/[pcre]/code/trunk/pcre_internal.h
ViewVC logotype

Diff of /code/trunk/pcre_internal.h

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 1197 by chpe, Thu Nov 1 19:23:27 2012 UTC revision 1688 by ph10, Fri Feb 24 17:30:30 2017 UTC
# Line 7  Line 7 
7  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.
8    
9                         Written by Philip Hazel                         Written by Philip Hazel
10             Copyright (c) 1997-2012 University of Cambridge             Copyright (c) 1997-2016 University of Cambridge
11    
12  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
13  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 194  preprocessor time in standard C environm Line 194  preprocessor time in standard C environm
194  typedef unsigned char pcre_uint8;  typedef unsigned char pcre_uint8;
195    
196  #if USHRT_MAX == 65535  #if USHRT_MAX == 65535
197    typedef unsigned short pcre_uint16;  typedef unsigned short pcre_uint16;
198    typedef short pcre_int16;  typedef short pcre_int16;
199    #define PCRE_UINT16_MAX USHRT_MAX
200    #define PCRE_INT16_MAX SHRT_MAX
201  #elif UINT_MAX == 65535  #elif UINT_MAX == 65535
202    typedef unsigned int pcre_uint16;  typedef unsigned int pcre_uint16;
203    typedef int pcre_int16;  typedef int pcre_int16;
204    #define PCRE_UINT16_MAX UINT_MAX
205    #define PCRE_INT16_MAX INT_MAX
206  #else  #else
207    #error Cannot determine a type for 16-bit unsigned integers  #error Cannot determine a type for 16-bit integers
208  #endif  #endif
209    
210  #if UINT_MAX == 4294967295  #if UINT_MAX == 4294967295U
211    typedef unsigned int pcre_uint32;  typedef unsigned int pcre_uint32;
212    typedef int pcre_int32;  typedef int pcre_int32;
213  #elif ULONG_MAX == 4294967295  #define PCRE_UINT32_MAX UINT_MAX
214    typedef unsigned long int pcre_uint32;  #define PCRE_INT32_MAX INT_MAX
215    typedef long int pcre_int32;  #elif ULONG_MAX == 4294967295UL
216    typedef unsigned long int pcre_uint32;
217    typedef long int pcre_int32;
218    #define PCRE_UINT32_MAX ULONG_MAX
219    #define PCRE_INT32_MAX LONG_MAX
220  #else  #else
221    #error Cannot determine a type for 32-bit unsigned integers  #error Cannot determine a type for 32-bit integers
222  #endif  #endif
223    
224  /* When checking for integer overflow in pcre_compile(), we need to handle  /* When checking for integer overflow in pcre_compile(), we need to handle
# Line 267  pcre.h(.in) and disable (comment out) th Line 275  pcre.h(.in) and disable (comment out) th
275    
276  typedef pcre_uint16 pcre_uchar;  typedef pcre_uint16 pcre_uchar;
277  #define UCHAR_SHIFT (1)  #define UCHAR_SHIFT (1)
278  #define IN_UCHARS(x) ((x) << UCHAR_SHIFT)  #define IN_UCHARS(x) ((x) * 2)
279  #define MAX_255(c) ((c) <= 255u)  #define MAX_255(c) ((c) <= 255u)
280  #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))  #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
281    
# Line 275  typedef pcre_uint16 pcre_uchar; Line 283  typedef pcre_uint16 pcre_uchar;
283    
284  typedef pcre_uint32 pcre_uchar;  typedef pcre_uint32 pcre_uchar;
285  #define UCHAR_SHIFT (2)  #define UCHAR_SHIFT (2)
286  #define IN_UCHARS(x) ((x) << UCHAR_SHIFT)  #define IN_UCHARS(x) ((x) * 4)
287  #define MAX_255(c) ((c) <= 255u)  #define MAX_255(c) ((c) <= 255u)
288  #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))  #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
289    
 /* Assert that pcre_uchar32 is a 32-bit type */  
 typedef int __assert_pcre_uchar32_size[sizeof(pcre_uchar) == 4 ? 1 : -1];  
   
290  #else  #else
291  #error Unsupported compiling mode  #error Unsupported compiling mode
292  #endif /* COMPILE_PCRE[8|16|32] */  #endif /* COMPILE_PCRE[8|16|32] */
# Line 311  start/end of string field names are. */ Line 316  start/end of string field names are. */
316         &(NLBLOCK->nllen), utf)) \         &(NLBLOCK->nllen), utf)) \
317      : \      : \
318      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
319       RAWUCHARTEST(p) == NLBLOCK->nl[0] && \       UCHAR21TEST(p) == NLBLOCK->nl[0] && \
320       (NLBLOCK->nllen == 1 || RAWUCHARTEST(p+1) == NLBLOCK->nl[1])       \       (NLBLOCK->nllen == 1 || UCHAR21TEST(p+1) == NLBLOCK->nl[1])       \
321      ) \      ) \
322    )    )
323    
# Line 325  start/end of string field names are. */ Line 330  start/end of string field names are. */
330         &(NLBLOCK->nllen), utf)) \         &(NLBLOCK->nllen), utf)) \
331      : \      : \
332      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
333       RAWUCHARTEST(p - NLBLOCK->nllen) == NLBLOCK->nl[0] &&              \       UCHAR21TEST(p - NLBLOCK->nllen) == NLBLOCK->nl[0] &&              \
334       (NLBLOCK->nllen == 1 || RAWUCHARTEST(p - NLBLOCK->nllen + 1) == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || UCHAR21TEST(p - NLBLOCK->nllen + 1) == NLBLOCK->nl[1]) \
335      ) \      ) \
336    )    )
337    
# Line 351  values. */ Line 356  values. */
356  #include "pcre.h"  #include "pcre.h"
357  #include "ucp.h"  #include "ucp.h"
358    
359    #ifdef COMPILE_PCRE32
360    /* Assert that the public PCRE_UCHAR32 is a 32-bit type */
361    typedef int __assert_pcre_uchar32_size[sizeof(PCRE_UCHAR32) == 4 ? 1 : -1];
362    #endif
363    
364  /* When compiling for use with the Virtual Pascal compiler, these functions  /* When compiling for use with the Virtual Pascal compiler, these functions
365  need to have their names changed. PCRE must be compiled with the -DVPCOMPAT  need to have their names changed. PCRE must be compiled with the -DVPCOMPAT
366  option on the command line. */  option on the command line. */
# Line 529  capturing parenthesis numbers in back re Line 539  capturing parenthesis numbers in back re
539  #define PUT2(a,n,d)   \  #define PUT2(a,n,d)   \
540    a[n] = (d) >> 8; \    a[n] = (d) >> 8; \
541    a[(n)+1] = (d) & 255    a[(n)+1] = (d) & 255
542    
543  /* For reasons that I do not understand, the expression in this GET2 macro is  /* For reasons that I do not understand, the expression in this GET2 macro is
544  treated by gcc as a signed expression, even when a is declared as unsigned. It  treated by gcc as a signed expression, even when a is declared as unsigned. It
545  seems that any kind of arithmetic results in a signed value. */  seems that any kind of arithmetic results in a signed value. */
546    
547  #define GET2(a,n) \  #define GET2(a,n) \
# Line 572  changed in future to be a fixed number o Line 582  changed in future to be a fixed number o
582  #define MAX_MARK ((1u << 8) - 1)  #define MAX_MARK ((1u << 8) - 1)
583  #endif  #endif
584    
585    /* There is a proposed future special "UTF-21" mode, in which only the lowest
586    21 bits of a 32-bit character are interpreted as UTF, with the remaining 11
587    high-order bits available to the application for other uses. In preparation for
588    the future implementation of this mode, there are macros that load a data item
589    and, if in this special mode, mask it to 21 bits. These macros all have names
590    starting with UCHAR21. In all other modes, including the normal 32-bit
591    library, the macros all have the same simple definitions. When the new mode is
592    implemented, it is expected that these definitions will be varied appropriately
593    using #ifdef when compiling the library that supports the special mode. */
594    
595    #define UCHAR21(eptr)        (*(eptr))
596    #define UCHAR21TEST(eptr)    (*(eptr))
597    #define UCHAR21INC(eptr)     (*(eptr)++)
598    #define UCHAR21INCTEST(eptr) (*(eptr)++)
599    
600  /* When UTF encoding is being used, a character is no longer just a single  /* When UTF encoding is being used, a character is no longer just a single
601  byte. The macros for character handling generate simple sequences when used in  byte in 8-bit mode or a single short in 16-bit mode. The macros for character
602  character-mode, and more complicated ones for UTF characters. GETCHARLENTEST  handling generate simple sequences when used in the basic mode, and more
603  and other macros are not used when UTF is not supported, so they are not  complicated ones for UTF characters. GETCHARLENTEST and other macros are not
604  defined. To make sure they can never even appear when UTF support is omitted,  used when UTF is not supported. To make sure they can never even appear when
605  we don't even define them. */  UTF support is omitted, we don't even define them. */
606    
607  #ifndef SUPPORT_UTF  #ifndef SUPPORT_UTF
608    
# Line 590  we don't even define them. */ Line 615  we don't even define them. */
615  #define GETCHARINC(c, eptr) c = *eptr++;  #define GETCHARINC(c, eptr) c = *eptr++;
616  #define GETCHARINCTEST(c, eptr) c = *eptr++;  #define GETCHARINCTEST(c, eptr) c = *eptr++;
617  #define GETCHARLEN(c, eptr, len) c = *eptr;  #define GETCHARLEN(c, eptr, len) c = *eptr;
 #define RAWUCHAR(eptr) (*(eptr))  
 #define RAWUCHARINC(eptr) (*(eptr)++)  
 #define RAWUCHARTEST(eptr) (*(eptr))  
 #define RAWUCHARINCTEST(eptr) (*(eptr)++)  
618  /* #define GETCHARLENTEST(c, eptr, len) */  /* #define GETCHARLENTEST(c, eptr, len) */
619  /* #define BACKCHAR(eptr) */  /* #define BACKCHAR(eptr) */
620  /* #define FORWARDCHAR(eptr) */  /* #define FORWARDCHAR(eptr) */
# Line 766  do not know if we are in UTF-8 mode. */ Line 787  do not know if we are in UTF-8 mode. */
787    c = *eptr; \    c = *eptr; \
788    if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len);    if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len);
789    
 /* Returns the next uchar, not advancing the pointer. This is called when  
 we know we are in UTF mode. */  
   
