE:        }
3237    switch(next)  
3238      {      case OP_ANYNL:
3239      case 0x0a:      case OP_VSPACE:
3240      case 0x0b:      case OP_NOT_VSPACE:
3241      case 0x0c:      switch(next)
3242      case 0x0d:        {
3243      case 0x85:        VSPACE_CASES:
3244      case 0x2028:        return op_code == OP_NOT_VSPACE;
3245      case 0x2029:  
3246      return op_code == OP_NOT_VSPACE;        default:
3247      default:        return op_code != OP_NOT_VSPACE;
3248      return op_code != OP_NOT_VSPACE;        }
     }  
3249    
3250  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3251    case OP_PROP:      case OP_PROP:
3252    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3253    
3254    case OP_NOTPROP:      case OP_NOTPROP:
3255    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3256  #endif  #endif
3257    
3258    default:      default:
3259    return FALSE;      return FALSE;
3260        }
3261    }    }
3262    
   
3263  /* 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
3264  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
3265  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 3269  replaced by OP_PROP codes when PCRE_UCP
3269  switch(op_code)  switch(op_code)
3270    {    {
3271    case OP_CHAR:    case OP_CHAR:
3272    case OP_CHARNC:    case OP_CHARI:
3273  #ifdef SUPPORT_UTF8    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
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_D:      case ESC_D:
3279      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 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_S:      case ESC_S:
3285      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 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_W:      case ESC_W:
3291      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3292    
3293      case ESC_h:      case ESC_h:
3294      case ESC_H:      case ESC_H:
3295      switch(c)      switch(c)
3296        {        {
3297        case 0x09:        HSPACE_CASES:
3298        case 0x20:        return escape != ESC_h;
3299        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;  
3300        default:        default:
3301        return -next == ESC_h;        return escape == ESC_h;
3302        }        }
3303    
3304      case ESC_v:      case ESC_v:
3305      case ESC_V:      case ESC_V:
3306      switch(c)      switch(c)
3307        {        {
3308        case 0x0a:        VSPACE_CASES:
3309        case 0x0b:        return escape != ESC_v;
3310        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3311        default:        default:
3312        return -next == ESC_v;        return escape == ESC_v;
3313        }        }
3314    
3315      /* 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
3316      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3317      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3318    
3319  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3320      case ESC_du:      case ESC_du:
# Line 2828  switch(op_code) Line 3325  switch(op_code)
3325      case ESC_SU:      case ESC_SU:
3326        {        {
3327        int temperrorcode = 0;        int temperrorcode = 0;
3328        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3329        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3330        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3331        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3332        }        }
# Line 2851  switch(op_code) Line 3348  switch(op_code)
3348        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. */
3349    
3350        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3351          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)
3352            return FALSE;            return FALSE;
3353    
3354        /* Do the property check. */        /* Do the property check. */
3355    
3356        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3357        }        }
3358  #endif  #endif
3359    
# Line 2871  switch(op_code) Line 3368  switch(op_code)
3368    these op-codes are never generated.) */    these op-codes are never generated.) */
3369    
3370    case OP_DIGIT:    case OP_DIGIT:
3371    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3372           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3373    
3374    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3375    return next == -ESC_d;    return escape == ESC_d;
3376    
3377    case OP_WHITESPACE:    case OP_WHITESPACE:
3378    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3379    
3380    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3381    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3382    
3383    case OP_HSPACE:    case OP_HSPACE:
3384    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3385           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3386    
3387    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3388    return next == -ESC_h;    return escape == ESC_h;
3389    
3390    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3391    case OP_ANYNL:    case OP_ANYNL:
3392    case OP_VSPACE:    case OP_VSPACE:
3393    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3394    
3395    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3396    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3397    
3398    case OP_WORDCHAR:    case OP_WORDCHAR:
3399    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3400           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3401    
3402    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3403    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3404    
3405    default:    default:
3406    return FALSE;    return FALSE;
# Line 2915  switch(op_code) Line 3412  switch(op_code)
3412    
3413    
3414  /*************************************************  /*************************************************
3415    *        Add a character or range to a class     *
3416    *************************************************/
3417    
3418    /* This function packages up the logic of adding a character or range of
3419    characters to a class. The character values in the arguments will be within the
3420    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3421    mutually recursive with the function immediately below.
