/[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 1248 by ph10, Wed Feb 13 17:36:38 2013 UTC revision 1454 by ph10, Sun Feb 9 18:55:03 2014 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) 1997-2014 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 308  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 322  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 574  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 592  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 768  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 887  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 972  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))  
   
 /* 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)++))  
   
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
# Line 1121  other. NOTE: The values also appear in p Line 1068  other. NOTE: The values also appear in p
1068    
1069    
1070  /* Private flags containing information about the compiled regex. They used to  /* Private flags containing information about the compiled regex. They used to
1071  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
1072  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
1073  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
1074    restrictions on partial matching have been lifted. It remains for backwards
1075  compatibility. */  compatibility. */
1076    
1077  #define PCRE_MODE8         0x0001  /* compiled in 8 bit mode */  #define PCRE_MODE8         0x00000001  /* compiled in 8 bit mode */
1078  #define PCRE_MODE16        0x0002  /* compiled in 16 bit mode */  #define PCRE_MODE16        0x00000002  /* compiled in 16 bit mode */
1079  #define PCRE_MODE32        0x0004  /* compiled in 32 bit mode */  #define PCRE_MODE32        0x00000004  /* compiled in 32 bit mode */
1080  #define PCRE_FIRSTSET      0x0010  /* first_char is set */  #define PCRE_FIRSTSET      0x00000010  /* first_char is set */
1081  #define PCRE_FCH_CASELESS  0x0020  /* caseless first char */  #define PCRE_FCH_CASELESS  0x00000020  /* caseless first char */
1082  #define PCRE_REQCHSET      0x0040  /* req_byte is set */  #define PCRE_REQCHSET      0x00000040  /* req_byte is set */
1083  #define PCRE_RCH_CASELESS  0x0080  /* caseless requested char */  #define PCRE_RCH_CASELESS  0x00000080  /* caseless requested char */
1084  #define PCRE_STARTLINE     0x0100  /* start after \n for multiline */  #define PCRE_STARTLINE     0x00000100  /* start after \n for multiline */
1085  #define PCRE_NOPARTIAL     0x0200  /* can't use partial with this regex */  #define PCRE_NOPARTIAL     0x00000200  /* can't use partial with this regex */
1086  #define PCRE_JCHANGED      0x0400  /* j option used in regex */  #define PCRE_JCHANGED      0x00000400  /* j option used in regex */
1087  #define PCRE_HASCRORLF     0x0800  /* explicit \r or \n in pattern */  #define PCRE_HASCRORLF     0x00000800  /* explicit \r or \n in pattern */
1088  #define PCRE_HASTHEN       0x1000  /* pattern contains (*THEN) */  #define PCRE_HASTHEN       0x00001000  /* pattern contains (*THEN) */
1089    #define PCRE_MLSET         0x00002000  /* match limit set by regex */
1090    #define PCRE_RLSET         0x00004000  /* recursion limit set by regex */
1091    #define PCRE_MATCH_EMPTY   0x00008000  /* pattern can match empty string */
1092    
1093  #if defined COMPILE_PCRE8  #if defined COMPILE_PCRE8
1094  #define PCRE_MODE          PCRE_MODE8  #define PCRE_MODE          PCRE_MODE8
# Line 1162  time, run time, or study time, respectiv Line 1113  time, run time, or study time, respectiv
1113  #define PUBLIC_COMPILE_OPTIONS \  #define PUBLIC_COMPILE_OPTIONS \
1114    (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \    (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
1115     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
1116     PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \     PCRE_NO_AUTO_CAPTURE|PCRE_NO_AUTO_POSSESS| \
1117       PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
1118     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \
1119     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE)     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE|PCRE_NEVER_UTF)
1120    
1121  #define PUBLIC_EXEC_OPTIONS \  #define PUBLIC_EXEC_OPTIONS \
1122    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \
# Line 1520  a positive value. */ Line 1472  a positive value. */
1472  #define STRING_xdigit               "xdigit"  #define STRING_xdigit               "xdigit"
1473    
1474  #define STRING_DEFINE               "DEFINE"  #define STRING_DEFINE               "DEFINE"
1475    #define STRING_WEIRD_STARTWORD      "[:<:]]"
1476    #define STRING_WEIRD_ENDWORD        "[:>:]]"
1477    
1478  #define STRING_CR_RIGHTPAR             "CR)"  #define STRING_CR_RIGHTPAR              "CR)"
1479  #define STRING_LF_RIGHTPAR             "LF)"  #define STRING_LF_RIGHTPAR              "LF)"
1480  #define STRING_CRLF_RIGHTPAR           "CRLF)"  #define STRING_CRLF_RIGHTPAR            "CRLF)"
1481  #define STRING_ANY_RIGHTPAR            "ANY)"  #define STRING_ANY_RIGHTPAR             "ANY)"
1482  #define STRING_ANYCRLF_RIGHTPAR        "ANYCRLF)"  #define STRING_ANYCRLF_RIGHTPAR         "ANYCRLF)"
1483  #define STRING_BSR_ANYCRLF_RIGHTPAR    "BSR_ANYCRLF)"  #define STRING_BSR_ANYCRLF_RIGHTPAR     "BSR_ANYCRLF)"
1484  #define STRING_BSR_UNICODE_RIGHTPAR    "BSR_UNICODE)"  #define STRING_BSR_UNICODE_RIGHTPAR     "BSR_UNICODE)"
1485  #define STRING_UTF8_RIGHTPAR           "UTF8)"  #define STRING_UTF8_RIGHTPAR            "UTF8)"
1486  #define STRING_UTF16_RIGHTPAR          "UTF16)"  #define STRING_UTF16_RIGHTPAR           "UTF16)"
