/[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 576 by ph10, Sun Nov 21 18:45:10 2010 UTC revision 1145 by ph10, Fri Oct 19 16:25:32 2012 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) 1997-2010 University of Cambridge             Copyright (c) 1997-2012 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 40  POSSIBILITY OF SUCH DAMAGE. Line 40  POSSIBILITY OF SUCH DAMAGE.
40    
41  /* This header contains definitions that are shared between the different  /* This header contains definitions that are shared between the different
42  modules, but which are not relevant to the exported API. This includes some  modules, but which are not relevant to the exported API. This includes some
43  functions whose names all begin with "_pcre_". */  functions whose names all begin with "_pcre_", "_pcre16_" or "_pcre32_"
44    depending on the PRIV macro. */
45    
46  #ifndef PCRE_INTERNAL_H  #ifndef PCRE_INTERNAL_H
47  #define PCRE_INTERNAL_H  #define PCRE_INTERNAL_H
# Line 51  functions whose names all begin with "_p Line 52  functions whose names all begin with "_p
52  #define PCRE_DEBUG  #define PCRE_DEBUG
53  #endif  #endif
54    
55  /* We do not support both EBCDIC and UTF-8 at the same time. The "configure"  /* PCRE is compiled as an 8 bit library if it is not requested otherwise. */
 script prevents both being selected, but not everybody uses "configure". */  
56    
57  #if defined EBCDIC && defined SUPPORT_UTF8  #if !defined COMPILE_PCRE16 && !defined COMPILE_PCRE32
58  #error The use of both EBCDIC and SUPPORT_UTF8 is not supported.  #define COMPILE_PCRE8
59  #endif  #endif
60    
61  /* If SUPPORT_UCP is defined, SUPPORT_UTF8 must also be defined. The  /* If SUPPORT_UCP is defined, SUPPORT_UTF must also be defined. The
62  "configure" script ensures this, but not everybody uses "configure". */  "configure" script ensures this, but not everybody uses "configure". */
63    
64  #if defined SUPPORT_UCP && !defined SUPPORT_UTF8  #if defined SUPPORT_UCP && !(defined SUPPORT_UTF)
65    #define SUPPORT_UTF 1
66    #endif
67    
68    /* We define SUPPORT_UTF if SUPPORT_UTF8 is enabled for compatibility
69    reasons with existing code. */
70    
71    #if defined SUPPORT_UTF8 && !(defined SUPPORT_UTF)
72    #define SUPPORT_UTF 1
73    #endif
74    
75    /* Fixme: SUPPORT_UTF8 should be eventually disappear from the code.
76    Until then we define it if SUPPORT_UTF is defined. */
77    
78    #if defined SUPPORT_UTF && !(defined SUPPORT_UTF8)
79  #define SUPPORT_UTF8 1  #define SUPPORT_UTF8 1
80  #endif  #endif
81    
82    /* We do not support both EBCDIC and UTF-8/16/32 at the same time. The "configure"
83    script prevents both being selected, but not everybody uses "configure". */
84    
85    #if defined EBCDIC && defined SUPPORT_UTF
86    #error The use of both EBCDIC and SUPPORT_UTF is not supported.
87    #endif
88    
89  /* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef  /* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef
90  inline, and there are *still* stupid compilers about that don't like indented  inline, and there are *still* stupid compilers about that don't like indented
91  pre-processor statements, or at least there were when I first wrote this. After  pre-processor statements, or at least there were when I first wrote this. After
# Line 158  set, we ensure here that it has no effec Line 179  set, we ensure here that it has no effec
179  #define PCRE_CALL_CONVENTION  #define PCRE_CALL_CONVENTION
180  #endif  #endif
181    
182  /* We need to have types that specify unsigned 16-bit and 32-bit integers. We  /* We need to have types that specify unsigned 8, 16 and 32-bit integers. We
183  cannot determine these outside the compilation (e.g. by running a program as  cannot determine these outside the compilation (e.g. by running a program as
184  part of "configure") because PCRE is often cross-compiled for use on other  part of "configure") because PCRE is often cross-compiled for use on other
185  systems. Instead we make use of the maximum sizes that are available at  systems. Instead we make use of the maximum sizes that are available at
186  preprocessor time in standard C environments. */  preprocessor time in standard C environments. */
187    
188    typedef unsigned char pcre_uint8;
189    
190  #if USHRT_MAX == 65535  #if USHRT_MAX == 65535
191    typedef unsigned short pcre_uint16;    typedef unsigned short pcre_uint16;
192    typedef short pcre_int16;    typedef short pcre_int16;
# Line 206  by "configure". */ Line 229  by "configure". */
229    
230  /* All character handling must be done as unsigned characters. Otherwise there  /* All character handling must be done as unsigned characters. Otherwise there
231  are problems with top-bit-set characters and functions such as isspace().  are problems with top-bit-set characters and functions such as isspace().
232  However, we leave the interface to the outside world as char *, because that  However, we leave the interface to the outside world as char * or short *,
233  should make things easier for callers. We define a short type for unsigned char  because that should make things easier for callers. This character type is
234  to save lots of typing. I tried "uchar", but it causes problems on Digital  called pcre_uchar.
235  Unix, where it is defined in sys/types, so use "uschar" instead. */  
236    The IN_UCHARS macro multiply its argument with the byte size of the current
237    pcre_uchar type. Useful for memcpy and such operations, whose require the
238    byte size of their input/output buffers.
239    
240    The MAX_255 macro checks whether its pcre_uchar input is less than 256.
241    
242    The TABLE_GET macro is designed for accessing elements of tables whose contain
243    exactly 256 items. When the character is able to contain more than 256
244    items, some check is needed before accessing these tables.
245    */
246    
247    #if defined COMPILE_PCRE8
248    
249    typedef unsigned char pcre_uchar;
250    #define IN_UCHARS(x) (x)
251    #define MAX_255(c) 1
252    #define TABLE_GET(c, table, default) ((table)[c])
253    
254    #elif defined COMPILE_PCRE16
255    
256    #if USHRT_MAX != 65535
257    /* This is a warning message. Change PCRE_UCHAR16 to a 16 bit data type in
258    pcre.h(.in) and disable (comment out) this message. */
259    #error Warning: PCRE_UCHAR16 is not a 16 bit data type.
260    #endif
261    
262    typedef pcre_uint16 pcre_uchar;
263    #define UCHAR_SHIFT (1)
264    #define IN_UCHARS(x) ((x) << UCHAR_SHIFT)
265    #define MAX_255(c) ((c) <= 255u)
266    #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
267    
268    #elif defined COMPILE_PCRE32
269    
270    typedef pcre_uint32 pcre_uchar;
271    #define UCHAR_SHIFT (2)
272    #define IN_UCHARS(x) ((x) << UCHAR_SHIFT)
273    #define MAX_255(c) ((c) <= 255u)
274    #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
275    
276  typedef unsigned char uschar;  /* Assert that pcre_uchar32 is a 32-bit type */
277    typedef int __assert_pcre_uchar32_size[sizeof(pcre_uchar) == 4 ? 1 : -1];
278    
279    #else
280    #error Unsupported compiling mode
281    #endif /* COMPILE_PCRE[8|16|32] */
282    
283  /* This is an unsigned int value that no character can ever have. UTF-8  /* This is an unsigned int value that no character can ever have. UTF-8
284  characters only go up to 0x7fffffff (though Unicode doesn't go beyond  characters only go up to 0x7fffffff (though Unicode doesn't go beyond
# Line 234  start/end of string field names are. */ Line 301  start/end of string field names are. */
301  #define IS_NEWLINE(p) \  #define IS_NEWLINE(p) \
302    ((NLBLOCK->nltype != NLTYPE_FIXED)? \    ((NLBLOCK->nltype != NLTYPE_FIXED)? \
303      ((p) < NLBLOCK->PSEND && \      ((p) < NLBLOCK->PSEND && \
304       _pcre_is_newline((p), NLBLOCK->nltype, NLBLOCK->PSEND, &(NLBLOCK->nllen),\       PRIV(is_newline)((p), NLBLOCK->nltype, NLBLOCK->PSEND, \
305         utf8)) \         &(NLBLOCK->nllen), utf)) \
306      : \      : \
307      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
308       (p)[0] == NLBLOCK->nl[0] && \       RAWUCHARTEST(p) == NLBLOCK->nl[0] && \
309       (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || RAWUCHARTEST(p+1) == NLBLOCK->nl[1])       \
310      ) \      ) \
311    )    )
312    
# Line 248  start/end of string field names are. */ Line 315  start/end of string field names are. */
315  #define WAS_NEWLINE(p) \  #define WAS_NEWLINE(p) \
316    ((NLBLOCK->nltype != NLTYPE_FIXED)? \    ((NLBLOCK->nltype != NLTYPE_FIXED)? \
317      ((p) > NLBLOCK->PSSTART && \      ((p) > NLBLOCK->PSSTART && \
318       _pcre_was_newline((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \       PRIV(was_newline)((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \
319         &(NLBLOCK->nllen), utf8)) \         &(NLBLOCK->nllen), utf)) \
320      : \      : \
321      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
322       (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \       RAWUCHARTEST(p - NLBLOCK->nllen) == NLBLOCK->nl[0] &&              \
323       (NLBLOCK->nllen == 1 || (p)[-NLBLOCK->nllen+1] == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || RAWUCHARTEST(p - NLBLOCK->nllen + 1) == NLBLOCK->nl[1]) \
324      ) \      ) \
325    )    )
326    
# Line 267  used for the external interface and appe Line 334  used for the external interface and appe
334  must begin with PCRE_. */  must begin with PCRE_. */
335    
336  #ifdef CUSTOM_SUBJECT_PTR  #ifdef CUSTOM_SUBJECT_PTR
337  #define PCRE_SPTR CUSTOM_SUBJECT_PTR  #define PCRE_PUCHAR CUSTOM_SUBJECT_PTR
 #define USPTR CUSTOM_SUBJECT_PTR  
338  #else  #else
339  #define PCRE_SPTR const char *  #define PCRE_PUCHAR const pcre_uchar *
 #define USPTR const unsigned char *  
340  #endif  #endif
341    
   
   
342  /* Include the public PCRE header and the definitions of UCP character property  /* Include the public PCRE header and the definitions of UCP character property
343  values. */  values. */
344    
# Line 343  The macros are controlled by the value o Line 406  The macros are controlled by the value o
406  the config.h file, but can be overridden by using -D on the command line. This  the config.h file, but can be overridden by using -D on the command line. This
407  is automated on Unix systems via the "configure" command. */  is automated on Unix systems via the "configure" command. */
408    
409    #if defined COMPILE_PCRE8
410    
411  #if LINK_SIZE == 2  #if LINK_SIZE == 2
412    
413  #define PUT(a,n,d)   \  #define PUT(a,n,d)   \
# Line 379  is automated on Unix systems via the "co Line 444  is automated on Unix systems via the "co
444  #define GET(a,n) \  #define GET(a,n) \
445    (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3])    (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3])
446    
447  #define MAX_PATTERN_SIZE (1 << 30)   /* Keep it positive */  /* Keep it positive */
448    #define MAX_PATTERN_SIZE (1 << 30)
449    
450    #else
451    #error LINK_SIZE must be either 2, 3, or 4
452    #endif
453    
454    #elif defined COMPILE_PCRE16
455    
456    #if LINK_SIZE == 2
457    
458    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
459    #undef LINK_SIZE
460    #define LINK_SIZE 1
461    
462    #define PUT(a,n,d)   \
463      (a[n] = (d))
464    
465    #define GET(a,n) \
466      (a[n])
467    
468    #define MAX_PATTERN_SIZE (1 << 16)
469    
470    #elif LINK_SIZE == 3 || LINK_SIZE == 4
471    
472    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
473    #undef LINK_SIZE
474    #define LINK_SIZE 2
475    
476    #define PUT(a,n,d)   \
477      (a[n] = (d) >> 16), \
478      (a[(n)+1] = (d) & 65535)
479    
480    #define GET(a,n) \
481      (((a)[n] << 16) | (a)[(n)+1])
482    
483    /* Keep it positive */
484    #define MAX_PATTERN_SIZE (1 << 30)
485    
486  #else  #else
487  #error LINK_SIZE must be either 2, 3, or 4  #error LINK_SIZE must be either 2, 3, or 4
488  #endif  #endif
489    
490    #elif defined COMPILE_PCRE32
491    
492    /* Only supported LINK_SIZE is 4 */
493    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
494    #undef LINK_SIZE
495    #define LINK_SIZE 1
496    
497    #define PUT(a,n,d)   \
498      (a[n] = (d))
499    
500    #define GET(a,n) \
501      (a[n])
502    
503    /* Keep it positive */
504    #define MAX_PATTERN_SIZE (1 << 30)
505    
506    #else
507    #error Unsupported compiling mode
508    #endif /* COMPILE_PCRE[8|16|32] */
509    
510  /* Convenience macro defined in terms of the others */  /* Convenience macro defined in terms of the others */
511    
# Line 396  is automated on Unix systems via the "co Line 516  is automated on Unix systems via the "co
516  offsets changes. There are used for repeat counts and for other things such as  offsets changes. There are used for repeat counts and for other things such as
517  capturing parenthesis numbers in back references. */  capturing parenthesis numbers in back references. */
518    
519    #if defined COMPILE_PCRE8
520    
521    #define IMM2_SIZE 2
522    
523  #define PUT2(a,n,d)   \  #define PUT2(a,n,d)   \
524    a[n] = (d) >> 8; \    a[n] = (d) >> 8; \
525    a[(n)+1] = (d) & 255    a[(n)+1] = (d) & 255
526    
527    /* For reasons that I do not understand, the expression in this GET2 macro is
528    treated by gcc as a signed expression, even when a is declared as unsigned. It
529    seems that any kind of arithmetic results in a signed value. */
530    
531  #define GET2(a,n) \  #define GET2(a,n) \
532    (((a)[n] << 8) | (a)[(n)+1])    (unsigned int)(((a)[n] << 8) | (a)[(n)+1])
533    
534  #define PUT2INC(a,n,d)  PUT2(a,n,d), a += 2  #elif defined COMPILE_PCRE16
535    
536    #define IMM2_SIZE 1
537    
538  /* When UTF-8 encoding is being used, a character is no longer just a single  #define PUT2(a,n,d)   \
539  byte. The macros for character handling generate simple sequences when used in     a[n] = d
540  byte-mode, and more complicated ones for UTF-8 characters. GETCHARLENTEST is  