3422    
3423    Arguments:
3424      classbits     the bit map for characters < 256
3425      uchardptr     points to the pointer for extra data
3426      options       the options word
3427      cd            contains pointers to tables etc.
3428      start         start of range character
3429      end           end of range character
3430    
3431    Returns:        the number of < 256 characters added
3432                    the pointer to extra data is updated
3433    */
3434    
3435    static int
3436    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3437      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3438    {
3439    pcre_uint32 c;
3440    int n8 = 0;
3441    
3442    /* If caseless matching is required, scan the range and process alternate
3443    cases. In Unicode, there are 8-bit characters that have alternate cases that
3444    are greater than 255 and vice-versa. Sometimes we can just extend the original
3445    range. */
3446    
3447    if ((options & PCRE_CASELESS) != 0)
3448      {
3449    #ifdef SUPPORT_UCP
3450      if ((options & PCRE_UTF8) != 0)
3451        {
3452        int rc;
3453        pcre_uint32 oc, od;
3454    
3455        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3456        c = start;
3457    
3458        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3459          {
3460          /* Handle a single character that has more than one other case. */
3461    
3462          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3463            PRIV(ucd_caseless_sets) + rc, oc);
3464    
3465          /* Do nothing if the other case range is within the original range. */
3466    
3467          else if (oc >= start && od <= end) continue;
3468    
3469          /* Extend the original range if there is overlap, noting that if oc < c, we
3470          can't have od > end because a subrange is always shorter than the basic
3471          range. Otherwise, use a recursive call to add the additional range. */
3472    
3473          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3474          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3475          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3476          }
3477        }
3478      else
3479    #endif  /* SUPPORT_UCP */
3480    
3481      /* Not UTF-mode, or no UCP */
3482    
3483      for (c = start; c <= end && c < 256; c++)
3484        {
3485        SETBIT(classbits, cd->fcc[c]);
3486        n8++;
3487        }
3488      }
3489    
3490    /* Now handle the original range. Adjust the final value according to the bit
3491    length - this means that the same lists of (e.g.) horizontal spaces can be used
3492    in all cases. */
3493    
3494    #if defined COMPILE_PCRE8
3495    #ifdef SUPPORT_UTF
3496      if ((options & PCRE_UTF8) == 0)
3497    #endif
3498      if (end > 0xff) end = 0xff;
3499    
3500    #elif defined COMPILE_PCRE16
3501    #ifdef SUPPORT_UTF
3502      if ((options & PCRE_UTF16) == 0)
3503    #endif
3504      if (end > 0xffff) end = 0xffff;
3505    
3506    #endif /* COMPILE_PCRE[8|16] */
3507    
3508    /* If all characters are less than 256, use the bit map. Otherwise use extra
3509    data. */
3510    
3511    if (end < 0x100)
3512      {
3513      for (c = start; c <= end; c++)
3514        {
3515        n8++;
3516        SETBIT(classbits, c);
3517        }
3518      }
3519    
3520    else
3521      {
3522      pcre_uchar *uchardata = *uchardptr;
3523    
3524    #ifdef SUPPORT_UTF
3525      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3526        {
3527        if (start < end)
3528          {
3529          *uchardata++ = XCL_RANGE;
3530          uchardata += PRIV(ord2utf)(start, uchardata);
3531          uchardata += PRIV(ord2utf)(end, uchardata);
3532          }
3533        else if (start == end)
3534          {
3535          *uchardata++ = XCL_SINGLE;
3536          uchardata += PRIV(ord2utf)(start, uchardata);
3537          }
3538        }
3539      else
3540    #endif  /* SUPPORT_UTF */
3541    
3542      /* Without UTF support, character values are constrained by the bit length,
3543      and can only be > 256 for 16-bit and 32-bit libraries. */
3544    
3545    #ifdef COMPILE_PCRE8
3546        {}
3547    #else
3548      if (start < end)
3549        {
3550        *uchardata++ = XCL_RANGE;
3551        *uchardata++ = start;
3552        *uchardata++ = end;
3553        }
3554      else if (start == end)
3555        {
3556        *uchardata++ = XCL_SINGLE;
3557        *uchardata++ = start;
3558        }
3559    #endif
3560    
3561      *uchardptr = uchardata;   /* Updata extra data pointer */
3562      }
3563    
3564    return n8;    /* Number of 8-bit characters */
3565    }
3566    
3567    
3568    
3569    
3570    /*************************************************
3571    *        Add a list of characters to a class     *
3572    *************************************************/
3573    
3574    /* This function is used for adding a list of case-equivalent characters to a
3575    class, and also for adding a list of horizontal or vertical whitespace. If the
3576    list is in order (which it should be), ranges of characters are detected and
3577    handled appropriately. This function is mutually recursive with the function
3578    above.