1487  #define STRING_UTF32_RIGHTPAR          "UTF32)"  #define STRING_UTF32_RIGHTPAR           "UTF32)"
1488  #define STRING_UTF_RIGHTPAR            "UTF)"  #define STRING_UTF_RIGHTPAR             "UTF)"
1489  #define STRING_UCP_RIGHTPAR            "UCP)"  #define STRING_UCP_RIGHTPAR             "UCP)"
1490  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"  #define STRING_NO_AUTO_POSSESS_RIGHTPAR "NO_AUTO_POSSESS)"
1491    #define STRING_NO_START_OPT_RIGHTPAR    "NO_START_OPT)"
1492    #define STRING_LIMIT_MATCH_EQ           "LIMIT_MATCH="
1493    #define STRING_LIMIT_RECURSION_EQ       "LIMIT_RECURSION="
1494    
1495  #else  /* SUPPORT_UTF */  #else  /* SUPPORT_UTF */
1496    
# Line 1781  only. */ Line 1738  only. */
1738  #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
1739    
1740  #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
1741    #define STRING_WEIRD_STARTWORD      STR_LEFT_SQUARE_BRACKET STR_COLON STR_LESS_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
1742    #define STRING_WEIRD_ENDWORD        STR_LEFT_SQUARE_BRACKET STR_COLON STR_GREATER_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
1743    
1744  #define STRING_CR_RIGHTPAR             STR_C STR_R STR_RIGHT_PARENTHESIS  #define STRING_CR_RIGHTPAR              STR_C STR_R STR_RIGHT_PARENTHESIS
1745  #define STRING_LF_RIGHTPAR             STR_L STR_F STR_RIGHT_PARENTHESIS  #define STRING_LF_RIGHTPAR              STR_L STR_F STR_RIGHT_PARENTHESIS
1746  #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
1747  #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
1748  #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
1749  #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
1750  #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
1751  #define STRING_UTF8_RIGHTPAR           STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS  #define STRING_UTF8_RIGHTPAR            STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS
1752  #define STRING_UTF16_RIGHTPAR          STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS  #define STRING_UTF16_RIGHTPAR           STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS
1753  #define STRING_UTF32_RIGHTPAR          STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS  #define STRING_UTF32_RIGHTPAR           STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS
1754  #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_RIGHT_PARENTHESIS  #define STRING_UTF_RIGHTPAR             STR_U STR_T STR_F STR_RIGHT_PARENTHESIS
1755  #define STRING_UCP_RIGHTPAR            STR_U STR_C STR_P STR_RIGHT_PARENTHESIS  #define STRING_UCP_RIGHTPAR             STR_U STR_C STR_P STR_RIGHT_PARENTHESIS
1756  #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  #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
1757    #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
1758    #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
1759    #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
1760    
1761  #endif  /* SUPPORT_UTF */  #endif  /* SUPPORT_UTF */
1762    
# Line 1835  only. */ Line 1797  only. */
1797  #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 */
1798  #define PT_WORD       8    /* Word - L plus N plus underscore */  #define PT_WORD       8    /* Word - L plus N plus underscore */
1799  #define PT_CLIST      9    /* Pseudo-property: match character list */  #define PT_CLIST      9    /* Pseudo-property: match character list */
1800    #define PT_UCNC      10    /* Universal Character nameable character */
1801    #define PT_TABSIZE   11    /* Size of square table for autopossessify tests */
1802    
1803    /* The following special properties are used only in XCLASS items, when POSIX
1804    classes are specified and PCRE_UCP is set - in other words, for Unicode
1805    handling of these classes. They are not available via the \p or \P escapes like
1806    those in the above list, and so they do not take part in the autopossessifying
1807    table. */
1808    
1809    #define PT_PXGRAPH   11    /* [:graph:] - characters that mark the paper */
1810    #define PT_PXPRINT   12    /* [:print:] - [:graph:] plus non-control spaces */
1811    #define PT_PXPUNCT   13    /* [:punct:] - punctuation characters */
1812    
1813  /* 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
1814  contain characters with values greater than 255. */  contain characters with values greater than 255. */
1815    
1816  #define XCL_NOT    0x01    /* Flag: this is a negative class */  #define XCL_NOT       0x01    /* Flag: this is a negative class */
1817  #define XCL_MAP    0x02    /* Flag: a 32-byte map is present */  #define XCL_MAP       0x02    /* Flag: a 32-byte map is present */
1818    #define XCL_HASPROP   0x04    /* Flag: property checks are present. */
1819    
1820  #define XCL_END       0    /* Marks end of individual items */  #define XCL_END       0    /* Marks end of individual items */
1821  #define XCL_SINGLE    1    /* Single item (one multibyte char) follows */  #define XCL_SINGLE    1    /* Single item (one multibyte char) follows */
# Line 1849  contain characters with values greater t Line 1824  contain characters with values greater t
1824  #define XCL_NOTPROP   4    /* Unicode inverted property (ditto) */  #define XCL_NOTPROP   4    /* Unicode inverted property (ditto) */
1825    
1826  /* 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
1827  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
1828  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
1829  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
1830  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
1831  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
1832  non-DOTALL mode, "." behaves like \N.  non-DOTALL mode, "." behaves like \N.