541  not used when UTF-8 is not supported, so it is not defined, and BACKCHAR should  #define GET2(a,n) \
542  never be called in byte mode. To make sure they can never even appear when     a[n]
 UTF-8 support is omitted, we don't even define them. */  
543    
544  #ifndef SUPPORT_UTF8  #elif defined COMPILE_PCRE32
545    
546    #define IMM2_SIZE 1
547    
548    #define PUT2(a,n,d)   \
549       a[n] = d
550    
551    #define GET2(a,n) \
552       a[n]
553    
554    #else
555    #error Unsupported compiling mode
556    #endif /* COMPILE_PCRE[8|16|32] */
557    
558    #define PUT2INC(a,n,d)  PUT2(a,n,d), a += IMM2_SIZE
559    
560    /* The maximum length of a MARK name is currently one data unit; it may be
561    changed in future to be a fixed number of bytes or to depend on LINK_SIZE. */
562    
563    #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
564    #define MAX_MARK ((1u << 16) - 1)
565    #else
566    #define MAX_MARK ((1u << 8) - 1)
567    #endif
568    
569    /* When UTF encoding is being used, a character is no longer just a single
570    byte. The macros for character handling generate simple sequences when used in
571    character-mode, and more complicated ones for UTF characters. GETCHARLENTEST
572    and other macros are not used when UTF is not supported, so they are not
573    defined. To make sure they can never even appear when UTF support is omitted,
574    we don't even define them. */
575    
576    #ifndef SUPPORT_UTF
577    
578    /* #define MAX_VALUE_FOR_SINGLE_CHAR */
579    /* #define HAS_EXTRALEN(c) */
580    /* #define GET_EXTRALEN(c) */
581    /* #define NOT_FIRSTCHAR(c) */
582  #define GETCHAR(c, eptr) c = *eptr;  #define GETCHAR(c, eptr) c = *eptr;
583  #define GETCHARTEST(c, eptr) c = *eptr;  #define GETCHARTEST(c, eptr) c = *eptr;
584  #define GETCHARINC(c, eptr) c = *eptr++;  #define GETCHARINC(c, eptr) c = *eptr++;
585  #define GETCHARINCTEST(c, eptr) c = *eptr++;  #define GETCHARINCTEST(c, eptr) c = *eptr++;
586  #define GETCHARLEN(c, eptr, len) c = *eptr;  #define GETCHARLEN(c, eptr, len) c = *eptr;
587    #define RAWUCHAR(eptr) (*(eptr))
588    #define RAWUCHARINC(eptr) (*(eptr)++)
589    #define RAWUCHARTEST(eptr) (*(eptr))
590    #define RAWUCHARINCTEST(eptr) (*(eptr)++)
591  /* #define GETCHARLENTEST(c, eptr, len) */  /* #define GETCHARLENTEST(c, eptr, len) */
592  /* #define BACKCHAR(eptr) */  /* #define BACKCHAR(eptr) */
593    /* #define FORWARDCHAR(eptr) */
594    /* #define ACROSSCHAR(condition, eptr, action) */
595    
596  #else   /* SUPPORT_UTF8 */  #else   /* SUPPORT_UTF */
597    
598  /* These macros were originally written in the form of loops that used data  /* Tests whether the code point needs extra characters to decode. */
599  from the tables whose names start with _pcre_utf8_table. They were rewritten by  
600  a user so as not to use loops, because in some environments this gives a  #define HASUTF8EXTRALEN(c) ((c) >= 0xc0)
 significant performance advantage, and it seems never to do any harm. */  
601    
602  /* Base macro to pick up the remaining bytes of a UTF-8 character, not  /* Base macro to pick up the remaining bytes of a UTF-8 character, not
603  advancing the pointer. */  advancing the pointer. */
# Line 451  advancing the pointer. */ Line 621  advancing the pointer. */
621            ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \            ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \
622      }      }
623    
 /* Get the next UTF-8 character, not advancing the pointer. This is called when  
 we know we are in UTF-8 mode. */  
   
 #define GETCHAR(c, eptr) \  
   c = *eptr; \  
   if (c >= 0xc0) GETUTF8(c, eptr);  
   
 /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the  
 pointer. */  
   
 #define GETCHARTEST(c, eptr) \  
   c = *eptr; \  
   if (utf8 && c >= 0xc0) GETUTF8(c, eptr);  
   
624  /* Base macro to pick up the remaining bytes of a UTF-8 character, advancing  /* Base macro to pick up the remaining bytes of a UTF-8 character, advancing
625  the pointer. */  the pointer. */
626    
# Line 499  the pointer. */ Line 655  the pointer. */
655        } \        } \
656      }      }
657    
658    #if defined COMPILE_PCRE8
659    
660    /* These macros were originally written in the form of loops that used data
661    from the tables whose names start with PRIV(utf8_table). They were rewritten by
662    a user so as not to use loops, because in some environments this gives a
663    significant performance advantage, and it seems never to do any harm. */
664    
665    /* Tells the biggest code point which can be encoded as a single character. */
666    
667    #define MAX_VALUE_FOR_SINGLE_CHAR 127
668    
669    /* Tests whether the code point needs extra characters to decode. */
670    
671    #define HAS_EXTRALEN(c) ((c) >= 0xc0)
672    
673    /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.
674    Otherwise it has an undefined behaviour. */
675    
676    #define GET_EXTRALEN(c) (PRIV(utf8_table4)[(c) & 0x3f])
677    
678    /* Returns TRUE, if the given character is not the first character
679    of a UTF sequence. */
680    
681    #define NOT_FIRSTCHAR(c) (((c) & 0xc0) == 0x80)
682    
683    /* Get the next UTF-8 character, not advancing the pointer. This is called when
684    we know we are in UTF-8 mode. */
685    
686    #define GETCHAR(c, eptr) \
687      c = *eptr; \
688      if (c >= 0xc0) GETUTF8(c, eptr);
689    
690    /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
691    pointer. */
692    
693    #define GETCHARTEST(c, eptr) \
694      c = *eptr; \
695      if (utf && c >= 0xc0) GETUTF8(c, eptr);
696    
697  /* Get the next UTF-8 character, advancing the pointer. This is called when we  /* Get the next UTF-8 character, advancing the pointer. This is called when we
698  know we are in UTF-8 mode. */  know we are in UTF-8 mode. */
699    
# Line 511  This is called when we don't know if we Line 706  This is called when we don't know if we
706    
707  #define GETCHARINCTEST(c, eptr) \  #define GETCHARINCTEST(c, eptr) \
708    c = *eptr++; \    c = *eptr++; \
709    if (utf8 && c >= 0xc0) GETUTF8INC(c, eptr);    if (utf && c >= 0xc0) GETUTF8INC(c, eptr);
710    
711  /* Base macro to pick up the remaining bytes of a UTF-8 character, not  /* Base macro to pick up the remaining bytes of a UTF-8 character, not
712  advancing the pointer, incrementing the length. */  advancing the pointer, incrementing the length. */
# Line 563  do not know if we are in UTF-8 mode. */ Line 758  do not know if we are in UTF-8 mode. */
758    
759  #define GETCHARLENTEST(c, eptr, len) \  #define GETCHARLENTEST(c, eptr, len) \
760    c = *eptr; \    c = *eptr; \
761    if (utf8 && c >= 0xc0) GETUTF8LEN(c, eptr, len);    if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len);
762    
763    /* Returns the next uchar, not advancing the pointer. This is called when
764    we know we are in UTF mode. */
765    
766    #define RAWUCHAR(eptr) \
767      (*(eptr))
768    
769    /* Returns the next uchar, advancing the pointer. This is called when
770    we know we are in UTF mode. */
771    
772    #define RAWUCHARINC(eptr) \
773      (*(eptr)++)
774    
775    /* Returns the next uchar, testing for UTF mode, and not advancing the
776    pointer. */
777    
778    #define RAWUCHARTEST(eptr) \
779      (*(eptr))
780    
781    /* Returns the next uchar, testing for UTF mode, advancing the
782    pointer. */
783    
784    #define RAWUCHARINCTEST(eptr) \
785      (*(eptr)++)
786    
787  /* 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
788  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
# Line 571  because almost all calls are already wit Line 790  because almost all calls are already wit
790    
791  #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--  #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--
792    
793  #endif  /* SUPPORT_UTF8 */  /* Same as above, just in the other direction. */
794    #define FORWARDCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr++
795    
796    /* Same as above, but it allows a fully customizable form. */
797    #define ACROSSCHAR(condition, eptr, action) \
798      while((condition) && ((eptr) & 0xc0) == 0x80) action
799    
800  /* In case there is no definition of offsetof() provided - though any proper  #elif defined COMPILE_PCRE16
 Standard C system should have one. */  
801    
802  #ifndef offsetof  /* Tells the biggest code point which can be encoded as a single character. */
803  #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))  
804  #endif  #define MAX_VALUE_FOR_SINGLE_CHAR 65535
805    
806    /* Tests whether the code point needs extra characters to decode. */
807    
808    #define HAS_EXTRALEN(c) (((c) & 0xfc00) == 0xd800)
809    
810    /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.
811    Otherwise it has an undefined behaviour. */
812    
813    #define GET_EXTRALEN(c) 1
814    
815    /* Returns TRUE, if the given character is not the first character
816    of a UTF sequence. */
817    
818    #define NOT_FIRSTCHAR(c) (((c) & 0xfc00) == 0xdc00)
819    
820    /* Base macro to pick up the low surrogate of a UTF-16 character, not
821    advancing the pointer. */
822    
823    #define GETUTF16(c, eptr) \
824       { c = (((c & 0x3ff) << 10) | (eptr[1] & 0x3ff)) + 0x10000; }
825    
826    /* Get the next UTF-16 character, not advancing the pointer. This is called when
827    we know we are in UTF-16 mode. */
828    
829    #define GETCHAR(c, eptr) \
830      c = *eptr; \
831      if ((c & 0xfc00) == 0xd800) GETUTF16(c, eptr);
832    
833    /* Get the next UTF-16 character, testing for UTF-16 mode, and not advancing the
834    pointer. */
835    
836    #define GETCHARTEST(c, eptr) \
837      c = *eptr; \
838      if (utf && (c & 0xfc00) == 0xd800) GETUTF16(c, eptr);
839    
840    /* Base macro to pick up the low surrogate of a UTF-16 character, advancing
841    the pointer. */
842    
843    #define GETUTF16INC(c, eptr) \
844       { c = (((c & 0x3ff) << 10) | (*eptr++ & 0x3ff)) + 0x10000; }
845    
846    /* Get the next UTF-16 character, advancing the pointer. This is called when we
847    know we are in UTF-16 mode. */
848    
849    #define GETCHARINC(c, eptr) \
850      c = *eptr++; \
851      if ((c & 0xfc00) == 0xd800) GETUTF16INC(c, eptr);
852    
853    /* Get the next character, testing for UTF-16 mode, and advancing the pointer.
854    This is called when we don't know if we are in UTF-16 mode. */
855    
856    #define GETCHARINCTEST(c, eptr) \
857      c = *eptr++; \
858      if (utf && (c & 0xfc00) == 0xd800) GETUTF16INC(c, eptr);
859    
860    /* Base macro to pick up the low surrogate of a UTF-16 character, not
861    advancing the pointer, incrementing the length. */
862    
863    #define GETUTF16LEN(c, eptr, len) \
864       { c = (((c & 0x3ff) << 10) | (eptr[1] & 0x3ff)) + 0x10000; len++; }
865    
866    /* Get the next UTF-16 character, not advancing the pointer, incrementing
867    length if there is a low surrogate. This is called when we know we are in
868    UTF-16 mode. */
869    
870    #define GETCHARLEN(c, eptr, len) \
871      c = *eptr; \
872      if ((c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);
873    
874    /* Get the next UTF-816character, testing for UTF-16 mode, not advancing the
875    pointer, incrementing length if there is a low surrogate. This is called when
876    we do not know if we are in UTF-16 mode. */
877    
878    #define GETCHARLENTEST(c, eptr, len) \
879      c = *eptr; \
880      if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);
881    
882    /* Returns the next uchar, not advancing the pointer. This is called when
883    we know we are in UTF mode. */
884    
885    #define RAWUCHAR(eptr) \
886      (*(eptr))
887    
888    /* Returns the next uchar, advancing the pointer. This is called when
889    we know we are in UTF mode. */
890    
891    #define RAWUCHARINC(eptr) \
892      (*(eptr)++)
893    
894    /* Returns the next uchar, testing for UTF mode, and not advancing the
895    pointer. */
896    
897    #define RAWUCHARTEST(eptr) \
898      (*(eptr))
899    
900    /* Returns the next uchar, testing for UTF mode, advancing the
901    pointer. */
902    
903    #define RAWUCHARINCTEST(eptr) \
904      (*(eptr)++)
905    
906    /* If the pointer is not at the start of a character, move it back until
907    it is. This is called only in UTF-16 mode - we don't put a test within the
908    macro because almost all calls are already within a block of UTF-16 only
909    code. */
910    
911    #define BACKCHAR(eptr) if ((*eptr & 0xfc00) == 0xdc00) eptr--
912    
913    /* Same as above, just in the other direction. */
914    #define FORWARDCHAR(eptr) if ((*eptr & 0xfc00) == 0xdc00) eptr++
915    
916    /* Same as above, but it allows a fully customizable form. */
917    #define ACROSSCHAR(condition, eptr, action) \
918      if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action
919    
920    #elif defined COMPILE_PCRE32
921    
922    /* These are trivial for the 32-bit library, since all UTF-32 characters fit
923    into one pcre_uchar unit. */
924    #define MAX_VALUE_FOR_SINGLE_CHAR (0x10ffffu)
925    #define HAS_EXTRALEN(c) (0)
926    #define GET_EXTRALEN(c) (0)
927    #define NOT_FIRSTCHAR(c) (0)
928    
929    #define UTF32_MASK (0x1fffffu)
930    
931    /* Get the next UTF-32 character, not advancing the pointer. This is called when
932    we know we are in UTF-32 mode. */
933    
934    #define GETCHAR(c, eptr) \
935      c = (*eptr) & UTF32_MASK;
936    
937    /* Get the next UTF-32 character, testing for UTF-32 mode, and not advancing the
938    pointer. */
939    
940    #define GETCHARTEST(c, eptr) \
941      c = *eptr; \
942      if (utf) c &= UTF32_MASK;
943    
944    /* Get the next UTF-32 character, advancing the pointer. This is called when we
945    know we are in UTF-32 mode. */
946    
947    #define GETCHARINC(c, eptr) \
948      c = (*eptr++) & UTF32_MASK;
949    
950    /* Get the next character, testing for UTF-32 mode, and advancing the pointer.
