/[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 723 by ph10, Sat Oct 8 15:55:23 2011 UTC revision 1147 by chpe, Sat Oct 20 20:17:03 2012 UTC
# Line 6  Line 6 
6  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
7    
8                         Written by Philip Hazel                         Written by Philip Hazel
9             Copyright (c) 1997-2011 University of Cambridge             Copyright (c) 1997-2012 University of Cambridge
10    
11  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
12  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 53  supporting internal functions that are n Line 53  supporting internal functions that are n
53  #include "pcre_internal.h"  #include "pcre_internal.h"
54    
55    
56  /* When PCRE_DEBUG is defined, we need the pcre_printint() function, which is  /* When PCRE_DEBUG is defined, we need the pcre(16|32)_printint() function, which
57  also used by pcretest. PCRE_DEBUG is not defined when building a production  is also used by pcretest. PCRE_DEBUG is not defined when building a production
58  library. */  library. We do not need to select pcre16_printint.c specially, because the
59    COMPILE_PCREx macro will already be appropriately set. */
60    
61  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
62  #include "pcre_printint.src"  /* pcre_printint.c should not include any headers */
63    #define PCRE_INCLUDED
64    #include "pcre_printint.c"
65    #undef PCRE_INCLUDED
66  #endif  #endif
67    
68    
69  /* Macro for setting individual bits in class bitmaps. */  /* Macro for setting individual bits in class bitmaps. */
70    
71  #define SETBIT(a,b) a[b/8] |= (1 << (b%8))  #define SETBIT(a,b) a[(b)/8] |= (1 << ((b)&7))
72    
73  /* Maximum length value to check against when making sure that the integer that  /* Maximum length value to check against when making sure that the integer that
74  holds the compiled pattern length does not overflow. We make it a bit less than  holds the compiled pattern length does not overflow. We make it a bit less than
# Line 73  to check them every time. */ Line 77  to check them every time. */
77    
78  #define OFLOW_MAX (INT_MAX - 20)  #define OFLOW_MAX (INT_MAX - 20)
79    
80    /* Definitions to allow mutual recursion */
81    
82    static int
83      add_list_to_class(pcre_uint8 *, pcre_uchar **, int, compile_data *,
84        const pcre_uint32 *, unsigned int);
85    
86    static BOOL
87      compile_regex(int, pcre_uchar **, const pcre_uchar **, int *, BOOL, BOOL, int, int,
88        pcre_uint32 *, pcre_int32 *, pcre_uint32 *, pcre_int32 *, branch_chain *,
89        compile_data *, int *);
90    
91    
92    
93  /*************************************************  /*************************************************
94  *      Code parameters and static tables         *  *      Code parameters and static tables         *
# Line 88  so this number is very generous. Line 104  so this number is very generous.
104  The same workspace is used during the second, actual compile phase for  The same workspace is used during the second, actual compile phase for
105  remembering forward references to groups so that they can be filled in at the  remembering forward references to groups so that they can be filled in at the
106  end. Each entry in this list occupies LINK_SIZE bytes, so even when LINK_SIZE  end. Each entry in this list occupies LINK_SIZE bytes, so even when LINK_SIZE
107  is 4 there is plenty of room. */  is 4 there is plenty of room for most patterns. However, the memory can get
108    filled up by repetitions of forward references, for example patterns like
109    /(?1){0,1999}(b)/, and one user did hit the limit. The code has been changed so
110    that the workspace is expanded using malloc() in this situation. The value
111    below is therefore a minimum, and we put a maximum on it for safety. The
112    minimum is now also defined in terms of LINK_SIZE so that the use of malloc()
113    kicks in at the same number of forward references in all cases. */
114    
115  #define COMPILE_WORK_SIZE (4096)  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)
116    #define COMPILE_WORK_SIZE_MAX (100*COMPILE_WORK_SIZE)
117    
118  /* The overrun tests check for a slightly smaller size so that they detect the  /* The overrun tests check for a slightly smaller size so that they detect the
119  overrun before it actually does run off the end of the data block. */  overrun before it actually does run off the end of the data block. */
120    
121  #define WORK_SIZE_CHECK (COMPILE_WORK_SIZE - 100)  #define WORK_SIZE_SAFETY_MARGIN (100)
122    
123    /* Private flags added to firstchar and reqchar. */
124    
125    #define REQ_CASELESS    (1 << 0)        /* Indicates caselessness */
126    #define REQ_VARY        (1 << 1)        /* Reqchar followed non-literal item */
127    /* Negative values for the firstchar and reqchar flags */
128    #define REQ_UNSET       (-2)
129    #define REQ_NONE        (-1)
130    
131    /* Repeated character flags. */
132    
133    #define UTF_LENGTH     0x10000000l      /* The char contains its length. */
134    
135  /* Table for handling escaped characters in the range '0'-'z'. Positive returns  /* Table for handling escaped characters in the range '0'-'z'. Positive returns
136  are simple data values; negative values are for special things like \d and so  are simple data values; negative values are for special things like \d and so
# Line 231  static const char posix_names[] = Line 265  static const char posix_names[] =
265    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0    STRING_graph0 STRING_print0 STRING_punct0 STRING_space0
266    STRING_word0  STRING_xdigit;    STRING_word0  STRING_xdigit;
267    
268  static const uschar posix_name_lengths[] = {  static const pcre_uint8 posix_name_lengths[] = {
269    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };    5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 5, 4, 6, 0 };
270    
271  /* Table of class bit maps for each POSIX class. Each class is formed from a  /* Table of class bit maps for each POSIX class. Each class is formed from a
# Line 266  substitutes must be in the order of the Line 300  substitutes must be in the order of the
300  both positive and negative cases. NULL means no substitute. */  both positive and negative cases. NULL means no substitute. */
301    
302  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
303  static const uschar *substitutes[] = {  static const pcre_uchar string_PNd[]  = {
304    (uschar *)"\\P{Nd}",    /* \D */    CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
305    (uschar *)"\\p{Nd}",    /* \d */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
306    (uschar *)"\\P{Xsp}",   /* \S */       /* NOTE: Xsp is Perl space */  static const pcre_uchar string_pNd[]  = {
307    (uschar *)"\\p{Xsp}",   /* \s */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
308    (uschar *)"\\P{Xwd}",   /* \W */    CHAR_N, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
309    (uschar *)"\\p{Xwd}"    /* \w */  static const pcre_uchar string_PXsp[] = {
310      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
311      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
312    static const pcre_uchar string_pXsp[] = {
313      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
314      CHAR_X, CHAR_s, CHAR_p, CHAR_RIGHT_CURLY_BRACKET, '\0' };
315    static const pcre_uchar string_PXwd[] = {
316      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
317      CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
318    static const pcre_uchar string_pXwd[] = {
319      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
320      CHAR_X, CHAR_w, CHAR_d, CHAR_RIGHT_CURLY_BRACKET, '\0' };
321    
322    static const pcre_uchar *substitutes[] = {
323      string_PNd,           /* \D */
324      string_pNd,           /* \d */
325      string_PXsp,          /* \S */       /* NOTE: Xsp is Perl space */
326      string_pXsp,          /* \s */
327      string_PXwd,          /* \W */
328      string_pXwd           /* \w */
329  };  };
330    
331  static const uschar *posix_substitutes[] = {  static const pcre_uchar string_pL[] =   {
332    (uschar *)"\\p{L}",     /* alpha */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
333    (uschar *)"\\p{Ll}",    /* lower */    CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
334    (uschar *)"\\p{Lu}",    /* upper */  static const pcre_uchar string_pLl[] =  {
335    (uschar *)"\\p{Xan}",   /* alnum */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
336    NULL,                   /* ascii */    CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
337    (uschar *)"\\h",        /* blank */  static const pcre_uchar string_pLu[] =  {
338    NULL,                   /* cntrl */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
339    (uschar *)"\\p{Nd}",    /* digit */    CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
340    NULL,                   /* graph */  static const pcre_uchar string_pXan[] = {
341    NULL,                   /* print */    CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
342    NULL,                   /* punct */    CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
343    (uschar *)"\\p{Xps}",   /* space */    /* NOTE: Xps is POSIX space */  static const pcre_uchar string_h[] =    {
344    (uschar *)"\\p{Xwd}",   /* word */    CHAR_BACKSLASH, CHAR_h, '\0' };
345    NULL,                   /* xdigit */  static const pcre_uchar string_pXps[] = {
346      CHAR_BACKSLASH, CHAR_p, CHAR_LEFT_CURLY_BRACKET,
347      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
348    static const pcre_uchar string_PL[] =   {
349      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
350      CHAR_L, CHAR_RIGHT_CURLY_BRACKET, '\0' };
351    static const pcre_uchar string_PLl[] =  {
352      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
353      CHAR_L, CHAR_l, CHAR_RIGHT_CURLY_BRACKET, '\0' };
354    static const pcre_uchar string_PLu[] =  {
355      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
356      CHAR_L, CHAR_u, CHAR_RIGHT_CURLY_BRACKET, '\0' };
357    static const pcre_uchar string_PXan[] = {
358      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
359      CHAR_X, CHAR_a, CHAR_n, CHAR_RIGHT_CURLY_BRACKET, '\0' };
360    static const pcre_uchar string_H[] =    {
361      CHAR_BACKSLASH, CHAR_H, '\0' };
362    static const pcre_uchar string_PXps[] = {
363      CHAR_BACKSLASH, CHAR_P, CHAR_LEFT_CURLY_BRACKET,
364      CHAR_X, CHAR_p, CHAR_s, CHAR_RIGHT_CURLY_BRACKET, '\0' };
365    
366    static const pcre_uchar *posix_substitutes[] = {
367      string_pL,            /* alpha */
368      string_pLl,           /* lower */
369      string_pLu,           /* upper */
370      string_pXan,          /* alnum */
371      NULL,                 /* ascii */
372      string_h,             /* blank */
373      NULL,                 /* cntrl */
374      string_pNd,           /* digit */
375      NULL,                 /* graph */
376      NULL,                 /* print */
377      NULL,                 /* punct */
378      string_pXps,          /* space */    /* NOTE: Xps is POSIX space */
379      string_pXwd,          /* word */
380      NULL,                 /* xdigit */
381    /* Negated cases */    /* Negated cases */
382    (uschar *)"\\P{L}",     /* ^alpha */    string_PL,            /* ^alpha */
383    (uschar *)"\\P{Ll}",    /* ^lower */    string_PLl,           /* ^lower */
384    (uschar *)"\\P{Lu}",    /* ^upper */    string_PLu,           /* ^upper */
385    (uschar *)"\\P{Xan}",   /* ^alnum */    string_PXan,          /* ^alnum */
386    NULL,                   /* ^ascii */    NULL,                 /* ^ascii */
387    (uschar *)"\\H",        /* ^blank */    string_H,             /* ^blank */
388    NULL,                   /* ^cntrl */    NULL,                 /* ^cntrl */
389    (uschar *)"\\P{Nd}",    /* ^digit */    string_PNd,           /* ^digit */
390    NULL,                   /* ^graph */    NULL,                 /* ^graph */
391    NULL,                   /* ^print */    NULL,                 /* ^print */
392    NULL,                   /* ^punct */    NULL,                 /* ^punct */
393    (uschar *)"\\P{Xps}",   /* ^space */   /* NOTE: Xps is POSIX space */    string_PXps,          /* ^space */   /* NOTE: Xps is POSIX space */
394    (uschar *)"\\P{Xwd}",   /* ^word */    string_PXwd,          /* ^word */
395    NULL                    /* ^xdigit */    NULL                  /* ^xdigit */
396  };  };
397  #define POSIX_SUBSIZE (sizeof(posix_substitutes)/sizeof(uschar *))  #define POSIX_SUBSIZE (sizeof(posix_substitutes) / sizeof(pcre_uchar *))
398  #endif  #endif
399    
400  #define STRING(a)  # a  #define STRING(a)  # a
# Line 365  static const char error_texts[] = Line 453  static const char error_texts[] =
453    /* 30 */    /* 30 */
454    "unknown POSIX class name\0"    "unknown POSIX class name\0"
455    "POSIX collating elements are not supported\0"    "POSIX collating elements are not supported\0"
456    "this version of PCRE is not compiled with PCRE_UTF8 support\0"    "this version of PCRE is compiled without UTF support\0"
457    "spare error\0"  /** DEAD **/    "spare error\0"  /** DEAD **/
458    "character value in \\x{...} sequence is too large\0"    "character value in \\x{...} sequence is too large\0"
459    /* 35 */    /* 35 */
# Line 388  static const char error_texts[] = Line 476  static const char error_texts[] =
476    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"    "too many named subpatterns (maximum " XSTRING(MAX_NAME_COUNT) ")\0"
477    /* 50 */    /* 50 */
478    "repeated subpattern is too long\0"    /** DEAD **/    "repeated subpattern is too long\0"    /** DEAD **/
479    "octal value is greater than \\377 (not in UTF-8 mode)\0"    "octal value is greater than \\377 in 8-bit non-UTF-8 mode\0"
480    "internal error: overran compiling workspace\0"    "internal error: overran compiling workspace\0"
481    "internal error: previously-checked referenced subpattern not found\0"    "internal error: previously-checked referenced subpattern not found\0"
482    "DEFINE group contains more than one branch\0"    "DEFINE group contains more than one branch\0"
# Line 407  static const char error_texts[] = Line 495  static const char error_texts[] =
495    /* 65 */    /* 65 */
496    "different names for subpatterns of the same number are not allowed\0"    "different names for subpatterns of the same number are not allowed\0"
497    "(*MARK) must have an argument\0"    "(*MARK) must have an argument\0"
498    "this version of PCRE is not compiled with PCRE_UCP support\0"    "this version of PCRE is not compiled with Unicode property support\0"
499    "\\c must be followed by an ASCII character\0"    "\\c must be followed by an ASCII character\0"
500    "\\k is not followed by a braced, angle-bracketed, or quoted name\0"    "\\k is not followed by a braced, angle-bracketed, or quoted name\0"
501      /* 70 */
502      "internal error: unknown opcode in find_fixedlength()\0"
503      "\\N is not supported in a class\0"
504      "too many forward references\0"
505      "disallowed Unicode code point (>= 0xd800 && <= 0xdfff)\0"
506      "invalid UTF-16 string\0"
507      /* 75 */
508      "name is too long in (*MARK), (*PRUNE), (*SKIP), or (*THEN)\0"
509      "character value in \\u.... sequence is too large\0"
510      "invalid UTF-32 string\0"
511    ;    ;
512    
513  /* Table to identify digits and hex digits. This is used when compiling  /* Table to identify digits and hex digits. This is used when compiling
# Line 428  For convenience, we use the same bit def Line 526  For convenience, we use the same bit def
526    
527  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
528    
529    /* Using a simple comparison for decimal numbers rather than a memory read
530    is much faster, and the resulting code is simpler (the compiler turns it
531    into a subtraction and unsigned comparison). */
532    
533    #define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9)
534    
535  #ifndef EBCDIC  #ifndef EBCDIC
536    
537  /* This is the "normal" case, for ASCII systems, and EBCDIC systems running in  /* This is the "normal" case, for ASCII systems, and EBCDIC systems running in
538  UTF-8 mode. */  UTF-8 mode. */
539    
540  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
541    {    {
542    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
543    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */
# Line 472  static const unsigned char digitab[] = Line 576  static const unsigned char digitab[] =
576    
577  /* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */  /* This is the "abnormal" case, for EBCDIC systems not running in UTF-8 mode. */
578    
579  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
580    {    {
581    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
582    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */
# Line 507  static const unsigned char digitab[] = Line 611  static const unsigned char digitab[] =
611    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */
612    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */
613    
614  static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */  static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */
615    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */
616    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */
617    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */
# Line 543  static const unsigned char ebcdic_charta Line 647  static const unsigned char ebcdic_charta
647  #endif  #endif
648    
649    
 /* Definition to allow mutual recursion */  
   
 static BOOL  
   compile_regex(int, uschar **, const uschar **, int *, BOOL, BOOL, int, int,  
     int *, int *, branch_chain *, compile_data *, int *);  
   
   
650    
651  /*************************************************  /*************************************************
652  *            Find an error text                  *  *            Find an error text                  *
# Line 578  return s; Line 675  return s;
675    
676    
677  /*************************************************  /*************************************************
678    *           Expand the workspace                 *
679    *************************************************/
680    
681    /* This function is called during the second compiling phase, if the number of
682    forward references fills the existing workspace, which is originally a block on
683    the stack. A larger block is obtained from malloc() unless the ultimate limit
684    has been reached or the increase will be rather small.