# Line 1874  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, E Line 1849  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, E
1849         ESC_E, ESC_Q, ESC_g, ESC_k,         ESC_E, ESC_Q, ESC_g, ESC_k,
1850         ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu };         ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu };
1851    
 /* 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.  
1852    
1853  *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions  /********************** Opcode definitions ******************/
1854  that follow must also be updated to match. There are also tables called  
1855  "coptable" and "poptable" in pcre_dfa_exec.c that must be updated. */  /****** NOTE NOTE NOTE ******
1856    
1857    Starting from 1 (i.e. after OP_END), the values up to OP_EOD must correspond in
1858    order to the list of escapes immediately above. Furthermore, values up to
1859    OP_DOLLM must not be changed without adjusting the table called autoposstab in
1860    pcre_compile.c
1861    
1862    Whenever this list is updated, the two macro definitions that follow must be
1863    updated to match. The possessification table called "opcode_possessify" in
1864    pcre_compile.c must also be updated, and also the tables called "coptable"
1865    and "poptable" in pcre_dfa_exec.c.
1866    
1867    ****** NOTE NOTE NOTE ******/
1868    
1869    
1870    /* The values between FIRST_AUTOTAB_OP and LAST_AUTOTAB_RIGHT_OP, inclusive,
1871    are used in a table for deciding whether a repeated character type can be
1872    auto-possessified. */
1873    
1874    #define FIRST_AUTOTAB_OP       OP_NOT_DIGIT
1875    #define LAST_AUTOTAB_LEFT_OP   OP_EXTUNI
1876    #define LAST_AUTOTAB_RIGHT_OP  OP_DOLLM
1877    
1878  enum {  enum {
1879    OP_END,            /* 0 End of pattern */    OP_END,            /* 0 End of pattern */
# Line 1912  enum { Line 1906  enum {
1906    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) */
1907    OP_EOD,            /* 24 End of data (\z) */    OP_EOD,            /* 24 End of data (\z) */
1908    
1909    OP_CIRC,           /* 25 Start of line - not multiline */    /* Line end assertions */
1910    OP_CIRCM,          /* 26 Start of line - multiline */  
1911    OP_DOLL,           /* 27 End of line - not multiline */    OP_DOLL,           /* 25 End of line - not multiline */
1912    OP_DOLLM,          /* 28 End of line - multiline */    OP_DOLLM,          /* 26 End of line - multiline */
1913      OP_CIRC,           /* 27 Start of line - not multiline */
1914      OP_CIRCM,          /* 28 Start of line - multiline */
1915    
1916      /* Single characters; caseful must precede the caseless ones */
1917    
1918    OP_CHAR,           /* 29 Match one character, casefully */    OP_CHAR,           /* 29 Match one character, casefully */
1919    OP_CHARI,          /* 30 Match one character, caselessly */    OP_CHARI,          /* 30 Match one character, caselessly */
1920    OP_NOT,            /* 31 Match one character, not the given one, casefully */    OP_NOT,            /* 31 Match one character, not the given one, casefully */
# Line 1924  enum { Line 1923  enum {
1923    /* 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
1924    the offset from the first one is used to generate the others. */    the offset from the first one is used to generate the others. */
1925    
1926    /**** Single characters, caseful, must precede the caseless ones ****/    /* Repeated characters; caseful must precede the caseless ones */
1927    
1928    OP_STAR,           /* 33 The maximizing and minimizing versions of */    OP_STAR,           /* 33 The maximizing and minimizing versions of */
1929    OP_MINSTAR,        /* 34 these six opcodes must come in pairs, with */    OP_MINSTAR,        /* 34 these six opcodes must come in pairs, with */
# Line 1942  enum { Line 1941  enum {
1941    OP_POSQUERY,       /* 44 Posesssified query, caseful */    OP_POSQUERY,       /* 44 Posesssified query, caseful */
1942    OP_POSUPTO,        /* 45 Possessified upto, caseful */    OP_POSUPTO,        /* 45 Possessified upto, caseful */
1943    
1944    /**** Single characters, caseless, must follow the caseful ones */    /* Repeated characters; caseless must follow the caseful ones */
1945    
1946    OP_STARI,          /* 46 */    OP_STARI,          /* 46 */
1947    OP_MINSTARI,       /* 47 */    OP_MINSTARI,       /* 47 */
# Line 1960  enum { Line 1959  enum {
1959    OP_POSQUERYI,      /* 57 Posesssified query, caseless */    OP_POSQUERYI,      /* 57 Posesssified query, caseless */
1960    OP_POSUPTOI,       /* 58 Possessified upto, caseless */    OP_POSUPTOI,       /* 58 Possessified upto, caseless */
1961    
1962    /**** The negated ones must follow the non-negated ones, and match them ****/    /* The negated ones must follow the non-negated ones, and match them */
1963    /**** Negated single character, caseful; must precede the caseless ones ****/    /* Negated repeated character, caseful; must precede the caseless ones */
1964    
1965    OP_NOTSTAR,        /* 59 The maximizing and minimizing versions of */    OP_NOTSTAR,        /* 59 The maximizing and minimizing versions of */
1966    OP_NOTMINSTAR,     /* 60 these six opcodes must come in pairs, with */    OP_NOTMINSTAR,     /* 60 these six opcodes must come in pairs, with */
# Line 1979  enum { Line 1978  enum {
1978    OP_NOTPOSQUERY,    /* 70 */    OP_NOTPOSQUERY,    /* 70 */
1979    OP_NOTPOSUPTO,     /* 71 */    OP_NOTPOSUPTO,     /* 71 */
1980    
1981    /**** Negated single character, caseless; must follow the caseful ones ****/    /* Negated repeated character, caseless; must follow the caseful ones */
1982    
1983    OP_NOTSTARI,       /* 72 */    OP_NOTSTARI,       /* 72 */
1984    OP_NOTMINSTARI,    /* 73 */    OP_NOTMINSTARI,    /* 73 */
# Line 1997  enum { Line 1996  enum {
1996    OP_NOTPOSQUERYI,   /* 83 */    OP_NOTPOSQUERYI,   /* 83 */
1997    OP_NOTPOSUPTOI,    /* 84 */    OP_NOTPOSUPTOI,    /* 84 */
1998    
1999    /**** Character types ****/    /* Character types */
2000    
2001    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */
2002    OP_TYPEMINSTAR,    /* 86 these six opcodes must come in pairs, with */    OP_TYPEMINSTAR,    /* 86 these six opcodes must come in pairs, with */
# Line 2028  enum { Line 2027  enum {
2027    OP_CRRANGE,        /* 104 These are different to the three sets above. */    OP_CRRANGE,        /* 104 These are different to the three sets above. */
2028    OP_CRMINRANGE,     /* 105 */    OP_CRMINRANGE,     /* 105 */
2029    
2030      OP_CRPOSSTAR,      /* 106 Possessified versions */
2031      OP_CRPOSPLUS,      /* 107 */
2032      OP_CRPOSQUERY,     /* 108 */
2033      OP_CRPOSRANGE,     /* 109 */
2034    
2035    /* End of quantifier opcodes */    /* End of quantifier opcodes */
2036    
2037    OP_CLASS,          /* 106 Match a character class, chars < 256 only */    OP_CLASS,          /* 110 Match a character class, chars < 256 only */
2038    OP_NCLASS,         /* 107 Same, but the bitmap was created from a negative    OP_NCLASS,         /* 111 Same, but the bitmap was created from a negative
2039                                class - the difference is relevant only when a                                class - the difference is relevant only when a
2040                                character > 255 is encountered. */                                character > 255 is encountered. */
2041    OP_XCLASS,         /* 108 Extended class for handling > 255 chars within the    OP_XCLASS,         /* 112 Extended class for handling > 255 chars within the
2042                                class. This does both positive and negative. */                                class. This does both positive and negative. */
2043    OP_REF,            /* 109 Match a back reference, casefully */    OP_REF,            /* 113 Match a back reference, casefully */
2044    OP_REFI,           /* 110 Match a back reference, caselessly */    OP_REFI,           /* 114 Match a back reference, caselessly */
2045    OP_RECURSE,        /* 111 Match a numbered subpattern (possibly recursive) */    OP_DNREF,          /* 115 Match a duplicate name backref, casefully */
2046    OP_CALLOUT,        /* 112 Call out to external function if provided */    OP_DNREFI,         /* 116 Match a duplicate name backref, caselessly */
2047      OP_RECURSE,        /* 117 Match a numbered subpattern (possibly recursive) */
2048    OP_ALT,            /* 113 Start of alternation */    OP_CALLOUT,        /* 118 Call out to external function if provided */
2049    OP_KET,            /* 114 End of group that doesn't have an unbounded repeat */  
2050    OP_KETRMAX,        /* 115 These two must remain together and in this */    OP_ALT,            /* 119 Start of alternation */
2051    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 */
2052    OP_KETRPOS,        /* 117 Possessive unlimited repeat. */    OP_KETRMAX,        /* 121 These two must remain together and in this */
2053      OP_KETRMIN,        /* 122 order. They are for groups the repeat for ever. */
2054      OP_KETRPOS,        /* 123 Possessive unlimited repeat. */
2055    
2056    /* 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
2057    asserts must remain in order. */    asserts must remain in order. */
2058    
2059    OP_REVERSE,        /* 118 Move pointer back - used in lookbehind assertions */    OP_REVERSE,        /* 124 Move pointer back - used in lookbehind assertions */
2060    OP_ASSERT,         /* 119 Positive lookahead */    OP_ASSERT,         /* 125 Positive lookahead */
2061    OP_ASSERT_NOT,     /* 120 Negative lookahead */    OP_ASSERT_NOT,     /* 126 Negative lookahead */
2062    OP_ASSERTBACK,     /* 121 Positive lookbehind */    OP_ASSERTBACK,     /* 127 Positive lookbehind */
2063    OP_ASSERTBACK_NOT, /* 122 Negative lookbehind */    OP_ASSERTBACK_NOT, /* 128 Negative lookbehind */
2064    
2065    /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately    /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately
2066    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
2067    subpattern that isn't an assertion. The POS versions must immediately follow    subpattern that isn't an assertion. The POS versions must immediately follow
2068    the non-POS versions in each case. */    the non-POS versions in each case. */
2069    
2070    OP_ONCE,           /* 123 Atomic group, contains captures */    OP_ONCE,           /* 129 Atomic group, contains captures */
2071    OP_ONCE_NC,        /* 124 Atomic group containing no captures */    OP_ONCE_NC,        /* 130 Atomic group containing no captures */
2072    OP_BRA,            /* 125 Start of non-capturing bracket */    OP_BRA,            /* 131 Start of non-capturing bracket */
2073    OP_BRAPOS,         /* 126 Ditto, with unlimited, possessive repeat */    OP_BRAPOS,         /* 132 Ditto, with unlimited, possessive repeat */
2074    OP_CBRA,           /* 127 Start of capturing bracket */    OP_CBRA,           /* 133 Start of capturing bracket */
2075    OP_CBRAPOS,        /* 128 Ditto, with unlimited, possessive repeat */    OP_CBRAPOS,        /* 134 Ditto, with unlimited, possessive repeat */
2076    OP_COND,           /* 129 Conditional group */    OP_COND,           /* 135 Conditional group */
2077    
2078    /* 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
2079    check for >= SBRA to distinguish the two sets. */    check for >= SBRA to distinguish the two sets. */
2080    
2081    OP_SBRA,           /* 130 Start of non-capturing bracket, check empty  */    OP_SBRA,           /* 136 Start of non-capturing bracket, check empty  */
2082    OP_SBRAPOS,        /* 131 Ditto, with unlimited, possessive repeat */    OP_SBRAPOS,        /* 137 Ditto, with unlimited, possessive repeat */
2083    OP_SCBRA,          /* 132 Start of capturing bracket, check empty */    OP_SCBRA,          /* 138 Start of capturing bracket, check empty */
2084    OP_SCBRAPOS,       /* 133 Ditto, with unlimited, possessive repeat */    OP_SCBRAPOS,       /* 139 Ditto, with unlimited, possessive repeat */
2085    OP_SCOND,          /* 134 Conditional group, check empty */    OP_SCOND,          /* 140 Conditional group, check empty */
2086    
