/[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 773 by ph10, Wed Nov 30 18:10:27 2011 UTC revision 1325 by ph10, Fri May 10 14:03:18 2013 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) 1997-2013 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 for most patterns. However, the memory can get  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  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  /(?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  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  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()  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. */  kicks in at the same number of forward references in all cases. */
114    
115  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)  #define COMPILE_WORK_SIZE (2048*LINK_SIZE)
# Line 104  overrun before it actually does run off Line 120  overrun before it actually does run off
120    
121  #define WORK_SIZE_SAFETY_MARGIN (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 238  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 273  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 372  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 395  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 406  static const char error_texts[] = Line 487  static const char error_texts[] =
487    "a numbered reference must not be zero\0"    "a numbered reference must not be zero\0"
488    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"    "an argument is not allowed for (*ACCEPT), (*FAIL), or (*COMMIT)\0"
489    /* 60 */    /* 60 */
490    "(*VERB) not recognized\0"    "(*VERB) not recognized or malformed\0"
491    "number is too big\0"    "number is too big\0"
492    "subpattern name expected\0"    "subpattern name expected\0"
493    "digit expected after (?+\0"    "digit expected after (?+\0"
# Line 414  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 */    /* 70 */
502    "internal error: unknown opcode in find_fixedlength()\0"    "internal error: unknown opcode in find_fixedlength()\0"
503    "\\N is not supported in a class\0"    "\\N is not supported in a class\0"
504    "too many forward references\0"    "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      "setting UTF is disabled by the application\0"
512    ;    ;
513    
514  /* 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 439  For convenience, we use the same bit def Line 527  For convenience, we use the same bit def
527    
528  Then we can use ctype_digit and ctype_xdigit in the code. */  Then we can use ctype_digit and ctype_xdigit in the code. */
529    
530    /* Using a simple comparison for decimal numbers rather than a memory read
531    is much faster, and the resulting code is simpler (the compiler turns it
532    into a subtraction and unsigned comparison). */
533    
534    #define IS_DIGIT(x) ((x) >= CHAR_0 && (x) <= CHAR_9)
535    
536  #ifndef EBCDIC  #ifndef EBCDIC
537    
538  /* 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
539  UTF-8 mode. */  UTF-8 mode. */
540    
541  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
542    {    {
543    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7 */
544    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15 */
# Line 483  static const unsigned char digitab[] = Line 577  static const unsigned char digitab[] =
577    
578  /* 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. */
579    
580  static const unsigned char digitab[] =  static const pcre_uint8 digitab[] =
581    {    {
582    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   0-  7  0 */
583    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */    0x00,0x00,0x00,0x00,0x00,0x00,0x00,0x00, /*   8- 15    */
# Line 518  static const unsigned char digitab[] = Line 612  static const unsigned char digitab[] =
612    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */    0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c,0x0c, /*  0 - 7  F0 */
613    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */    0x0c,0x0c,0x00,0x00,0x00,0x00,0x00,0x00};/*  8 -255    */
614    
615  static const unsigned char ebcdic_chartab[] = { /* chartable partial dup */  static const pcre_uint8 ebcdic_chartab[] = { /* chartable partial dup */
616    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */    0x80,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*   0-  7 */
617    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */    0x00,0x00,0x00,0x00,0x01,0x01,0x00,0x00, /*   8- 15 */
618    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */    0x00,0x00,0x00,0x00,0x00,0x01,0x00,0x00, /*  16- 23 */
# Line 554  static const unsigned char ebcdic_charta Line 648  static const unsigned char ebcdic_charta
648  #endif  #endif
649    
650    
 /* 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 *);  
   
   
651    
652  /*************************************************  /*************************************************
653  *            Find an error text                  *  *            Find an error text                  *
# Line 581  find_error_text(int n) Line 668  find_error_text(int n)
668  const char *s = error_texts;  const char *s = error_texts;
669  for (; n > 0; n--)  for (; n > 0; n--)
670    {    {
671    while (*s++ != 0) {};    while (*s++ != CHAR_NULL) {};
672    if (*s == 0) return "Error text not found (please report)";    if (*s == CHAR_NULL) return "Error text not found (please report)";
673    }    }
674  return s;  return s;
675  }  }
# Line 592  return s; Line 679  return s;
679  *           Expand the workspace                 *  *           Expand the workspace                 *
680  *************************************************/  *************************************************/
681    
682  /* This function is called during the second compiling phase, if the number of  /* This function is called during the second compiling phase, if the number of
683  forward references fills the existing workspace, which is originally a block on  forward references fills the existing workspace, which is originally a block on
684  the stack. A larger block is obtained from malloc() unless the ultimate limit  the stack. A larger block is obtained from malloc() unless the ultimate limit
685  has been reached or the increase will be rather small.  has been reached or the increase will be rather small.
686    
687  Argument: pointer to the compile data block  Argument: pointer to the compile data block
# Line 604  Returns:  0 if all went well, else an er Line 691  Returns:  0 if all went well, else an er
691  static int  static int
692  expand_workspace(compile_data *cd)  expand_workspace(compile_data *cd)
693  {  {
694  uschar *newspace;  pcre_uchar *newspace;
695  int newsize = cd->workspace_size * 2;  int newsize = cd->workspace_size * 2;
696    
697  if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;  if (newsize > COMPILE_WORK_SIZE_MAX) newsize = COMPILE_WORK_SIZE_MAX;
# Line 612  if (cd->workspace_size >= COMPILE_WORK_S Line 699  if (cd->workspace_size >= COMPILE_WORK_S
699      newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)      newsize - cd->workspace_size < WORK_SIZE_SAFETY_MARGIN)
700   return ERR72;   return ERR72;
701    
702  newspace = (pcre_malloc)(newsize);  newspace = (PUBL(malloc))(IN_UCHARS(newsize));
703  if (newspace == NULL) return ERR21;  if (newspace == NULL) return ERR21;
704    memcpy(newspace, cd->start_workspace, cd->workspace_size * sizeof(pcre_uchar));
705  memcpy(newspace, cd->start_workspace, cd->workspace_size);  cd->hwm = (pcre_uchar *)newspace + (cd->hwm - cd->start_workspace);
706  cd->hwm = (uschar *)newspace + (cd->hwm - cd->start_workspace);  if (cd->workspace_size > COMPILE_WORK_SIZE)
707  if (cd->workspace_size > COMPILE_WORK_SIZE)    (PUBL(free))((void *)cd->start_workspace);
   (pcre_free)((void *)cd->start_workspace);  
708  cd->start_workspace = newspace;  cd->start_workspace = newspace;
709  cd->workspace_size = newsize;  cd->workspace_size = newsize;
710  return 0;  return 0;
# Line 642  Returns:    TRUE or FALSE Line 728  Returns:    TRUE or FALSE
728  */  */
729    
730  static BOOL  static BOOL
731  is_counted_repeat(const uschar *p)  is_counted_repeat(const pcre_uchar *p)
732  {  {
733  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
734  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
735    while (IS_DIGIT(*p)) p++;
736  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
737    
738  if (*p++ != CHAR_COMMA) return FALSE;  if (*p++ != CHAR_COMMA) return FALSE;
739  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;  if (*p == CHAR_RIGHT_CURLY_BRACKET) return TRUE;
740    
741  if ((digitab[*p++] & ctype_digit) == 0) return FALSE;  if (!IS_DIGIT(*p)) return FALSE;
742  while ((digitab[*p] & ctype_digit) != 0) p++;  p++;
743    while (IS_DIGIT(*p)) p++;
744    
745  return (*p == CHAR_RIGHT_CURLY_BRACKET);  return (*p == CHAR_RIGHT_CURLY_BRACKET);
746  }  }
# Line 664  return (*p == CHAR_RIGHT_CURLY_BRACKET); Line 752  return (*p == CHAR_RIGHT_CURLY_BRACKET);
752  *************************************************/  *************************************************/
753    
754  /* 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
755  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
756  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
757  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
758  UTF-8 is enabled, a positive value greater than 255 may be returned. On entry,  be returned in chptr.
759  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
760  sequence.  escape sequence.
761    
762  Arguments:  Arguments:
763    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
764      chptr          points to the data character
765    errorcodeptr   points to the errorcode variable    errorcodeptr   points to the errorcode variable
766    bracount       number of previous extracting brackets    bracount       number of previous extracting brackets
767    options        the options bits    options        the options bits
768    isclass        TRUE if inside a character class    isclass        TRUE if inside a character class
769    
770  Returns:         zero or positive => a data character  Returns:         zero => a data character
771                   negative => a special escape sequence                   positive => a special escape sequence
772                     negative => a back reference
773                   on error, errorcodeptr is set                   on error, errorcodeptr is set
774  */  */
775    
776  static int  static int
777  check_escape(const uschar **ptrptr, int *errorcodeptr, int bracount,  check_escape(const pcre_uchar **ptrptr, pcre_uint32 *chptr, int *errorcodeptr,
778    int options, BOOL isclass)    int bracount, int options, BOOL isclass)
779  {  {
780  BOOL utf8 = (options & PCRE_UTF8) != 0;  /* PCRE_UTF16 has the same value as PCRE_UTF8. */
781  const uschar *ptr = *ptrptr + 1;  BOOL utf = (options & PCRE_UTF8) != 0;
782  int c, i;  const pcre_uchar *ptr = *ptrptr + 1;
783    pcre_uint32 c;
784    int escape = 0;
785    int i;
786    
787  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */  GETCHARINCTEST(c, ptr);           /* Get character value, increment pointer */
788  ptr--;                            /* Set pointer back to the last byte */  ptr--;                            /* Set pointer back to the last byte */
789    
790  /* If backslash is at the end of the pattern, it's an error. */  /* If backslash is at the end of the pattern, it's an error. */
791    
792  if (c == 0) *errorcodeptr = ERR1;  if (c == CHAR_NULL) *errorcodeptr = ERR1;
793    
794  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup  /* Non-alphanumerics are literals. For digits or letters, do an initial lookup
795  in a table. A non-zero result is something that can be returned immediately.  in a table. A non-zero result is something that can be returned immediately.
796  Otherwise further processing may be required. */  Otherwise further processing may be required. */
797    
798  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
799  else if (c < CHAR_0 || c > CHAR_z) {}                     /* Not alphanumeric */  /* Not alphanumeric */
800  else if ((i = escapes[c - CHAR_0]) != 0) c = i;  else if (c < CHAR_0 || c > CHAR_z) {}
801    else if ((i = escapes[c - CHAR_0]) != 0)
802      { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
803    
804  #else           /* EBCDIC coding */  #else           /* EBCDIC coding */
805  else if (c < 'a' || (ebcdic_chartab[c] & 0x0E) == 0) {}   /* Not alphanumeric */  /* Not alphanumeric */
806  else if ((i = escapes[c - 0x48]) != 0)  c = i;  else if (c < CHAR_a || (!MAX_255(c) || (ebcdic_chartab[c] & 0x0E) == 0)) {}
807    else if ((i = escapes[c - 0x48]) != 0)  { if (i > 0) c = (pcre_uint32)i; else escape = -i; }
808  #endif  #endif
809    
810  /* Escapes that need further processing, or are illegal. */  /* Escapes that need further processing, or are illegal. */
811    
812  else  else
813    {    {
814    const uschar *oldptr;    const pcre_uchar *oldptr;
815    BOOL braced, negated;    BOOL braced, negated, overflow;
816      int s;
817    
818    switch (c)    switch (c)
819      {      {
# Line 733  else Line 830  else
830        {        {
831        /* In JavaScript, \u must be followed by four hexadecimal numbers.        /* In JavaScript, \u must be followed by four hexadecimal numbers.
832        Otherwise it is a lowercase u letter. */        Otherwise it is a lowercase u letter. */
833        if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
834             && (digitab[ptr[3]] & ctype_xdigit) != 0 && (digitab[ptr[4]] & ctype_xdigit) != 0)          && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0
835            && MAX_255(ptr[3]) && (digitab[ptr[3]] & ctype_xdigit) != 0
836            && MAX_255(ptr[4]) && (digitab[ptr[4]] & ctype_xdigit) != 0)
837          {          {
838          c = 0;          c = 0;
839          for (i = 0; i < 4; ++i)          for (i = 0; i < 4; ++i)
840            {            {
841            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
842  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
843            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
844            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 748  else Line 847  else
847            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
848  #endif  #endif
849            }            }
850    
851    #if defined COMPILE_PCRE8
852            if (c > (utf ? 0x10ffffU : 0xffU))
853    #elif defined COMPILE_PCRE16
854            if (c > (utf ? 0x10ffffU : 0xffffU))
855    #elif defined COMPILE_PCRE32
856            if (utf && c > 0x10ffffU)
857    #endif
858              {
859              *errorcodeptr = ERR76;
860              }
861            else if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
862          }          }
863        }        }
864      else      else
# Line 774  else Line 885  else
885      (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
886      number either in angle brackets or in single quotes. However, these are      number either in angle brackets or in single quotes. However, these are
887      (possibly recursive) subroutine calls, _not_ backreferences. Just return      (possibly recursive) subroutine calls, _not_ backreferences. Just return
888      the -ESC_g code (cf \k). */      the ESC_g code (cf \k). */
889    
890      case CHAR_g:      case CHAR_g:
891      if (isclass) break;      if (isclass) break;
892      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)      if (ptr[1] == CHAR_LESS_THAN_SIGN || ptr[1] == CHAR_APOSTROPHE)
893        {        {
894        c = -ESC_g;        escape = ESC_g;
895        break;        break;
896        }        }
897    
# Line 788  else Line 899  else
899    
900      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
901        {        {
902        const uschar *p;        const pcre_uchar *p;
903        for (p = ptr+2; *p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET; p++)        for (p = ptr+2; *p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET; p++)
904          if (*p != CHAR_MINUS && (digitab[*p] & ctype_digit) == 0) break;          if (*p != CHAR_MINUS && !IS_DIGIT(*p)) break;
905        if (*p != 0 && *p != CHAR_RIGHT_CURLY_BRACKET)        if (*p != CHAR_NULL && *p != CHAR_RIGHT_CURLY_BRACKET)
906          {          {
907          c = -ESC_k;          escape = ESC_k;
908          break;          break;
909          }          }
910        braced = TRUE;        braced = TRUE;
# Line 808  else Line 919  else
919        }        }
920      else negated = FALSE;      else negated = FALSE;
921    
922      c = 0;      /* The integer range is limited by the machine's int representation. */
923      while ((digitab[ptr[1]] & ctype_digit) != 0)      s = 0;
924        c = c * 10 + *(++ptr) - CHAR_0;      overflow = FALSE;
925        while (IS_DIGIT(ptr[1]))
     if (c < 0)   /* Integer overflow */  
926        {        {
927          if (s > INT_MAX / 10 - 1) /* Integer overflow */
928            {
929            overflow = TRUE;
930            break;
931            }
932          s = s * 10 + (int)(*(++ptr) - CHAR_0);
933          }
934        if (overflow) /* Integer overflow */
935          {
936          while (IS_DIGIT(ptr[1]))
937            ptr++;
938        *errorcodeptr = ERR61;        *errorcodeptr = ERR61;
939        break;        break;
940        }        }
# Line 824  else Line 945  else
945        break;        break;
946        }        }
947    
948      if (c == 0)      if (s == 0)
949        {        {
950        *errorcodeptr = ERR58;        *errorcodeptr = ERR58;
951        break;        break;
# Line 832  else Line 953  else
953    
954      if (negated)      if (negated)
955        {        {
956        if (c > bracount)        if (s > bracount)
957          {          {
958          *errorcodeptr = ERR15;          *errorcodeptr = ERR15;
959          break;          break;
960          }          }
961        c = bracount - (c - 1);        s = bracount - (s - 1);
962        }        }
963    
964      c = -(ESC_REF + c);      escape = -s;
965      break;      break;
966    
967      /* The handling of escape sequences consisting of a string of digits      /* The handling of escape sequences consisting of a string of digits
# Line 861  else Line 982  else
982      if (!isclass)      if (!isclass)
983        {        {
984        oldptr = ptr;        oldptr = ptr;
985        c -= CHAR_0;        /* The integer range is limited by the machine's int representation. */
986        while ((digitab[ptr[1]] & ctype_digit) != 0)        s = (int)(c -CHAR_0);
987          c = c * 10 + *(++ptr) - CHAR_0;        overflow = FALSE;
988        if (c < 0)    /* Integer overflow */        while (IS_DIGIT(ptr[1]))
989            {
990            if (s > INT_MAX / 10 - 1) /* Integer overflow */
991              {
992              overflow = TRUE;
993              break;
994              }
995            s = s * 10 + (int)(*(++ptr) - CHAR_0);
996            }
997          if (overflow) /* Integer overflow */
998          {          {
999            while (IS_DIGIT(ptr[1]))
1000              ptr++;
1001          *errorcodeptr = ERR61;          *errorcodeptr = ERR61;
1002          break;          break;
1003          }          }
1004        if (c < 10 || c <= bracount)        if (s < 10 || s <= bracount)
1005          {          {
1006          c = -(ESC_REF + c);          escape = -s;
1007          break;          break;
1008          }          }
1009        ptr = oldptr;      /* Put the pointer back and fall through */        ptr = oldptr;      /* Put the pointer back and fall through */
# Line 891  else Line 1023  else
1023      /* \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
1024      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
1025      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
1026      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,
1027      than 3 octal digits. */      but no more than 3 octal digits. */
1028    
1029      case CHAR_0:      case CHAR_0:
1030      c -= CHAR_0;      c -= CHAR_0;
1031      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)      while(i++ < 2 && ptr[1] >= CHAR_0 && ptr[1] <= CHAR_7)
1032          c = c * 8 + *(++ptr) - CHAR_0;          c = c * 8 + *(++ptr) - CHAR_0;
1033      if (!utf8 && c > 255) *errorcodeptr = ERR51;  #ifdef COMPILE_PCRE8
1034        if (!utf && c > 0xff) *errorcodeptr = ERR51;
1035    #endif
1036      break;      break;
1037    
1038      /* \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
1039      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.
1040      treated as a data character. */      If not, { is treated as a data character. */
1041    
1042      case CHAR_x:      case CHAR_x:
1043      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)      if ((options & PCRE_JAVASCRIPT_COMPAT) != 0)
1044        {        {
1045        /* In JavaScript, \x must be followed by two hexadecimal numbers.        /* In JavaScript, \x must be followed by two hexadecimal numbers.
1046        Otherwise it is a lowercase x letter. */        Otherwise it is a lowercase x letter. */
1047        if ((digitab[ptr[1]] & ctype_xdigit) != 0 && (digitab[ptr[2]] & ctype_xdigit) != 0)        if (MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0
1048            && MAX_255(ptr[2]) && (digitab[ptr[2]] & ctype_xdigit) != 0)
1049          {          {
1050          c = 0;          c = 0;
1051          for (i = 0; i < 2; ++i)          for (i = 0; i < 2; ++i)
1052            {            {
1053            register int cc = *(++ptr);            register pcre_uint32 cc = *(++ptr);
1054  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1055            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */            if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
1056            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));            c = (c << 4) + cc - ((cc < CHAR_A)? CHAR_0 : (CHAR_A - 10));
# Line 930  else Line 1065  else
1065    
1066      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)      if (ptr[1] == CHAR_LEFT_CURLY_BRACKET)
1067        {        {
1068        const uschar *pt = ptr + 2;        const pcre_uchar *pt = ptr + 2;
       int count = 0;  
1069    
1070        c = 0;        c = 0;
1071        while ((digitab[*pt] & ctype_xdigit) != 0)        overflow = FALSE;
1072          while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0)
1073          {          {
1074          register int cc = *pt++;          register pcre_uint32 cc = *pt++;
1075          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */          if (c == 0 && cc == CHAR_0) continue;     /* Leading zeroes */
1076          count++;  
1077    #ifdef COMPILE_PCRE32
1078            if (c >= 0x10000000l) { overflow = TRUE; break; }
1079    #endif
1080    
1081  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1082          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */          if (cc >= CHAR_a) cc -= 32;               /* Convert to upper case */
# Line 947  else Line 1085  else
1085          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 */
1086          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));          c = (c << 4) + cc - ((cc >= CHAR_0)? CHAR_0 : (CHAR_A - 10));
1087  #endif  #endif
1088    
1089    #if defined COMPILE_PCRE8
1090            if (c > (utf ? 0x10ffffU : 0xffU)) { overflow = TRUE; break; }
1091    #elif defined COMPILE_PCRE16
1092            if (c > (utf ? 0x10ffffU : 0xffffU)) { overflow = TRUE; break; }
1093    #elif defined COMPILE_PCRE32
1094            if (utf && c > 0x10ffffU) { overflow = TRUE; break; }
1095    #endif
1096            }
1097    
1098          if (overflow)
1099            {
1100            while (MAX_255(*pt) && (digitab[*pt] & ctype_xdigit) != 0) pt++;
1101            *errorcodeptr = ERR34;
1102          }          }
1103    
1104        if (*pt == CHAR_RIGHT_CURLY_BRACKET)        if (*pt == CHAR_RIGHT_CURLY_BRACKET)
1105          {          {
1106          if (c < 0 || count > (utf8? 8 : 2)) *errorcodeptr = ERR34;          if (utf && c >= 0xd800 && c <= 0xdfff) *errorcodeptr = ERR73;
1107          ptr = pt;          ptr = pt;
1108          break;          break;
1109          }          }
# Line 963  else Line 1115  else
1115      /* Read just a single-byte hex-defined char */      /* Read just a single-byte hex-defined char */
1116    
1117      c = 0;      c = 0;
1118      while (i++ < 2 && (digitab[ptr[1]] & ctype_xdigit) != 0)      while (i++ < 2 && MAX_255(ptr[1]) && (digitab[ptr[1]] & ctype_xdigit) != 0)
1119        {        {
1120        int cc;                                  /* Some compilers don't like */        pcre_uint32 cc;                          /* Some compilers don't like */
1121        cc = *(++ptr);                           /* ++ in initializers */        cc = *(++ptr);                           /* ++ in initializers */
1122  #ifndef EBCDIC  /* ASCII/UTF-8 coding */  #ifndef EBCDIC  /* ASCII/UTF-8 coding */
1123        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */        if (cc >= CHAR_a) cc -= 32;              /* Convert to upper case */
# Line 984  else Line 1136  else
1136    
1137      case CHAR_c:      case CHAR_c:
1138      c = *(++ptr);      c = *(++ptr);
1139      if (c == 0)      if (c == CHAR_NULL)
1140        {        {
1141        *errorcodeptr = ERR2;        *errorcodeptr = ERR2;
1142        break;        break;
# Line 1024  else Line 1176  else
1176  newline". PCRE does not support \N{name}. However, it does support  newline". PCRE does not support \N{name}. However, it does support
1177  quantification such as \N{2,3}. */  quantification such as \N{2,3}. */
1178    
1179  if (c == -ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&  if (escape == ESC_N && ptr[1] == CHAR_LEFT_CURLY_BRACKET &&
1180       !is_counted_repeat(ptr+2))       !is_counted_repeat(ptr+2))
1181    *errorcodeptr = ERR37;    *errorcodeptr = ERR37;
1182    
1183  /* If PCRE_UCP is set, we change the values for \d etc. */  /* If PCRE_UCP is set, we change the values for \d etc. */
1184    
1185  if ((options & PCRE_UCP) != 0 && c <= -ESC_D && c >= -ESC_w)  if ((options & PCRE_UCP) != 0 && escape >= ESC_D && escape <= ESC_w)
1186    c -= (ESC_DU - ESC_D);    escape += (ESC_DU - ESC_D);
1187    
1188  /* Set the pointer to the final character before returning. */  /* Set the pointer to the final character before returning. */
1189    
1190  *ptrptr = ptr;  *ptrptr = ptr;
1191  return c;  *chptr = c;
1192    return escape;
1193  }  }
1194    
   