3579    
3580    Arguments:
3581      classbits     the bit map for characters < 256
3582      uchardptr     points to the pointer for extra data
3583      options       the options word
3584      cd            contains pointers to tables etc.
3585      p             points to row of 32-bit values, terminated by NOTACHAR
3586      except        character to omit; this is used when adding lists of
3587                      case-equivalent characters to avoid including the one we
3588                      already know about
3589    
3590    Returns:        the number of < 256 characters added
3591                    the pointer to extra data is updated
3592    */
3593    
3594    static int
3595    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3596      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3597    {
3598    int n8 = 0;
3599    while (p[0] < NOTACHAR)
3600      {
3601      int n = 0;
3602      if (p[0] != except)
3603        {
3604        while(p[n+1] == p[0] + n + 1) n++;
3605        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3606        }
3607      p += n + 1;
3608      }
3609    return n8;
3610    }
3611    
3612    
3613    
3614    /*************************************************
3615    *    Add characters not in a list to a class     *
3616    *************************************************/
3617    
3618    /* This function is used for adding the complement of a list of horizontal or
3619    vertical whitespace to a class. The list must be in order.
3620    
3621    Arguments:
3622      classbits     the bit map for characters < 256
3623      uchardptr     points to the pointer for extra data
3624      options       the options word
3625      cd            contains pointers to tables etc.
3626      p             points to row of 32-bit values, terminated by NOTACHAR
3627    
3628    Returns:        the number of < 256 characters added
3629                    the pointer to extra data is updated
3630    */
3631    
3632    static int
3633    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3634      int options, compile_data *cd, const pcre_uint32 *p)
3635    {
3636    BOOL utf = (options & PCRE_UTF8) != 0;
3637    int n8 = 0;
3638    if (p[0] > 0)
3639      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3640    while (p[0] < NOTACHAR)
3641      {
3642      while (p[1] == p[0] + 1) p++;
3643      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3644        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3645      p++;
3646      }
3647    return n8;
3648    }
3649    
3650    
3651    
3652    /*************************************************
3653  *           Compile one branch                   *  *           Compile one branch                   *
3654  *************************************************/  *************************************************/
3655    
# Line 2929  Arguments: Line 3664  Arguments:
3664    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3665    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3666    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3667    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)
3668    reqbyteptr     set to the last literal character required, else < 0    reqcharptr     set to the last literal character required, else < 0
3669    bcptr          points to current branch chain    bcptr          points to current branch chain
3670      cond_depth     conditional nesting depth
3671    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
3672    lengthptr      NULL during the real compile phase    lengthptr      NULL during the real compile phase
3673                   points to length accumulator during pre-compile phase                   points to length accumulator during pre-compile phase
# Line 2941  Returns:         TRUE on success Line 3677  Returns:         TRUE on success
3677  */  */
3678    
3679  static BOOL  static BOOL
3680  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3681    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,
3682      pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,
3683    compile_data *cd, int *lengthptr)    compile_data *cd, int *lengthptr)
3684  {  {
3685  int repeat_type, op_type;  int repeat_type, op_type;
3686  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3687  int bravalue = 0;  int bravalue = 0;
3688  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3689  int firstbyte, reqbyte;  pcre_int32 firstchar, reqchar;
3690  int zeroreqbyte, zerofirstbyte;  pcre_int32 zeroreqchar, zerofirstchar;
3691  int req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3692  int options = *optionsptr;  int options = *optionsptr;               /* May change dynamically */
3693  int after_manual_callout = 0;  int after_manual_callout = 0;
3694  int length_prevgroup = 0;  int length_prevgroup = 0;
3695  register int c;  register int c;
3696  register uschar *code = *codeptr;  int escape;
3697  uschar *last_code = code;  register pcre_uchar *code = *codeptr;
3698  uschar *orig_code = code;  pcre_uchar *last_code = code;
3699  uschar *tempcode;  pcre_uchar *orig_code = code;
3700    pcre_uchar *tempcode;
3701  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3702  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3703  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3704  const uschar *tempptr;  const pcre_uchar *tempptr;
3705  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3706  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3707  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3708  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3709  uschar classbits[32];  pcre_uint8 classbits[32];
3710    
3711  #ifdef SUPPORT_UTF8  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we
3712  BOOL class_utf8;  must not do this for other options (e.g. PCRE_EXTENDED) because they may change
3713  BOOL utf8 = (options & PCRE_UTF8) != 0;  dynamically as we process the pattern. */
3714  uschar *class_utf8data;  
3715  uschar *class_utf8data_base;  #ifdef SUPPORT_UTF
3716  uschar utf8_char[6];  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3717    BOOL utf = (options & PCRE_UTF8) != 0;
3718    pcre_uchar utf_chars[6];
3719  #else  #else
3720  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3721  uschar *utf8_char = NULL;  #endif
3722    