685    
686    Argument: pointer to the compile data block
687    Returns:  0 if all went well, else an error number
688    */
689    
690    static int
691    expand_workspace(compile_data *cd)
692    {
693    pcre_uchar *newspace;
694    int newsize = cd->workspace_size * 2;
695    
696    if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;
697    if (cd->workspace_size >= COMPILE_WORK_SIZE_MAX ||
698        newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)
699     return ERR72;
700    
701    newspace = (PUBL(malloc))(IN_UCHARS(newsize));
702    if (newspace == NULL) return ERR21;
703    memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar));
704    cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace);
705    if (cd->workspace_size > COMPILE_WORK_SIZE)
706      (PUBL(free))((void *)cd->start_workspace);
707    cd->start_workspace = newspace;
708    cd->workspace_size = newsize;
709    return 0;
710    }
711    
712    
713    
714    /*************************************************
715  *            Check for counted repeat            *  *            Check for counted repeat            *
716  *************************************************/  *************************************************/
717    
# Line 593  Returns:    TRUE or FALSE Line 727  Returns:    TRUE or FALSE
727  */  */
728    
729  static BOOL  static BOOL
730  is_counted_repeat(const uschar *p)  is_counted_repeat(const pcre_uchar *p)
731  {  {
732  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
733  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
734    while (IS_DIGIT(*p)) p++;
735  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
736    
737  if (*p++ != CHAR_COMMA) return FALSE;  if (*p++ != CHAR_COMMA) return FALSE;
738  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
739    
740  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
741  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
742    while (IS_DIGIT(*p)) p++;
743    
744  return (*p == CHAR_RIGHT_CURLY_BRACKET);  return (*p == CHAR_RIGHT_CURLY_BRACKET);
745  }  }
# Line 615  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 751  return (*p == CHAR_RIGHT_CURLY_BRACKET);
751  *************************************************/  *************************************************/
752    
753  /* This function is called when a \ has been encountered. It either returns a  /* This function is called when a \ has been encountered. It either returns a
754  positive value for a simple escape such as \n, or a negative value which  positive value for a simple escape such as \n, or 0 for a data character
755  encodes one of the more complicated things such as \d. A backreference to group  which will be placed in chptr. A backreference to group n is returned as
756  n is returned as -(ESC_REF + n); ESC_REF is the highest ESC_xxx macro. When  negative n. When UTF-8 is enabled, a positive value greater than 255 may
757  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
758  ptr is pointing at the \. On exit, it is on the final character of the escape  On entry,ptr is pointing at the \. On exit, it is on the final character of the
759  sequence.  escape sequence.
760    
761  Arguments:  Arguments:
762    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
763      chptr          points to the data character
764    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
765    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
766    options        the options bits    options        the options bits
767    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
768    
769  Returns:         zero or positive => a data character  Returns:         zero => a data character
770                   negative => a special escape sequence                   positive => a special escape sequence
771                     negative => a back reference
772                   on error, errorcodeptr is set                   on error, errorcodeptr is set
773  */  */
774    
775  static int  static int
776  check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
777    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
778  {  {
779  BOOL utf8 = (options & PCRE_UTF8) != 0;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
780  const uschar *ptr = *ptrptr + 1;  BOOL utf = (options & PCRE_UTF8) != 0;
781  int c, i;  const pcre_uchar *ptr = *ptrptr + 1;
782    pcre_uint32 c;
783    int escape = 0;
784    int i;
785    
786  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
787  ptr--;                            /* Set pointer back to the last byte */  ptr--;                            /* Set pointer back to the last byte */
# Line 654  in a table. A non-zero result is somethi Line 795  in a table. A non-zero result is somethi
795  Otherwise further processing may be required. */  Otherwise further processing may be required. */
796    
797  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
798  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */  /* Not alphanumeric */
799  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if (c < CHAR_0 || c > CHAR_z) {}
800    else if ((i = escapes[c - CHAR_0]) != 0) { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
801    
802  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
803  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */  /* Not alphanumeric */
804  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
805    else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
806  #endif  #endif
807    
808  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
809    
810  else  else
811    {    {
812    const uschar *oldptr;    const pcre_uchar *oldptr;
813    BOOL braced, negated;    BOOL braced, negated, overflow;
814      int s;
815    
816    switch (c)    switch (c)
817      {      {
# Line 676  else Line 820  else
820    
821      case CHAR_l:      case CHAR_l:
822      case CHAR_L:      case CHAR_L:
823        *errorcodeptr = ERR37;
824        break;
825    
826      case CHAR_u:      case CHAR_u:
827        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
828          {
829          /* In JavaScript, \u must be followed by four hexadecimal numbers.
830          Otherwise it is a lowercase u letter. */
831          if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
832            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
833            && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
834            && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
835            {
836            c = 0;
837            for (i = 0; i < 4; ++i)
838              {
839              register pcre_uint32 cc = *(++ptr);
840    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
841              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
842              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
843    #else           /* EBCDIC coding */
844              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
845              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
846    #endif
847              }
848    
849    #if defined COMPILE_PCRE8
850            if (c > (utf ? 0x10ffff : 0xff))
851    #elif defined COMPILE_PCRE16
852            if (c > (utf ? 0x10ffff : 0xffff))
853    #elif defined COMPILE_PCRE32
854            if (utf && c > 0x10ffff)
855    #endif
856              {
857              *errorcodeptr = ERR76;
858              }
859            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
860            }
861          }
862        else
863          *errorcodeptr = ERR37;
864        break;
865    
866      case CHAR_U:      case CHAR_U:
867      *errorcodeptr = ERR37;      /* In JavaScript, \U is an uppercase U letter. */
868        if ((options & PCRE_JAVASCRIPT_COMPAT) == 0) *errorcodeptr = ERR37;
869      break;      break;
870    
871      /* In a character class, \g is just a literal "g". Outside a character      /* In a character class, \g is just a literal "g". Outside a character
# Line 696  else Line 883  else
883      (3) For Oniguruma compatibility we also support \g followed by a name or a      (3) For Oniguruma compatibility we also support \g followed by a name or a
884      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
885      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
886      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
887    
888      case CHAR_g:      case CHAR_g:
889      if (isclass) break;      if (isclass) break;
890      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
891        {        {
892        c = -ESC_g;        escape = ESC_g;
893        break;        break;
894        }        }
895    
# Line 710  else Line 897  else
897    
898      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
899        {        {
900        const uschar *p;        const pcre_uchar *p;
901        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
902          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
903        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)
904          {          {
905          c = -ESC_k;          escape = ESC_k;
906          break;          break;
907          }          }
908        braced = TRUE;        braced = TRUE;
# Line 730  else Line 917  else
917        }        }
918      else negated = FALSE;      else negated = FALSE;
919    
920      c = 0;      /* The integer range is limited by the machine's int representation. */
921      while ((digitab[ptr[1]] & ctype_digit) != 0)      s = 0;
922        c = c * 10 + *(++ptr) - CHAR_0;      overflow = FALSE;
923        while (IS_DIGIT(ptr[1]))
924      if (c < 0)   /* Integer overflow */        {
925          if (s > INT_MAX / 10 - 1) /* Integer overflow */
926            {
927            overflow = TRUE;
928            break;
929            }
930          s = s * 10 + (int)(*(++ptr) - CHAR_0);
931          }
932        if (overflow) /* Integer overflow */
933        {        {
934          while (IS_DIGIT(ptr[1]))
935            ptr++;
936        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
937        break;        break;
938        }        }
# Line 746  else Line 943  else
943        break;        break;
944        }        }
945    
946      if (c == 0)      if (s == 0)
947        {        {
948        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
949        break;        break;
# Line 754  else Line 951  else
951    
952      if (negated)      if (negated)
953        {        {
954        if (c > bracount)        if (s > bracount)
955          {          {
956          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
957          break;          break;
958          }          }
959        c = bracount - (c - 1);        s = bracount - (s - 1);
960        }        }
961    
962      c = -(ESC_REF + c);      escape = -s;
963      break;      break;
964    
965      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 783  else Line 980  else
980      if (!isclass)      if (!isclass)
981        {        {
982        oldptr = ptr;        oldptr = ptr;
983        c -= CHAR_0;        /* The integer range is limited by the machine's int representation. */
984        while ((digitab[ptr[1]] & ctype_digit) != 0)        s = (int)(c -CHAR_0);
985          c = c * 10 + *(++ptr) - CHAR_0;        overflow = FALSE;
986        if (c < 0)    /* Integer overflow */        while (IS_DIGIT(ptr[1]))
987          {          {
988            if (s > INT_MAX / 10 - 1) /* Integer overflow */
989              {
990              overflow = TRUE;
991              break;
992              }
993            s = s * 10 + (int)(*(++ptr) - CHAR_0);
994            }
995          if (overflow) /* Integer overflow */
996            {
997            while (IS_DIGIT(ptr[1]))
998              ptr++;
999          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1000          break;          break;
1001          }          }
1002        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1003          {          {
1004          c = -(ESC_REF + c);          escape = -s;
1005          break;          break;
1006          }          }
1007        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 813  else Line 1021  else
1021      /* \0 always starts an octal number, but we may drop through to here with a      /* \0 always starts an octal number, but we may drop through to here with a
1022      larger first octal digit. The original code used just to take the least      larger first octal digit. The original code used just to take the least
1023      significant 8 bits of octal numbers (I think this is what early Perls used      significant 8 bits of octal numbers (I think this is what early Perls used
1024      to do). Nowadays we allow for larger numbers in UTF-8 mode, but no more      to do). Nowadays we allow for larger numbers in UTF-8 mode and 16-bit mode,
1025      than 3 octal digits. */      but no more than 3 octal digits. */
1026    
1027      case CHAR_0:      case CHAR_0:
1028      c -= CHAR_0;      c -= CHAR_0;
1029      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1030          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1031      if (!utf8 && c > 255) *errorcodeptr = ERR51;  #ifdef COMPILE_PCRE8
1032        if (!utf && c > 0xff) *errorcodeptr = ERR51;
1033    #endif
1034      break;      break;
1035    
1036      /* \x is complicated. \x{ddd} is a character number which can be greater      /* \x is complicated. \x{ddd} is a character number which can be greater
1037      than 0xff in utf8 mode, but only if the ddd are hex digits. If not, { is      than 0xff in utf or non-8bit mode, but only if the ddd are hex digits.
1038      treated as a data character. */      If not, { is treated as a data character. */
1039    
1040      case CHAR_x:      case CHAR_x:
1041        if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1042          {
1043          /* In JavaScript, \x must be followed by two hexadecimal numbers.