   
1195  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
1196  /*************************************************  /*************************************************
1197  *               Handle \P and \p                 *  *               Handle \P and \p                 *
# Line 1054  escape sequence. Line 1205  escape sequence.
1205  Argument:  Argument:
1206    ptrptr         points to the pattern position pointer    ptrptr         points to the pattern position pointer
1207    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
1208    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
1209      pdataptr       points to an unsigned int that is set to the detailed property value
1210    errorcodeptr   points to the error code variable    errorcodeptr   points to the error code variable
1211    
1212  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
1213  */  */
1214    
1215  static int  static BOOL
1216  get_ucp(const uschar **ptrptr, BOOL *negptr, int *dptr, int *errorcodeptr)  get_ucp(const pcre_uchar **ptrptr, BOOL *negptr, unsigned int *ptypeptr,
1217      unsigned int *pdataptr, int *errorcodeptr)
1218  {  {
1219  int c, i, bot, top;  pcre_uchar c;
1220  const uschar *ptr = *ptrptr;  int i, bot, top;
1221  char name[32];  const pcre_uchar *ptr = *ptrptr;
1222    pcre_uchar name[32];
1223    
1224  c = *(++ptr);  c = *(++ptr);
1225  if (c == 0) goto ERROR_RETURN;  if (c == CHAR_NULL) goto ERROR_RETURN;
1226    
1227  *negptr = FALSE;  *negptr = FALSE;
1228    
# Line 1082  if (c == CHAR_LEFT_CURLY_BRACKET) Line 1236  if (c == CHAR_LEFT_CURLY_BRACKET)
1236      *negptr = TRUE;      *negptr = TRUE;
1237      ptr++;      ptr++;
1238      }      }
1239    for (i = 0; i < (int)sizeof(name) - 1; i++)    for (i = 0; i < (int)(sizeof(name) / sizeof(pcre_uchar)) - 1; i++)
1240      {      {
1241      c = *(++ptr);      c = *(++ptr);
1242      if (c == 0) goto ERROR_RETURN;      if (c == CHAR_NULL) goto ERROR_RETURN;
1243      if (c == CHAR_RIGHT_CURLY_BRACKET) break;      if (c == CHAR_RIGHT_CURLY_BRACKET) break;
1244      name[i] = c;      name[i] = c;
1245      }      }
# Line 1106  else Line 1260  else
1260  /* Search for a recognized property name using binary chop */  /* Search for a recognized property name using binary chop */
1261    
1262  bot = 0;  bot = 0;
1263  top = _pcre_utt_size;  top = PRIV(utt_size);
1264    
1265  while (bot < top)  while (bot < top)
1266    {    {
1267      int r;
1268    i = (bot + top) >> 1;    i = (bot + top) >> 1;
1269    c = strcmp(name, _pcre_utt_names + _pcre_utt[i].name_offset);    r = STRCMP_UC_C8(name, PRIV(utt_names) + PRIV(utt)[i].name_offset);
1270    if (c == 0)    if (r == 0)
1271      {      {
1272      *dptr = _pcre_utt[i].value;      *ptypeptr = PRIV(utt)[i].type;
1273      return _pcre_utt[i].type;      *pdataptr = PRIV(utt)[i].value;
1274        return TRUE;
1275      }      }
1276    if (c > 0) bot = i + 1; else top = i;    if (r > 0) bot = i + 1; else top = i;
1277    }    }
1278    
1279  *errorcodeptr = ERR47;  *errorcodeptr = ERR47;
1280  *ptrptr = ptr;  *ptrptr = ptr;
1281  return -1;  return FALSE;
1282    
1283  ERROR_RETURN:  ERROR_RETURN:
1284  *errorcodeptr = ERR46;  *errorcodeptr = ERR46;
1285  *ptrptr = ptr;  *ptrptr = ptr;
1286  return -1;  return FALSE;
1287  }  }
1288  #endif  #endif
1289    
# Line 1153  Returns:         pointer to '}' on succe Line 1309  Returns:         pointer to '}' on succe
1309                   current ptr on error, with errorcodeptr set non-zero                   current ptr on error, with errorcodeptr set non-zero
1310  */  */
1311    
1312  static const uschar *  static const pcre_uchar *
1313  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)
1314  {  {
1315  int min = 0;  int min = 0;
1316  int max = -1;  int max = -1;
# Line 1162  int max = -1; Line 1318  int max = -1;
1318  /* 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
1319  an integer overflow. */  an integer overflow. */
1320    
1321  while ((digitab[*p] & ctype_digit) != 0) min = min * 10 + *p++ - CHAR_0;  while (IS_DIGIT(*p)) min = min * 10 + (int)(*p++ - CHAR_0);
1322  if (min < 0 || min > 65535)  if (min < 0 || min > 65535)
1323    {    {
1324    *errorcodeptr = ERR5;    *errorcodeptr = ERR5;
# Line 1177  if (*p == CHAR_RIGHT_CURLY_BRACKET) max Line 1333  if (*p == CHAR_RIGHT_CURLY_BRACKET) max
1333    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)    if (*(++p) != CHAR_RIGHT_CURLY_BRACKET)
1334      {      {
1335      max = 0;      max = 0;
1336      while((digitab[*p] & ctype_digit) != 0) max = max * 10 + *p++ - CHAR_0;      while(IS_DIGIT(*p)) max = max * 10 + (int)(*p++ - CHAR_0);
1337      if (max < 0 || max > 65535)      if (max < 0 || max > 65535)
1338        {        {
1339        *errorcodeptr = ERR5;        *errorcodeptr = ERR5;
# Line 1232  Arguments: Line 1388  Arguments:
1388    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1389    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1390    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1391    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1392    count        pointer to the current capturing subpattern number (updated)    count        pointer to the current capturing subpattern number (updated)
1393    
1394  Returns:       the number of the named subpattern, or -1 if not found  Returns:       the number of the named subpattern, or -1 if not found
1395  */  */
1396    
1397  static int  static int
1398  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,
1399    BOOL xmode, BOOL utf8, int *count)    BOOL xmode, BOOL utf, int *count)
1400  {  {
1401  uschar *ptr = *ptrptr;  pcre_uchar *ptr = *ptrptr;
1402  int start_count = *count;  int start_count = *count;
1403  int hwm_count = start_count;  int hwm_count = start_count;
1404  BOOL dup_parens = FALSE;  BOOL dup_parens = FALSE;
# Line 1254  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1410  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1410    {    {
1411    /* Handle specials such as (*SKIP) or (*UTF8) etc. */    /* Handle specials such as (*SKIP) or (*UTF8) etc. */
1412    
1413    if (ptr[1] == CHAR_ASTERISK) ptr += 2;    if (ptr[1] == CHAR_ASTERISK)
1414        {
1415        ptr += 2;
1416        while (ptr < cd->end_pattern && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1417        }
1418    
1419    /* Handle a normal, unnamed capturing parenthesis. */    /* Handle a normal, unnamed capturing parenthesis. */
1420    
# Line 1278  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1438  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1438    
1439    else if (ptr[2] == CHAR_NUMBER_SIGN)    else if (ptr[2] == CHAR_NUMBER_SIGN)
1440      {      {
1441      for (ptr += 3; *ptr != 0; ptr++) if (*ptr == CHAR_RIGHT_PARENTHESIS) break;      for (ptr += 3; *ptr != CHAR_NULL; ptr++)
1442          if (*ptr == CHAR_RIGHT_PARENTHESIS) break;
1443      goto FAIL_EXIT;      goto FAIL_EXIT;
1444      }      }
1445    
# Line 1291  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1452  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1452      ptr += 2;      ptr += 2;
1453      if (ptr[1] != CHAR_QUESTION_MARK)      if (ptr[1] != CHAR_QUESTION_MARK)
1454        {        {
1455        while (*ptr != 0 && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;        while (*ptr != CHAR_NULL && *ptr != CHAR_RIGHT_PARENTHESIS) ptr++;
1456        if (*ptr != 0) ptr++;        if (*ptr != CHAR_NULL) ptr++;
1457        }        }
1458      }      }
1459    
# Line 1308  if (ptr[0] == CHAR_LEFT_PARENTHESIS) Line 1469  if (ptr[0] == CHAR_LEFT_PARENTHESIS)
1469      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&      if ((*ptr == CHAR_LESS_THAN_SIGN && ptr[1] != CHAR_EXCLAMATION_MARK &&
1470          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)          ptr[1] != CHAR_EQUALS_SIGN) || *ptr == CHAR_APOSTROPHE)
1471        {        {
1472        int term;        pcre_uchar term;
1473        const uschar *thisname;        const pcre_uchar *thisname;
1474        *count += 1;        *count += 1;
1475        if (name == NULL && *count == lorn) return *count;        if (name == NULL && *count == lorn) return *count;
1476        term = *ptr++;        term = *ptr++;
1477        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;        if (term == CHAR_LESS_THAN_SIGN) term = CHAR_GREATER_THAN_SIGN;
1478        thisname = ptr;        thisname = ptr;
1479        while (*ptr != term) ptr++;        while (*ptr != term) ptr++;
1480        if (name != NULL && lorn == ptr - thisname &&        if (name != NULL && lorn == (int)(ptr - thisname) &&
1481            strncmp((const char *)name, (const char *)thisname, lorn) == 0)            STRNCMP_UC_UC(name, thisname, (unsigned int)lorn) == 0)
1482          return *count;          return *count;
1483        term++;        term++;
1484        }        }
# Line 1335  for (; ptr < cd->end_pattern; ptr++) Line 1496  for (; ptr < cd->end_pattern; ptr++)
1496    
1497    if (*ptr == CHAR_BACKSLASH)    if (*ptr == CHAR_BACKSLASH)
1498      {      {
1499      if (*(++ptr) == 0) goto FAIL_EXIT;      if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1500      if (*ptr == CHAR_Q) for (;;)      if (*ptr == CHAR_Q) for (;;)
1501        {        {
1502        while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};        while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1503        if (*ptr == 0) goto FAIL_EXIT;        if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1504        if (*(++ptr) == CHAR_E) break;        if (*(++ptr) == CHAR_E) break;
1505        }        }
1506      continue;      continue;
# Line 1360  for (; ptr < cd->end_pattern; ptr++) Line 1521  for (; ptr < cd->end_pattern; ptr++)
1521          {          {
1522          if (ptr[2] == CHAR_E)          if (ptr[2] == CHAR_E)
1523            ptr+= 2;            ptr+= 2;
1524          else if (strncmp((const char *)ptr+2,          else if (STRNCMP_UC_C8(ptr + 2,
1525                   STR_Q STR_BACKSLASH STR_E, 3) == 0)                   STR_Q STR_BACKSLASH STR_E, 3) == 0)
1526            ptr += 4;            ptr += 4;
1527          else          else
# Line 1383  for (; ptr < cd->end_pattern; ptr++) Line 1544  for (; ptr < cd->end_pattern; ptr++)
1544    
1545      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)      while (*(++ptr) != CHAR_RIGHT_SQUARE_BRACKET)
1546        {        {
1547        if (*ptr == 0) return -1;        if (*ptr == CHAR_NULL) return -1;
1548        if (*ptr == CHAR_BACKSLASH)        if (*ptr == CHAR_BACKSLASH)
1549          {          {
1550          if (*(++ptr) == 0) goto FAIL_EXIT;          if (*(++ptr) == CHAR_NULL) goto FAIL_EXIT;
1551          if (*ptr == CHAR_Q) for (;;)          if (*ptr == CHAR_Q) for (;;)
1552            {            {
1553            while (*(++ptr) != 0 && *ptr != CHAR_BACKSLASH) {};            while (*(++ptr) != CHAR_NULL && *ptr != CHAR_BACKSLASH) {};
1554            if (*ptr == 0) goto FAIL_EXIT;            if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1555            if (*(++ptr) == CHAR_E) break;            if (*(++ptr) == CHAR_E) break;
1556            }            }
1557          continue;          continue;
# Line 1404  for (; ptr < cd->end_pattern; ptr++) Line 1565  for (; ptr < cd->end_pattern; ptr++)
1565    if (xmode && *ptr == CHAR_NUMBER_SIGN)    if (xmode && *ptr == CHAR_NUMBER_SIGN)
1566      {      {
1567      ptr++;      ptr++;
1568      while (*ptr != 0)      while (*ptr != CHAR_NULL)
1569        {        {
1570        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }        if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
1571        ptr++;        ptr++;
1572  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1573        if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;        if (utf) FORWARDCHAR(ptr);
1574  #endif  #endif
1575        }        }
1576      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1577      continue;      continue;
1578      }      }
1579    
# Line 1420  for (; ptr < cd->end_pattern; ptr++) Line 1581  for (; ptr < cd->end_pattern; ptr++)
1581    
1582    if (*ptr == CHAR_LEFT_PARENTHESIS)    if (*ptr == CHAR_LEFT_PARENTHESIS)
1583      {      {
1584      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, count);      int rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, count);
1585      if (rc > 0) return rc;      if (rc > 0) return rc;
1586      if (*ptr == 0) goto FAIL_EXIT;      if (*ptr == CHAR_NULL) goto FAIL_EXIT;
1587      }      }
1588    
1589    else if (*ptr == CHAR_RIGHT_PARENTHESIS)    else if (*ptr == CHAR_RIGHT_PARENTHESIS)
# Line 1466  Arguments: Line 1627  Arguments:
1627    name         name to seek, or NULL if seeking a numbered subpattern    name         name to seek, or NULL if seeking a numbered subpattern
1628    lorn         name length, or subpattern number if name is NULL    lorn         name length, or subpattern number if name is NULL
1629    xmode        TRUE if we are in /x mode    xmode        TRUE if we are in /x mode
1630    utf8         TRUE if we are in UTF-8 mode    utf          TRUE if we are in UTF-8 / UTF-16 / UTF-32 mode
1631    
1632  Returns:       the number of the found subpattern, or -1 if not found  Returns:       the number of the found subpattern, or -1 if not found
1633  */  */
1634    
1635  static int  static int
1636  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,
1637    BOOL utf8)    BOOL utf)
1638  {  {
1639  uschar *ptr = (uschar *)cd->start_pattern;  pcre_uchar *ptr = (pcre_uchar *)cd->start_pattern;
1640  int count = 0;  int count = 0;
1641  int rc;  int rc;
1642    
# Line 1486  matching closing parens. That is why we Line 1647  matching closing parens. That is why we
1647    
1648  for (;;)  for (;;)
1649    {    {
1650    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf8, &count);    rc = find_parens_sub(&ptr, cd, name, lorn, xmode, utf, &count);
1651    if (rc > 0 || *ptr++ == 0) break;    if (rc > 0 || *ptr++ == CHAR_NULL) break;
1652    }    }
1653    
1654  return rc;  return rc;
# Line 1513  Arguments: Line 1674  Arguments:
1674  Returns:       pointer to the first significant opcode  Returns:       pointer to the first significant opcode
1675  */  */
1676    
1677  static const uschar*  static const pcre_uchar*
1678  first_significant_code(const uschar *code, BOOL skipassert)  first_significant_code(const pcre_uchar *code, BOOL skipassert)
1679  {  {
1680  for (;;)  for (;;)
1681    {    {
# Line 1525  for (;;) Line 1686  for (;;)
1686      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1687      if (!skipassert) return code;      if (!skipassert) return code;
1688      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
1689      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1690      break;      break;
1691    
1692      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
# Line 1539  for (;;) Line 1700  for (;;)
1700      case OP_RREF:      case OP_RREF:
1701      case OP_NRREF:      case OP_NRREF:
1702      case OP_DEF:      case OP_DEF:
1703      code += _pcre_OP_lengths[*code];      code += PRIV(OP_lengths)[*code];
1704      break;      break;
1705    
1706      default:      default:
# Line 1569  and doing the check at the end; a flag s Line 1730  and doing the check at the end; a flag s
1730    
1731  Arguments:  Arguments:
1732    code     points to the start of the pattern (the bracket)    code     points to the start of the pattern (the bracket)
1733    utf8     TRUE in UTF-8 mode    utf      TRUE in UTF-8 / UTF-16 / UTF-32 mode
1734    atend    TRUE if called when the pattern is complete    atend    TRUE if called when the pattern is complete
1735    cd       the "compile data" structure    cd       the "compile data" structure
1736    
# Line 1581  Returns:   the fixed length, Line 1742  Returns:   the fixed length,
1742  */  */
1743    
1744  static int  static int
1745  find_fixedlength(uschar *code, BOOL utf8, BOOL atend, compile_data *cd)  find_fixedlength(pcre_uchar *code, BOOL utf, BOOL atend, compile_data *cd)
1746  {  {
1747  int length = -1;  int length = -1;
1748    
1749  register int branchlength = 0;  register int branchlength = 0;
1750  register uschar *cc = code + 1 + LINK_SIZE;  register pcre_uchar *cc = code + 1 + LINK_SIZE;
1751    
1752  /* 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
1753  branch, check the length against that of the other branches. */  branch, check the length against that of the other branches. */
# Line 1594  branch, check the length against that of Line 1755  branch, check the length against that of
1755  for (;;)  for (;;)
1756    {    {
1757    int d;    int d;
1758    uschar *ce, *cs;    pcre_uchar *ce, *cs;
1759    register int op = *cc;    register pcre_uchar op = *cc;
1760    
1761    switch (op)    switch (op)
1762      {      {
1763      /* We only need to continue for OP_CBRA (normal capturing bracket) and      /* We only need to continue for OP_CBRA (normal capturing bracket) and
# Line 1608  for (;;) Line 1770  for (;;)
1770      case OP_ONCE:      case OP_ONCE:
1771      case OP_ONCE_NC:      case OP_ONCE_NC:
1772      case OP_COND:      case OP_COND:
1773      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);
1774      if (d < 0) return d;      if (d < 0) return d;
1775      branchlength += d;      branchlength += d;
1776      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
# Line 1639  for (;;) Line 1801  for (;;)
1801    
1802      case OP_RECURSE:      case OP_RECURSE:
1803      if (!atend) return -3;      if (!atend) return -3;
1804      cs = ce = (uschar *)cd->start_code + GET(cc, 1);  /* Start subpattern */      cs = ce = (pcre_uchar *)cd->start_code + GET(cc, 1);  /* Start subpattern */
1805      do ce += GET(ce, 1); while (*ce == OP_ALT);       /* End subpattern */      do ce += GET(ce, 1); while (*ce == OP_ALT);           /* End subpattern */
1806      if (cc > cs && cc < ce) return -1;                /* Recursion */      if (cc > cs && cc < ce) return -1;                    /* Recursion */
1807      d = find_fixedlength(cs + 2, utf8, atend, cd);      d = find_fixedlength(cs + IMM2_SIZE, utf, atend, cd);
1808      if (d < 0) return d;      if (d < 0) return d;
1809      branchlength += d;      branchlength += d;
1810      cc += 1 + LINK_SIZE;      cc += 1 + LINK_SIZE;
# Line 1655  for (;;) Line 1817  for (;;)
1817      case OP_ASSERTBACK:      case OP_ASSERTBACK:
1818      case OP_ASSERTBACK_NOT:      case OP_ASSERTBACK_NOT:
1819      do cc += GET(cc, 1); while (*cc == OP_ALT);      do cc += GET(cc, 1); while (*cc == OP_ALT);
1820      /* Fall through */      cc += PRIV(OP_lengths)[*cc];
1821        break;
1822    
1823      /* Skip over things that don't match chars */      /* Skip over things that don't match chars */
1824    
# Line 1663  for (;;) Line 1826  for (;;)
1826      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
1827      case OP_SKIP_ARG:      case OP_SKIP_ARG:
1828      case OP_THEN_ARG:      case OP_THEN_ARG:
1829      cc += cc[1] + _pcre_OP_lengths[*cc];      cc += cc[1] + PRIV(OP_lengths)[*cc];
1830      break;      break;
1831    
1832      case OP_CALLOUT:      case OP_CALLOUT:
# Line 1690  for (;;) Line 1853  for (;;)
1853      case OP_SOM:      case OP_SOM:
1854      case OP_THEN:      case OP_THEN:
1855      case OP_WORD_BOUNDARY:      case OP_WORD_BOUNDARY:
1856      cc += _pcre_OP_lengths[*cc];      cc += PRIV(OP_lengths)[*cc];
1857      break;      break;
1858    
1859      /* Handle literal characters */      /* Handle literal characters */
# Line 1701  for (;;) Line 1864  for (;;)
1864      case OP_NOTI:      case OP_NOTI:
1865      branchlength++;      branchlength++;
1866      cc += 2;      cc += 2;
1867  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1868      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1869  #endif  #endif
1870      break;      break;
1871    
# Line 1713  for (;;) Line 1876  for (;;)
1876      case OP_EXACTI:      case OP_EXACTI:
1877      case OP_NOTEXACT:      case OP_NOTEXACT:
1878      case OP_NOTEXACTI:      case OP_NOTEXACTI:
1879      branchlength += GET2(cc,1);      branchlength += (int)GET2(cc,1);
1880      cc += 4;      cc += 2 + IMM2_SIZE;
1881  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
1882      if (utf8 && cc[-1] >= 0xc0) cc += _pcre_utf8_table4[cc[-1] & 0x3f];      if (utf && HAS_EXTRALEN(cc[-1])) cc += GET_EXTRALEN(cc[-1]);
1883  #endif  #endif
1884      break;      break;
1885    
1886      case OP_TYPEEXACT:      case OP_TYPEEXACT:
1887      branchlength += GET2(cc,1);      branchlength += GET2(cc,1);
1888      if (cc[3] == OP_PROP || cc[3] == OP_NOTPROP) cc += 2;      if (cc[1 + IMM2_SIZE] == OP_PROP || cc[1 + IMM2_SIZE] == OP_NOTPROP)
1889      cc += 4;        cc += 2;
1890        cc += 1 + IMM2_SIZE + 1;
1891      break;      break;
1892    
1893      /* Handle single-char matchers */      /* Handle single-char matchers */
# Line 1749  for (;;) Line 1913  for (;;)
1913      cc++;      cc++;
1914      break;      break;
1915    
1916      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;      /* The single-byte matcher isn't allowed. This only happens in UTF-8 mode;
1917      otherwise \C is coded as OP_ALLANY. */      otherwise \C is coded as OP_ALLANY. */
1918    
1919      case OP_ANYBYTE:      case OP_ANYBYTE:
# Line 1757  for (;;) Line 1921  for (;;)
1921    
1922      /* Check a class for variable quantification */      /* Check a class for variable quantification */
1923    
 #ifdef SUPPORT_UTF8  
     case OP_XCLASS:  
     cc += GET(cc, 1) - 33;  
     /* Fall through */  
 #endif  
   