3723    /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3724    class_uchardata always so that it can be passed to add_to_class() always,
3725    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3726    alternative calls for the different cases. */
3727    
3728    pcre_uchar *class_uchardata;
3729    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3730    BOOL xclass;
3731    pcre_uchar *class_uchardata_base;
3732  #endif  #endif
3733    
3734  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 2992  greedy_non_default = greedy_default ^ 1; Line 3742  greedy_non_default = greedy_default ^ 1;
3742    
3743  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3744  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
3745  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
3746  find one.  find one.
3747    
3748  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
3749  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
3750  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3751  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3752    
3753  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;
3754    
3755  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3756  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
3757  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
3758  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3759    value. This is used only for ASCII characters. */
3760    
3761  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3762    
3763  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3764    
# Line 3019  for (;; ptr++) Line 3770  for (;; ptr++)
3770    BOOL is_quantifier;    BOOL is_quantifier;
3771    BOOL is_recurse;    BOOL is_recurse;
3772    BOOL reset_bracount;    BOOL reset_bracount;
3773    int class_charcount;    int class_has_8bitchar;
3774    int class_lastchar;    int class_one_char;
3775    int newoptions;    int newoptions;
3776    int recno;    int recno;
3777    int refsign;    int refsign;
3778    int skipbytes;    int skipbytes;
3779    int subreqbyte;    int subreqchar;
3780    int subfirstbyte;    int subfirstchar;
3781    int terminator;    int terminator;
3782    int mclength;    int mclength;
3783    uschar mcbuffer[8];    int tempbracount;
3784      int ec; // FIXMEchpe pcre_uint32
3785      pcre_uchar mcbuffer[8];
3786    
3787    /* Get next byte in the pattern */    /* Get next character in the pattern */
3788    
3789    c = *ptr;    c = *ptr;
3790    
# Line 3053  for (;; ptr++) Line 3806  for (;; ptr++)
3806  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3807      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3808  #endif  #endif
3809      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */      if (code > cd->start_workspace + cd->workspace_size -
3810            WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3811        {        {
3812        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3813        goto FAILED;        goto FAILED;
# Line 3076  for (;; ptr++) Line 3830  for (;; ptr++)
3830        }        }
3831    
3832      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3833      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3834          (int)(code - last_code), c, c));
3835    
3836      /* 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
3837      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 3841  for (;; ptr++)
3841        {        {
3842        if (previous > orig_code)        if (previous > orig_code)
3843          {          {
3844          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3845          code -= previous - orig_code;          code -= previous - orig_code;
3846          previous = orig_code;          previous = orig_code;
3847          }          }
# Line 3102  for (;; ptr++) Line 3857  for (;; ptr++)
3857    /* 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
3858    reference list. */    reference list. */
3859    
3860    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3861               WORK_SIZE_SAFETY_MARGIN)
3862      {      {
3863      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3864      goto FAILED;      goto FAILED;
# Line 3150  for (;; ptr++) Line 3906  for (;; ptr++)
3906      previous_callout = NULL;      previous_callout = NULL;
3907      }      }
3908    
3909    /* In extended mode, skip white space and comments */    /* In extended mode, skip white space and comments. */
3910    
3911    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
3912      {      {
3913      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
3914      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3915        {        {
3916        ptr++;        ptr++;
3917        while (*ptr != 0)        while (*ptr != 0)
3918          {          {
3919          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3920          ptr++;          ptr++;
3921  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3922          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3923  #endif  #endif
3924          }          }
3925        if (*ptr != 0) continue;        if (*ptr != 0) continue;
# Line 3187  for (;; ptr++) Line 3943  for (;; ptr++)
3943      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
3944      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3945      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3946      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
3947      *reqbyteptr = reqbyte;      *reqcharptr = reqchar;
3948      *codeptr = code;      *codeptr = code;
3949      *ptrptr = ptr;      *ptrptr = ptr;
3950      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3209  for (;; ptr++) Line 3965  for (;; ptr++)
3965      the setting of any following char as a first character. */      the setting of any following char as a first character. */
3966    
3967      case CHAR_CIRCUMFLEX_ACCENT:      case CHAR_CIRCUMFLEX_ACCENT:
3968        previous = NULL;
3969      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3970        {        {
3971        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3972          *code++ = OP_CIRCM;
3973        }        }
3974      previous = NULL;      else *code++ = OP_CIRC;
     *code++ = OP_CIRC;  
3975      break;      break;
3976    
3977      case CHAR_DOLLAR_SIGN:      case CHAR_DOLLAR_SIGN:
3978      previous = NULL;      previous = NULL;
3979      *code++ = OP_DOLL;      *code++ = ((options & PCRE_MULTILINE) != 0)? OP_DOLLM : OP_DOLL;
3980      break;      break;
3981    
3982      /* 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
3983      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
3984    
3985      case CHAR_DOT:      case CHAR_DOT:
3986      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3987      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
3988      zeroreqbyte = reqbyte;      zeroreqchar = reqchar;
3989      previous = code;      previous = code;
3990      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
3991      break;      break;
# Line 3283  for (;; ptr++) Line 4040  for (;; ptr++)
4040          {          {
4041          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
4042            ptr++;            ptr++;
4043          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)  
4044            ptr += 3;            ptr += 3;
4045          else          else
4046            break;            break;
# Line 3303  for (;; ptr++) Line 4059  for (;; ptr++)
4059          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4060        {        {
4061        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4062        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4063        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
4064        break;        break;
4065        }        }
4066    
# Line 3314  for (;; ptr++) Line 4070  for (;; ptr++)
4070    
4071      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4072    
4073      /* 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:
4074      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 <
4075      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
4076        character. */
4077    
4078      class_charcount = 0;      class_has_8bitchar = 0;
4079      class_lastchar = -1;      class_one_char = 0;
4080    
4081      /* 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
4082      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
4083      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
4084      */      map. */
4085    
4086      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4087    
4088  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4089      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;
4090      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4091      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4092  #endif  #endif
4093    
4094      /* 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 4097  for (;; ptr++)
4097    
4098      if (c != 0) do      if (c != 0) do
4099        {        {
4100        const uschar *oldptr;        const pcre_uchar *oldptr;
4101    
4102  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4103        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
4104          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4105          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
4106          }          }
4107    #endif
4108    
4109        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4110          /* In the pre-compile phase, accumulate the length of any extra
4111        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
4112        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4113        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4114          however. */
4115        if (lengthptr != NULL)  
4116          {        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4117          *lengthptr += class_utf8data - class_utf8data_base;          {
4118          class_utf8data = class_utf8data_base;          xclass = TRUE;
4119            *lengthptr += class_uchardata - class_uchardata_base;
4120            class_uchardata = class_uchardata_base;
4121          }          }
   
4122  #endif  #endif
4123    
4124        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3386  for (;; ptr++) Line 4146  for (;; ptr++)
4146          {          {
4147          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
4148          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
4149          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
4150          uschar pbits[32];          pcre_uint8 pbits[32];
4151    
4152          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
4153            {            {
# Line 3414  for (;; ptr++) Line 4174  for (;; ptr++)
4174          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4175          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4176    
4177          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4178            posix_class = 0;            posix_class = 0;
4179    
4180          /* 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 4202  for (;; ptr++)
4202          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
4203    
4204          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
4205            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
4206    
4207          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
4208    
# Line 3457  for (;; ptr++) Line 4217  for (;; ptr++)
4217              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4218            }            }
4219    
4220          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4221          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4222    
4223          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3473  for (;; ptr++) Line 4233  for (;; ptr++)
4233            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4234    
4235          ptr = tempptr + 1;          ptr = tempptr + 1;
4236          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 character. */
4237            class_has_8bitchar = 1;
4238            /* Every class contains at least two characters. */
4239            class_one_char = 2;
4240          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4241          }          }
4242    
4243        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
4244        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
4245        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
4246        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
4247        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_one_char bigger