 #define RAWUCHAR(eptr) \  
   (*(eptr))  
   
 /* Returns the next uchar, advancing the pointer. This is called when  
 we know we are in UTF mode. */  
   
 #define RAWUCHARINC(eptr) \  
   (*((eptr)++))  
   
 /* Returns the next uchar, testing for UTF mode, and not advancing the  
 pointer. */  
   
 #define RAWUCHARTEST(eptr) \  
   (*(eptr))  
   
 /* Returns the next uchar, testing for UTF mode, advancing the  
 pointer. */  
   
 #define RAWUCHARINCTEST(eptr) \  
   (*((eptr)++))  
   
790  /* If the pointer is not at the start of a character, move it back until  /* If the pointer is not at the start of a character, move it back until
791  it is. This is called only in UTF-8 mode - we don't put a test within the macro  it is. This is called only in UTF-8 mode - we don't put a test within the macro
792  because almost all calls are already within a block of UTF-8 only code. */  because almost all calls are already within a block of UTF-8 only code. */
# Line 885  we do not know if we are in UTF-16 mode. Line 882  we do not know if we are in UTF-16 mode.
882    c = *eptr; \    c = *eptr; \
883    if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);    if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);
884    
 /* Returns the next uchar, not advancing the pointer. This is called when  
 we know we are in UTF mode. */  
   
 #define RAWUCHAR(eptr) \  
   (*(eptr))  
   
 /* Returns the next uchar, advancing the pointer. This is called when  
 we know we are in UTF mode. */  
   
 #define RAWUCHARINC(eptr) \  
   (*((eptr)++))  
   
 /* Returns the next uchar, testing for UTF mode, and not advancing the  
 pointer. */  
   
 #define RAWUCHARTEST(eptr) \  
   (*(eptr))  
   
 /* Returns the next uchar, testing for UTF mode, advancing the  
 pointer. */  
   
 #define RAWUCHARINCTEST(eptr) \  
   (*((eptr)++))  
   
885  /* If the pointer is not at the start of a character, move it back until  /* If the pointer is not at the start of a character, move it back until
886  it is. This is called only in UTF-16 mode - we don't put a test within the  it is. This is called only in UTF-16 mode - we don't put a test within the
887  macro because almost all calls are already within a block of UTF-16 only  macro because almost all calls are already within a block of UTF-16 only
# Line 932  into one pcre_uchar unit. */ Line 905  into one pcre_uchar unit. */
905  #define GET_EXTRALEN(c) (0)  #define GET_EXTRALEN(c) (0)
906  #define NOT_FIRSTCHAR(c) (0)  #define NOT_FIRSTCHAR(c) (0)
907    
 #define UTF32_MASK (0x1fffffu)  
   
908  /* Get the next UTF-32 character, not advancing the pointer. This is called when  /* Get the next UTF-32 character, not advancing the pointer. This is called when
909  we know we are in UTF-32 mode. */  we know we are in UTF-32 mode. */
910    
911  #define GETCHAR(c, eptr) \  #define GETCHAR(c, eptr) \
912    c = (*eptr) & UTF32_MASK;    c = *(eptr);
913    
914  /* Get the next UTF-32 character, testing for UTF-32 mode, and not advancing the  /* Get the next UTF-32 character, testing for UTF-32 mode, and not advancing the
915  pointer. */  pointer. */
916    
917  #define GETCHARTEST(c, eptr) \  #define GETCHARTEST(c, eptr) \
918    c = *eptr; \    c = *(eptr);
   if (utf) c &= UTF32_MASK;  
919    
920  /* Get the next UTF-32 character, advancing the pointer. This is called when we  /* Get the next UTF-32 character, advancing the pointer. This is called when we
921  know we are in UTF-32 mode. */  know we are in UTF-32 mode. */
922    
923  #define GETCHARINC(c, eptr) \  #define GETCHARINC(c, eptr) \
924    c = (*eptr++) & UTF32_MASK;    c = *((eptr)++);
925    
926  /* Get the next character, testing for UTF-32 mode, and advancing the pointer.  /* Get the next character, testing for UTF-32 mode, and advancing the pointer.
927  This is called when we don't know if we are in UTF-32 mode. */  This is called when we don't know if we are in UTF-32 mode. */
928    
929  #define GETCHARINCTEST(c, eptr) \  #define GETCHARINCTEST(c, eptr) \
930    c = *eptr++; \    c = *((eptr)++);
   if (utf) c &= UTF32_MASK;  
931    
932  /* Get the next UTF-32 character, not advancing the pointer, not incrementing  /* Get the next UTF-32 character, not advancing the pointer, not incrementing
933  length (since all UTF-32 is of length 1). This is called when we know we are in  length (since all UTF-32 is of length 1). This is called when we know we are in
# Line 974  This is called when we do not know if we Line 943  This is called when we do not know if we
943  #define GETCHARLENTEST(c, eptr, len) \  #define GETCHARLENTEST(c, eptr, len) \
944    GETCHARTEST(c, eptr)    GETCHARTEST(c, eptr)
945    
 /* Returns the next uchar, not advancing the pointer. This is called when  
 we know we are in UTF mode. */  
   
 #define RAWUCHAR(eptr) \  
   (*(eptr) & UTF32_MASK)  
   
 /* Returns the next uchar, advancing the pointer. This is called when  
 we know we are in UTF mode. */  
   
 #define RAWUCHARINC(eptr) \  
   (*((eptr)++) & UTF32_MASK)  
   
 /* Returns the next uchar, testing for UTF mode, and not advancing the  
 pointer. */  
   
 #define RAWUCHARTEST(eptr) \  
   (utf ? (*(eptr) & UTF32_MASK) : *(eptr))  
   
 /* Returns the next uchar, testing for UTF mode, advancing the  
 pointer. */  
   
 #define RAWUCHARINCTEST(eptr) \  
   (utf ? (*((eptr)++) & UTF32_MASK) : *((eptr)++))  
   