1044          Otherwise it is a lowercase x letter. */
1045          if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1046            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1047            {
1048            c = 0;
1049            for (i = 0; i < 2; ++i)
1050              {
1051              register pcre_uint32 cc = *(++ptr);
1052    #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1053              if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1054              c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
1055    #else           /* EBCDIC coding */
1056              if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
1057              c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1058    #endif
1059              }
1060            }
1061          break;
1062          }
1063    
1064      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1065        {        {
1066        const uschar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
       int count = 0;  
1067    
1068        c = 0;        c = 0;
1069        while ((digitab[*pt] & ctype_xdigit) != 0)        overflow = FALSE;
1070          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1071          {          {
1072          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1073          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1074          count++;  
1075    #ifdef COMPILE_PCRE32
1076            if (c >= 0x10000000l) { overflow = TRUE; break; }
1077    #endif
1078    
1079  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1080          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
# Line 847  else Line 1083  else
1083          if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */          if (cc >= CHAR_a && cc <= CHAR_z) cc += 64;  /* Convert to upper case */
1084          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1085  #endif  #endif
1086    
1087    #if defined COMPILE_PCRE8
1088            if (c > (utf ? 0x10ffff : 0xff)) { overflow = TRUE; break; }
1089    #elif defined COMPILE_PCRE16
1090            if (c > (utf ? 0x10ffff : 0xffff)) { overflow = TRUE; break; }
1091    #elif defined COMPILE_PCRE32
1092            if (utf && c > 0x10ffff) { overflow = TRUE; break; }
1093    #endif
1094            }
1095    
1096          if (overflow)
1097            {
1098            while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1099            *errorcodeptr = ERR34;
1100          }          }
1101    
1102        if (*pt == CHAR_RIGHT_CURLY_BRACKET)        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
1103          {          {
1104          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1105          ptr = pt;          ptr = pt;
1106          break;          break;
1107          }          }
# Line 863  else Line 1113  else
1113      /* Read just a single-byte hex-defined char */      /* Read just a single-byte hex-defined char */
1114    
1115      c = 0;      c = 0;
1116      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1117        {        {
1118        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1119        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1120  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1121        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 924  else Line 1174  else
1174  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1175  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1176    
1177  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1178       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1179    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1180    
1181  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1182    
1183  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1184    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1185    
1186  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1187    
1188  *ptrptr = ptr;  *ptrptr = ptr;
1189  return c;  *chptr = c;
1190    return escape;
1191  }  }
1192    
   
   
1193  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1194  /*************************************************  /*************************************************
1195  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 954  escape sequence. Line 1203  escape sequence.
1203  Argument:  Argument:
1204    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1205    negptr         points to a boolean that is set TRUE for negation else FALSE    negptr         points to a boolean that is set TRUE for negation else FALSE
1206    dptr           points to an int that is set to the detailed property value    ptypeptr       points to an unsigned int that is set to the type value
1207      pdataptr       points to an unsigned int that is set to the detailed property value
1208    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1209    
1210  Returns:         type value from ucp_type_table, or -1 for an invalid type  Returns:         TRUE if the type value was found, or FALSE for an invalid type
1211  */  */
1212    
1213  static int  static BOOL
1214  get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1215      unsigned int *pdataptr, int *errorcodeptr)
1216  {  {
1217  int c, i, bot, top;  pcre_uchar c;
1218  const uschar *ptr = *ptrptr;  int i, bot, top;
1219  char name[32];  const pcre_uchar *ptr = *ptrptr;
1220    pcre_uchar name[32];
1221    
1222  c = *(++ptr);  c = *(++ptr);
1223  if (c == 0) goto ERROR_RETURN;  if (c == 0) goto ERROR_RETURN;
# Line 982  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1234  if (c == CHAR_LEFT_CURLY_BRACKET)
1234      *negptr = TRUE;      *negptr = TRUE;
1235      ptr++;      ptr++;
1236      }      }
1237    for (i = 0; i < (int)sizeof(name) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1238      {      {
1239      c = *(++ptr);      c = *(++ptr);
1240      if (c == 0) goto ERROR_RETURN;      if (c == 0) goto ERROR_RETURN;
# Line 1006  else Line 1258  else
1258  /* Search for a recognized property name using binary chop */  /* Search for a recognized property name using binary chop */
1259    
1260  bot = 0;  bot = 0;
1261  top = _pcre_utt_size;  top = PRIV(utt_size);
1262    
1263  while (bot < top)  while (bot < top)
1264    {    {
1265      int r;
1266    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1267    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1268    if (c == 0)    if (r == 0)
1269      {      {
1270      *dptr = _pcre_utt[i].value;      *ptypeptr = PRIV(utt)[i].type;
1271      return _pcre_utt[i].type;      *pdataptr = PRIV(utt)[i].value;
1272        return TRUE;
1273      }      }
1274    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1275    }    }
1276    
1277  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1278  *ptrptr = ptr;  *ptrptr = ptr;
1279  return -1;  return FALSE;
1280    
1281  ERROR_RETURN:  ERROR_RETURN:
1282  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1283  *ptrptr = ptr;  *ptrptr = ptr;
1284  return -1;  return FALSE;
1285  }  }
1286  #endif  #endif
1287    
# Line 1053  Returns:         pointer to '}' on succe Line 1307  Returns:         pointer to '}' on succe
1307                   current ptr on error, with errorcodeptr set non-zero                   current ptr on error, with errorcodeptr set non-zero
1308  */  */
1309    
1310  static const uschar *  static const pcre_uchar *
1311  read_repeat_counts(const uschar *p, int *minp, int *maxp, int *errorcodeptr)  read_repeat_counts(const pcre_uchar *p, int *minp, int *maxp, int *errorcodeptr)
1312  {  {
1313  int min = 0;  int min = 0;
1314  int max = -1;  int max = -1;
# Line 1062  int max = -1; Line 1316  int max = -1;
1316  /* Read the minimum value and do a paranoid check: a negative value indicates  /* Read the minimum value and do a paranoid check: a negative value indicates
1317  an integer overflow. */  an integer overflow. */
1318    
1319  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1320  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1321    {    {
1322    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1077  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1331  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1331    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1332      {      {
1333      max = 0;      max = 0;
1334      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1335      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1336        {        {
1337        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1132  Arguments: Line 1386  Arguments:
1386    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1387    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1388    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1389    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1390    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1391    
1392  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1393  */  */
1394    
1395  static int  static int
1396  find_parens_sub(uschar **ptrptr, compile_data *cd, const uschar *name, int lorn,  find_parens_sub(pcre_uchar **ptrptr, compile_data *cd, const pcre_uchar *name, int lorn,
1397    BOOL xmode, BOOL utf8, int *count)    BOOL xmode, BOOL utf, int *count)
1398  {  {
1399  uschar *ptr = *ptrptr;  pcre_uchar *ptr = *ptrptr;
1400  int start_count = *count;  int start_count = *count;
1401  int hwm_count = start_count;  int hwm_count = start_count;
1402  BOOL dup_parens = FALSE;  BOOL dup_parens = FALSE;
# Line 1208  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1462  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1462      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1463          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1464        {        {
1465        int term;        pcre_uchar term;
1466        const uschar *thisname;        const pcre_uchar *thisname;
1467        *count += 1;        *count += 1;
1468        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1469        term = *ptr++;        term = *ptr++;
1470        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1471        thisname = ptr;        thisname = ptr;
1472        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1473        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1474            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1475          return *count;          return *count;
1476        term++;        term++;
1477        }        }
# Line 1260  for (; ptr < cd->end_pattern; ptr++) Line 1514  for (; ptr < cd->end_pattern; ptr++)
1514          {          {
1515          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1516            ptr+= 2;            ptr+= 2;
1517          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1518                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1519            ptr += 4;            ptr += 4;
1520          else          else
# Line 1308  for (; ptr < cd->end_pattern; ptr++) Line 1562  for (; ptr < cd->end_pattern; ptr++)
1562        {        {
1563        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1564        ptr++;        ptr++;
1565  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1566        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1567  #endif  #endif
1568        }        }
1569      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
# Line 1320  for (; ptr < cd->end_pattern; ptr++) Line 1574  for (; ptr < cd->end_pattern; ptr++)
1574    
1575    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1576      {      {
1577      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1578      if (rc > 0) return rc;      if (rc > 0) return rc;
1579      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == 0) goto FAIL_EXIT;
1580      }      }
# Line 1366  Arguments: Line 1620  Arguments:
1620    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1621    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1622    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1623    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1624    
1625  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1626  */  */
1627    
1628  static int  static int
1629  find_parens(compile_data *cd, const uschar *name, int lorn, BOOL xmode,  find_parens(compile_data *cd, const pcre_uchar *name, int lorn, BOOL xmode,
1630    BOOL utf8)    BOOL utf)
1631  {  {
1632  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1633  int count = 0;  int count = 0;
1634  int rc;  int rc;
1635    
# Line 1386  matching closing parens. That is why we Line 1640  matching closing parens. That is why we
1640    
1641  for (;;)  for (;;)
1642    {    {
1643    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1644    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == 0) break;
1645    }    }
1646    
# Line 1413  Arguments: Line 1667  Arguments:
1667  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1668  */  */
1669    
1670  static const uschar*  static const pcre_uchar*
1671  first_significant_code(const uschar *code, BOOL skipassert)  first_significant_code(const pcre_uchar *code, BOOL skipassert)
1672  {  {
1673  for (;;)  for (;;)
1674    {    {
# Line 1425  for (;;) Line 1679  for (;;)
1679      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1680      if (!skipassert) return code;      if (!skipassert) return code;
1681      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1682      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1683      break;      break;
1684    
1685      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1439  for (;;) Line 1693  for (;;)
1693      case OP_RREF:      case OP_RREF:
1694      case OP_NRREF:      case OP_NRREF:
1695      case OP_DEF:      case OP_DEF:
1696      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1697      break;      break;
1698    
1699      default:      default:
# Line 1469  and doing the check at the end; a flag s Line 1723  and doing the check at the end; a flag s
1723    
1724  Arguments:  Arguments:
1725    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1726    utf8     TRUE in UTF-8 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1727    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1728    cd       the "compile data" structure    cd       the "compile data" structure
1729    
1730  Returns:   the fixed length,  Returns:   the fixed length,
1731               or -1 if there is no fixed length,               or -1 if there is no fixed length,
1732               or -2 if \C was encountered               or -2 if \C was encountered (in UTF-8 mode only)
1733               or -3 if an OP_RECURSE item was encountered and atend is FALSE               or -3 if an OP_RECURSE item was encountered and atend is FALSE
1734                 or -4 if an unknown opcode was encountered (internal error)
1735  */  */
1736    
1737  static int  static int
1738  find_fixedlength(uschar *code, BOOL utf8, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1739  {  {
1740  int length = -1;  int length = -1;
1741    
1742  register int branchlength = 0;  register int branchlength = 0;
1743  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1744    
1745  /* Scan along the opcodes for this branch. If we get to the end of the  /* Scan along the opcodes for this branch. If we get to the end of the
1746  branch, check the length against that of the other branches. */  branch, check the length against that of the other branches. */
# Line 1493  branch, check the length against that of Line 1748  branch, check the length against that of
1748  for (;;)  for (;;)
1749    {    {
1750    int d;    int d;
1751    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1752    register int op = *cc;    register pcre_uchar op = *cc;
1753    
1754    switch (op)    switch (op)
1755      {      {
1756      /* We only need to continue for OP_CBRA (normal capturing bracket) and      /* We only need to continue for OP_CBRA (normal capturing bracket) and
1757      OP_BRA (normal non-capturing bracket) because the other variants of these      OP_BRA (normal non-capturing bracket) because the other variants of these
1758      opcodes are all concerned with unlimited repeated groups, which of course      opcodes are all concerned with unlimited repeated groups, which of course
1759      are not of fixed length. They will cause a -1 response from the default      are not of fixed length. */
     case of this switch. */  
1760    
1761      case OP_CBRA:      case OP_CBRA:
1762      case OP_BRA:      case OP_BRA:
1763      case OP_ONCE:      case OP_ONCE:
1764      case OP_ONCE_NC:      case OP_ONCE_NC:
1765      case OP_COND:      case OP_COND:
1766      d = find_fixedlength(cc + ((op == OP_CBRA)? 2:0), utf8, atend, cd);      d = find_fixedlength(cc + ((op == OP_CBRA)? IMM2_SIZE : 0), utf, atend, cd);
1767      if (d < 0) return d;      if (d < 0) return d;
1768      branchlength += d;      branchlength += d;
1769      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1770      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
1771      break;      break;
1772    
1773      /* Reached end of a branch; if it's a ket it is the end of a nested      /* Reached end of a branch; if it's a ket it is the end of a nested call.