4248        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
4249          as literal characters (by default), or are faulted if
4250        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4251    
4252        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4253          {          {
4254          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4255    
4256          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4257    
4258          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4259          else if (-c == ESC_Q)            /* Handle start of quoted string */            c = ec;
4260            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4261            else if (escape == ESC_N)            /* \N is not supported in a class */
4262              {
4263              *errorcodeptr = ERR71;
4264              goto FAILED;
4265              }
4266            else if (escape == ESC_Q)            /* Handle start of quoted string */
4267            {            {
4268            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4269              {              {
# Line 3500  for (;; ptr++) Line 4272  for (;; ptr++)
4272            else inescq = TRUE;            else inescq = TRUE;
4273            continue;            continue;
4274            }            }
4275          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4276    
4277          if (c < 0)          else
4278            {            {
4279            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4280            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
4281              class_has_8bitchar++;
4282              /* Every class contains at least two characters. */
4283              class_one_char += 2;
4284    
4285            switch (-c)            switch (escape)
4286              {              {
4287  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4288              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 4292  for (;; ptr++)
4292              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4293              case ESC_SU:              case ESC_SU:
4294              nestptr = ptr;              nestptr = ptr;
4295              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4296              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4297              continue;              continue;
4298  #endif  #endif
4299              case ESC_d:              case ESC_d:
# Line 3540  for (;; ptr++) Line 4315  for (;; ptr++)
4315              continue;              continue;
4316    
4317              /* 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
4318              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4319              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4320                EBCDIC, so we lazily just adjust the appropriate bit. */
4321    
4322              case ESC_s:              case ESC_s:
4323              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
4324              classbits[1] |= cbits[cbit_space+1] & ~0x08;              classbits[1] |= cbits[cbit_space+1] & ~0x08;
4325              for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];              for (c = 2; c < 32; c++) classbits[c] |= cbits[c+cbit_space];
4326              continue;              continue;
4327    
# Line 3554  for (;; ptr++) Line 4330  for (;; ptr++)
4330              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4331              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4332              continue;              continue;
4333    
4334                /* The rest apply in both UCP and non-UCP cases. */
4335    
4336              case ESC_h:              case ESC_h:
4337              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4338              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  
4339              continue;              continue;
4340    
4341              case ESC_H:              case ESC_H:
4342              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4343                {                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  
4344              continue;              continue;
4345    
4346              case ESC_v:              case ESC_v:
4347              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4348              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  
4349              continue;              continue;
4350    
4351              case ESC_V:              case ESC_V:
4352              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4353                {                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  
4354              continue;              continue;
4355    
4356  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3679  for (;; ptr++) Line 4361  for (;; ptr++)
4361                int pdata;                int pdata;
4362                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
4363                if (ptype < 0) goto FAILED;                if (ptype < 0) goto FAILED;
4364                class_utf8 = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_utf8data++ = ((-c == ESC_p) != negated)?  
4365                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4366                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4367                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4368                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4369                continue;                continue;
4370                }                }
4371  #endif  #endif
# Line 3698  for (;; ptr++) Line 4379  for (;; ptr++)
4379                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4380                goto FAILED;                goto FAILED;
4381                }                }
4382              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4383              c = *ptr;              /* Get the final character and fall through */              class_one_char -= 2;     /* Undo the speculative increase. */
4384                c = *ptr;                /* Get the final character and fall through */
4385              break;              break;
4386              }              }
4387            }            }
4388    
4389          /* 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).
