/[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 745 by ph10, Mon Nov 14 11:41:03 2011 UTC revision 1061 by chpe, Tue Oct 16 15:54:02 2012 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2011 University of Cambridge             Copyright (c) 1997-2012 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. */  library. We do not need to select pcre16_printint.c specially, because the
59    COMPILE_PCREx macro will already be appropriately set. */
60    
61  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
62  #include "pcre_printint.src"  /* pcre_printint.c should not include any headers */
63    #define PCRE_INCLUDED
64    #include "pcre_printint.c"
65    #undef PCRE_INCLUDED
66  #endif  #endif
67    
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 73  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL,
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"
# 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"    "\\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 428  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 472  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 507  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 543  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, uschar **, const uschar **, int *, BOOL, BOOL, int, int,  
     int *, int *, branch_chain *, compile_data *, int *);  
   
647    
648    
649  /*************************************************  /*************************************************
# Line 578  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            *  *            Check for counted repeat            *
714  *************************************************/  *************************************************/
715    
# Line 593  Returns:    TRUE or FALSE Line 725  Returns:    TRUE or FALSE
725  */  */
726    
727  static BOOL  static BOOL
728  is_counted_repeat(const uschar *p)  is_counted_repeat(const pcre_uchar *p)
729  {  {
730  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
731  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
732    while (IS_DIGIT(*p)) p++;
733  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
734    
735  if (*p++ != CHAR_COMMA) return FALSE;  if (*p++ != CHAR_COMMA) return FALSE;
736  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
737    
738  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
739  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
740    while (IS_DIGIT(*p)) p++;
741    
742  return (*p == CHAR_RIGHT_CURLY_BRACKET);  return (*p == CHAR_RIGHT_CURLY_BRACKET);
743  }  }
# Line 615  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 749  return (*p == CHAR_RIGHT_CURLY_BRACKET);
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, int *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 654  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 = 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 = 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;
812    
813    switch (c)    switch (c)
# Line 684  else Line 825  else
825        {        {
826        /* In JavaScript, \u must be followed by four hexadecimal numbers.        /* In JavaScript, \u must be followed by four hexadecimal numbers.
827        Otherwise it is a lowercase u letter. */        Otherwise it is a lowercase u letter. */
828        if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
829             && (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0)          && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
830            && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
831            && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
832          {          {
833          c = 0;          c = 0;
834          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
835            {            {
836            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
837  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
838            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
839            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 699  else Line 842  else
842            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
843  #endif  #endif
844            }            }
845    
846    #if defined COMPILE_PCRE8
847            if (c > (utf ? 0x10ffff : 0xff))
848    #elif defined COMPILE_PCRE16
849            if (c > (utf ? 0x10ffff : 0xffff))
850    #elif defined COMPILE_PCRE32
851            if (utf && c > 0x10ffff)
852    #endif
853              {
854              *errorcodeptr = ERR76;
855              }
856            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
857          }          }
858        }        }
859      else      else
# Line 725  else Line 880  else
880      (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
881      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
882      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
883      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
884    
885      case CHAR_g:      case CHAR_g:
886      if (isclass) break;      if (isclass) break;
887      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
888        {        {
889        c = -ESC_g;        escape = ESC_g;
890        break;        break;
891        }        }
892    
# Line 739  else Line 894  else
894    
895      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
896        {        {
897        const uschar *p;        const pcre_uchar *p;
898        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
899          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
900        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
901          {          {
902          c = -ESC_k;          escape = ESC_k;
903          break;          break;
904          }          }
905        braced = TRUE;        braced = TRUE;
# Line 759  else Line 914  else
914        }        }
915      else negated = FALSE;      else negated = FALSE;
916    
917        /* The integer range is limited by the machine's int representation. */
918      c = 0;      c = 0;
919      while ((digitab[ptr[1]] & ctype_digit) != 0)      while (IS_DIGIT(ptr[1]))
920          {
921          if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */
922            {
923            c = -1;
924            break;
925            }
926        c = c * 10 + *(++ptr) - CHAR_0;        c = c * 10 + *(++ptr) - CHAR_0;
927          }
928      if (c < 0)   /* Integer overflow */      if (((unsigned int)c) > INT_MAX) /* Integer overflow */
929        {        {
930          while (IS_DIGIT(ptr[1]))
931            ptr++;
932        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
933        break;        break;
934        }        }
# Line 791  else Line 955  else
955        c = bracount - (c - 1);        c = bracount - (c - 1);
956        }        }
957    
958      c = -(ESC_REF + c);      escape = -c;
959      break;      break;
960    
961      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 812  else Line 976  else
976      if (!isclass)      if (!isclass)
977        {        {
978        oldptr = ptr;        oldptr = ptr;
979          /* The integer range is limited by the machine's int representation. */
980        c -= CHAR_0;        c -= CHAR_0;
981        while ((digitab[ptr[1]] & ctype_digit) != 0)        while (IS_DIGIT(ptr[1]))
982            {
983            if (((unsigned int)c) > INT_MAX / 10) /* Integer overflow */
984              {
985              c = -1;
986              break;
987              }
988          c = c * 10 + *(++ptr) - CHAR_0;          c = c * 10 + *(++ptr) - CHAR_0;
989        if (c < 0)    /* Integer overflow */          }
990          if (((unsigned int)c) > INT_MAX) /* Integer overflow */
991          {          {
992            while (IS_DIGIT(ptr[1]))
993              ptr++;
994          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
995          break;          break;
996          }          }
997        if (c < 10 || c <= bracount)        if (c < 10 || c <= bracount)
998          {          {
999          c = -(ESC_REF + c);          escape = -c;
1000          break;          break;
1001          }          }
1002        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 842  else Line 1016  else
1016      /* \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
1017      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
1018      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
1019      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,
1020      than 3 octal digits. */      but no more than 3 octal digits. */
1021    
1022      case CHAR_0:      case CHAR_0:
1023      c -= CHAR_0;      c -= CHAR_0;
1024      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1025          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1026      if (!utf8 && c > 255) *errorcodeptr = ERR51;  #ifdef COMPILE_PCRE8
1027        if (!utf && c > 0xff) *errorcodeptr = ERR51;
1028    #endif
1029      break;      break;
1030    
1031      /* \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
1032      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.
1033      treated as a data character. */      If not, { is treated as a data character. */
1034    
1035      case CHAR_x:      case CHAR_x:
1036      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1037        {        {
1038        /* In JavaScript, \x must be followed by two hexadecimal numbers.        /* In JavaScript, \x must be followed by two hexadecimal numbers.
1039        Otherwise it is a lowercase x letter. */        Otherwise it is a lowercase x letter. */
1040        if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0)        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1041            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1042          {          {
1043          c = 0;          c = 0;
1044          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1045            {            {
1046            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1047  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1048            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1049            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 881  else Line 1058  else
1058    
1059      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1060        {        {
1061        const uschar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
1062        int count = 0;        BOOL overflow;
1063    
1064        c = 0;        c = 0;
1065        while ((digitab[*pt] & ctype_xdigit) != 0)        overflow = FALSE;
1066          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1067          {          {
1068          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1069          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1070          count++;  
1071    #ifdef COMPILE_PCRE32
1072            if (c >= 0x10000000l) { overflow = TRUE; break; }
1073    #endif
1074    
1075  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1076          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
# Line 898  else Line 1079  else
1079          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 */
1080          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1081  #endif  #endif
1082    
1083    #if defined COMPILE_PCRE8
1084            if (c > (utf ? 0x10ffff : 0xff)) { overflow = TRUE; break; }
1085    #elif defined COMPILE_PCRE16
1086            if (c > (utf ? 0x10ffff : 0xffff)) { overflow = TRUE; break; }
1087    #elif defined COMPILE_PCRE32
1088            if (utf && c > 0x10ffff) { overflow = TRUE; break; }
1089    #endif
1090            }
1091    
1092          if (overflow)
1093            {
1094            while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1095            *errorcodeptr = ERR34;
1096          }          }
1097    
1098        if (*pt == CHAR_RIGHT_CURLY_BRACKET)        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
1099          {          {
1100          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1101          ptr = pt;          ptr = pt;
1102          break;          break;
1103          }          }
# Line 914  else Line 1109  else
1109      /* Read just a single-byte hex-defined char */      /* Read just a single-byte hex-defined char */
1110    
1111      c = 0;      c = 0;
1112      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1113        {        {
1114        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1115        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1116  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1117        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 975  else Line 1170  else
1170  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1171  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1172    
1173  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1174       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1175    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1176    
1177  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1178    
1179  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1180    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1181    
1182  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1183    
1184  *ptrptr = ptr;  *ptrptr = ptr;
1185  return c;  *chptr = c;
1186    return escape;
1187  }  }
1188    
   