951    This is called when we don't know if we are in UTF-32 mode. */
952    
953    #define GETCHARINCTEST(c, eptr) \
954      c = *eptr++; \
955      if (utf) c &= UTF32_MASK;
956    
957    /* Get the next UTF-32 character, not advancing the pointer, not incrementing
958    length (since all UTF-32 is of length 1). This is called when we know we are in
959    UTF-32 mode. */
960    
961    #define GETCHARLEN(c, eptr, len) \
962      GETCHAR(c, eptr)
963    
964    /* Get the next UTF-32character, testing for UTF-32 mode, not advancing the
965    pointer, not incrementing the length (since all UTF-32 is of length 1).
966    This is called when we do not know if we are in UTF-32 mode. */
967    
968    #define GETCHARLENTEST(c, eptr, len) \
969      GETCHARTEST(c, eptr)
970    
971    /* Returns the next uchar, not advancing the pointer. This is called when
972    we know we are in UTF mode. */
973    
974    #define RAWUCHAR(eptr) \
975      (*(eptr) & UTF32_MASK)
976    
977    /* Returns the next uchar, advancing the pointer. This is called when
978    we know we are in UTF mode. */
979    
980    #define RAWUCHARINC(eptr) \
981      (*(eptr)++ & UTF32_MASK)
982    
983    /* Returns the next uchar, testing for UTF mode, and not advancing the
984    pointer. */
985    
986    #define RAWUCHARTEST(eptr) \
987      (utf ? (*(eptr) & UTF32_MASK) : *(eptr))
988    
989    /* Returns the next uchar, testing for UTF mode, advancing the
990    pointer. */
991    
992    #define RAWUCHARINCTEST(eptr) \
993      (utf ? (*(eptr)++ & UTF32_MASK) : *(eptr)++)
994    
995    /* If the pointer is not at the start of a character, move it back until
996    it is. This is called only in UTF-32 mode - we don't put a test within the
997    macro because almost all calls are already within a block of UTF-32 only
998    code.
999    These are all no-ops since all UTF-32 characters fit into one pcre_uchar. */
1000    
1001    #define BACKCHAR(eptr) do { } while (0)
1002    
1003    /* Same as above, just in the other direction. */
1004    #define FORWARDCHAR(eptr) do { } while (0)
1005    
1006    /* Same as above, but it allows a fully customizable form. */
1007    #define ACROSSCHAR(condition, eptr, action) do { } while (0)
1008    
1009    #else
1010    #error Unsupported compiling mode
1011    #endif /* COMPILE_PCRE[8|16|32] */
1012    
1013    #endif  /* SUPPORT_UTF */
1014    
1015    /* Tests for Unicode horizontal and vertical whitespace characters must check a
1016    number of different values. Using a switch statement for this generates the
1017    fastest code (no loop, no memory access), and there are several places in the
1018    interpreter code where this happens. In order to ensure that all the case lists
1019    remain in step, we use macros so that there is only one place where the lists
1020    are defined.
1021    
1022    These values are also required as lists in pcre_compile.c when processing \h,
1023    \H, \v and \V in a character class. The lists are defined in pcre_tables.c, but
1024    macros that define the values are here so that all the definitions are
1025    together. The lists must be in ascending character order, terminated by
1026    NOTACHAR (which is 0xffffffff).
1027    
1028    Any changes should ensure that the various macros are kept in step with each
1029    other. NOTE: The values also appear in pcre_jit_compile.c. */
1030    
1031    /* ------ ASCII/Unicode environments ------ */
1032    
1033    #ifndef EBCDIC
1034    
1035    #define HSPACE_LIST \
1036      CHAR_HT, CHAR_SPACE, 0xa0, \
1037      0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \
1038      0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \
1039      NOTACHAR
1040    
1041    #define HSPACE_MULTIBYTE_CASES \
1042      case 0x1680:  /* OGHAM SPACE MARK */ \
1043      case 0x180e:  /* MONGOLIAN VOWEL SEPARATOR */ \
1044      case 0x2000:  /* EN QUAD */ \
1045      case 0x2001:  /* EM QUAD */ \
1046      case 0x2002:  /* EN SPACE */ \
1047      case 0x2003:  /* EM SPACE */ \
1048      case 0x2004:  /* THREE-PER-EM SPACE */ \
1049      case 0x2005:  /* FOUR-PER-EM SPACE */ \
1050      case 0x2006:  /* SIX-PER-EM SPACE */ \
1051      case 0x2007:  /* FIGURE SPACE */ \
1052      case 0x2008:  /* PUNCTUATION SPACE */ \
1053      case 0x2009:  /* THIN SPACE */ \
1054      case 0x200A:  /* HAIR SPACE */ \
1055      case 0x202f:  /* NARROW NO-BREAK SPACE */ \
1056      case 0x205f:  /* MEDIUM MATHEMATICAL SPACE */ \
1057      case 0x3000   /* IDEOGRAPHIC SPACE */
1058    
1059    #define HSPACE_BYTE_CASES \
1060      case CHAR_HT: \
1061      case CHAR_SPACE: \
1062      case 0xa0     /* NBSP */
1063    
1064    #define HSPACE_CASES \
1065      HSPACE_BYTE_CASES: \
1066      HSPACE_MULTIBYTE_CASES
1067    
1068    #define VSPACE_LIST \
1069      CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR
1070    
1071    #define VSPACE_MULTIBYTE_CASES \
1072      case 0x2028:    /* LINE SEPARATOR */ \
1073      case 0x2029     /* PARAGRAPH SEPARATOR */
1074    
1075    #define VSPACE_BYTE_CASES \
1076      case CHAR_LF: \
1077      case CHAR_VT: \
1078      case CHAR_FF: \
1079      case CHAR_CR: \
1080      case CHAR_NEL
1081    
1082    #define VSPACE_CASES \
1083      VSPACE_BYTE_CASES: \
1084      VSPACE_MULTIBYTE_CASES
1085    
1086    /* ------ EBCDIC environments ------ */
1087    
1088    #else
1089    #define HSPACE_LIST CHAR_HT, CHAR_SPACE
1090    
1091    #define HSPACE_BYTE_CASES \
1092      case CHAR_HT: \
1093      case CHAR_SPACE
1094    
1095    #define HSPACE_CASES HSPACE_BYTE_CASES
1096    
1097    #ifdef EBCDIC_NL25
1098    #define VSPACE_LIST \
1099      CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR
1100    #else
1101    #define VSPACE_LIST \
1102      CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR
1103    #endif
1104    
1105    #define VSPACE_BYTE_CASES \
1106      case CHAR_LF: \
1107      case CHAR_VT: \
1108      case CHAR_FF: \
1109      case CHAR_CR: \
1110      case CHAR_NEL
1111    
1112    #define VSPACE_CASES VSPACE_BYTE_CASES
1113    #endif  /* EBCDIC */
1114    
1115    /* ------ End of whitespace macros ------ */
1116    
 /* These are the public options that can change during matching. */  
1117    
 #define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)  
1118    
1119  /* Private flags containing information about the compiled regex. They used to  /* Private flags containing information about the compiled regex. They used to
1120  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 now they
# Line 592  are in a 16-bit flags word. From release Line 1122  are in a 16-bit flags word. From release
1122  the restrictions on partial matching have been lifted. It remains for backwards  the restrictions on partial matching have been lifted. It remains for backwards
1123  compatibility. */  compatibility. */
1124    
1125  #define PCRE_NOPARTIAL     0x0001  /* can't use partial with this regex */  #define PCRE_MODE8         0x0001  /* compiled in 8 bit mode */
1126  #define PCRE_FIRSTSET      0x0002  /* first_byte is set */  #define PCRE_MODE16        0x0002  /* compiled in 16 bit mode */
1127  #define PCRE_REQCHSET      0x0004  /* req_byte is set */  #define PCRE_MODE32        0x0004  /* compiled in 32 bit mode */
1128  #define PCRE_STARTLINE     0x0008  /* start after \n for multiline */  #define PCRE_FIRSTSET      0x0010  /* first_char is set */
1129  #define PCRE_JCHANGED      0x0010  /* j option used in regex */  #define PCRE_FCH_CASELESS  0x0020  /* caseless first char */
1130  #define PCRE_HASCRORLF     0x0020  /* explicit \r or \n in pattern */  #define PCRE_REQCHSET      0x0040  /* req_byte is set */
1131    #define PCRE_RCH_CASELESS  0x0080  /* caseless requested char */
1132    #define PCRE_STARTLINE     0x0100  /* start after \n for multiline */
1133    #define PCRE_NOPARTIAL     0x0200  /* can't use partial with this regex */
1134    #define PCRE_JCHANGED      0x0400  /* j option used in regex */
1135    #define PCRE_HASCRORLF     0x0800  /* explicit \r or \n in pattern */
1136    #define PCRE_HASTHEN       0x1000  /* pattern contains (*THEN) */
1137    
1138    #if defined COMPILE_PCRE8
1139    #define PCRE_MODE          PCRE_MODE8
1140    #elif defined COMPILE_PCRE16
1141    #define PCRE_MODE          PCRE_MODE16
1142    #elif defined COMPILE_PCRE32
1143    #define PCRE_MODE          PCRE_MODE32
1144    #endif
1145    #define PCRE_MODE_MASK     (PCRE_MODE8 | PCRE_MODE16 | PCRE_MODE32)
1146    
1147  /* Options for the "extra" block produced by pcre_study(). */  /* Flags for the "extra" block produced by pcre_study(). */
1148    
1149  #define PCRE_STUDY_MAPPED   0x01     /* a map of starting chars exists */  #define PCRE_STUDY_MAPPED  0x0001  /* a map of starting chars exists */
1150  #define PCRE_STUDY_MINLEN   0x02     /* a minimum length field exists */  #define PCRE_STUDY_MINLEN  0x0002  /* a minimum length field exists */
1151    
1152  /* Masks for identifying the public options that are permitted at compile  /* Masks for identifying the public options that are permitted at compile
1153  time, run time, or study time, respectively. */  time, run time, or study time, respectively. */
# Line 628  time, run time, or study time, respectiv Line 1173  time, run time, or study time, respectiv
1173     PCRE_DFA_RESTART|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \     PCRE_DFA_RESTART|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \
1174     PCRE_NO_START_OPTIMIZE)     PCRE_NO_START_OPTIMIZE)
1175    
1176  #define PUBLIC_STUDY_OPTIONS 0   /* None defined */  #define PUBLIC_STUDY_OPTIONS \
1177       (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \
1178        PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE|PCRE_STUDY_EXTRA_NEEDED)
1179    
1180  /* Magic number to provide a small check against being handed junk. Also used  /* Magic number to provide a small check against being handed junk. */
 to detect whether a pattern was compiled on a host of different endianness. */  
1181    
1182  #define MAGIC_NUMBER  0x50435245UL   /* 'PCRE' */  #define MAGIC_NUMBER  0x50435245UL   /* 'PCRE' */
1183    
1184  /* Negative values for the firstchar and reqchar variables */  /* This variable is used to detect a loaded regular expression
1185    in different endianness. */
1186    
1187  #define REQ_UNSET (-2)  #define REVERSED_MAGIC_NUMBER  0x45524350UL   /* 'ERCP' */
 #define REQ_NONE  (-1)  
1188    
1189  /* The maximum remaining length of subject we are prepared to search for a  /* The maximum remaining length of subject we are prepared to search for a
1190  req_byte match. */  req_byte match. */
1191    
1192  #define REQ_BYTE_MAX 1000  #define REQ_BYTE_MAX 1000
1193    
 /* Flags added to firstbyte or reqbyte; a "non-literal" item is either a  
 variable-length repeat, or a anything other than literal characters. */  
   
 #define REQ_CASELESS 0x0100    /* indicates caselessness */  
 #define REQ_VARY     0x0200    /* reqbyte followed non-literal item */  
   
1194  /* Miscellaneous definitions. The #ifndef is to pacify compiler warnings in  /* Miscellaneous definitions. The #ifndef is to pacify compiler warnings in
1195  environments where these macros are defined elsewhere. Unfortunately, there  environments where these macros are defined elsewhere. Unfortunately, there
1196  is no way to do the same for the typedef. */  is no way to do the same for the typedef. */
# Line 679  for) in a minority area (EBCDIC platform Line 1219  for) in a minority area (EBCDIC platform
1219  application that did need both could compile two versions of the library, using  application that did need both could compile two versions of the library, using
1220  macros to give the functions distinct names. */  macros to give the functions distinct names. */
1221    
1222  #ifndef SUPPORT_UTF8  #ifndef SUPPORT_UTF
1223    
1224  /* UTF-8 support is not enabled; use the platform-dependent character literals  /* UTF-8 support is not enabled; use the platform-dependent character literals
1225  so that PCRE works on both ASCII and EBCDIC platforms, in non-UTF-mode only. */  so that PCRE works in both ASCII and EBCDIC environments, but only in non-UTF
1226    mode. Newline characters are problematic in EBCDIC. Though it has CR and LF
1227    characters, a common practice has been to use its NL (0x15) character as the
1228    line terminator in C-like processing environments. However, sometimes the LF
1229    (0x25) character is used instead, according to this Unicode document:
1230    
1231    http://unicode.org/standard/reports/tr13/tr13-5.html
1232    
1233    PCRE defaults EBCDIC NL to 0x15, but has a build-time option to select 0x25
1234    instead. Whichever is *not* chosen is defined as NEL.