1774      call. If it's ALT it is an alternation in a nested call. If it is      If it's ALT it is an alternation in a nested call. An ACCEPT is effectively
1775      END it's the end of the outer call. All can be handled by the same code.      an ALT. If it is END it's the end of the outer call. All can be handled by
1776      Note that we must not include the OP_KETRxxx opcodes here, because they      the same code. Note that we must not include the OP_KETRxxx opcodes here,
1777      all imply an unlimited repeat. */      because they all imply an unlimited repeat. */
1778    
1779      case OP_ALT:      case OP_ALT:
1780      case OP_KET:      case OP_KET:
1781      case OP_END:      case OP_END:
1782        case OP_ACCEPT:
1783        case OP_ASSERT_ACCEPT:
1784      if (length < 0) length = branchlength;      if (length < 0) length = branchlength;
1785        else if (length != branchlength) return -1;        else if (length != branchlength) return -1;
1786      if (*cc != OP_ALT) return length;      if (*cc != OP_ALT) return length;
# Line 1537  for (;;) Line 1794  for (;;)
1794    
1795      case OP_RECURSE:      case OP_RECURSE:
1796      if (!atend) return -3;      if (!atend) return -3;
1797      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1798      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1799      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1800      d = find_fixedlength(cs + 2, utf8, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1801      if (d < 0) return d;      if (d < 0) return d;
1802      branchlength += d;      branchlength += d;
1803      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1553  for (;;) Line 1810  for (;;)
1810      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1811      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1812      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1813      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1814        break;
1815    
1816      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1817    
1818      case OP_REVERSE:      case OP_MARK:
1819      case OP_CREF:      case OP_PRUNE_ARG:
1820      case OP_NCREF:      case OP_SKIP_ARG:
1821      case OP_RREF:      case OP_THEN_ARG:
1822      case OP_NRREF:      cc += cc[1] + PRIV(OP_lengths)[*cc];
1823      case OP_DEF:      break;
1824    
1825      case OP_CALLOUT:      case OP_CALLOUT:
     case OP_SOD:  
     case OP_SOM:  
     case OP_SET_SOM:  
     case OP_EOD:  
     case OP_EODN:  
1826      case OP_CIRC:      case OP_CIRC:
1827      case OP_CIRCM:      case OP_CIRCM:
1828        case OP_CLOSE:
1829        case OP_COMMIT:
1830        case OP_CREF:
1831        case OP_DEF:
1832      case OP_DOLL:      case OP_DOLL:
1833      case OP_DOLLM:      case OP_DOLLM:
1834        case OP_EOD:
1835        case OP_EODN:
1836        case OP_FAIL:
1837        case OP_NCREF:
1838        case OP_NRREF:
1839      case OP_NOT_WORD_BOUNDARY:      case OP_NOT_WORD_BOUNDARY:
1840        case OP_PRUNE:
1841        case OP_REVERSE:
1842        case OP_RREF:
1843        case OP_SET_SOM:
1844        case OP_SKIP:
1845        case OP_SOD:
1846        case OP_SOM:
1847        case OP_THEN:
1848      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1849      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1850      break;      break;
1851    
1852      /* Handle literal characters */      /* Handle literal characters */
# Line 1586  for (;;) Line 1857  for (;;)
1857      case OP_NOTI:      case OP_NOTI:
1858      branchlength++;      branchlength++;
1859      cc += 2;      cc += 2;
1860  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1861      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1862  #endif  #endif
1863      break;      break;
1864    
# Line 1595  for (;;) Line 1866  for (;;)
1866      need to skip over a multibyte character in UTF8 mode.  */      need to skip over a multibyte character in UTF8 mode.  */
1867    
1868      case OP_EXACT:      case OP_EXACT:
1869      branchlength += GET2(cc,1);      case OP_EXACTI:
1870      cc += 4;      case OP_NOTEXACT:
1871  #ifdef SUPPORT_UTF8      case OP_NOTEXACTI:
1872      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      branchlength += (int)GET2(cc,1);
1873        cc += 2 + IMM2_SIZE;
1874    #ifdef SUPPORT_UTF
1875        if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1876  #endif  #endif
1877      break;      break;
1878    
1879      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1880      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1881      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1882      cc += 4;        cc += 2;
1883        cc += 1 + IMM2_SIZE + 1;
1884      break;      break;
1885    
1886      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1615  for (;;) Line 1890  for (;;)
1890      cc += 2;      cc += 2;
1891      /* Fall through */      /* Fall through */
1892    
1893        case OP_HSPACE:
1894        case OP_VSPACE:
1895        case OP_NOT_HSPACE:
1896        case OP_NOT_VSPACE:
1897      case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
1898      case OP_DIGIT:      case OP_DIGIT:
1899      case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
# Line 1627  for (;;) Line 1906  for (;;)
1906      cc++;      cc++;
1907      break;      break;
1908    
1909      /* The single-byte matcher isn't allowed */      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;
1910        otherwise \C is coded as OP_ALLANY. */
1911    
1912      case OP_ANYBYTE:      case OP_ANYBYTE:
1913      return -2;      return -2;
1914    
1915      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1916    
1917  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1918      case OP_XCLASS:      case OP_XCLASS:
     cc += GET(cc, 1) - 33;  
     /* Fall through */  
1919  #endif  #endif
   
1920      case OP_CLASS:      case OP_CLASS:
1921      case OP_NCLASS:      case OP_NCLASS:
1922      cc += 33;  
1923        switch (op)
1924          {
1925    #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1926          case OP_XCLASS:
1927          cc += GET(cc, 1);
1928          break;
1929    #endif
1930    
1931          case OP_CLASS:
1932          case OP_NCLASS:
1933          cc += PRIV(OP_lengths)[OP_CLASS];
1934          break;
1935          }
1936    
1937      switch (*cc)      switch (*cc)
1938        {        {
1939          case OP_CRPLUS:
1940          case OP_CRMINPLUS:
1941        case OP_CRSTAR:        case OP_CRSTAR:
1942        case OP_CRMINSTAR:        case OP_CRMINSTAR:
1943        case OP_CRQUERY:        case OP_CRQUERY:
# Line 1654  for (;;) Line 1946  for (;;)
1946    
1947        case OP_CRRANGE:        case OP_CRRANGE:
1948        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1949        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1950        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1951        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1952        break;        break;
1953    
1954        default:        default:
# Line 1666  for (;;) Line 1958  for (;;)
1958    
1959      /* Anything else is variable length */      /* Anything else is variable length */
1960    
1961      default:      case OP_ANYNL:
1962        case OP_BRAMINZERO:
1963        case OP_BRAPOS:
1964        case OP_BRAPOSZERO:
1965        case OP_BRAZERO:
1966        case OP_CBRAPOS:
1967        case OP_EXTUNI:
1968        case OP_KETRMAX:
1969        case OP_KETRMIN:
1970        case OP_KETRPOS:
1971        case OP_MINPLUS:
1972        case OP_MINPLUSI:
1973        case OP_MINQUERY:
1974        case OP_MINQUERYI:
1975        case OP_MINSTAR:
1976        case OP_MINSTARI:
1977        case OP_MINUPTO:
1978        case OP_MINUPTOI:
1979        case OP_NOTMINPLUS:
1980        case OP_NOTMINPLUSI:
1981        case OP_NOTMINQUERY:
1982        case OP_NOTMINQUERYI:
1983        case OP_NOTMINSTAR:
1984        case OP_NOTMINSTARI:
1985        case OP_NOTMINUPTO:
1986        case OP_NOTMINUPTOI:
1987        case OP_NOTPLUS:
1988        case OP_NOTPLUSI:
1989        case OP_NOTPOSPLUS:
1990        case OP_NOTPOSPLUSI:
1991        case OP_NOTPOSQUERY:
1992        case OP_NOTPOSQUERYI:
1993        case OP_NOTPOSSTAR:
1994        case OP_NOTPOSSTARI:
1995        case OP_NOTPOSUPTO:
1996        case OP_NOTPOSUPTOI:
1997        case OP_NOTQUERY:
1998        case OP_NOTQUERYI:
1999        case OP_NOTSTAR:
2000        case OP_NOTSTARI:
2001        case OP_NOTUPTO:
2002        case OP_NOTUPTOI:
2003        case OP_PLUS:
2004        case OP_PLUSI:
2005        case OP_POSPLUS:
2006        case OP_POSPLUSI:
2007        case OP_POSQUERY:
2008        case OP_POSQUERYI:
2009        case OP_POSSTAR:
2010        case OP_POSSTARI:
2011        case OP_POSUPTO:
2012        case OP_POSUPTOI:
2013        case OP_QUERY:
2014        case OP_QUERYI:
2015        case OP_REF:
2016        case OP_REFI:
2017        case OP_SBRA:
2018        case OP_SBRAPOS:
2019        case OP_SCBRA:
2020        case OP_SCBRAPOS:
2021        case OP_SCOND:
2022        case OP_SKIPZERO:
2023        case OP_STAR:
2024        case OP_STARI:
2025        case OP_TYPEMINPLUS:
2026        case OP_TYPEMINQUERY:
2027        case OP_TYPEMINSTAR:
2028        case OP_TYPEMINUPTO:
2029        case OP_TYPEPLUS:
2030        case OP_TYPEPOSPLUS:
2031        case OP_TYPEPOSQUERY:
2032        case OP_TYPEPOSSTAR:
2033        case OP_TYPEPOSUPTO:
2034        case OP_TYPEQUERY:
2035        case OP_TYPESTAR:
2036        case OP_TYPEUPTO:
2037        case OP_UPTO:
2038        case OP_UPTOI:
2039      return -1;      return -1;
2040    
2041        /* Catch unrecognized opcodes so that when new ones are added they
2042        are not forgotten, as has happened in the past. */
2043    
2044        default:
2045        return -4;
2046      }      }
2047    }    }
2048  /* Control never gets here */  /* Control never gets here */
# Line 1688  length. Line 2063  length.
2063    
2064  Arguments:  Arguments:
2065    code        points to start of expression    code        points to start of expression
2066    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2067    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2068    
2069  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
2070  */  */
2071    
2072  const uschar *  const pcre_uchar *
2073  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2074  {  {
2075  for (;;)  for (;;)
2076    {    {
2077    register int c = *code;    register pcre_uchar c = *code;
2078    
2079    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2080    
# Line 1713  for (;;) Line 2088  for (;;)
2088    
2089    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2090      {      {
2091      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2092      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2093      }      }
2094    
2095    /* Handle capturing bracket */    /* Handle capturing bracket */
# Line 1722  for (;;) Line 2097  for (;;)
2097    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2098             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2099      {      {
2100      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2101      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2102      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2103      }      }
2104    
2105    /* 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 1752  for (;;) Line 2127  for (;;)
2127        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2128        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2129        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2130        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2131            code += 2;
2132        break;        break;
2133    
2134        case OP_MARK:        case OP_MARK:
# Line 1768  for (;;) Line 2144  for (;;)
2144    
2145      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2146    
2147      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2148    
2149    /* 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
2150    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
2151    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2152    
2153  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2154      if (utf8) switch(c)      if (utf) switch(c)
2155        {        {
2156        case OP_CHAR:        case OP_CHAR:
2157        case OP_CHARI:        case OP_CHARI:
# Line 1805  for (;;) Line 2181  for (;;)
2181        case OP_MINQUERYI:        case OP_MINQUERYI:
2182        case OP_POSQUERY:        case OP_POSQUERY:
2183        case OP_POSQUERYI:        case OP_POSQUERYI:
2184        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2185        break;        break;
2186        }        }
2187  #else  #else
2188      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2189  #endif  #endif