   
1189  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1190  /*************************************************  /*************************************************
1191  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1012  Returns:         type value from ucp_typ Line 1206  Returns:         type value from ucp_typ
1206  */  */
1207    
1208  static int  static int
1209  get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)
1210  {  {
1211  int c, i, bot, top;  int c, i, bot, top;
1212  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
1213  char name[32];  pcre_uchar name[32];
1214    
1215  c = *(++ptr);  c = *(++ptr);
1216  if (c == 0) goto ERROR_RETURN;  if (c == 0) goto ERROR_RETURN;
# Line 1033  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1227  if (c == CHAR_LEFT_CURLY_BRACKET)
1227      *negptr = TRUE;      *negptr = TRUE;
1228      ptr++;      ptr++;
1229      }      }
1230    for (i = 0; i < (int)sizeof(name) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1231      {      {
1232      c = *(++ptr);      c = *(++ptr);
1233      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 1057  else Line 1251  else
1251  /* Search for a recognized property name using binary chop */  /* Search for a recognized property name using binary chop */
1252    
1253  bot = 0;  bot = 0;
1254  top = _pcre_utt_size;  top = PRIV(utt_size);
1255    
1256  while (bot < top)  while (bot < top)
1257    {    {
1258    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1259    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);    c = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1260    if (c == 0)    if (c == 0)
1261      {      {
1262      *dptr = _pcre_utt[i].value;      *dptr = PRIV(utt)[i].value;
1263      return _pcre_utt[i].type;      return PRIV(utt)[i].type;
1264      }      }
1265    if (c > 0) bot = i + 1; else top = i;    if (c > 0) bot = i + 1; else top = i;
1266    }    }
# Line 1104  Returns:         pointer to '}' on succe Line 1298  Returns:         pointer to '}' on succe
1298                   current ptr on error, with errorcodeptr set non-zero                   current ptr on error, with errorcodeptr set non-zero
1299  */  */
1300    
1301  static const uschar *  static const pcre_uchar *
1302  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)
1303  {  {
1304  int min = 0;  int min = 0;
1305  int max = -1;  int max = -1;
# Line 1113  int max = -1; Line 1307  int max = -1;
1307  /* 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
1308  an integer overflow. */  an integer overflow. */
1309    
1310  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + *p++ - CHAR_0;
1311  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1312    {    {
1313    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1128  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1322  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1322    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1323      {      {
1324      max = 0;      max = 0;
1325      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + *p++ - CHAR_0;
1326      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1327        {        {
1328        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1183  Arguments: Line 1377  Arguments:
1377    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1378    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1379    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1380    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1381    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1382    
1383  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1384  */  */
1385    
1386  static int  static int
1387  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,
1388    BOOL xmode, BOOL utf8, int *count)    BOOL xmode, BOOL utf, int *count)
1389  {  {
1390  uschar *ptr = *ptrptr;  pcre_uchar *ptr = *ptrptr;
1391  int start_count = *count;  int start_count = *count;
1392  int hwm_count = start_count;  int hwm_count = start_count;
1393  BOOL dup_parens = FALSE;  BOOL dup_parens = FALSE;
# Line 1260  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1454  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1454          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1455        {        {
1456        int term;        int term;
1457        const uschar *thisname;        const pcre_uchar *thisname;
1458        *count += 1;        *count += 1;
1459        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1460        term = *ptr++;        term = *ptr++;
# Line 1268  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1462  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1462        thisname = ptr;        thisname = ptr;
1463        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1464        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == ptr - thisname &&
1465            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, lorn) == 0)
1466          return *count;          return *count;
1467        term++;        term++;
1468        }        }
# Line 1311  for (; ptr < cd->end_pattern; ptr++) Line 1505  for (; ptr < cd->end_pattern; ptr++)
1505          {          {
1506          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1507            ptr+= 2;            ptr+= 2;
1508          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1509                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1510            ptr += 4;            ptr += 4;
1511          else          else
# Line 1359  for (; ptr < cd->end_pattern; ptr++) Line 1553  for (; ptr < cd->end_pattern; ptr++)
1553        {        {
1554        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1555        ptr++;        ptr++;
1556  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1557        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1558  #endif  #endif
1559        }        }
1560      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
# Line 1371  for (; ptr < cd->end_pattern; ptr++) Line 1565  for (; ptr < cd->end_pattern; ptr++)
1565    
1566    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1567      {      {
1568      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1569      if (rc > 0) return rc;      if (rc > 0) return rc;
1570      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1571      }      }
# Line 1417  Arguments: Line 1611  Arguments:
1611    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1612    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1613    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1614    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1615    
1616  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1617  */  */
1618    
1619  static int  static int
1620  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,
1621    BOOL utf8)    BOOL utf)
1622  {  {
1623  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1624  int count = 0;  int count = 0;
1625  int rc;  int rc;
1626    
# Line 1437  matching closing parens. That is why we Line 1631  matching closing parens. That is why we
1631    
1632  for (;;)  for (;;)
1633    {    {
1634    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1635    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == 0) break;
1636    }    }
1637    
# Line 1464  Arguments: Line 1658  Arguments:
1658  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1659  */  */
1660    
1661  static const uschar*  static const pcre_uchar*
1662  first_significant_code(const uschar *code, BOOL skipassert)  first_significant_code(const pcre_uchar *code, BOOL skipassert)
1663  {  {
1664  for (;;)  for (;;)
1665    {    {
# Line 1476  for (;;) Line 1670  for (;;)
1670      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1671      if (!skipassert) return code;      if (!skipassert) return code;
1672      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1673      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1674      break;      break;
1675    
1676      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1490  for (;;) Line 1684  for (;;)
1684      case OP_RREF:      case OP_RREF:
1685      case OP_NRREF:      case OP_NRREF:
1686      case OP_DEF:      case OP_DEF:
1687      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1688      break;      break;
1689    
1690      default:      default:
# Line 1520  and doing the check at the end; a flag s Line 1714  and doing the check at the end; a flag s
1714    
1715  Arguments:  Arguments:
1716    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1717    utf8     TRUE in UTF-8 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1718    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1719    cd       the "compile data" structure    cd       the "compile data" structure
1720    
1721  Returns:   the fixed length,  Returns:   the fixed length,
1722               or -1 if there is no fixed length,               or -1 if there is no fixed length,
1723               or -2 if \C was encountered               or -2 if \C was encountered (in UTF-8 mode only)
1724               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
1725                 or -4 if an unknown opcode was encountered (internal error)
1726  */  */
1727    
1728  static int  static int
1729  find_fixedlength(uschar *code, BOOL utf8, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1730  {  {
1731  int length = -1;  int length = -1;
1732    
1733  register int branchlength = 0;  register int branchlength = 0;
1734  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1735    
1736  /* 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
1737  branch, check the length against that of the other branches. */  branch, check the length against that of the other branches. */
# Line 1544  branch, check the length against that of Line 1739  branch, check the length against that of
1739  for (;;)  for (;;)
1740    {    {
1741    int d;    int d;
1742    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1743    register int op = *cc;    register int op = *cc;
1744    
1745    switch (op)    switch (op)
1746      {      {
1747      /* We only need to continue for OP_CBRA (normal capturing bracket) and      /* We only need to continue for OP_CBRA (normal capturing bracket) and
1748      OP_BRA (normal non-capturing bracket) because the other variants of these      OP_BRA (normal non-capturing bracket) because the other variants of these
1749      opcodes are all concerned with unlimited repeated groups, which of course      opcodes are all concerned with unlimited repeated groups, which of course
1750      are not of fixed length. They will cause a -1 response from the default      are not of fixed length. */
     case of this switch. */  
1751    
1752      case OP_CBRA:      case OP_CBRA:
1753      case OP_BRA:      case OP_BRA:
1754      case OP_ONCE:      case OP_ONCE:
1755      case OP_ONCE_NC:      case OP_ONCE_NC:
1756      case OP_COND:      case OP_COND:
1757      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd);      d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd);
1758      if (d < 0) return d;      if (d < 0) return d;
1759      branchlength += d;      branchlength += d;
1760      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1761      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
1762      break;      break;
1763    
1764      /* 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.
1765      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
1766      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
1767      Note that we must not include the OP_KETRxxx opcodes here, because they      the same code. Note that we must not include the OP_KETRxxx opcodes here,
1768      all imply an unlimited repeat. */      because they all imply an unlimited repeat. */
1769    
1770      case OP_ALT:      case OP_ALT:
1771      case OP_KET:      case OP_KET:
1772      case OP_END:      case OP_END:
1773        case OP_ACCEPT:
1774        case OP_ASSERT_ACCEPT:
1775      if (length < 0) length = branchlength;      if (length < 0) length = branchlength;
1776        else if (length != branchlength) return -1;        else if (length != branchlength) return -1;
1777      if (*cc != OP_ALT) return length;      if (*cc != OP_ALT) return length;
# Line 1588  for (;;) Line 1785  for (;;)
1785    
1786      case OP_RECURSE:      case OP_RECURSE:
1787      if (!atend) return -3;      if (!atend) return -3;
1788      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1789      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1790      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1791      d = find_fixedlength(cs + 2, utf8, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1792      if (d < 0) return d;      if (d < 0) return d;
1793      branchlength += d;      branchlength += d;
1794      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1604  for (;;) Line 1801  for (;;)
1801      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1802      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1803      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1804      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1805        break;
1806    
1807      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1808    
1809      case OP_REVERSE:      case OP_MARK:
1810      case OP_CREF:      case OP_PRUNE_ARG:
1811      case OP_NCREF:      case OP_SKIP_ARG:
1812      case OP_RREF:      case OP_THEN_ARG:
1813      case OP_NRREF:      cc += cc[1] + PRIV(OP_lengths)[*cc];
1814      case OP_DEF:      break;
1815    
1816      case OP_CALLOUT:      case OP_CALLOUT:
     case OP_SOD:  
     case OP_SOM:  
     case OP_SET_SOM:  
     case OP_EOD:  
     case OP_EODN:  
1817      case OP_CIRC:      case OP_CIRC:
1818      case OP_CIRCM:      case OP_CIRCM:
1819        case OP_CLOSE:
1820        case OP_COMMIT:
1821        case OP_CREF:
1822        case OP_DEF:
1823      case OP_DOLL:      case OP_DOLL:
1824      case OP_DOLLM:      case OP_DOLLM:
1825        case OP_EOD:
1826        case OP_EODN:
1827        case OP_FAIL:
1828        case OP_NCREF:
1829        case OP_NRREF:
1830      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1831        case OP_PRUNE:
1832        case OP_REVERSE:
1833        case OP_RREF:
1834        case OP_SET_SOM:
1835        case OP_SKIP:
1836        case OP_SOD:
1837        case OP_SOM:
1838        case OP_THEN:
1839      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1840      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1841      break;      break;
1842    
1843      /* Handle literal characters */      /* Handle literal characters */
# Line 1637  for (;;) Line 1848  for (;;)
1848      case OP_NOTI:      case OP_NOTI:
1849      branchlength++;      branchlength++;
1850      cc += 2;      cc += 2;
1851  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1852      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1853  #endif  #endif
1854      break;      break;
1855    
# Line 1646  for (;;) Line 1857  for (;;)
1857      need to skip over a multibyte character in UTF8 mode.  */      need to skip over a multibyte character in UTF8 mode.  */
1858    
1859      case OP_EXACT:      case OP_EXACT:
1860        case OP_EXACTI:
1861        case OP_NOTEXACT:
1862        case OP_NOTEXACTI:
1863      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1864      cc += 4;      cc += 2 + IMM2_SIZE;
1865  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
1866      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1867  #endif  #endif
1868      break;      break;
1869    
1870      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1871      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1872      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1873      cc += 4;        cc += 2;
1874        cc += 1 + IMM2_SIZE + 1;
1875      break;      break;
1876    
1877      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1666  for (;;) Line 1881  for (;;)
1881      cc += 2;      cc += 2;
1882      /* Fall through */      /* Fall through */
1883    
1884        case OP_HSPACE:
1885        case OP_VSPACE:
1886        case OP_NOT_HSPACE:
1887        case OP_NOT_VSPACE:
1888      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
1889      case OP_DIGIT:      case OP_DIGIT:
1890      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
# Line 1678  for (;;) Line 1897  for (;;)
1897      cc++;      cc++;
1898      break;      break;
1899    
1900      /* The single-byte matcher isn't allowed */      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;
1901        otherwise \C is coded as OP_ALLANY. */
1902    
1903      case OP_ANYBYTE:      case OP_ANYBYTE:
1904      return -2;      return -2;
1905    
1906      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1907    
1908  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1909      case OP_XCLASS:      case OP_XCLASS:
1910      cc += GET(cc, 1) - 33;      cc += GET(cc, 1) - PRIV(OP_lengths)[OP_CLASS];
1911      /* Fall through */      /* Fall through */
1912  #endif  #endif
1913    
1914      case OP_CLASS:      case OP_CLASS:
1915      case OP_NCLASS:      case OP_NCLASS:
1916      cc += 33;      cc += PRIV(OP_lengths)[OP_CLASS];
1917    
1918      switch (*cc)      switch (*cc)
1919        {        {
1920          case OP_CRPLUS:
1921          case OP_CRMINPLUS:
1922        case OP_CRSTAR:        case OP_CRSTAR:
1923        case OP_CRMINSTAR:        case OP_CRMINSTAR:
1924        case OP_CRQUERY:        case OP_CRQUERY:
# Line 1705  for (;;) Line 1927  for (;;)
1927    
1928        case OP_CRRANGE:        case OP_CRRANGE:
1929        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1930        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1931        branchlength += GET2(cc,1);        branchlength += GET2(cc,1);
1932        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1933        break;        break;
1934    
1935        default:        default:
# Line 1717  for (;;) Line 1939  for (;;)
1939    
1940      /* Anything else is variable length */      /* Anything else is variable length */
1941    
1942      default:      case OP_ANYNL:
1943        case OP_BRAMINZERO:
1944        case OP_BRAPOS:
1945        case OP_BRAPOSZERO:
1946        case OP_BRAZERO:
1947        case OP_CBRAPOS:
1948        case OP_EXTUNI:
1949        case OP_KETRMAX:
1950        case OP_KETRMIN:
1951        case OP_KETRPOS:
1952        case OP_MINPLUS:
1953        case OP_MINPLUSI:
1954        case OP_MINQUERY:
1955        case OP_MINQUERYI:
1956        case OP_MINSTAR:
1957        case OP_MINSTARI:
1958        case OP_MINUPTO:
1959        case OP_MINUPTOI:
1960        case OP_NOTMINPLUS:
1961        case OP_NOTMINPLUSI:
1962        case OP_NOTMINQUERY:
1963        case OP_NOTMINQUERYI:
1964        case OP_NOTMINSTAR:
1965        case OP_NOTMINSTARI:
1966        case OP_NOTMINUPTO:
1967        case OP_NOTMINUPTOI:
1968        case OP_NOTPLUS:
1969        case OP_NOTPLUSI:
1970        case OP_NOTPOSPLUS:
1971        case OP_NOTPOSPLUSI:
1972        case OP_NOTPOSQUERY:
1973        case OP_NOTPOSQUERYI:
1974        case OP_NOTPOSSTAR:
1975        case OP_NOTPOSSTARI:
1976        case OP_NOTPOSUPTO:
1977        case OP_NOTPOSUPTOI:
1978        case OP_NOTQUERY:
1979        case OP_NOTQUERYI:
1980        case OP_NOTSTAR:
1981        case OP_NOTSTARI:
1982        case OP_NOTUPTO:
1983        case OP_NOTUPTOI:
1984        case OP_PLUS:
1985        case OP_PLUSI:
1986        case OP_POSPLUS:
1987        case OP_POSPLUSI:
1988        case OP_POSQUERY:
1989        case OP_POSQUERYI:
1990        case OP_POSSTAR:
1991        case OP_POSSTARI:
1992        case OP_POSUPTO:
1993        case OP_POSUPTOI:
1994        case OP_QUERY:
1995        case OP_QUERYI:
1996        case OP_REF:
1997        case OP_REFI:
1998        case OP_SBRA:
1999        case OP_SBRAPOS:
2000        case OP_SCBRA:
2001        case OP_SCBRAPOS:
2002        case OP_SCOND:
2003        case OP_SKIPZERO:
2004        case OP_STAR:
2005        case OP_STARI:
2006        case OP_TYPEMINPLUS:
2007        case OP_TYPEMINQUERY:
2008        case OP_TYPEMINSTAR:
2009        case OP_TYPEMINUPTO:
2010        case OP_TYPEPLUS:
2011        case OP_TYPEPOSPLUS:
2012        case OP_TYPEPOSQUERY:
2013        case OP_TYPEPOSSTAR:
2014        case OP_TYPEPOSUPTO:
2015        case OP_TYPEQUERY:
2016        case OP_TYPESTAR:
2017        case OP_TYPEUPTO:
2018        case OP_UPTO:
2019        case OP_UPTOI:
2020      return -1;      return -1;
2021    
2022        /* Catch unrecognized opcodes so that when new ones are added they
2023        are not forgotten, as has happened in the past. */
2024    
2025        default:
2026        return -4;
2027      }      }
2028    }    }
2029  /* Control never gets here */  /* Control never gets here */
# Line 1739  length. Line 2044  length.
2044    
2045  Arguments:  Arguments:
2046    code        points to start of expression    code        points to start of expression
2047    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2048    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2049    
2050  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
2051  */  */
2052    
2053  const uschar *  const pcre_uchar *
2054  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2055  {  {
2056  for (;;)  for (;;)
2057    {    {
# Line 1764  for (;;) Line 2069  for (;;)
2069    
2070    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2071      {      {
2072      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2073      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2074      }      }
2075    
2076    /* Handle capturing bracket */    /* Handle capturing bracket */
# Line 1774  for (;;) Line 2079  for (;;)
2079             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2080      {      {
2081      int n = GET2(code, 1+LINK_SIZE);      int n = GET2(code, 1+LINK_SIZE);
2082      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2083      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2084      }      }
2085    
2086    /* 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 1803  for (;;) Line 2108  for (;;)
2108        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2109        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2110        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2111        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2112            code += 2;
2113        break;        break;
2114    
2115        case OP_MARK:        case OP_MARK:
# Line 1819  for (;;) Line 2125  for (;;)
2125    
2126      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2127    
2128      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2129    
2130    /* 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
2131    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
2132    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2133    
2134  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2135      if (utf8) switch(c)      if (utf) switch(c)
2136        {        {
2137        case OP_CHAR:        case OP_CHAR:
2138        case OP_CHARI:        case OP_CHARI:
# Line 1856  for (;;) Line 2162  for (;;)
2162        case OP_MINQUERYI:        case OP_MINQUERYI:
2163        case OP_POSQUERY:        case OP_POSQUERY:
2164        case OP_POSQUERYI:        case OP_POSQUERYI:
2165        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2166        break;        break;
2167        }        }
2168  #else  #else
2169      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2170  #endif  #endif
2171      }      }
2172    }    }
# Line 1877  instance of OP_RECURSE. Line 2183  instance of OP_RECURSE.
2183    
2184  Arguments:  Arguments:
2185    code        points to start of expression    code        points to start of expression
2186    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2187    
2188  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
2189  */  */
2190    
2191  static const uschar *  static const pcre_uchar *
2192  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2193  {  {
2194  for (;;)  for (;;)
2195    {    {
# Line 1922  for (;;) Line 2228  for (;;)
2228        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2229        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2230        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2231        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2232            code += 2;
2233        break;        break;
2234    
2235        case OP_MARK:        case OP_MARK:
# Line 1938  for (;;) Line 2245  for (;;)
2245    
2246      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2247    
2248      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2249    
2250      /* 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
2251      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
2252      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2253    
2254  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2255      if (utf8) switch(c)      if (utf) switch(c)
2256        {        {
2257        case OP_CHAR:        case OP_CHAR:
2258        case OP_CHARI:        case OP_CHARI:
2259          case OP_NOT:
2260          case OP_NOTI:
2261        case OP_EXACT:        case OP_EXACT:
2262        case OP_EXACTI:        case OP_EXACTI:
2263          case OP_NOTEXACT:
2264          case OP_NOTEXACTI:
2265        case OP_UPTO:        case OP_UPTO:
2266        case OP_UPTOI:        case OP_UPTOI:
2267          case OP_NOTUPTO:
2268          case OP_NOTUPTOI:
2269        case OP_MINUPTO:        case OP_MINUPTO:
2270        case OP_MINUPTOI:        case OP_MINUPTOI:
2271          case OP_NOTMINUPTO:
2272          case OP_NOTMINUPTOI:
2273        case OP_POSUPTO:        case OP_POSUPTO:
2274        case OP_POSUPTOI:        case OP_POSUPTOI:
2275          case OP_NOTPOSUPTO:
2276          case OP_NOTPOSUPTOI:
2277        case OP_STAR:        case OP_STAR:
2278        case OP_STARI:        case OP_STARI:
2279          case OP_NOTSTAR:
2280          case OP_NOTSTARI:
2281        case OP_MINSTAR:        case OP_MINSTAR:
2282        case OP_MINSTARI:        case OP_MINSTARI:
2283          case OP_NOTMINSTAR:
2284          case OP_NOTMINSTARI:
2285        case OP_POSSTAR:        case OP_POSSTAR:
2286        case OP_POSSTARI:        case OP_POSSTARI:
2287          case OP_NOTPOSSTAR:
2288          case OP_NOTPOSSTARI:
2289        case OP_PLUS:        case OP_PLUS:
2290        case OP_PLUSI:        case OP_PLUSI:
2291          case OP_NOTPLUS:
2292          case OP_NOTPLUSI:
2293        case OP_MINPLUS:        case OP_MINPLUS:
2294        case OP_MINPLUSI:        case OP_MINPLUSI:
2295          case OP_NOTMINPLUS:
2296          case OP_NOTMINPLUSI:
2297        case OP_POSPLUS:        case OP_POSPLUS:
2298        case OP_POSPLUSI:        case OP_POSPLUSI:
2299          case OP_NOTPOSPLUS:
2300          case OP_NOTPOSPLUSI:
2301        case OP_QUERY:        case OP_QUERY:
2302        case OP_QUERYI:        case OP_QUERYI:
2303          case OP_NOTQUERY:
2304          case OP_NOTQUERYI:
2305        case OP_MINQUERY:        case OP_MINQUERY:
2306        case OP_MINQUERYI:        case OP_MINQUERYI:
2307          case OP_NOTMINQUERY:
2308          case OP_NOTMINQUERYI:
2309        case OP_POSQUERY:        case OP_POSQUERY:
2310        case OP_POSQUERYI:        case OP_POSQUERYI:
2311        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_NOTPOSQUERY:
2312          case OP_NOTPOSQUERYI:
2313          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2314        break;        break;
2315        }        }
2316  #else  #else
2317      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2318  #endif  #endif
2319      }      }
2320    }    }
# Line 2002  bracket whose current branch will alread Line 2337  bracket whose current branch will alread
2337  Arguments:  Arguments:
2338    code        points to start of search    code        points to start of search
2339    endcode     points to where to stop    endcode     points to where to stop
2340    utf8        TRUE if in UTF8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2341    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2342    
2343  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2344  */  */
2345    
2346  static BOOL  static BOOL
2347  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2348    compile_data *cd)    BOOL utf, compile_data *cd)
2349  {  {
2350  register int c;  register int c;
2351  for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2352       code < endcode;       code < endcode;
2353       code = first_significant_code(code + _pcre_OP_lengths[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2354    {    {
2355    const uschar *ccode;    const pcre_uchar *ccode;
2356    
2357    c = *code;    c = *code;
2358    
# Line 2040  for (code = first_significant_code(code Line 2375  for (code = first_significant_code(code
2375    
2376    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2377      {      {
2378      const uschar *scode;      const pcre_uchar *scode;
2379      BOOL empty_branch;      BOOL empty_branch;
2380    
2381      /* Test for forward reference */      /* Test for forward reference */
# Line 2058  for (code = first_significant_code(code Line 2393  for (code = first_significant_code(code
2393    
2394      do      do
2395        {        {
2396        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd))
2397          {          {
2398          empty_branch = TRUE;          empty_branch = TRUE;
2399          break;          break;
# Line 2076  for (code = first_significant_code(code Line 2411  for (code = first_significant_code(code
2411    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2412        c == OP_BRAPOSZERO)        c == OP_BRAPOSZERO)
2413      {      {
2414      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2415      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
2416      c = *code;      c = *code;
2417      continue;      continue;
# Line 2114  for (code = first_significant_code(code Line 2449  for (code = first_significant_code(code
2449        empty_branch = FALSE;        empty_branch = FALSE;
2450        do        do
2451          {          {
2452          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
2453            empty_branch = TRUE;            empty_branch = TRUE;
2454          code += GET(code, 1);          code += GET(code, 1);
2455          }          }
# Line 2132  for (code = first_significant_code(code Line 2467  for (code = first_significant_code(code
2467      {      {
2468      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2469      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2470      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2471      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"
2472      here. */      here. */
2473    
2474  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2475      case OP_XCLASS:      case OP_XCLASS:
2476      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2477      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2144  for (code = first_significant_code(code Line 2479  for (code = first_significant_code(code
2479    
2480      case OP_CLASS:      case OP_CLASS:
2481      case OP_NCLASS:      case OP_NCLASS:
2482      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2483    
2484  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2485      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2486  #endif  #endif
2487    
# Line 2219  for (code = first_significant_code(code Line 2554  for (code = first_significant_code(code
2554      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2555      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2556      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2557      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2558          code += 2;
2559      break;      break;
2560    
2561      /* End of branch */      /* End of branch */
# Line 2234  for (code = first_significant_code(code Line 2570  for (code = first_significant_code(code
2570      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2571      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2572    
2573  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2574      case OP_STAR:      case OP_STAR:
2575      case OP_STARI:      case OP_STARI:
2576      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2247  for (code = first_significant_code(code Line 2583  for (code = first_significant_code(code
2583      case OP_MINQUERYI:      case OP_MINQUERYI:
2584      case OP_POSQUERY:      case OP_POSQUERY:
2585      case OP_POSQUERYI:      case OP_POSQUERYI:
2586      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2587      break;      break;
2588    
2589      case OP_UPTO:      case OP_UPTO:
# Line 2256  for (code = first_significant_code(code Line 2592  for (code = first_significant_code(code
2592      case OP_MINUPTOI:      case OP_MINUPTOI:
2593      case OP_POSUPTO:      case OP_POSUPTO:
2594      case OP_POSUPTOI:      case OP_POSUPTOI:
2595      if (utf8 && code[3] >= 0xc0) code += _pcre_utf8_table4[code[3] & 0x3f];      if (utf && HAS_EXTRALEN(code[1 + IMM2_SIZE])) code += GET_EXTRALEN(code[1 + IMM2_SIZE]);
2596      break;      break;
2597  #endif  #endif
2598    
# Line 2300  Arguments: Line 2636  Arguments:
2636    code        points to start of the recursion    code        points to start of the recursion
2637    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2638    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2639    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2640    cd          pointers to tables etc    cd          pointers to tables etc
2641    
2642  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2643  */  */
2644    
2645  static BOOL  static BOOL
2646  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2647    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2648  {  {
2649  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2650    {    {
2651    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
2652      return FALSE;      return FALSE;
2653    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2654    }    }
# Line 2364  Returns:   TRUE or FALSE Line 2700  Returns:   TRUE or FALSE
2700  */  */
2701    
2702  static BOOL  static BOOL
2703  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2704  {  {
2705  int terminator;          /* Don't combine these lines; the Solaris cc */  int terminator;          /* Don't combine these lines; the Solaris cc */
2706  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
# Line 2408  Returns:     a value representing the na Line 2744  Returns:     a value representing the na
2744  */  */
2745    
2746  static int  static int
2747  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2748  {  {
2749  const char *pn = posix_names;  const char *pn = posix_names;
2750  register int yield = 0;  register int yield = 0;
2751  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2752    {    {
2753    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2754      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, len) == 0) return yield;
2755    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2756    yield++;    yield++;
2757    }    }
# Line 2447  value in the reference (which is a group Line 2783  value in the reference (which is a group
2783  Arguments:  Arguments:
2784    group      points to the start of the group    group      points to the start of the group
2785    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2786    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2787    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2788    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
2789    
# Line 2455  Returns:     nothing Line 2791  Returns:     nothing
2791  */  */
2792    
2793  static void  static void
2794  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2795    uschar *save_hwm)    pcre_uchar *save_hwm)
2796  {  {
2797  uschar *ptr = group;  pcre_uchar *ptr = group;
2798    
2799  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2800    {    {
2801    int offset;    int offset;
2802    uschar *hc;    pcre_uchar *hc;
2803    
2804    /* 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
2805    reference. */    reference. */
# Line 2508  Arguments: Line 2844  Arguments:
2844  Returns:         new code pointer  Returns:         new code pointer
2845  */  */
2846    
2847  static uschar *  static pcre_uchar *
2848  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2849  {  {
2850  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2851  *code++ = 255;  *code++ = 255;
2852  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2853  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2854  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2855  }  }
2856    
2857    
# Line 2537  Returns:             nothing Line 2873  Returns:             nothing
2873  */  */
2874    
2875  static void  static void
2876  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2877  {  {
2878  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2879  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2551  PUT(previous_callout, 2 + LINK_SIZE, len Line 2887  PUT(previous_callout, 2 + LINK_SIZE, len
2887  *************************************************/  *************************************************/
2888    
2889  /* 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
2890  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
2891  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
2892  start address.  start address. A character with multiple other cases is returned on its own
2893    with a special return value.
2894    
2895  Arguments:  Arguments:
2896    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2561  Arguments: Line 2898  Arguments:
2898    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2899    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2900    
2901  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2902                   0 when a range is returned
2903                  >0 the CASESET offset for char with multiple other cases
2904                    in this case, ocptr contains the original
2905  */  */
2906    
2907  static BOOL  static int
2908  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,
2909    unsigned int *odptr)    unsigned int *odptr)
2910  {  {
2911  unsigned int c, othercase, next;  unsigned int c, othercase, next;
2912    int co;
2913    
2914    /* Find the first character that has an other case. If it has multiple other
2915    cases, return its case offset value. */
2916    
2917  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2918    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2919      if ((co = UCD_CASESET(c)) != 0)
2920        {
2921        *ocptr = c++;   /* Character that has the set */
2922        *cptr = c;      /* Rest of input range */
2923        return co;
2924        }
2925      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2926      }
2927    
2928  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2929    
2930  *ocptr = othercase;  *ocptr = othercase;
2931  next = othercase + 1;  next = othercase + 1;
# Line 2584  for (++c; c <= d; c++) Line 2936  for (++c; c <= d; c++)
2936    next++;    next++;
2937    }    }
2938    
2939  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2940  *cptr = c;  *cptr = c;             /* Rest of input range */
2941    return 0;
 return TRUE;  
2942  }  }
2943    
2944    
# Line 2611  Returns:       TRUE if auto-possessifyin Line 2962  Returns:       TRUE if auto-possessifyin
2962  static BOOL  static BOOL
2963  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(int c, int ptype, int pdata, BOOL negated)
2964  {  {
2965    #ifdef SUPPORT_UCP
2966    const pcre_uint32 *p;
2967    #endif
2968    
2969  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2970    
2971  switch(ptype)  switch(ptype)
2972    {    {
2973    case PT_LAMP:    case PT_LAMP:
# Line 2620  switch(ptype) Line 2976  switch(ptype)
2976            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
2977    
2978    case PT_GC:    case PT_GC:
2979    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2980    
2981    case PT_PC:    case PT_PC:
2982    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2631  switch(ptype) Line 2987  switch(ptype)
2987    /* These are specials */    /* These are specials */
2988    
2989    case PT_ALNUM:    case PT_ALNUM:
2990    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
2991            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
2992    
2993    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
2994    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
2995            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
2996            == negated;            == negated;
2997    
2998    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
2999    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3000            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
3001            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
3002            == negated;            == negated;
3003    
3004    case PT_WORD:    case PT_WORD:
3005    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3006            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3007            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3008    
3009    #ifdef SUPPORT_UCP
3010      case PT_CLIST:
3011      p = PRIV(ucd_caseless_sets) + prop->caseset;
3012      for (;;)
3013        {
3014        if ((unsigned int)c < *p) return !negated;
3015        if ((unsigned int)c == *p++) return negated;
3016        }
3017      break;  /* Control never reaches here */
3018    #endif
3019    }    }
3020    
3021  return FALSE;  return FALSE;
3022  }  }
3023  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2666  sense to automatically possessify the re Line 3034  sense to automatically possessify the re
3034    
3035  Arguments:  Arguments:
3036    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3037    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3038    ptr           next character in pattern    ptr           next character in pattern
3039    options       options bits    options       options bits
3040    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2675  Returns:        TRUE if possessifying is Line 3043  Returns:        TRUE if possessifying is
3043  */  */
3044    
3045  static BOOL  static BOOL
3046  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3047    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3048  {  {
3049  int c, next;  pcre_int32 c = NOTACHAR; // FIXMEchpe pcre_uint32
3050    pcre_int32 next;
3051    int escape;
3052  int op_code = *previous++;  int op_code = *previous++;
3053    
3054  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2687  if ((options & PCRE_EXTENDED) != 0) Line 3057  if ((options & PCRE_EXTENDED) != 0)
3057    {    {
3058    for (;;)    for (;;)
3059      {      {
3060      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3061      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3062        {        {
3063        ptr++;        ptr++;
# Line 2695  if ((options & PCRE_EXTENDED) != 0) Line 3065  if ((options & PCRE_EXTENDED) != 0)
3065          {          {
3066          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3067          ptr++;          ptr++;
3068  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3069          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3070  #endif  #endif
3071          }          }
3072        }        }
# Line 2710  value is a character, a negative value i Line 3080  value is a character, a negative value i
3080  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3081    {    {
3082    int temperrorcode = 0;    int temperrorcode = 0;
3083    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3084    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3085    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3086    }    }
3087    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
3088    {    {
3089  #ifdef SUPPORT_UTF8    escape = 0;
3090    if (utf8) { GETCHARINC(next, ptr); } else  #ifdef SUPPORT_UTF
3091      if (utf) { GETCHARINC(next, ptr); } else
3092  #endif  #endif
3093    next = *ptr++;    next = *ptr++;
3094    }    }
   