2087    /* The next two pairs must (respectively) be kept together. */    /* The next two pairs must (respectively) be kept together. */
2088    
2089    OP_CREF,           /* 135 Used to hold a capture number as condition */    OP_CREF,           /* 141 Used to hold a capture number as condition */
2090    OP_NCREF,          /* 136 Same, but generated by a name reference*/    OP_DNCREF,         /* 142 Used to point to duplicate names as a condition */
2091    OP_RREF,           /* 137 Used to hold a recursion number as condition */    OP_RREF,           /* 143 Used to hold a recursion number as condition */
2092    OP_NRREF,          /* 138 Same, but generated by a name reference*/    OP_DNRREF,         /* 144 Used to point to duplicate names as a condition */
2093    OP_DEF,            /* 139 The DEFINE condition */    OP_DEF,            /* 145 The DEFINE condition */
2094    
2095    OP_BRAZERO,        /* 140 These two must remain together and in this */    OP_BRAZERO,        /* 146 These two must remain together and in this */
2096    OP_BRAMINZERO,     /* 141 order. */    OP_BRAMINZERO,     /* 147 order. */
2097    OP_BRAPOSZERO,     /* 142 */    OP_BRAPOSZERO,     /* 148 */
2098    
2099    /* These are backtracking control verbs */    /* These are backtracking control verbs */
2100    
2101    OP_MARK,           /* 143 always has an argument */    OP_MARK,           /* 149 always has an argument */
2102    OP_PRUNE,          /* 144 */    OP_PRUNE,          /* 150 */
2103    OP_PRUNE_ARG,      /* 145 same, but with argument */    OP_PRUNE_ARG,      /* 151 same, but with argument */
2104    OP_SKIP,           /* 146 */    OP_SKIP,           /* 152 */
2105    OP_SKIP_ARG,       /* 147 same, but with argument */    OP_SKIP_ARG,       /* 153 same, but with argument */
2106    OP_THEN,           /* 148 */    OP_THEN,           /* 154 */
2107    OP_THEN_ARG,       /* 149 same, but with argument */    OP_THEN_ARG,       /* 155 same, but with argument */
2108    OP_COMMIT,         /* 150 */    OP_COMMIT,         /* 156 */
2109    
2110    /* These are forced failure and success verbs */    /* These are forced failure and success verbs */
2111    
2112    OP_FAIL,           /* 151 */    OP_FAIL,           /* 157 */
2113    OP_ACCEPT,         /* 152 */    OP_ACCEPT,         /* 158 */
2114    OP_ASSERT_ACCEPT,  /* 153 Used inside assertions */    OP_ASSERT_ACCEPT,  /* 159 Used inside assertions */
2115    OP_CLOSE,          /* 154 Used before OP_ACCEPT to close open captures */    OP_CLOSE,          /* 160 Used before OP_ACCEPT to close open captures */
2116    
2117    /* This is used to skip a subpattern with a {0} quantifier */    /* This is used to skip a subpattern with a {0} quantifier */
2118    
2119    OP_SKIPZERO,       /* 155 */    OP_SKIPZERO,       /* 161 */
2120    
2121    /* 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
2122    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 2121  enum { Line 2127  enum {
2127    
2128  /* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro  /* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro
2129  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
2130  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
2131    pcre_dfa_exec.c that must be updated. */
2132    
2133    
2134  /* 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 2134  some cases doesn't actually use these na Line 2141  some cases doesn't actually use these na
2141    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \
2142    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \
2143    "extuni",  "\\Z", "\\z",                                        \    "extuni",  "\\Z", "\\z",                                        \
2144    "^", "^", "$", "$", "char", "chari", "not", "noti",             \    "$", "$", "^", "^", "char", "chari", "not", "noti",             \
2145    "*", "*?", "+", "+?", "?", "??",                                \    "*", "*?", "+", "+?", "?", "??",                                \
2146    "{", "{", "{",                                                  \    "{", "{", "{",                                                  \
2147    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
# Line 2150  some cases doesn't actually use these na Line 2157  some cases doesn't actually use these na
2157    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \
2158    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
2159    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \
2160    "class", "nclass", "xclass", "Ref", "Refi",                     \    "*+","++", "?+", "{",                                           \
2161      "class", "nclass", "xclass", "Ref", "Refi", "DnRef", "DnRefi",  \
2162    "Recurse", "Callout",                                           \    "Recurse", "Callout",                                           \
2163    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \
2164    "Reverse", "Assert", "Assert not", "AssertB", "AssertB not",    \    "Reverse", "Assert", "Assert not", "AssertB", "AssertB not",    \
# Line 2159  some cases doesn't actually use these na Line 2167  some cases doesn't actually use these na
2167    "Cond",                                                         \    "Cond",                                                         \
2168    "SBra", "SBraPos", "SCBra", "SCBraPos",                         \    "SBra", "SBraPos", "SCBra", "SCBraPos",                         \
2169    "SCond",                                                        \    "SCond",                                                        \
2170    "Cond ref", "Cond nref", "Cond rec", "Cond nrec", "Cond def",   \    "Cond ref", "Cond dnref", "Cond rec", "Cond dnrec", "Cond def", \
2171    "Brazero", "Braminzero", "Braposzero",                          \    "Brazero", "Braminzero", "Braposzero",                          \
2172    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \
2173    "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \    "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \
# Line 2184  in UTF-8 mode. The code that uses this t Line 2192  in UTF-8 mode. The code that uses this t
2192    3, 3,                          /* \P, \p                                 */ \    3, 3,                          /* \P, \p                                 */ \