946  /* If the pointer is not at the start of a character, move it back until  /* If the pointer is not at the start of a character, move it back until
947  it is. This is called only in UTF-32 mode - we don't put a test within the  it is. This is called only in UTF-32 mode - we don't put a test within the
948  macro because almost all calls are already within a block of UTF-32 only  macro because almost all calls are already within a block of UTF-32 only
949  code.  code.
950  These are all no-ops since all UTF-32 characters fit into one pcre_uchar. */  These are all no-ops since all UTF-32 characters fit into one pcre_uchar. */
951    
952  #define BACKCHAR(eptr) do { } while (0)  #define BACKCHAR(eptr) do { } while (0)
# Line 1028  are defined. Line 973  are defined.
973  These values are also required as lists in pcre_compile.c when processing \h,  These values are also required as lists in pcre_compile.c when processing \h,
974  \H, \v and \V in a character class. The lists are defined in pcre_tables.c, but  \H, \v and \V in a character class. The lists are defined in pcre_tables.c, but
975  macros that define the values are here so that all the definitions are  macros that define the values are here so that all the definitions are
976  together. The lists must be in ascending character order, terminated by  together. The lists must be in ascending character order, terminated by
977  NOTACHAR (which is 0xffffffff).  NOTACHAR (which is 0xffffffff).
978    
979  Any changes should ensure that the various macros are kept in step with each  Any changes should ensure that the various macros are kept in step with each
# Line 1039  other. NOTE: The values also appear in p Line 984  other. NOTE: The values also appear in p
984  #ifndef EBCDIC  #ifndef EBCDIC
985    
986  #define HSPACE_LIST \  #define HSPACE_LIST \
987    CHAR_HT, CHAR_SPACE, 0xa0, \    CHAR_HT, CHAR_SPACE, CHAR_NBSP, \
988    0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \    0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \
989    0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \    0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \
990    NOTACHAR    NOTACHAR
991    
992  #define HSPACE_MULTIBYTE_CASES \  #define HSPACE_MULTIBYTE_CASES \
993    case 0x1680:  /* OGHAM SPACE MARK */ \    case 0x1680:  /* OGHAM SPACE MARK */ \
# Line 1065  other. NOTE: The values also appear in p Line 1010  other. NOTE: The values also appear in p
1010  #define HSPACE_BYTE_CASES \  #define HSPACE_BYTE_CASES \
1011    case CHAR_HT: \    case CHAR_HT: \
1012    case CHAR_SPACE: \    case CHAR_SPACE: \
1013    case 0xa0     /* NBSP */    case CHAR_NBSP
1014    
1015  #define HSPACE_CASES \  #define HSPACE_CASES \
1016    HSPACE_BYTE_CASES: \    HSPACE_BYTE_CASES: \
1017    HSPACE_MULTIBYTE_CASES    HSPACE_MULTIBYTE_CASES
1018    
1019  #define VSPACE_LIST \  #define VSPACE_LIST \
1020    CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR    CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR
1021    
1022  #define VSPACE_MULTIBYTE_CASES \  #define VSPACE_MULTIBYTE_CASES \
1023    case 0x2028:    /* LINE SEPARATOR */ \    case 0x2028:    /* LINE SEPARATOR */ \
# Line 1092  other. NOTE: The values also appear in p Line 1037  other. NOTE: The values also appear in p
1037  /* ------ EBCDIC environments ------ */  /* ------ EBCDIC environments ------ */
1038    
1039  #else  #else
1040  #define HSPACE_LIST CHAR_HT, CHAR_SPACE  #define HSPACE_LIST CHAR_HT, CHAR_SPACE, CHAR_NBSP, NOTACHAR
1041    
1042  #define HSPACE_BYTE_CASES \  #define HSPACE_BYTE_CASES \
1043    case CHAR_HT: \    case CHAR_HT: \
1044    case CHAR_SPACE    case CHAR_SPACE: \
1045      case CHAR_NBSP
1046    
1047  #define HSPACE_CASES HSPACE_BYTE_CASES  #define HSPACE_CASES HSPACE_BYTE_CASES
1048    
1049  #ifdef EBCDIC_NL25  #ifdef EBCDIC_NL25
1050  #define VSPACE_LIST \  #define VSPACE_LIST \
1051    CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR    CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR
1052  #else  #else
1053  #define VSPACE_LIST \  #define VSPACE_LIST \
1054    CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR    CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR
1055  #endif  #endif
1056    
1057  #define VSPACE_BYTE_CASES \  #define VSPACE_BYTE_CASES \
1058    case CHAR_LF: \    case CHAR_LF: \
# Line 1123  other. NOTE: The values also appear in p Line 1069  other. NOTE: The values also appear in p
1069    
1070    
1071  /* Private flags containing information about the compiled regex. They used to  /* Private flags containing information about the compiled regex. They used to
1072  live at the top end of the options word, but that got almost full, so now they  live at the top end of the options word, but that got almost full, so they were
1073  are in a 16-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as  moved to a 16-bit flags word - which got almost full, so now they are in a
1074  the restrictions on partial matching have been lifted. It remains for backwards  32-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as the
1075    restrictions on partial matching have been lifted. It remains for backwards
1076  compatibility. */  compatibility. */
1077    
1078  #define PCRE_MODE8         0x0001  /* compiled in 8 bit mode */  #define PCRE_MODE8         0x00000001  /* compiled in 8 bit mode */
1079  #define PCRE_MODE16        0x0002  /* compiled in 16 bit mode */  #define PCRE_MODE16        0x00000002  /* compiled in 16 bit mode */
1080  #define PCRE_MODE32        0x0004  /* compiled in 32 bit mode */  #define PCRE_MODE32        0x00000004  /* compiled in 32 bit mode */
1081  #define PCRE_FIRSTSET      0x0010  /* first_char is set */  #define PCRE_FIRSTSET      0x00000010  /* first_char is set */
1082  #define PCRE_FCH_CASELESS  0x0020  /* caseless first char */  #define PCRE_FCH_CASELESS  0x00000020  /* caseless first char */
1083  #define PCRE_REQCHSET      0x0040  /* req_byte is set */  #define PCRE_REQCHSET      0x00000040  /* req_byte is set */
1084  #define PCRE_RCH_CASELESS  0x0080  /* caseless requested char */  #define PCRE_RCH_CASELESS  0x00000080  /* caseless requested char */
1085  #define PCRE_STARTLINE     0x0100  /* start after \n for multiline */  #define PCRE_STARTLINE     0x00000100  /* start after \n for multiline */
1086  #define PCRE_NOPARTIAL     0x0200  /* can't use partial with this regex */  #define PCRE_NOPARTIAL     0x00000200  /* can't use partial with this regex */
1087  #define PCRE_JCHANGED      0x0400  /* j option used in regex */  #define PCRE_JCHANGED      0x00000400  /* j option used in regex */
1088  #define PCRE_HASCRORLF     0x0800  /* explicit \r or \n in pattern */  #define PCRE_HASCRORLF     0x00000800  /* explicit \r or \n in pattern */
1089  #define PCRE_HASTHEN       0x1000  /* pattern contains (*THEN) */  #define PCRE_HASTHEN       0x00001000  /* pattern contains (*THEN) */
1090    #define PCRE_MLSET         0x00002000  /* match limit set by regex */
1091    #define PCRE_RLSET         0x00004000  /* recursion limit set by regex */
1092    #define PCRE_MATCH_EMPTY   0x00008000  /* pattern can match empty string */
1093    
1094  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
1095  #define PCRE_MODE          PCRE_MODE8  #define PCRE_MODE          PCRE_MODE8
# Line 1164  time, run time, or study time, respectiv Line 1114  time, run time, or study time, respectiv
1114  #define PUBLIC_COMPILE_OPTIONS \  #define PUBLIC_COMPILE_OPTIONS \
1115    (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \    (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
1116     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
1117     PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \     PCRE_NO_AUTO_CAPTURE|PCRE_NO_AUTO_POSSESS| \
1118       PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
1119     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \
1120     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE)     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE|PCRE_NEVER_UTF)
1121    
1122  #define PUBLIC_EXEC_OPTIONS \  #define PUBLIC_EXEC_OPTIONS \
1123    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \
# Line 1265  same code point. */ Line 1216  same code point. */
1216    
1217  #define CHAR_ESC                    '\047'  #define CHAR_ESC                    '\047'
1218  #define CHAR_DEL                    '\007'  #define CHAR_DEL                    '\007'
1219    #define CHAR_NBSP                   '\x41'
1220  #define STR_ESC                     "\047"  #define STR_ESC                     "\047"
1221  #define STR_DEL                     "\007"  #define STR_DEL                     "\007"
1222    
# Line 1279  a positive value. */ Line 1231  a positive value. */
1231  #define CHAR_NEL                    ((unsigned char)'\x85')  #define CHAR_NEL                    ((unsigned char)'\x85')
1232  #define CHAR_ESC                    '\033'  #define CHAR_ESC                    '\033'
1233  #define CHAR_DEL                    '\177'  #define CHAR_DEL                    '\177'
1234    #define CHAR_NBSP                   ((unsigned char)'\xa0')
1235    
1236  #define STR_LF                      "\n"  #define STR_LF                      "\n"
1237  #define STR_NL                      STR_LF  #define STR_NL                      STR_LF
# Line 1522  a positive value. */ Line 1475  a positive value. */
1475  #define STRING_xdigit               "xdigit"  #define STRING_xdigit               "xdigit"
1476    
1477  #define STRING_DEFINE               "DEFINE"  #define STRING_DEFINE               "DEFINE"
1478    #define STRING_WEIRD_STARTWORD      "[:<:]]"
1479    #define STRING_WEIRD_ENDWORD        "[:>:]]"
1480    
1481  #define STRING_CR_RIGHTPAR             "CR)"  #define STRING_CR_RIGHTPAR              "CR)"
1482  #define STRING_LF_RIGHTPAR             "LF)"  #define STRING_LF_RIGHTPAR              "LF)"
1483  #define STRING_CRLF_RIGHTPAR           "CRLF)"  #define STRING_CRLF_RIGHTPAR            "CRLF)"
1484  #define STRING_ANY_RIGHTPAR            "ANY)"  #define STRING_ANY_RIGHTPAR             "ANY)"
1485  #define STRING_ANYCRLF_RIGHTPAR        "ANYCRLF)"  #define STRING_ANYCRLF_RIGHTPAR         "ANYCRLF)"
1486  #define STRING_BSR_ANYCRLF_RIGHTPAR    "BSR_ANYCRLF)"  #define STRING_BSR_ANYCRLF_RIGHTPAR     "BSR_ANYCRLF)"
1487  #define STRING_BSR_UNICODE_RIGHTPAR    "BSR_UNICODE)"  #define STRING_BSR_UNICODE_RIGHTPAR     "BSR_UNICODE)"
1488  #ifdef COMPILE_PCRE8  #define STRING_UTF8_RIGHTPAR            "UTF8)"
1489  #define STRING_UTF_RIGHTPAR            "UTF8)"  #define STRING_UTF16_RIGHTPAR           "UTF16)"
1490  #endif  #define STRING_UTF32_RIGHTPAR           "UTF32)"
1491  #ifdef COMPILE_PCRE16  #define STRING_UTF_RIGHTPAR             "UTF)"
1492  #define STRING_UTF_RIGHTPAR            "UTF16)"  #define STRING_UCP_RIGHTPAR             "UCP)"
1493  #endif  #define STRING_NO_AUTO_POSSESS_RIGHTPAR "NO_AUTO_POSSESS)"
1494  #ifdef COMPILE_PCRE32  #define STRING_NO_START_OPT_RIGHTPAR    "NO_START_OPT)"
1495  #define STRING_UTF_RIGHTPAR            "UTF32)"  #define STRING_LIMIT_MATCH_EQ           "LIMIT_MATCH="
1496  #endif  #define STRING_LIMIT_RECURSION_EQ       "LIMIT_RECURSION="
 #define STRING_UCP_RIGHTPAR            "UCP)"  
 #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"  
1497    
1498  #else  /* SUPPORT_UTF */  #else  /* SUPPORT_UTF */
1499    