1235    
1236    In both ASCII and EBCDIC environments, CHAR_NL and CHAR_LF are synonyms for the
1237    same code point. */
1238    
1239    #ifdef EBCDIC
1240    
1241    #ifndef EBCDIC_NL25
1242    #define CHAR_NL                     '\x15'
1243    #define CHAR_NEL                    '\x25'
1244    #define STR_NL                      "\x15"
1245    #define STR_NEL                     "\x25"
1246    #else
1247    #define CHAR_NL                     '\x25'
1248    #define CHAR_NEL                    '\x15'
1249    #define STR_NL                      "\x25"
1250    #define STR_NEL                     "\x15"
1251    #endif
1252    
1253    #define CHAR_LF                     CHAR_NL
1254    #define STR_LF                      STR_NL
1255    
1256    #define CHAR_ESC                    '\047'
1257    #define CHAR_DEL                    '\007'
1258    #define STR_ESC                     "\047"
1259    #define STR_DEL                     "\007"
1260    
1261    #else  /* Not EBCDIC */
1262    
1263    /* In ASCII/Unicode, linefeed is '\n' and we equate this to NL for
1264    compatibility. NEL is the Unicode newline character; make sure it is
1265    a positive value. */
1266    
1267    #define CHAR_LF                     '\n'
1268    #define CHAR_NL                     CHAR_LF
1269    #define CHAR_NEL                    ((unsigned char)'\x85')
1270    #define CHAR_ESC                    '\033'
1271    #define CHAR_DEL                    '\177'
1272    
1273    #define STR_LF                      "\n"
1274    #define STR_NL                      STR_LF
1275    #define STR_NEL                     "\x85"
1276    #define STR_ESC                     "\033"
1277    #define STR_DEL                     "\177"
1278    
1279    #endif  /* EBCDIC */
1280    
1281    /* The remaining definitions work in both environments. */
1282    
1283  #define CHAR_HT                     '\t'  #define CHAR_HT                     '\t'
1284  #define CHAR_VT                     '\v'  #define CHAR_VT                     '\v'
1285  #define CHAR_FF                     '\f'  #define CHAR_FF                     '\f'
1286  #define CHAR_CR                     '\r'  #define CHAR_CR                     '\r'
 #define CHAR_NL                     '\n'  
1287  #define CHAR_BS                     '\b'  #define CHAR_BS                     '\b'
1288  #define CHAR_BEL                    '\a'  #define CHAR_BEL                    '\a'
 #ifdef EBCDIC  
 #define CHAR_ESC                    '\047'  
 #define CHAR_DEL                    '\007'  
 #else  
 #define CHAR_ESC                    '\033'  
 #define CHAR_DEL                    '\177'  
 #endif  
1289    
1290  #define CHAR_SPACE                  ' '  #define CHAR_SPACE                  ' '
1291  #define CHAR_EXCLAMATION_MARK       '!'  #define CHAR_EXCLAMATION_MARK       '!'
# Line 799  so that PCRE works on both ASCII and EBC Line 1387  so that PCRE works on both ASCII and EBC
1387  #define STR_VT                      "\v"  #define STR_VT                      "\v"
1388  #define STR_FF                      "\f"  #define STR_FF                      "\f"
1389  #define STR_CR                      "\r"  #define STR_CR                      "\r"
 #define STR_NL                      "\n"  
1390  #define STR_BS                      "\b"  #define STR_BS                      "\b"
1391  #define STR_BEL                     "\a"  #define STR_BEL                     "\a"
 #ifdef EBCDIC  
 #define STR_ESC                     "\047"  
 #define STR_DEL                     "\007"  
 #else  
 #define STR_ESC                     "\033"  
 #define STR_DEL                     "\177"  
 #endif  
1392    
1393  #define STR_SPACE                   " "  #define STR_SPACE                   " "
1394  #define STR_EXCLAMATION_MARK        "!"  #define STR_EXCLAMATION_MARK        "!"
# Line 939  so that PCRE works on both ASCII and EBC Line 1519  so that PCRE works on both ASCII and EBC
1519  #define STRING_ANYCRLF_RIGHTPAR        "ANYCRLF)"  #define STRING_ANYCRLF_RIGHTPAR        "ANYCRLF)"
1520  #define STRING_BSR_ANYCRLF_RIGHTPAR    "BSR_ANYCRLF)"  #define STRING_BSR_ANYCRLF_RIGHTPAR    "BSR_ANYCRLF)"
1521  #define STRING_BSR_UNICODE_RIGHTPAR    "BSR_UNICODE)"  #define STRING_BSR_UNICODE_RIGHTPAR    "BSR_UNICODE)"
1522  #define STRING_UTF8_RIGHTPAR           "UTF8)"  #ifdef COMPILE_PCRE8
1523    #define STRING_UTF_RIGHTPAR            "UTF8)"
1524    #endif
1525    #ifdef COMPILE_PCRE16
1526    #define STRING_UTF_RIGHTPAR            "UTF16)"
1527    #endif
1528    #ifdef COMPILE_PCRE32
1529    #define STRING_UTF_RIGHTPAR            "UTF32)"
1530    #endif
1531  #define STRING_UCP_RIGHTPAR            "UCP)"  #define STRING_UCP_RIGHTPAR            "UCP)"
1532  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"
1533    
1534  #else  /* SUPPORT_UTF8 */  #else  /* SUPPORT_UTF */
1535    
1536  /* UTF-8 support is enabled; always use UTF-8 (=ASCII) character codes. This  /* UTF-8 support is enabled; always use UTF-8 (=ASCII) character codes. This
1537  works in both modes non-EBCDIC platforms, and on EBCDIC platforms in UTF-8 mode  works in both modes non-EBCDIC platforms, and on EBCDIC platforms in UTF-8 mode
# Line 953  only. */ Line 1541  only. */
1541  #define CHAR_VT                     '\013'  #define CHAR_VT                     '\013'
1542  #define CHAR_FF                     '\014'  #define CHAR_FF                     '\014'
1543  #define CHAR_CR                     '\015'  #define CHAR_CR                     '\015'
1544  #define CHAR_NL                     '\012'  #define CHAR_LF                     '\012'
1545    #define CHAR_NL                     CHAR_LF
1546    #define CHAR_NEL                    ((unsigned char)'\x85')
1547  #define CHAR_BS                     '\010'  #define CHAR_BS                     '\010'
1548  #define CHAR_BEL                    '\007'  #define CHAR_BEL                    '\007'
1549  #define CHAR_ESC                    '\033'  #define CHAR_ESC                    '\033'
# Line 1194  only. */ Line 1784  only. */
1784  #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
1785  #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
1786  #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
1787  #define STRING_UTF8_RIGHTPAR           STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS  #ifdef COMPILE_PCRE8
1788    #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS
1789    #endif
1790    #ifdef COMPILE_PCRE16
1791    #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS
1792    #endif
1793    #ifdef COMPILE_PCRE32
1794    #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS
1795    #endif
1796  #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
1797  #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_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
1798    
1799  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
1800    
1801  /* Escape items that are just an encoding of a particular data value. */  /* Escape items that are just an encoding of a particular data value. */
1802    
# Line 1211  only. */ Line 1809  only. */
1809  #endif  #endif
1810    
1811  #ifndef ESC_n  #ifndef ESC_n
1812  #define ESC_n CHAR_NL  #define ESC_n CHAR_LF
1813  #endif  #endif
1814    
1815  #ifndef ESC_r  #ifndef ESC_r
# Line 1236  only. */ Line 1834  only. */
1834  #define PT_SPACE      6    /* Perl space - Z plus 9,10,12,13 */  #define PT_SPACE      6    /* Perl space - Z plus 9,10,12,13 */
1835  #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 */
1836  #define PT_WORD       8    /* Word - L plus N plus underscore */  #define PT_WORD       8    /* Word - L plus N plus underscore */
1837    #define PT_CLIST      9    /* Pseudo-property: match character list */
1838    
1839  /* 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
1840  contain UTF-8 characters with values greater than 255. */  contain characters with values greater than 255. */
1841    
1842  #define XCL_NOT    0x01    /* Flag: this is a negative class */  #define XCL_NOT    0x01    /* Flag: this is a negative class */
1843  #define XCL_MAP    0x02    /* Flag: a 32-byte map is present */  #define XCL_MAP    0x02    /* Flag: a 32-byte map is present */
# Line 1251  contain UTF-8 characters with values gre Line 1850  contain UTF-8 characters with values gre
1850    
1851  /* 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
1852  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
1853  their negation. Also, they must appear in the same order as in the opcode  0 for a data character.  Also, they must appear in the same order as in the opcode
1854  definitions below, up to ESC_z. There's a dummy for OP_ALLANY because it  definitions below, up to ESC_z. There's a dummy for OP_ALLANY because it
1855  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
1856  used for [^] in JavaScript compatibility mode. In non-DOTALL mode, "." behaves  used for [^] in JavaScript compatibility mode, and for \C in non-utf mode. In
1857  like \N.  non-DOTALL mode, "." behaves like \N.
1858    
1859  The special values ESC_DU, ESC_du, etc. are used instead of ESC_D, ESC_d, etc.  The special values ESC_DU, ESC_du, etc. are used instead of ESC_D, ESC_d, etc.
1860  when PCRE_UCP is set, when replacement of \d etc by \p sequences is required.  when PCRE_UCP is set and replacement of \d etc by \p sequences is required.
1861  They must be contiguous, and remain in order so that the replacements can be  They must be contiguous, and remain in order so that the replacements can be
1862  looked up from a table.  looked up from a table.
1863    
1864  The final escape must be ESC_REF as subsequent values are used for  Negative numbers are used to encode a backreference (\1, \2, \3, etc.) in
1865  backreferences (\1, \2, \3, etc). There are two tests in the code for an escape  check_escape(). There are two tests in the code for an escape
1866  greater than ESC_b and less than ESC_Z to detect the types that may be  greater than ESC_b and less than ESC_Z to detect the types that may be
1867  repeated. These are the types that consume characters. If any new escapes are  repeated. These are the types that consume characters. If any new escapes are
1868  put in between that don't consume a character, that code will have to change.  put in between that don't consume a character, that code will have to change.
# Line 1273  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, E Line 1872  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, E
1872         ESC_W, ESC_w, ESC_N, ESC_dum, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H,         ESC_W, ESC_w, ESC_N, ESC_dum, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H,
1873         ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z,         ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z,
1874         ESC_E, ESC_Q, ESC_g, ESC_k,         ESC_E, ESC_Q, ESC_g, ESC_k,
1875         ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu,         ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu };
        ESC_REF };  
1876    
1877  /* Opcode table: Starting from 1 (i.e. after OP_END), the values up to  /* Opcode table: Starting from 1 (i.e. after OP_END), the values up to
1878  OP_EOD must correspond in order to the list of escapes immediately above.  OP_EOD must correspond in order to the list of escapes immediately above.