2193    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \
2194    1,                             /* \X                                     */ \    1,                             /* \X                                     */ \
2195    1, 1, 1, 1, 1, 1,              /* \Z, \z, ^, ^M, $, $M                   */ \    1, 1, 1, 1, 1, 1,              /* \Z, \z, $, $M ^, ^M                    */ \
2196    2,                             /* Char  - the minimum length             */ \    2,                             /* Char  - the minimum length             */ \
2197    2,                             /* Chari  - the minimum length            */ \    2,                             /* Chari  - the minimum length            */ \
2198    2,                             /* not                                    */ \    2,                             /* not                                    */ \
# Line 2215  in UTF-8 mode. The code that uses this t Line 2223  in UTF-8 mode. The code that uses this t
2223    /* Character class & ref repeats                                         */ \    /* Character class & ref repeats                                         */ \
2224    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \
2225    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \
2226      1, 1, 1, 1+2*IMM2_SIZE,        /* Possessive *+, ++, ?+, CRPOSRANGE      */ \
2227    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \
2228    1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \    1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \
2229    0,                             /* XCLASS - variable length               */ \    0,                             /* XCLASS - variable length               */ \
2230    1+IMM2_SIZE,                   /* REF                                    */ \    1+IMM2_SIZE,                   /* REF                                    */ \
2231    1+IMM2_SIZE,                   /* REFI                                   */ \    1+IMM2_SIZE,                   /* REFI                                   */ \
2232      1+2*IMM2_SIZE,                 /* DNREF                                  */ \
2233      1+2*IMM2_SIZE,                 /* DNREFI                                 */ \
2234    1+LINK_SIZE,                   /* RECURSE                                */ \    1+LINK_SIZE,                   /* RECURSE                                */ \
2235    2+2*LINK_SIZE,                 /* CALLOUT                                */ \    2+2*LINK_SIZE,                 /* CALLOUT                                */ \
2236    1+LINK_SIZE,                   /* Alt                                    */ \    1+LINK_SIZE,                   /* Alt                                    */ \
# Line 2244  in UTF-8 mode. The code that uses this t Line 2255  in UTF-8 mode. The code that uses this t
2255    1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \    1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \
2256    1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \    1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \
2257    1+LINK_SIZE,                   /* SCOND                                  */ \    1+LINK_SIZE,                   /* SCOND                                  */ \
2258    1+IMM2_SIZE, 1+IMM2_SIZE,      /* CREF, NCREF                            */ \    1+IMM2_SIZE, 1+2*IMM2_SIZE,    /* CREF, DNCREF                           */ \
2259    1+IMM2_SIZE, 1+IMM2_SIZE,      /* RREF, NRREF                            */ \    1+IMM2_SIZE, 1+2*IMM2_SIZE,    /* RREF, DNRREF                           */ \
2260    1,                             /* DEF                                    */ \    1,                             /* DEF                                    */ \
2261    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \
2262    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \
# Line 2254  in UTF-8 mode. The code that uses this t Line 2265  in UTF-8 mode. The code that uses this t
2265    1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \    1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \
2266    1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */    1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */
2267    
2268  /* 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. */  
2269    
2270  #define RREF_ANY  0xffff  #define RREF_ANY  0xffff
2271    
# Line 2270  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4, Line 2280  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,
2280         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
2281         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
2282         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,
2283         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERRCOUNT };         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERR78, ERR79,
2284           ERR80, ERR81, ERR82, ERR83, ERR84, ERR85, ERRCOUNT };
2285    
2286  /* JIT compiling modes. The function list is indexed by them. */  /* JIT compiling modes. The function list is indexed by them. */
2287    
2288  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,
2289         JIT_NUMBER_OF_COMPILE_MODES };         JIT_NUMBER_OF_COMPILE_MODES };
2290    
# Line 2280  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COM Line 2292  enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COM
2292  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
2293  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
2294  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
2295  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  
 */  
2296    
2297  #if defined COMPILE_PCRE8  The size of the structure must be a multiple of 8 bytes. For the case of
2298  #define REAL_PCRE real_pcre  compiled-on-4 and run-on-8, we include an extra pointer that is always NULL so
2299  #elif defined COMPILE_PCRE16  that there are an even number of pointers which therefore are a multiple of 8
2300  #define REAL_PCRE real_pcre16  bytes.
