/[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 602 by ph10, Wed May 25 08:29:03 2011 UTC revision 1033 by ph10, Mon Sep 10 11:02:48 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_" or "_pcre16_" depending on
44    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. */
56  script prevents both being selected, but not everybody uses "configure". */  #ifndef COMPILE_PCRE16
57    #define COMPILE_PCRE8
 #if defined EBCDIC && defined SUPPORT_UTF8  
 #error The use of both EBCDIC and SUPPORT_UTF8 is not supported.  
58  #endif  #endif
59    
60  /* If SUPPORT_UCP is defined, SUPPORT_UTF8 must also be defined. The  /* If SUPPORT_UCP is defined, SUPPORT_UTF must also be defined. The
61  "configure" script ensures this, but not everybody uses "configure". */  "configure" script ensures this, but not everybody uses "configure". */
62    
63  #if defined SUPPORT_UCP && !defined SUPPORT_UTF8  #if defined SUPPORT_UCP && !(defined SUPPORT_UTF)
64    #define SUPPORT_UTF 1
65    #endif
66    
67    /* We define SUPPORT_UTF if SUPPORT_UTF8 is enabled for compatibility
68    reasons with existing code. */
69    
70    #if defined SUPPORT_UTF8 && !(defined SUPPORT_UTF)
71    #define SUPPORT_UTF 1
72    #endif
73    
74    /* Fixme: SUPPORT_UTF8 should be eventually disappear from the code.
75    Until then we define it if SUPPORT_UTF is defined. */
76    
77    #if defined SUPPORT_UTF && !(defined SUPPORT_UTF8)
78  #define SUPPORT_UTF8 1  #define SUPPORT_UTF8 1
79  #endif  #endif
80    
81    /* We do not support both EBCDIC and UTF-8/16 at the same time. The "configure"
82    script prevents both being selected, but not everybody uses "configure". */
83    
84    #if defined EBCDIC && defined SUPPORT_UTF
85    #error The use of both EBCDIC and SUPPORT_UTF8/16 is not supported.
86    #endif
87    
88  /* 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
89  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
90  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 178  set, we ensure here that it has no effec
178  #define PCRE_CALL_CONVENTION  #define PCRE_CALL_CONVENTION
179  #endif  #endif
180    
181  /* 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
182  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
183  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
184  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
185  preprocessor time in standard C environments. */  preprocessor time in standard C environments. */
186    
187    typedef unsigned char pcre_uint8;
188    
189  #if USHRT_MAX == 65535  #if USHRT_MAX == 65535
190    typedef unsigned short pcre_uint16;    typedef unsigned short pcre_uint16;
191    typedef short pcre_int16;    typedef short pcre_int16;
# Line 206  by "configure". */ Line 228  by "configure". */
228    
229  /* All character handling must be done as unsigned characters. Otherwise there  /* All character handling must be done as unsigned characters. Otherwise there
230  are problems with top-bit-set characters and functions such as isspace().  are problems with top-bit-set characters and functions such as isspace().
231  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 *,
232  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
233  to save lots of typing. I tried "uchar", but it causes problems on Digital  called pcre_uchar.
234  Unix, where it is defined in sys/types, so use "uschar" instead. */  
235    The IN_UCHARS macro multiply its argument with the byte size of the current
236    pcre_uchar type. Useful for memcpy and such operations, whose require the
237    byte size of their input/output buffers.
238    
239    The MAX_255 macro checks whether its pcre_uchar input is less than 256.
240    
241    The TABLE_GET macro is designed for accessing elements of tables whose contain
242    exactly 256 items. When the character is able to contain more than 256
243    items, some check is needed before accessing these tables.
244    */
245    
246    #ifdef COMPILE_PCRE8
247    
248    typedef unsigned char pcre_uchar;
249    #define IN_UCHARS(x) (x)
250    #define MAX_255(c) 1
251    #define TABLE_GET(c, table, default) ((table)[c])
252    
253    #else
254    
255    #ifdef COMPILE_PCRE16
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 unsigned char uschar;  typedef pcre_uint16 pcre_uchar;
263    #define IN_UCHARS(x) ((x) << 1)
264    #define MAX_255(c) ((c) <= 255u)
265    #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
266    
267    #else
268    #error Unsupported compiling mode
269    #endif /* COMPILE_PCRE16 */
270    
271    #endif /* COMPILE_PCRE8 */
272    
273  /* 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
274  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 291  start/end of string field names are. */
291  #define IS_NEWLINE(p) \  #define IS_NEWLINE(p) \
292    ((NLBLOCK->nltype != NLTYPE_FIXED)? \    ((NLBLOCK->nltype != NLTYPE_FIXED)? \
293      ((p) < NLBLOCK->PSEND && \      ((p) < NLBLOCK->PSEND && \
294       _pcre_is_newline((p), NLBLOCK->nltype, NLBLOCK->PSEND, &(NLBLOCK->nllen),\       PRIV(is_newline)((p), NLBLOCK->nltype, NLBLOCK->PSEND, \
295         utf8)) \         &(NLBLOCK->nllen), utf)) \
296      : \      : \
297      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
298       (p)[0] == NLBLOCK->nl[0] && \       (p)[0] == NLBLOCK->nl[0] && \
# Line 248  start/end of string field names are. */ Line 305  start/end of string field names are. */
305  #define WAS_NEWLINE(p) \  #define WAS_NEWLINE(p) \
306    ((NLBLOCK->nltype != NLTYPE_FIXED)? \    ((NLBLOCK->nltype != NLTYPE_FIXED)? \
307      ((p) > NLBLOCK->PSSTART && \      ((p) > NLBLOCK->PSSTART && \
308       _pcre_was_newline((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \       PRIV(was_newline)((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \
309         &(NLBLOCK->nllen), utf8)) \         &(NLBLOCK->nllen), utf)) \
310      : \      : \
311      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
312       (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \       (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \
# Line 267  used for the external interface and appe Line 324  used for the external interface and appe
324  must begin with PCRE_. */  must begin with PCRE_. */
325    
326  #ifdef CUSTOM_SUBJECT_PTR  #ifdef CUSTOM_SUBJECT_PTR
327  #define PCRE_SPTR CUSTOM_SUBJECT_PTR  #define PCRE_PUCHAR CUSTOM_SUBJECT_PTR
 #define USPTR CUSTOM_SUBJECT_PTR  
328  #else  #else
329  #define PCRE_SPTR const char *  #define PCRE_PUCHAR const pcre_uchar *
 #define USPTR const unsigned char *  
330  #endif  #endif
331    
   
   
332  /* Include the public PCRE header and the definitions of UCP character property  /* Include the public PCRE header and the definitions of UCP character property
333  values. */  values. */
334    
# Line 343  The macros are controlled by the value o Line 396  The macros are controlled by the value o
396  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
397  is automated on Unix systems via the "configure" command. */  is automated on Unix systems via the "configure" command. */
398    
399    #ifdef COMPILE_PCRE8
400    
401  #if LINK_SIZE == 2  #if LINK_SIZE == 2
402    
403  #define PUT(a,n,d)   \  #define PUT(a,n,d)   \
# Line 379  is automated on Unix systems via the "co Line 434  is automated on Unix systems via the "co
434  #define GET(a,n) \  #define GET(a,n) \
435    (((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])
436    
437  #define MAX_PATTERN_SIZE (1 << 30)   /* Keep it positive */  /* Keep it positive */
438    #define MAX_PATTERN_SIZE (1 << 30)
439    
440    #else
441    #error LINK_SIZE must be either 2, 3, or 4
442    #endif
443    
444    #else /* COMPILE_PCRE8 */
445    
446    #ifdef COMPILE_PCRE16
447    
448    #if LINK_SIZE == 2
449    
450    #undef LINK_SIZE
451    #define LINK_SIZE 1
452    
453    #define PUT(a,n,d)   \
454      (a[n] = (d))
455    
456    #define GET(a,n) \
457      (a[n])
458    
459    #define MAX_PATTERN_SIZE (1 << 16)
460    
461    #elif LINK_SIZE == 3 || LINK_SIZE == 4
462    
463    #undef LINK_SIZE
464    #define LINK_SIZE 2
465    
466    #define PUT(a,n,d)   \
467      (a[n] = (d) >> 16), \
468      (a[(n)+1] = (d) & 65535)
469    
470    #define GET(a,n) \
471      (((a)[n] << 16) | (a)[(n)+1])
472    
473    /* Keep it positive */
474    #define MAX_PATTERN_SIZE (1 << 30)
475    
476  #else  #else
477  #error LINK_SIZE must be either 2, 3, or 4  #error LINK_SIZE must be either 2, 3, or 4
478  #endif  #endif
479    
480    #else
481    #error Unsupported compiling mode
482    #endif /* COMPILE_PCRE16 */
483    
484    #endif /* COMPILE_PCRE8 */
485    
486  /* Convenience macro defined in terms of the others */  /* Convenience macro defined in terms of the others */
487    
# Line 396  is automated on Unix systems via the "co Line 492  is automated on Unix systems via the "co
492  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
493  capturing parenthesis numbers in back references. */  capturing parenthesis numbers in back references. */
494    
495    #ifdef COMPILE_PCRE8
496    
497    #define IMM2_SIZE 2
498    
499  #define PUT2(a,n,d)   \  #define PUT2(a,n,d)   \
500    a[n] = (d) >> 8; \    a[n] = (d) >> 8; \
501    a[(n)+1] = (d) & 255    a[(n)+1] = (d) & 255
# Line 403  capturing parenthesis numbers in back re Line 503  capturing parenthesis numbers in back re
503  #define GET2(a,n) \  #define GET2(a,n) \
504    (((a)[n] << 8) | (a)[(n)+1])    (((a)[n] << 8) | (a)[(n)+1])
505    
506  #define PUT2INC(a,n,d)  PUT2(a,n,d), a += 2  #else /* COMPILE_PCRE8 */
507    
508    #ifdef COMPILE_PCRE16
509    
510  /* When UTF-8 encoding is being used, a character is no longer just a single  #define IMM2_SIZE 1
 byte. The macros for character handling generate simple sequences when used in  
 byte-mode, and more complicated ones for UTF-8 characters. GETCHARLENTEST is  
 not used when UTF-8 is not supported, so it is not defined, and BACKCHAR should  
 never be called in byte mode. To make sure they can never even appear when  
 UTF-8 support is omitted, we don't even define them. */  
511    
512  #ifndef SUPPORT_UTF8  #define PUT2(a,n,d)   \
513       a[n] = d
514    
515    #define GET2(a,n) \
516       a[n]
517    
518    #else
519    #error Unsupported compiling mode
520    #endif /* COMPILE_PCRE16 */
521    
522    #endif /* COMPILE_PCRE8 */
523    
524    #define PUT2INC(a,n,d)  PUT2(a,n,d), a += IMM2_SIZE
525    
526    /* The maximum length of a MARK name is currently one data unit; it may be
527    changed in future to be a fixed number of bytes or to depend on LINK_SIZE. */
528    
529    #define MAX_MARK ((1 << (sizeof(pcre_uchar)*8)) - 1)
530    
531    /* When UTF encoding is being used, a character is no longer just a single
532    character. The macros for character handling generate simple sequences when
533    used in character-mode, and more complicated ones for UTF characters.