# Line 1299  enum { Line 1897  enum {
1897    OP_WHITESPACE,         /*  9 \s */    OP_WHITESPACE,         /*  9 \s */
1898    OP_NOT_WORDCHAR,       /* 10 \W */    OP_NOT_WORDCHAR,       /* 10 \W */
1899    OP_WORDCHAR,           /* 11 \w */    OP_WORDCHAR,           /* 11 \w */
1900    OP_ANY,            /* 12 Match any character except newline */  
1901      OP_ANY,            /* 12 Match any character except newline (\N) */
1902    OP_ALLANY,         /* 13 Match any character */    OP_ALLANY,         /* 13 Match any character */
1903    OP_ANYBYTE,        /* 14 Match any byte (\C); different to OP_ANY for UTF-8 */    OP_ANYBYTE,        /* 14 Match any byte (\C); different to OP_ANY for UTF-8 */
1904    OP_NOTPROP,        /* 15 \P (not Unicode property) */    OP_NOTPROP,        /* 15 \P (not Unicode property) */
# Line 1310  enum { Line 1909  enum {
1909    OP_NOT_VSPACE,     /* 20 \V (not vertical whitespace) */    OP_NOT_VSPACE,     /* 20 \V (not vertical whitespace) */
1910    OP_VSPACE,         /* 21 \v (vertical whitespace) */    OP_VSPACE,         /* 21 \v (vertical whitespace) */
1911    OP_EXTUNI,         /* 22 \X (extended Unicode sequence */    OP_EXTUNI,         /* 22 \X (extended Unicode sequence */
1912    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) */
1913    OP_EOD,            /* 24 End of data: \z */    OP_EOD,            /* 24 End of data (\z) */
1914    
1915    OP_OPT,            /* 25 Set runtime options */    OP_CIRC,           /* 25 Start of line - not multiline */
1916    OP_CIRC,           /* 26 Start of line - varies with multiline switch */    OP_CIRCM,          /* 26 Start of line - multiline */
1917    OP_DOLL,           /* 27 End of line - varies with multiline switch */    OP_DOLL,           /* 27 End of line - not multiline */
1918    OP_CHAR,           /* 28 Match one character, casefully */    OP_DOLLM,          /* 28 End of line - multiline */
1919    OP_CHARNC,         /* 29 Match one character, caselessly */    OP_CHAR,           /* 29 Match one character, casefully */
1920    OP_NOT,            /* 30 Match one character, not the following one */    OP_CHARI,          /* 30 Match one character, caselessly */
1921      OP_NOT,            /* 31 Match one character, not the given one, casefully */
1922    OP_STAR,           /* 31 The maximizing and minimizing versions of */    OP_NOTI,           /* 32 Match one character, not the given one, caselessly */
1923    OP_MINSTAR,        /* 32 these six opcodes must come in pairs, with */  
1924    OP_PLUS,           /* 33 the minimizing one second. */    /* The following sets of 13 opcodes must always be kept in step because
1925    OP_MINPLUS,        /* 34 This first set applies to single characters.*/    the offset from the first one is used to generate the others. */
1926    OP_QUERY,          /* 35 */  
1927    OP_MINQUERY,       /* 36 */    /**** Single characters, caseful, must precede the caseless ones ****/
1928    
1929    OP_UPTO,           /* 37 From 0 to n matches */    OP_STAR,           /* 33 The maximizing and minimizing versions of */
1930    OP_MINUPTO,        /* 38 */    OP_MINSTAR,        /* 34 these six opcodes must come in pairs, with */
1931    OP_EXACT,          /* 39 Exactly n matches */    OP_PLUS,           /* 35 the minimizing one second. */
1932      OP_MINPLUS,        /* 36 */
1933    OP_POSSTAR,        /* 40 Possessified star */    OP_QUERY,          /* 37 */
1934    OP_POSPLUS,        /* 41 Possessified plus */    OP_MINQUERY,       /* 38 */
1935    OP_POSQUERY,       /* 42 Posesssified query */  
1936    OP_POSUPTO,        /* 43 Possessified upto */    OP_UPTO,           /* 39 From 0 to n matches of one character, caseful*/
1937      OP_MINUPTO,        /* 40 */
1938    OP_NOTSTAR,        /* 44 The maximizing and minimizing versions of */    OP_EXACT,          /* 41 Exactly n matches */
1939    OP_NOTMINSTAR,     /* 45 these six opcodes must come in pairs, with */  
1940    OP_NOTPLUS,        /* 46 the minimizing one second. They must be in */    OP_POSSTAR,        /* 42 Possessified star, caseful */
1941    OP_NOTMINPLUS,     /* 47 exactly the same order as those above. */    OP_POSPLUS,        /* 43 Possessified plus, caseful */
1942    OP_NOTQUERY,       /* 48 This set applies to "not" single characters. */    OP_POSQUERY,       /* 44 Posesssified query, caseful */
1943    OP_NOTMINQUERY,    /* 49 */    OP_POSUPTO,        /* 45 Possessified upto, caseful */
1944    
1945    OP_NOTUPTO,        /* 50 From 0 to n matches */    /**** Single characters, caseless, must follow the caseful ones */
1946    OP_NOTMINUPTO,     /* 51 */  
1947    OP_NOTEXACT,       /* 52 Exactly n matches */    OP_STARI,          /* 46 */
1948      OP_MINSTARI,       /* 47 */
1949    OP_NOTPOSSTAR,     /* 53 Possessified versions */    OP_PLUSI,          /* 48 */
1950    OP_NOTPOSPLUS,     /* 54 */    OP_MINPLUSI,       /* 49 */
1951    OP_NOTPOSQUERY,    /* 55 */    OP_QUERYI,         /* 50 */
1952    OP_NOTPOSUPTO,     /* 56 */    OP_MINQUERYI,      /* 51 */
1953    
1954    OP_TYPESTAR,       /* 57 The maximizing and minimizing versions of */    OP_UPTOI,          /* 52 From 0 to n matches of one character, caseless */
1955    OP_TYPEMINSTAR,    /* 58 these six opcodes must come in pairs, with */    OP_MINUPTOI,       /* 53 */
1956    OP_TYPEPLUS,       /* 59 the minimizing one second. These codes must */    OP_EXACTI,         /* 54 */
1957    OP_TYPEMINPLUS,    /* 60 be in exactly the same order as those above. */  
1958    OP_TYPEQUERY,      /* 61 This set applies to character types such as \d */    OP_POSSTARI,       /* 55 Possessified star, caseless */
1959    OP_TYPEMINQUERY,   /* 62 */    OP_POSPLUSI,       /* 56 Possessified plus, caseless */
1960      OP_POSQUERYI,      /* 57 Posesssified query, caseless */
1961    OP_TYPEUPTO,       /* 63 From 0 to n matches */    OP_POSUPTOI,       /* 58 Possessified upto, caseless */
1962    OP_TYPEMINUPTO,    /* 64 */  
1963    OP_TYPEEXACT,      /* 65 Exactly n matches */    /**** The negated ones must follow the non-negated ones, and match them ****/
1964      /**** Negated single character, caseful; must precede the caseless ones ****/
1965    OP_TYPEPOSSTAR,    /* 66 Possessified versions */  
1966    OP_TYPEPOSPLUS,    /* 67 */    OP_NOTSTAR,        /* 59 The maximizing and minimizing versions of */
1967    OP_TYPEPOSQUERY,   /* 68 */    OP_NOTMINSTAR,     /* 60 these six opcodes must come in pairs, with */
1968    OP_TYPEPOSUPTO,    /* 69 */    OP_NOTPLUS,        /* 61 the minimizing one second. They must be in */
1969      OP_NOTMINPLUS,     /* 62 exactly the same order as those above. */
1970    OP_CRSTAR,         /* 70 The maximizing and minimizing versions of */    OP_NOTQUERY,       /* 63 */
1971    OP_CRMINSTAR,      /* 71 all these opcodes must come in pairs, with */    OP_NOTMINQUERY,    /* 64 */
1972    OP_CRPLUS,         /* 72 the minimizing one second. These codes must */  
1973    OP_CRMINPLUS,      /* 73 be in exactly the same order as those above. */    OP_NOTUPTO,        /* 65 From 0 to n matches, caseful */
1974    OP_CRQUERY,        /* 74 These are for character classes and back refs */    OP_NOTMINUPTO,     /* 66 */
1975    OP_CRMINQUERY,     /* 75 */    OP_NOTEXACT,       /* 67 Exactly n matches */
1976    OP_CRRANGE,        /* 76 These are different to the three sets above. */  
1977    OP_CRMINRANGE,     /* 77 */    OP_NOTPOSSTAR,     /* 68 Possessified versions, caseful */
1978      OP_NOTPOSPLUS,     /* 69 */
1979    OP_CLASS,          /* 78 Match a character class, chars < 256 only */    OP_NOTPOSQUERY,    /* 70 */
1980    OP_NCLASS,         /* 79 Same, but the bitmap was created from a negative    OP_NOTPOSUPTO,     /* 71 */
1981                             class - the difference is relevant only when a UTF-8  
1982                             character > 255 is encountered. */    /**** Negated single character, caseless; must follow the caseful ones ****/
1983    
1984    OP_XCLASS,         /* 80 Extended class for handling UTF-8 chars within the    OP_NOTSTARI,       /* 72 */
1985                             class. This does both positive and negative. */    OP_NOTMINSTARI,    /* 73 */
1986      OP_NOTPLUSI,       /* 74 */
1987    OP_REF,            /* 81 Match a back reference */    OP_NOTMINPLUSI,    /* 75 */
1988    OP_RECURSE,        /* 82 Match a numbered subpattern (possibly recursive) */    OP_NOTQUERYI,      /* 76 */
1989    OP_CALLOUT,        /* 83 Call out to external function if provided */    OP_NOTMINQUERYI,   /* 77 */
1990    
1991    OP_ALT,            /* 84 Start of alternation */    OP_NOTUPTOI,       /* 78 From 0 to n matches, caseless */
1992    OP_KET,            /* 85 End of group that doesn't have an unbounded repeat */    OP_NOTMINUPTOI,    /* 79 */
1993    OP_KETRMAX,        /* 86 These two must remain together and in this */    OP_NOTEXACTI,      /* 80 Exactly n matches */
1994    OP_KETRMIN,        /* 87 order. They are for groups the repeat for ever. */  
1995      OP_NOTPOSSTARI,    /* 81 Possessified versions, caseless */
1996    /* The assertions must come before BRA, CBRA, ONCE, and COND.*/    OP_NOTPOSPLUSI,    /* 82 */
1997      OP_NOTPOSQUERYI,   /* 83 */
1998    OP_ASSERT,         /* 88 Positive lookahead */    OP_NOTPOSUPTOI,    /* 84 */
1999    OP_ASSERT_NOT,     /* 89 Negative lookahead */  
2000    OP_ASSERTBACK,     /* 90 Positive lookbehind */    /**** Character types ****/
2001    OP_ASSERTBACK_NOT, /* 91 Negative lookbehind */  
2002    OP_REVERSE,        /* 92 Move pointer back - used in lookbehind assertions */    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */
2003      OP_TYPEMINSTAR,    /* 86 these six opcodes must come in pairs, with */
2004    /* ONCE, BRA, CBRA, and COND must come after the assertions, with ONCE first,    OP_TYPEPLUS,       /* 87 the minimizing one second. These codes must */
2005    as there's a test for >= ONCE for a subpattern that isn't an assertion. */    OP_TYPEMINPLUS,    /* 88 be in exactly the same order as those above. */
2006      OP_TYPEQUERY,      /* 89 */
2007    OP_ONCE,           /* 93 Atomic group */    OP_TYPEMINQUERY,   /* 90 */
2008    OP_BRA,            /* 94 Start of non-capturing bracket */  
2009    OP_CBRA,           /* 95 Start of capturing bracket */    OP_TYPEUPTO,       /* 91 From 0 to n matches */
2010    OP_COND,           /* 96 Conditional group */    OP_TYPEMINUPTO,    /* 92 */
2011      OP_TYPEEXACT,      /* 93 Exactly n matches */
2012    
2013      OP_TYPEPOSSTAR,    /* 94 Possessified versions */
2014      OP_TYPEPOSPLUS,    /* 95 */
2015      OP_TYPEPOSQUERY,   /* 96 */
2016      OP_TYPEPOSUPTO,    /* 97 */
2017    
2018      /* These are used for character classes and back references; only the
2019      first six are the same as the sets above. */
2020    
2021      OP_CRSTAR,         /* 98 The maximizing and minimizing versions of */
2022      OP_CRMINSTAR,      /* 99 all these opcodes must come in pairs, with */
2023      OP_CRPLUS,         /* 100 the minimizing one second. These codes must */
2024      OP_CRMINPLUS,      /* 101 be in exactly the same order as those above. */
2025      OP_CRQUERY,        /* 102 */
2026      OP_CRMINQUERY,     /* 103 */
2027    
2028      OP_CRRANGE,        /* 104 These are different to the three sets above. */
2029      OP_CRMINRANGE,     /* 105 */
2030    
2031      /* End of quantifier opcodes */
2032    
2033      OP_CLASS,          /* 106 Match a character class, chars < 256 only */
2034      OP_NCLASS,         /* 107 Same, but the bitmap was created from a negative
2035                                  class - the difference is relevant only when a
2036                                  character > 255 is encountered. */
2037      OP_XCLASS,         /* 108 Extended class for handling > 255 chars within the
2038                                  class. This does both positive and negative. */
2039      OP_REF,            /* 109 Match a back reference, casefully */
2040      OP_REFI,           /* 110 Match a back reference, caselessly */
2041      OP_RECURSE,        /* 111 Match a numbered subpattern (possibly recursive) */
2042      OP_CALLOUT,        /* 112 Call out to external function if provided */
2043    
2044      OP_ALT,            /* 113 Start of alternation */
2045      OP_KET,            /* 114 End of group that doesn't have an unbounded repeat */
2046      OP_KETRMAX,        /* 115 These two must remain together and in this */
2047      OP_KETRMIN,        /* 116 order. They are for groups the repeat for ever. */
2048      OP_KETRPOS,        /* 117 Possessive unlimited repeat. */
2049    
2050      /* The assertions must come before BRA, CBRA, ONCE, and COND, and the four
2051      asserts must remain in order. */
2052    
2053      OP_REVERSE,        /* 118 Move pointer back - used in lookbehind assertions */
2054      OP_ASSERT,         /* 119 Positive lookahead */
2055      OP_ASSERT_NOT,     /* 120 Negative lookahead */
2056      OP_ASSERTBACK,     /* 121 Positive lookbehind */
2057      OP_ASSERTBACK_NOT, /* 122 Negative lookbehind */
2058    
2059      /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately
2060      after the assertions, with ONCE first, as there's a test for >= ONCE for a
2061      subpattern that isn't an assertion. The POS versions must immediately follow
2062      the non-POS versions in each case. */
2063    
2064      OP_ONCE,           /* 123 Atomic group, contains captures */
2065      OP_ONCE_NC,        /* 124 Atomic group containing no captures */
2066      OP_BRA,            /* 125 Start of non-capturing bracket */
2067      OP_BRAPOS,         /* 126 Ditto, with unlimited, possessive repeat */
2068      OP_CBRA,           /* 127 Start of capturing bracket */
2069      OP_CBRAPOS,        /* 128 Ditto, with unlimited, possessive repeat */
2070      OP_COND,           /* 129 Conditional group */
2071    
2072    /* These three must follow the previous three, in the same order. There's a    /* These five must follow the previous five, in the same order. There's a
2073    check for >= SBRA to distinguish the two sets. */    check for >= SBRA to distinguish the two sets. */
2074    
2075    OP_SBRA,           /* 97 Start of non-capturing bracket, check empty  */    OP_SBRA,           /* 130 Start of non-capturing bracket, check empty  */
2076    OP_SCBRA,          /* 98 Start of capturing bracket, check empty */    OP_SBRAPOS,        /* 131 Ditto, with unlimited, possessive repeat */
2077    OP_SCOND,          /* 99 Conditional group, check empty */    OP_SCBRA,          /* 132 Start of capturing bracket, check empty */
2078      OP_SCBRAPOS,       /* 133 Ditto, with unlimited, possessive repeat */
2079      OP_SCOND,          /* 134 Conditional group, check empty */
2080    
2081    /* The next two pairs must (respectively) be kept together. */    /* The next two pairs must (respectively) be kept together. */
2082    
2083    OP_CREF,           /* 100 Used to hold a capture number as condition */    OP_CREF,           /* 135 Used to hold a capture number as condition */
2084    OP_NCREF,          /* 101 Same, but generaged by a name reference*/    OP_NCREF,          /* 136 Same, but generated by a name reference*/
2085    OP_RREF,           /* 102 Used to hold a recursion number as condition */    OP_RREF,           /* 137 Used to hold a recursion number as condition */
2086    OP_NRREF,          /* 103 Same, but generaged by a name reference*/    OP_NRREF,          /* 138 Same, but generated by a name reference*/
2087    OP_DEF,            /* 104 The DEFINE condition */    OP_DEF,            /* 139 The DEFINE condition */
2088    
2089    OP_BRAZERO,        /* 105 These two must remain together and in this */    OP_BRAZERO,        /* 140 These two must remain together and in this */