3095  else return FALSE;  else return FALSE;
3096    
3097  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2731  if ((options & PCRE_EXTENDED) != 0) Line 3100  if ((options & PCRE_EXTENDED) != 0)
3100    {    {
3101    for (;;)    for (;;)
3102      {      {
3103      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3104      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3105        {        {
3106        ptr++;        ptr++;
# Line 2739  if ((options & PCRE_EXTENDED) != 0) Line 3108  if ((options & PCRE_EXTENDED) != 0)
3108          {          {
3109          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3110          ptr++;          ptr++;
3111  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3112          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3113  #endif  #endif
3114          }          }
3115        }        }
# Line 2751  if ((options & PCRE_EXTENDED) != 0) Line 3120  if ((options & PCRE_EXTENDED) != 0)
3120  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3121    
3122  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3123    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)
3124      return FALSE;      return FALSE;
3125    
3126  /* 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. */  
3127    
3128  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3129        op_code == OP_NOT || op_code == OP_NOTI)
3130      //if (escape == 0) switch(op_code)
3131    {    {
3132    case OP_CHAR:  #ifdef SUPPORT_UTF
 #ifdef SUPPORT_UTF8  
3133    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3134  #else  #else
3135    c = *previous;    c = *previous;
3136  #endif  #endif
3137    return c != next;    }
3138    
3139    /* For CHARI (caseless character) we must check the other case. If we have  /* Now compare the next item with the previous opcode. First, handle cases when
3140    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3141    
3142    case OP_CHARI:  if (escape == 0)
3143  #ifdef SUPPORT_UTF8    {
3144    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3145  #else    case, which maps to the special PT_CLIST property. Check this first. */
3146    c = *previous;  
3147    #ifdef SUPPORT_UCP
3148      if (utf && (unsigned int)c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3149        {
3150        int ocs = UCD_CASESET(next);
3151        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3152        }
3153  #endif  #endif
3154    if (c == next) return FALSE;  
3155  #ifdef SUPPORT_UTF8    switch(op_code)
   if (utf8)  
3156      {      {
3157      unsigned int othercase;      case OP_CHAR:
3158      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3159    
3160        /* For CHARI (caseless character) we must check the other case. If we have
3161        Unicode property support, we can use it to test the other case of
3162        high-valued characters. We know that next can have only one other case,
3163        because multi-other-case characters are dealt with above. */
3164    
3165        case OP_CHARI:
3166        if (c == next) return FALSE;
3167    #ifdef SUPPORT_UTF
3168        if (utf)
3169          {
3170          unsigned int othercase;
3171          if (next < 128) othercase = cd->fcc[next]; else
3172  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3173      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE((unsigned int)next);
3174  #else  #else
3175      othercase = NOTACHAR;        othercase = NOTACHAR;
3176  #endif  #endif
3177      return (unsigned int)c != othercase;        return (unsigned int)c != othercase;
3178      }        }
3179    else      else
3180  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3181    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */      return (c != TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3182    
3183    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These      case OP_NOT:
3184    opcodes are not used for multi-byte characters, because they are coded using      return c == next;
3185    an XCLASS instead. */  
3186        case OP_NOTI:
3187    case OP_NOT:      if (c == next) return TRUE;
3188    return (c = *previous) == next;  #ifdef SUPPORT_UTF
3189        if (utf)
3190    case OP_NOTI:        {
3191    if ((c = *previous) == next) return TRUE;        unsigned int othercase;
3192  #ifdef SUPPORT_UTF8        if (next < 128) othercase = cd->fcc[next]; else
   if (utf8)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3193  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3194      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE((unsigned int)next);
3195  #else  #else
3196      othercase = NOTACHAR;        othercase = NOTACHAR;
3197  #endif  #endif
3198      return (unsigned int)c == othercase;        return (unsigned int)c == othercase;
3199      }        }
3200    else      else
3201  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3202    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */      return (c == TABLE_GET((unsigned int)next, cd->fcc, next));  /* Not UTF */
3203    
3204    /* 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.
3205    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. */
3206    
3207    case OP_DIGIT:      case OP_DIGIT:
3208    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3209    
3210    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3211    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3212    
3213    case OP_WHITESPACE:      case OP_WHITESPACE:
3214    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3215    
3216    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3217    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3218    
3219    case OP_WORDCHAR:      case OP_WORDCHAR:
3220    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3221    
3222    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3223    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3224    
3225    case OP_HSPACE:      case OP_HSPACE:
3226    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3227    switch(next)      switch(next)
3228      {        {
3229      case 0x09:        HSPACE_CASES:
3230      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;  
     }  
3231    
3232    case OP_ANYNL:        default:
3233    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3234    case OP_NOT_VSPACE:        }
3235    switch(next)  
3236      {      case OP_ANYNL:
3237      case 0x0a:      case OP_VSPACE:
3238      case 0x0b:      case OP_NOT_VSPACE:
3239      case 0x0c:      switch(next)
3240      case 0x0d:        {
3241      case 0x85:        VSPACE_CASES:
3242      case 0x2028:        return op_code == OP_NOT_VSPACE;
3243      case 0x2029:  
3244      return op_code == OP_NOT_VSPACE;        default:
3245      default:        return op_code != OP_NOT_VSPACE;
3246      return op_code != OP_NOT_VSPACE;        }
     }  
3247    
3248  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3249    case OP_PROP:      case OP_PROP:
3250    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3251    
3252    case OP_NOTPROP:      case OP_NOTPROP:
3253    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3254  #endif  #endif
3255    
3256    default:      default:
3257    return FALSE;      return FALSE;
3258        }
3259    }    }
3260    
   