534    GETCHARLENTEST and other macros are not used when UTF is not supported,
535    so they are not defined. To make sure they can never even appear when
536    UTF support is omitted, we don't even define them. */
537    
538    #ifndef SUPPORT_UTF
539    
540    /* #define MAX_VALUE_FOR_SINGLE_CHAR */
541    /* #define HAS_EXTRALEN(c) */
542    /* #define GET_EXTRALEN(c) */
543    /* #define NOT_FIRSTCHAR(c) */
544  #define GETCHAR(c, eptr) c = *eptr;  #define GETCHAR(c, eptr) c = *eptr;
545  #define GETCHARTEST(c, eptr) c = *eptr;  #define GETCHARTEST(c, eptr) c = *eptr;
546  #define GETCHARINC(c, eptr) c = *eptr++;  #define GETCHARINC(c, eptr) c = *eptr++;
# Line 421  UTF-8 support is omitted, we don't even Line 548  UTF-8 support is omitted, we don't even
548  #define GETCHARLEN(c, eptr, len) c = *eptr;  #define GETCHARLEN(c, eptr, len) c = *eptr;
549  /* #define GETCHARLENTEST(c, eptr, len) */  /* #define GETCHARLENTEST(c, eptr, len) */
550  /* #define BACKCHAR(eptr) */  /* #define BACKCHAR(eptr) */
551    /* #define FORWARDCHAR(eptr) */
552    /* #define ACROSSCHAR(condition, eptr, action) */
553    
554    #else   /* SUPPORT_UTF */
555    
556  #else   /* SUPPORT_UTF8 */  #ifdef COMPILE_PCRE8
557    
558  /* These macros were originally written in the form of loops that used data  /* These macros were originally written in the form of loops that used data
559  from the tables whose names start with _pcre_utf8_table. They were rewritten by  from the tables whose names start with PRIV(utf8_table). They were rewritten by
560  a user so as not to use loops, because in some environments this gives a  a user so as not to use loops, because in some environments this gives a
561  significant performance advantage, and it seems never to do any harm. */  significant performance advantage, and it seems never to do any harm. */
562    
563    /* Tells the biggest code point which can be encoded as a single character. */
564    
565    #define MAX_VALUE_FOR_SINGLE_CHAR 127
566    
567    /* Tests whether the code point needs extra characters to decode. */
568    
569    #define HAS_EXTRALEN(c) ((c) >= 0xc0)
570    
571    /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.
572    Otherwise it has an undefined behaviour. */
573    
574    #define GET_EXTRALEN(c) (PRIV(utf8_table4)[(c) & 0x3f])
575    
576    /* Returns TRUE, if the given character is not the first character
577    of a UTF sequence. */
578    
579    #define NOT_FIRSTCHAR(c) (((c) & 0xc0) == 0x80)
580    
581  /* 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
582  advancing the pointer. */  advancing the pointer. */
583    
# Line 463  pointer. */ Line 612  pointer. */
612    
613  #define GETCHARTEST(c, eptr) \  #define GETCHARTEST(c, eptr) \
614    c = *eptr; \    c = *eptr; \
615    if (utf8 && c >= 0xc0) GETUTF8(c, eptr);    if (utf && c >= 0xc0) GETUTF8(c, eptr);
616    
617  /* 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
618  the pointer. */  the pointer. */
# Line 511  This is called when we don't know if we Line 660  This is called when we don't know if we
660    
661  #define GETCHARINCTEST(c, eptr) \  #define GETCHARINCTEST(c, eptr) \
662    c = *eptr++; \    c = *eptr++; \
663    if (utf8 && c >= 0xc0) GETUTF8INC(c, eptr);    if (utf && c >= 0xc0) GETUTF8INC(c, eptr);
664    
665  /* 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
666  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 712  do not know if we are in UTF-8 mode. */
712    
713  #define GETCHARLENTEST(c, eptr, len) \  #define GETCHARLENTEST(c, eptr, len) \
714    c = *eptr; \    c = *eptr; \
715    if (utf8 && c >= 0xc0) GETUTF8LEN(c, eptr, len);    if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len);
716    
717  /* 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
718  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 720  because almost all calls are already wit
720    
721  #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--  #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--
722    
723  #endif  /* SUPPORT_UTF8 */  /* Same as above, just in the other direction. */
724    #define FORWARDCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr++
725    
726    /* Same as above, but it allows a fully customizable form. */
727    #define ACROSSCHAR(condition, eptr, action) \
728      while((condition) && ((eptr) & 0xc0) == 0x80) action
729    
730    #else /* COMPILE_PCRE8 */
731    
732    #ifdef COMPILE_PCRE16
733    
734    /* Tells the biggest code point which can be encoded as a single character. */
735    
736    #define MAX_VALUE_FOR_SINGLE_CHAR 65535
737    
738    /* Tests whether the code point needs extra characters to decode. */
739    
740    #define HAS_EXTRALEN(c) (((c) & 0xfc00) == 0xd800)
741    
742    /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.
743    Otherwise it has an undefined behaviour. */
744    
745    #define GET_EXTRALEN(c) 1
746    
747    /* Returns TRUE, if the given character is not the first character
748    of a UTF sequence. */
749    
750    #define NOT_FIRSTCHAR(c) (((c) & 0xfc00) == 0xdc00)
751    
752    /* Base macro to pick up the low surrogate of a UTF-16 character, not
753    advancing the pointer. */
754    
755    #define GETUTF16(c, eptr) \
756       { c = (((c & 0x3ff) << 10) | (eptr[1] & 0x3ff)) + 0x10000; }
757    
758    /* Get the next UTF-16 character, not advancing the pointer. This is called when
759    we know we are in UTF-16 mode. */
760    
761    #define GETCHAR(c, eptr) \
762      c = *eptr; \
763      if ((c & 0xfc00) == 0xd800) GETUTF16(c, eptr);
764    
765    /* Get the next UTF-16 character, testing for UTF-16 mode, and not advancing the
766    pointer. */
767    
768    #define GETCHARTEST(c, eptr) \
769      c = *eptr; \
770      if (utf && (c & 0xfc00) == 0xd800) GETUTF16(c, eptr);
771    
772    /* Base macro to pick up the low surrogate of a UTF-16 character, advancing
773    the pointer. */
774    
775    #define GETUTF16INC(c, eptr) \
776       { c = (((c & 0x3ff) << 10) | (*eptr++ & 0x3ff)) + 0x10000; }
777    
778    /* Get the next UTF-16 character, advancing the pointer. This is called when we
779    know we are in UTF-16 mode. */
780    
781    #define GETCHARINC(c, eptr) \
782      c = *eptr++; \
783      if ((c & 0xfc00) == 0xd800) GETUTF16INC(c, eptr);
784    
785    /* Get the next character, testing for UTF-16 mode, and advancing the pointer.