1924      case OP_CLASS:      case OP_CLASS:
1925      case OP_NCLASS:      case OP_NCLASS:
1926      cc += 33;  #if defined SUPPORT_UTF || defined COMPILE_PCRE16 || defined COMPILE_PCRE32
1927        case OP_XCLASS:
1928        /* The original code caused an unsigned overflow in 64 bit systems,
1929        so now we use a conditional statement. */
1930        if (op == OP_XCLASS)
1931          cc += GET(cc, 1);
1932        else
1933          cc += PRIV(OP_lengths)[OP_CLASS];
1934    #else
1935        cc += PRIV(OP_lengths)[OP_CLASS];
1936    #endif
1937    
1938      switch (*cc)      switch (*cc)
1939        {        {
# Line 1779  for (;;) Line 1947  for (;;)
1947    
1948        case OP_CRRANGE:        case OP_CRRANGE:
1949        case OP_CRMINRANGE:        case OP_CRMINRANGE:
1950        if (GET2(cc,1) != GET2(cc,3)) return -1;        if (GET2(cc,1) != GET2(cc,1+IMM2_SIZE)) return -1;
1951        branchlength += GET2(cc,1);        branchlength += (int)GET2(cc,1);
1952        cc += 5;        cc += 1 + 2 * IMM2_SIZE;
1953        break;        break;
1954    
1955        default:        default:
# Line 1896  length. Line 2064  length.
2064    
2065  Arguments:  Arguments:
2066    code        points to start of expression    code        points to start of expression
2067    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2068    number      the required bracket number or negative to find a lookbehind    number      the required bracket number or negative to find a lookbehind
2069    
2070  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
2071  */  */
2072    
2073  const uschar *  const pcre_uchar *
2074  _pcre_find_bracket(const uschar *code, BOOL utf8, int number)  PRIV(find_bracket)(const pcre_uchar *code, BOOL utf, int number)
2075  {  {
2076  for (;;)  for (;;)
2077    {    {
2078    register int c = *code;    register pcre_uchar c = *code;
2079    
2080    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2081    
# Line 1921  for (;;) Line 2089  for (;;)
2089    
2090    else if (c == OP_REVERSE)    else if (c == OP_REVERSE)
2091      {      {
2092      if (number < 0) return (uschar *)code;      if (number < 0) return (pcre_uchar *)code;
2093      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2094      }      }
2095    
2096    /* Handle capturing bracket */    /* Handle capturing bracket */
# Line 1930  for (;;) Line 2098  for (;;)
2098    else if (c == OP_CBRA || c == OP_SCBRA ||    else if (c == OP_CBRA || c == OP_SCBRA ||
2099             c == OP_CBRAPOS || c == OP_SCBRAPOS)             c == OP_CBRAPOS || c == OP_SCBRAPOS)
2100      {      {
2101      int n = GET2(code, 1+LINK_SIZE);      int n = (int)GET2(code, 1+LINK_SIZE);
2102      if (n == number) return (uschar *)code;      if (n == number) return (pcre_uchar *)code;
2103      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2104      }      }
2105    
2106    /* 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 1960  for (;;) Line 2128  for (;;)
2128        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2129        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2130        case OP_TYPEPOSUPTO:        case OP_TYPEPOSUPTO:
2131        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2132            code += 2;
2133        break;        break;
2134    
2135        case OP_MARK:        case OP_MARK:
2136        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2137        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2138        case OP_THEN_ARG:        case OP_THEN_ARG:
2139        code += code[1];        code += code[1];
2140        break;        break;
# Line 1976  for (;;) Line 2142  for (;;)
2142    
2143      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2144    
2145      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2146    
2147    /* 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
2148    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
2149    arrange to skip the extra bytes. */    arrange to skip the extra bytes. */
2150    
2151  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2152      if (utf8) switch(c)      if (utf) switch(c)
2153        {        {
2154        case OP_CHAR:        case OP_CHAR:
2155        case OP_CHARI:        case OP_CHARI:
# Line 2013  for (;;) Line 2179  for (;;)
2179        case OP_MINQUERYI:        case OP_MINQUERYI:
2180        case OP_POSQUERY:        case OP_POSQUERY:
2181        case OP_POSQUERYI:        case OP_POSQUERYI:
2182        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2183        break;        break;
2184        }        }
2185  #else  #else
2186      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2187  #endif  #endif
2188      }      }
2189    }    }
# Line 2034  instance of OP_RECURSE. Line 2200  instance of OP_RECURSE.
2200    
2201  Arguments:  Arguments:
2202    code        points to start of expression    code        points to start of expression
2203    utf8        TRUE in UTF-8 mode    utf         TRUE in UTF-8 / UTF-16 / UTF-32 mode
2204    
2205  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
2206  */  */
2207    
2208  static const uschar *  static const pcre_uchar *
2209  find_recurse(const uschar *code, BOOL utf8)  find_recurse(const pcre_uchar *code, BOOL utf)
2210  {  {
2211  for (;;)  for (;;)
2212    {    {
2213    register int c = *code;    register pcre_uchar c = *code;
2214    if (c == OP_END) return NULL;    if (c == OP_END) return NULL;
2215    if (c == OP_RECURSE) return code;    if (c == OP_RECURSE) return code;
2216    
# Line 2079  for (;;) Line 2245  for (;;)
2245        case OP_TYPEUPTO:        case OP_TYPEUPTO:
2246        case OP_TYPEMINUPTO:        case OP_TYPEMINUPTO:
2247        case OP_TYPEEXACT:        case OP_TYPEEXACT:
2248        if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;        if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2249            code += 2;
2250        break;        break;
2251    
2252        case OP_MARK:        case OP_MARK:
2253        case OP_PRUNE_ARG:        case OP_PRUNE_ARG:
2254        case OP_SKIP_ARG:        case OP_SKIP_ARG:
       code += code[1];  
       break;  
   