2190      }      }
2191    }    }
# Line 1826  instance of OP_RECURSE. Line 2202  instance of OP_RECURSE.
2202    
2203  Arguments:  Arguments:
2204    code        points to start of expression    code        points to start of expression
2205    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2206    
2207  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
2208  */  */
2209    
2210  static const uschar *  static const pcre_uchar *
2211  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2212  {  {
2213  for (;;)  for (;;)
2214    {    {
2215    register int c = *code;    register pcre_uchar c = *code;
2216    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2217    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2218    
# Line 1871  for (;;) Line 2247  for (;;)
2247        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2248        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2249        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2250        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2251            code += 2;
2252        break;        break;
2253    
2254        case OP_MARK:        case OP_MARK:
# Line 1887  for (;;) Line 2264  for (;;)
2264    
2265      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2266    
2267      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2268    
2269      /* 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
2270      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
2271      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2272    
2273  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2274      if (utf8) switch(c)      if (utf) switch(c)
2275        {        {
2276        case OP_CHAR:        case OP_CHAR:
2277        case OP_CHARI:        case OP_CHARI:
2278          case OP_NOT:
2279          case OP_NOTI:
2280        case OP_EXACT:        case OP_EXACT:
2281        case OP_EXACTI:        case OP_EXACTI:
2282          case OP_NOTEXACT:
2283          case OP_NOTEXACTI:
2284        case OP_UPTO:        case OP_UPTO:
2285        case OP_UPTOI:        case OP_UPTOI:
2286          case OP_NOTUPTO:
2287          case OP_NOTUPTOI:
2288        case OP_MINUPTO:        case OP_MINUPTO:
2289        case OP_MINUPTOI:        case OP_MINUPTOI:
2290          case OP_NOTMINUPTO:
2291          case OP_NOTMINUPTOI:
2292        case OP_POSUPTO:        case OP_POSUPTO:
2293        case OP_POSUPTOI:        case OP_POSUPTOI:
2294          case OP_NOTPOSUPTO:
2295          case OP_NOTPOSUPTOI:
2296        case OP_STAR:        case OP_STAR:
2297        case OP_STARI:        case OP_STARI:
2298          case OP_NOTSTAR:
2299          case OP_NOTSTARI:
2300        case OP_MINSTAR:        case OP_MINSTAR:
2301        case OP_MINSTARI:        case OP_MINSTARI:
2302          case OP_NOTMINSTAR:
2303          case OP_NOTMINSTARI:
2304        case OP_POSSTAR:        case OP_POSSTAR:
2305        case OP_POSSTARI:        case OP_POSSTARI:
2306          case OP_NOTPOSSTAR:
2307          case OP_NOTPOSSTARI:
2308        case OP_PLUS:        case OP_PLUS:
2309        case OP_PLUSI:        case OP_PLUSI:
2310          case OP_NOTPLUS:
2311          case OP_NOTPLUSI:
2312        case OP_MINPLUS:        case OP_MINPLUS:
2313        case OP_MINPLUSI:        case OP_MINPLUSI:
2314          case OP_NOTMINPLUS:
2315          case OP_NOTMINPLUSI:
2316        case OP_POSPLUS:        case OP_POSPLUS:
2317        case OP_POSPLUSI:        case OP_POSPLUSI:
2318          case OP_NOTPOSPLUS:
2319          case OP_NOTPOSPLUSI:
2320        case OP_QUERY:        case OP_QUERY:
2321        case OP_QUERYI:        case OP_QUERYI:
2322          case OP_NOTQUERY:
2323          case OP_NOTQUERYI:
2324        case OP_MINQUERY:        case OP_MINQUERY:
2325        case OP_MINQUERYI:        case OP_MINQUERYI:
2326          case OP_NOTMINQUERY:
2327          case OP_NOTMINQUERYI:
2328        case OP_POSQUERY:        case OP_POSQUERY:
2329        case OP_POSQUERYI:        case OP_POSQUERYI:
2330        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_NOTPOSQUERY:
2331          case OP_NOTPOSQUERYI:
2332          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2333        break;        break;
2334        }        }
2335  #else  #else
2336      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2337  #endif  #endif
2338      }      }
2339    }    }
# Line 1951  bracket whose current branch will alread Line 2356  bracket whose current branch will alread
2356  Arguments:  Arguments:
2357    code        points to start of search    code        points to start of search
2358    endcode     points to where to stop    endcode     points to where to stop
2359    utf8        TRUE if in UTF8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2360    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2361    
2362  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2363  */  */
2364    
2365  static BOOL  static BOOL
2366  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2367    compile_data *cd)    BOOL utf, compile_data *cd)
2368  {  {
2369  register int c;  register pcre_uchar c;
2370  for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2371       code < endcode;       code < endcode;
2372       code = first_significant_code(code + _pcre_OP_lengths[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2373    {    {
2374    const uschar *ccode;    const pcre_uchar *ccode;
2375    
2376    c = *code;    c = *code;
2377    
# Line 1989  for (code = first_significant_code(code Line 2394  for (code = first_significant_code(code
2394    
2395    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2396      {      {
2397      const uschar *scode;      const pcre_uchar *scode;
2398      BOOL empty_branch;      BOOL empty_branch;
2399    
2400      /* Test for forward reference */      /* Test for forward reference */
2401    
2402      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2403        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2404    
2405      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2406    
# Line 2007  for (code = first_significant_code(code Line 2412  for (code = first_significant_code(code
2412    
2413      do      do
2414        {        {
2415        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd))
2416          {          {
2417          empty_branch = TRUE;          empty_branch = TRUE;
2418          break;          break;
# Line 2025  for (code = first_significant_code(code Line 2430  for (code = first_significant_code(code
2430    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2431        c == OP_BRAPOSZERO)        c == OP_BRAPOSZERO)
2432      {      {
2433      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2434      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
2435      c = *code;      c = *code;
2436      continue;      continue;
# Line 2063  for (code = first_significant_code(code Line 2468  for (code = first_significant_code(code
2468        empty_branch = FALSE;        empty_branch = FALSE;
2469        do        do
2470          {          {
2471          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
2472            empty_branch = TRUE;            empty_branch = TRUE;
2473          code += GET(code, 1);          code += GET(code, 1);
2474          }          }
# Line 2081  for (code = first_significant_code(code Line 2486  for (code = first_significant_code(code
2486      {      {
2487      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2488      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2489      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2490      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"
2491      here. */      here. */
2492    
2493  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2494      case OP_XCLASS:      case OP_XCLASS:
2495      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2496      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2093  for (code = first_significant_code(code Line 2498  for (code = first_significant_code(code
2498    
2499      case OP_CLASS:      case OP_CLASS:
2500      case OP_NCLASS:      case OP_NCLASS:
2501      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2502    
2503  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2504      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2505  #endif  #endif
2506    
# Line 2168  for (code = first_significant_code(code Line 2573  for (code = first_significant_code(code
2573      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2574      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2575      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2576      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2577          code += 2;
2578      break;      break;
2579    
2580      /* End of branch */      /* End of branch */
# Line 2183  for (code = first_significant_code(code Line 2589  for (code = first_significant_code(code
2589      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2590      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2591    
2592  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2593      case OP_STAR:      case OP_STAR:
2594      case OP_STARI:      case OP_STARI:
2595      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2196  for (code = first_significant_code(code Line 2602  for (code = first_significant_code(code
2602      case OP_MINQUERYI:      case OP_MINQUERYI:
2603      case OP_POSQUERY:      case OP_POSQUERY:
2604      case OP_POSQUERYI:      case OP_POSQUERYI:
2605      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2606      break;      break;
2607    
2608      case OP_UPTO:      case OP_UPTO:
# Line 2205  for (code = first_significant_code(code Line 2611  for (code = first_significant_code(code
2611      case OP_MINUPTOI:      case OP_MINUPTOI:
2612      case OP_POSUPTO:      case OP_POSUPTO:
2613      case OP_POSUPTOI:      case OP_POSUPTOI:
2614      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]);
2615      break;      break;
2616  #endif  #endif
2617    
# Line 2249  Arguments: Line 2655  Arguments:
2655    code        points to start of the recursion    code        points to start of the recursion
2656    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2657    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2658    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2659    cd          pointers to tables etc    cd          pointers to tables etc
2660    
2661  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2662  */  */
2663    
2664  static BOOL  static BOOL
2665  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2666    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2667  {  {
2668  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2669    {    {
2670    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
2671      return FALSE;      return FALSE;
2672    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2673    }    }
# Line 2313  Returns:   TRUE or FALSE Line 2719  Returns:   TRUE or FALSE
2719  */  */
2720    
2721  static BOOL  static BOOL
2722  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2723  {  {
2724  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2725  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2726  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != 0; ptr++)
2727    {    {
# Line 2357  Returns:     a value representing the na Line 2763  Returns:     a value representing the na
2763  */  */
2764    
2765  static int  static int
2766  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2767  {  {
2768  const char *pn = posix_names;  const char *pn = posix_names;
2769  register int yield = 0;  register int yield = 0;
2770  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2771    {    {
2772    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2773      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2774    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2775    yield++;    yield++;
2776    }    }
# Line 2396  value in the reference (which is a group Line 2802  value in the reference (which is a group
2802  Arguments:  Arguments:
2803    group      points to the start of the group    group      points to the start of the group
2804    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2805    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2806    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2807    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
2808    
# Line 2404  Returns:     nothing Line 2810  Returns:     nothing
2810  */  */
2811    
2812  static void  static void
2813  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2814    uschar *save_hwm)    pcre_uchar *save_hwm)
2815  {  {
2816  uschar *ptr = group;  pcre_uchar *ptr = group;
2817    
2818  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2819    {    {
2820    int offset;    int offset;
2821    uschar *hc;    pcre_uchar *hc;
2822    
2823    /* 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
2824    reference. */    reference. */
2825    
2826    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2827      {      {
2828      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2829      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2830        {        {
2831        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2432  while ((ptr = (uschar *)find_recurse(ptr Line 2838  while ((ptr = (uschar *)find_recurse(ptr
2838    
2839    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2840      {      {
2841      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2842      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2843      }      }
2844    
# Line 2457  Arguments: Line 2863  Arguments:
2863  Returns:         new code pointer  Returns:         new code pointer
2864  */  */
2865    
2866  static uschar *  static pcre_uchar *
2867  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2868  {  {
2869  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2870  *code++ = 255;  *code++ = 255;
2871  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2872  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2873  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2874  }  }
2875    
2876    
# Line 2486  Returns:             nothing Line 2892  Returns:             nothing
2892  */  */
2893    
2894  static void  static void
2895  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2896  {  {
2897  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2898  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2500  PUT(previous_callout, 2 + LINK_SIZE, len Line 2906  PUT(previous_callout, 2 + LINK_SIZE, len
2906  *************************************************/  *************************************************/
2907    
2908  /* 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
2909  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
2910  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
2911  start address.  start address. A character with multiple other cases is returned on its own
2912    with a special return value.
2913    
2914  Arguments:  Arguments:
2915    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2510  Arguments: Line 2917  Arguments:
2917    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2918    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2919    
2920  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2921                   0 when a range is returned
2922                  >0 the CASESET offset for char with multiple other cases
2923                    in this case, ocptr contains the original
2924  */  */
2925    
2926  static BOOL  static int
2927  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2928    unsigned int *odptr)    pcre_uint32 *odptr)
2929  {  {
2930  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2931    unsigned int co;
2932    
2933    /* Find the first character that has an other case. If it has multiple other
2934    cases, return its case offset value. */
2935    
2936  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2937    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2938      if ((co = UCD_CASESET(c)) != 0)
2939        {
2940        *ocptr = c++;   /* Character that has the set */
2941        *cptr = c;      /* Rest of input range */
2942        return (int)co;
2943        }
2944      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2945      }
2946    
2947  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2948    
2949  *ocptr = othercase;  *ocptr = othercase;
2950  next = othercase + 1;  next = othercase + 1;
# Line 2533  for (++c; c <= d; c++) Line 2955  for (++c; c <= d; c++)
2955    next++;    next++;
2956    }    }
2957    
2958  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2959  *cptr = c;  *cptr = c;             /* Rest of input range */
2960    return 0;
 return TRUE;  
2961  }  }
2962    
2963    
# Line 2558  Returns:       TRUE if auto-possessifyin Line 2979  Returns:       TRUE if auto-possessifyin
2979  */  */
2980    
2981  static BOOL  static BOOL
2982  check_char_prop(int c, int ptype, int pdata, BOOL negated)  check_char_prop(pcre_uint32 c, unsigned int ptype, unsigned int pdata, BOOL negated)
2983  {  {
2984    #ifdef SUPPORT_UCP
2985    const pcre_uint32 *p;
2986    #endif
2987    
2988  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2989    
2990  switch(ptype)  switch(ptype)
2991    {    {
2992    case PT_LAMP:    case PT_LAMP:
# Line 2569  switch(ptype) Line 2995  switch(ptype)
2995            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
2996    
2997    case PT_GC:    case PT_GC:
2998    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2999    
3000    case PT_PC:    case PT_PC:
3001    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2580  switch(ptype) Line 3006  switch(ptype)
3006    /* These are specials */    /* These are specials */
3007    
3008    case PT_ALNUM:    case PT_ALNUM:
3009    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3010            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
3011    
3012    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
3013    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3014            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
3015            == negated;            == negated;
3016    
3017    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
3018    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3019            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
3020            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
3021            == negated;            == negated;
3022    
3023    case PT_WORD:    case PT_WORD:
3024    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3025            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3026            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3027    
3028    #ifdef SUPPORT_UCP
3029      case PT_CLIST:
3030      p = PRIV(ucd_caseless_sets) + prop->caseset;
3031      for (;;)
3032        {
3033        if ((unsigned int)c < *p) return !negated;
3034        if ((unsigned int)c == *p++) return negated;
3035        }
3036      break;  /* Control never reaches here */
3037    #endif
3038    }    }
3039    
3040  return FALSE;  return FALSE;
3041  }  }
3042  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2615  sense to automatically possessify the re Line 3053  sense to automatically possessify the re
3053    
3054  Arguments:  Arguments:
3055    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3056    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3057    ptr           next character in pattern    ptr           next character in pattern
3058    options       options bits    options       options bits
3059    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2624  Returns:        TRUE if possessifying is Line 3062  Returns:        TRUE if possessifying is
3062  */  */
3063    
3064  static BOOL  static BOOL
3065  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3066    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3067  {  {
3068  int c, next;  pcre_uint32 c = NOTACHAR;
3069  int op_code = *previous++;  pcre_uint32 next;
3070    int escape;
3071    pcre_uchar op_code = *previous++;
3072    
3073  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3074    
# Line 2636  if ((options & PCRE_EXTENDED) != 0) Line 3076  if ((options & PCRE_EXTENDED) != 0)
3076    {    {
3077    for (;;)    for (;;)
3078      {      {
3079      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3080      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3081        {        {
3082        ptr++;        ptr++;
# Line 2644  if ((options & PCRE_EXTENDED) != 0) Line 3084  if ((options & PCRE_EXTENDED) != 0)
3084          {          {
3085          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3086          ptr++;          ptr++;
3087  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3088          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3089  #endif  #endif
3090          }          }
3091        }        }
# Line 2659  value is a character, a negative value i Line 3099  value is a character, a negative value i
3099  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3100    {    {
3101    int temperrorcode = 0;    int temperrorcode = 0;
3102    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options, FALSE);
3103    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3104    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3105    }    }
3106    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
3107    {    {
3108  #ifdef SUPPORT_UTF8    escape = 0;
3109    if (utf8) { GETCHARINC(next, ptr); } else  #ifdef SUPPORT_UTF
3110      if (utf) { GETCHARINC(next, ptr); } else
3111  #endif  #endif
3112    next = *ptr++;    next = *ptr++;
3113    }    }
   
3114  else return FALSE;  else return FALSE;
3115    
3116  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2680  if ((options & PCRE_EXTENDED) != 0) Line 3119  if ((options & PCRE_EXTENDED) != 0)
3119    {    {
3120    for (;;)    for (;;)
3121      {      {
3122      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3123      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3124        {        {
3125        ptr++;        ptr++;
# Line 2688  if ((options & PCRE_EXTENDED) != 0) Line 3127  if ((options & PCRE_EXTENDED) != 0)
3127          {          {
3128          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3129          ptr++;          ptr++;
3130  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3131          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3132  #endif  #endif
3133          }          }
3134        }        }
# Line 2700  if ((options & PCRE_EXTENDED) != 0) Line 3139  if ((options & PCRE_EXTENDED) != 0)
3139  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3140    
3141  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3142    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)
3143      return FALSE;      return FALSE;
3144    
3145  /* 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. */  
3146    
3147  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3148        op_code == OP_NOT || op_code == OP_NOTI)
3149    {    {
3150    case OP_CHAR:  #ifdef SUPPORT_UTF
 #ifdef SUPPORT_UTF8  
3151    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3152  #else  #else
3153    c = *previous;    c = *previous;
3154  #endif  #endif
3155    return c != next;    }
3156    
3157    /* 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
3158    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3159    
3160    case OP_CHARI:  if (escape == 0)
3161  #ifdef SUPPORT_UTF8    {
3162    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3163  #else    case, which maps to the special PT_CLIST property. Check this first. */
3164    c = *previous;  
3165    #ifdef SUPPORT_UCP
3166      if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3167        {
3168        unsigned int ocs = UCD_CASESET(next);
3169        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3170        }
3171  #endif  #endif
3172    if (c == next) return FALSE;  
3173  #ifdef SUPPORT_UTF8    switch(op_code)
   if (utf8)  
3174      {      {
3175      unsigned int othercase;      case OP_CHAR:
3176      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3177    
3178        /* For CHARI (caseless character) we must check the other case. If we have
3179        Unicode property support, we can use it to test the other case of
3180        high-valued characters. We know that next can have only one other case,
3181        because multi-other-case characters are dealt with above. */
3182    
3183        case OP_CHARI:
3184        if (c == next) return FALSE;
3185    #ifdef SUPPORT_UTF
3186        if (utf)
3187          {
3188          pcre_uint32 othercase;
3189          if (next < 128) othercase = cd->fcc[next]; else
3190  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3191      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3192  #else  #else
3193      othercase = NOTACHAR;        othercase = NOTACHAR;
3194  #endif  #endif
3195      return (unsigned int)c != othercase;        return c != othercase;
3196      }        }
3197    else      else
3198  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3199    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3200    
3201    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These      case OP_NOT:
3202    opcodes are not used for multi-byte characters, because they are coded using      return c == next;
3203    an XCLASS instead. */  
3204        case OP_NOTI:
3205    case OP_NOT:      if (c == next) return TRUE;
3206    return (c = *previous) == next;  #ifdef SUPPORT_UTF
3207        if (utf)
3208    case OP_NOTI:        {
3209    if ((c = *previous) == next) return TRUE;        pcre_uint32 othercase;
3210  #ifdef SUPPORT_UTF8        if (next < 128) othercase = cd->fcc[next]; else
   if (utf8)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3211  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3212      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE(next);
3213  #else  #else
3214      othercase = NOTACHAR;        othercase = NOTACHAR;
3215  #endif  #endif
3216      return (unsigned int)c == othercase;        return c == othercase;
3217      }        }
3218    else      else
3219  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3220    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3221    
3222    /* 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.