# Line 1656  only. */ Line 1609  only. */
1609  #define CHAR_VERTICAL_LINE          '\174'  #define CHAR_VERTICAL_LINE          '\174'
1610  #define CHAR_RIGHT_CURLY_BRACKET    '\175'  #define CHAR_RIGHT_CURLY_BRACKET    '\175'
1611  #define CHAR_TILDE                  '\176'  #define CHAR_TILDE                  '\176'
1612    #define CHAR_NBSP                   ((unsigned char)'\xa0')
1613    
1614  #define STR_HT                      "\011"  #define STR_HT                      "\011"
1615  #define STR_VT                      "\013"  #define STR_VT                      "\013"
# Line 1788  only. */ Line 1742  only. */
1742  #define STRING_xdigit               STR_x STR_d STR_i STR_g STR_i STR_t  #define STRING_xdigit               STR_x STR_d STR_i STR_g STR_i STR_t
1743    
1744  #define STRING_DEFINE               STR_D STR_E STR_F STR_I STR_N STR_E  #define STRING_DEFINE               STR_D STR_E STR_F STR_I STR_N STR_E
1745    #define STRING_WEIRD_STARTWORD      STR_LEFT_SQUARE_BRACKET STR_COLON STR_LESS_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
1746    #define STRING_WEIRD_ENDWORD        STR_LEFT_SQUARE_BRACKET STR_COLON STR_GREATER_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
1747    
1748  #define STRING_CR_RIGHTPAR             STR_C STR_R STR_RIGHT_PARENTHESIS  #define STRING_CR_RIGHTPAR              STR_C STR_R STR_RIGHT_PARENTHESIS
1749  #define STRING_LF_RIGHTPAR             STR_L STR_F STR_RIGHT_PARENTHESIS  #define STRING_LF_RIGHTPAR              STR_L STR_F STR_RIGHT_PARENTHESIS
1750  #define STRING_CRLF_RIGHTPAR           STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS  #define STRING_CRLF_RIGHTPAR            STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
1751  #define STRING_ANY_RIGHTPAR            STR_A STR_N STR_Y STR_RIGHT_PARENTHESIS  #define STRING_ANY_RIGHTPAR             STR_A STR_N STR_Y STR_RIGHT_PARENTHESIS
1752  #define STRING_ANYCRLF_RIGHTPAR        STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS  #define STRING_ANYCRLF_RIGHTPAR         STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
1753  #define STRING_BSR_ANYCRLF_RIGHTPAR    STR_B STR_S STR_R STR_UNDERSCORE STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS  #define STRING_BSR_ANYCRLF_RIGHTPAR     STR_B STR_S STR_R STR_UNDERSCORE STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
1754  #define STRING_BSR_UNICODE_RIGHTPAR    STR_B STR_S STR_R STR_UNDERSCORE STR_U STR_N STR_I STR_C STR_O STR_D STR_E STR_RIGHT_PARENTHESIS  #define STRING_BSR_UNICODE_RIGHTPAR     STR_B STR_S STR_R STR_UNDERSCORE STR_U STR_N STR_I STR_C STR_O STR_D STR_E STR_RIGHT_PARENTHESIS
1755  #ifdef COMPILE_PCRE8  #define STRING_UTF8_RIGHTPAR            STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS
1756  #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS  #define STRING_UTF16_RIGHTPAR           STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS
1757  #endif  #define STRING_UTF32_RIGHTPAR           STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS
1758  #ifdef COMPILE_PCRE16  #define STRING_UTF_RIGHTPAR             STR_U STR_T STR_F STR_RIGHT_PARENTHESIS
1759  #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS  #define STRING_UCP_RIGHTPAR             STR_U STR_C STR_P STR_RIGHT_PARENTHESIS
1760  #endif  #define STRING_NO_AUTO_POSSESS_RIGHTPAR STR_N STR_O STR_UNDERSCORE STR_A STR_U STR_T STR_O STR_UNDERSCORE STR_P STR_O STR_S STR_S STR_E STR_S STR_S STR_RIGHT_PARENTHESIS
1761  #ifdef COMPILE_PCRE32  #define STRING_NO_START_OPT_RIGHTPAR    STR_N STR_O STR_UNDERSCORE STR_S STR_T STR_A STR_R STR_T STR_UNDERSCORE STR_O STR_P STR_T STR_RIGHT_PARENTHESIS
1762  #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS  #define STRING_LIMIT_MATCH_EQ           STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_M STR_A STR_T STR_C STR_H STR_EQUALS_SIGN
1763  #endif  #define STRING_LIMIT_RECURSION_EQ       STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_R STR_E STR_C STR_U STR_R STR_S STR_I STR_O STR_N STR_EQUALS_SIGN
 #define STRING_UCP_RIGHTPAR            STR_U STR_C STR_P STR_RIGHT_PARENTHESIS  
 #define STRING_NO_START_OPT_RIGHTPAR   STR_N STR_O STR_UNDERSCORE STR_S STR_T STR_A STR_R STR_T STR_UNDERSCORE STR_O STR_P STR_T STR_RIGHT_PARENTHESIS  
1764    
1765  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
1766    
1767  /* Escape items that are just an encoding of a particular data value. */  /* Escape items that are just an encoding of a particular data value. */
1768    
1769    #ifndef ESC_a
1770    #define ESC_a CHAR_BEL
1771    #endif
1772    
1773  #ifndef ESC_e  #ifndef ESC_e
1774  #define ESC_e CHAR_ESC  #define ESC_e CHAR_ESC
1775  #endif  #endif
# Line 1847  only. */ Line 1805  only. */
1805  #define PT_PXSPACE    7    /* POSIX space - Z plus 9,10,11,12,13 */  #define PT_PXSPACE    7    /* POSIX space - Z plus 9,10,11,12,13 */
1806  #define PT_WORD       8    /* Word - L plus N plus underscore */  #define PT_WORD       8    /* Word - L plus N plus underscore */
1807  #define PT_CLIST      9    /* Pseudo-property: match character list */  #define PT_CLIST      9    /* Pseudo-property: match character list */
1808    #define PT_UCNC      10    /* Universal Character nameable character */
1809    #define PT_TABSIZE   11    /* Size of square table for autopossessify tests */
1810    
1811    /* The following special properties are used only in XCLASS items, when POSIX
1812    classes are specified and PCRE_UCP is set - in other words, for Unicode
1813    handling of these classes. They are not available via the \p or \P escapes like
1814    those in the above list, and so they do not take part in the autopossessifying
1815    table. */
1816    
1817    #define PT_PXGRAPH   11    /* [:graph:] - characters that mark the paper */
1818    #define PT_PXPRINT   12    /* [:print:] - [:graph:] plus non-control spaces */
1819    #define PT_PXPUNCT   13    /* [:punct:] - punctuation characters */
1820    
1821  /* Flag bits and data types for the extended class (OP_XCLASS) for classes that  /* Flag bits and data types for the extended class (OP_XCLASS) for classes that
1822  contain characters with values greater than 255. */  contain characters with values greater than 255. */
1823    
1824  #define XCL_NOT    0x01    /* Flag: this is a negative class */  #define XCL_NOT       0x01    /* Flag: this is a negative class */
1825  #define XCL_MAP    0x02    /* Flag: a 32-byte map is present */  #define XCL_MAP       0x02    /* Flag: a 32-byte map is present */
1826    #define XCL_HASPROP   0x04    /* Flag: property checks are present. */
1827    
1828  #define XCL_END       0    /* Marks end of individual items */  #define XCL_END       0    /* Marks end of individual items */
1829  #define XCL_SINGLE    1    /* Single item (one multibyte char) follows */  #define XCL_SINGLE    1    /* Single item (one multibyte char) follows */
# Line 1861  contain characters with values greater t Line 1832  contain characters with values greater t
1832  #define XCL_NOTPROP   4    /* Unicode inverted property (ditto) */  #define XCL_NOTPROP   4    /* Unicode inverted property (ditto) */
1833    
1834  /* These are escaped items that aren't just an encoding of a particular data  /* These are escaped items that aren't just an encoding of a particular data
1835  value such as \n. They must have non-zero values, as check_escape() returns  value such as \n. They must have non-zero values, as check_escape() returns 0
1836  0 for a data character.  Also, they must appear in the same order as in the opcode  for a data character.  Also, they must appear in the same order as in the
1837  definitions below, up to ESC_z. There's a dummy for OP_ALLANY because it  opcode definitions below, up to ESC_z. There's a dummy for OP_ALLANY because it
1838  corresponds to "." in DOTALL mode rather than an escape sequence. It is also  corresponds to "." in DOTALL mode rather than an escape sequence. It is also
1839  used for [^] in JavaScript compatibility mode, and for \C in non-utf mode. In  used for [^] in JavaScript compatibility mode, and for \C in non-utf mode. In
1840  non-DOTALL mode, "." behaves like \N.  non-DOTALL mode, "." behaves like \N.
# Line 1886  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, E Line 1857  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, E
1857         ESC_E, ESC_Q, ESC_g, ESC_k,         ESC_E, ESC_Q, ESC_g, ESC_k,
1858         ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu };         ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu };
1859    
 /* Opcode table: Starting from 1 (i.e. after OP_END), the values up to  
 OP_EOD must correspond in order to the list of escapes immediately above.  
1860    
1861  *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions  /********************** Opcode definitions ******************/
1862  that follow must also be updated to match. There are also tables called  
1863  "coptable" and "poptable" in pcre_dfa_exec.c that must be updated. */  /****** NOTE NOTE NOTE ******
1864    
1865    Starting from 1 (i.e. after OP_END), the values up to OP_EOD must correspond in
1866    order to the list of escapes immediately above. Furthermore, values up to
1867    OP_DOLLM must not be changed without adjusting the table called autoposstab in
1868    pcre_compile.c
1869    
1870    Whenever this list is updated, the two macro definitions that follow must be
1871    updated to match. The possessification table called "opcode_possessify" in
1872    pcre_compile.c must also be updated, and also the tables called "coptable"
1873    and "poptable" in pcre_dfa_exec.c.
1874    
1875    ****** NOTE NOTE NOTE ******/
1876    
1877    
1878    /* The values between FIRST_AUTOTAB_OP and LAST_AUTOTAB_RIGHT_OP, inclusive,
1879    are used in a table for deciding whether a repeated character type can be
1880    auto-possessified. */
1881    
1882    #define FIRST_AUTOTAB_OP       OP_NOT_DIGIT
1883    #define LAST_AUTOTAB_LEFT_OP   OP_EXTUNI
1884    #define LAST_AUTOTAB_RIGHT_OP  OP_DOLLM
1885    
1886  enum {  enum {
1887    OP_END,            /* 0 End of pattern */    OP_END,            /* 0 End of pattern */
# Line 1924  enum { Line 1914  enum {
1914    OP_EODN,           /* 23 End of data or \n at end of data (\Z) */    OP_EODN,           /* 23 End of data or \n at end of data (\Z) */
1915    OP_EOD,            /* 24 End of data (\z) */    OP_EOD,            /* 24 End of data (\z) */
1916    
1917    OP_CIRC,           /* 25 Start of line - not multiline */    /* Line end assertions */
1918    OP_CIRCM,          /* 26 Start of line - multiline */  
1919    OP_DOLL,           /* 27 End of line - not multiline */    OP_DOLL,           /* 25 End of line - not multiline */
1920    OP_DOLLM,          /* 28 End of line - multiline */    OP_DOLLM,          /* 26 End of line - multiline */
1921      OP_CIRC,           /* 27 Start of line - not multiline */
1922      OP_CIRCM,          /* 28 Start of line - multiline */
1923    
1924      /* Single characters; caseful must precede the caseless ones */
1925    
1926    OP_CHAR,           /* 29 Match one character, casefully */    OP_CHAR,           /* 29 Match one character, casefully */
1927    OP_CHARI,          /* 30 Match one character, caselessly */    OP_CHARI,          /* 30 Match one character, caselessly */
1928    OP_NOT,            /* 31 Match one character, not the given one, casefully */    OP_NOT,            /* 31 Match one character, not the given one, casefully */