2255        case OP_THEN_ARG:        case OP_THEN_ARG:
2256        code += code[1];        code += code[1];
2257        break;        break;
# Line 2095  for (;;) Line 2259  for (;;)
2259    
2260      /* Add in the fixed length from the table */      /* Add in the fixed length from the table */
2261    
2262      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2263    
2264      /* 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
2265      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
2266      to arrange to skip the extra bytes. */      to arrange to skip the extra bytes. */
2267    
2268  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2269      if (utf8) switch(c)      if (utf) switch(c)
2270        {        {
2271        case OP_CHAR:        case OP_CHAR:
2272        case OP_CHARI:        case OP_CHARI:
2273          case OP_NOT:
2274          case OP_NOTI:
2275        case OP_EXACT:        case OP_EXACT:
2276        case OP_EXACTI:        case OP_EXACTI:
2277          case OP_NOTEXACT:
2278          case OP_NOTEXACTI:
2279        case OP_UPTO:        case OP_UPTO:
2280        case OP_UPTOI:        case OP_UPTOI:
2281          case OP_NOTUPTO:
2282          case OP_NOTUPTOI:
2283        case OP_MINUPTO:        case OP_MINUPTO:
2284        case OP_MINUPTOI:        case OP_MINUPTOI:
2285          case OP_NOTMINUPTO:
2286          case OP_NOTMINUPTOI:
2287        case OP_POSUPTO:        case OP_POSUPTO:
2288        case OP_POSUPTOI:        case OP_POSUPTOI:
2289          case OP_NOTPOSUPTO:
2290          case OP_NOTPOSUPTOI:
2291        case OP_STAR:        case OP_STAR:
2292        case OP_STARI:        case OP_STARI:
2293          case OP_NOTSTAR:
2294          case OP_NOTSTARI:
2295        case OP_MINSTAR:        case OP_MINSTAR:
2296        case OP_MINSTARI:        case OP_MINSTARI:
2297          case OP_NOTMINSTAR:
2298          case OP_NOTMINSTARI:
2299        case OP_POSSTAR:        case OP_POSSTAR:
2300        case OP_POSSTARI:        case OP_POSSTARI:
2301          case OP_NOTPOSSTAR:
2302          case OP_NOTPOSSTARI:
2303        case OP_PLUS:        case OP_PLUS:
2304        case OP_PLUSI:        case OP_PLUSI:
2305          case OP_NOTPLUS:
2306          case OP_NOTPLUSI:
2307        case OP_MINPLUS:        case OP_MINPLUS:
2308        case OP_MINPLUSI:        case OP_MINPLUSI:
2309          case OP_NOTMINPLUS:
2310          case OP_NOTMINPLUSI:
2311        case OP_POSPLUS:        case OP_POSPLUS:
2312        case OP_POSPLUSI:        case OP_POSPLUSI:
2313          case OP_NOTPOSPLUS:
2314          case OP_NOTPOSPLUSI:
2315        case OP_QUERY:        case OP_QUERY:
2316        case OP_QUERYI:        case OP_QUERYI:
2317          case OP_NOTQUERY:
2318          case OP_NOTQUERYI:
2319        case OP_MINQUERY:        case OP_MINQUERY:
2320        case OP_MINQUERYI:        case OP_MINQUERYI:
2321          case OP_NOTMINQUERY:
2322          case OP_NOTMINQUERYI:
2323        case OP_POSQUERY:        case OP_POSQUERY:
2324        case OP_POSQUERYI:        case OP_POSQUERYI:
2325        if (code[-1] >= 0xc0) code += _pcre_utf8_table4[code[-1] & 0x3f];        case OP_NOTPOSQUERY:
2326          case OP_NOTPOSQUERYI:
2327          if (HAS_EXTRALEN(code[-1])) code += GET_EXTRALEN(code[-1]);
2328        break;        break;
2329        }        }
2330  #else  #else
2331      (void)(utf8);  /* Keep compiler happy by referencing function argument */      (void)(utf);  /* Keep compiler happy by referencing function argument */
2332  #endif  #endif
2333      }      }
2334    }    }
# Line 2159  bracket whose current branch will alread Line 2351  bracket whose current branch will alread
2351  Arguments:  Arguments:
2352    code        points to start of search    code        points to start of search
2353    endcode     points to where to stop    endcode     points to where to stop
2354    utf8        TRUE if in UTF8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2355    cd          contains pointers to tables etc.    cd          contains pointers to tables etc.
2356    
2357  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2358  */  */
2359    
2360  static BOOL  static BOOL
2361  could_be_empty_branch(const uschar *code, const uschar *endcode, BOOL utf8,  could_be_empty_branch(const pcre_uchar *code, const pcre_uchar *endcode,
2362    compile_data *cd)    BOOL utf, compile_data *cd)
2363  {  {
2364  register int c;  register pcre_uchar c;
2365  for (code = first_significant_code(code + _pcre_OP_lengths[*code], TRUE);  for (code = first_significant_code(code + PRIV(OP_lengths)[*code], TRUE);
2366       code < endcode;       code < endcode;
2367       code = first_significant_code(code + _pcre_OP_lengths[c], TRUE))       code = first_significant_code(code + PRIV(OP_lengths)[c], TRUE))
2368    {    {
2369    const uschar *ccode;    const pcre_uchar *ccode;
2370    
2371    c = *code;    c = *code;
2372    
# Line 2197  for (code = first_significant_code(code Line 2389  for (code = first_significant_code(code
2389    
2390    if (c == OP_RECURSE)    if (c == OP_RECURSE)
2391      {      {
2392      const uschar *scode;      const pcre_uchar *scode;
2393      BOOL empty_branch;      BOOL empty_branch;
2394    
2395      /* Test for forward reference */      /* Test for forward reference */
2396    
2397      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)      for (scode = cd->start_workspace; scode < cd->hwm; scode += LINK_SIZE)
2398        if (GET(scode, 0) == code + 1 - cd->start_code) return TRUE;        if ((int)GET(scode, 0) == (int)(code + 1 - cd->start_code)) return TRUE;
2399    
2400      /* Not a forward reference, test for completed backward reference */      /* Not a forward reference, test for completed backward reference */
2401    
# Line 2215  for (code = first_significant_code(code Line 2407  for (code = first_significant_code(code
2407    
2408      do      do
2409        {        {
2410        if (could_be_empty_branch(scode, endcode, utf8, cd))        if (could_be_empty_branch(scode, endcode, utf, cd))
2411          {          {
2412          empty_branch = TRUE;          empty_branch = TRUE;
2413          break;          break;
# Line 2233  for (code = first_significant_code(code Line 2425  for (code = first_significant_code(code
2425    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||    if (c == OP_BRAZERO || c == OP_BRAMINZERO || c == OP_SKIPZERO ||
2426        c == OP_BRAPOSZERO)        c == OP_BRAPOSZERO)
2427      {      {
2428      code += _pcre_OP_lengths[c];      code += PRIV(OP_lengths)[c];
2429      do code += GET(code, 1); while (*code == OP_ALT);      do code += GET(code, 1); while (*code == OP_ALT);
2430      c = *code;      c = *code;
2431      continue;      continue;
# Line 2271  for (code = first_significant_code(code Line 2463  for (code = first_significant_code(code
2463        empty_branch = FALSE;        empty_branch = FALSE;
2464        do        do
2465          {          {
2466          if (!empty_branch && could_be_empty_branch(code, endcode, utf8, cd))          if (!empty_branch && could_be_empty_branch(code, endcode, utf, cd))
2467            empty_branch = TRUE;            empty_branch = TRUE;
2468          code += GET(code, 1);          code += GET(code, 1);
2469          }          }
# Line 2289  for (code = first_significant_code(code Line 2481  for (code = first_significant_code(code
2481      {      {
2482      /* Check for quantifiers after a class. XCLASS is used for classes that      /* Check for quantifiers after a class. XCLASS is used for classes that
2483      cannot be represented just by a bit map. This includes negated single      cannot be represented just by a bit map. This includes negated single
2484      high-valued characters. The length in _pcre_OP_lengths[] is zero; the      high-valued characters. The length in PRIV(OP_lengths)[] is zero; the
2485      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"
2486      here. */      here. */
2487    
2488  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2489      case OP_XCLASS:      case OP_XCLASS:
2490      ccode = code += GET(code, 1);      ccode = code += GET(code, 1);
2491      goto CHECK_CLASS_REPEAT;      goto CHECK_CLASS_REPEAT;
# Line 2301  for (code = first_significant_code(code Line 2493  for (code = first_significant_code(code
2493    
2494      case OP_CLASS:      case OP_CLASS:
2495      case OP_NCLASS:      case OP_NCLASS:
2496      ccode = code + 33;      ccode = code + PRIV(OP_lengths)[OP_CLASS];
2497    
2498  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
2499      CHECK_CLASS_REPEAT:      CHECK_CLASS_REPEAT:
2500  #endif  #endif
2501    
# Line 2376  for (code = first_significant_code(code Line 2568  for (code = first_significant_code(code
2568      case OP_TYPEUPTO:      case OP_TYPEUPTO:
2569      case OP_TYPEMINUPTO:      case OP_TYPEMINUPTO:
2570      case OP_TYPEPOSUPTO:      case OP_TYPEPOSUPTO:
2571      if (code[3] == OP_PROP || code[3] == OP_NOTPROP) code += 2;      if (code[1 + IMM2_SIZE] == OP_PROP || code[1 + IMM2_SIZE] == OP_NOTPROP)
2572          code += 2;
2573      break;      break;
2574    
2575      /* End of branch */      /* End of branch */
# Line 2391  for (code = first_significant_code(code Line 2584  for (code = first_significant_code(code
2584      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,      /* In UTF-8 mode, STAR, MINSTAR, POSSTAR, QUERY, MINQUERY, POSQUERY, UPTO,
2585      MINUPTO, and POSUPTO may be followed by a multibyte character */      MINUPTO, and POSUPTO may be followed by a multibyte character */
2586    
2587  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
2588      case OP_STAR:      case OP_STAR:
2589      case OP_STARI:      case OP_STARI:
2590      case OP_MINSTAR:      case OP_MINSTAR:
# Line 2404  for (code = first_significant_code(code Line 2597  for (code = first_significant_code(code
2597      case OP_MINQUERYI:      case OP_MINQUERYI:
2598      case OP_POSQUERY:      case OP_POSQUERY:
2599      case OP_POSQUERYI:      case OP_POSQUERYI:
2600      if (utf8 && code[1] >= 0xc0) code += _pcre_utf8_table4[code[1] & 0x3f];      if (utf && HAS_EXTRALEN(code[1])) code += GET_EXTRALEN(code[1]);
2601      break;      break;
2602    
2603      case OP_UPTO:      case OP_UPTO:
# Line 2413  for (code = first_significant_code(code Line 2606  for (code = first_significant_code(code
2606      case OP_MINUPTOI:      case OP_MINUPTOI:
2607      case OP_POSUPTO:      case OP_POSUPTO:
2608      case OP_POSUPTOI:      case OP_POSUPTOI:
2609      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]);
2610      break;      break;
2611  #endif  #endif
2612    
# Line 2423  for (code = first_significant_code(code Line 2616  for (code = first_significant_code(code
2616      case OP_MARK:      case OP_MARK:
2617      case OP_PRUNE_ARG:      case OP_PRUNE_ARG:
2618      case OP_SKIP_ARG:      case OP_SKIP_ARG:
     code += code[1];  
     break;  
   
2619      case OP_THEN_ARG:      case OP_THEN_ARG:
2620      code += code[1];      code += code[1];
2621      break;      break;
# Line 2457  Arguments: Line 2647  Arguments:
2647    code        points to start of the recursion    code        points to start of the recursion
2648    endcode     points to where to stop (current RECURSE item)    endcode     points to where to stop (current RECURSE item)
2649    bcptr       points to the chain of current (unclosed) branch starts    bcptr       points to the chain of current (unclosed) branch starts
2650    utf8        TRUE if in UTF-8 mode    utf         TRUE if in UTF-8 / UTF-16 / UTF-32 mode
2651    cd          pointers to tables etc    cd          pointers to tables etc
2652    
2653  Returns:      TRUE if what is matched could be empty  Returns:      TRUE if what is matched could be empty
2654  */  */
2655    
2656  static BOOL  static BOOL
2657  could_be_empty(const uschar *code, const uschar *endcode, branch_chain *bcptr,  could_be_empty(const pcre_uchar *code, const pcre_uchar *endcode,
2658    BOOL utf8, compile_data *cd)    branch_chain *bcptr, BOOL utf, compile_data *cd)
2659  {  {
2660  while (bcptr != NULL && bcptr->current_branch >= code)  while (bcptr != NULL && bcptr->current_branch >= code)
2661    {    {
2662    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf8, cd))    if (!could_be_empty_branch(bcptr->current_branch, endcode, utf, cd))
2663      return FALSE;      return FALSE;
2664    bcptr = bcptr->outer;    bcptr = bcptr->outer;
2665    }    }
# Line 2521  Returns:   TRUE or FALSE Line 2711  Returns:   TRUE or FALSE
2711  */  */
2712    
2713  static BOOL  static BOOL
2714  check_posix_syntax(const uschar *ptr, const uschar **endptr)  check_posix_syntax(const pcre_uchar *ptr, const pcre_uchar **endptr)
2715  {  {
2716  int terminator;          /* Don't combine these lines; the Solaris cc */  pcre_uchar terminator;          /* Don't combine these lines; the Solaris cc */
2717  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */  terminator = *(++ptr);   /* compiler warns about "non-constant" initializer. */
2718  for (++ptr; *ptr != 0; ptr++)  for (++ptr; *ptr != CHAR_NULL; ptr++)
2719    {    {
2720    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)    if (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
2721      ptr++;      ptr++;
# Line 2565  Returns:     a value representing the na Line 2755  Returns:     a value representing the na
2755  */  */
2756    
2757  static int  static int
2758  check_posix_name(const uschar *ptr, int len)  check_posix_name(const pcre_uchar *ptr, int len)
2759  {  {
2760  const char *pn = posix_names;  const char *pn = posix_names;
2761  register int yield = 0;  register int yield = 0;
2762  while (posix_name_lengths[yield] != 0)  while (posix_name_lengths[yield] != 0)
2763    {    {
2764    if (len == posix_name_lengths[yield] &&    if (len == posix_name_lengths[yield] &&
2765      strncmp((const char *)ptr, pn, len) == 0) return yield;      STRNCMP_UC_C8(ptr, pn, (unsigned int)len) == 0) return yield;
2766    pn += posix_name_lengths[yield] + 1;    pn += posix_name_lengths[yield] + 1;
2767    yield++;    yield++;
2768    }    }
# Line 2604  value in the reference (which is a group Line 2794  value in the reference (which is a group
2794  Arguments:  Arguments:
2795    group      points to the start of the group    group      points to the start of the group
2796    adjust     the amount by which the group is to be moved    adjust     the amount by which the group is to be moved
2797    utf8       TRUE in UTF-8 mode    utf        TRUE in UTF-8 / UTF-16 / UTF-32 mode
2798    cd         contains pointers to tables etc.    cd         contains pointers to tables etc.
2799    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
2800    
# Line 2612  Returns:     nothing Line 2802  Returns:     nothing
2802  */  */
2803    
2804  static void  static void
2805  adjust_recurse(uschar *group, int adjust, BOOL utf8, compile_data *cd,  adjust_recurse(pcre_uchar *group, int adjust, BOOL utf, compile_data *cd,
2806    uschar *save_hwm)    pcre_uchar *save_hwm)
2807  {  {
2808  uschar *ptr = group;  pcre_uchar *ptr = group;
2809    
2810  while ((ptr = (uschar *)find_recurse(ptr, utf8)) != NULL)  while ((ptr = (pcre_uchar *)find_recurse(ptr, utf)) != NULL)
2811    {    {
2812    int offset;    int offset;
2813    uschar *hc;    pcre_uchar *hc;
2814    
2815    /* 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
2816    reference. */    reference. */
2817    
2818    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)    for (hc = save_hwm; hc < cd->hwm; hc += LINK_SIZE)
2819      {      {
2820      offset = GET(hc, 0);      offset = (int)GET(hc, 0);
2821      if (cd->start_code + offset == ptr + 1)      if (cd->start_code + offset == ptr + 1)
2822        {        {
2823        PUT(hc, 0, offset + adjust);        PUT(hc, 0, offset + adjust);
# Line 2640  while ((ptr = (uschar *)find_recurse(ptr Line 2830  while ((ptr = (uschar *)find_recurse(ptr
2830    
2831    if (hc >= cd->hwm)    if (hc >= cd->hwm)
2832      {      {
2833      offset = GET(ptr, 1);      offset = (int)GET(ptr, 1);
2834      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);      if (cd->start_code + offset >= group) PUT(ptr, 1, offset + adjust);
2835      }      }
2836    
# Line 2665  Arguments: Line 2855  Arguments:
2855  Returns:         new code pointer  Returns:         new code pointer
2856  */  */
2857    
2858  static uschar *  static pcre_uchar *
2859  auto_callout(uschar *code, const uschar *ptr, compile_data *cd)  auto_callout(pcre_uchar *code, const pcre_uchar *ptr, compile_data *cd)
2860  {  {
2861  *code++ = OP_CALLOUT;  *code++ = OP_CALLOUT;
2862  *code++ = 255;  *code++ = 255;
2863  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */  PUT(code, 0, (int)(ptr - cd->start_pattern));  /* Pattern offset */
2864  PUT(code, LINK_SIZE, 0);                       /* Default length */  PUT(code, LINK_SIZE, 0);                       /* Default length */
2865  return code + 2*LINK_SIZE;  return code + 2 * LINK_SIZE;
2866  }  }
2867    
2868    
# Line 2694  Returns:             nothing Line 2884  Returns:             nothing
2884  */  */
2885    
2886  static void  static void
2887  complete_callout(uschar *previous_callout, const uschar *ptr, compile_data *cd)  complete_callout(pcre_uchar *previous_callout, const pcre_uchar *ptr, compile_data *cd)
2888  {  {
2889  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));  int length = (int)(ptr - cd->start_pattern - GET(previous_callout, 2));
2890  PUT(previous_callout, 2 + LINK_SIZE, length);  PUT(previous_callout, 2 + LINK_SIZE, length);
# Line 2708  PUT(previous_callout, 2 + LINK_SIZE, len Line 2898  PUT(previous_callout, 2 + LINK_SIZE, len
2898  *************************************************/  *************************************************/
2899    
2900  /* 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
2901  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
2902  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
2903  start address.  start address. A character with multiple other cases is returned on its own
2904    with a special return value.
2905    
2906  Arguments:  Arguments:
2907    cptr        points to starting character value; updated    cptr        points to starting character value; updated
# Line 2718  Arguments: Line 2909  Arguments:
2909    ocptr       where to put start of othercase range    ocptr       where to put start of othercase range
2910    odptr       where to put end of othercase range    odptr       where to put end of othercase range
2911    
2912  Yield:        TRUE when range returned; FALSE when no more  Yield:        -1 when no more
2913                   0 when a range is returned
2914                  >0 the CASESET offset for char with multiple other cases
2915                    in this case, ocptr contains the original
2916  */  */
2917    
2918  static BOOL  static int
2919  get_othercase_range(unsigned int *cptr, unsigned int d, unsigned int *ocptr,  get_othercase_range(pcre_uint32 *cptr, pcre_uint32 d, pcre_uint32 *ocptr,
2920    unsigned int *odptr)    pcre_uint32 *odptr)
2921  {  {
2922  unsigned int c, othercase, next;  pcre_uint32 c, othercase, next;
2923    unsigned int co;
2924    
2925    /* Find the first character that has an other case. If it has multiple other
2926    cases, return its case offset value. */
2927    
2928  for (c = *cptr; c <= d; c++)  for (c = *cptr; c <= d; c++)
2929    { if ((othercase = UCD_OTHERCASE(c)) != c) break; }    {
2930      if ((co = UCD_CASESET(c)) != 0)
2931        {
2932        *ocptr = c++;   /* Character that has the set */
2933        *cptr = c;      /* Rest of input range */
2934        return (int)co;
2935        }
2936      if ((othercase = UCD_OTHERCASE(c)) != c) break;
2937      }
2938    
2939  if (c > d) return FALSE;  if (c > d) return -1;  /* Reached end of range */
2940    
2941  *ocptr = othercase;  *ocptr = othercase;
2942  next = othercase + 1;  next = othercase + 1;
# Line 2741  for (++c; c <= d; c++) Line 2947  for (++c; c <= d; c++)
2947    next++;    next++;
2948    }    }
2949    
2950  *odptr = next - 1;  *odptr = next - 1;     /* End of othercase range */
2951  *cptr = c;  *cptr = c;             /* Rest of input range */
2952    return 0;
 return TRUE;  
2953  }  }
2954    
2955    
# Line 2766  Returns:       TRUE if auto-possessifyin Line 2971  Returns:       TRUE if auto-possessifyin
2971  */  */
2972    
2973  static BOOL  static BOOL
2974  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)
2975  {  {
2976    #ifdef SUPPORT_UCP
2977    const pcre_uint32 *p;
2978    #endif
2979    
2980  const ucd_record *prop = GET_UCD(c);  const ucd_record *prop = GET_UCD(c);
2981    
2982  switch(ptype)  switch(ptype)
2983    {    {
2984    case PT_LAMP:    case PT_LAMP:
# Line 2777  switch(ptype) Line 2987  switch(ptype)
2987            prop->chartype == ucp_Lt) == negated;            prop->chartype == ucp_Lt) == negated;
2988    
2989    case PT_GC:    case PT_GC:
2990    return (pdata == _pcre_ucp_gentype[prop->chartype]) == negated;    return (pdata == PRIV(ucp_gentype)[prop->chartype]) == negated;
2991    
2992    case PT_PC:    case PT_PC:
2993    return (pdata == prop->chartype) == negated;    return (pdata == prop->chartype) == negated;
# Line 2788  switch(ptype) Line 2998  switch(ptype)
2998    /* These are specials */    /* These are specials */
2999    
3000    case PT_ALNUM:    case PT_ALNUM:
3001    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3002            _pcre_ucp_gentype[prop->chartype] == ucp_N) == negated;            PRIV(ucp_gentype)[prop->chartype] == ucp_N) == negated;
3003    
3004    case PT_SPACE:    /* Perl space */    case PT_SPACE:    /* Perl space */
3005    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3006            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)            c == CHAR_HT || c == CHAR_NL || c == CHAR_FF || c == CHAR_CR)
3007            == negated;            == negated;
3008    
3009    case PT_PXSPACE:  /* POSIX space */    case PT_PXSPACE:  /* POSIX space */
3010    return (_pcre_ucp_gentype[prop->chartype] == ucp_Z ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_Z ||
3011            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||            c == CHAR_HT || c == CHAR_NL || c == CHAR_VT ||
3012            c == CHAR_FF || c == CHAR_CR)            c == CHAR_FF || c == CHAR_CR)
3013            == negated;            == negated;
3014    
3015    case PT_WORD:    case PT_WORD:
3016    return (_pcre_ucp_gentype[prop->chartype] == ucp_L ||    return (PRIV(ucp_gentype)[prop->chartype] == ucp_L ||
3017            _pcre_ucp_gentype[prop->chartype] == ucp_N ||            PRIV(ucp_gentype)[prop->chartype] == ucp_N ||
3018            c == CHAR_UNDERSCORE) == negated;            c == CHAR_UNDERSCORE) == negated;
3019    
3020    #ifdef SUPPORT_UCP
3021      case PT_CLIST:
3022      p = PRIV(ucd_caseless_sets) + prop->caseset;
3023      for (;;)
3024        {
3025        if (c < *p) return !negated;
3026        if (c == *p++) return negated;
3027        }
3028      break;  /* Control never reaches here */
3029    #endif
3030    }    }
3031    
3032  return FALSE;  return FALSE;
3033  }  }
3034  #endif  /* SUPPORT_UCP */  #endif  /* SUPPORT_UCP */
# Line 2823  sense to automatically possessify the re Line 3045  sense to automatically possessify the re
3045    
3046  Arguments:  Arguments:
3047    previous      pointer to the repeated opcode    previous      pointer to the repeated opcode
3048    utf8          TRUE in UTF-8 mode    utf           TRUE in UTF-8 / UTF-16 / UTF-32 mode
3049    ptr           next character in pattern    ptr           next character in pattern
3050    options       options bits    options       options bits
3051    cd            contains pointers to tables etc.    cd            contains pointers to tables etc.
# Line 2832  Returns:        TRUE if possessifying is Line 3054  Returns:        TRUE if possessifying is
3054  */  */
3055    
3056  static BOOL  static BOOL
3057  check_auto_possessive(const uschar *previous, BOOL utf8, const uschar *ptr,  check_auto_possessive(const pcre_uchar *previous, BOOL utf,
3058    int options, compile_data *cd)    const pcre_uchar *ptr, int options, compile_data *cd)
3059  {  {
3060  int c, next;  pcre_uint32 c = NOTACHAR;
3061  int op_code = *previous++;  pcre_uint32 next;
3062    int escape;
3063    pcre_uchar op_code = *previous++;
3064    
3065  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
3066    
# Line 2844  if ((options & PCRE_EXTENDED) != 0) Line 3068  if ((options & PCRE_EXTENDED) != 0)
3068    {    {
3069    for (;;)    for (;;)
3070      {      {
3071      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3072      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3073        {        {
3074        ptr++;        ptr++;
3075        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3076          {          {
3077          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3078          ptr++;          ptr++;
3079  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3080          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3081  #endif  #endif
3082          }          }
3083        }        }
# Line 2867  value is a character, a negative value i Line 3091  value is a character, a negative value i
3091  if (*ptr == CHAR_BACKSLASH)  if (*ptr == CHAR_BACKSLASH)
3092    {    {
3093    int temperrorcode = 0;    int temperrorcode = 0;
3094    next = check_escape(&ptr, &temperrorcode, cd->bracount, options, FALSE);    escape = check_escape(&ptr, &next, &temperrorcode, cd->bracount, options,
3095        FALSE);
3096    if (temperrorcode != 0) return FALSE;    if (temperrorcode != 0) return FALSE;
3097    ptr++;    /* Point after the escape sequence */    ptr++;    /* Point after the escape sequence */
3098    }    }
3099    else if (!MAX_255(*ptr) || (cd->ctypes[*ptr] & ctype_meta) == 0)
 else if ((cd->ctypes[*ptr] & ctype_meta) == 0)  
3100    {    {
3101  #ifdef SUPPORT_UTF8    escape = 0;
3102    if (utf8) { GETCHARINC(next, ptr); } else  #ifdef SUPPORT_UTF
3103      if (utf) { GETCHARINC(next, ptr); } else
3104  #endif  #endif
3105    next = *ptr++;    next = *ptr++;
3106    }    }
   