786    This is called when we don't know if we are in UTF-16 mode. */
787    
788    #define GETCHARINCTEST(c, eptr) \
789      c = *eptr++; \
790      if (utf && (c & 0xfc00) == 0xd800) GETUTF16INC(c, eptr);
791    
792    /* Base macro to pick up the low surrogate of a UTF-16 character, not
793    advancing the pointer, incrementing the length. */
794    
795    #define GETUTF16LEN(c, eptr, len) \
796       { c = (((c & 0x3ff) << 10) | (eptr[1] & 0x3ff)) + 0x10000; len++; }
797    
798    /* Get the next UTF-16 character, not advancing the pointer, incrementing
799    length if there is a low surrogate. This is called when we know we are in
800    UTF-16 mode. */
801    
802    #define GETCHARLEN(c, eptr, len) \
803      c = *eptr; \
804      if ((c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);
805    
806    /* Get the next UTF-816character, testing for UTF-16 mode, not advancing the
807    pointer, incrementing length if there is a low surrogate. This is called when
808    we do not know if we are in UTF-16 mode. */
809    
810    #define GETCHARLENTEST(c, eptr, len) \
811      c = *eptr; \
812      if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);
813    
814    /* If the pointer is not at the start of a character, move it back until
815    it is. This is called only in UTF-16 mode - we don't put a test within the
816    macro because almost all calls are already within a block of UTF-16 only
817    code. */
818    
819    #define BACKCHAR(eptr) if ((*eptr & 0xfc00) == 0xdc00) eptr--
820    
821    /* Same as above, just in the other direction. */
822    #define FORWARDCHAR(eptr) if ((*eptr & 0xfc00) == 0xdc00) eptr++
823    
824    /* Same as above, but it allows a fully customizable form. */
825    #define ACROSSCHAR(condition, eptr, action) \
826      if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action
827    
828    #endif
829    
830    #endif /* COMPILE_PCRE8 */
831    
832    #endif  /* SUPPORT_UTF */
833    
834    
835  /* In case there is no definition of offsetof() provided - though any proper  /* In case there is no definition of offsetof() provided - though any proper
# Line 582  Standard C system should have one. */ Line 840  Standard C system should have one. */
840  #endif  #endif
841    
842    
 /* These are the public options that can change during matching. */  
   
 #define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)  
   
843  /* Private flags containing information about the compiled regex. They used to  /* Private flags containing information about the compiled regex. They used to
844  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
845  are in a 16-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as  are in a 16-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as
846  the restrictions on partial matching have been lifted. It remains for backwards  the restrictions on partial matching have been lifted. It remains for backwards
847  compatibility. */  compatibility. */
848    
849  #define PCRE_NOPARTIAL     0x0001  /* can't use partial with this regex */  #ifdef COMPILE_PCRE8
850  #define PCRE_FIRSTSET      0x0002  /* first_byte is set */  #define PCRE_MODE          0x0001  /* compiled in 8 bit mode */
851  #define PCRE_REQCHSET      0x0004  /* req_byte is set */  #endif
852  #define PCRE_STARTLINE     0x0008  /* start after \n for multiline */  #ifdef COMPILE_PCRE16
853  #define PCRE_JCHANGED      0x0010  /* j option used in regex */  #define PCRE_MODE          0x0002  /* compiled in 16 bit mode */
854  #define PCRE_HASCRORLF     0x0020  /* explicit \r or \n in pattern */  #endif
855    #define PCRE_FIRSTSET      0x0010  /* first_char is set */
856    #define PCRE_FCH_CASELESS  0x0020  /* caseless first char */
857    #define PCRE_REQCHSET      0x0040  /* req_byte is set */
858    #define PCRE_RCH_CASELESS  0x0080  /* caseless requested char */
859    #define PCRE_STARTLINE     0x0100  /* start after \n for multiline */
860    #define PCRE_NOPARTIAL     0x0200  /* can't use partial with this regex */
861    #define PCRE_JCHANGED      0x0400  /* j option used in regex */
862    #define PCRE_HASCRORLF     0x0800  /* explicit \r or \n in pattern */
863    #define PCRE_HASTHEN       0x1000  /* pattern contains (*THEN) */
864    
865  /* Options for the "extra" block produced by pcre_study(). */  /* Flags for the "extra" block produced by pcre_study(). */
866    
867  #define PCRE_STUDY_MAPPED   0x01     /* a map of starting chars exists */  #define PCRE_STUDY_MAPPED  0x0001  /* a map of starting chars exists */
868  #define PCRE_STUDY_MINLEN   0x02     /* a minimum length field exists */  #define PCRE_STUDY_MINLEN  0x0002  /* a minimum length field exists */
869    
870  /* Masks for identifying the public options that are permitted at compile  /* Masks for identifying the public options that are permitted at compile
871  time, run time, or study time, respectively. */  time, run time, or study time, respectively. */
# Line 628  time, run time, or study time, respectiv Line 891  time, run time, or study time, respectiv
891     PCRE_DFA_RESTART|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \     PCRE_DFA_RESTART|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \
892     PCRE_NO_START_OPTIMIZE)     PCRE_NO_START_OPTIMIZE)
893    
894  #define PUBLIC_STUDY_OPTIONS 0   /* None defined */  #define PUBLIC_STUDY_OPTIONS \
895       (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \
896        PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE|PCRE_STUDY_EXTRA_NEEDED)
897    
898  /* 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. */  
899    
900  #define MAGIC_NUMBER  0x50435245UL   /* 'PCRE' */  #define MAGIC_NUMBER  0x50435245UL   /* 'PCRE' */
901    
902    /* This variable is used to detect a loaded regular expression
903    in different endianness. */
904    
905    #define REVERSED_MAGIC_NUMBER  0x45524350UL   /* 'ERCP' */
906    
907  /* Negative values for the firstchar and reqchar variables */  /* Negative values for the firstchar and reqchar variables */
908    
909  #define REQ_UNSET (-2)  #define REQ_UNSET (-2)
# Line 645  req_byte match. */ Line 914  req_byte match. */
914    
915  #define REQ_BYTE_MAX 1000  #define REQ_BYTE_MAX 1000
916    
 /* 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 */  
   
917  /* Miscellaneous definitions. The #ifndef is to pacify compiler warnings in  /* Miscellaneous definitions. The #ifndef is to pacify compiler warnings in
918  environments where these macros are defined elsewhere. Unfortunately, there  environments where these macros are defined elsewhere. Unfortunately, there
919  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 942  for) in a minority area (EBCDIC platform
942  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
943  macros to give the functions distinct names. */  macros to give the functions distinct names. */
944    
945  #ifndef SUPPORT_UTF8  #ifndef SUPPORT_UTF
946    
947  /* UTF-8 support is not enabled; use the platform-dependent character literals  /* UTF-8 support is not enabled; use the platform-dependent character literals
948  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
949    mode. Newline characters are problematic in EBCDIC. Though it has CR and LF
950    characters, a common practice has been to use its NL (0x15) character as the
951    line terminator in C-like processing environments. However, sometimes the LF
952    (0x25) character is used instead, according to this Unicode document:
953    
954    http://unicode.org/standard/reports/tr13/tr13-5.html
955    
956    PCRE defaults EBCDIC NL to 0x15, but has a build-time option to select 0x25
957    instead. Whichever is *not* chosen is defined as NEL.
958    
959    In both ASCII and EBCDIC environments, CHAR_NL and CHAR_LF are synonyms for the
960    same code point. */
961    
962    #ifdef EBCDIC
963    
964    #ifndef EBCDIC_NL25
965    #define CHAR_NL                     '\x15'
966    #define CHAR_NEL                    '\x25'
967    #define STR_NL                      "\x15"
968    #define STR_NEL                     "\x25"
969    #else
970    #define CHAR_NL                     '\x25'
971    #define CHAR_NEL                    '\x15'
972    #define STR_NL                      "\x25"
973    #define STR_NEL                     "\x15"
974    #endif
975    
976    #define CHAR_LF                     CHAR_NL
977    #define STR_LF                      STR_NL
978    
979    #define CHAR_ESC                    '\047'
980    #define CHAR_DEL                    '\007'
981    #define STR_ESC                     "\047"
982    #define STR_DEL                     "\007"
983    
984    #else  /* Not EBCDIC */
985    
986    /* In ASCII/Unicode, linefeed is '\n' and we equate this to NL for
987    compatibility. NEL is the Unicode newline character; make sure it is
988    a positive value. */
989    
990    #define CHAR_LF                     '\n'
991    #define CHAR_NL                     CHAR_LF
992    #define CHAR_NEL                    ((unsigned char)'\x85')
993    #define CHAR_ESC                    '\033'
994    #define CHAR_DEL                    '\177'
995    
996    #define STR_LF                      "\n"
997    #define STR_NL                      STR_LF
998    #define STR_NEL                     "\x85"
999    #define STR_ESC                     "\033"
1000    #define STR_DEL                     "\177"
1001    
1002    #endif  /* EBCDIC */
1003    
1004    /* The remaining definitions work in both environments. */
1005    
1006  #define CHAR_HT                     '\t'  #define CHAR_HT                     '\t'
1007  #define CHAR_VT                     '\v'  #define CHAR_VT                     '\v'
1008  #define CHAR_FF                     '\f'  #define CHAR_FF                     '\f'
1009  #define CHAR_CR                     '\r'  #define CHAR_CR                     '\r'
 #define CHAR_NL                     '\n'  
1010  #define CHAR_BS                     '\b'  #define CHAR_BS                     '\b'
1011  #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  
1012    
1013  #define CHAR_SPACE                  ' '  #define CHAR_SPACE                  ' '
1014  #define CHAR_EXCLAMATION_MARK       '!'  #define CHAR_EXCLAMATION_MARK       '!'
# Line 799  so that PCRE works on both ASCII and EBC Line 1110  so that PCRE works on both ASCII and EBC
1110  #define STR_VT                      "\v"  #define STR_VT                      "\v"
1111  #define STR_FF                      "\f"  #define STR_FF                      "\f"
1112  #define STR_CR                      "\r"  #define STR_CR                      "\r"
 #define STR_NL                      "\n"  
1113  #define STR_BS                      "\b"  #define STR_BS                      "\b"
1114  #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  
1115    
1116  #define STR_SPACE                   " "  #define STR_SPACE                   " "
1117  #define STR_EXCLAMATION_MARK        "!"  #define STR_EXCLAMATION_MARK        "!"