2301  #elif defined COMPILE_PCRE32  
2302  #define REAL_PCRE real_pcre32  It is necessary to fork the struct for the 32 bit library, since it needs to
2303  #endif  use pcre_uint32 for first_char and req_char. We can't put an ifdef inside the
2304    typedef because pcretest needs access to the struct of the 8-, 16- and 32-bit
2305  /* It is necessary to fork the struct for 32 bit, since it needs to use  variants.
2306   * pcre_uchar for first_char and req_char. Can't put an ifdef inside the  
2307   * typedef since pcretest needs access to  the struct of the 8-, 16-  *** WARNING ***
2308   * and 32-bit variants. */  When new fields are added to these structures, remember to adjust the code in
2309    pcre_byte_order.c that is concerned with swapping the byte order of the fields
2310    when a compiled regex is reloaded on a host with different endianness.
2311    *** WARNING ***
2312    There is also similar byte-flipping code in pcretest.c, which is used for
2313    testing the byte-flipping features. It must also be kept in step.
2314    *** WARNING ***
2315    */
2316    
2317  typedef struct real_pcre8_or_16 {  typedef struct real_pcre8_or_16 {
2318    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2319    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2320    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2321    pcre_uint16 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2322      pcre_uint32 limit_match;        /* Limit set from regex */
2323      pcre_uint32 limit_recursion;    /* Limit set from regex */
2324      pcre_uint16 first_char;         /* Starting character */
2325      pcre_uint16 req_char;           /* This character must be seen */
2326    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2327    pcre_uint16 top_bracket;        /* Highest numbered group */    pcre_uint16 top_bracket;        /* Highest numbered group */
2328    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 */  
2329    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2330    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2331    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2332    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2333      pcre_uint16 dummy1;             /* To ensure size is a multiple of 8 */
2334      pcre_uint16 dummy2;             /* To ensure size is a multiple of 8 */
2335      pcre_uint16 dummy3;             /* To ensure size is a multiple of 8 */
2336    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2337    const pcre_uint8 *nullpad;      /* NULL padding */    void             *nullpad;      /* NULL padding */
2338  } real_pcre8_or_16;  } real_pcre8_or_16;
2339    
2340  typedef struct real_pcre8_or_16 real_pcre;  typedef struct real_pcre8_or_16 real_pcre;
# Line 2331  typedef struct real_pcre32 { Line 2344  typedef struct real_pcre32 {
2344    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2345    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2346    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2347    pcre_uint16 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2348      pcre_uint32 limit_match;        /* Limit set from regex */
2349      pcre_uint32 limit_recursion;    /* Limit set from regex */
2350      pcre_uint32 first_char;         /* Starting character */
2351      pcre_uint32 req_char;           /* This character must be seen */
2352    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2353    pcre_uint16 top_bracket;        /* Highest numbered group */    pcre_uint16 top_bracket;        /* Highest numbered group */
2354    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 */  
2355    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2356    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2357    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2358    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2359    pcre_uint16 dummy1;             /* for later expansion */    pcre_uint16 dummy;              /* To ensure size is a multiple of 8 */
   pcre_uint16 dummy2;             /* for later expansion */  
2360    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */    const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2361    void *nullpad;                  /* for later expansion */    void             *nullpad;      /* NULL padding */
2362  } real_pcre32;  } real_pcre32;
2363    
2364    #if defined COMPILE_PCRE8
2365    #define REAL_PCRE real_pcre
2366    #elif defined COMPILE_PCRE16
2367    #define REAL_PCRE real_pcre16
2368    #elif defined COMPILE_PCRE32
2369    #define REAL_PCRE real_pcre32
2370    #endif
2371    
2372  /* Assert that the size of REAL_PCRE is divisible by 8 */  /* Assert that the size of REAL_PCRE is divisible by 8 */
2373  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];
2374    
# Line 2380  typedef struct open_capitem { Line 2402  typedef struct open_capitem {
2402    pcre_uint16 flag;             /* Set TRUE if recursive back ref */    pcre_uint16 flag;             /* Set TRUE if recursive back ref */
2403  } open_capitem;  } open_capitem;
2404    
2405    /* Structure for building a list of named groups during the first pass of
2406    compiling. */
2407    
2408    typedef struct named_group {
2409      const pcre_uchar  *name;          /* Points to the name in the pattern */
2410      int                length;        /* Length of the name */
2411      pcre_uint32        number;        /* Group number */
2412    } named_group;
2413    
2414  /* Structure for passing "static" information around between the functions  /* Structure for passing "static" information around between the functions
2415  doing the compiling, so that they are thread-safe. */  doing the compiling, so that they are thread-safe. */
2416    
# Line 2392  typedef struct compile_data { Line 2423  typedef struct compile_data {
2423    const pcre_uchar *start_code;     /* The start of the compiled code */    const pcre_uchar *start_code;     /* The start of the compiled code */
2424    const pcre_uchar *start_pattern;  /* The start of the pattern */    const pcre_uchar *start_pattern;  /* The start of the pattern */
2425    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 */  