3107  else return FALSE;  else return FALSE;
3108    
3109  /* Skip whitespace and comments in extended mode */  /* Skip whitespace and comments in extended mode */
# Line 2888  if ((options & PCRE_EXTENDED) != 0) Line 3112  if ((options & PCRE_EXTENDED) != 0)
3112    {    {
3113    for (;;)    for (;;)
3114      {      {
3115      while ((cd->ctypes[*ptr] & ctype_space) != 0) ptr++;      while (MAX_255(*ptr) && (cd->ctypes[*ptr] & ctype_space) != 0) ptr++;
3116      if (*ptr == CHAR_NUMBER_SIGN)      if (*ptr == CHAR_NUMBER_SIGN)
3117        {        {
3118        ptr++;        ptr++;
3119        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3120          {          {
3121          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen; break; }
3122          ptr++;          ptr++;
3123  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3124          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3125  #endif  #endif
3126          }          }
3127        }        }
# Line 2908  if ((options & PCRE_EXTENDED) != 0) Line 3132  if ((options & PCRE_EXTENDED) != 0)
3132  /* If the next thing is itself optional, we have to give up. */  /* If the next thing is itself optional, we have to give up. */
3133    
3134  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||  if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3135    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)
3136      return FALSE;      return FALSE;
3137    
3138  /* 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. */  
3139    
3140  if (next >= 0) switch(op_code)  if (op_code == OP_CHAR || op_code == OP_CHARI ||
3141        op_code == OP_NOT || op_code == OP_NOTI)
3142    {    {
3143    case OP_CHAR:  #ifdef SUPPORT_UTF
 #ifdef SUPPORT_UTF8  
3144    GETCHARTEST(c, previous);    GETCHARTEST(c, previous);
3145  #else  #else
3146    c = *previous;    c = *previous;
3147  #endif  #endif
3148    return c != next;    }
3149    
3150    /* 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
3151    Unicode property support, we can use it to test the other case of  the next item is a character. */
   high-valued characters. */  
3152    
3153    case OP_CHARI:  if (escape == 0)
3154  #ifdef SUPPORT_UTF8    {
3155    GETCHARTEST(c, previous);    /* For a caseless UTF match, the next character may have more than one other
3156  #else    case, which maps to the special PT_CLIST property. Check this first. */
3157    c = *previous;  
3158    #ifdef SUPPORT_UCP
3159      if (utf && c != NOTACHAR && (options & PCRE_CASELESS) != 0)
3160        {
3161        unsigned int ocs = UCD_CASESET(next);
3162        if (ocs > 0) return check_char_prop(c, PT_CLIST, ocs, op_code >= OP_NOT);
3163        }
3164  #endif  #endif
3165    if (c == next) return FALSE;  
3166  #ifdef SUPPORT_UTF8    switch(op_code)
   if (utf8)  
3167      {      {
3168      unsigned int othercase;      case OP_CHAR:
3169      if (next < 128) othercase = cd->fcc[next]; else      return c != next;
3170    
3171        /* For CHARI (caseless character) we must check the other case. If we have
3172        Unicode property support, we can use it to test the other case of
3173        high-valued characters. We know that next can have only one other case,
3174        because multi-other-case characters are dealt with above. */
3175    
3176        case OP_CHARI:
3177        if (c == next) return FALSE;
3178    #ifdef SUPPORT_UTF
3179        if (utf)
3180          {
3181          pcre_uint32 othercase;
3182          if (next < 128) othercase = cd->fcc[next]; else
3183  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3184      othercase = UCD_OTHERCASE((unsigned int)next);        othercase = UCD_OTHERCASE(next);
3185  #else  #else
3186      othercase = NOTACHAR;        othercase = NOTACHAR;
3187  #endif  #endif
3188      return (unsigned int)c != othercase;        return c != othercase;
3189      }        }
3190    else      else
3191  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3192    return (c != cd->fcc[next]);  /* Non-UTF-8 mode */      return (c != TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3193    
3194    /* For OP_NOT and OP_NOTI, the data is always a single-byte character. These      case OP_NOT:
3195    opcodes are not used for multi-byte characters, because they are coded using      return c == next;
3196    an XCLASS instead. */  
3197        case OP_NOTI:
3198    case OP_NOT:      if (c == next) return TRUE;
3199    return (c = *previous) == next;  #ifdef SUPPORT_UTF
3200        if (utf)
3201    case OP_NOTI:        {
3202    if ((c = *previous) == next) return TRUE;        pcre_uint32 othercase;
3203  #ifdef SUPPORT_UTF8        if (next < 128) othercase = cd->fcc[next]; else
   if (utf8)  
     {  
     unsigned int othercase;  
     if (next < 128) othercase = cd->fcc[next]; else  
3204  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3205      othercase = UCD_OTHERCASE(next);        othercase = UCD_OTHERCASE(next);
3206  #else  #else
3207      othercase = NOTACHAR;        othercase = NOTACHAR;
3208  #endif  #endif
3209      return (unsigned int)c == othercase;        return c == othercase;
3210      }        }
3211    else      else
3212  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
3213    return (c == cd->fcc[next]);  /* Non-UTF-8 mode */      return (c == TABLE_GET(next, cd->fcc, next));  /* Not UTF */
3214    
3215    /* 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.
3216    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. */
3217    
3218    case OP_DIGIT:      case OP_DIGIT:
3219    return next > 127 || (cd->ctypes[next] & ctype_digit) == 0;      return next > 255 || (cd->ctypes[next] & ctype_digit) == 0;
3220    
3221    case OP_NOT_DIGIT:      case OP_NOT_DIGIT:
3222    return next <= 127 && (cd->ctypes[next] & ctype_digit) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_digit) != 0;
3223    
3224    case OP_WHITESPACE:      case OP_WHITESPACE:
3225    return next > 127 || (cd->ctypes[next] & ctype_space) == 0;      return next > 255 || (cd->ctypes[next] & ctype_space) == 0;
3226    
3227    case OP_NOT_WHITESPACE:      case OP_NOT_WHITESPACE:
3228    return next <= 127 && (cd->ctypes[next] & ctype_space) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_space) != 0;
3229    
3230    case OP_WORDCHAR:      case OP_WORDCHAR:
3231    return next > 127 || (cd->ctypes[next] & ctype_word) == 0;      return next > 255 || (cd->ctypes[next] & ctype_word) == 0;
3232    
3233    case OP_NOT_WORDCHAR:      case OP_NOT_WORDCHAR:
3234    return next <= 127 && (cd->ctypes[next] & ctype_word) != 0;      return next <= 255 && (cd->ctypes[next] & ctype_word) != 0;
3235    
3236    case OP_HSPACE:      case OP_HSPACE:
3237    case OP_NOT_HSPACE:      case OP_NOT_HSPACE:
3238    switch(next)      switch(next)
3239      {        {
3240      case 0x09:        HSPACE_CASES:
3241      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;  
     }  
3242    
3243    case OP_ANYNL:        default:
3244    case OP_VSPACE:        return op_code != OP_NOT_HSPACE;
3245    case OP_NOT_VSPACE:        }
3246    switch(next)  
3247      {      case OP_ANYNL:
3248      case 0x0a:      case OP_VSPACE:
3249      case 0x0b:      case OP_NOT_VSPACE:
3250      case 0x0c:      switch(next)
3251      case 0x0d:        {
3252      case 0x85:        VSPACE_CASES:
3253      case 0x2028:        return op_code == OP_NOT_VSPACE;
3254      case 0x2029:  
3255      return op_code == OP_NOT_VSPACE;        default:
3256      default:        return op_code != OP_NOT_VSPACE;
3257      return op_code != OP_NOT_VSPACE;        }
     }  
3258    
3259  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3260    case OP_PROP:      case OP_PROP:
3261    return check_char_prop(next, previous[0], previous[1], FALSE);      return check_char_prop(next, previous[0], previous[1], FALSE);
3262    
3263    case OP_NOTPROP:      case OP_NOTPROP:
3264    return check_char_prop(next, previous[0], previous[1], TRUE);      return check_char_prop(next, previous[0], previous[1], TRUE);
3265  #endif  #endif
3266    
3267    default:      default:
3268    return FALSE;      return FALSE;
3269        }
3270    }    }
3271    
   