# Line 1936  enum { Line 1931  enum {
1931    /* The following sets of 13 opcodes must always be kept in step because    /* The following sets of 13 opcodes must always be kept in step because
1932    the offset from the first one is used to generate the others. */    the offset from the first one is used to generate the others. */
1933    
1934    /**** Single characters, caseful, must precede the caseless ones ****/    /* Repeated characters; caseful must precede the caseless ones */
1935    
1936    OP_STAR,           /* 33 The maximizing and minimizing versions of */    OP_STAR,           /* 33 The maximizing and minimizing versions of */
1937    OP_MINSTAR,        /* 34 these six opcodes must come in pairs, with */    OP_MINSTAR,        /* 34 these six opcodes must come in pairs, with */
# Line 1954  enum { Line 1949  enum {
1949    OP_POSQUERY,       /* 44 Posesssified query, caseful */    OP_POSQUERY,       /* 44 Posesssified query, caseful */
1950    OP_POSUPTO,        /* 45 Possessified upto, caseful */    OP_POSUPTO,        /* 45 Possessified upto, caseful */
1951    
1952    /**** Single characters, caseless, must follow the caseful ones */    /* Repeated characters; caseless must follow the caseful ones */
1953    
1954    OP_STARI,          /* 46 */    OP_STARI,          /* 46 */
1955    OP_MINSTARI,       /* 47 */    OP_MINSTARI,       /* 47 */
# Line 1972  enum { Line 1967  enum {
1967    OP_POSQUERYI,      /* 57 Posesssified query, caseless */    OP_POSQUERYI,      /* 57 Posesssified query, caseless */
1968    OP_POSUPTOI,       /* 58 Possessified upto, caseless */    OP_POSUPTOI,       /* 58 Possessified upto, caseless */
1969    
1970    /**** The negated ones must follow the non-negated ones, and match them ****/    /* The negated ones must follow the non-negated ones, and match them */
1971    /**** Negated single character, caseful; must precede the caseless ones ****/    /* Negated repeated character, caseful; must precede the caseless ones */
1972    
1973    OP_NOTSTAR,        /* 59 The maximizing and minimizing versions of */    OP_NOTSTAR,        /* 59 The maximizing and minimizing versions of */
1974    OP_NOTMINSTAR,     /* 60 these six opcodes must come in pairs, with */    OP_NOTMINSTAR,     /* 60 these six opcodes must come in pairs, with */
# Line 1991  enum { Line 1986  enum {
1986    OP_NOTPOSQUERY,    /* 70 */    OP_NOTPOSQUERY,    /* 70 */
1987    OP_NOTPOSUPTO,     /* 71 */    OP_NOTPOSUPTO,     /* 71 */
1988    
1989    /**** Negated single character, caseless; must follow the caseful ones ****/    /* Negated repeated character, caseless; must follow the caseful ones */
1990    
1991    OP_NOTSTARI,       /* 72 */    OP_NOTSTARI,       /* 72 */
1992    OP_NOTMINSTARI,    /* 73 */    OP_NOTMINSTARI,    /* 73 */
# Line 2009  enum { Line 2004  enum {
2004    OP_NOTPOSQUERYI,   /* 83 */    OP_NOTPOSQUERYI,   /* 83 */
2005    OP_NOTPOSUPTOI,    /* 84 */    OP_NOTPOSUPTOI,    /* 84 */
2006    
2007    /**** Character types ****/    /* Character types */
2008    
2009    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */
2010    OP_TYPEMINSTAR,    /* 86 these six opcodes must come in pairs, with */    OP_TYPEMINSTAR,    /* 86 these six opcodes must come in pairs, with */
# Line 2040  enum { Line 2035  enum {
2035    OP_CRRANGE,        /* 104 These are different to the three sets above. */    OP_CRRANGE,        /* 104 These are different to the three sets above. */
2036    OP_CRMINRANGE,     /* 105 */    OP_CRMINRANGE,     /* 105 */
2037    
2038      OP_CRPOSSTAR,      /* 106 Possessified versions */
2039      OP_CRPOSPLUS,      /* 107 */
2040      OP_CRPOSQUERY,     /* 108 */
2041      OP_CRPOSRANGE,     /* 109 */
2042    
2043    /* End of quantifier opcodes */    /* End of quantifier opcodes */
2044    
2045    OP_CLASS,          /* 106 Match a character class, chars < 256 only */    OP_CLASS,          /* 110 Match a character class, chars < 256 only */
2046    OP_NCLASS,         /* 107 Same, but the bitmap was created from a negative    OP_NCLASS,         /* 111 Same, but the bitmap was created from a negative
2047                                class - the difference is relevant only when a                                class - the difference is relevant only when a
2048                                character > 255 is encountered. */                                character > 255 is encountered. */
2049    OP_XCLASS,         /* 108 Extended class for handling > 255 chars within the    OP_XCLASS,         /* 112 Extended class for handling > 255 chars within the
2050                                class. This does both positive and negative. */                                class. This does both positive and negative. */
2051    OP_REF,            /* 109 Match a back reference, casefully */    OP_REF,            /* 113 Match a back reference, casefully */
2052    OP_REFI,           /* 110 Match a back reference, caselessly */    OP_REFI,           /* 114 Match a back reference, caselessly */
2053    OP_RECURSE,        /* 111 Match a numbered subpattern (possibly recursive) */    OP_DNREF,          /* 115 Match a duplicate name backref, casefully */
2054    OP_CALLOUT,        /* 112 Call out to external function if provided */    OP_DNREFI,         /* 116 Match a duplicate name backref, caselessly */
2055      OP_RECURSE,        /* 117 Match a numbered subpattern (possibly recursive) */
2056    OP_ALT,            /* 113 Start of alternation */    OP_CALLOUT,        /* 118 Call out to external function if provided */
2057    OP_KET,            /* 114 End of group that doesn't have an unbounded repeat */  
2058    OP_KETRMAX,        /* 115 These two must remain together and in this */    OP_ALT,            /* 119 Start of alternation */
2059    OP_KETRMIN,        /* 116 order. They are for groups the repeat for ever. */    OP_KET,            /* 120 End of group that doesn't have an unbounded repeat */
2060    OP_KETRPOS,        /* 117 Possessive unlimited repeat. */    OP_KETRMAX,        /* 121 These two must remain together and in this */
2061      OP_KETRMIN,        /* 122 order. They are for groups the repeat for ever. */
2062      OP_KETRPOS,        /* 123 Possessive unlimited repeat. */
2063    
2064    /* The assertions must come before BRA, CBRA, ONCE, and COND, and the four    /* The assertions must come before BRA, CBRA, ONCE, and COND, and the four
2065    asserts must remain in order. */    asserts must remain in order. */
2066    
2067    OP_REVERSE,        /* 118 Move pointer back - used in lookbehind assertions */    OP_REVERSE,        /* 124 Move pointer back - used in lookbehind assertions */
2068    OP_ASSERT,         /* 119 Positive lookahead */    OP_ASSERT,         /* 125 Positive lookahead */
2069    OP_ASSERT_NOT,     /* 120 Negative lookahead */    OP_ASSERT_NOT,     /* 126 Negative lookahead */
2070    OP_ASSERTBACK,     /* 121 Positive lookbehind */    OP_ASSERTBACK,     /* 127 Positive lookbehind */
2071    OP_ASSERTBACK_NOT, /* 122 Negative lookbehind */    OP_ASSERTBACK_NOT, /* 128 Negative lookbehind */
2072    
2073    /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately    /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately
2074    after the assertions, with ONCE first, as there's a test for >= ONCE for a    after the assertions, with ONCE first, as there's a test for >= ONCE for a
2075    subpattern that isn't an assertion. The POS versions must immediately follow    subpattern that isn't an assertion. The POS versions must immediately follow
2076    the non-POS versions in each case. */    the non-POS versions in each case. */
2077    
2078    OP_ONCE,           /* 123 Atomic group, contains captures */    OP_ONCE,           /* 129 Atomic group, contains captures */
2079    OP_ONCE_NC,        /* 124 Atomic group containing no captures */    OP_ONCE_NC,        /* 130 Atomic group containing no captures */
2080    OP_BRA,            /* 125 Start of non-capturing bracket */    OP_BRA,            /* 131 Start of non-capturing bracket */
2081    OP_BRAPOS,         /* 126 Ditto, with unlimited, possessive repeat */    OP_BRAPOS,         /* 132 Ditto, with unlimited, possessive repeat */
2082    OP_CBRA,           /* 127 Start of capturing bracket */    OP_CBRA,           /* 133 Start of capturing bracket */
2083    OP_CBRAPOS,        /* 128 Ditto, with unlimited, possessive repeat */    OP_CBRAPOS,        /* 134 Ditto, with unlimited, possessive repeat */
2084    OP_COND,           /* 129 Conditional group */    OP_COND,           /* 135 Conditional group */
2085    
2086    /* These five must follow the previous five, in the same order. There's a    /* These five must follow the previous five, in the same order. There's a
2087    check for >= SBRA to distinguish the two sets. */    check for >= SBRA to distinguish the two sets. */
2088    
2089    OP_SBRA,           /* 130 Start of non-capturing bracket, check empty  */    OP_SBRA,           /* 136 Start of non-capturing bracket, check empty  */
2090    OP_SBRAPOS,        /* 131 Ditto, with unlimited, possessive repeat */    OP_SBRAPOS,        /* 137 Ditto, with unlimited, possessive repeat */
2091    OP_SCBRA,          /* 132 Start of capturing bracket, check empty */    OP_SCBRA,          /* 138 Start of capturing bracket, check empty */
2092    OP_SCBRAPOS,       /* 133 Ditto, with unlimited, possessive repeat */    OP_SCBRAPOS,       /* 139 Ditto, with unlimited, possessive repeat */
2093    OP_SCOND,          /* 134 Conditional group, check empty */    OP_SCOND,          /* 140 Conditional group, check empty */
2094    
2095    /* The next two pairs must (respectively) be kept together. */    /* The next two pairs must (respectively) be kept together. */
2096    
2097    OP_CREF,           /* 135 Used to hold a capture number as condition */    OP_CREF,           /* 141 Used to hold a capture number as condition */
2098    OP_NCREF,          /* 136 Same, but generated by a name reference*/    OP_DNCREF,         /* 142 Used to point to duplicate names as a condition */
2099    OP_RREF,           /* 137 Used to hold a recursion number as condition */    OP_RREF,           /* 143 Used to hold a recursion number as condition */
2100    OP_NRREF,          /* 138 Same, but generated by a name reference*/    OP_DNRREF,         /* 144 Used to point to duplicate names as a condition */
2101    OP_DEF,            /* 139 The DEFINE condition */    OP_DEF,            /* 145 The DEFINE condition */
2102    
2103    OP_BRAZERO,        /* 140 These two must remain together and in this */    OP_BRAZERO,        /* 146 These two must remain together and in this */
2104    OP_BRAMINZERO,     /* 141 order. */    OP_BRAMINZERO,     /* 147 order. */
2105    OP_BRAPOSZERO,     /* 142 */    OP_BRAPOSZERO,     /* 148 */
2106    
2107    /* These are backtracking control verbs */    /* These are backtracking control verbs */
2108    
2109    OP_MARK,           /* 143 always has an argument */    OP_MARK,           /* 149 always has an argument */
2110    OP_PRUNE,          /* 144 */    OP_PRUNE,          /* 150 */
2111    OP_PRUNE_ARG,      /* 145 same, but with argument */    OP_PRUNE_ARG,      /* 151 same, but with argument */
2112    OP_SKIP,           /* 146 */    OP_SKIP,           /* 152 */