# Line 939  so that PCRE works on both ASCII and EBC Line 1242  so that PCRE works on both ASCII and EBC
1242  #define STRING_ANYCRLF_RIGHTPAR        "ANYCRLF)"  #define STRING_ANYCRLF_RIGHTPAR        "ANYCRLF)"
1243  #define STRING_BSR_ANYCRLF_RIGHTPAR    "BSR_ANYCRLF)"  #define STRING_BSR_ANYCRLF_RIGHTPAR    "BSR_ANYCRLF)"
1244  #define STRING_BSR_UNICODE_RIGHTPAR    "BSR_UNICODE)"  #define STRING_BSR_UNICODE_RIGHTPAR    "BSR_UNICODE)"
1245  #define STRING_UTF8_RIGHTPAR           "UTF8)"  #ifdef COMPILE_PCRE8
1246    #define STRING_UTF_RIGHTPAR            "UTF8)"
1247    #endif
1248    #ifdef COMPILE_PCRE16
1249    #define STRING_UTF_RIGHTPAR            "UTF16)"
1250    #endif
1251  #define STRING_UCP_RIGHTPAR            "UCP)"  #define STRING_UCP_RIGHTPAR            "UCP)"
1252  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"  #define STRING_NO_START_OPT_RIGHTPAR   "NO_START_OPT)"
1253    
1254  #else  /* SUPPORT_UTF8 */  #else  /* SUPPORT_UTF */
1255    
1256  /* 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
1257  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 1261  only. */
1261  #define CHAR_VT                     '\013'  #define CHAR_VT                     '\013'
1262  #define CHAR_FF                     '\014'  #define CHAR_FF                     '\014'
1263  #define CHAR_CR                     '\015'  #define CHAR_CR                     '\015'
1264  #define CHAR_NL                     '\012'  #define CHAR_LF                     '\012'
1265    #define CHAR_NL                     CHAR_LF
1266    #define CHAR_NEL                    ((unsigned char)'\x85')
1267  #define CHAR_BS                     '\010'  #define CHAR_BS                     '\010'
1268  #define CHAR_BEL                    '\007'  #define CHAR_BEL                    '\007'
1269  #define CHAR_ESC                    '\033'  #define CHAR_ESC                    '\033'
# Line 1194  only. */ Line 1504  only. */
1504  #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
1505  #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
1506  #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
1507  #define STRING_UTF8_RIGHTPAR           STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS  #ifdef COMPILE_PCRE8
1508    #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS
1509    #endif
1510    #ifdef COMPILE_PCRE16
1511    #define STRING_UTF_RIGHTPAR            STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS
1512    #endif
1513  #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
1514  #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
1515    
1516  #endif  /* SUPPORT_UTF8 */  #endif  /* SUPPORT_UTF */
1517    
1518  /* Escape items that are just an encoding of a particular data value. */  /* Escape items that are just an encoding of a particular data value. */
1519    
# Line 1211  only. */ Line 1526  only. */
1526  #endif  #endif
1527    
1528  #ifndef ESC_n  #ifndef ESC_n
1529  #define ESC_n CHAR_NL  #define ESC_n CHAR_LF
1530  #endif  #endif
1531    
1532  #ifndef ESC_r  #ifndef ESC_r
# Line 1238  only. */ Line 1553  only. */
1553  #define PT_WORD       8    /* Word - L plus N plus underscore */  #define PT_WORD       8    /* Word - L plus N plus underscore */
1554    
1555  /* 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
1556  contain UTF-8 characters with values greater than 255. */  contain characters with values greater than 255. */
1557    
1558  #define XCL_NOT    0x01    /* Flag: this is a negative class */  #define XCL_NOT    0x01    /* Flag: this is a negative class */
1559  #define XCL_MAP    0x02    /* Flag: a 32-byte map is present */  #define XCL_MAP    0x02    /* Flag: a 32-byte map is present */
# Line 1254  value such as \n. They must have non-zer Line 1569  value such as \n. They must have non-zer
1569  their negation. Also, they must appear in the same order as in the opcode  their negation. Also, they must appear in the same order as in the opcode
1570  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
1571  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
1572  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
1573  like \N.  non-DOTALL mode, "." behaves like \N.
1574    
1575  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.
1576  when PCRE_UCP is set, when replacement of \d etc by \p sequences is required.  when PCRE_UCP is set, when replacement of \d etc by \p sequences is required.
# Line 1325  enum { Line 1640  enum {
1640    
1641    /* The following sets of 13 opcodes must always be kept in step because    /* The following sets of 13 opcodes must always be kept in step because
1642    the offset from the first one is used to generate the others. */    the offset from the first one is used to generate the others. */
1643    
1644    /**** Single characters, caseful, must precede the caseless ones ****/    /**** Single characters, caseful, must precede the caseless ones ****/
1645    
1646    OP_STAR,           /* 33 The maximizing and minimizing versions of */    OP_STAR,           /* 33 The maximizing and minimizing versions of */
# Line 1345  enum { Line 1660  enum {
1660    OP_POSUPTO,        /* 45 Possessified upto, caseful */    OP_POSUPTO,        /* 45 Possessified upto, caseful */
1661    
1662    /**** Single characters, caseless, must follow the caseful ones */    /**** Single characters, caseless, must follow the caseful ones */
1663    
1664    OP_STARI,          /* 46 */    OP_STARI,          /* 46 */
1665    OP_MINSTARI,       /* 47 */    OP_MINSTARI,       /* 47 */
1666    OP_PLUSI,          /* 48 */    OP_PLUSI,          /* 48 */
# Line 1382  enum { Line 1697  enum {
1697    OP_NOTPOSUPTO,     /* 71 */    OP_NOTPOSUPTO,     /* 71 */
1698    
1699    /**** Negated single character, caseless; must follow the caseful ones ****/    /**** Negated single character, caseless; must follow the caseful ones ****/
1700    
1701    OP_NOTSTARI,       /* 72 */    OP_NOTSTARI,       /* 72 */
1702    OP_NOTMINSTARI,    /* 73 */    OP_NOTMINSTARI,    /* 73 */
1703    OP_NOTPLUSI,       /* 74 */    OP_NOTPLUSI,       /* 74 */
# Line 1400  enum { Line 1715  enum {
1715    OP_NOTPOSUPTOI,    /* 84 */    OP_NOTPOSUPTOI,    /* 84 */
1716    
1717    /**** Character types ****/    /**** Character types ****/
1718    
1719    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */    OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */
1720    OP_TYPEMINSTAR,    /* 86 these six opcodes must come in pairs, with */    OP_TYPEMINSTAR,    /* 86 these six opcodes must come in pairs, with */
1721    OP_TYPEPLUS,       /* 87 the minimizing one second. These codes must */    OP_TYPEPLUS,       /* 87 the minimizing one second. These codes must */
# Line 1418  enum { Line 1733  enum {
1733    OP_TYPEPOSUPTO,    /* 97 */    OP_TYPEPOSUPTO,    /* 97 */
1734    
1735    /* These are used for character classes and back references; only the    /* These are used for character classes and back references; only the
1736    first six are the same as the sets above. */    first six are the same as the sets above. */
1737    
1738    OP_CRSTAR,         /* 98 The maximizing and minimizing versions of */    OP_CRSTAR,         /* 98 The maximizing and minimizing versions of */
1739    OP_CRMINSTAR,      /* 99 all these opcodes must come in pairs, with */    OP_CRMINSTAR,      /* 99 all these opcodes must come in pairs, with */
# Line 1435  enum { Line 1750  enum {
1750    OP_CLASS,          /* 106 Match a character class, chars < 256 only */    OP_CLASS,          /* 106 Match a character class, chars < 256 only */
1751    OP_NCLASS,         /* 107 Same, but the bitmap was created from a negative    OP_NCLASS,         /* 107 Same, but the bitmap was created from a negative
1752                                class - the difference is relevant only when a                                class - the difference is relevant only when a
1753                                UTF-8 character > 255 is encountered. */                                character > 255 is encountered. */
1754    OP_XCLASS,         /* 108 Extended class for handling UTF-8 chars within the    OP_XCLASS,         /* 108 Extended class for handling > 255 chars within the
1755                                class. This does both positive and negative. */                                class. This does both positive and negative. */
1756    OP_REF,            /* 109 Match a back reference, casefully */    OP_REF,            /* 109 Match a back reference, casefully */
1757    OP_REFI,           /* 110 Match a back reference, caselessly */    OP_REFI,           /* 110 Match a back reference, caselessly */
# Line 1447  enum { Line 1762  enum {
1762    OP_KET,            /* 114 End of group that doesn't have an unbounded repeat */    OP_KET,            /* 114 End of group that doesn't have an unbounded repeat */
1763    OP_KETRMAX,        /* 115 These two must remain together and in this */    OP_KETRMAX,        /* 115 These two must remain together and in this */
1764    OP_KETRMIN,        /* 116 order. They are for groups the repeat for ever. */    OP_KETRMIN,        /* 116 order. They are for groups the repeat for ever. */
1765      OP_KETRPOS,        /* 117 Possessive unlimited repeat. */
1766    
1767    /* The assertions must come before BRA, CBRA, ONCE, and COND.*/    /* The assertions must come before BRA, CBRA, ONCE, and COND, and the four
1768      asserts must remain in order. */
1769    
1770    OP_ASSERT,         /* 117 Positive lookahead */    OP_REVERSE,        /* 118 Move pointer back - used in lookbehind assertions */
1771    OP_ASSERT_NOT,     /* 118 Negative lookahead */    OP_ASSERT,         /* 119 Positive lookahead */
1772    OP_ASSERTBACK,     /* 119 Positive lookbehind */    OP_ASSERT_NOT,     /* 120 Negative lookahead */
1773    OP_ASSERTBACK_NOT, /* 120 Negative lookbehind */    OP_ASSERTBACK,     /* 121 Positive lookbehind */
1774    OP_REVERSE,        /* 121 Move pointer back - used in lookbehind assertions */    OP_ASSERTBACK_NOT, /* 122 Negative lookbehind */
1775    
1776    /* ONCE, BRA, CBRA, and COND must come after the assertions, with ONCE first,    /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately
1777    as there's a test for >= ONCE for a subpattern that isn't an assertion. */    after the assertions, with ONCE first, as there's a test for >= ONCE for a
1778      subpattern that isn't an assertion. The POS versions must immediately follow
1779    OP_ONCE,           /* 122 Atomic group */    the non-POS versions in each case. */
1780    OP_BRA,            /* 123 Start of non-capturing bracket */  
1781    OP_CBRA,           /* 124 Start of capturing bracket */    OP_ONCE,           /* 123 Atomic group, contains captures */