3261  /* 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
3262  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
3263  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 2908  switch(op_code) Line 3268  switch(op_code)
3268    {    {
3269    case OP_CHAR:    case OP_CHAR:
3270    case OP_CHARI:    case OP_CHARI:
3271  #ifdef SUPPORT_UTF8    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3272      {      {
3273      case ESC_d:      case ESC_d:
3274      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3275    
3276      case ESC_D:      case ESC_D:
3277      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3278    
3279      case ESC_s:      case ESC_s:
3280      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3281    
3282      case ESC_S:      case ESC_S:
3283      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3284    
3285      case ESC_w:      case ESC_w:
3286      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3287    
3288      case ESC_W:      case ESC_W:
3289      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3290    
3291      case ESC_h:      case ESC_h:
3292      case ESC_H:      case ESC_H:
3293      switch(c)      switch(c)
3294        {        {
3295        case 0x09:        HSPACE_CASES:
3296        case 0x20:        return escape != ESC_h;
3297        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;  
3298        default:        default:
3299        return -next == ESC_h;        return escape == ESC_h;
3300        }        }
3301    
3302      case ESC_v:      case ESC_v:
3303      case ESC_V:      case ESC_V:
3304      switch(c)      switch(c)
3305        {        {
3306        case 0x0a:        VSPACE_CASES:
3307        case 0x0b:        return escape != ESC_v;
3308        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3309        default:        default:
3310        return -next == ESC_v;        return escape == ESC_v;
3311        }        }
3312    
3313      /* 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
3314      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3315      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3316    
3317  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3318      case ESC_du:      case ESC_du:
# Line 2990  switch(op_code) Line 3323  switch(op_code)
3323      case ESC_SU:      case ESC_SU:
3324        {        {
3325        int temperrorcode = 0;        int temperrorcode = 0;
3326        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3327        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3328        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3329        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3330        }        }
# Line 3013  switch(op_code) Line 3346  switch(op_code)
3346        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. */
3347    
3348        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3349          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)
3350            return FALSE;            return FALSE;
3351    
3352        /* Do the property check. */        /* Do the property check. */
3353    
3354        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3355        }        }
3356  #endif  #endif
3357    
# Line 3033  switch(op_code) Line 3366  switch(op_code)
3366    these op-codes are never generated.) */    these op-codes are never generated.) */
3367    
3368    case OP_DIGIT:    case OP_DIGIT:
3369    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3370           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3371    
3372    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3373    return next == -ESC_d;    return escape == ESC_d;
3374    
3375    case OP_WHITESPACE:    case OP_WHITESPACE:
3376    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3377    
3378    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3379    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3380    
3381    case OP_HSPACE:    case OP_HSPACE:
3382    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3383           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3384    
3385    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3386    return next == -ESC_h;    return escape == ESC_h;
3387    
3388    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3389    case OP_ANYNL:    case OP_ANYNL:
3390    case OP_VSPACE:    case OP_VSPACE:
3391    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3392    
3393    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3394    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3395    
3396    case OP_WORDCHAR:    case OP_WORDCHAR:
3397    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3398           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3399    
3400      case OP_NOT_WORDCHAR:
3401      return escape == ESC_w || escape == ESC_d;
3402    
3403      default:
3404      return FALSE;
3405      }
3406    
3407    /* Control does not reach here */
3408    }
3409    
3410    
3411    
3412    /*************************************************
3413    *        Add a character or range to a class     *
3414    *************************************************/
3415    
3416    /* This function packages up the logic of adding a character or range of
3417    characters to a class. The character values in the arguments will be within the
3418    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3419    mutually recursive with the function immediately below.
3420    
3421    Arguments:
3422      classbits     the bit map for characters < 256
3423      uchardptr     points to the pointer for extra data
3424      options       the options word
3425      cd            contains pointers to tables etc.
3426      start         start of range character
3427      end           end of range character
3428    
3429    Returns:        the number of < 256 characters added
3430                    the pointer to extra data is updated
3431    */
3432    
3433    static int
3434    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3435      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3436    {
3437    pcre_uint32 c;
3438    int n8 = 0;
3439    
3440    /* If caseless matching is required, scan the range and process alternate
3441    cases. In Unicode, there are 8-bit characters that have alternate cases that
3442    are greater than 255 and vice-versa. Sometimes we can just extend the original
3443    range. */
3444    
3445    if ((options & PCRE_CASELESS) != 0)
3446      {
3447    #ifdef SUPPORT_UCP
3448      if ((options & PCRE_UTF8) != 0)
3449        {
3450        int rc;
3451        pcre_uint32 oc, od;
3452    
3453        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3454        c = start;
3455    
3456        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3457          {
3458          /* Handle a single character that has more than one other case. */
3459    
3460          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3461            PRIV(ucd_caseless_sets) + rc, oc);
3462    
3463          /* Do nothing if the other case range is within the original range. */
3464    
3465          else if (oc >= start && od <= end) continue;
3466    
3467          /* Extend the original range if there is overlap, noting that if oc < c, we
3468          can't have od > end because a subrange is always shorter than the basic
3469          range. Otherwise, use a recursive call to add the additional range. */
3470    
3471          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3472          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3473          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3474          }
3475        }
3476      else
3477    #endif  /* SUPPORT_UCP */
3478    
3479      /* Not UTF-mode, or no UCP */
3480    
3481      for (c = start; c <= end && c < 256; c++)
3482        {
3483        SETBIT(classbits, cd->fcc[c]);
3484        n8++;
3485        }
3486      }
3487    
3488    /* Now handle the original range. Adjust the final value according to the bit
3489    length - this means that the same lists of (e.g.) horizontal spaces can be used
3490    in all cases. */
3491    
3492    #if defined COMPILE_PCRE8
3493    #ifdef SUPPORT_UTF
3494      if ((options & PCRE_UTF8) == 0)
3495    #endif
3496      if (end > 0xff) end = 0xff;
3497    
3498    #elif defined COMPILE_PCRE16
3499    #ifdef SUPPORT_UTF
3500      if ((options & PCRE_UTF16) == 0)
3501    #endif
3502      if (end > 0xffff) end = 0xffff;
3503    
3504    #endif /* COMPILE_PCRE[8|16] */
3505    
3506    /* If all characters are less than 256, use the bit map. Otherwise use extra
3507    data. */
3508    
3509    if (end < 0x100)
3510      {
3511      for (c = start; c <= end; c++)
3512        {
3513        n8++;
3514        SETBIT(classbits, c);
3515        }
3516      }
3517    
3518    else
3519      {
3520      pcre_uchar *uchardata = *uchardptr;
3521    
3522    #ifdef SUPPORT_UTF
3523      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3524        {
3525        if (start < end)
3526          {
3527          *uchardata++ = XCL_RANGE;
3528          uchardata += PRIV(ord2utf)(start, uchardata);
3529          uchardata += PRIV(ord2utf)(end, uchardata);
3530          }
3531        else if (start == end)
3532          {
3533          *uchardata++ = XCL_SINGLE;
3534          uchardata += PRIV(ord2utf)(start, uchardata);
3535          }
3536        }
3537      else
3538    #endif  /* SUPPORT_UTF */
3539    
3540      /* Without UTF support, character values are constrained by the bit length,
3541      and can only be > 256 for 16-bit and 32-bit libraries. */
3542    
3543    #ifdef COMPILE_PCRE8
3544        {}
3545    #else
3546      if (start < end)
3547        {
3548        *uchardata++ = XCL_RANGE;
3549        *uchardata++ = start;
3550        *uchardata++ = end;
3551        }
3552      else if (start == end)
3553        {
3554        *uchardata++ = XCL_SINGLE;
3555        *uchardata++ = start;
3556        }
3557    #endif
3558    
3559      *uchardptr = uchardata;   /* Updata extra data pointer */
3560      }
3561    
3562    return n8;    /* Number of 8-bit characters */
3563    }
3564    
3565    
3566    
3567    
3568    /*************************************************
3569    *        Add a list of characters to a class     *
3570    *************************************************/
3571    
3572    /* This function is used for adding a list of case-equivalent characters to a
3573    class, and also for adding a list of horizontal or vertical whitespace. If the
3574    list is in order (which it should be), ranges of characters are detected and
3575    handled appropriately. This function is mutually recursive with the function
3576    above.
3577    
3578    Arguments:
3579      classbits     the bit map for characters < 256
3580      uchardptr     points to the pointer for extra data
3581      options       the options word
3582      cd            contains pointers to tables etc.
3583      p             points to row of 32-bit values, terminated by NOTACHAR
3584      except        character to omit; this is used when adding lists of
3585                      case-equivalent characters to avoid including the one we
3586                      already know about
3587    
3588    Returns:        the number of < 256 characters added
3589                    the pointer to extra data is updated
3590    */
3591    
3592    static int
3593    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3594      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3595    {
3596    int n8 = 0;
3597    while (p[0] < NOTACHAR)
3598      {
3599      int n = 0;
3600      if (p[0] != except)
3601        {
3602        while(p[n+1] == p[0] + n + 1) n++;
3603        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3604        }
3605      p += n + 1;
3606      }
3607    return n8;
3608    }
3609    
3610    
3611    
3612    /*************************************************
3613    *    Add characters not in a list to a class     *
3614    *************************************************/
3615    
3616    case OP_NOT_WORDCHAR:  /* This function is used for adding the complement of a list of horizontal or
3617    return next == -ESC_w || next == -ESC_d;  vertical whitespace to a class. The list must be in order.
3618    
3619    default:  Arguments:
3620    return FALSE;    classbits     the bit map for characters < 256
3621    }    uchardptr     points to the pointer for extra data
3622      options       the options word
3623      cd            contains pointers to tables etc.
3624      p             points to row of 32-bit values, terminated by NOTACHAR
3625    
3626    Returns:        the number of < 256 characters added
3627                    the pointer to extra data is updated
3628    */
3629    
3630  /* Control does not reach here */  static int
3631  }  add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3632      int options, compile_data *cd, const pcre_uint32 *p)
3633    {
3634    BOOL utf = (options & PCRE_UTF8) != 0;
3635    int n8 = 0;
3636    if (p[0] > 0)
3637      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3638    while (p[0] < NOTACHAR)
3639      {
3640      while (p[1] == p[0] + 1) p++;
3641      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3642        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3643      p++;
3644      }
3645    return n8;
3646    }
3647    
3648    
3649    
# Line 3091  Arguments: Line 3662  Arguments:
3662    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3663    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3664    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3665    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)
3666    reqbyteptr     set to the last literal character required, else < 0    reqcharptr     set to the last literal character required, else < 0
3667    bcptr          points to current branch chain    bcptr          points to current branch chain
3668    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3669    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3104  Returns:         TRUE on success Line 3675  Returns:         TRUE on success
3675  */  */
3676    
3677  static BOOL  static BOOL
3678  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3679    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr, pcre_int32 *firstcharptr,
3680    int cond_depth, compile_data *cd, int *lengthptr)    pcre_int32 *reqcharptr, branch_chain *bcptr, int cond_depth,
3681      compile_data *cd, int *lengthptr)
3682  {  {
3683  int repeat_type, op_type;  int repeat_type, op_type;
3684  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3685  int bravalue = 0;  int bravalue = 0;
3686  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3687  int firstbyte, reqbyte;  pcre_int32 firstchar, reqchar;
3688  int zeroreqbyte, zerofirstbyte;  pcre_int32 zeroreqchar, zerofirstchar;
3689  int req_caseopt, reqvary, tempreqvary;  pcre_int32 req_caseopt, reqvary, tempreqvary;
3690  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3691  int after_manual_callout = 0;  int after_manual_callout = 0;
3692  int length_prevgroup = 0;  int length_prevgroup = 0;
3693  register int c;  register int c;
3694  register uschar *code = *codeptr;  int escape;
3695  uschar *last_code = code;  register pcre_uchar *code = *codeptr;
3696  uschar *orig_code = code;  pcre_uchar *last_code = code;
3697  uschar *tempcode;  pcre_uchar *orig_code = code;
3698    pcre_uchar *tempcode;
3699  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3700  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3701  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3702  const uschar *tempptr;  const pcre_uchar *tempptr;
3703  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3704  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3705  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3706  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3707  uschar classbits[32];  pcre_uint8 classbits[32];
3708    
3709  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we  /* We can fish out the UTF-8 setting once and for all into a BOOL, but we
3710  must not do this for other options (e.g. PCRE_EXTENDED) because they may change  must not do this for other options (e.g. PCRE_EXTENDED) because they may change
3711  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3712    
3713  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3714  BOOL class_utf8;  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3715  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3716  uschar *class_utf8data;  pcre_uchar utf_chars[6];
 uschar *class_utf8data_base;  
 uschar utf8_char[6];  
3717  #else  #else
3718  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3719    #endif
3720    
3721    /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3722    class_uchardata always so that it can be passed to add_to_class() always,
3723    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3724    alternative calls for the different cases. */
3725    
3726    pcre_uchar *class_uchardata;
3727    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3728    BOOL xclass;
3729    pcre_uchar *class_uchardata_base;
3730  #endif  #endif
3731    
3732  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3158  greedy_non_default = greedy_default ^ 1; Line 3740  greedy_non_default = greedy_default ^ 1;
3740    
3741  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3742  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
3743  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
3744  find one.  find one.
3745    
3746  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
3747  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
3748  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3749  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3750    
3751  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = REQ_UNSET;
3752    
3753  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3754  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
3755  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
3756  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3757    value. This is used only for ASCII characters. */
3758    
3759  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3760    
3761  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3762    
# Line 3185  for (;; ptr++) Line 3768  for (;; ptr++)
3768    BOOL is_quantifier;    BOOL is_quantifier;
3769    BOOL is_recurse;    BOOL is_recurse;
3770    BOOL reset_bracount;    BOOL reset_bracount;
3771    int class_charcount;    int class_has_8bitchar;
3772    int class_lastchar;    int class_one_char;
3773    int newoptions;    int newoptions;
3774    int recno;    int recno;
3775    int refsign;    int refsign;
3776    int skipbytes;    int skipbytes;
3777    int subreqbyte;    int subreqchar;
3778    int subfirstbyte;    int subfirstchar;
3779    int terminator;    int terminator;
3780    int mclength;    int mclength;
3781    int tempbracount;    int tempbracount;
3782    uschar mcbuffer[8];    int ec; // FIXMEchpe pcre_uint32
3783      pcre_uchar mcbuffer[8];
3784    
3785    /* Get next byte in the pattern */    /* Get next character in the pattern */
3786    
3787    c = *ptr;    c = *ptr;
3788    
# Line 3220  for (;; ptr++) Line 3804  for (;; ptr++)
3804  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3805      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3806  #endif  #endif
3807      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */      if (code > cd->start_workspace + cd->workspace_size -
3808            WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3809        {        {
3810        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3811        goto FAILED;        goto FAILED;
# Line 3243  for (;; ptr++) Line 3828  for (;; ptr++)
3828        }        }
3829    
3830      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3831      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3832          (int)(code - last_code), c, c));
3833    
3834      /* 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
3835      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 3253  for (;; ptr++) Line 3839  for (;; ptr++)
3839        {        {
3840        if (previous > orig_code)        if (previous > orig_code)
3841          {          {
3842          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3843          code -= previous - orig_code;          code -= previous - orig_code;
3844          previous = orig_code;          previous = orig_code;
3845          }          }
# Line 3269  for (;; ptr++) Line 3855  for (;; ptr++)
3855    /* 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
3856    reference list. */    reference list. */
3857    
3858    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3859               WORK_SIZE_SAFETY_MARGIN)
3860      {      {
3861      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3862      goto FAILED;      goto FAILED;
# Line 3321  for (;; ptr++) Line 3908  for (;; ptr++)
3908    
3909    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
3910      {      {
3911      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
3912      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3913        {        {
3914        ptr++;        ptr++;
# Line 3329  for (;; ptr++) Line 3916  for (;; ptr++)
3916          {          {
3917          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3918          ptr++;          ptr++;
3919  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3920          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3921  #endif  #endif
3922          }          }
3923        if (*ptr != 0) continue;        if (*ptr != 0) continue;
# Line 3354  for (;; ptr++) Line 3941  for (;; ptr++)
3941      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
3942      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3943      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3944      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
3945      *reqbyteptr = reqbyte;      *reqcharptr = reqchar;
3946      *codeptr = code;      *codeptr = code;
3947      *ptrptr = ptr;      *ptrptr = ptr;
3948      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3379  for (;; ptr++) Line 3966  for (;; ptr++)
3966      previous = NULL;      previous = NULL;
3967      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3968        {        {
3969        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3970        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3971        }        }
3972      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3391  for (;; ptr++) Line 3978  for (;; ptr++)
3978      break;      break;
3979    
3980      /* 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
3981      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
3982    
3983      case CHAR_DOT:      case CHAR_DOT:
3984      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
3985      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
3986      zeroreqbyte = reqbyte;      zeroreqchar = reqchar;
3987      previous = code;      previous = code;
3988      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
3989      break;      break;
# Line 3451  for (;; ptr++) Line 4038  for (;; ptr++)
4038          {          {
4039          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
4040            ptr++;            ptr++;
4041          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)  
4042            ptr += 3;            ptr += 3;
4043          else          else
4044            break;            break;
# Line 3471  for (;; ptr++) Line 4057  for (;; ptr++)
4057          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4058        {        {
4059        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4060        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4061        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
4062        break;        break;
4063        }        }
4064    
# Line 3482  for (;; ptr++) Line 4068  for (;; ptr++)
4068    
4069      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4070    
4071      /* 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:
4072      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 <
4073      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
4074        character. */
4075    
4076      class_charcount = 0;      class_has_8bitchar = 0;
4077      class_lastchar = -1;      class_one_char = 0;
4078    
4079      /* 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
4080      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
4081      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
4082      */      map. */
4083    
4084      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4085    
4086  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4087      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;
4088      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4089      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4090  #endif  #endif
4091    
4092      /* 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 3508  for (;; ptr++) Line 4095  for (;; ptr++)
4095    
4096      if (c != 0) do      if (c != 0) do
4097        {        {
4098        const uschar *oldptr;        const pcre_uchar *oldptr;
4099    
4100  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4101        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
4102          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4103          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
4104          }          }
4105    #endif
4106    
4107        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4108          /* In the pre-compile phase, accumulate the length of any extra
4109        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
4110        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4111        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4112          however. */
4113        if (lengthptr != NULL)  
4114          {        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4115          *lengthptr += class_utf8data - class_utf8data_base;          {
4116          class_utf8data = class_utf8data_base;          xclass = TRUE;
4117            *lengthptr += class_uchardata - class_uchardata_base;
4118            class_uchardata = class_uchardata_base;
4119          }          }
   