2113    OP_SKIP_ARG,       /* 147 same, but with argument */    OP_SKIP_ARG,       /* 153 same, but with argument */
2114    OP_THEN,           /* 148 */    OP_THEN,           /* 154 */
2115    OP_THEN_ARG,       /* 149 same, but with argument */    OP_THEN_ARG,       /* 155 same, but with argument */
2116    OP_COMMIT,         /* 150 */    OP_COMMIT,         /* 156 */
2117    
2118    /* These are forced failure and success verbs */    /* These are forced failure and success verbs */
2119    
2120    OP_FAIL,           /* 151 */    OP_FAIL,           /* 157 */
2121    OP_ACCEPT,         /* 152 */    OP_ACCEPT,         /* 158 */
2122    OP_ASSERT_ACCEPT,  /* 153 Used inside assertions */    OP_ASSERT_ACCEPT,  /* 159 Used inside assertions */
2123    OP_CLOSE,          /* 154 Used before OP_ACCEPT to close open captures */    OP_CLOSE,          /* 160 Used before OP_ACCEPT to close open captures */
2124    
2125    /* This is used to skip a subpattern with a {0} quantifier */    /* This is used to skip a subpattern with a {0} quantifier */
2126    
2127    OP_SKIPZERO,       /* 155 */    OP_SKIPZERO,       /* 161 */
2128    
2129    /* This is not an opcode, but is used to check that tables indexed by opcode    /* This is not an opcode, but is used to check that tables indexed by opcode
2130    are the correct length, in order to catch updating errors - there have been    are the correct length, in order to catch updating errors - there have been
# Line 2133  enum { Line 2135  enum {
2135    
2136  /* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro  /* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro
2137  definitions that follow must also be updated to match. There are also tables  definitions that follow must also be updated to match. There are also tables
2138  called "coptable" and "poptable" in pcre_dfa_exec.c that must be updated. */  called "opcode_possessify" in pcre_compile.c and "coptable" and "poptable" in
2139    pcre_dfa_exec.c that must be updated. */
2140    
2141    
2142  /* This macro defines textual names for all the opcodes. These are used only  /* This macro defines textual names for all the opcodes. These are used only
# Line 2146  some cases doesn't actually use these na Line 2149  some cases doesn't actually use these na
2149    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \
2150    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \
2151    "extuni",  "\\Z", "\\z",                                        \    "extuni",  "\\Z", "\\z",                                        \
2152    "^", "^", "$", "$", "char", "chari", "not", "noti",             \    "$", "$", "^", "^", "char", "chari", "not", "noti",             \
2153    "*", "*?", "+", "+?", "?", "??",                                \    "*", "*?", "+", "+?", "?", "??",                                \
2154    "{", "{", "{",                                                  \    "{", "{", "{",                                                  \
2155    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
# Line 2162  some cases doesn't actually use these na Line 2165  some cases doesn't actually use these na
2165    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \
2166    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
2167    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \
2168    "class", "nclass", "xclass", "Ref", "Refi",                     \    "*+","++", "?+", "{",                                           \
2169      "class", "nclass", "xclass", "Ref", "Refi", "DnRef", "DnRefi",  \
2170    "Recurse", "Callout",                                           \    "Recurse", "Callout",                                           \
2171    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \
2172    "Reverse", "Assert", "Assert not", "AssertB", "AssertB not",    \    "Reverse", "Assert", "Assert not", "AssertB", "AssertB not",    \
# Line 2171  some cases doesn't actually use these na Line 2175  some cases doesn't actually use these na
2175    "Cond",                                                         \    "Cond",                                                         \
2176    "SBra", "SBraPos", "SCBra", "SCBraPos",                         \    "SBra", "SBraPos", "SCBra", "SCBraPos",                         \
2177    "SCond",                                                        \    "SCond",                                                        \
2178    "Cond ref", "Cond nref", "Cond rec", "Cond nrec", "Cond def",   \    "Cond ref", "Cond dnref", "Cond rec", "Cond dnrec", "Cond def", \
2179    "Brazero", "Braminzero", "Braposzero",                          \    "Brazero", "Braminzero", "Braposzero",                          \
2180    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \
2181    "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \    "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \
# Line 2196  in UTF-8 mode. The code that uses this t Line 2200  in UTF-8 mode. The code that uses this t
2200    3, 3,                          /* \P, \p                                 */ \    3, 3,                          /* \P, \p                                 */ \
2201    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \
2202    1,                             /* \X                                     */ \    1,                             /* \X                                     */ \
2203    1, 1, 1, 1, 1, 1,              /* \Z, \z, ^, ^M, $, $M                   */ \    1, 1, 1, 1, 1, 1,              /* \Z, \z, $, $M ^, ^M                    */ \
2204    2,                             /* Char  - the minimum length             */ \    2,                             /* Char  - the minimum length             */ \
2205    2,                             /* Chari  - the minimum length            */ \    2,                             /* Chari  - the minimum length            */ \
2206    2,                             /* not                                    */ \    2,                             /* not                                    */ \
# Line 2227  in UTF-8 mode. The code that uses this t Line 2231  in UTF-8 mode. The code that uses this t
2231    /* Character class & ref repeats                                         */ \    /* Character class & ref repeats                                         */ \
2232    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \
2233    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \
2234      1, 1, 1, 1+2*IMM2_SIZE,        /* Possessive *+, ++, ?+, CRPOSRANGE      */ \
2235    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \
2236    1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \    1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \
2237    0,                             /* XCLASS - variable length               */ \    0,                             /* XCLASS - variable length               */ \
2238    1+IMM2_SIZE,                   /* REF                                    */ \    1+IMM2_SIZE,                   /* REF                                    */ \
2239    1+IMM2_SIZE,                   /* REFI                                   */ \    1+IMM2_SIZE,                   /* REFI                                   */ \
2240      1+2*IMM2_SIZE,                 /* DNREF                                  */ \
2241      1+2*IMM2_SIZE,                 /* DNREFI                                 */ \
2242    1+LINK_SIZE,                   /* RECURSE                                */ \    1+LINK_SIZE,                   /* RECURSE                                */ \
2243    2+2*LINK_SIZE,                 /* CALLOUT                                */ \    2+2*LINK_SIZE,                 /* CALLOUT                                */ \
2244    1+LINK_SIZE,                   /* Alt                                    */ \    1+LINK_SIZE,                   /* Alt                                    */ \
# Line 2256  in UTF-8 mode. The code that uses this t Line 2263  in UTF-8 mode. The code that uses this t
2263    1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \    1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \
2264    1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \    1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \
2265    1+LINK_SIZE,                   /* SCOND                                  */ \    1+LINK_SIZE,                   /* SCOND                                  */ \
2266    1+IMM2_SIZE, 1+IMM2_SIZE,      /* CREF, NCREF                            */ \    1+IMM2_SIZE, 1+2*IMM2_SIZE,    /* CREF, DNCREF                           */ \
2267    1+IMM2_SIZE, 1+IMM2_SIZE,      /* RREF, NRREF                            */ \    1+IMM2_SIZE, 1+2*IMM2_SIZE,    /* RREF, DNRREF                           */ \
2268    1,                             /* DEF                                    */ \    1,                             /* DEF                                    */ \
2269    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \
2270    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \
# Line 2266  in UTF-8 mode. The code that uses this t Line 2273  in UTF-8 mode. The code that uses this t
2273    1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \    1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \
2274    1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */    1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */
2275    
2276  /* A magic value for OP_RREF and OP_NRREF to indicate the "any recursion"  /* A magic value for OP_RREF to indicate the "any recursion" condition. */
 condition. */  
2277    
2278  #define RREF_ANY  0xffff  #define RREF_ANY  0xffff
2279    
# Line 2282  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4, Line 2288  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,
2288         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
2289         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
2290         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,
2291         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERRCOUNT };         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERR78, ERR79,
2292           ERR80, ERR81, ERR82, ERR83, ERR84, ERR85, ERR86, ERR87, ERRCOUNT };
2293    
2294  /* JIT compiling modes. The function list is indexed by them. */  /* JIT compiling modes. The function list is indexed by them. */
2295    
2296  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,
2297         JIT_NUMBER_OF_COMPILE_MODES };         JIT_NUMBER_OF_COMPILE_MODES };
2298    
# Line 2292  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COM Line 2300  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COM
2300  code vector run on as long as necessary after the end. We store an explicit  code vector run on as long as necessary after the end. We store an explicit
2301  offset to the name table so that if a regex is compiled on one host, saved, and  offset to the name table so that if a regex is compiled on one host, saved, and
2302  then run on another where the size of pointers is different, all might still  then run on another where the size of pointers is different, all might still
2303  be well. For the case of compiled-on-4 and run-on-8, we include an extra  be well.
 pointer that is always NULL. For future-proofing, a few dummy fields were  
 originally included - even though you can never get this planning right - but  
 there is only one left now.  
   