1782    OP_COND,           /* 125 Conditional group */    OP_ONCE_NC,        /* 124 Atomic group containing no captures */
1783      OP_BRA,            /* 125 Start of non-capturing bracket */
1784      OP_BRAPOS,         /* 126 Ditto, with unlimited, possessive repeat */
1785      OP_CBRA,           /* 127 Start of capturing bracket */
1786      OP_CBRAPOS,        /* 128 Ditto, with unlimited, possessive repeat */
1787      OP_COND,           /* 129 Conditional group */
1788    
1789    /* 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
1790    check for >= SBRA to distinguish the two sets. */    check for >= SBRA to distinguish the two sets. */
1791    
1792    OP_SBRA,           /* 126 Start of non-capturing bracket, check empty  */    OP_SBRA,           /* 130 Start of non-capturing bracket, check empty  */
1793    OP_SCBRA,          /* 127 Start of capturing bracket, check empty */    OP_SBRAPOS,        /* 131 Ditto, with unlimited, possessive repeat */
1794    OP_SCOND,          /* 128 Conditional group, check empty */    OP_SCBRA,          /* 132 Start of capturing bracket, check empty */
1795      OP_SCBRAPOS,       /* 133 Ditto, with unlimited, possessive repeat */
1796      OP_SCOND,          /* 134 Conditional group, check empty */
1797    
1798    /* The next two pairs must (respectively) be kept together. */    /* The next two pairs must (respectively) be kept together. */
1799    
1800    OP_CREF,           /* 129 Used to hold a capture number as condition */    OP_CREF,           /* 135 Used to hold a capture number as condition */
1801    OP_NCREF,          /* 130 Same, but generated by a name reference*/    OP_NCREF,          /* 136 Same, but generated by a name reference*/
1802    OP_RREF,           /* 131 Used to hold a recursion number as condition */    OP_RREF,           /* 137 Used to hold a recursion number as condition */
1803    OP_NRREF,          /* 132 Same, but generated by a name reference*/    OP_NRREF,          /* 138 Same, but generated by a name reference*/
1804    OP_DEF,            /* 133 The DEFINE condition */    OP_DEF,            /* 139 The DEFINE condition */
1805    
1806    OP_BRAZERO,        /* 134 These two must remain together and in this */    OP_BRAZERO,        /* 140 These two must remain together and in this */
1807    OP_BRAMINZERO,     /* 135 order. */    OP_BRAMINZERO,     /* 141 order. */
1808      OP_BRAPOSZERO,     /* 142 */
1809    
1810    /* These are backtracking control verbs */    /* These are backtracking control verbs */
1811    
1812    OP_MARK,           /* 136 always has an argument */    OP_MARK,           /* 143 always has an argument */
1813    OP_PRUNE,          /* 137 */    OP_PRUNE,          /* 144 */
1814    OP_PRUNE_ARG,      /* 138 same, but with argument */    OP_PRUNE_ARG,      /* 145 same, but with argument */
1815    OP_SKIP,           /* 139 */    OP_SKIP,           /* 146 */
1816    OP_SKIP_ARG,       /* 140 same, but with argument */    OP_SKIP_ARG,       /* 147 same, but with argument */
1817    OP_THEN,           /* 141 */    OP_THEN,           /* 148 */
1818    OP_THEN_ARG,       /* 142 same, but with argument */    OP_THEN_ARG,       /* 149 same, but with argument */
1819    OP_COMMIT,         /* 143 */    OP_COMMIT,         /* 150 */
1820    
1821    /* These are forced failure and success verbs */    /* These are forced failure and success verbs */
1822    
1823    OP_FAIL,           /* 144 */    OP_FAIL,           /* 151 */
1824    OP_ACCEPT,         /* 145 */    OP_ACCEPT,         /* 152 */
1825    OP_CLOSE,          /* 146 Used before OP_ACCEPT to close open captures */    OP_ASSERT_ACCEPT,  /* 153 Used inside assertions */
1826      OP_CLOSE,          /* 154 Used before OP_ACCEPT to close open captures */
1827    
1828    /* This is used to skip a subpattern with a {0} quantifier */    /* This is used to skip a subpattern with a {0} quantifier */
1829    
1830    OP_SKIPZERO,       /* 147 */    OP_SKIPZERO,       /* 155 */
1831    
1832    /* 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
1833    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 1517  called "coptable" and "poptable" in pcre Line 1843  called "coptable" and "poptable" in pcre
1843    
1844  /* 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
1845  for debugging, and some of them are only partial names. The macro is referenced  for debugging, and some of them are only partial names. The macro is referenced
1846  only in pcre_printint.c, which fills out the full names in many cases (and in  only in pcre_printint.c, which fills out the full names in many cases (and in
1847  some cases doesn't actually use these names at all). */  some cases doesn't actually use these names at all). */
1848    
1849  #define OP_NAME_LIST \  #define OP_NAME_LIST \
# Line 1543  some cases doesn't actually use these na Line 1869  some cases doesn't actually use these na
1869    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \
1870    "class", "nclass", "xclass", "Ref", "Refi",                     \    "class", "nclass", "xclass", "Ref", "Refi",                     \
1871    "Recurse", "Callout",                                           \    "Recurse", "Callout",                                           \
1872    "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not",     \    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \
1873    "AssertB", "AssertB not", "Reverse",                            \    "Reverse", "Assert", "Assert not", "AssertB", "AssertB not",    \
1874    "Once", "Bra", "CBra", "Cond", "SBra", "SCBra", "SCond",        \    "Once", "Once_NC",                                              \
1875      "Bra", "BraPos", "CBra", "CBraPos",                             \
1876      "Cond",                                                         \
1877      "SBra", "SBraPos", "SCBra", "SCBraPos",                         \
1878      "SCond",                                                        \
1879    "Cond ref", "Cond nref", "Cond rec", "Cond nrec", "Cond def",   \    "Cond ref", "Cond nref", "Cond rec", "Cond nrec", "Cond def",   \
1880    "Brazero", "Braminzero",                                        \    "Brazero", "Braminzero", "Braposzero",                          \
1881    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \    "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \
1882    "*THEN", "*THEN", "*COMMIT", "*FAIL", "*ACCEPT",                \    "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \
1883      "*ACCEPT", "*ASSERT_ACCEPT",                                    \
1884    "Close", "Skip zero"    "Close", "Skip zero"
1885    
1886    
# Line 1577  in UTF-8 mode. The code that uses this t Line 1908  in UTF-8 mode. The code that uses this t
1908    2,                             /* noti                                   */ \    2,                             /* noti                                   */ \
1909    /* Positive single-char repeats                             ** These are */ \    /* Positive single-char repeats                             ** These are */ \
1910    2, 2, 2, 2, 2, 2,              /* *, *?, +, +?, ?, ??       ** minima in */ \    2, 2, 2, 2, 2, 2,              /* *, *?, +, +?, ?, ??       ** minima in */ \
1911    4, 4, 4,                       /* upto, minupto, exact      ** mode      */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* upto, minupto             ** mode      */ \
1912    2, 2, 2, 4,                    /* *+, ++, ?+, upto+                      */ \    2+IMM2_SIZE,                   /* exact                                  */ \
1913      2, 2, 2, 2+IMM2_SIZE,          /* *+, ++, ?+, upto+                      */ \
1914    2, 2, 2, 2, 2, 2,              /* *I, *?I, +I, +?I, ?I, ??I ** UTF-8     */ \    2, 2, 2, 2, 2, 2,              /* *I, *?I, +I, +?I, ?I, ??I ** UTF-8     */ \
1915    4, 4, 4,                       /* upto I, minupto I, exact I             */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* upto I, minupto I                      */ \
1916    2, 2, 2, 4,                    /* *+I, ++I, ?+I, upto+I                  */ \    2+IMM2_SIZE,                   /* exact I                                */ \
1917      2, 2, 2, 2+IMM2_SIZE,          /* *+I, ++I, ?+I, upto+I                  */ \
1918    /* Negative single-char repeats - only for chars < 256                   */ \    /* Negative single-char repeats - only for chars < 256                   */ \
1919    2, 2, 2, 2, 2, 2,              /* NOT *, *?, +, +?, ?, ??                */ \    2, 2, 2, 2, 2, 2,              /* NOT *, *?, +, +?, ?, ??                */ \
1920    4, 4, 4,                       /* NOT upto, minupto, exact               */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* NOT upto, minupto                      */ \
1921    2, 2, 2, 4,                    /* Possessive NOT *, +, ?, upto           */ \    2+IMM2_SIZE,                   /* NOT exact                              */ \
1922      2, 2, 2, 2+IMM2_SIZE,          /* Possessive NOT *, +, ?, upto           */ \
1923    2, 2, 2, 2, 2, 2,              /* NOT *I, *?I, +I, +?I, ?I, ??I          */ \    2, 2, 2, 2, 2, 2,              /* NOT *I, *?I, +I, +?I, ?I, ??I          */ \
1924    4, 4, 4,                       /* NOT upto I, minupto I, exact I         */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* NOT upto I, minupto I                  */ \
1925    2, 2, 2, 4,                    /* Possessive NOT *I, +I, ?I, upto I      */ \    2+IMM2_SIZE,                   /* NOT exact I                            */ \
1926      2, 2, 2, 2+IMM2_SIZE,          /* Possessive NOT *I, +I, ?I, upto I      */ \
1927    /* Positive type repeats                                                 */ \    /* Positive type repeats                                                 */ \
1928    2, 2, 2, 2, 2, 2,              /* Type *, *?, +, +?, ?, ??               */ \    2, 2, 2, 2, 2, 2,              /* Type *, *?, +, +?, ?, ??               */ \
1929    4, 4, 4,                       /* Type upto, minupto, exact              */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* Type upto, minupto                     */ \
1930    2, 2, 2, 4,                    /* Possessive *+, ++, ?+, upto+           */ \    2+IMM2_SIZE,                   /* Type exact                             */ \
1931      2, 2, 2, 2+IMM2_SIZE,          /* Possessive *+, ++, ?+, upto+           */ \
1932    /* Character class & ref repeats                                         */ \    /* Character class & ref repeats                                         */ \