3272  /* 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
3273  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
3274  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 3065  switch(op_code) Line 3279  switch(op_code)
3279    {    {
3280    case OP_CHAR:    case OP_CHAR:
3281    case OP_CHARI:    case OP_CHARI:
3282  #ifdef SUPPORT_UTF8    switch(escape)
   GETCHARTEST(c, previous);  
 #else  
   c = *previous;  
 #endif  
   switch(-next)  
3283      {      {
3284      case ESC_d:      case ESC_d:
3285      return c > 127 || (cd->ctypes[c] & ctype_digit) == 0;      return c > 255 || (cd->ctypes[c] & ctype_digit) == 0;
3286    
3287      case ESC_D:      case ESC_D:
3288      return c <= 127 && (cd->ctypes[c] & ctype_digit) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_digit) != 0;
3289    
3290      case ESC_s:      case ESC_s:
3291      return c > 127 || (cd->ctypes[c] & ctype_space) == 0;      return c > 255 || (cd->ctypes[c] & ctype_space) == 0;
3292    
3293      case ESC_S:      case ESC_S:
3294      return c <= 127 && (cd->ctypes[c] & ctype_space) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_space) != 0;
3295    
3296      case ESC_w:      case ESC_w:
3297      return c > 127 || (cd->ctypes[c] & ctype_word) == 0;      return c > 255 || (cd->ctypes[c] & ctype_word) == 0;
3298    
3299      case ESC_W:      case ESC_W:
3300      return c <= 127 && (cd->ctypes[c] & ctype_word) != 0;      return c <= 255 && (cd->ctypes[c] & ctype_word) != 0;
3301    
3302      case ESC_h:      case ESC_h:
3303      case ESC_H:      case ESC_H:
3304      switch(c)      switch(c)
3305        {        {
3306        case 0x09:        HSPACE_CASES:
3307        case 0x20:        return escape != ESC_h;
3308        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;  
3309        default:        default:
3310        return -next == ESC_h;        return escape == ESC_h;
3311        }        }
3312    
3313      case ESC_v:      case ESC_v:
3314      case ESC_V:      case ESC_V:
3315      switch(c)      switch(c)
3316        {        {
3317        case 0x0a:        VSPACE_CASES:
3318        case 0x0b:        return escape != ESC_v;
3319        case 0x0c:  
       case 0x0d:  
       case 0x85:  
       case 0x2028:  
       case 0x2029:  
       return -next != ESC_v;  
3320        default:        default:
3321        return -next == ESC_v;        return escape == ESC_v;
3322        }        }
3323    
3324      /* 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
3325      their substitutions and process them. The result will always be either      their substitutions and process them. The result will always be either
3326      -ESC_p or -ESC_P. Then fall through to process those values. */      ESC_p or ESC_P. Then fall through to process those values. */
3327    
3328  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
3329      case ESC_du:      case ESC_du:
# Line 3147  switch(op_code) Line 3334  switch(op_code)
3334      case ESC_SU:      case ESC_SU:
3335        {        {
3336        int temperrorcode = 0;        int temperrorcode = 0;
3337        ptr = substitutes[-next - ESC_DU];        ptr = substitutes[escape - ESC_DU];
3338        next = check_escape(&ptr, &temperrorcode, 0, options, FALSE);        escape = check_escape(&ptr, &next, &temperrorcode, 0, options, FALSE);
3339        if (temperrorcode != 0) return FALSE;        if (temperrorcode != 0) return FALSE;
3340        ptr++;    /* For compatibility */        ptr++;    /* For compatibility */
3341        }        }
# Line 3157  switch(op_code) Line 3344  switch(op_code)
3344      case ESC_p:      case ESC_p:
3345      case ESC_P:      case ESC_P:
3346        {        {
3347        int ptype, pdata, errorcodeptr;        unsigned int ptype = 0, pdata = 0;
3348          int errorcodeptr;
3349        BOOL negated;        BOOL negated;
3350    
3351        ptr--;      /* Make ptr point at the p or P */        ptr--;      /* Make ptr point at the p or P */
3352        ptype = get_ucp(&ptr, &negated, &pdata, &errorcodeptr);        if (!get_ucp(&ptr, &negated, &ptype, &pdata, &errorcodeptr))
3353        if (ptype < 0) return FALSE;          return FALSE;
3354        ptr++;      /* Point past the final curly ket */        ptr++;      /* Point past the final curly ket */
3355    
3356        /* 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 3170  switch(op_code) Line 3358  switch(op_code)
3358        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. */
3359    
3360        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||        if (*ptr == CHAR_ASTERISK || *ptr == CHAR_QUESTION_MARK ||
3361          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)
3362            return FALSE;            return FALSE;
3363    
3364        /* Do the property check. */        /* Do the property check. */
3365    
3366        return check_char_prop(c, ptype, pdata, (next == -ESC_P) != negated);        return check_char_prop(c, ptype, pdata, (escape == ESC_P) != negated);
3367        }        }
3368  #endif  #endif
3369    
# Line 3190  switch(op_code) Line 3378  switch(op_code)
3378    these op-codes are never generated.) */    these op-codes are never generated.) */
3379    
3380    case OP_DIGIT:    case OP_DIGIT:
3381    return next == -ESC_D || next == -ESC_s || next == -ESC_W ||    return escape == ESC_D || escape == ESC_s || escape == ESC_W ||
3382           next == -ESC_h || next == -ESC_v || next == -ESC_R;           escape == ESC_h || escape == ESC_v || escape == ESC_R;
3383    
3384    case OP_NOT_DIGIT:    case OP_NOT_DIGIT:
3385    return next == -ESC_d;    return escape == ESC_d;
3386    
3387    case OP_WHITESPACE:    case OP_WHITESPACE:
3388    return next == -ESC_S || next == -ESC_d || next == -ESC_w || next == -ESC_R;    return escape == ESC_S || escape == ESC_d || escape == ESC_w;
3389    
3390    case OP_NOT_WHITESPACE:    case OP_NOT_WHITESPACE:
3391    return next == -ESC_s || next == -ESC_h || next == -ESC_v;    return escape == ESC_s || escape == ESC_h || escape == ESC_v || escape == ESC_R;
3392    
3393    case OP_HSPACE:    case OP_HSPACE:
3394    return next == -ESC_S || next == -ESC_H || next == -ESC_d ||    return escape == ESC_S || escape == ESC_H || escape == ESC_d ||
3395           next == -ESC_w || next == -ESC_v || next == -ESC_R;           escape == ESC_w || escape == ESC_v || escape == ESC_R;
3396    
3397    case OP_NOT_HSPACE:    case OP_NOT_HSPACE:
3398    return next == -ESC_h;    return escape == ESC_h;
3399    
3400    /* Can't have \S in here because VT matches \S (Perl anomaly) */    /* Can't have \S in here because VT matches \S (Perl anomaly) */
3401    case OP_ANYNL:    case OP_ANYNL:
3402    case OP_VSPACE:    case OP_VSPACE:
3403    return next == -ESC_V || next == -ESC_d || next == -ESC_w;    return escape == ESC_V || escape == ESC_d || escape == ESC_w;
3404    
3405    case OP_NOT_VSPACE:    case OP_NOT_VSPACE:
3406    return next == -ESC_v || next == -ESC_R;    return escape == ESC_v || escape == ESC_R;
3407    
3408    case OP_WORDCHAR:    case OP_WORDCHAR:
3409    return next == -ESC_W || next == -ESC_s || next == -ESC_h ||    return escape == ESC_W || escape == ESC_s || escape == ESC_h ||
3410           next == -ESC_v || next == -ESC_R;           escape == ESC_v || escape == ESC_R;
3411    
3412    case OP_NOT_WORDCHAR:    case OP_NOT_WORDCHAR:
3413    return next == -ESC_w || next == -ESC_d;    return escape == ESC_w || escape == ESC_d;
3414    
3415    default:    default:
3416    return FALSE;    return FALSE;
# Line 3234  switch(op_code) Line 3422  switch(op_code)
3422    
3423    
3424  /*************************************************  /*************************************************
3425    *        Add a character or range to a class     *
3426    *************************************************/
3427    
3428    /* This function packages up the logic of adding a character or range of
3429    characters to a class. The character values in the arguments will be within the
3430    valid values for the current mode (8-bit, 16-bit, UTF, etc). This function is
3431    mutually recursive with the function immediately below.
3432    
3433    Arguments:
3434      classbits     the bit map for characters < 256
3435      uchardptr     points to the pointer for extra data
3436      options       the options word
3437      cd            contains pointers to tables etc.
3438      start         start of range character
3439      end           end of range character
3440    
3441    Returns:        the number of < 256 characters added
3442                    the pointer to extra data is updated
3443    */
3444    
3445    static int
3446    add_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3447      compile_data *cd, pcre_uint32 start, pcre_uint32 end)
3448    {
3449    pcre_uint32 c;
3450    int n8 = 0;
3451    
3452    /* If caseless matching is required, scan the range and process alternate
3453    cases. In Unicode, there are 8-bit characters that have alternate cases that
3454    are greater than 255 and vice-versa. Sometimes we can just extend the original
3455    range. */
3456    
3457    if ((options & PCRE_CASELESS) != 0)
3458      {
3459    #ifdef SUPPORT_UCP
3460      if ((options & PCRE_UTF8) != 0)
3461        {
3462        int rc;
3463        pcre_uint32 oc, od;
3464    
3465        options &= ~PCRE_CASELESS;   /* Remove for recursive calls */
3466        c = start;
3467    
3468        while ((rc = get_othercase_range(&c, end, &oc, &od)) >= 0)
3469          {
3470          /* Handle a single character that has more than one other case. */
3471    
3472          if (rc > 0) n8 += add_list_to_class(classbits, uchardptr, options, cd,
3473            PRIV(ucd_caseless_sets) + rc, oc);
3474    
3475          /* Do nothing if the other case range is within the original range. */
3476    
3477          else if (oc >= start && od <= end) continue;
3478    
3479          /* Extend the original range if there is overlap, noting that if oc < c, we
3480          can't have od > end because a subrange is always shorter than the basic
3481          range. Otherwise, use a recursive call to add the additional range. */
3482    
3483          else if (oc < start && od >= start - 1) start = oc; /* Extend downwards */
3484          else if (od > end && oc <= end + 1) end = od;       /* Extend upwards */
3485          else n8 += add_to_class(classbits, uchardptr, options, cd, oc, od);
3486          }
3487        }
3488      else
3489    #endif  /* SUPPORT_UCP */
3490    
3491      /* Not UTF-mode, or no UCP */
3492    
3493      for (c = start; c <= end && c < 256; c++)
3494        {
3495        SETBIT(classbits, cd->fcc[c]);
3496        n8++;
3497        }
3498      }
3499    
3500    /* Now handle the original range. Adjust the final value according to the bit
3501    length - this means that the same lists of (e.g.) horizontal spaces can be used
3502    in all cases. */
3503    
3504    #if defined COMPILE_PCRE8
3505    #ifdef SUPPORT_UTF
3506      if ((options & PCRE_UTF8) == 0)
3507    #endif
3508      if (end > 0xff) end = 0xff;
3509    
3510    #elif defined COMPILE_PCRE16
3511    #ifdef SUPPORT_UTF
3512      if ((options & PCRE_UTF16) == 0)
3513    #endif
3514      if (end > 0xffff) end = 0xffff;
3515    
3516    #endif /* COMPILE_PCRE[8|16] */
3517    
3518    /* If all characters are less than 256, use the bit map. Otherwise use extra
3519    data. */
3520    
3521    if (end < 0x100)
3522      {
3523      for (c = start; c <= end; c++)
3524        {
3525        n8++;
3526        SETBIT(classbits, c);
3527        }
3528      }
3529    
3530    else
3531      {
3532      pcre_uchar *uchardata = *uchardptr;
3533    
3534    #ifdef SUPPORT_UTF
3535      if ((options & PCRE_UTF8) != 0)  /* All UTFs use the same flag bit */
3536        {
3537        if (start < end)
3538          {
3539          *uchardata++ = XCL_RANGE;
3540          uchardata += PRIV(ord2utf)(start, uchardata);
3541          uchardata += PRIV(ord2utf)(end, uchardata);
3542          }
3543        else if (start == end)
3544          {
3545          *uchardata++ = XCL_SINGLE;
3546          uchardata += PRIV(ord2utf)(start, uchardata);
3547          }
3548        }
3549      else
3550    #endif  /* SUPPORT_UTF */
3551    
3552      /* Without UTF support, character values are constrained by the bit length,
3553      and can only be > 256 for 16-bit and 32-bit libraries. */
3554    
3555    #ifdef COMPILE_PCRE8
3556        {}
3557    #else
3558      if (start < end)
3559        {
3560        *uchardata++ = XCL_RANGE;
3561        *uchardata++ = start;
3562        *uchardata++ = end;
3563        }
3564      else if (start == end)
3565        {
3566        *uchardata++ = XCL_SINGLE;
3567        *uchardata++ = start;
3568        }
3569    #endif
3570    
3571      *uchardptr = uchardata;   /* Updata extra data pointer */
3572      }
3573    
3574    return n8;    /* Number of 8-bit characters */
3575    }
3576    
3577    
3578    
3579    
3580    /*************************************************
3581    *        Add a list of characters to a class     *
3582    *************************************************/
3583    
3584    /* This function is used for adding a list of case-equivalent characters to a
3585    class, and also for adding a list of horizontal or vertical whitespace. If the
3586    list is in order (which it should be), ranges of characters are detected and
3587    handled appropriately. This function is mutually recursive with the function
3588    above.
3589    
3590    Arguments:
3591      classbits     the bit map for characters < 256
3592      uchardptr     points to the pointer for extra data
3593      options       the options word
3594      cd            contains pointers to tables etc.
3595      p             points to row of 32-bit values, terminated by NOTACHAR
3596      except        character to omit; this is used when adding lists of
3597                      case-equivalent characters to avoid including the one we
3598                      already know about
3599    
3600    Returns:        the number of < 256 characters added
3601                    the pointer to extra data is updated
3602    */
3603    
3604    static int
3605    add_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr, int options,
3606      compile_data *cd, const pcre_uint32 *p, unsigned int except)
3607    {
3608    int n8 = 0;
3609    while (p[0] < NOTACHAR)
3610      {
3611      int n = 0;
3612      if (p[0] != except)
3613        {
3614        while(p[n+1] == p[0] + n + 1) n++;
3615        n8 += add_to_class(classbits, uchardptr, options, cd, p[0], p[n]);
3616        }
3617      p += n + 1;
3618      }
3619    return n8;
3620    }
3621    
3622    
3623    
3624    /*************************************************
3625    *    Add characters not in a list to a class     *
3626    *************************************************/
3627    
3628    /* This function is used for adding the complement of a list of horizontal or
3629    vertical whitespace to a class. The list must be in order.
3630    
3631    Arguments:
3632      classbits     the bit map for characters < 256
3633      uchardptr     points to the pointer for extra data
3634      options       the options word
3635      cd            contains pointers to tables etc.
3636      p             points to row of 32-bit values, terminated by NOTACHAR
3637    
3638    Returns:        the number of < 256 characters added
3639                    the pointer to extra data is updated
3640    */
3641    
3642    static int
3643    add_not_list_to_class(pcre_uint8 *classbits, pcre_uchar **uchardptr,
3644      int options, compile_data *cd, const pcre_uint32 *p)
3645    {
3646    BOOL utf = (options & PCRE_UTF8) != 0;
3647    int n8 = 0;
3648    if (p[0] > 0)
3649      n8 += add_to_class(classbits, uchardptr, options, cd, 0, p[0] - 1);
3650    while (p[0] < NOTACHAR)
3651      {
3652      while (p[1] == p[0] + 1) p++;
3653      n8 += add_to_class(classbits, uchardptr, options, cd, p[0] + 1,
3654        (p[1] == NOTACHAR) ? (utf ? 0x10ffffu : 0xffffffffu) : p[1] - 1);
3655      p++;
3656      }
3657    return n8;
3658    }
3659    
3660    
3661    
3662    /*************************************************
3663  *           Compile one branch                   *  *           Compile one branch                   *
3664  *************************************************/  *************************************************/
3665    
# Line 3248  Arguments: Line 3674  Arguments:
3674    codeptr        points to the pointer to the current code point    codeptr        points to the pointer to the current code point
3675    ptrptr         points to the current pattern pointer    ptrptr         points to the current pattern pointer
3676    errorcodeptr   points to error code variable    errorcodeptr   points to error code variable
3677    firstbyteptr   set to initial literal character, or < 0 (REQ_UNSET, REQ_NONE)    firstcharptr    place to put the first required character
3678    reqbyteptr     set to the last literal character required, else < 0    firstcharflagsptr place to put the first character flags, or a negative number
3679      reqcharptr     place to put the last required character
3680      reqcharflagsptr place to put the last required character flags, or a negative number
3681    bcptr          points to current branch chain    bcptr          points to current branch chain
3682    cond_depth     conditional nesting depth    cond_depth     conditional nesting depth
3683    cd             contains pointers to tables etc.    cd             contains pointers to tables etc.
# Line 3261  Returns:         TRUE on success Line 3689  Returns:         TRUE on success
3689  */  */
3690    
3691  static BOOL  static BOOL
3692  compile_branch(int *optionsptr, uschar **codeptr, const uschar **ptrptr,  compile_branch(int *optionsptr, pcre_uchar **codeptr,
3693    int *errorcodeptr, int *firstbyteptr, int *reqbyteptr, branch_chain *bcptr,    const pcre_uchar **ptrptr, int *errorcodeptr,
3694    int cond_depth, compile_data *cd, int *lengthptr)    pcre_uint32 *firstcharptr, pcre_int32 *firstcharflagsptr,
3695      pcre_uint32 *reqcharptr, pcre_int32 *reqcharflagsptr,
3696      branch_chain *bcptr, int cond_depth,
3697      compile_data *cd, int *lengthptr)
3698  {  {
3699  int repeat_type, op_type;  int repeat_type, op_type;
3700  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */  int repeat_min = 0, repeat_max = 0;      /* To please picky compilers */
3701  int bravalue = 0;  int bravalue = 0;
3702  int greedy_default, greedy_non_default;  int greedy_default, greedy_non_default;
3703  int firstbyte, reqbyte;  pcre_uint32 firstchar, reqchar;
3704  int zeroreqbyte, zerofirstbyte;  pcre_int32 firstcharflags, reqcharflags;
3705  int req_caseopt, reqvary, tempreqvary;  pcre_uint32 zeroreqchar, zerofirstchar;
3706    pcre_int32 zeroreqcharflags, zerofirstcharflags;
3707    pcre_int32 req_caseopt, reqvary, tempreqvary;
3708  int options = *optionsptr;               /* May change dynamically */  int options = *optionsptr;               /* May change dynamically */
3709  int after_manual_callout = 0;  int after_manual_callout = 0;
3710  int length_prevgroup = 0;  int length_prevgroup = 0;
3711  register int c;  register pcre_uint32 c;
3712  register uschar *code = *codeptr;  int escape;
3713  uschar *last_code = code;  register pcre_uchar *code = *codeptr;
3714  uschar *orig_code = code;  pcre_uchar *last_code = code;
3715  uschar *tempcode;  pcre_uchar *orig_code = code;
3716    pcre_uchar *tempcode;
3717  BOOL inescq = FALSE;  BOOL inescq = FALSE;
3718  BOOL groupsetfirstbyte = FALSE;  BOOL groupsetfirstchar = FALSE;
3719  const uschar *ptr = *ptrptr;  const pcre_uchar *ptr = *ptrptr;
3720  const uschar *tempptr;  const pcre_uchar *tempptr;
3721  const uschar *nestptr = NULL;  const pcre_uchar *nestptr = NULL;
3722  uschar *previous = NULL;  pcre_uchar *previous = NULL;
3723  uschar *previous_callout = NULL;  pcre_uchar *previous_callout = NULL;
3724  uschar *save_hwm = NULL;  pcre_uchar *save_hwm = NULL;
3725  uschar classbits[32];  pcre_uint8 classbits[32];
3726    
3727  /* 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
3728  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
3729  dynamically as we process the pattern. */  dynamically as we process the pattern. */
3730    
3731  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3732  BOOL class_utf8;  /* PCRE_UTF[16|32] have the same value as PCRE_UTF8. */
3733  BOOL utf8 = (options & PCRE_UTF8) != 0;  BOOL utf = (options & PCRE_UTF8) != 0;
3734  uschar *class_utf8data;  #ifndef COMPILE_PCRE32
3735  uschar *class_utf8data_base;  pcre_uchar utf_chars[6];
3736  uschar utf8_char[6];  #endif
3737  #else  #else
3738  BOOL utf8 = FALSE;  BOOL utf = FALSE;
3739    #endif
3740    
3741    /* Helper variables for OP_XCLASS opcode (for characters > 255). We define
3742    class_uchardata always so that it can be passed to add_to_class() always,
3743    though it will not be used in non-UTF 8-bit cases. This avoids having to supply
3744    alternative calls for the different cases. */
3745    
3746    pcre_uchar *class_uchardata;
3747    #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
3748    BOOL xclass;
3749    pcre_uchar *class_uchardata_base;
3750  #endif  #endif
3751    
3752  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
# Line 3315  greedy_non_default = greedy_default ^ 1; Line 3760  greedy_non_default = greedy_default ^ 1;
3760    
3761  /* Initialize no first byte, no required byte. REQ_UNSET means "no char  /* Initialize no first byte, no required byte. REQ_UNSET means "no char
3762  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
3763  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
3764  find one.  find one.
3765    
3766  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
3767  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
3768  zerofirstbyte and zeroreqbyte when such a repeat is encountered. The individual  zerofirstbyte and zeroreqchar when such a repeat is encountered. The individual
3769  item types that can be repeated set these backoff variables appropriately. */  item types that can be repeated set these backoff variables appropriately. */
3770    
3771  firstbyte = reqbyte = zerofirstbyte = zeroreqbyte = REQ_UNSET;  firstchar = reqchar = zerofirstchar = zeroreqchar = 0;
3772    firstcharflags = reqcharflags = zerofirstcharflags = zeroreqcharflags = REQ_UNSET;
3773    
3774  /* The variable req_caseopt contains either the REQ_CASELESS value or zero,  /* The variable req_caseopt contains either the REQ_CASELESS value
3775  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
3776  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
3777  case status of the value. This is used only for ASCII characters. */  firstchar or reqchar variables to record the case status of the
3778    value. This is used only for ASCII characters. */
3779    
3780  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS : 0;  req_caseopt = ((options & PCRE_CASELESS) != 0)? REQ_CASELESS:0;
3781    
3782  /* Switch on next character until the end of the branch */  /* Switch on next character until the end of the branch */
3783    
# Line 3342  for (;; ptr++) Line 3789  for (;; ptr++)
3789    BOOL is_quantifier;    BOOL is_quantifier;
3790    BOOL is_recurse;    BOOL is_recurse;
3791    BOOL reset_bracount;    BOOL reset_bracount;
3792    int class_charcount;    int class_has_8bitchar;
3793    int class_lastchar;    int class_one_char;
3794    int newoptions;    int newoptions;
3795    int recno;    int recno;
3796    int refsign;    int refsign;
3797    int skipbytes;    int skipbytes;
3798    int subreqbyte;    pcre_uint32 subreqchar, subfirstchar;
3799    int subfirstbyte;    pcre_int32 subreqcharflags, subfirstcharflags;
3800    int terminator;    int terminator;
3801    int mclength;    unsigned int mclength;
3802    int tempbracount;    unsigned int tempbracount;
3803    uschar mcbuffer[8];    pcre_uint32 ec;
3804      pcre_uchar mcbuffer[8];
3805    
3806    /* Get next byte in the pattern */    /* Get next character in the pattern */
3807    
3808    c = *ptr;    c = *ptr;
3809    
3810    /* If we are at the end of a nested substitution, revert to the outer level    /* If we are at the end of a nested substitution, revert to the outer level
3811    string. Nesting only happens one level deep. */    string. Nesting only happens one level deep. */
3812    
3813    if (c == 0 && nestptr != NULL)    if (c == CHAR_NULL && nestptr != NULL)
3814      {      {
3815      ptr = nestptr;      ptr = nestptr;
3816      nestptr = NULL;      nestptr = NULL;
# Line 3377  for (;; ptr++) Line 3825  for (;; ptr++)
3825  #ifdef PCRE_DEBUG  #ifdef PCRE_DEBUG
3826      if (code > cd->hwm) cd->hwm = code;                 /* High water info */      if (code > cd->hwm) cd->hwm = code;                 /* High water info */
3827  #endif  #endif
3828      if (code > cd->start_workspace + cd->workspace_size -      if (code > cd->start_workspace + cd->workspace_size -
3829          WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */          WORK_SIZE_SAFETY_MARGIN)                       /* Check for overrun */
3830        {        {
3831        *errorcodeptr = ERR52;        *errorcodeptr = ERR52;
# Line 3401  for (;; ptr++) Line 3849  for (;; ptr++)
3849        }        }
3850    
3851      *lengthptr += (int)(code - last_code);      *lengthptr += (int)(code - last_code);
3852      DPRINTF(("length=%d added %d c=%c\n", *lengthptr, (int)(code - last_code),      DPRINTF(("length=%d added %d c=%c (0x%x)\n", *lengthptr,
3853        c));        (int)(code - last_code), c, c));
3854    
3855      /* 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
3856      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 3412  for (;; ptr++) Line 3860  for (;; ptr++)
3860        {        {
3861        if (previous > orig_code)        if (previous > orig_code)
3862          {          {
3863          memmove(orig_code, previous, code - previous);          memmove(orig_code, previous, IN_UCHARS(code - previous));
3864          code -= previous - orig_code;          code -= previous - orig_code;
3865          previous = orig_code;          previous = orig_code;
3866          }          }
# Line 3428  for (;; ptr++) Line 3876  for (;; ptr++)
3876    /* 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
3877    reference list. */    reference list. */
3878    
3879    else if (cd->hwm > cd->start_workspace + cd->workspace_size -    else if (cd->hwm > cd->start_workspace + cd->workspace_size -
3880             WORK_SIZE_SAFETY_MARGIN)             WORK_SIZE_SAFETY_MARGIN)
3881      {      {
3882      *errorcodeptr = ERR52;      *errorcodeptr = ERR52;
# Line 3437  for (;; ptr++) Line 3885  for (;; ptr++)
3885    
3886    /* If in \Q...\E, check for the end; if not, we have a literal */    /* If in \Q...\E, check for the end; if not, we have a literal */
3887    
3888    if (inescq && c != 0)    if (inescq && c != CHAR_NULL)
3889      {      {
3890      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)      if (c == CHAR_BACKSLASH && ptr[1] == CHAR_E)
3891        {        {
# Line 3481  for (;; ptr++) Line 3929  for (;; ptr++)
3929    
3930    if ((options & PCRE_EXTENDED) != 0)    if ((options & PCRE_EXTENDED) != 0)
3931      {      {
3932      if ((cd->ctypes[c] & ctype_space) != 0) continue;      if (MAX_255(*ptr) && (cd->ctypes[c] & ctype_space) != 0) continue;
3933      if (c == CHAR_NUMBER_SIGN)      if (c == CHAR_NUMBER_SIGN)
3934        {        {
3935        ptr++;        ptr++;
3936        while (*ptr != 0)        while (*ptr != CHAR_NULL)
3937          {          {
3938          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }          if (IS_NEWLINE(ptr)) { ptr += cd->nllen - 1; break; }
3939          ptr++;          ptr++;
3940  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
3941          if (utf8) while ((*ptr & 0xc0) == 0x80) ptr++;          if (utf) FORWARDCHAR(ptr);
3942  #endif  #endif
3943          }          }
3944        if (*ptr != 0) continue;        if (*ptr != CHAR_NULL) continue;
3945    
3946        /* Else fall through to handle end of string */        /* Else fall through to handle end of string */
3947        c = 0;        c = 0;
# Line 3514  for (;; ptr++) Line 3962  for (;; ptr++)
3962      case 0:                        /* The branch terminates at string end */      case 0:                        /* The branch terminates at string end */
3963      case CHAR_VERTICAL_LINE:       /* or | or ) */      case CHAR_VERTICAL_LINE:       /* or | or ) */
3964      case CHAR_RIGHT_PARENTHESIS:      case CHAR_RIGHT_PARENTHESIS:
3965      *firstbyteptr = firstbyte;      *firstcharptr = firstchar;
3966      *reqbyteptr = reqbyte;      *firstcharflagsptr = firstcharflags;
3967        *reqcharptr = reqchar;
3968        *reqcharflagsptr = reqcharflags;
3969      *codeptr = code;      *codeptr = code;
3970      *ptrptr = ptr;      *ptrptr = ptr;
3971      if (lengthptr != NULL)      if (lengthptr != NULL)
# Line 3539  for (;; ptr++) Line 3989  for (;; ptr++)
3989      previous = NULL;      previous = NULL;
3990      if ((options & PCRE_MULTILINE) != 0)      if ((options & PCRE_MULTILINE) != 0)
3991        {        {
3992        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
3993        *code++ = OP_CIRCM;        *code++ = OP_CIRCM;
3994        }        }
3995      else *code++ = OP_CIRC;      else *code++ = OP_CIRC;
# Line 3551  for (;; ptr++) Line 4001  for (;; ptr++)
4001      break;      break;
4002    
4003      /* 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
4004      repeats. The value of reqbyte doesn't change either. */      repeats. The value of reqchar doesn't change either. */
4005    
4006      case CHAR_DOT:      case CHAR_DOT:
4007      if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;      if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4008      zerofirstbyte = firstbyte;      zerofirstchar = firstchar;
4009      zeroreqbyte = reqbyte;      zerofirstcharflags = firstcharflags;
4010        zeroreqchar = reqchar;
4011        zeroreqcharflags = reqcharflags;
4012      previous = code;      previous = code;
4013      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;      *code++ = ((options & PCRE_DOTALL) != 0)? OP_ALLANY: OP_ANY;
4014      break;      break;
# Line 3611  for (;; ptr++) Line 4063  for (;; ptr++)
4063          {          {
4064          if (ptr[1] == CHAR_E)          if (ptr[1] == CHAR_E)
4065            ptr++;            ptr++;
4066          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)  
4067            ptr += 3;            ptr += 3;
4068          else          else
4069            break;            break;
# Line 3631  for (;; ptr++) Line 4082  for (;; ptr++)
4082          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)          (cd->external_options & PCRE_JAVASCRIPT_COMPAT) != 0)
4083        {        {
4084        *code++ = negate_class? OP_ALLANY : OP_FAIL;        *code++ = negate_class? OP_ALLANY : OP_FAIL;
4085        if (firstbyte == REQ_UNSET) firstbyte = REQ_NONE;        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4086        zerofirstbyte = firstbyte;        zerofirstchar = firstchar;
4087          zerofirstcharflags = firstcharflags;
4088        break;        break;
4089        }        }
4090    
# Line 3642  for (;; ptr++) Line 4094  for (;; ptr++)
4094    
4095      should_flip_negation = FALSE;      should_flip_negation = FALSE;
4096    
4097      /* 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:
4098      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 <
4099      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
4100        character. */
4101    
4102      class_charcount = 0;      class_has_8bitchar = 0;
4103      class_lastchar = -1;      class_one_char = 0;
4104    
4105      /* 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
4106      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
4107      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
4108      */      map. */
4109    
4110      memset(classbits, 0, 32 * sizeof(uschar));      memset(classbits, 0, 32 * sizeof(pcre_uint8));
4111    
4112  #ifdef SUPPORT_UTF8  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4113      class_utf8 = FALSE;                       /* No chars >= 256 */      xclass = FALSE;
4114      class_utf8data = code + LINK_SIZE + 2;    /* For UTF-8 items */      class_uchardata = code + LINK_SIZE + 2;   /* For XCLASS items */
4115      class_utf8data_base = class_utf8data;     /* For resetting in pass 1 */      class_uchardata_base = class_uchardata;   /* Save the start */
4116  #endif  #endif
4117    
4118      /* Process characters until ] is reached. By writing this as a "do" it      /* Process characters until ] is reached. By writing this as a "do" it
4119      means that an initial ] is taken as a data character. At the start of the      means that an initial ] is taken as a data character. At the start of the
4120      loop, c contains the first byte of the character. */      loop, c contains the first byte of the character. */
4121    
4122      if (c != 0) do      if (c != CHAR_NULL) do
4123        {        {
4124        const uschar *oldptr;        const pcre_uchar *oldptr;
4125    
4126  #ifdef SUPPORT_UTF8  #ifdef SUPPORT_UTF
4127        if (utf8 && c > 127)        if (utf && HAS_EXTRALEN(c))
4128          {                           /* Braces are required because the */          {                           /* Braces are required because the */
4129          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */          GETCHARLEN(c, ptr, ptr);    /* macro generates multiple statements */
4130          }          }
4131    #endif
4132    
4133        /* In the pre-compile phase, accumulate the length of any UTF-8 extra  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4134          /* In the pre-compile phase, accumulate the length of any extra
4135        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
4136        contain a zillion UTF-8 characters no longer overwrite the work space        contain a zillion > 255 characters no longer overwrite the work space
4137        (which is on the stack). */        (which is on the stack). We have to remember that there was XCLASS data,
4138          however. */
4139        if (lengthptr != NULL)  
4140          {        if (lengthptr != NULL && class_uchardata > class_uchardata_base)
4141          *lengthptr += class_utf8data - class_utf8data_base;          {
4142          class_utf8data = class_utf8data_base;          xclass = TRUE;
4143            *lengthptr += class_uchardata - class_uchardata_base;
4144            class_uchardata = class_uchardata_base;
4145          }          }
   