4120  #endif  #endif
4121    
4122        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3554  for (;; ptr++) Line 4144  for (;; ptr++)
4144          {          {
4145          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
4146          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
4147          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
4148          uschar pbits[32];          pcre_uint8 pbits[32];
4149    
4150          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
4151            {            {
# Line 3582  for (;; ptr++) Line 4172  for (;; ptr++)
4172          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4173          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4174    
4175          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4176            posix_class = 0;            posix_class = 0;
4177    
4178          /* 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 3610  for (;; ptr++) Line 4200  for (;; ptr++)
4200          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
4201    
4202          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
4203            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
4204    
4205          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
4206    
# Line 3625  for (;; ptr++) Line 4215  for (;; ptr++)
4215              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4216            }            }
4217    
4218          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4219          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4220    
4221          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3641  for (;; ptr++) Line 4231  for (;; ptr++)
4231            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4232    
4233          ptr = tempptr + 1;          ptr = tempptr + 1;
4234          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 character. */
4235            class_has_8bitchar = 1;
4236            /* Every class contains at least two characters. */
4237            class_one_char = 2;
4238          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4239          }          }
4240    
4241        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
4242        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
4243        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
4244        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
4245        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_one_char bigger
4246        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
4247          as literal characters (by default), or are faulted if
4248        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4249    
4250        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4251          {          {
4252          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4253    
4254          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4255    
4256          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4257          else if (-c == ESC_Q)            /* Handle start of quoted string */            c = ec;
4258            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4259            else if (escape == ESC_N)            /* \N is not supported in a class */
4260              {
4261              *errorcodeptr = ERR71;
4262              goto FAILED;
4263              }
4264            else if (escape == ESC_Q)            /* Handle start of quoted string */
4265            {            {
4266            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4267              {              {
# Line 3668  for (;; ptr++) Line 4270  for (;; ptr++)
4270            else inescq = TRUE;            else inescq = TRUE;
4271            continue;            continue;
4272            }            }
4273          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4274    
4275          if (c < 0)          else
4276            {            {
4277            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4278            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
4279              class_has_8bitchar++;
4280              /* Every class contains at least two characters. */
4281              class_one_char += 2;
4282    
4283            switch (-c)            switch (escape)
4284              {              {
4285  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4286              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3685  for (;; ptr++) Line 4290  for (;; ptr++)
4290              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4291              case ESC_SU:              case ESC_SU:
4292              nestptr = ptr;              nestptr = ptr;
4293              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4294              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4295              continue;              continue;
4296  #endif  #endif
4297              case ESC_d:              case ESC_d:
# Line 3709  for (;; ptr++) Line 4314  for (;; ptr++)
4314    
4315              /* 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
4316              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4317              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4318                EBCDIC, so we lazily just adjust the appropriate bit. */
4319    
4320              case ESC_s:              case ESC_s:
4321              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 3722  for (;; ptr++) Line 4328  for (;; ptr++)
4328              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4329              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4330              continue;              continue;
4331    
4332                /* The rest apply in both UCP and non-UCP cases. */
4333    
4334              case ESC_h:              case ESC_h:
4335              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4336              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  
4337              continue;              continue;
4338    
4339              case ESC_H:              case ESC_H:
4340              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4341                {                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  
4342              continue;              continue;
4343    
4344              case ESC_v:              case ESC_v:
4345              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4346              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  
4347              continue;              continue;
4348    
4349              case ESC_V:              case ESC_V:
4350              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4351                {                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  
4352              continue;              continue;
4353    
4354  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3847  for (;; ptr++) Line 4359  for (;; ptr++)
4359                int pdata;                int pdata;
4360                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);
4361                if (ptype < 0) goto FAILED;                if (ptype < 0) goto FAILED;
4362                class_utf8 = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_utf8data++ = ((-c == ESC_p) != negated)?  
4363                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4364                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4365                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4366                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4367                continue;                continue;
4368                }                }
4369  #endif  #endif
# Line 3866  for (;; ptr++) Line 4377  for (;; ptr++)
4377                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4378                goto FAILED;                goto FAILED;
4379                }                }
4380              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4381              c = *ptr;              /* Get the final character and fall through */              class_one_char -= 2;     /* Undo the speculative increase. */
4382                c = *ptr;                /* Get the final character and fall through */
4383              break;              break;
4384              }              }
4385            }            }
4386    
4387          /* 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).
4388          greater than 256 in UTF-8 mode. */          This may be greater than 256. */
4389    
4390            escape = 0;
4391    
4392          }   /* End of backslash handling */          }   /* End of backslash handling */
4393    
4394        /* 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
4395        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
4396        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4397        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4398    
4399        CHECK_RANGE:        CHECK_RANGE:
4400        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3888  for (;; ptr++) Line 4402  for (;; ptr++)
4402          inescq = FALSE;          inescq = FALSE;
4403          ptr += 2;          ptr += 2;
4404          }          }
   