4390          greater than 256 in UTF-8 mode. */          This may be greater than 256. */
4391    
4392            escape = 0;
4393    
4394          }   /* End of backslash handling */          }   /* End of backslash handling */
4395    
4396        /* 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
4397        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
4398        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4399        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4400    
4401        CHECK_RANGE:        CHECK_RANGE:
4402        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3720  for (;; ptr++) Line 4404  for (;; ptr++)
4404          inescq = FALSE;          inescq = FALSE;
4405          ptr += 2;          ptr += 2;
4406          }          }
   
4407        oldptr = ptr;        oldptr = ptr;
4408    
4409        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4410    
4411        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4412    
# Line 3746  for (;; ptr++) Line 4429  for (;; ptr++)
4429            inescq = TRUE;            inescq = TRUE;
4430            break;            break;
4431            }            }
4432    
4433            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4434            back the pointer and jump to handle the character that preceded it. */
4435    
4436          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4437            {            {
4438            ptr = oldptr;            ptr = oldptr;
4439            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4440            }            }
4441    
4442            /* Otherwise, we have a potential range; pick up the next character */
4443    
4444  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4445          if (utf8)          if (utf)
4446            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4447            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4448            }            }
# Line 3768  for (;; ptr++) Line 4456  for (;; ptr++)
4456    
4457          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4458            {            {
4459            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4460              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4461            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4462    
4463            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4464    
4465            if (d < 0)            if (descape != 0)
4466              {              {
4467              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4468                {                {
4469                ptr = oldptr;                ptr = oldptr;
4470                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4471                }                }
4472              }              }
4473            }            }
   
         /* 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);  
               }  
             }  
 #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;  
4474    
4475          /* 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
4476            one-character ranges. */
4477    
4478          if (lengthptr == NULL) for (; c <= d; c++)          if (d < c)
4479            {            {
4480            classbits[c/8] |= (1 << (c&7));            *errorcodeptr = ERR8;
4481            if ((options & PCRE_CASELESS) != 0)            goto FAILED;
             {  
             int uc = cd->fcc[c];           /* flip case */  
             classbits[uc/8] |= (1 << (uc&7));  
             }  
4482            }            }
4483            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4484    
4485          continue;   /* Go get the next char in the class */          /* We have found a character range, so single character optimizations
4486          }          cannot be done anymore. Any value greater than 1 indicates that there
4487            is more than one character. */
4488    
4489            class_one_char = 2;
4490    
4491        /* 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. */  
4492    
4493        LONE_SINGLE_CHARACTER:          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4494    
4495            class_has_8bitchar +=
4496              add_to_class(classbits, &class_uchardata, options, cd, c, d);
4497    
4498            continue;   /* Go get the next char in the class */
4499            }
4500    
4501        /* Handle a character that cannot go in the bit map */        /* Handle a single character - we can get here for a normal non-escape
4502          char, or after \ that introduces a single character or for an apparent
4503          range that isn't. Only the value 1 matters for class_one_char, so don't
4504          increase it if it is already 2 or more ... just in case there's a class
4505          with a zillion characters in it. */
4506    
4507          CLASS_SINGLE_CHARACTER:
4508          if (class_one_char < 2) class_one_char++;
4509    
4510          /* If class_one_char is 1, we have the first single character in the
4511          class, and there have been no prior ranges, or XCLASS items generated by
4512          escapes. If this is the final character in the class, we can optimize by
4513          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4514          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4515          to be set. Otherwise, there can be no first char if this item is first,
4516          whatever repeat count may follow. In the case of reqchar, save the
4517          previous value for reinstating. */
4518    
4519  #ifdef SUPPORT_UTF8        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
       if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
4520          {          {
4521          class_utf8 = TRUE;          ptr++;
4522          *class_utf8data++ = XCL_SINGLE;          zeroreqchar = reqchar;
         class_utf8data += _pcre_ord2utf8(c, class_utf8data);  
4523    
4524  #ifdef SUPPORT_UCP          if (negate_class)
         if ((options & PCRE_CASELESS) != 0)  
4525            {            {
4526            unsigned int othercase;  #ifdef SUPPORT_UCP
4527            if ((othercase = UCD_OTHERCASE(c)) != c)              // FIXMEchpe pcreuint32?
4528              int d;
4529    #endif
4530              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4531              zerofirstchar = firstchar;
4532    
4533              /* For caseless UTF-8 mode when UCP support is available, check
4534              whether this character has more than one other case. If so, generate
4535              a special OP_NOTPROP item instead of OP_NOTI. */
4536    
4537    #ifdef SUPPORT_UCP
4538              if (utf && (options & PCRE_CASELESS) != 0 &&
4539                  (d = UCD_CASESET(c)) != 0)
4540                {
4541                *code++ = OP_NOTPROP;
4542                *code++ = PT_CLIST;
4543                *code++ = d;
4544                }
4545              else
4546    #endif
4547              /* Char has only one other case, or UCP not available */
4548    
4549              {              {
4550              *class_utf8data++ = XCL_SINGLE;              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4551              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4552                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4553                  code += PRIV(ord2utf)(c, code);
4554                else
4555    #endif
4556                  *code++ = c;
4557              }              }