3223    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. */
3224    
3225    case OP_DIGIT:      case OP_DIGIT:
3226    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3227    
3228    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3229    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3230    
3231    case OP_WHITESPACE:      case OP_WHITESPACE:
3232    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3233    
3234    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3235    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3236    
3237    case OP_WORDCHAR:      case OP_WORDCHAR:
3238    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3239    
3240    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3241    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3242    
3243    case OP_HSPACE:      case OP_HSPACE:
3244    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3245    switch(next)      switch(next)
3246      {        {
3247      case 0x09:        HSPACE_CASES:
3248      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;  
     }  
3249    
3250    case OP_ANYNL:        default:
3251    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3252    case OP_NOT_VSPACE:        }
3253    switch(next)  
3254      {      case OP_ANYNL:
3255      case 0x0a:      case OP_VSPACE:
3256      case 0x0b:      case OP_NOT_VSPACE:
3257      case 0x0c:      switch(next)
3258      case 0x0d:        {
3259      case 0x85:        VSPACE_CASES:
3260      case 0x2028:        return op_code == OP_NOT_VSPACE;
3261      case 0x2029:  
3262      return op_code == OP_NOT_VSPACE;        default:
3263      default:        return op_code != OP_NOT_VSPACE;
3264      return op_code != OP_NOT_VSPACE;        }
     }  
3265    
3266  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3267    case OP_PROP:      case OP_PROP:
3268    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3269    
3270    case OP_NOTPROP:      case OP_NOTPROP:
3271    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3272  #endif  #endif
3273    
3274    default:      default:
3275    return FALSE;      return FALSE;
3276        }
3277    }    }
3278    
   
3279  /* 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
3280  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
3281  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 2857  switch(op_code) Line 3286  switch(op_code)
3286    {    {
3287    case OP_CHAR:    case OP_CHAR:
3288    case OP_CHARI:    case OP_CHARI:
3289  #ifdef SUPPORT_UTF8    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3290      {      {
3291      case ESC_d:      case ESC_d:
3292      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3293    
3294      case ESC_D:      case ESC_D:
3295      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3296    
3297      case ESC_s:      case ESC_s:
3298      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3299    
3300      case ESC_S:      case ESC_S:
3301      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3302    
3303      case ESC_w:      case ESC_w:
3304      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3305    
3306      case ESC_W:      case ESC_W:
3307      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3308    
3309      case ESC_h:      case ESC_h:
3310      case ESC_H:      case ESC_H:
3311      switch(c)      switch(c)
3312        {        {
3313        case 0x09:        HSPACE_CASES:
3314        case 0x20:        return escape != ESC_h;
3315        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;  
3316        default:        default:
3317        return -next == ESC_h;        return escape == ESC_h;
3318        }        }
3319    
3320      case ESC_v:      case ESC_v:
3321      case ESC_V:      case ESC_V:
3322      switch(c)      switch(c)
3323        {        {
3324        case 0x0a:        VSPACE_CASES:
3325        case 0x0b:        return escape != ESC_v;
3326        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3327        default:        default:
3328        return -next == ESC_v;        return escape == ESC_v;
3329        }        }
3330    
3331      /* 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
3332      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3333      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3334    
3335  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3336      case ESC_du:      case ESC_du:
# Line 2939  switch(op_code) Line 3341  switch(op_code)
3341      case ESC_SU:      case ESC_SU:
3342        {        {
3343        int temperrorcode = 0;        int temperrorcode = 0;
3344        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3345        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3346        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3347        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3348        }        }
# Line 2949  switch(op_code) Line 3351  switch(op_code)
3351      case ESC_p:      case ESC_p:
3352      case ESC_P:      case ESC_P:
3353        {        {
3354        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3355          int errorcodeptr;
3356        BOOL negated;        BOOL negated;
3357    
3358        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3359        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3360        if (ptype < 0) return FALSE;          return FALSE;
3361        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3362    
3363        /* If the property item is optional, we have to give up. (When generated        /* If the property item is optional, we have to give up. (When generated
# Line 2962  switch(op_code) Line 3365  switch(op_code)
3365        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. */
3366    
3367        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3368          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)
3369            return FALSE;            return FALSE;
3370    
3371        /* Do the property check. */        /* Do the property check. */
3372    
3373        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3374        }        }
3375  #endif  #endif
3376    
# Line 2982  switch(op_code) Line 3385  switch(op_code)
3385    these op-codes are never generated.) */    these op-codes are never generated.) */
3386    
3387    case OP_DIGIT:    case OP_DIGIT:
3388    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3389           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3390    
3391    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3392    return next == -ESC_d;    return escape == ESC_d;
3393    
3394    case OP_WHITESPACE:    case OP_WHITESPACE:
3395    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3396    
3397    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3398    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3399    
3400    case OP_HSPACE:    case OP_HSPACE:
3401    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3402           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3403    
3404    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3405    return next == -ESC_h;    return escape == ESC_h;
3406    
3407    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3408    case OP_ANYNL:    case OP_ANYNL:
3409    case OP_VSPACE:    case OP_VSPACE:
3410    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3411    
3412    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3413    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3414    
3415    case OP_WORDCHAR:    case OP_WORDCHAR:
3416    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3417           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3418    
3419    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3420    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3421    
3422    default:    default:
3423    return FALSE;    return FALSE;
# Line 3026  switch(op_code) Line 3429  switch(op_code)
3429    
3430    
3431  /*************************************************  /*************************************************
3432    *        Add a character or range to a class     *
3433    *************************************************/
3434    
3435    /* This function packages up the logic of adding a character or range of
3436    characters to a class. The character values in the arguments will be within the
3437    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3438    mutually recursive with the function immediately below.
3439    
3440    Arguments:
3441      classbits     the bit map for characters < 256
3442      uchardptr     points to the pointer for extra data
3443      options       the options word
3444      cd            contains pointers to tables etc.
3445      start         start of range character
3446      end           end of range character
3447    
3448    Returns:        the number of < 256 characters added
3449                    the pointer to extra data is updated
3450    */
3451    
3452    static int
3453    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3454      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3455    {
3456    pcre_uint32 c;
3457    int n8 = 0;
3458    
3459    /* If caseless matching is required, scan the range and process alternate
3460    cases. In Unicode, there are 8-bit characters that have alternate cases that
3461    are greater than 255 and vice-versa. Sometimes we can just extend the original
3462    range. */
3463    
3464    if ((options & PCRE_CASELESS) != 0)
3465      {
3466    #ifdef SUPPORT_UCP
3467      if ((options & PCRE_UTF8) != 0)
3468        {
3469        int rc;
3470        pcre_uint32 oc, od;
3471    
3472        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3473        c = start;
3474    
3475        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3476          {
3477          /* Handle a single character that has more than one other case. */
3478    
3479          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3480            PRIV(ucd_caseless_sets) + rc, oc);
3481    
3482          /* Do nothing if the other case range is within the original range. */
3483    
3484          else if (oc >= start && od <= end) continue;
3485    
3486          /* Extend the original range if there is overlap, noting that if oc < c, we
3487          can't have od > end because a subrange is always shorter than the basic
3488          range. Otherwise, use a recursive call to add the additional range. */
3489    
3490          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3491          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3492          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3493          }
3494        }
3495      else
3496    #endif  /* SUPPORT_UCP */
3497    
3498      /* Not UTF-mode, or no UCP */
3499    
3500      for (c = start; c <= end && c < 256; c++)
3501        {
3502        SETBIT(classbits, cd->fcc[c]);
3503        n8++;
3504        }
3505      }
3506    
3507    /* Now handle the original range. Adjust the final value according to the bit
3508    length - this means that the same lists of (e.g.) horizontal spaces can be used
3509    in all cases. */
3510    
3511    #if defined COMPILE_PCRE8
3512    #ifdef SUPPORT_UTF
3513      if ((options & PCRE_UTF8) == 0)
3514    #endif
3515      if (end > 0xff) end = 0xff;
3516    
3517    #elif defined COMPILE_PCRE16
3518    #ifdef SUPPORT_UTF
3519      if ((options & PCRE_UTF16) == 0)
3520    #endif
3521      if (end > 0xffff) end = 0xffff;
3522    
3523    #endif /* COMPILE_PCRE[8|16] */
3524    
3525    /* If all characters are less than 256, use the bit map. Otherwise use extra
3526    data. */
3527    
3528    if (end < 0x100)
3529      {
3530      for (c = start; c <= end; c++)
3531        {
3532        n8++;
3533        SETBIT(classbits, c);
3534        }
3535      }
3536    
3537    else
3538      {
3539      pcre_uchar *uchardata = *uchardptr;
3540    
3541    #ifdef SUPPORT_UTF
3542      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3543        {
3544        if (start < end)
3545          {
3546          *uchardata++ = XCL_RANGE;
3547          uchardata += PRIV(ord2utf)(start, uchardata);
3548          uchardata += PRIV(ord2utf)(end, uchardata);
3549          }
3550        else if (start == end)
3551          {
3552          *uchardata++ = XCL_SINGLE;
3553          uchardata += PRIV(ord2utf)(start, uchardata);
3554          }
3555        }
3556      else
3557    #endif  /* SUPPORT_UTF */
3558    
3559      /* Without UTF support, character values are constrained by the bit length,
3560      and can only be > 256 for 16-bit and 32-bit libraries. */
3561    
3562    #ifdef COMPILE_PCRE8
3563        {}
3564    #else
3565      if (start < end)
3566        {
3567        *uchardata++ = XCL_RANGE;
3568        *uchardata++ = start;
3569        *uchardata++ = end;
3570        }
3571      else if (start == end)
3572        {
3573        *uchardata++ = XCL_SINGLE;
3574        *uchardata++ = start;
3575        }
3576    #endif
3577    
3578      *uchardptr = uchardata;   /* Updata extra data pointer */
3579      }
3580    
3581    return n8;    /* Number of 8-bit characters */
3582    }
3583    
3584    
3585    
3586    
3587    /*************************************************
3588    *        Add a list of characters to a class     *
3589    *************************************************/
3590    
3591    /* This function is used for adding a list of case-equivalent characters to a
3592    class, and also for adding a list of horizontal or vertical whitespace. If the
3593    list is in order (which it should be), ranges of characters are detected and
3594    handled appropriately. This function is mutually recursive with the function
3595    above.
3596    
3597    Arguments:
3598      classbits     the bit map for characters < 256
3599      uchardptr     points to the pointer for extra data
3600      options       the options word
3601      cd            contains pointers to tables etc.
3602      p             points to row of 32-bit values, terminated by NOTACHAR
3603      except        character to omit; this is used when adding lists of
3604                      case-equivalent characters to avoid including the one we
3605                      already know about
3606    
3607    Returns:        the number of < 256 characters added
3608                    the pointer to extra data is updated
3609    */
3610    
3611    static int
3612    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3613      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3614    {
3615    int n8 = 0;
3616    while (p[0] < NOTACHAR)
3617      {
3618      int n = 0;
3619      if (p[0] != except)
3620        {
3621        while(p[n+1] == p[0] + n + 1) n++;
3622        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3623        }
3624      p += n + 1;
3625      }
3626    return n8;
3627    }
3628    
3629    
3630    
3631    /*************************************************
3632    *    Add characters not in a list to a class     *
3633    *************************************************/
3634    
3635    /* This function is used for adding the complement of a list of horizontal or
3636    vertical whitespace to a class. The list must be in order.
3637    
3638    Arguments:
3639      classbits     the bit map for characters < 256
3640      uchardptr     points to the pointer for extra data
3641      options       the options word
3642      cd            contains pointers to tables etc.