 NOTE NOTE NOTE:  
 Because people can now save and re-use compiled patterns, any additions to this  
 structure should be made at the end, and something earlier (e.g. a new  
 flag in the options or one of the dummy fields) should indicate that the new  
 fields are present. Currently PCRE always sets the dummy fields to zero.  
 NOTE NOTE NOTE  
 */  
2304    
2305  #if defined COMPILE_PCRE8  The size of the structure must be a multiple of 8 bytes. For the case of
2306  #define REAL_PCRE real_pcre  compiled-on-4 and run-on-8, we include an extra pointer that is always NULL so
2307  #elif defined COMPILE_PCRE16  that there are an even number of pointers which therefore are a multiple of 8
2308  #define REAL_PCRE real_pcre16  bytes.
2309  #elif defined COMPILE_PCRE32  
2310  #define REAL_PCRE real_pcre32  It is necessary to fork the struct for the 32 bit library, since it needs to
2311  #endif  use pcre_uint32 for first_char and req_char. We can't put an ifdef inside the
2312    typedef because pcretest needs access to the struct of the 8-, 16- and 32-bit
2313  /* It is necessary to fork the struct for 32 bit, since it needs to use  variants.
2314   * pcre_uchar for first_char and req_char. Can't put an ifdef inside the  
2315   * typedef since pcretest needs access to  the struct of the 8-, 16-  *** WARNING ***
2316   * and 32-bit variants. */  When new fields are added to these structures, remember to adjust the code in
2317    pcre_byte_order.c that is concerned with swapping the byte order of the fields
2318    when a compiled regex is reloaded on a host with different endianness.
2319    *** WARNING ***
2320    There is also similar byte-flipping code in pcretest.c, which is used for
2321    testing the byte-flipping features. It must also be kept in step.
2322    *** WARNING ***
2323    */
2324    
2325  typedef struct real_pcre8_or_16 {  typedef struct real_pcre8_or_16 {
2326    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2327    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2328    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2329    pcre_uint16 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2330      pcre_uint32 limit_match;        /* Limit set from regex */
2331      pcre_uint32 limit_recursion;    /* Limit set from regex */
2332      pcre_uint16 first_char;         /* Starting character */
2333      pcre_uint16 req_char;           /* This character must be seen */
2334    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2335    pcre_uint16 top_bracket;        /* Highest numbered group */    pcre_uint16 top_bracket;        /* Highest numbered group */
2336    pcre_uint16 top_backref;        /* Highest numbered back reference */    pcre_uint16 top_backref;        /* Highest numbered back reference */
   pcre_uint16 first_char;         /* Starting character */  
   pcre_uint16 req_char;           /* This character must be seen */  
2337    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2338    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2339    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2340    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2341      pcre_uint16 dummy1;             /* To ensure size is a multiple of 8 */
2342      pcre_uint16 dummy2;             /* To ensure size is a multiple of 8 */
2343      pcre_uint16 dummy3;             /* To ensure size is a multiple of 8 */
2344    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2345    const pcre_uint8 *nullpad;      /* NULL padding */    void             *nullpad;      /* NULL padding */
2346  } real_pcre8_or_16;  } real_pcre8_or_16;
2347    
2348  typedef struct real_pcre8_or_16 real_pcre;  typedef struct real_pcre8_or_16 real_pcre;
# Line 2343  typedef struct real_pcre32 { Line 2352  typedef struct real_pcre32 {
2352    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2353    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2354    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2355    pcre_uint16 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2356      pcre_uint32 limit_match;        /* Limit set from regex */
2357      pcre_uint32 limit_recursion;    /* Limit set from regex */
2358      pcre_uint32 first_char;         /* Starting character */
2359      pcre_uint32 req_char;           /* This character must be seen */
2360    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2361    pcre_uint16 top_bracket;        /* Highest numbered group */    pcre_uint16 top_bracket;        /* Highest numbered group */
2362    pcre_uint16 top_backref;        /* Highest numbered back reference */    pcre_uint16 top_backref;        /* Highest numbered back reference */
   pcre_uint32 first_char;         /* Starting character */  
   pcre_uint32 req_char;           /* This character must be seen */  
2363    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2364    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2365    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2366    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2367    pcre_uint16 dummy1;             /* for later expansion */    pcre_uint16 dummy;              /* To ensure size is a multiple of 8 */
   pcre_uint16 dummy2;             /* for later expansion */  
2368    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2369    void *nullpad;                  /* for later expansion */    void             *nullpad;      /* NULL padding */
2370  } real_pcre32;  } real_pcre32;
2371    
2372    #if defined COMPILE_PCRE8
2373    #define REAL_PCRE real_pcre
2374    #elif defined COMPILE_PCRE16
2375    #define REAL_PCRE real_pcre16
2376    #elif defined COMPILE_PCRE32
2377    #define REAL_PCRE real_pcre32
2378    #endif
2379    
2380  /* Assert that the size of REAL_PCRE is divisible by 8 */  /* Assert that the size of REAL_PCRE is divisible by 8 */
2381  typedef int __assert_real_pcre_size_divisible_8[(sizeof(REAL_PCRE) % 8) == 0 ? 1 : -1];  typedef int __assert_real_pcre_size_divisible_8[(sizeof(REAL_PCRE) % 8) == 0 ? 1 : -1];
2382    
# Line 2392  typedef struct open_capitem { Line 2410  typedef struct open_capitem {
2410    pcre_uint16 flag;             /* Set TRUE if recursive back ref */    pcre_uint16 flag;             /* Set TRUE if recursive back ref */
2411  } open_capitem;  } open_capitem;
2412    
2413    /* Structure for building a list of named groups during the first pass of
2414    compiling. */
2415    
2416    typedef struct named_group {
2417      const pcre_uchar  *name;          /* Points to the name in the pattern */
2418      int                length;        /* Length of the name */
2419      pcre_uint32        number;        /* Group number */
2420    } named_group;
2421    
2422  /* Structure for passing "static" information around between the functions  /* Structure for passing "static" information around between the functions
2423  doing the compiling, so that they are thread-safe. */  doing the compiling, so that they are thread-safe. */
2424    
# Line 2404  typedef struct compile_data { Line 2431  typedef struct compile_data {
2431    const pcre_uchar *start_code;     /* The start of the compiled code */    const pcre_uchar *start_code;     /* The start of the compiled code */
2432    const pcre_uchar *start_pattern;  /* The start of the pattern */    const pcre_uchar *start_pattern;  /* The start of the pattern */
2433    const pcre_uchar *end_pattern;    /* The end of the pattern */    const pcre_uchar *end_pattern;    /* The end of the pattern */
   open_capitem *open_caps;          /* Chain of open capture items */  
2434    pcre_uchar *hwm;                  /* High watermark of workspace */    pcre_uchar *hwm;                  /* High watermark of workspace */
2435      open_capitem *open_caps;          /* Chain of open capture items */
2436      named_group *named_groups;        /* Points to vector in pre-compile */
2437    pcre_uchar *name_table;           /* The name/number table */    pcre_uchar *name_table;           /* The name/number table */
2438    int  names_found;                 /* Number of entries so far */    int  names_found;                 /* Number of entries so far */
2439    int  name_entry_size;             /* Size of each entry */    int  name_entry_size;             /* Size of each entry */
2440      int  named_group_list_size;       /* Number of entries in the list */
2441    int  workspace_size;              /* Size of workspace */    int  workspace_size;              /* Size of workspace */
2442    unsigned int  bracount;           /* Count of capturing parens as we compile */    unsigned int bracount;            /* Count of capturing parens as we compile */
2443    int  final_bracount;              /* Saved value after first pass */    int  final_bracount;              /* Saved value after first pass */
2444    int  max_lookbehind;              /* Maximum lookbehind (characters) */    int  max_lookbehind;              /* Maximum lookbehind (characters) */