2426    pcre_uchar *hwm;                  /* High watermark of workspace */    pcre_uchar *hwm;                  /* High watermark of workspace */
2427      open_capitem *open_caps;          /* Chain of open capture items */
2428      named_group *named_groups;        /* Points to vector in pre-compile */
2429    pcre_uchar *name_table;           /* The name/number table */    pcre_uchar *name_table;           /* The name/number table */
2430    int  names_found;                 /* Number of entries so far */    int  names_found;                 /* Number of entries so far */
2431    int  name_entry_size;             /* Size of each entry */    int  name_entry_size;             /* Size of each entry */
2432      int  named_group_list_size;       /* Number of entries in the list */
2433    int  workspace_size;              /* Size of workspace */    int  workspace_size;              /* Size of workspace */
2434    unsigned int  bracount;           /* Count of capturing parens as we compile */    unsigned int bracount;            /* Count of capturing parens as we compile */
2435    int  final_bracount;              /* Saved value after first pass */    int  final_bracount;              /* Saved value after first pass */
2436    int  max_lookbehind;              /* Maximum lookbehind (characters) */    int  max_lookbehind;              /* Maximum lookbehind (characters) */
2437    int  top_backref;                 /* Maximum back reference */    int  top_backref;                 /* Maximum back reference */
2438    unsigned int backref_map;         /* Bitmap of low back refs */    unsigned int backref_map;         /* Bitmap of low back refs */
2439      unsigned int namedrefcount;       /* Number of backreferences by name */
2440      int  parens_depth;                /* Depth of nested parentheses */
2441    int  assert_depth;                /* Depth of nested assertions */    int  assert_depth;                /* Depth of nested assertions */
2442    int  external_options;            /* External (initial) options */    pcre_uint32 external_options;     /* External (initial) options */
2443    int  external_flags;              /* External flag bits to be set */    pcre_uint32 external_flags;       /* External flag bits to be set */
2444    int  req_varyopt;                 /* "After variable item" flag for reqbyte */    int  req_varyopt;                 /* "After variable item" flag for reqbyte */
2445    BOOL had_accept;                  /* (*ACCEPT) encountered */    BOOL had_accept;                  /* (*ACCEPT) encountered */
2446    BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */    BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */
2447    BOOL check_lookbehind;            /* Lookbehinds need later checking */    BOOL check_lookbehind;            /* Lookbehinds need later checking */
2448      BOOL dupnames;                    /* Duplicate names exist */
2449    int  nltype;                      /* Newline type */    int  nltype;                      /* Newline type */
2450    int  nllen;                       /* Newline string length */    int  nllen;                       /* Newline string length */
2451    pcre_uchar nl[4];                 /* Newline string when fixed length */    pcre_uchar nl[4];                 /* Newline string when fixed length */
# Line 2431  typedef struct recursion_info { Line 2467  typedef struct recursion_info {
2467    unsigned int group_num;         /* Number of group that was called */    unsigned int group_num;         /* Number of group that was called */
2468    int *offset_save;               /* Pointer to start of saved offsets */    int *offset_save;               /* Pointer to start of saved offsets */
2469    int saved_max;                  /* Number of saved offsets */    int saved_max;                  /* Number of saved offsets */
2470    int saved_capture_last;         /* Last capture number */    int saved_capture_last;         /* Last capture number */
2471    PCRE_PUCHAR subject_position;   /* Position at start of recursion */    PCRE_PUCHAR subject_position;   /* Position at start of recursion */
2472  } recursion_info;  } recursion_info;
2473    
# Line 2468  typedef struct match_data { Line 2504  typedef struct match_data {
2504    int    nllen;                   /* Newline string length */    int    nllen;                   /* Newline string length */
2505    int    name_count;              /* Number of names in name table */    int    name_count;              /* Number of names in name table */
2506    int    name_entry_size;         /* Size of entry in names table */    int    name_entry_size;         /* Size of entry in names table */
2507      unsigned int skip_arg_count;    /* For counting SKIP_ARGs */
2508      unsigned int ignore_skip_arg;   /* For re-run when SKIP arg name not found */
2509    pcre_uchar *name_table;         /* Table of names */    pcre_uchar *name_table;         /* Table of names */
2510    pcre_uchar nl[4];               /* Newline string when fixed */    pcre_uchar nl[4];               /* Newline string when fixed */
2511    const  pcre_uint8 *lcc;         /* Points to lower casing table */    const  pcre_uint8 *lcc;         /* Points to lower casing table */
# Line 2484  typedef struct match_data { Line 2522  typedef struct match_data {
2522    BOOL   hitend;                  /* Hit the end of the subject at some point */    BOOL   hitend;                  /* Hit the end of the subject at some point */
2523    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */
2524    BOOL   hasthen;                 /* Pattern contains (*THEN) */    BOOL   hasthen;                 /* Pattern contains (*THEN) */
   BOOL   ignore_skip_arg;         /* For re-run when SKIP name not found */  
2525    const  pcre_uchar *start_code;  /* For use when recursing */    const  pcre_uchar *start_code;  /* For use when recursing */
2526    PCRE_PUCHAR start_subject;      /* Start of the subject string */    PCRE_PUCHAR start_subject;      /* Start of the subject string */
2527    PCRE_PUCHAR end_subject;        /* End of the subject string */    PCRE_PUCHAR end_subject;        /* End of the subject string */

Legend:
Removed from v.1248  
changed lines
  Added in v.1454

  ViewVC Help
Powered by ViewVC 1.1.5