2090    OP_BRAMINZERO,     /* 106 order. */    OP_BRAMINZERO,     /* 141 order. */
2091      OP_BRAPOSZERO,     /* 142 */
2092    
2093    /* These are backtracking control verbs */    /* These are backtracking control verbs */
2094    
2095    OP_MARK,           /* 107 always has an argument */    OP_MARK,           /* 143 always has an argument */
2096    OP_PRUNE,          /* 108 */    OP_PRUNE,          /* 144 */
2097    OP_PRUNE_ARG,      /* 109 same, but with argument */    OP_PRUNE_ARG,      /* 145 same, but with argument */
2098    OP_SKIP,           /* 110 */    OP_SKIP,           /* 146 */
2099    OP_SKIP_ARG,       /* 111 same, but with argument */    OP_SKIP_ARG,       /* 147 same, but with argument */
2100    OP_THEN,           /* 112 */    OP_THEN,           /* 148 */
2101    OP_THEN_ARG,       /* 113 same, but with argument */    OP_THEN_ARG,       /* 149 same, but with argument */
2102    OP_COMMIT,         /* 114 */    OP_COMMIT,         /* 150 */
2103    
2104    /* These are forced failure and success verbs */    /* These are forced failure and success verbs */
2105    
2106    OP_FAIL,           /* 115 */    OP_FAIL,           /* 151 */
2107    OP_ACCEPT,         /* 116 */    OP_ACCEPT,         /* 152 */
2108    OP_CLOSE,          /* 117 Used before OP_ACCEPT to close open captures */    OP_ASSERT_ACCEPT,  /* 153 Used inside assertions */
2109      OP_CLOSE,          /* 154 Used before OP_ACCEPT to close open captures */
2110    
2111    /* This is used to skip a subpattern with a {0} quantifier */    /* This is used to skip a subpattern with a {0} quantifier */
2112    
2113    OP_SKIPZERO,       /* 118 */    OP_SKIPZERO,       /* 155 */
2114    
2115    /* 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
2116    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 1462  called "coptable" and "poptable" in pcre Line 2125  called "coptable" and "poptable" in pcre
2125    
2126    
2127  /* 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
2128  for debugging. The macro is referenced only in pcre_printint.c. */  for debugging, and some of them are only partial names. The macro is referenced
2129    only in pcre_printint.c, which fills out the full names in many cases (and in
2130    some cases doesn't actually use these names at all). */
2131    
2132  #define OP_NAME_LIST \  #define OP_NAME_LIST \
2133    "End", "\\A", "\\G", "\\K", "\\B", "\\b", "\\D", "\\d",         \    "End", "\\A", "\\G", "\\K", "\\B", "\\b", "\\D", "\\d",         \
2134    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \
2135    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \
2136    "extuni",  "\\Z", "\\z",                                        \    "extuni",  "\\Z", "\\z",                                        \
2137    "Opt", "^", "$", "char", "charnc", "not",                       \    "^", "^", "$", "$", "char", "chari", "not", "noti",             \
2138    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \    "*", "*?", "+", "+?", "?", "??",                                \
2139      "{", "{", "{",                                                  \
2140    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
2141    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \    "*", "*?", "+", "+?", "?", "??",                                \
2142      "{", "{", "{",                                                  \
2143      "*+","++", "?+", "{",                                           \
2144      "*", "*?", "+", "+?", "?", "??",                                \
2145      "{", "{", "{",                                                  \
2146      "*+","++", "?+", "{",                                           \
2147      "*", "*?", "+", "+?", "?", "??",                                \
2148      "{", "{", "{",                                                  \
2149    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
2150    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \
2151    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
2152    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \
2153    "class", "nclass", "xclass", "Ref", "Recurse", "Callout",       \    "class", "nclass", "xclass", "Ref", "Refi",                     \
2154    "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not",     \    "Recurse", "Callout",                                           \
2155    "AssertB", "AssertB not", "Reverse",                            \    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \
2156    "Once", "Bra", "CBra", "Cond", "SBra", "SCBra", "SCond",        \    "Reverse", "Assert", "Assert not", "AssertB", "AssertB not",    \
2157      "Once", "Once_NC",                                              \
2158      "Bra", "BraPos", "CBra", "CBraPos",                             \
2159      "Cond",                                                         \
2160      "SBra", "SBraPos", "SCBra", "SCBraPos",                         \
2161      "SCond",                                                        \
2162    "Cond ref", "Cond nref", "Cond rec", "Cond nrec", "Cond def",   \    "Cond ref", "Cond nref", "Cond rec", "Cond nrec", "Cond def",   \
2163    "Brazero", "Braminzero",                                        \    "Brazero", "Braminzero", "Braposzero",                          \
2164    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \
2165    "*THEN", "*THEN", "*COMMIT", "*FAIL", "*ACCEPT",                \    "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \
2166      "*ACCEPT", "*ASSERT_ACCEPT",                                    \
2167    "Close", "Skip zero"    "Close", "Skip zero"
2168    
2169    
# Line 1505  in UTF-8 mode. The code that uses this t Line 2184  in UTF-8 mode. The code that uses this t
2184    3, 3,                          /* \P, \p                                 */ \    3, 3,                          /* \P, \p                                 */ \
2185    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \
2186    1,                             /* \X                                     */ \    1,                             /* \X                                     */ \
2187    1, 1, 2, 1, 1,                 /* \Z, \z, Opt, ^, $                      */ \    1, 1, 1, 1, 1, 1,              /* \Z, \z, ^, ^M, $, $M                   */ \
2188    2,                             /* Char  - the minimum length             */ \    2,                             /* Char  - the minimum length             */ \
2189    2,                             /* Charnc  - the minimum length           */ \    2,                             /* Chari  - the minimum length            */ \
2190    2,                             /* not                                    */ \    2,                             /* not                                    */ \
2191    /* Positive single-char repeats                            ** These are  */ \    2,                             /* noti                                   */ \
2192    2, 2, 2, 2, 2, 2,              /* *, *?, +, +?, ?, ??      ** minima in  */ \    /* Positive single-char repeats                             ** These are */ \
2193    4, 4, 4,                       /* upto, minupto, exact     ** UTF-8 mode */ \    2, 2, 2, 2, 2, 2,              /* *, *?, +, +?, ?, ??       ** minima in */ \
2194    2, 2, 2, 4,                    /* *+, ++, ?+, upto+                      */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* upto, minupto             ** mode      */ \
2195      2+IMM2_SIZE,                   /* exact                                  */ \
2196      2, 2, 2, 2+IMM2_SIZE,          /* *+, ++, ?+, upto+                      */ \
2197      2, 2, 2, 2, 2, 2,              /* *I, *?I, +I, +?I, ?I, ??I ** UTF-8     */ \
2198      2+IMM2_SIZE, 2+IMM2_SIZE,      /* upto I, minupto I                      */ \
2199      2+IMM2_SIZE,                   /* exact I                                */ \
2200      2, 2, 2, 2+IMM2_SIZE,          /* *+I, ++I, ?+I, upto+I                  */ \
2201    /* Negative single-char repeats - only for chars < 256                   */ \    /* Negative single-char repeats - only for chars < 256                   */ \
2202    2, 2, 2, 2, 2, 2,              /* NOT *, *?, +, +?, ?, ??                */ \    2, 2, 2, 2, 2, 2,              /* NOT *, *?, +, +?, ?, ??                */ \
2203    4, 4, 4,                       /* NOT upto, minupto, exact               */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* NOT upto, minupto                      */ \
2204    2, 2, 2, 4,                    /* Possessive *, +, ?, upto               */ \    2+IMM2_SIZE,                   /* NOT exact                              */ \
2205      2, 2, 2, 2+IMM2_SIZE,          /* Possessive NOT *, +, ?, upto           */ \
2206      2, 2, 2, 2, 2, 2,              /* NOT *I, *?I, +I, +?I, ?I, ??I          */ \
2207      2+IMM2_SIZE, 2+IMM2_SIZE,      /* NOT upto I, minupto I                  */ \
2208      2+IMM2_SIZE,                   /* NOT exact I                            */ \
2209      2, 2, 2, 2+IMM2_SIZE,          /* Possessive NOT *I, +I, ?I, upto I      */ \
2210    /* Positive type repeats                                                 */ \    /* Positive type repeats                                                 */ \
2211    2, 2, 2, 2, 2, 2,              /* Type *, *?, +, +?, ?, ??               */ \    2, 2, 2, 2, 2, 2,              /* Type *, *?, +, +?, ?, ??               */ \
2212    4, 4, 4,                       /* Type upto, minupto, exact              */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* Type upto, minupto                     */ \
2213    2, 2, 2, 4,                    /* Possessive *+, ++, ?+, upto+           */ \    2+IMM2_SIZE,                   /* Type exact                             */ \
2214      2, 2, 2, 2+IMM2_SIZE,          /* Possessive *+, ++, ?+, upto+           */ \
2215    /* Character class & ref repeats                                         */ \    /* Character class & ref repeats                                         */ \
2216    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \
2217    5, 5,                          /* CRRANGE, CRMINRANGE                    */ \    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \
2218   33,                             /* CLASS                                  */ \    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \
2219   33,                             /* NCLASS                                 */ \    1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \
2220    0,                             /* XCLASS - variable length               */ \    0,                             /* XCLASS - variable length               */ \
2221    3,                             /* REF                                    */ \    1+IMM2_SIZE,                   /* REF                                    */ \
2222      1+IMM2_SIZE,                   /* REFI                                   */ \
2223    1+LINK_SIZE,                   /* RECURSE                                */ \    1+LINK_SIZE,                   /* RECURSE                                */ \
2224    2+2*LINK_SIZE,                 /* CALLOUT                                */ \    2+2*LINK_SIZE,                 /* CALLOUT                                */ \
2225    1+LINK_SIZE,                   /* Alt                                    */ \    1+LINK_SIZE,                   /* Alt                                    */ \
2226    1+LINK_SIZE,                   /* Ket                                    */ \    1+LINK_SIZE,                   /* Ket                                    */ \
2227    1+LINK_SIZE,                   /* KetRmax                                */ \    1+LINK_SIZE,                   /* KetRmax                                */ \
2228    1+LINK_SIZE,                   /* KetRmin                                */ \    1+LINK_SIZE,                   /* KetRmin                                */ \
2229      1+LINK_SIZE,                   /* KetRpos                                */ \
2230      1+LINK_SIZE,                   /* Reverse                                */ \
2231    1+LINK_SIZE,                   /* Assert                                 */ \    1+LINK_SIZE,                   /* Assert                                 */ \
2232    1+LINK_SIZE,                   /* Assert not                             */ \    1+LINK_SIZE,                   /* Assert not                             */ \
2233    1+LINK_SIZE,                   /* Assert behind                          */ \    1+LINK_SIZE,                   /* Assert behind                          */ \
2234    1+LINK_SIZE,                   /* Assert behind not                      */ \    1+LINK_SIZE,                   /* Assert behind not                      */ \
   1+LINK_SIZE,                   /* Reverse                                */ \  
2235    1+LINK_SIZE,                   /* ONCE                                   */ \    1+LINK_SIZE,                   /* ONCE                                   */ \
2236      1+LINK_SIZE,                   /* ONCE_NC                                */ \
2237    1+LINK_SIZE,                   /* BRA                                    */ \    1+LINK_SIZE,                   /* BRA                                    */ \
2238    3+LINK_SIZE,                   /* CBRA                                   */ \    1+LINK_SIZE,                   /* BRAPOS                                 */ \
2239      1+LINK_SIZE+IMM2_SIZE,         /* CBRA                                   */ \
2240      1+LINK_SIZE+IMM2_SIZE,         /* CBRAPOS                                */ \
2241    1+LINK_SIZE,                   /* COND                                   */ \    1+LINK_SIZE,                   /* COND                                   */ \
2242    1+LINK_SIZE,                   /* SBRA                                   */ \    1+LINK_SIZE,                   /* SBRA                                   */ \
2243    3+LINK_SIZE,                   /* SCBRA                                  */ \    1+LINK_SIZE,                   /* SBRAPOS                                */ \
2244      1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \
2245      1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \
2246    1+LINK_SIZE,                   /* SCOND                                  */ \    1+LINK_SIZE,                   /* SCOND                                  */ \
2247    3, 3,                          /* CREF, NCREF                            */ \    1+IMM2_SIZE, 1+IMM2_SIZE,      /* CREF, NCREF                            */ \
2248    3, 3,                          /* RREF, NRREF                            */ \    1+IMM2_SIZE, 1+IMM2_SIZE,      /* RREF, NRREF                            */ \
2249    1,                             /* DEF                                    */ \    1,                             /* DEF                                    */ \
2250    1, 1,                          /* BRAZERO, BRAMINZERO                    */ \    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \
2251    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \
2252    1, 3,                          /* SKIP, SKIP_ARG                         */ \    1, 3,                          /* SKIP, SKIP_ARG                         */ \
2253    1+LINK_SIZE, 3+LINK_SIZE,      /* THEN, THEN_ARG                         */ \    1, 3,                          /* THEN, THEN_ARG                         */ \
2254    1, 1, 1, 3, 1                  /* COMMIT, FAIL, ACCEPT, CLOSE, SKIPZERO  */    1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \
2255      1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */
2256    
2257  /* A magic value for OP_RREF and OP_NRREF to indicate the "any recursion"  /* A magic value for OP_RREF and OP_NRREF to indicate the "any recursion"
2258  condition. */  condition. */
# Line 1571  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4, Line 2269  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,
2269         ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,         ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
2270         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
2271         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
2272         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68,         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,
2273         ERRCOUNT };         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERRCOUNT };
2274    
2275    /* JIT compiling modes. The function list is indexed by them. */
2276    enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,
2277           JIT_NUMBER_OF_COMPILE_MODES };
2278    
2279  /* The real format of the start of the pcre block; the index of names and the  /* The real format of the start of the pcre block; the index of names and the
2280  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
# Line 1591  fields are present. Currently PCRE alway Line 2293  fields are present. Currently PCRE alway
2293  NOTE NOTE NOTE  NOTE NOTE NOTE
2294  */  */
2295    
2296  typedef struct real_pcre {  #if defined COMPILE_PCRE8
2297    #define REAL_PCRE real_pcre
2298    #elif defined COMPILE_PCRE16
2299    #define REAL_PCRE real_pcre16
2300    #elif defined COMPILE_PCRE32
2301    #define REAL_PCRE real_pcre32
2302    #endif
2303    
2304    /* It is necessary to fork the struct for 32 bit, since it needs to use
2305     * pcre_uchar for first_char and req_char. Can't put an ifdef inside the
2306     * typedef since pcretest needs access to  the struct of the 8-, 16-
2307     * and 32-bit variants. */
2308    
2309    typedef struct real_pcre8_or_16 {
2310    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2311    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2312    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2313    pcre_uint16 flags;              /* Private flags */    pcre_uint16 flags;              /* Private flags */
2314    pcre_uint16 dummy1;             /* For future use */    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2315    pcre_uint16 top_bracket;    pcre_uint16 top_bracket;        /* Highest numbered group */
2316    pcre_uint16 top_backref;    pcre_uint16 top_backref;        /* Highest numbered back reference */
2317    pcre_uint16 first_byte;    pcre_uint16 first_char;         /* Starting character */
2318    pcre_uint16 req_byte;    pcre_uint16 req_char;           /* This character must be seen */
2319    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2320    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2321    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2322    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2323      const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2324      const pcre_uint8 *nullpad;      /* NULL padding */
2325    } real_pcre8_or_16;
2326    
2327    const unsigned char *tables;    /* Pointer to tables or NULL for std */  typedef struct real_pcre8_or_16 real_pcre;
2328    const unsigned char *nullpad;   /* NULL padding */  typedef struct real_pcre8_or_16 real_pcre16;
2329  } real_pcre;  
2330    typedef struct real_pcre32 {
2331      pcre_uint32 magic_number;
2332      pcre_uint32 size;               /* Total that was malloced */
2333      pcre_uint32 options;            /* Public options */
2334      pcre_uint16 flags;              /* Private flags */
2335      pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2336      pcre_uint16 top_bracket;        /* Highest numbered group */
2337      pcre_uint16 top_backref;        /* Highest numbered back reference */
2338      pcre_uint32 first_char;         /* Starting character */
2339      pcre_uint32 req_char;           /* This character must be seen */
2340      pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2341      pcre_uint16 name_entry_size;    /* Size of any name items */
2342      pcre_uint16 name_count;         /* Number of name items */
2343      pcre_uint16 ref_count;          /* Reference count */
2344      pcre_uint16 dummy1;             /* for later expansion */
2345      pcre_uint16 dummy2;             /* for later expansion */
2346      const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2347      void *nullpad;                  /* for later expansion */
2348    } real_pcre32;
2349    
2350    /* Assert that the size of REAL_PCRE is divisible by 8 */
2351    typedef int __assert_real_pcre_size_divisible_8[(sizeof(REAL_PCRE) % 8) == 0 ? 1 : -1];
2352    
2353    /* Needed in pcretest to access some fields in the real_pcre* structures
2354     * directly. They're unified for 8/16/32 bits since the structs only differ
2355     * after these fields; if that ever changes, need to fork those defines into
2356     * 8/16 and 32 bit versions. */
2357    #define REAL_PCRE_MAGIC(re)     (((REAL_PCRE*)re)->magic_number)
2358    #define REAL_PCRE_SIZE(re)      (((REAL_PCRE*)re)->size)
2359    #define REAL_PCRE_OPTIONS(re)   (((REAL_PCRE*)re)->options)
2360    #define REAL_PCRE_FLAGS(re)     (((REAL_PCRE*)re)->flags)
2361    
2362  /* The format of the block used to store data from pcre_study(). The same  /* The format of the block used to store data from pcre_study(). The same
2363  remark (see NOTE above) about extending this structure applies. */  remark (see NOTE above) about extending this structure applies. */
# Line 1616  remark (see NOTE above) about extending Line 2365  remark (see NOTE above) about extending
2365  typedef struct pcre_study_data {  typedef struct pcre_study_data {
2366    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2367    pcre_uint32 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2368    uschar start_bits[32];          /* Starting char bits */    pcre_uint8 start_bits[32];      /* Starting char bits */
2369    pcre_uint32 minlength;          /* Minimum subject length */    pcre_uint32 minlength;          /* Minimum subject length */
2370  } pcre_study_data;  } pcre_study_data;
2371    
# Line 1635  typedef struct open_capitem { Line 2384  typedef struct open_capitem {
2384  doing the compiling, so that they are thread-safe. */  doing the compiling, so that they are thread-safe. */
2385    
2386  typedef struct compile_data {  typedef struct compile_data {
2387    const uschar *lcc;            /* Points to lower casing table */    const pcre_uint8 *lcc;            /* Points to lower casing table */
2388    const uschar *fcc;            /* Points to case-flipping table */    const pcre_uint8 *fcc;            /* Points to case-flipping table */
2389    const uschar *cbits;          /* Points to character type table */    const pcre_uint8 *cbits;          /* Points to character type table */
2390    const uschar *ctypes;         /* Points to table of type maps */    const pcre_uint8 *ctypes;         /* Points to table of type maps */
2391    const uschar *start_workspace;/* The start of working space */    const pcre_uchar *start_workspace;/* The start of working space */
2392    const uschar *start_code;     /* The start of the compiled code */    const pcre_uchar *start_code;     /* The start of the compiled code */
2393    const uschar *start_pattern;  /* The start of the pattern */    const pcre_uchar *start_pattern;  /* The start of the pattern */
2394    const uschar *end_pattern;    /* The end of the pattern */    const pcre_uchar *end_pattern;    /* The end of the pattern */
2395    open_capitem *open_caps;      /* Chain of open capture items */    open_capitem *open_caps;          /* Chain of open capture items */
2396    uschar *hwm;                  /* High watermark of workspace */    pcre_uchar *hwm;                  /* High watermark of workspace */
2397    uschar *name_table;           /* The name/number table */    pcre_uchar *name_table;           /* The name/number table */
2398    int  names_found;             /* Number of entries so far */    int  names_found;                 /* Number of entries so far */
2399    int  name_entry_size;         /* Size of each entry */    int  name_entry_size;             /* Size of each entry */
2400    int  bracount;                /* Count of capturing parens as we compile */    int  workspace_size;              /* Size of workspace */
2401    int  final_bracount;          /* Saved value after first pass */    unsigned int  bracount;           /* Count of capturing parens as we compile */
2402    int  top_backref;             /* Maximum back reference */    int  final_bracount;              /* Saved value after first pass */
2403    unsigned int backref_map;     /* Bitmap of low back refs */    int  max_lookbehind;              /* Maximum lookbehind (characters) */
2404    int  external_options;        /* External (initial) options */    int  top_backref;                 /* Maximum back reference */
2405    int  external_flags;          /* External flag bits to be set */    unsigned int backref_map;         /* Bitmap of low back refs */
2406    int  req_varyopt;             /* "After variable item" flag for reqbyte */    int  assert_depth;                /* Depth of nested assertions */
2407    BOOL had_accept;              /* (*ACCEPT) encountered */    int  external_options;            /* External (initial) options */
2408    BOOL check_lookbehind;        /* Lookbehinds need later checking */    int  external_flags;              /* External flag bits to be set */
2409    int  nltype;                  /* Newline type */    int  req_varyopt;                 /* "After variable item" flag for reqbyte */
2410    int  nllen;                   /* Newline string length */    BOOL had_accept;                  /* (*ACCEPT) encountered */
2411    uschar nl[4];                 /* Newline string when fixed length */    BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */
2412      BOOL check_lookbehind;            /* Lookbehinds need later checking */
2413      int  nltype;                      /* Newline type */
2414      int  nllen;                       /* Newline string length */
2415      pcre_uchar nl[4];                 /* Newline string when fixed length */
2416  } compile_data;  } compile_data;
2417    
2418  /* Structure for maintaining a chain of pointers to the currently incomplete  /* Structure for maintaining a chain of pointers to the currently incomplete
2419  branches, for testing for left recursion. */  branches, for testing for left recursion while compiling. */
2420    
2421  typedef struct branch_chain {  typedef struct branch_chain {
2422    struct branch_chain *outer;    struct branch_chain *outer;
2423    uschar *current_branch;    pcre_uchar *current_branch;
2424  } branch_chain;  } branch_chain;
2425    
2426  /* Structure for items in a linked list that represents an explicit recursive  /* Structure for items in a linked list that represents an explicit recursive
2427  call within the pattern. */  call within the pattern; used by pcre_exec(). */
2428    
2429  typedef struct recursion_info {  typedef struct recursion_info {
2430    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
2431    int group_num;                /* Number of group that was called */    unsigned int group_num;         /* Number of group that was called */
2432    const uschar *after_call;     /* "Return value": points after the call in the expr */    int *offset_save;               /* Pointer to start of saved offsets */
2433    int *offset_save;             /* Pointer to start of saved offsets */    int saved_max;                  /* Number of saved offsets */
2434    int saved_max;                /* Number of saved offsets */    PCRE_PUCHAR subject_position;   /* Position at start of recursion */
   int save_offset_top;          /* Current value of offset_top */  
2435  } recursion_info;  } recursion_info;
2436    
2437    /* A similar structure for pcre_dfa_exec(). */
2438    
2439    typedef struct dfa_recursion_info {
2440      struct dfa_recursion_info *prevrec;
2441      int group_num;
2442      PCRE_PUCHAR subject_position;
2443    } dfa_recursion_info;
2444    
2445  /* Structure for building a chain of data for holding the values of the subject  /* Structure for building a chain of data for holding the values of the subject
2446  pointer at the start of each subpattern, so as to detect when an empty string  pointer at the start of each subpattern, so as to detect when an empty string
2447  has been matched by a subpattern - to break infinite loops. */  has been matched by a subpattern - to break infinite loops; used by
2448    pcre_exec(). */
2449    
2450  typedef struct eptrblock {  typedef struct eptrblock {
2451    struct eptrblock *epb_prev;    struct eptrblock *epb_prev;
2452    USPTR epb_saved_eptr;    PCRE_PUCHAR epb_saved_eptr;
2453  } eptrblock;  } eptrblock;
2454    
2455    
# Line 1699  typedef struct match_data { Line 2460  typedef struct match_data {
2460    unsigned long int match_call_count;      /* As it says */    unsigned long int match_call_count;      /* As it says */
2461    unsigned long int match_limit;           /* As it says */    unsigned long int match_limit;           /* As it says */
2462    unsigned long int match_limit_recursion; /* As it says */    unsigned long int match_limit_recursion; /* As it says */
2463    int   *offset_vector;         /* Offset vector */    int   *offset_vector;           /* Offset vector */
2464    int    offset_end;            /* One past the end */    int    offset_end;              /* One past the end */
2465    int    offset_max;            /* The maximum usable for return data */    int    offset_max;              /* The maximum usable for return data */
2466    int    nltype;                /* Newline type */    int    nltype;                  /* Newline type */
2467    int    nllen;                 /* Newline string length */    int    nllen;                   /* Newline string length */
2468    int    name_count;            /* Number of names in name table */    int    name_count;              /* Number of names in name table */
2469    int    name_entry_size;       /* Size of entry in names table */    int    name_entry_size;         /* Size of entry in names table */
2470    uschar *name_table;           /* Table of names */    pcre_uchar *name_table;         /* Table of names */
2471    uschar nl[4];                 /* Newline string when fixed */    pcre_uchar nl[4];               /* Newline string when fixed */
2472    const uschar *lcc;            /* Points to lower casing table */    const  pcre_uint8 *lcc;         /* Points to lower casing table */
2473    const uschar *ctypes;         /* Points to table of type maps */    const  pcre_uint8 *fcc;         /* Points to case-flipping table */
2474    BOOL   offset_overflow;       /* Set if too many extractions */    const  pcre_uint8 *ctypes;      /* Points to table of type maps */
2475    BOOL   notbol;                /* NOTBOL flag */    BOOL   offset_overflow;         /* Set if too many extractions */
2476    BOOL   noteol;                /* NOTEOL flag */    BOOL   notbol;                  /* NOTBOL flag */
2477    BOOL   utf8;                  /* UTF8 flag */    BOOL   noteol;                  /* NOTEOL flag */
2478    BOOL   jscript_compat;        /* JAVASCRIPT_COMPAT flag */    BOOL   utf;                     /* UTF-8 / UTF-16 flag */