4146  #endif  #endif
4147    
4148        /* Inside \Q...\E everything is literal except \E */        /* Inside \Q...\E everything is literal except \E */
# Line 3714  for (;; ptr++) Line 4170  for (;; ptr++)
4170          {          {
4171          BOOL local_negate = FALSE;          BOOL local_negate = FALSE;
4172          int posix_class, taboffset, tabopt;          int posix_class, taboffset, tabopt;
4173          register const uschar *cbits = cd->cbits;          register const pcre_uint8 *cbits = cd->cbits;
4174          uschar pbits[32];          pcre_uint8 pbits[32];
4175    
4176          if (ptr[1] != CHAR_COLON)          if (ptr[1] != CHAR_COLON)
4177            {            {
# Line 3770  for (;; ptr++) Line 4226  for (;; ptr++)
4226          /* Copy in the first table (always present) */          /* Copy in the first table (always present) */
4227    
4228          memcpy(pbits, cbits + posix_class_maps[posix_class],          memcpy(pbits, cbits + posix_class_maps[posix_class],
4229            32 * sizeof(uschar));            32 * sizeof(pcre_uint8));
4230    
4231          /* If there is a second table, add or remove it as required. */          /* If there is a second table, add or remove it as required. */
4232    
# Line 3785  for (;; ptr++) Line 4241  for (;; ptr++)
4241              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];              for (c = 0; c < 32; c++) pbits[c] &= ~cbits[c + taboffset];
4242            }            }
4243    
4244          /* Not see if we need to remove any special characters. An option          /* Now see if we need to remove any special characters. An option
4245          value of 1 removes vertical space and 2 removes underscore. */          value of 1 removes vertical space and 2 removes underscore. */
4246    
4247          if (tabopt < 0) tabopt = -tabopt;          if (tabopt < 0) tabopt = -tabopt;
# Line 3801  for (;; ptr++) Line 4257  for (;; ptr++)
4257            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];            for (c = 0; c < 32; c++) classbits[c] |= pbits[c];
4258    
4259          ptr = tempptr + 1;          ptr = tempptr + 1;
4260          class_charcount = 10;  /* Set > 1; assumes more than 1 per class */          /* Every class contains at least one < 256 character. */
4261            class_has_8bitchar = 1;
4262            /* Every class contains at least two characters. */
4263            class_one_char = 2;
4264          continue;    /* End of POSIX syntax handling */          continue;    /* End of POSIX syntax handling */
4265          }          }
4266    
4267        /* Backslash may introduce a single character, or it may introduce one        /* Backslash may introduce a single character, or it may introduce one
4268        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
4269        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
4270        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
4271        class_charcount bigger than one. Unrecognized escapes fall through and        speculatively set both class_has_8bitchar and class_one_char bigger
4272        are either treated as literal characters (by default), or are faulted if        than one. Unrecognized escapes fall through and are either treated
4273          as literal characters (by default), or are faulted if
4274        PCRE_EXTRA is set. */        PCRE_EXTRA is set. */
4275    
4276        if (c == CHAR_BACKSLASH)        if (c == CHAR_BACKSLASH)
4277          {          {
4278          c = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);          escape = check_escape(&ptr, &ec, errorcodeptr, cd->bracount, options,
4279              TRUE);
4280          if (*errorcodeptr != 0) goto FAILED;          if (*errorcodeptr != 0) goto FAILED;
4281            if (escape == 0) c = ec;
4282          if (-c == ESC_b) c = CHAR_BS;    /* \b is backspace in a class */          else if (escape == ESC_b) c = CHAR_BS; /* \b is backspace in a class */
4283          else if (-c == ESC_N)            /* \N is not supported in a class */          else if (escape == ESC_N)          /* \N is not supported in a class */
4284            {            {
4285            *errorcodeptr = ERR71;            *errorcodeptr = ERR71;
4286            goto FAILED;            goto FAILED;
4287            }            }
4288          else if (-c == ESC_Q)            /* Handle start of quoted string */          else if (escape == ESC_Q)            /* Handle start of quoted string */
4289            {            {
4290            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)            if (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
4291              {              {
# Line 3833  for (;; ptr++) Line 4294  for (;; ptr++)
4294            else inescq = TRUE;            else inescq = TRUE;
4295            continue;            continue;
4296            }            }
4297          else if (-c == ESC_E) continue;  /* Ignore orphan \E */          else if (escape == ESC_E) continue;  /* Ignore orphan \E */
4298    
4299          if (c < 0)          else
4300            {            {
4301            register const uschar *cbits = cd->cbits;            register const pcre_uint8 *cbits = cd->cbits;
4302            class_charcount += 2;     /* Greater than 1 is what matters */            /* Every class contains at least two < 256 characters. */
4303              class_has_8bitchar++;
4304              /* Every class contains at least two characters. */
4305              class_one_char += 2;
4306    
4307            switch (-c)            switch (escape)
4308              {              {
4309  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
4310              case ESC_du:     /* These are the values given for \d etc */              case ESC_du:     /* These are the values given for \d etc */
# Line 3850  for (;; ptr++) Line 4314  for (;; ptr++)
4314              case ESC_su:     /* of the default ASCII testing. */              case ESC_su:     /* of the default ASCII testing. */
4315              case ESC_SU:              case ESC_SU:
4316              nestptr = ptr;              nestptr = ptr;
4317              ptr = substitutes[-c - ESC_DU] - 1;  /* Just before substitute */              ptr = substitutes[escape - ESC_DU] - 1;  /* Just before substitute */
4318              class_charcount -= 2;                /* Undo! */              class_has_8bitchar--;                /* Undo! */
4319              continue;              continue;
4320  #endif  #endif
4321              case ESC_d:              case ESC_d:
# Line 3874  for (;; ptr++) Line 4338  for (;; ptr++)
4338    
4339              /* 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
4340              if it was previously set by something earlier in the character              if it was previously set by something earlier in the character
4341              class. */              class. Luckily, the value of CHAR_VT is 0x0b in both ASCII and
4342                EBCDIC, so we lazily just adjust the appropriate bit. */
4343    
4344              case ESC_s:              case ESC_s:
4345              classbits[0] |= cbits[cbit_space];              classbits[0] |= cbits[cbit_space];
# Line 3888  for (;; ptr++) Line 4353  for (;; ptr++)
4353              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */              classbits[1] |= 0x08;    /* Perl 5.004 onwards omits VT from \s */
4354              continue;              continue;
4355    
4356                /* The rest apply in both UCP and non-UCP cases. */
4357    
4358              case ESC_h:              case ESC_h:
4359              SETBIT(classbits, 0x09); /* VT */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4360              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  
4361              continue;              continue;
4362    
4363              case ESC_H:              case ESC_H:
4364              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4365                {                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  
4366              continue;              continue;
4367    
4368              case ESC_v:              case ESC_v:
4369              SETBIT(classbits, 0x0a); /* LF */              (void)add_list_to_class(classbits, &class_uchardata, options, cd,
4370              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  
4371              continue;              continue;
4372    
4373              case ESC_V:              case ESC_V:
4374              for (c = 0; c < 32; c++)              (void)add_not_list_to_class(classbits, &class_uchardata, options,
4375                {                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  
4376              continue;              continue;
4377    
4378  #ifdef SUPPORT_UCP  #ifdef SUPPORT_UCP
# Line 4009  for (;; ptr++) Line 4380  for (;; ptr++)
4380              case ESC_P:              case ESC_P:
4381                {                {
4382                BOOL negated;                BOOL negated;
4383                int pdata;                unsigned int ptype = 0, pdata = 0;
4384                int ptype = get_ucp(&ptr, &negated, &pdata, errorcodeptr);                if (!get_ucp(&ptr, &negated, &ptype, &pdata, errorcodeptr))
4385                if (ptype < 0) goto FAILED;                  goto FAILED;
4386                class_utf8 = TRUE;                *class_uchardata++ = ((escape == ESC_p) != negated)?
               *class_utf8data++ = ((-c == ESC_p) != negated)?  
4387                  XCL_PROP : XCL_NOTPROP;                  XCL_PROP : XCL_NOTPROP;
4388                *class_utf8data++ = ptype;                *class_uchardata++ = ptype;
4389                *class_utf8data++ = pdata;                *class_uchardata++ = pdata;
4390                class_charcount -= 2;   /* Not a < 256 character */                class_has_8bitchar--;                /* Undo! */
4391                continue;                continue;
4392                }                }
4393  #endif  #endif
# Line 4031  for (;; ptr++) Line 4401  for (;; ptr++)
4401                *errorcodeptr = ERR7;                *errorcodeptr = ERR7;
4402                goto FAILED;                goto FAILED;
4403                }                }
4404              class_charcount -= 2;  /* Undo the default count from above */              class_has_8bitchar--;    /* Undo the speculative increase. */
4405              c = *ptr;              /* Get the final character and fall through */              class_one_char -= 2;     /* Undo the speculative increase. */
4406                c = *ptr;                /* Get the final character and fall through */
4407              break;              break;
4408              }              }
4409            }            }
4410    
4411          /* 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).
4412          greater than 256 in UTF-8 mode. */          This may be greater than 256. */
4413    
4414            escape = 0;
4415    
4416          }   /* End of backslash handling */          }   /* End of backslash handling */
4417    
4418        /* 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
4419        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
4420        at the end is treated as a literal. Perl ignores orphaned \E sequences        treated as a literal. Perl ignores orphaned \E sequences entirely. The
4421        entirely. The code for handling \Q and \E is messy. */        code for handling \Q and \E is messy. */
4422    
4423        CHECK_RANGE:        CHECK_RANGE:
4424        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)        while (ptr[1] == CHAR_BACKSLASH && ptr[2] == CHAR_E)
# Line 4053  for (;; ptr++) Line 4426  for (;; ptr++)
4426          inescq = FALSE;          inescq = FALSE;
4427          ptr += 2;          ptr += 2;
4428          }          }
   