4405        oldptr = ptr;        oldptr = ptr;
4406    
4407        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4408    
4409        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4410    
# Line 3914  for (;; ptr++) Line 4427  for (;; ptr++)
4427            inescq = TRUE;            inescq = TRUE;
4428            break;            break;
4429            }            }
4430    
4431            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4432            back the pointer and jump to handle the character that preceded it. */
4433    
4434          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4435            {            {
4436            ptr = oldptr;            ptr = oldptr;
4437            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4438            }            }
4439    
4440            /* Otherwise, we have a potential range; pick up the next character */
4441    
4442  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4443          if (utf8)          if (utf)
4444            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4445            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4446            }            }
# Line 3936  for (;; ptr++) Line 4454  for (;; ptr++)
4454    
4455          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4456            {            {
4457            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4458              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4459            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4460    
4461            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4462    
4463            if (d < 0)            if (descape != 0)
4464              {              {
4465              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4466                {                {
4467                ptr = oldptr;                ptr = oldptr;
4468                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4469                }                }
4470              }              }
4471            }            }
4472    
4473          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4474          one-character ranges */          one-character ranges. */
4475    
4476          if (d < c)          if (d < c)
4477            {            {
4478            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4479            goto FAILED;            goto FAILED;
4480            }            }
4481            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4482    
4483          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4484            cannot be done anymore. Any value greater than 1 indicates that there
4485            is more than one character. */
4486    
4487            class_one_char = 2;
4488    
4489          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4490    
4491          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4492    
4493          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless          class_has_8bitchar +=
4494          matching, we have to use an XCLASS with extra data items. Caseless            add_to_class(classbits, &class_uchardata, options, cd, c, d);
4495          matching for characters > 127 is available only if UCP support is  
         available. */  
   