2479    BOOL   use_ucp;               /* PCRE_UCP flag */    BOOL   jscript_compat;          /* JAVASCRIPT_COMPAT flag */
2480    BOOL   endonly;               /* Dollar not before final \n */    BOOL   use_ucp;                 /* PCRE_UCP flag */
2481    BOOL   notempty;              /* Empty string match not wanted */    BOOL   endonly;                 /* Dollar not before final \n */
2482    BOOL   notempty_atstart;      /* Empty string match at start not wanted */    BOOL   notempty;                /* Empty string match not wanted */
2483    BOOL   hitend;                /* Hit the end of the subject at some point */    BOOL   notempty_atstart;        /* Empty string match at start not wanted */
2484    BOOL   bsr_anycrlf;           /* \R is just any CRLF, not full Unicode */    BOOL   hitend;                  /* Hit the end of the subject at some point */
2485    const uschar *start_code;     /* For use when recursing */    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */
2486    USPTR  start_subject;         /* Start of the subject string */    BOOL   hasthen;                 /* Pattern contains (*THEN) */
2487    USPTR  end_subject;           /* End of the subject string */    BOOL   ignore_skip_arg;         /* For re-run when SKIP name not found */
2488    USPTR  start_match_ptr;       /* Start of matched string */    const  pcre_uchar *start_code;  /* For use when recursing */
2489    USPTR  end_match_ptr;         /* Subject position at end match */    PCRE_PUCHAR start_subject;      /* Start of the subject string */
2490    USPTR  start_used_ptr;        /* Earliest consulted character */    PCRE_PUCHAR end_subject;        /* End of the subject string */
2491    int    partial;               /* PARTIAL options */    PCRE_PUCHAR start_match_ptr;    /* Start of matched string */
2492    int    end_offset_top;        /* Highwater mark at end of match */    PCRE_PUCHAR end_match_ptr;      /* Subject position at end match */
2493    int    capture_last;          /* Most recent capture number */    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */
2494    int    start_offset;          /* The start offset value */    int    partial;                 /* PARTIAL options */
2495    eptrblock *eptrchain;         /* Chain of eptrblocks for tail recursions */    int    end_offset_top;          /* Highwater mark at end of match */
2496    int    eptrn;                 /* Next free eptrblock */    int    capture_last;            /* Most recent capture number */
2497    recursion_info *recursive;    /* Linked list of recursion data */    int    start_offset;            /* The start offset value */
2498    void  *callout_data;          /* To pass back to callouts */    int    match_function_type;     /* Set for certain special calls of MATCH() */
2499    const uschar *mark;           /* Mark pointer to pass back */    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */
2500      int    eptrn;                   /* Next free eptrblock */
2501      recursion_info *recursive;      /* Linked list of recursion data */
2502      void  *callout_data;            /* To pass back to callouts */
2503      const  pcre_uchar *mark;        /* Mark pointer to pass back on success */
2504      const  pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */
2505      const  pcre_uchar *once_target; /* Where to back up to for atomic groups */
2506    #ifdef NO_RECURSE
2507      void  *match_frames_base;       /* For remembering malloc'd frames */
2508    #endif
2509  } match_data;  } match_data;
2510    
2511  /* A similar structure is used for the same purpose by the DFA matching  /* A similar structure is used for the same purpose by the DFA matching
2512  functions. */  functions. */
2513    
2514  typedef struct dfa_match_data {  typedef struct dfa_match_data {
2515    const uschar *start_code;     /* Start of the compiled pattern */    const pcre_uchar *start_code;     /* Start of the compiled pattern */
2516    const uschar *start_subject;  /* Start of the subject string */    const pcre_uchar *start_subject ; /* Start of the subject string */
2517    const uschar *end_subject;    /* End of subject string */    const pcre_uchar *end_subject;    /* End of subject string */
2518    const uschar *start_used_ptr; /* Earliest consulted character */    const pcre_uchar *start_used_ptr; /* Earliest consulted character */
2519    const uschar *tables;         /* Character tables */    const pcre_uint8 *tables;         /* Character tables */
2520    int   start_offset;           /* The start offset value */    int   start_offset;               /* The start offset value */
2521    int   moptions;               /* Match options */    int   moptions;                   /* Match options */
2522    int   poptions;               /* Pattern options */    int   poptions;                   /* Pattern options */
2523    int    nltype;                /* Newline type */    int   nltype;                     /* Newline type */
2524    int    nllen;                 /* Newline string length */    int   nllen;                      /* Newline string length */
2525    uschar nl[4];                 /* Newline string when fixed */    pcre_uchar nl[4];                 /* Newline string when fixed */
2526    void  *callout_data;          /* To pass back to callouts */    void *callout_data;               /* To pass back to callouts */
2527      dfa_recursion_info *recursive;    /* Linked list of recursion data */
2528  } dfa_match_data;  } dfa_match_data;
2529    
2530  /* Bit definitions for entries in the pcre_ctypes table. */  /* Bit definitions for entries in the pcre_ctypes table. */
# Line 1789  total length. */ Line 2560  total length. */
2560  #define ctypes_offset (cbits_offset + cbit_length)  #define ctypes_offset (cbits_offset + cbit_length)
2561  #define tables_length (ctypes_offset + 256)  #define tables_length (ctypes_offset + 256)
2562    
2563    /* Internal function and data prefixes. */
2564    
2565    #if defined COMPILE_PCRE8
2566    #ifndef PUBL
2567    #define PUBL(name) pcre_##name
2568    #endif
2569    #ifndef PRIV
2570    #define PRIV(name) _pcre_##name
2571    #endif
2572    #elif defined COMPILE_PCRE16
2573    #ifndef PUBL
2574    #define PUBL(name) pcre16_##name
2575    #endif
2576    #ifndef PRIV
2577    #define PRIV(name) _pcre16_##name
2578    #endif
2579    #elif defined COMPILE_PCRE32
2580    #ifndef PUBL
2581    #define PUBL(name) pcre32_##name
2582    #endif
2583    #ifndef PRIV
2584    #define PRIV(name) _pcre32_##name
2585    #endif
2586    #else
2587    #error Unsupported compiling mode
2588    #endif /* COMPILE_PCRE[8|16|32] */
2589    
2590  /* Layout of the UCP type table that translates property names into types and  /* Layout of the UCP type table that translates property names into types and
2591  codes. Each entry used to point directly to a name, but to reduce the number of  codes. Each entry used to point directly to a name, but to reduce the number of
2592  relocations in shared libraries, it now has an offset into a single string  relocations in shared libraries, it now has an offset into a single string
# Line 1806  of the exported public functions. They h Line 2604  of the exported public functions. They h
2604  but are not part of the PCRE public API. The data for these tables is in the  but are not part of the PCRE public API. The data for these tables is in the
2605  pcre_tables.c module. */  pcre_tables.c module. */
2606    
2607  extern const int    _pcre_utf8_table1[];  #ifdef COMPILE_PCRE8
2608  extern const int    _pcre_utf8_table2[];  extern const int            PRIV(utf8_table1)[];
2609  extern const int    _pcre_utf8_table3[];  extern const int            PRIV(utf8_table1_size);
2610  extern const uschar _pcre_utf8_table4[];  extern const int            PRIV(utf8_table2)[];
2611    extern const int            PRIV(utf8_table3)[];
2612  extern const int    _pcre_utf8_table1_size;  extern const pcre_uint8     PRIV(utf8_table4)[];
2613    #endif /* COMPILE_PCRE8 */
2614    
2615    extern const char           PRIV(utt_names)[];
2616    extern const ucp_type_table PRIV(utt)[];
2617    extern const int            PRIV(utt_size);
2618    
2619  extern const char   _pcre_utt_names[];  extern const pcre_uint8     PRIV(OP_lengths)[];
2620  extern const ucp_type_table _pcre_utt[];  extern const pcre_uint8     PRIV(default_tables)[];
 extern const int _pcre_utt_size;  
2621    
2622  extern const uschar _pcre_default_tables[];  extern const pcre_uint32    PRIV(hspace_list)[];
2623    extern const pcre_uint32    PRIV(vspace_list)[];
 extern const uschar _pcre_OP_lengths[];  
2624    
2625    
2626  /* Internal shared functions. These are functions that are used by more than  /* Internal shared functions. These are functions that are used by more than
2627  one of the exported public functions. They have to be "external" in the C  one of the exported public functions. They have to be "external" in the C
2628  sense, but are not part of the PCRE public API. */  sense, but are not part of the PCRE public API. */
2629    
2630  extern const uschar *_pcre_find_bracket(const uschar *, BOOL, int);  /* String comparison functions. */
2631  extern BOOL          _pcre_is_newline(USPTR, int, USPTR, int *, BOOL);  #if defined COMPILE_PCRE8
 extern int           _pcre_ord2utf8(int, uschar *);  
 extern real_pcre    *_pcre_try_flipped(const real_pcre *, real_pcre *,  
                        const pcre_study_data *, pcre_study_data *);  
 extern int           _pcre_valid_utf8(USPTR, int);  
 extern BOOL          _pcre_was_newline(USPTR, int, USPTR, int *, BOOL);  
 extern BOOL          _pcre_xclass(int, const uschar *);  
2632    
2633    #define STRCMP_UC_UC(str1, str2) \
2634      strcmp((char *)(str1), (char *)(str2))
2635    #define STRCMP_UC_C8(str1, str2) \
2636      strcmp((char *)(str1), (str2))
2637    #define STRNCMP_UC_UC(str1, str2, num) \
2638      strncmp((char *)(str1), (char *)(str2), (num))
2639    #define STRNCMP_UC_C8(str1, str2, num) \
2640      strncmp((char *)(str1), (str2), (num))
2641    #define STRLEN_UC(str) strlen((const char *)str)
2642    
2643    #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2644    
2645    extern int               PRIV(strcmp_uc_uc)(const pcre_uchar *,
2646                               const pcre_uchar *);
2647    extern int               PRIV(strcmp_uc_c8)(const pcre_uchar *,
2648                               const char *);
2649    extern int               PRIV(strncmp_uc_uc)(const pcre_uchar *,
2650                               const pcre_uchar *, unsigned int num);
2651    extern int               PRIV(strncmp_uc_c8)(const pcre_uchar *,
2652                               const char *, unsigned int num);
2653    extern unsigned int      PRIV(strlen_uc)(const pcre_uchar *str);
2654    
2655    #define STRCMP_UC_UC(str1, str2) \
2656      PRIV(strcmp_uc_uc)((str1), (str2))
2657    #define STRCMP_UC_C8(str1, str2) \
2658      PRIV(strcmp_uc_c8)((str1), (str2))
2659    #define STRNCMP_UC_UC(str1, str2, num) \
2660      PRIV(strncmp_uc_uc)((str1), (str2), (num))
2661    #define STRNCMP_UC_C8(str1, str2, num) \
2662      PRIV(strncmp_uc_c8)((str1), (str2), (num))
2663    #define STRLEN_UC(str) PRIV(strlen_uc)(str)
2664    
2665    #endif /* COMPILE_PCRE[8|16|32] */
2666    
2667    #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16
2668    
2669    #define STRCMP_UC_UC_TEST(str1, str2) STRCMP_UC_UC(str1, str2)
2670    #define STRCMP_UC_C8_TEST(str1, str2) STRCMP_UC_C8(str1, str2)
2671    
2672    #elif defined COMPILE_PCRE32
2673    
2674    extern int               PRIV(strcmp_uc_uc_utf)(const pcre_uchar *,
2675                               const pcre_uchar *);
2676    extern int               PRIV(strcmp_uc_c8_utf)(const pcre_uchar *,
2677                               const char *);
2678    
2679    #define STRCMP_UC_UC_TEST(str1, str2) \
2680      (utf ? PRIV(strcmp_uc_uc_utf)((str1), (str2)) : PRIV(strcmp_uc_uc)((str1), (str2)))
2681    #define STRCMP_UC_C8_TEST(str1, str2) \
2682      (utf ? PRIV(strcmp_uc_c8_utf)((str1), (str2)) : PRIV(strcmp_uc_c8)((str1), (str2)))
2683    
2684    #endif /* COMPILE_PCRE[8|16|32] */
2685    
2686    extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);
2687    extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2688                               int *, BOOL);
2689    extern unsigned int      PRIV(ord2utf)(pcre_uint32, pcre_uchar *);
2690    extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);
2691    extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2692                               int *, BOOL);
2693    extern BOOL              PRIV(xclass)(pcre_uint32, const pcre_uchar *, BOOL);
2694    
2695    #ifdef SUPPORT_JIT
2696    extern void              PRIV(jit_compile)(const REAL_PCRE *,
2697                               PUBL(extra) *, int);
2698    extern int               PRIV(jit_exec)(const REAL_PCRE *, const PUBL(extra) *,
2699                               const pcre_uchar *, int, int, int, int *, int);
2700    extern void              PRIV(jit_free)(void *);
2701    extern int               PRIV(jit_get_size)(void *);
2702    extern const char*       PRIV(jit_get_target)(void);
2703    #endif
2704    
2705  /* Unicode character database (UCD) */  /* Unicode character database (UCD) */
2706    
2707  typedef struct {  typedef struct {
2708    uschar script;    pcre_uint8 script;     /* ucp_Arabic, etc. */
2709    uschar chartype;    pcre_uint8 chartype;   /* ucp_Cc, etc. (general categories) */
2710    pcre_int32 other_case;    pcre_uint8 gbprop;     /* ucp_gbControl, etc. (grapheme break property) */
2711      pcre_uint8 caseset;    /* offset to multichar other cases or zero */
2712      pcre_int32 other_case; /* offset to other case, or zero if none */
2713  } ucd_record;  } ucd_record;
2714    
2715  extern const ucd_record  _pcre_ucd_records[];  extern const pcre_uint32 PRIV(ucd_caseless_sets)[];
2716  extern const uschar      _pcre_ucd_stage1[];  extern const ucd_record  PRIV(ucd_records)[];
2717  extern const pcre_uint16 _pcre_ucd_stage2[];  extern const pcre_uint8  PRIV(ucd_stage1)[];
2718  extern const int         _pcre_ucp_gentype[];  extern const pcre_uint16 PRIV(ucd_stage2)[];
2719    extern const pcre_uint32 PRIV(ucp_gentype)[];
2720    extern const pcre_uint32 PRIV(ucp_gbtable)[];
2721    #ifdef SUPPORT_JIT
2722    extern const int         PRIV(ucp_typerange)[];
2723    #endif
2724    
2725    #ifdef SUPPORT_UCP
2726  /* UCD access macros */  /* UCD access macros */
2727    
2728  #define UCD_BLOCK_SIZE 128  #define UCD_BLOCK_SIZE 128
2729  #define GET_UCD(ch) (_pcre_ucd_records + \  #define GET_UCD(ch) (PRIV(ucd_records) + \
2730          _pcre_ucd_stage2[_pcre_ucd_stage1[(ch) / UCD_BLOCK_SIZE] * \          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(int)(ch) / UCD_BLOCK_SIZE] * \
2731          UCD_BLOCK_SIZE + ch % UCD_BLOCK_SIZE])          UCD_BLOCK_SIZE + (int)(ch) % UCD_BLOCK_SIZE])
2732    
2733  #define UCD_CHARTYPE(ch)  GET_UCD(ch)->chartype  #define UCD_CHARTYPE(ch)    GET_UCD(ch)->chartype
2734  #define UCD_SCRIPT(ch)    GET_UCD(ch)->script  #define UCD_SCRIPT(ch)      GET_UCD(ch)->script
2735  #define UCD_CATEGORY(ch)  _pcre_ucp_gentype[UCD_CHARTYPE(ch)]  #define UCD_CATEGORY(ch)    PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]
2736  #define UCD_OTHERCASE(ch) (ch + GET_UCD(ch)->other_case)  #define UCD_GRAPHBREAK(ch)  GET_UCD(ch)->gbprop
2737    #define UCD_CASESET(ch)     GET_UCD(ch)->caseset
2738    #define UCD_OTHERCASE(ch)   ((pcre_uint32)((int)ch + (int)(GET_UCD(ch)->other_case)))
2739    
2740    #endif /* SUPPORT_UCP */
2741    
2742  #endif  #endif
2743    

Legend:
Removed from v.576  
changed lines
  Added in v.1145

  ViewVC Help
Powered by ViewVC 1.1.5