2445    int  top_backref;                 /* Maximum back reference */    int  top_backref;                 /* Maximum back reference */
2446    unsigned int backref_map;         /* Bitmap of low back refs */    unsigned int backref_map;         /* Bitmap of low back refs */
2447      unsigned int namedrefcount;       /* Number of backreferences by name */
2448      int  parens_depth;                /* Depth of nested parentheses */
2449    int  assert_depth;                /* Depth of nested assertions */    int  assert_depth;                /* Depth of nested assertions */
2450    int  external_options;            /* External (initial) options */    pcre_uint32 external_options;     /* External (initial) options */
2451    int  external_flags;              /* External flag bits to be set */    pcre_uint32 external_flags;       /* External flag bits to be set */
2452    int  req_varyopt;                 /* "After variable item" flag for reqbyte */    int  req_varyopt;                 /* "After variable item" flag for reqbyte */
2453    BOOL had_accept;                  /* (*ACCEPT) encountered */    BOOL had_accept;                  /* (*ACCEPT) encountered */
2454    BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */    BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */
2455    BOOL check_lookbehind;            /* Lookbehinds need later checking */    BOOL check_lookbehind;            /* Lookbehinds need later checking */
2456      BOOL dupnames;                    /* Duplicate names exist */
2457      BOOL dupgroups;                   /* Duplicate groups exist: (?| found */
2458      BOOL iscondassert;                /* Next assert is a condition */
2459    int  nltype;                      /* Newline type */    int  nltype;                      /* Newline type */
2460    int  nllen;                       /* Newline string length */    int  nllen;                       /* Newline string length */
2461    pcre_uchar nl[4];                 /* Newline string when fixed length */    pcre_uchar nl[4];                 /* Newline string when fixed length */
# Line 2435  typedef struct branch_chain { Line 2469  typedef struct branch_chain {
2469    pcre_uchar *current_branch;    pcre_uchar *current_branch;
2470  } branch_chain;  } branch_chain;
2471    
2472    /* Structure for mutual recursion detection. */
2473    
2474    typedef struct recurse_check {
2475      struct recurse_check *prev;
2476      const pcre_uchar *group;
2477    } recurse_check;
2478    
2479  /* Structure for items in a linked list that represents an explicit recursive  /* Structure for items in a linked list that represents an explicit recursive
2480  call within the pattern; used by pcre_exec(). */  call within the pattern; used by pcre_exec(). */
2481    
# Line 2443  typedef struct recursion_info { Line 2484  typedef struct recursion_info {
2484    unsigned int group_num;         /* Number of group that was called */    unsigned int group_num;         /* Number of group that was called */
2485    int *offset_save;               /* Pointer to start of saved offsets */    int *offset_save;               /* Pointer to start of saved offsets */
2486    int saved_max;                  /* Number of saved offsets */    int saved_max;                  /* Number of saved offsets */
2487      int saved_capture_last;         /* Last capture number */
2488    PCRE_PUCHAR subject_position;   /* Position at start of recursion */    PCRE_PUCHAR subject_position;   /* Position at start of recursion */
2489  } recursion_info;  } recursion_info;
2490    
# Line 2479  typedef struct match_data { Line 2521  typedef struct match_data {
2521    int    nllen;                   /* Newline string length */    int    nllen;                   /* Newline string length */
2522    int    name_count;              /* Number of names in name table */    int    name_count;              /* Number of names in name table */
2523    int    name_entry_size;         /* Size of entry in names table */    int    name_entry_size;         /* Size of entry in names table */
2524      unsigned int skip_arg_count;    /* For counting SKIP_ARGs */
2525      unsigned int ignore_skip_arg;   /* For re-run when SKIP arg name not found */
2526    pcre_uchar *name_table;         /* Table of names */    pcre_uchar *name_table;         /* Table of names */
2527    pcre_uchar nl[4];               /* Newline string when fixed */    pcre_uchar nl[4];               /* Newline string when fixed */
2528    const  pcre_uint8 *lcc;         /* Points to lower casing table */    const  pcre_uint8 *lcc;         /* Points to lower casing table */
2529    const  pcre_uint8 *fcc;         /* Points to case-flipping table */    const  pcre_uint8 *fcc;         /* Points to case-flipping table */
2530    const  pcre_uint8 *ctypes;      /* Points to table of type maps */    const  pcre_uint8 *ctypes;      /* Points to table of type maps */
   BOOL   offset_overflow;         /* Set if too many extractions */  
2531    BOOL   notbol;                  /* NOTBOL flag */    BOOL   notbol;                  /* NOTBOL flag */
2532    BOOL   noteol;                  /* NOTEOL flag */    BOOL   noteol;                  /* NOTEOL flag */
2533    BOOL   utf;                     /* UTF-8 / UTF-16 flag */    BOOL   utf;                     /* UTF-8 / UTF-16 flag */
# Line 2496  typedef struct match_data { Line 2539  typedef struct match_data {
2539    BOOL   hitend;                  /* Hit the end of the subject at some point */    BOOL   hitend;                  /* Hit the end of the subject at some point */
2540    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */
2541    BOOL   hasthen;                 /* Pattern contains (*THEN) */    BOOL   hasthen;                 /* Pattern contains (*THEN) */
   BOOL   ignore_skip_arg;         /* For re-run when SKIP name not found */  
2542    const  pcre_uchar *start_code;  /* For use when recursing */    const  pcre_uchar *start_code;  /* For use when recursing */
2543    PCRE_PUCHAR start_subject;      /* Start of the subject string */    PCRE_PUCHAR start_subject;      /* Start of the subject string */
2544    PCRE_PUCHAR end_subject;        /* End of the subject string */    PCRE_PUCHAR end_subject;        /* End of the subject string */
# Line 2505  typedef struct match_data { Line 2547  typedef struct match_data {
2547    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */
2548    int    partial;                 /* PARTIAL options */    int    partial;                 /* PARTIAL options */
2549    int    end_offset_top;          /* Highwater mark at end of match */    int    end_offset_top;          /* Highwater mark at end of match */
2550    int    capture_last;            /* Most recent capture number */    pcre_int32 capture_last;        /* Most recent capture number + overflow flag */
2551    int    start_offset;            /* The start offset value */    int    start_offset;            /* The start offset value */
2552    int    match_function_type;     /* Set for certain special calls of MATCH() */    int    match_function_type;     /* Set for certain special calls of MATCH() */
2553    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */
# Line 2730  extern const pcre_uint8  PRIV(ucd_stage1 Line 2772  extern const pcre_uint8  PRIV(ucd_stage1
2772  extern const pcre_uint16 PRIV(ucd_stage2)[];  extern const pcre_uint16 PRIV(ucd_stage2)[];
2773  extern const pcre_uint32 PRIV(ucp_gentype)[];  extern const pcre_uint32 PRIV(ucp_gentype)[];
2774  extern const pcre_uint32 PRIV(ucp_gbtable)[];  extern const pcre_uint32 PRIV(ucp_gbtable)[];
2775    #ifdef COMPILE_PCRE32
2776    extern const ucd_record  PRIV(dummy_ucd_record)[];
2777    #endif
2778  #ifdef SUPPORT_JIT  #ifdef SUPPORT_JIT
2779  extern const int         PRIV(ucp_typerange)[];  extern const int         PRIV(ucp_typerange)[];
2780  #endif  #endif
# Line 2738  extern const int         PRIV(ucp_typera Line 2783  extern const int         PRIV(ucp_typera
2783  /* UCD access macros */  /* UCD access macros */
2784    
2785  #define UCD_BLOCK_SIZE 128  #define UCD_BLOCK_SIZE 128
2786  #define GET_UCD(ch) (PRIV(ucd_records) + \  #define REAL_GET_UCD(ch) (PRIV(ucd_records) + \
2787          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(int)(ch) / UCD_BLOCK_SIZE] * \          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(int)(ch) / UCD_BLOCK_SIZE] * \
2788          UCD_BLOCK_SIZE + (int)(ch) % UCD_BLOCK_SIZE])          UCD_BLOCK_SIZE + (int)(ch) % UCD_BLOCK_SIZE])
2789    
2790    #ifdef COMPILE_PCRE32
2791    #define GET_UCD(ch) ((ch > 0x10ffff)? PRIV(dummy_ucd_record) : REAL_GET_UCD(ch))
2792    #else
2793    #define GET_UCD(ch) REAL_GET_UCD(ch)
2794    #endif
2795    
2796  #define UCD_CHARTYPE(ch)    GET_UCD(ch)->chartype  #define UCD_CHARTYPE(ch)    GET_UCD(ch)->chartype
2797  #define UCD_SCRIPT(ch)      GET_UCD(ch)->script  #define UCD_SCRIPT(ch)      GET_UCD(ch)->script

Legend:
Removed from v.1197  
changed lines
  Added in v.1688

  ViewVC Help
Powered by ViewVC 1.1.5