 #ifdef SUPPORT_UTF8  
         if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))  
           {  
           class_utf8 = TRUE;  
   
           /* With UCP support, we can find the other case equivalents of  
           the relevant characters. There may be several ranges. Optimize how  
           they fit with the basic range. */  
   
 #ifdef SUPPORT_UCP  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             unsigned int occ, ocd;  
             unsigned int cc = c;  
             unsigned int origd = d;  
             while (get_othercase_range(&cc, origd, &occ, &ocd))  
               {  
               if (occ >= (unsigned int)c &&  
                   ocd <= (unsigned int)d)  
                 continue;                          /* Skip embedded ranges */  
   
               if (occ < (unsigned int)c  &&  
                   ocd >= (unsigned int)c - 1)      /* Extend the basic range */  
                 {                                  /* if there is overlap,   */  
                 c = occ;                           /* noting that if occ < c */  
                 continue;                          /* we can't have ocd > d  */  
                 }                                  /* because a subrange is  */  
               if (ocd > (unsigned int)d &&  
                   occ <= (unsigned int)d + 1)      /* always shorter than    */  
                 {                                  /* the basic range.       */  
                 d = ocd;  
                 continue;  
                 }  
   
               if (occ == ocd)  
                 {  
                 *class_utf8data++ = XCL_SINGLE;  
                 }  
               else  
                 {  
                 *class_utf8data++ = XCL_RANGE;  
                 class_utf8data += _pcre_ord2utf8(occ, class_utf8data);  
                 }  
               class_utf8data += _pcre_ord2utf8(ocd, class_utf8data);  
               }  
             }  
 #endif  /* SUPPORT_UCP */  
   
           /* Now record the original range, possibly modified for UCP caseless  
           overlapping ranges. */  
   
           *class_utf8data++ = XCL_RANGE;  
           class_utf8data += _pcre_ord2utf8(c, class_utf8data);  
           class_utf8data += _pcre_ord2utf8(d, class_utf8data);  
   
           /* With UCP support, we are done. Without UCP support, there is no  
           caseless matching for UTF-8 characters > 127; we can use the bit map  
           for the smaller ones. */  
   
 #ifdef SUPPORT_UCP  
           continue;    /* With next character in the class */  
 #else  
           if ((options & PCRE_CASELESS) == 0 || c > 127) continue;  
   
           /* Adjust upper limit and fall through to set up the map */  
   
           d = 127;  
   
 #endif  /* SUPPORT_UCP */  
           }  
 #endif  /* SUPPORT_UTF8 */  
   
         /* We use the bit map for all cases when not in UTF-8 mode; else  
         ranges that lie entirely within 0-127 when there is UCP support; else  
         for partial ranges without UCP support. */  
   
         class_charcount += d - c + 1;  
         class_lastchar = d;  
   
         /* We can save a bit of time by skipping this in the pre-compile. */  
   
         if (lengthptr == NULL) for (; c <= d; c++)  
           {  
           classbits[c/8] |= (1 << (c&7));  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             int uc = cd->fcc[c];           /* flip case */  
             classbits[uc/8] |= (1 << (uc&7));  
             }  
           }  
   
4496          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4497          }          }
4498    
4499        /* Handle a lone single character - we can get here for a normal        /* Handle a single character - we can get here for a normal non-escape
4500        non-escape char, or after \ that introduces a single character or for an        char, or after \ that introduces a single character or for an apparent
4501        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4502          increase it if it is already 2 or more ... just in case there's a class
4503        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4504    
4505          CLASS_SINGLE_CHARACTER:
4506          if (class_one_char < 2) class_one_char++;
4507    
4508          /* If class_one_char is 1, we have the first single character in the
4509          class, and there have been no prior ranges, or XCLASS items generated by
4510          escapes. If this is the final character in the class, we can optimize by
4511          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4512          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4513          to be set. Otherwise, there can be no first char if this item is first,
4514          whatever repeat count may follow. In the case of reqchar, save the
4515          previous value for reinstating. */
4516    
4517        /* Handle a character that cannot go in the bit map */        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
   