1933    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \
1934    5, 5,                          /* CRRANGE, CRMINRANGE                    */ \    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \
1935   33,                             /* CLASS                                  */ \    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \
1936   33,                             /* NCLASS                                 */ \    1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \
1937    0,                             /* XCLASS - variable length               */ \    0,                             /* XCLASS - variable length               */ \
1938    3,                             /* REF                                    */ \    1+IMM2_SIZE,                   /* REF                                    */ \
1939    3,                             /* REFI                                   */ \    1+IMM2_SIZE,                   /* REFI                                   */ \
1940    1+LINK_SIZE,                   /* RECURSE                                */ \    1+LINK_SIZE,                   /* RECURSE                                */ \
1941    2+2*LINK_SIZE,                 /* CALLOUT                                */ \    2+2*LINK_SIZE,                 /* CALLOUT                                */ \
1942    1+LINK_SIZE,                   /* Alt                                    */ \    1+LINK_SIZE,                   /* Alt                                    */ \
1943    1+LINK_SIZE,                   /* Ket                                    */ \    1+LINK_SIZE,                   /* Ket                                    */ \
1944    1+LINK_SIZE,                   /* KetRmax                                */ \    1+LINK_SIZE,                   /* KetRmax                                */ \
1945    1+LINK_SIZE,                   /* KetRmin                                */ \    1+LINK_SIZE,                   /* KetRmin                                */ \
1946      1+LINK_SIZE,                   /* KetRpos                                */ \
1947      1+LINK_SIZE,                   /* Reverse                                */ \
1948    1+LINK_SIZE,                   /* Assert                                 */ \    1+LINK_SIZE,                   /* Assert                                 */ \
1949    1+LINK_SIZE,                   /* Assert not                             */ \    1+LINK_SIZE,                   /* Assert not                             */ \
1950    1+LINK_SIZE,                   /* Assert behind                          */ \    1+LINK_SIZE,                   /* Assert behind                          */ \
1951    1+LINK_SIZE,                   /* Assert behind not                      */ \    1+LINK_SIZE,                   /* Assert behind not                      */ \
   1+LINK_SIZE,                   /* Reverse                                */ \  
1952    1+LINK_SIZE,                   /* ONCE                                   */ \    1+LINK_SIZE,                   /* ONCE                                   */ \
1953      1+LINK_SIZE,                   /* ONCE_NC                                */ \
1954    1+LINK_SIZE,                   /* BRA                                    */ \    1+LINK_SIZE,                   /* BRA                                    */ \
1955    3+LINK_SIZE,                   /* CBRA                                   */ \    1+LINK_SIZE,                   /* BRAPOS                                 */ \
1956      1+LINK_SIZE+IMM2_SIZE,         /* CBRA                                   */ \
1957      1+LINK_SIZE+IMM2_SIZE,         /* CBRAPOS                                */ \
1958    1+LINK_SIZE,                   /* COND                                   */ \    1+LINK_SIZE,                   /* COND                                   */ \
1959    1+LINK_SIZE,                   /* SBRA                                   */ \    1+LINK_SIZE,                   /* SBRA                                   */ \
1960    3+LINK_SIZE,                   /* SCBRA                                  */ \    1+LINK_SIZE,                   /* SBRAPOS                                */ \
1961      1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \
1962      1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \
1963    1+LINK_SIZE,                   /* SCOND                                  */ \    1+LINK_SIZE,                   /* SCOND                                  */ \
1964    3, 3,                          /* CREF, NCREF                            */ \    1+IMM2_SIZE, 1+IMM2_SIZE,      /* CREF, NCREF                            */ \
1965    3, 3,                          /* RREF, NRREF                            */ \    1+IMM2_SIZE, 1+IMM2_SIZE,      /* RREF, NRREF                            */ \
1966    1,                             /* DEF                                    */ \    1,                             /* DEF                                    */ \
1967    1, 1,                          /* BRAZERO, BRAMINZERO                    */ \    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \
1968    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \
1969    1, 3,                          /* SKIP, SKIP_ARG                         */ \    1, 3,                          /* SKIP, SKIP_ARG                         */ \
1970    1+LINK_SIZE, 3+LINK_SIZE,      /* THEN, THEN_ARG                         */ \    1, 3,                          /* THEN, THEN_ARG                         */ \
1971    1, 1, 1, 3, 1                  /* COMMIT, FAIL, ACCEPT, CLOSE, SKIPZERO  */ \    1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \
1972      1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */
1973    
1974  /* 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"
1975  condition. */  condition. */
# Line 1643  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4, Line 1986  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,
1986         ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,         ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
1987         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
1988         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
1989         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68,         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,
1990         ERRCOUNT };         ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERRCOUNT };
1991    
1992    /* JIT compiling modes. The function list is indexed by them. */
1993    enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,
1994           JIT_NUMBER_OF_COMPILE_MODES };
1995    
1996  /* 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
1997  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 1663  fields are present. Currently PCRE alway Line 2010  fields are present. Currently PCRE alway
2010  NOTE NOTE NOTE  NOTE NOTE NOTE
2011  */  */
2012    
2013  typedef struct real_pcre {  #ifdef COMPILE_PCRE8
2014    #define REAL_PCRE real_pcre
2015    #else
2016    #define REAL_PCRE real_pcre16
2017    #endif
2018    
2019    typedef struct REAL_PCRE {
2020    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2021    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2022    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2023    pcre_uint16 flags;              /* Private flags */    pcre_uint16 flags;              /* Private flags */
2024    pcre_uint16 dummy1;             /* For future use */    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2025    pcre_uint16 top_bracket;    pcre_uint16 top_bracket;        /* Highest numbered group */
2026    pcre_uint16 top_backref;    pcre_uint16 top_backref;        /* Highest numbered back reference */
2027    pcre_uint16 first_byte;    pcre_uint16 first_char;         /* Starting character */
2028    pcre_uint16 req_byte;    pcre_uint16 req_char;           /* This character must be seen */
2029    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2030    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2031    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2032    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2033      const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2034    const unsigned char *tables;    /* Pointer to tables or NULL for std */    const pcre_uint8 *nullpad;      /* NULL padding */
2035    const unsigned char *nullpad;   /* NULL padding */  } REAL_PCRE;
 } real_pcre;  
2036    
2037  /* 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
2038  remark (see NOTE above) about extending this structure applies. */  remark (see NOTE above) about extending this structure applies. */
# Line 1688  remark (see NOTE above) about extending Line 2040  remark (see NOTE above) about extending
2040  typedef struct pcre_study_data {  typedef struct pcre_study_data {
2041    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2042    pcre_uint32 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2043    uschar start_bits[32];          /* Starting char bits */    pcre_uint8 start_bits[32];      /* Starting char bits */
2044    pcre_uint32 minlength;          /* Minimum subject length */    pcre_uint32 minlength;          /* Minimum subject length */
2045  } pcre_study_data;  } pcre_study_data;
2046    
# Line 1707  typedef struct open_capitem { Line 2059  typedef struct open_capitem {
2059  doing the compiling, so that they are thread-safe. */  doing the compiling, so that they are thread-safe. */
2060    
2061  typedef struct compile_data {  typedef struct compile_data {
2062    const uschar *lcc;            /* Points to lower casing table */    const pcre_uint8 *lcc;            /* Points to lower casing table */
2063    const uschar *fcc;            /* Points to case-flipping table */    const pcre_uint8 *fcc;            /* Points to case-flipping table */
2064    const uschar *cbits;          /* Points to character type table */    const pcre_uint8 *cbits;          /* Points to character type table */
2065    const uschar *ctypes;         /* Points to table of type maps */    const pcre_uint8 *ctypes;         /* Points to table of type maps */
2066    const uschar *start_workspace;/* The start of working space */    const pcre_uchar *start_workspace;/* The start of working space */
2067    const uschar *start_code;     /* The start of the compiled code */    const pcre_uchar *start_code;     /* The start of the compiled code */
2068    const uschar *start_pattern;  /* The start of the pattern */    const pcre_uchar *start_pattern;  /* The start of the pattern */
2069    const uschar *end_pattern;    /* The end of the pattern */    const pcre_uchar *end_pattern;    /* The end of the pattern */