4429        oldptr = ptr;        oldptr = ptr;
4430    
4431        /* Remember \r or \n */        /* Remember if \r or \n were explicitly used */
4432    
4433        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;        if (c == CHAR_CR || c == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4434    
# Line 4064  for (;; ptr++) Line 4436  for (;; ptr++)
4436    
4437        if (!inescq && ptr[1] == CHAR_MINUS)        if (!inescq && ptr[1] == CHAR_MINUS)
4438          {          {
4439          int d;          pcre_uint32 d;
4440          ptr += 2;          ptr += 2;
4441          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;          while (*ptr == CHAR_BACKSLASH && ptr[1] == CHAR_E) ptr += 2;
4442    
# Line 4080  for (;; ptr++) Line 4452  for (;; ptr++)
4452            break;            break;
4453            }            }
4454    
4455          if (*ptr == 0 || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))          /* Minus (hyphen) at the end of a class is treated as a literal, so put
4456            back the pointer and jump to handle the character that preceded it. */
4457    
4458            if (*ptr == CHAR_NULL || (!inescq && *ptr == CHAR_RIGHT_SQUARE_BRACKET))
4459            {            {
4460            ptr = oldptr;            ptr = oldptr;
4461            goto LONE_SINGLE_CHARACTER;            goto CLASS_SINGLE_CHARACTER;
4462            }            }
4463    
4464  #ifdef SUPPORT_UTF8          /* Otherwise, we have a potential range; pick up the next character */
4465          if (utf8)  
4466    #ifdef SUPPORT_UTF
4467            if (utf)
4468            {                           /* Braces are required because the */            {                           /* Braces are required because the */
4469            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */            GETCHARLEN(d, ptr, ptr);    /* macro generates multiple statements */
4470            }            }
# Line 4101  for (;; ptr++) Line 4478  for (;; ptr++)
4478    
4479          if (!inescq && d == CHAR_BACKSLASH)          if (!inescq && d == CHAR_BACKSLASH)
4480            {            {
4481            d = check_escape(&ptr, errorcodeptr, cd->bracount, options, TRUE);            int descape;
4482              descape = check_escape(&ptr, &d, errorcodeptr, cd->bracount, options, TRUE);
4483            if (*errorcodeptr != 0) goto FAILED;            if (*errorcodeptr != 0) goto FAILED;
4484    
4485            /* \b is backspace; any other special means the '-' was literal */            /* \b is backspace; any other special means the '-' was literal. */
4486    
4487            if (d < 0)            if (descape != 0)
4488              {              {
4489              if (d == -ESC_b) d = CHAR_BS; else              if (descape == ESC_b) d = CHAR_BS; else
4490                {                {
4491                ptr = oldptr;                ptr = oldptr;
4492                goto LONE_SINGLE_CHARACTER;  /* A few lines below */                goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
               }  
             }  
           }  
   
         /* Check that the two values are in the correct order. Optimize  
         one-character ranges */  
   
         if (d < c)  
           {  
           *errorcodeptr = ERR8;  
           goto FAILED;  
           }  
   
         if (d == c) goto LONE_SINGLE_CHARACTER;  /* A few lines below */  
   
         /* Remember \r or \n */  
   
         if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;  
   
         /* In UTF-8 mode, if the upper limit is > 255, or > 127 for caseless  
         matching, we have to use an XCLASS with extra data items. Caseless  
         matching for characters > 127 is available only if UCP support is  
         available. */  
   
 #ifdef SUPPORT_UTF8  
         if (utf8 && (d > 255 || ((options & PCRE_CASELESS) != 0 && d > 127)))  
           {  
           class_utf8 = TRUE;  
   
           /* With UCP support, we can find the other case equivalents of  
           the relevant characters. There may be several ranges. Optimize how  
           they fit with the basic range. */  
   
 #ifdef SUPPORT_UCP  
           if ((options & PCRE_CASELESS) != 0)  
             {  
             unsigned int occ, ocd;  
             unsigned int cc = c;  
             unsigned int origd = d;  
             while (get_othercase_range(&cc, origd, &occ, &ocd))  
               {  
               if (occ >= (unsigned int)c &&  
                   ocd <= (unsigned int)d)  
                 continue;                          /* Skip embedded ranges */  
   
               if (occ < (unsigned int)c  &&  
                   ocd >= (unsigned int)c - 1)      /* Extend the basic range */  
                 {                                  /* if there is overlap,   */  
                 c = occ;                           /* noting that if occ < c */  
                 continue;                          /* we can't have ocd > d  */  
                 }                                  /* because a subrange is  */  
               if (ocd > (unsigned int)d &&  
                   occ <= (unsigned int)d + 1)      /* always shorter than    */  
                 {                                  /* the basic range.       */  
                 d = ocd;  
                 continue;  
                 }  
   
               if (occ == ocd)  
                 {  
                 *class_utf8data++ = XCL_SINGLE;  
                 }  
               else  
                 {  
                 *class_utf8data++ = XCL_RANGE;  
                 class_utf8data += _pcre_ord2utf8(occ, class_utf8data);  
                 }  
               class_utf8data += _pcre_ord2utf8(ocd, class_utf8data);  
4493                }                }
4494              }              }
4495  #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 */  
4496    
4497            d = 127;          /* Check that the two values are in the correct order. Optimize
4498            one-character ranges. */
4499    
4500  #endif  /* SUPPORT_UCP */          if (d < c)
4501              {
4502              *errorcodeptr = ERR8;
4503              goto FAILED;
4504            }            }
4505  #endif  /* SUPPORT_UTF8 */          if (d == c) goto CLASS_SINGLE_CHARACTER;  /* A few lines below */
4506    
4507          /* We use the bit map for all cases when not in UTF-8 mode; else          /* We have found a character range, so single character optimizations
4508          ranges that lie entirely within 0-127 when there is UCP support; else          cannot be done anymore. Any value greater than 1 indicates that there
4509          for partial ranges without UCP support. */          is more than one character. */
4510    
4511          class_charcount += d - c + 1;          class_one_char = 2;
         class_lastchar = d;  
4512    
4513          /* We can save a bit of time by skipping this in the pre-compile. */          /* Remember an explicit \r or \n, and add the range to the class. */
4514    
4515          if (lengthptr == NULL) for (; c <= d; c++)          if (d == CHAR_CR || d == CHAR_NL) cd->external_flags |= PCRE_HASCRORLF;
4516            {  
4517            classbits[c/8] |= (1 << (c&7));          class_has_8bitchar +=
4518            if ((options & PCRE_CASELESS) != 0)            add_to_class(classbits, &class_uchardata, options, cd, c, d);
             {  
             int uc = cd->fcc[c];           /* flip case */  
             classbits[uc/8] |= (1 << (uc&7));  
             }  
           }  
4519    
4520          continue;   /* Go get the next char in the class */          continue;   /* Go get the next char in the class */
4521          }          }
4522    
4523        /* 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
4524        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
4525        apparent range that isn't. */        range that isn't. Only the value 1 matters for class_one_char, so don't
4526          increase it if it is already 2 or more ... just in case there's a class
4527        LONE_SINGLE_CHARACTER:        with a zillion characters in it. */
4528    
4529        /* Handle a character that cannot go in the bit map */        CLASS_SINGLE_CHARACTER:
4530          if (class_one_char < 2) class_one_char++;
4531    
4532          /* If class_one_char is 1, we have the first single character in the
4533          class, and there have been no prior ranges, or XCLASS items generated by
4534          escapes. If this is the final character in the class, we can optimize by
4535          turning the item into a 1-character OP_CHAR[I] if it's positive, or
4536          OP_NOT[I] if it's negative. In the positive case, it can cause firstchar
4537          to be set. Otherwise, there can be no first char if this item is first,
4538          whatever repeat count may follow. In the case of reqchar, save the
4539          previous value for reinstating. */
4540    
4541  #ifdef SUPPORT_UTF8        if (class_one_char == 1 && ptr[1] == CHAR_RIGHT_SQUARE_BRACKET)
       if (utf8 && (c > 255 || ((options & PCRE_CASELESS) != 0 && c > 127)))  
4542          {          {
4543          class_utf8 = TRUE;          ptr++;
4544          *class_utf8data++ = XCL_SINGLE;          zeroreqchar = reqchar;
4545          class_utf8data += _pcre_ord2utf8(c, class_utf8data);          zeroreqcharflags = reqcharflags;
4546    
4547  #ifdef SUPPORT_UCP          if (negate_class)
         if ((options & PCRE_CASELESS) != 0)  
4548            {            {
4549            unsigned int othercase;  #ifdef SUPPORT_UCP
4550            if ((othercase = UCD_OTHERCASE(c)) != c)            int d;
4551    #endif
4552              if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4553              zerofirstchar = firstchar;
4554              zerofirstcharflags = firstcharflags;
4555    
4556              /* For caseless UTF-8 mode when UCP support is available, check
4557              whether this character has more than one other case. If so, generate
4558              a special OP_NOTPROP item instead of OP_NOTI. */
4559    
4560    #ifdef SUPPORT_UCP
4561              if (utf && (options & PCRE_CASELESS) != 0 &&
4562                  (d = UCD_CASESET(c)) != 0)
4563                {
4564                *code++ = OP_NOTPROP;
4565                *code++ = PT_CLIST;
4566                *code++ = d;
4567                }
4568              else
4569    #endif
4570              /* Char has only one other case, or UCP not available */
4571    
4572              {              {
4573              *class_utf8data++ = XCL_SINGLE;              *code++ = ((options & PCRE_CASELESS) != 0)? OP_NOTI: OP_NOT;
4574              class_utf8data += _pcre_ord2utf8(othercase, class_utf8data);  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4575                if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4576                  code += PRIV(ord2utf)(c, code);
4577                else
4578    #endif
4579                  *code++ = c;
4580              }              }
4581    
4582              /* We are finished with this character class */
4583    
4584              goto END_CLASS;
4585            }            }
 #endif  /* SUPPORT_UCP */  
4586    
4587          }          /* For a single, positive character, get the value into mcbuffer, and
4588        else          then we can handle this with the normal one-character code. */
 #endif  /* SUPPORT_UTF8 */  
4589    
4590        /* Handle a single-byte character */  #if defined SUPPORT_UTF && !defined COMPILE_PCRE32
4591          {          if (utf && c > MAX_VALUE_FOR_SINGLE_CHAR)
4592          classbits[c/8] |= (1 << (c&7));            mclength = PRIV(ord2utf)(c, mcbuffer);
4593          if ((options & PCRE_CASELESS) != 0)          else
4594    #endif
4595            {            {
4596            c = cd->fcc[c];   /* flip case */            mcbuffer[0] = c;
4597            classbits[c/8] |= (1 << (c&7));            mclength = 1;
4598            }            }
4599          class_charcount++;          goto ONE_CHAR;
4600          class_lastchar = c;          }       /* End of 1-char optimization */
4601          }  
4602          /* There is more than one character in the class, or an XCLASS item
4603          has been generated. Add this character to the class. */
4604    
4605          class_has_8bitchar +=
4606            add_to_class(classbits, &class_uchardata, options, cd, c, c);
4607        }        }
4608    
4609      /* 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.
4610      If we are at the end of an internal nested string, revert to the outer      If we are at the end of an internal nested string, revert to the outer
4611      string. */      string. */
4612    
4613      while (((c = *(++ptr)) != 0 ||      while (((c = *(++ptr)) != CHAR_NULL ||
4614             (nestptr != NULL &&             (nestptr != NULL &&
4615               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != 0)) &&               (ptr = nestptr, nestptr = NULL, c = *(++ptr)) != CHAR_NULL)) &&
4616             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));             (c != CHAR_RIGHT_SQUARE_BRACKET || inescq));
4617    
4618      /* Check for missing terminating ']' */      /* Check for missing terminating ']' */
4619    
4620      if (c == 0)      if (c == CHAR_NULL)
4621        {        {
4622        *errorcodeptr = ERR6;        *errorcodeptr = ERR6;
4623        goto FAILED;        goto FAILED;
4624        }        }
4625    
4626      /* 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
4627      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
4628      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
4629      optimize.      only if the very last character in the class needs XCLASS will it contain
4630        anything at this point. For this reason, xclass gets set TRUE above when
4631      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
4632      characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR      instead of just doing a test on class_uchardata below. */
4633      operate on single-bytes characters only. This is an historical hangover.  
4634      Maybe one day we can tidy these opcodes to handle multi-byte characters.  #if defined SUPPORT_UTF || !defined COMPILE_PCRE8
4635        if (class_uchardata > class_uchardata_base) xclass = TRUE;
4636      The optimization throws away the bit map. We turn the item into a  #endif
4637      1-character OP_CHAR[I] if it's positive, or OP_NOT[I] if it's negative.  
4638      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
4639      case, it can cause firstbyte to be set. Otherwise, there can be no first      setting, whatever the repeat count. Any reqchar setting must remain
4640      char if this item is first, whatever repeat count may follow. In the case      unchanged after any kind of repeat. */
4641      of reqbyte, save the previous value for reinstating. */  
4642        if (firstcharflags == REQ_UNSET) firstcharflags = REQ_NONE;
4643  #ifdef SUPPORT_UTF8      zerofirstchar = firstchar;
4644      if (class_charcount == 1 && !class_utf8 &&      zerofirstcharflags = firstcharflags;
4645        (!utf8 || !negate_class || class_lastchar < 128))      zeroreqchar = reqchar;
4646  #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;  
         zerofirstbyt