 #ifdef SUPPORT_UTF8  
       if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
4518          {          {
4519          class_utf8 = TRUE;          ptr++;
4520          *class_utf8data++ = XCL_SINGLE;          zeroreqchar = reqchar;
         class_utf8data += _pcre_ord2utf8(c, class_utf8data);  
4521    
4522  #ifdef SUPPORT_UCP          if (negate_class)
         if ((options & PCRE_CASELESS) != 0)  
4523            {            {
4524            unsigned int othercase;  #ifdef SUPPORT_UCP
4525            if ((othercase = UCD_OTHERCASE(c)) != c)              // FIXMEchpe pcreuint32?
4526              int d;
4527    #endif
4528              if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4529              zerofirstchar = firstchar;
4530    
4531              /* For caseless UTF-8 mode when UCP support is available, check
4532              whether this character has more than one other case. If so, generate
4533              a special OP_NOTPROP item instead of OP_NOTI. */
4534    
4535    #ifdef SUPPORT_UCP
4536              if (utf && (options & PCRE_CASELESS) != 0 &&
4537                  (d = UCD_CASESET(c)) != 0)
4538                {
4539                *code++ = OP_NOTPROP;
4540                *code++ = PT_CLIST;
4541                *code++ = d;
4542                }
4543              else
4544    #endif
4545              /* Char has only one other case, or UCP not available */
4546    
4547              {              {
4548              *class_utf8data++ = XCL_SINGLE;              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4549              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4550                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4551                  code += PRIV(ord2utf)(c, code);
4552                else
4553    #endif
4554                  *code++ = c;
4555              }              }
4556    
4557              /* We are finished with this character class */
4558    
4559              goto END_CLASS;
4560            }            }
 #endif  /* SUPPORT_UCP */  
4561    
4562          }          /* For a single, positive character, get the value into mcbuffer, and
4563        else          then we can handle this with the normal one-character code. */
 #endif  /* SUPPORT_UTF8 */  
4564    
4565        /* Handle a single-byte character */  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4566          {          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4567          classbits[c/8] |= (1 << (c&7));            mclength = PRIV(ord2utf)(c, mcbuffer);
4568          if ((options & PCRE_CASELESS) != 0)          else
4569    #endif
4570            {            {
4571            c = cd->fcc[c];   /* flip case */            mcbuffer[0] = c;
4572            classbits[c/8] |= (1 << (c&7));            mclength = 1;
4573            }            }
4574          class_charcount++;          goto ONE_CHAR;
4575          class_lastchar = c;          }       /* End of 1-char optimization */
4576          }  
4577          /* There is more than one character in the class, or an XCLASS item
4578          has been generated. Add this character to the class. */
4579    
4580          class_has_8bitchar +=
4581            add_to_class(classbits, &class_uchardata, options, cd, c, c);
4582        }        }
4583    
4584      /* Loop until ']' reached. This "while" is the end of the "do" far above.      /* Loop until ']' reached. This "while" is the end of the "do" far above.
# Line 4126  for (;; ptr++) Line 4598  for (;; ptr++)
4598        goto FAILED;        goto FAILED;
4599        }        }
4600    
4601      /* If class_charcount is 1, we saw precisely one character whose value is      /* We will need an XCLASS if data has been placed in class_uchardata. In
4602      less than 256. As long as there were no characters >= 128 and there was no      the second phase this is a sufficient test. However, in the pre-compile
4603      use of \p or \P, in other words, no use of any XCLASS features, we can      phase, class_uchardata gets emptied to prevent workspace overflow, so it
4604      optimize.      only if the very last character in the class needs XCLASS will it contain
4605        anything at this point. For this reason, xclass gets set TRUE above when
4606      In UTF-8 mode, we can optimize the negative case only if there were no      uchar_classdata is emptied, and that's why this code is the way it is here
4607      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      instead of just doing a test on class_uchardata below. */
4608      operate on single-bytes characters only. This is an historical hangover.  
4609      Maybe one day we can tidy these opcodes to handle multi-byte characters.  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4610        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4611      The optimization throws away the bit map. We turn the item into a  #endif
4612      1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
4613      Note that OP_NOT[I] does not support multibyte characters. In the positive      /* If this is the first thing in the branch, there can be no first char
4614      case, it can cause firstbyte to be set. Otherwise, there can be no first      setting, whatever the repeat count. Any reqchar setting must remain
4615      char if this item is first, whatever repeat count may follow. In the case      unchanged after any kind of repeat. */
4616      of reqbyte, save the previous value for reinstating. */  
4617        if (firstchar == REQ_UNSET) firstchar = REQ_NONE;
4618  #ifdef SUPPORT_UTF8      zerofirstchar = firstchar;
4619      if (class_charcount == 1 && !class_utf8 &&      zeroreqchar = reqchar;
       (!utf8 || !negate_class || class_lastchar < 128))  
 #else  
     if (class_charcount == 1)  
 #endif  
       {  
       zeroreqbyte = reqbyte;  
   
       /* The OP_NOT[I] opcodes work on one-byte characters only. */  
   
       if (negate_class)  
         {  
         if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;  
         zerofirstbyte = firstbyte;  
         *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;  
         *code++ = class_lastchar;  
         break;  
         }  
   
       /* For a single, positive character, get the value into mcbuffer, and  
       then we can handle this with the normal one-character code. */  
   
 #ifdef SUPPORT_UTF8  
       if (utf8 && class_lastchar > 127)  
         mclength = _pcre_ord2utf8(class_lastchar, mcbuffer);  
       else  
 #endif  
         {  
         mcbuffer[0] = class_lastchar;  
         mclength = 1;  
         }  
       goto ONE_CHAR;  
       }       /* End of 1-char optimization */  
   
     /* The general case - not the one-char optimization. If this is the first  
     thing in the branch, there can be no first char setting, whatever the  
     repeat count. Any reqbyte setting must remain unchanged after any kind of  
     repeat. */  
   
     if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;  
     zerofirstbyte = firstbyte;  
     zeroreqbyte = reqbyte;  
4620    
4621      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4622      extended class, with its own opcode, unless there was a negated special      extended class, with its own opcode, unless there was a negated special
# Line 4195  for (;; ptr++) Line 4626  for (;; ptr++)
4626      be listed) there are no characters < 256, we can omit the bitmap in the      be listed) there are no characters < 256, we can omit the bitmap in the
4627      actual compiled code. */      actual compiled code. */
4628    
4629  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4630      if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))      if (xclass && (!should_flip_negation || (options & PCRE_UCP) != 0))
4631    #elif !defined COMPILE_PCRE8
4632        if (xclass && !should_flip_negation)
4633    #endif
4634    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4635        {        {
4636        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_uchardata++ = XCL_END;    /* Marks the end of extra data */
4637        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
4638        code += LINK_SIZE;        code += LINK_SIZE;
4639        *code = negate_class? XCL_NOT : 0;        *code = negate_class? XCL_NOT:0;
4640    
4641        /* If the map is required, move up the extra data to make room for it;        /* If the map is required, move up the extra data to make room for it;
4642        otherwise just move the code pointer to the end of the extra data. */        otherwise just move the code pointer to the end of the extra data. */
4643    
4644        if (class_charcount > 0)        if (class_has_8bitchar > 0)
4645          {          {
4646          *code++ |= XCL_MAP;          *code++ |= XCL_MAP;
4647          memmove(code + 32, code, class_utf8data - code);          memmove(code + (32 / sizeof(pcre_uchar)), code,
4648              IN_UCHARS(class_uchardata - code));
4649          memcpy(code, classbits, 32);          memcpy(code, classbits, 32);
4650          code = class_utf8data + 32;          code = class_uchardata + (32 / sizeof(pcre_uchar));
4651          }          }
4652        else code = class_utf8data;        else code = class_uchardata;
4653    
4654        /* Now fill in the complete length of the item */        /* Now fill in the complete length of the item */
4655    
4656        PUT(previous, 1, code - previous);        PUT(previous, 1, (int)(code - previous));
4657        break;   /* End of class handling */        break;   /* End of class handling */
4658        }        }
4659  #endif  #endif
# Line 4229  for (;; ptr++) Line 4665  for (;; ptr++)
4665      negating it if necessary. */      negating it if necessary. */
4666    
4667      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
4668      if (negate_class)      if (lengthptr == NULL)    /* Save time in the pre-compile phase */
       {  
       if (lengthptr == NULL)    /* Save time in the pre-compile phase */  
         for (c = 0; c < 32; c++) code[c] = ~classbits[c];  
       }  
     else  
4669        {        {
4670          if (negate_class)
4671            for (c = 0; c < 32; c++) classbits[c] = ~classbits[c];
4672        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4673        }        }
4674      code += 32;      code += 32 / sizeof(pcre_uchar);
4675    
4676        END_CLASS:
4677      break;      break;
4678    
4679    
# Line 4275  for (;; ptr++) Line 4710  for (;; ptr++)
4710    
4711      if (repeat_min == 0)      if (repeat_min == 0)
4712        {        {
4713        firstbyte = zerofirstbyte;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4714        reqbyte = zeroreqbyte;        /* Ditto */        reqchar = zeroreqchar;        /* Ditto */
4715        }        }
4716    
4717      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4318  for (;; ptr++) Line 4753  for (;; ptr++)
4753    
4754      if (*previous == OP_RECURSE)      if (*previous == OP_RECURSE)
4755        {        {
4756        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);        memmove(previous + 1 + LINK_SIZE, previous, IN_UCHARS(1 + LINK_SIZE));
4757        *previous = OP_ONCE;        *previous = OP_ONCE;
4758        PUT(previous, 1, 2 + 2*LINK_SIZE);        PUT(previous, 1, 2 + 2*LINK_SIZE);
4759        previous[2 + 2*LINK_SIZE] = OP_KET;        previous[2 + 2*LINK_SIZE] = OP_KET;
# Line 4339  for (;; ptr++) Line 4774  for (;; ptr++)
4774    
4775      /* Now handle repetition for the different types of item. */      /* Now handle repetition for the different types of item. */
4776    
4777      /* If previous was a character match, abolish the item and generate a      /* If previous was a character or negated character match, abolish the item
4778      repeat item instead. If a char item has a minumum of more than one, ensure      and generate a repeat item instead. If a char item has a minimum of more
4779      that it is set in reqbyte - it might not be if a sequence such as x{3} is      than one, ensure that it is set in reqchar - it might not be if a sequence
4780      the first thing in a branch because the x will have gone into firstbyte      such as x{3} is the first thing in a branch because the x will have gone
4781      instead.  */      into firstchar instead.  */
4782    
4783      if (*previous == OP_CHAR || *previous == OP_CHARI)      if (*previous == OP_CHAR || *previous == OP_CHARI
4784            || *previous == OP_NOT || *previous == OP_NOTI)
4785        {        {
4786        op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;        switch (*previous)
4787            {
4788            default: /* Make compiler happy. */