2070    open_capitem *open_caps;      /* Chain of open capture items */    open_capitem *open_caps;          /* Chain of open capture items */
2071    uschar *hwm;                  /* High watermark of workspace */    pcre_uchar *hwm;                  /* High watermark of workspace */
2072    uschar *name_table;           /* The name/number table */    pcre_uchar *name_table;           /* The name/number table */
2073    int  names_found;             /* Number of entries so far */    int  names_found;                 /* Number of entries so far */
2074    int  name_entry_size;         /* Size of each entry */    int  name_entry_size;             /* Size of each entry */
2075    int  bracount;                /* Count of capturing parens as we compile */    int  workspace_size;              /* Size of workspace */
2076    int  final_bracount;          /* Saved value after first pass */    int  bracount;                    /* Count of capturing parens as we compile */
2077    int  top_backref;             /* Maximum back reference */    int  final_bracount;              /* Saved value after first pass */
2078    unsigned int backref_map;     /* Bitmap of low back refs */    int  max_lookbehind;              /* Maximum lookbehind (characters) */
2079    int  external_options;        /* External (initial) options */    int  top_backref;                 /* Maximum back reference */
2080    int  external_flags;          /* External flag bits to be set */    unsigned int backref_map;         /* Bitmap of low back refs */
2081    int  req_varyopt;             /* "After variable item" flag for reqbyte */    int  assert_depth;                /* Depth of nested assertions */
2082    BOOL had_accept;              /* (*ACCEPT) encountered */    int  external_options;            /* External (initial) options */
2083    BOOL check_lookbehind;        /* Lookbehinds need later checking */    int  external_flags;              /* External flag bits to be set */
2084    int  nltype;                  /* Newline type */    int  req_varyopt;                 /* "After variable item" flag for reqbyte */
2085    int  nllen;                   /* Newline string length */    BOOL had_accept;                  /* (*ACCEPT) encountered */
2086    uschar nl[4];                 /* Newline string when fixed length */    BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */
2087      BOOL check_lookbehind;            /* Lookbehinds need later checking */
2088      int  nltype;                      /* Newline type */
2089      int  nllen;                       /* Newline string length */
2090      pcre_uchar nl[4];                 /* Newline string when fixed length */
2091  } compile_data;  } compile_data;
2092    
2093  /* Structure for maintaining a chain of pointers to the currently incomplete  /* Structure for maintaining a chain of pointers to the currently incomplete
2094  branches, for testing for left recursion. */  branches, for testing for left recursion while compiling. */
2095    
2096  typedef struct branch_chain {  typedef struct branch_chain {
2097    struct branch_chain *outer;    struct branch_chain *outer;
2098    uschar *current_branch;    pcre_uchar *current_branch;
2099  } branch_chain;  } branch_chain;
2100    
2101  /* Structure for items in a linked list that represents an explicit recursive  /* Structure for items in a linked list that represents an explicit recursive
2102  call within the pattern. */  call within the pattern; used by pcre_exec(). */
2103    
2104  typedef struct recursion_info {  typedef struct recursion_info {
2105    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
2106    int group_num;                /* Number of group that was called */    int group_num;                  /* Number of group that was called */
2107    const uschar *after_call;     /* "Return value": points after the call in the expr */    int *offset_save;               /* Pointer to start of saved offsets */
2108    int *offset_save;             /* Pointer to start of saved offsets */    int saved_max;                  /* Number of saved offsets */
2109    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 */  
2110  } recursion_info;  } recursion_info;
2111    
2112    /* A similar structure for pcre_dfa_exec(). */
2113    
2114    typedef struct dfa_recursion_info {
2115      struct dfa_recursion_info *prevrec;
2116      int group_num;
2117      PCRE_PUCHAR subject_position;
2118    } dfa_recursion_info;
2119    
2120  /* 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
2121  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
2122  has been matched by a subpattern - to break infinite loops. */  has been matched by a subpattern - to break infinite loops; used by
2123    pcre_exec(). */
2124    
2125  typedef struct eptrblock {  typedef struct eptrblock {
2126    struct eptrblock *epb_prev;    struct eptrblock *epb_prev;
2127    USPTR epb_saved_eptr;    PCRE_PUCHAR epb_saved_eptr;
2128  } eptrblock;  } eptrblock;
2129    
2130    
# Line 1771  typedef struct match_data { Line 2135  typedef struct match_data {
2135    unsigned long int match_call_count;      /* As it says */    unsigned long int match_call_count;      /* As it says */
2136    unsigned long int match_limit;           /* As it says */    unsigned long int match_limit;           /* As it says */
2137    unsigned long int match_limit_recursion; /* As it says */    unsigned long int match_limit_recursion; /* As it says */
2138    int   *offset_vector;         /* Offset vector */    int   *offset_vector;           /* Offset vector */
2139    int    offset_end;            /* One past the end */    int    offset_end;              /* One past the end */
2140    int    offset_max;            /* The maximum usable for return data */    int    offset_max;              /* The maximum usable for return data */
2141    int    nltype;                /* Newline type */    int    nltype;                  /* Newline type */
2142    int    nllen;                 /* Newline string length */    int    nllen;                   /* Newline string length */
2143    int    name_count;            /* Number of names in name table */    int    name_count;              /* Number of names in name table */
2144    int    name_entry_size;       /* Size of entry in names table */    int    name_entry_size;         /* Size of entry in names table */
2145    uschar *name_table;           /* Table of names */    pcre_uchar *name_table;         /* Table of names */
2146    uschar nl[4];                 /* Newline string when fixed */    pcre_uchar nl[4];               /* Newline string when fixed */
2147    const uschar *lcc;            /* Points to lower casing table */    const  pcre_uint8 *lcc;         /* Points to lower casing table */
2148    const uschar *ctypes;         /* Points to table of type maps */    const  pcre_uint8 *fcc;         /* Points to case-flipping table */
2149    BOOL   offset_overflow;       /* Set if too many extractions */    const  pcre_uint8 *ctypes;      /* Points to table of type maps */
2150    BOOL   notbol;                /* NOTBOL flag */    BOOL   offset_overflow;         /* Set if too many extractions */
2151    BOOL   noteol;                /* NOTEOL flag */    BOOL   notbol;                  /* NOTBOL flag */
2152    BOOL   utf8;                  /* UTF8 flag */    BOOL   noteol;                  /* NOTEOL flag */
2153    BOOL   jscript_compat;        /* JAVASCRIPT_COMPAT flag */    BOOL   utf;                     /* UTF-8 / UTF-16 flag */
2154    BOOL   use_ucp;               /* PCRE_UCP flag */    BOOL   jscript_compat;          /* JAVASCRIPT_COMPAT flag */
2155    BOOL   endonly;               /* Dollar not before final \n */    BOOL   use_ucp;                 /* PCRE_UCP flag */
2156    BOOL   notempty;              /* Empty string match not wanted */    BOOL   endonly;                 /* Dollar not before final \n */
2157    BOOL   notempty_atstart;      /* Empty string match at start not wanted */    BOOL   notempty;                /* Empty string match not wanted */
2158    BOOL   hitend;                /* Hit the end of the subject at some point */    BOOL   notempty_atstart;        /* Empty string match at start not wanted */
2159    BOOL   bsr_anycrlf;           /* \R is just any CRLF, not full Unicode */    BOOL   hitend;                  /* Hit the end of the subject at some point */
2160    const uschar *start_code;     /* For use when recursing */    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */
2161    USPTR  start_subject;         /* Start of the subject string */    BOOL   hasthen;                 /* Pattern contains (*THEN) */
2162    USPTR  end_subject;           /* End of the subject string */    BOOL   ignore_skip_arg;         /* For re-run when SKIP name not found */
2163    USPTR  start_match_ptr;       /* Start of matched string */    const  pcre_uchar *start_code;  /* For use when recursing */
2164    USPTR  end_match_ptr;         /* Subject position at end match */    PCRE_PUCHAR start_subject;      /* Start of the subject string */
2165    USPTR  start_used_ptr;        /* Earliest consulted character */    PCRE_PUCHAR end_subject;        /* End of the subject string */
2166    int    partial;               /* PARTIAL options */    PCRE_PUCHAR start_match_ptr;    /* Start of matched string */
2167    int    end_offset_top;        /* Highwater mark at end of match */    PCRE_PUCHAR end_match_ptr;      /* Subject position at end match */
2168    int    capture_last;          /* Most recent capture number */    PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */
2169    int    start_offset;          /* The start offset value */    int    partial;                 /* PARTIAL options */
2170    eptrblock *eptrchain;         /* Chain of eptrblocks for tail recursions */    int    end_offset_top;          /* Highwater mark at end of match */
2171    int    eptrn;                 /* Next free eptrblock */    int    capture_last;            /* Most recent capture number */
2172    recursion_info *recursive;    /* Linked list of recursion data */    int    start_offset;            /* The start offset value */
2173    void  *callout_data;          /* To pass back to callouts */    int    match_function_type;     /* Set for certain special calls of MATCH() */
2174    const uschar *mark;           /* Mark pointer to pass back */    eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */
2175      int    eptrn;                   /* Next free eptrblock */
2176      recursion_info *recursive;      /* Linked list of recursion data */