3643      p             points to row of 32-bit values, terminated by NOTACHAR
3644    
3645    Returns:        the number of < 256 characters added
3646                    the pointer to extra data is updated
3647    */
3648    
3649    static int
3650    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3651      int options, compile_data *cd, const pcre_uint32 *p)
3652    {
3653    BOOL utf = (options & PCRE_UTF8) != 0;
3654    int n8 = 0;
3655    if (p[0] > 0)
3656      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3657    while (p[0] < NOTACHAR)
3658      {
3659      while (p[1] == p[0] + 1) p++;
3660      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3661        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3662      p++;
3663      }
3664    return n8;
3665    }
3666    
3667    
3668    
3669    /*************************************************
3670  *           Compile one branch                   *  *           Compile one branch                   *
3671  *************************************************/  *************************************************/
3672    
# Line 3040  Arguments: Line 3681  Arguments:
3681    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3682    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3683    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3684    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3685    reqbyteptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3686      reqcharptr     place to put the last required character
3687      reqcharflagsptr place to put the last required character flags, or a negative number
3688    bcptr          points to current branch chain    bcptr          points to current branch chain
3689    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3690    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3053  Returns:         TRUE on success Line 3696  Returns:         TRUE on success
3696  */  */
3697    
3698  static BOOL  static BOOL
3699  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3700    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3701    int cond_depth, compile_data *cd, int *lengthptr)    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3702      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3703      branch_chain *bcptr, int cond_depth,
3704      compile_data *cd, int *lengthptr)
3705  {  {
3706  int repeat_type, op_type;  int repeat_type, op_type;
3707  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3708  int bravalue = 0;  int bravalue = 0;
3709  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3710  int firstbyte, reqbyte;  pcre_uint32 firstchar, reqchar;
3711  int zeroreqbyte, zerofirstbyte;  pcre_int32 firstcharflags, reqcharflags;
3712  int req_caseopt, reqvary, tempreqvary;  pcre_uint32 zeroreqchar, zerofirstchar;
3713    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3714    pcre_int32 req_caseopt, reqvary, tempreqvary;
3715  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3716  int after_manual_callout = 0;  int after_manual_callout = 0;
3717  int length_prevgroup = 0;  int length_prevgroup = 0;
3718  register int c;  register pcre_uint32 c;
3719  register uschar *code = *codeptr;  int escape;
3720  uschar *last_code = code;  register pcre_uchar *code = *codeptr;
3721  uschar *orig_code = code;  pcre_uchar *last_code = code;
3722  uschar *tempcode;  pcre_uchar *orig_code = code;
3723    pcre_uchar *tempcode;
3724  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3725  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3726  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3727  const uschar *tempptr;  const pcre_uchar *tempptr;
3728  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3729  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3730  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3731  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3732  uschar classbits[32];  pcre_uint8 classbits[32];
3733    
3734  /* 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
3735  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
3736  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3737    
3738  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3739  BOOL class_utf8;  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3740  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3741  uschar *class_utf8data;  #ifndef COMPILE_PCRE32
3742  uschar *class_utf8data_base;  pcre_uchar utf_chars[6];
3743  uschar utf8_char[6];  #endif
3744  #else  #else
3745  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3746    #endif
3747    
3748    /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3749    class_uchardata always so that it can be passed to add_to_class() always,
3750    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3751    alternative calls for the different cases. */
3752    
3753    pcre_uchar *class_uchardata;
3754    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3755    BOOL xclass;
3756    pcre_uchar *class_uchardata_base;
3757  #endif  #endif
3758    
3759  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3107  greedy_non_default = greedy_default ^ 1; Line 3767  greedy_non_default = greedy_default ^ 1;
3767    
3768  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3769  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
3770  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
3771  find one.  find one.
3772    
3773  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
3774  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
3775  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3776  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3777    
3778  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3779    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3780    
3781  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3782  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
3783  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
3784  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3785    value. This is used only for ASCII characters. */
3786    
3787  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3788    
3789  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3790    
# Line 3134  for (;; ptr++) Line 3796  for (;; ptr++)
3796    BOOL is_quantifier;    BOOL is_quantifier;
3797    BOOL is_recurse;    BOOL is_recurse;
3798    BOOL reset_bracount;    BOOL reset_bracount;
3799    int class_charcount;    int class_has_8bitchar;
3800    int class_lastchar;    int class_one_char;
3801    int newoptions;    int newoptions;
3802    int recno;    int recno;
3803    int refsign;    int refsign;
3804    int skipbytes;    int skipbytes;
3805    int subreqbyte;    pcre_uint32 subreqchar, subfirstchar;
3806    int subfirstbyte;    pcre_int32 subreqcharflags, subfirstcharflags;
3807    int terminator;    int terminator;
3808    int mclength;    unsigned int mclength;
3809    int tempbracount;    unsigned int tempbracount;
3810    uschar mcbuffer[8];    pcre_uint32 ec;
3811      pcre_uchar mcbuffer[8];
3812    
3813    /* Get next byte in the pattern */    /* Get next character in the pattern */
3814    
3815    c = *ptr;    c = *ptr;
3816    
# Line 3169  for (;; ptr++) Line 3832  for (;; ptr++)
3832  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3833      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3834  #endif  #endif
3835      if (code > cd->start_workspace + WORK_SIZE_CHECK)   /* Check for overrun */      if (code > cd->start_workspace + cd->workspace_size -
3836            WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3837        {        {
3838        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
3839        goto FAILED;        goto FAILED;
# Line 3192  for (;; ptr++) Line 3856  for (;; ptr++)
3856        }        }
3857    
3858      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3859      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, code - last_code, c));      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3860          (int)(code - last_code), c, c));
3861    
3862      /* 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
3863      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 3202  for (;; ptr++) Line 3867  for (;; ptr++)
3867        {        {
3868        if (previous > orig_code)        if (previous > orig_code)
3869          {          {
3870          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3871          code -= previous - orig_code;          code -= previous - orig_code;
3872          previous = orig_code;          previous = orig_code;
3873          }          }
# Line 3218  for (;; ptr++) Line 3883  for (;; ptr++)
3883    /* 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
3884    reference list. */    reference list. */
3885    
3886    else if (cd->hwm > cd->start_workspace + WORK_SIZE_CHECK)    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3887               WORK_SIZE_SAFETY_MARGIN)
3888      {      {
3889      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
3890      goto FAILED;      goto FAILED;
# Line 3270  for (;; ptr++) Line 3936  for (;; ptr++)
3936    
3937    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
3938      {      {
3939      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
3940      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3941        {        {
3942        ptr++;        ptr++;
# Line 3278  for (;; ptr++) Line 3944  for (;; ptr++)
3944          {          {
3945          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3946          ptr++;          ptr++;
3947  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3948          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3949  #endif  #endif
3950          }          }
3951        if (*ptr != 0) continue;        if (*ptr != 0) continue;
# Line 3303  for (;; ptr++) Line 3969  for (;; ptr++)
3969      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
3970      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3971      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3972      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
3973      *reqbyteptr = reqbyte;      *firstcharflagsptr = firstcharflags;
3974        *reqcharptr = reqchar;
3975        *reqcharflagsptr = reqcharflags;
3976      *codeptr = code;      *codeptr = code;
3977      *ptrptr = ptr;      *ptrptr = ptr;
3978      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3328  for (;; ptr++) Line 3996  for (;; ptr++)
3996      previous = NULL;      previous = NULL;
3997      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3998        {        {
3999        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4000        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
4001        }        }
4002      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3340  for (;; ptr++) Line 4008  for (;; ptr++)
4008      break;      break;
4009    
4010      /* 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
4011      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4012    
4013      case CHAR_DOT:      case CHAR_DOT:
4014      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4015      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
4016      zeroreqbyte = reqbyte;      zerofirstcharflags = firstcharflags;
4017        zeroreqchar = reqchar;
4018        zeroreqcharflags = reqcharflags;
4019      previous = code;      previous = code;
4020      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4021      break;      break;
# Line 3400  for (;; ptr++) Line 4070  for (;; ptr++)
4070          {          {
4071          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
4072            ptr++;            ptr++;
4073          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)  
4074            ptr += 3;            ptr += 3;
4075          else          else
4076            break;            break;
# Line 3420  for (;; ptr++) Line 4089  for (;; ptr++)
4089          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4090        {        {
4091        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4092        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4093        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
4094          zerofirstcharflags = firstcharflags;
4095        break;        break;
4096        }        }
4097    
# Line 3431  for (;; ptr++) Line 4101  for (;; ptr++)
4101    
4102      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4103    
4104      /* 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:
4105      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 <
4106      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
4107        character. */
4108    
4109      class_charcount = 0;      class_has_8bitchar = 0;
4110      class_lastchar = -1;      class_one_char = 0;
4111    
4112      /* 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
4113      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
4114      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
4115      */      map. */
4116    
4117      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4118    
4119  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4120      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;
4121      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4122      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4123  #endif  #endif
4124    
4125      /* 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 3457  for (;; ptr++) Line 4128  for (;; ptr++)
4128    
4129      if (c != 0) do      if (c != 0) do
4130        {        {
4131        const uschar *oldptr;        const pcre_uchar *oldptr;
4132    
4133  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4134        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
4135          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4136          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
4137          }          }
4138    #endif
4139    
4140        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4141          /* In the pre-compile phase, accumulate the length of any extra
4142        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
4143        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4144        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4145          however. */
4146        if (lengthptr != NULL)  
4147          {        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4148          *lengthptr += class_utf8data - class_utf8data_base;          {
4149          class_utf8data = class_utf8data_base;          xclass = TRUE;
4150            *lengthptr += class_uchardata - class_uchardata_base;
4151            class_uchardata = class_uchardata_base;
4152          }          }
   
4153  #endif  #endif
4154    
4155        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3503  for (;; ptr++) Line 4177  for (;; ptr++)
4177          {          {
4178          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
4179          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
4180          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
4181          uschar pbits[32];          pcre_uint8 pbits[32];
4182    
4183          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
4184            {            {
# Line 3531  for (;; ptr++) Line 4205  for (;; ptr++)
4205          alpha. This relies on the fact that the class table starts with          alpha. This relies on the fact that the class table starts with
4206          alpha, lower, upper as the first 3 entries. */          alpha, lower, upper as the first 3 entries. */
4207    
4208          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)          if ((options & PCRE_CASELESS) != 0 && posix_class <= 2)
4209            posix_class = 0;            posix_class = 0;
4210    
4211          /* 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 3559  for (;; ptr++) Line 4233  for (;; ptr++)
4233          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
4234    
4235          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
4236            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
4237    
4238          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
4239    
# Line 3574  for (;; ptr++) Line 4248  for (;; ptr++)
4248              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4249            }            }
4250    
4251          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4252          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4253    
4254          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3590  for (;; ptr++) Line 4264  for (;; ptr++)
4264            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4265    
4266          ptr = tempptr + 1;          ptr = tempptr + 1;
4267          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 character. */
4268            class_has_8bitchar = 1;
4269            /* Every class contains at least two characters. */
4270            class_one_char = 2;
4271          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4272          }          }
4273    
4274        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
4275        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
4276        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
4277        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
4278        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_one_char bigger
4279        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
4280          as literal characters (by default), or are faulted if
4281        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4282    
4283        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4284          {          {
4285          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options, TRUE);
4286    
4287          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4288    
4289          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          if (escape == 0)
4290          else if (-c == ESC_Q)            /* Handle start of quoted string */            c = ec;
4291            else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4292            else if (escape == ESC_N)            /* \N is not supported in a class */
4293              {
4294              *errorcodeptr = ERR71;
4295              goto FAILED;
4296              }
4297            else if (escape == ESC_Q)            /* Handle start of quoted string */
4298            {            {
4299            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4300              {              {
# Line 3617  for (;; ptr++) Line 4303  for (;; ptr++)
4303            else inescq = TRUE;            else inescq = TRUE;
4304            continue;            continue;
4305            }            }
4306          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4307    
4308          if (c < 0)          else
4309            {            {
4310            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4311            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
4312              class_has_8bitchar++;
4313              /* Every class contains at least two characters. */
4314              class_one_char += 2;
4315    
4316            switch (-c)            switch (escape)
4317              {              {
4318  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4319              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3634  for (;; ptr++) Line 4323  for (;; ptr++)
4323              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4324              case ESC_SU:              case ESC_SU:
4325              nestptr = ptr;              nestptr = ptr;
4326              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4327              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4328              continue;              continue;
4329  #endif  #endif
4330              case ESC_d:              case ESC_d:
# Line 3658  for (;; ptr++) Line 4347  for (;; ptr++)
4347    
4348              /* 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
4349              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4350              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4351                EBCDIC, so we lazily just adjust the appropriate bit. */
4352    
4353              case ESC_s:              case ESC_s:
4354              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 3671  for (;; ptr++) Line 4361  for (;; ptr++)
4361              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];              for (c = 0; c < 32; c++) classbits[c] |= ~cbits[c+cbit_space];
4362              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4363              continue;              continue;
4364    
4365                /* The rest apply in both UCP and non-UCP cases. */
4366    
4367              case ESC_h:              case ESC_h:
4368              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4369              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  
4370              continue;              continue;
4371    
4372              case ESC_H:              case ESC_H:
4373              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4374                {                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  
4375              continue;              continue;
4376    
4377              case ESC_v:              case ESC_v:
4378              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4379              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  
4380              continue;              continue;
4381    
4382              case ESC_V:              case ESC_V:
4383              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4384                {                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  
4385              continue;              continue;
4386    
4387  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 3793  for (;; ptr++) Line 4389  for (;; ptr++)
4389              case ESC_P:              case ESC_P:
4390                {                {
4391                BOOL negated;                BOOL negated;
4392                int pdata;                unsigned int ptype = 0, pdata = 0;
4393                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4394                if (ptype < 0) goto FAILED;                  goto FAILED;
4395                class_utf8 = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_utf8data++ = ((-c == ESC_p) != negated)?  