2177      void  *callout_data;            /* To pass back to callouts */
2178      const  pcre_uchar *mark;        /* Mark pointer to pass back on success */
2179      const  pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */
2180      const  pcre_uchar *once_target; /* Where to back up to for atomic groups */
2181    #ifdef NO_RECURSE
2182      void  *match_frames_base;       /* For remembering malloc'd frames */
2183    #endif
2184  } match_data;  } match_data;
2185    
2186  /* 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
2187  functions. */  functions. */
2188    
2189  typedef struct dfa_match_data {  typedef struct dfa_match_data {
2190    const uschar *start_code;     /* Start of the compiled pattern */    const pcre_uchar *start_code;     /* Start of the compiled pattern */
2191    const uschar *start_subject;  /* Start of the subject string */    const pcre_uchar *start_subject ; /* Start of the subject string */
2192    const uschar *end_subject;    /* End of subject string */    const pcre_uchar *end_subject;    /* End of subject string */
2193    const uschar *start_used_ptr; /* Earliest consulted character */    const pcre_uchar *start_used_ptr; /* Earliest consulted character */
2194    const uschar *tables;         /* Character tables */    const pcre_uint8 *tables;         /* Character tables */
2195    int   start_offset;           /* The start offset value */    int   start_offset;               /* The start offset value */
2196    int   moptions;               /* Match options */    int   moptions;                   /* Match options */
2197    int   poptions;               /* Pattern options */    int   poptions;                   /* Pattern options */
2198    int    nltype;                /* Newline type */    int   nltype;                     /* Newline type */
2199    int    nllen;                 /* Newline string length */    int   nllen;                      /* Newline string length */
2200    uschar nl[4];                 /* Newline string when fixed */    pcre_uchar nl[4];                 /* Newline string when fixed */
2201    void  *callout_data;          /* To pass back to callouts */    void *callout_data;               /* To pass back to callouts */
2202      dfa_recursion_info *recursive;    /* Linked list of recursion data */
2203  } dfa_match_data;  } dfa_match_data;
2204    
2205  /* Bit definitions for entries in the pcre_ctypes table. */  /* Bit definitions for entries in the pcre_ctypes table. */
# Line 1861  total length. */ Line 2235  total length. */
2235  #define ctypes_offset (cbits_offset + cbit_length)  #define ctypes_offset (cbits_offset + cbit_length)
2236  #define tables_length (ctypes_offset + 256)  #define tables_length (ctypes_offset + 256)
2237    
2238    /* Internal function and data prefixes. */
2239    
2240    #ifdef COMPILE_PCRE8
2241    #ifndef PUBL
2242    #define PUBL(name) pcre_##name
2243    #endif
2244    #ifndef PRIV
2245    #define PRIV(name) _pcre_##name
2246    #endif
2247    #else /* COMPILE_PCRE8 */
2248    #ifdef COMPILE_PCRE16
2249    #ifndef PUBL
2250    #define PUBL(name) pcre16_##name
2251    #endif
2252    #ifndef PRIV
2253    #define PRIV(name) _pcre16_##name
2254    #endif
2255    #else
2256    #error Unsupported compiling mode
2257    #endif /* COMPILE_PCRE16 */
2258    #endif /* COMPILE_PCRE8 */
2259    
2260  /* 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
2261  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
2262  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 1878  of the exported public functions. They h Line 2274  of the exported public functions. They h
2274  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
2275  pcre_tables.c module. */  pcre_tables.c module. */
2276    
2277  extern const int    _pcre_utf8_table1[];  #ifdef COMPILE_PCRE8
2278  extern const int    _pcre_utf8_table2[];  
2279  extern const int    _pcre_utf8_table3[];  extern const int            PRIV(utf8_table1)[];
2280  extern const uschar _pcre_utf8_table4[];  extern const int            PRIV(utf8_table1_size);
2281    extern const int            PRIV(utf8_table2)[];
2282    extern const int            PRIV(utf8_table3)[];
2283    extern const pcre_uint8     PRIV(utf8_table4)[];
2284    
2285  extern const int    _pcre_utf8_table1_size;  #endif /* COMPILE_PCRE8 */
2286    
2287  extern const char   _pcre_utt_names[];  extern const char           PRIV(utt_names)[];
2288  extern const ucp_type_table _pcre_utt[];  extern const ucp_type_table PRIV(utt)[];
2289  extern const int _pcre_utt_size;  extern const int            PRIV(utt_size);
2290    
2291  extern const uschar _pcre_default_tables[];  extern const pcre_uint8     PRIV(default_tables)[];
2292    
2293  extern const uschar _pcre_OP_lengths[];  extern const pcre_uint8     PRIV(OP_lengths)[];
2294    
2295    
2296  /* Internal shared functions. These are functions that are used by more than  /* Internal shared functions. These are functions that are used by more than
2297  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
2298  sense, but are not part of the PCRE public API. */  sense, but are not part of the PCRE public API. */
2299    
2300  extern const uschar *_pcre_find_bracket(const uschar *, BOOL, int);  /* String comparison functions. */
2301  extern BOOL          _pcre_is_newline(USPTR, int, USPTR, int *, BOOL);  #ifdef COMPILE_PCRE8
2302  extern int           _pcre_ord2utf8(int, uschar *);  
2303  extern real_pcre    *_pcre_try_flipped(const real_pcre *, real_pcre *,  #define STRCMP_UC_UC(str1, str2) \
2304                         const pcre_study_data *, pcre_study_data *);    strcmp((char *)(str1), (char *)(str2))
2305  extern int           _pcre_valid_utf8(USPTR, int, int *);  #define STRCMP_UC_C8(str1, str2) \
2306  extern BOOL          _pcre_was_newline(USPTR, int, USPTR, int *, BOOL);    strcmp((char *)(str1), (str2))
2307  extern BOOL          _pcre_xclass(int, const uschar *);  #define STRNCMP_UC_UC(str1, str2, num) \
2308      strncmp((char *)(str1), (char *)(str2), (num))
2309    #define STRNCMP_UC_C8(str1, str2, num) \
2310      strncmp((char *)(str1), (str2), (num))
2311    #define STRLEN_UC(str) strlen((const char *)str)
2312    
2313    #else
2314    
2315    extern int               PRIV(strcmp_uc_uc)(const pcre_uchar *,
2316                               const pcre_uchar *);
2317    extern int               PRIV(strcmp_uc_c8)(const pcre_uchar *,
2318                               const char *);
2319    extern int               PRIV(strncmp_uc_uc)(const pcre_uchar *,
2320                               const pcre_uchar *, unsigned int num);
2321    extern int               PRIV(strncmp_uc_c8)(const pcre_uchar *,
2322                               const char *, unsigned int num);
2323    extern unsigned int      PRIV(strlen_uc)(const pcre_uchar *str);
2324    
2325    #define STRCMP_UC_UC(str1, str2) \
2326      PRIV(strcmp_uc_uc)((str1), (str2))
2327    #define STRCMP_UC_C8(str1, str2) \
2328      PRIV(strcmp_uc_c8)((str1), (str2))
2329    #define STRNCMP_UC_UC(str1, str2, num) \
2330      PRIV(strncmp_uc_uc)((str1), (str2), (num))
2331    #define STRNCMP_UC_C8(str1, str2, num) \
2332      PRIV(strncmp_uc_c8)((str1), (str2), (num))
2333    #define STRLEN_UC(str) PRIV(strlen_uc)(str)
2334    
2335    #endif /* COMPILE_PCRE8 */
2336    
2337    extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);
2338    extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2339                               int *, BOOL);
2340    extern int               PRIV(ord2utf)(pcre_uint32, pcre_uchar *);
2341    extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);
2342    extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2343                               int *, BOOL);
2344    extern BOOL              PRIV(xclass)(int, const pcre_uchar *, BOOL);
2345    
2346    #ifdef SUPPORT_JIT
2347    extern void              PRIV(jit_compile)(const REAL_PCRE *,
2348                               PUBL(extra) *, int);
2349    extern int               PRIV(jit_exec)(const REAL_PCRE *, const PUBL(extra) *,
2350                               const pcre_uchar *, int, int, int, int *, int);
2351    extern void              PRIV(jit_free)(void *);
2352    extern int               PRIV(jit_get_size)(void *);
2353    extern const char*       PRIV(jit_get_target)(void);
2354    #endif
2355    
2356  /* Unicode character database (UCD) */  /* Unicode character database (UCD) */
2357    
2358  typedef struct {  typedef struct {
2359    uschar script;    pcre_uint8 script;     /* ucp_Arabic, etc. */
2360    uschar chartype;    pcre_uint8 chartype;   /* ucp_Cc, etc. (general categories) */
2361    pcre_int32 other_case;    pcre_uint8 gbprop;     /* ucp_gbControl, etc. (grapheme break property) */
2362      pcre_int32 other_case; /* offset to other case, or zero if none */
2363  } ucd_record;  } ucd_record;
2364    
2365  extern const ucd_record  _pcre_ucd_records[];  extern const ucd_record  PRIV(ucd_records)[];
2366  extern const uschar      _pcre_ucd_stage1[];  extern const pcre_uint8  PRIV(ucd_stage1)[];
2367  extern const pcre_uint16 _pcre_ucd_stage2[];  extern const pcre_uint16 PRIV(ucd_stage2)[];
2368  extern const int         _pcre_ucp_gentype[];  extern const int         PRIV(ucp_gentype)[];
2369    extern const pcre_uint32 PRIV(ucp_gbtable)[];
2370    #ifdef SUPPORT_JIT
2371    extern const int         PRIV(ucp_typerange)[];
2372    #endif
2373    
2374    #ifdef SUPPORT_UCP
2375  /* UCD access macros */  /* UCD access macros */
2376    
2377  #define UCD_BLOCK_SIZE 128  #define UCD_BLOCK_SIZE 128
2378  #define GET_UCD(ch) (_pcre_ucd_records + \  #define GET_UCD(ch) (PRIV(ucd_records) + \
2379          _pcre_ucd_stage2[_pcre_ucd_stage1[(ch) / UCD_BLOCK_SIZE] * \          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(ch) / UCD_BLOCK_SIZE] * \
2380          UCD_BLOCK_SIZE + (ch) % UCD_BLOCK_SIZE])          UCD_BLOCK_SIZE + (ch) % UCD_BLOCK_SIZE])
2381    
2382  #define UCD_CHARTYPE(ch)  GET_UCD(ch)->chartype  #define UCD_CHARTYPE(ch)    GET_UCD(ch)->chartype
2383  #define UCD_SCRIPT(ch)    GET_UCD(ch)->script  #define UCD_SCRIPT(ch)      GET_UCD(ch)->script
2384  #define UCD_CATEGORY(ch)  _pcre_ucp_gentype[UCD_CHARTYPE(ch)]  #define UCD_CATEGORY(ch)    PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]
2385  #define UCD_OTHERCASE(ch) (ch + GET_UCD(ch)->other_case)  #define UCD_GRAPHBREAK(ch)  GET_UCD(ch)->gbprop
2386    #define UCD_OTHERCASE(ch)   (ch + GET_UCD(ch)->other_case)
2387    
2388    #endif /* SUPPORT_UCP */
2389    
2390  #endif  #endif
2391    

Legend:
Removed from v.602  
changed lines
  Added in v.1033

  ViewVC Help
Powered by ViewVC 1.1.5