4396                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4397                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4398                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4399                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4400                continue;                continue;
4401                }                }
4402  #endif  #endif
# Line 3815  for (;; ptr++) Line 4410  for (;; ptr++)
4410                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4411                goto FAILED;                goto FAILED;
4412                }                }
4413              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4414              c = *ptr;              /* Get the final character and fall through */              class_one_char -= 2;     /* Undo the speculative increase. */
4415                c = *ptr;                /* Get the final character and fall through */
4416              break;              break;
4417              }              }
4418            }            }
4419    
4420          /* 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).
4421          greater than 256 in UTF-8 mode. */          This may be greater than 256. */
4422    
4423            escape = 0;
4424    
4425          }   /* End of backslash handling */          }   /* End of backslash handling */
4426    
4427        /* 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
4428        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
4429        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4430        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4431    
4432        CHECK_RANGE:        CHECK_RANGE:
4433        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 3837  for (;; ptr++) Line 4435  for (;; ptr++)
4435          inescq = FALSE;          inescq = FALSE;
4436          ptr += 2;          ptr += 2;
4437          }          }
   
4438        oldptr = ptr;        oldptr = ptr;
4439    
4440        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4441    
4442        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4443    
# Line 3848  for (;; ptr++) Line 4445  for (;; ptr++)
4445    
4446        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4447          {          {
4448          int d;          pcre_uint32 d;
4449          ptr += 2;          ptr += 2;
4450          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4451    
# Line 3863  for (;; ptr++) Line 4460  for (;; ptr++)
4460            inescq = TRUE;            inescq = TRUE;
4461            break;            break;
4462            }            }
4463    
4464            /* Minus (hyphen) at the end of a class is treated as a literal, so put
4465            back the pointer and jump to handle the character that preceded it. */
4466    
4467          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4468            {            {
4469            ptr = oldptr;            ptr = oldptr;
4470            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4471            }            }
4472    
4473            /* Otherwise, we have a potential range; pick up the next character */
4474    
4475  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4476          if (utf8)          if (utf)
4477            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4478            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4479            }            }
# Line 3885  for (;; ptr++) Line 4487  for (;; ptr++)
4487    
4488          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4489            {            {
4490            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4491              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4492            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4493    
4494            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4495    
4496            if (d < 0)            if (descape != 0)
4497              {              {
4498              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4499                {                {
4500                ptr = oldptr;                ptr = oldptr;
4501                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4502                }                }
4503              }              }
4504            }            }
4505    
4506          /* Check that the two values are in the correct order. Optimize          /* Check that the two values are in the correct order. Optimize
4507          one-character ranges */          one-character ranges. */
4508    
4509          if (d < c)          if (d < c)
4510            {            {
4511            *errorcodeptr = ERR8;            *errorcodeptr = ERR8;
4512            goto FAILED;            goto FAILED;
4513            }            }
4514            if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4515    
4516          if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */          /* We have found a character range, so single character optimizations
4517            cannot be done anymore. Any value greater than 1 indicates that there
4518            is more than one character. */
4519    
4520            class_one_char = 2;
4521    
4522          /* Remember \r or \n */          /* Remember an explicit \r or \n, and add the range to the class. */
4523    
4524          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4525    
4526          /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless          class_has_8bitchar +=
4527          matching, we have to use an XCLASS with extra data items. Caseless            add_to_class(classbits, &class_uchardata, options, cd, c, d);
4528          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));  
             }  
           }  
   
4529          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4530          }          }
4531    
4532        /* 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
4533        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
4534        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4535          increase it if it is already 2 or more ... just in case there's a class
4536        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4537    
4538        /* Handle a character that cannot go in the bit map */        CLASS_SINGLE_CHARACTER:
4539          if (class_one_char < 2) class_one_char++;
4540    
4541          /* If class_one_char is 1, we have the first single character in the
4542          class, and there have been no prior ranges, or XCLASS items generated by
4543          escapes. If this is the final character in the class, we can optimize by
4544          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4545          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4546          to be set. Otherwise, there can be no first char if this item is first,
4547          whatever repeat count may follow. In the case of reqchar, save the
4548          previous value for reinstating. */
4549    
4550  #ifdef SUPPORT_UTF8        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
       if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
4551          {          {
4552          class_utf8 = TRUE;          ptr++;
4553          *class_utf8data++ = XCL_SINGLE;          zeroreqchar = reqchar;
4554          class_utf8data += _pcre_ord2utf8(c, class_utf8data);          zeroreqcharflags = reqcharflags;
4555    
4556  #ifdef SUPPORT_UCP          if (negate_class)
         if ((options & PCRE_CASELESS) != 0)  
4557            {            {
4558            unsigned int othercase;  #ifdef SUPPORT_UCP
4559            if ((othercase = UCD_OTHERCASE(c)) != c)            int d;
4560    #endif
4561              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4562              zerofirstchar = firstchar;
4563              zerofirstcharflags = firstcharflags;
4564    
4565              /* For caseless UTF-8 mode when UCP support is available, check
4566              whether this character has more than one other case. If so, generate
4567              a special OP_NOTPROP item instead of OP_NOTI. */
4568    
4569    #ifdef SUPPORT_UCP
4570              if (utf && (options & PCRE_CASELESS) != 0 &&
4571                  (d = UCD_CASESET(c)) != 0)
4572                {
4573                *code++ = OP_NOTPROP;
4574                *code++ = PT_CLIST;
4575                *code++ = d;
4576                }
4577              else
4578    #endif
4579              /* Char has only one other case, or UCP not available */
4580    
4581              {              {
4582              *class_utf8data++ = XCL_SINGLE;              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4583              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4584                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4585                  code += PRIV(ord2utf)(c, code);
4586                else
4587    #endif
4588                  *code++ = c;
4589              }              }
4590    
4591              /* We are finished with this character class */
4592    
4593              goto END_CLASS;
4594            }            }
 #endif  /* SUPPORT_UCP */  
4595    
4596          }          /* For a single, positive character, get the value into mcbuffer, and
4597        else          then we can handle this with the normal one-character code. */
 #endif  /* SUPPORT_UTF8 */  
4598    
4599        /* Handle a single-byte character */  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4600          {          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4601          classbits[c/8] |= (1 << (c&7));            mclength = PRIV(ord2utf)(c, mcbuffer);
4602          if ((options & PCRE_CASELESS) != 0)          else
4603    #endif
4604            {            {
4605            c = cd->fcc[c];   /* flip case */            mcbuffer[0] = c;
4606            classbits[c/8] |= (1 << (c&7));            mclength = 1;
4607            }            }
4608          class_charcount++;          goto ONE_CHAR;
4609          class_lastchar = c;          }       /* End of 1-char optimization */
4610          }  
4611          /* There is more than one character in the class, or an XCLASS item
4612          has been generated. Add this character to the class. */
4613    
4614          class_has_8bitchar +=
4615            add_to_class(classbits, &class_uchardata, options, cd, c, c);
4616        }        }
4617    
4618      /* 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 4075  for (;; ptr++) Line 4632  for (;; ptr++)
4632        goto FAILED;        goto FAILED;
4633        }        }
4634    
4635      /* 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
4636      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
4637      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
4638      optimize.      only if the very last character in the class needs XCLASS will it contain
4639        anything at this point. For this reason, xclass gets set TRUE above when
4640      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
4641      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      instead of just doing a test on class_uchardata below. */
4642      operate on single-bytes characters only. This is an historical hangover.  
4643      Maybe one day we can tidy these opcodes to handle multi-byte characters.  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4644        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4645      The optimization throws away the bit map. We turn the item into a  #endif
4646      1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
4647      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
4648      case, it can cause firstbyte to be set. Otherwise, there can be no first      setting, whatever the repeat count. Any reqchar setting must remain
4649      char if this item is first, whatever repeat count may follow. In the case      unchanged after any kind of repeat. */
4650      of reqbyte, save the previous value for reinstating. */  
4651        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4652  #ifdef SUPPORT_UTF8      zerofirstchar = firstchar;
4653      if (class_charcount == 1 && !class_utf8 &&      zerofirstcharflags = firstcharflags;
4654        (!utf8 || !negate_class || class_lastchar < 128))      zeroreqchar = reqchar;
4655  #else      zeroreqcharflags = reqcharflags;
     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;  
4656    
4657      /* If there are characters with values > 255, we have to compile an      /* If there are characters with values > 255, we have to compile an
4658      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 4144  for (;; ptr++) Line 4662  for (;; ptr++)
4662      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
4663      actual compiled code. */      actual compiled code. */
4664    
4665  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4666      if (class_utf8 && (!should_flip_negation || (options & PCRE_UCP) != 0))      if (xclass && (!should_flip_negation || (options & PCRE_UCP) != 0))
4667    #elif !defined COMPILE_PCRE8
4668        if (xclass && !should_flip_negation)
4669    #endif
4670    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4671        {        {
4672        *class_utf8data++ = XCL_END;    /* Marks the end of extra data */        *class_uchardata++ = XCL_END;    /* Marks the end of extra data */
4673        *code++ = OP_XCLASS;        *code++ = OP_XCLASS;
4674        code += LINK_SIZE;        code += LINK_SIZE;
4675        *code = negate_class? XCL_NOT : 0;        *code = negate_class? XCL_NOT:0;
4676    
4677        /* 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;
4678        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. */
4679    
4680        if (class_charcount > 0)        if (class_has_8bitchar > 0)
4681          {          {
4682          *code++ |= XCL_MAP;          *code++ |= XCL_MAP;
4683          memmove(code + 32, code, class_utf8data - code);          memmove(code + (32 / sizeof(pcre_uchar)), code,
4684              IN_UCHARS(class_uchardata - code));
4685          memcpy(code, classbits, 32);          memcpy(code, classbits, 32);
4686          code = class_utf8data + 32;          code = class_uchardata + (32 / sizeof(pcre_uchar));
4687          }          }
4688        else code = class_utf8data;        else code = class_uchardata;
4689    
4690        /* Now fill in the complete length of the item */        /* Now fill in the complete length of the item */
4691    
4692        PUT(previous, 1, code - previous);        PUT(previous, 1, (int)(code - previous));
4693        break;   /* End of class handling */        break;   /* End of class handling */
4694        }        }
4695  #endif  #endif
# Line 4178  for (;; ptr++) Line 4701  for (;; ptr++)
4701      negating it if necessary. */      negating it if necessary. */
4702    
4703      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;      *code++ = (negate_class == should_flip_negation) ? OP_CLASS : OP_NCLASS;
4704      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  
4705        {        {
4706          if (negate_class)
4707            for (c = 0; c < 32; c++) classbits[c] = ~classbits[c];
4708        memcpy(code, classbits, 32);        memcpy(code, classbits, 32);
4709        }        }
4710      code += 32;      code += 32 / sizeof(pcre_uchar);
4711    
4712        END_CLASS:
4713      break;      break;
4714    
4715    
# Line 4224  for (;; ptr++) Line 4746  for (;; ptr++)
4746    
4747      if (repeat_min == 0)      if (repeat_min == 0)
4748        {        {
4749        firstbyte = zerofirstbyte;    /* Adjust for zero repeat */        firstchar = zerofirstchar;    /* Adjust for zero repeat */
4750        reqbyte = zeroreqbyte;        /* Ditto */        firstcharflags = zerofirstcharflags;
4751          reqchar = zeroreqchar;        /* Ditto */
4752          reqcharflags = zeroreqcharflags;
4753        }        }
4754    
4755      /* Remember whether this is a variable length repeat */      /* Remember whether this is a variable length repeat */
# Line 4267  for (;; ptr++) Line 4791  for (;; ptr++)
4791    
4792      if (*previous == OP_RECURSE)      if (*previous == OP_RECURSE)
4793        {        {
4794        memmove(previous + 1 + LINK_SIZE, previous, 1 + LINK_SIZE);        memmove(previous + 1 + LINK_SIZE, previous, IN_UCHARS(1 + LINK_SIZE));
4795        *previous = OP_ONCE;        *previous = OP_ONCE;
4796        PUT(previous, 1, 2 + 2*LINK_SIZE);        PUT(previous, 1, 2 + 2*LINK_SIZE);
4797        previous[2 + 2*LINK_SIZE] = OP_KET;        previous[2 + 2*LINK_SIZE] = OP_KET;
# Line 4288  for (;; ptr++) Line 4812  for (;; ptr++)
4812    
4813      /* Now handle repetition for the different types of item. */      /* Now handle repetition for the different types of item. */
4814    
4815      /* If previous was a character match, abolish the item and generate a      /* If previous was a character or negated character match, abolish the item
4816      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
4817      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
4818      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
4819      instead.  */      into firstchar instead.  */
4820    
4821      if (*previous == OP_CHAR || *previous == OP_CHARI)      if (*previous == OP_CHAR || *previous == OP_CHARI
4822            || *previous == OP_NOT || *previous == OP_NOTI)
4823        {        {
4824        op_type = (*previous == OP_CHAR)? 0 : OP_STARI - OP_STAR;        switch (*previous)
4825            {
4826            default: /* Make compiler happy. */
4827            case OP_CHAR:  op_type = OP_STAR - OP_STAR; break;
4828            case OP_CHARI: op_type = OP_STARI - OP_STAR; break;
4829            case OP_NOT:   op_type = OP_NOTSTAR - OP_STAR; break;
4830            case OP_NOTI:  op_type = OP_NOTSTARI - OP_STAR; break;
4831            }
4832    
4833        /* Deal with UTF-8 characters that take up more than one byte. It's        /* Deal with UTF characters that take up more than one character. It's
4834        easier to write this out separately than try to macrify it. Use c to        easier to write this out separately than try to macrify it. Use c to
4835        hold the length of the