/[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 349 by ph10, Wed Jul 2 18:42:11 2008 UTC revision 1688 by ph10, Fri Feb 24 17:30:30 2017 UTC
# Line 7  Line 7 
7  and semantics are as close as possible to those of the Perl 5 language.  and semantics are as close as possible to those of the Perl 5 language.
8    
9                         Written by Philip Hazel                         Written by Philip Hazel
10             Copyright (c) 1997-2008 University of Cambridge             Copyright (c) 1997-2016 University of Cambridge
11    
12  -----------------------------------------------------------------------------  -----------------------------------------------------------------------------
13  Redistribution and use in source and binary forms, with or without  Redistribution and use in source and binary forms, with or without
# Line 40  POSSIBILITY OF SUCH DAMAGE. Line 40  POSSIBILITY OF SUCH DAMAGE.
40    
41  /* This header contains definitions that are shared between the different  /* This header contains definitions that are shared between the different
42  modules, but which are not relevant to the exported API. This includes some  modules, but which are not relevant to the exported API. This includes some
43  functions whose names all begin with "_pcre_". */  functions whose names all begin with "_pcre_", "_pcre16_" or "_pcre32_"
44    depending on the PRIV macro. */
45    
46  #ifndef PCRE_INTERNAL_H  #ifndef PCRE_INTERNAL_H
47  #define PCRE_INTERNAL_H  #define PCRE_INTERNAL_H
48    
49  /* Define DEBUG to get debugging output on stdout. */  /* Define PCRE_DEBUG to get debugging output on stdout. */
50    
51  #if 0  #if 0
52  #define DEBUG  #define PCRE_DEBUG
53    #endif
54    
55    /* PCRE is compiled as an 8 bit library if it is not requested otherwise. */
56    
57    #if !defined COMPILE_PCRE16 && !defined COMPILE_PCRE32
58    #define COMPILE_PCRE8
59    #endif
60    
61    /* If SUPPORT_UCP is defined, SUPPORT_UTF must also be defined. The
62    "configure" script ensures this, but not everybody uses "configure". */
63    
64    #if defined SUPPORT_UCP && !(defined SUPPORT_UTF)
65    #define SUPPORT_UTF 1
66    #endif
67    
68    /* We define SUPPORT_UTF if SUPPORT_UTF8 is enabled for compatibility
69    reasons with existing code. */
70    
71    #if defined SUPPORT_UTF8 && !(defined SUPPORT_UTF)
72    #define SUPPORT_UTF 1
73    #endif
74    
75    /* Fixme: SUPPORT_UTF8 should be eventually disappear from the code.
76    Until then we define it if SUPPORT_UTF is defined. */
77    
78    #if defined SUPPORT_UTF && !(defined SUPPORT_UTF8)
79    #define SUPPORT_UTF8 1
80    #endif
81    
82    /* We do not support both EBCDIC and UTF-8/16/32 at the same time. The "configure"
83    script prevents both being selected, but not everybody uses "configure". */
84    
85    #if defined EBCDIC && defined SUPPORT_UTF
86    #error The use of both EBCDIC and SUPPORT_UTF is not supported.
87  #endif  #endif
88    
89  /* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef  /* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef
# Line 60  It turns out that the Mac Debugging.h he Line 95  It turns out that the Mac Debugging.h he
95  be absolutely sure we get our version. */  be absolutely sure we get our version. */
96    
97  #undef DPRINTF  #undef DPRINTF
98  #ifdef DEBUG  #ifdef PCRE_DEBUG
99  #define DPRINTF(p) printf p  #define DPRINTF(p) printf p
100  #else  #else
101  #define DPRINTF(p) /* Nothing */  #define DPRINTF(p) /* Nothing */
# Line 72  setjmp and stdarg are used is when NO_RE Line 107  setjmp and stdarg are used is when NO_RE
107    
108  #include <ctype.h>  #include <ctype.h>
109  #include <limits.h>  #include <limits.h>
 #include <setjmp.h>  
 #include <stdarg.h>  
110  #include <stddef.h>  #include <stddef.h>
111  #include <stdio.h>  #include <stdio.h>
112  #include <stdlib.h>  #include <stdlib.h>
113  #include <string.h>  #include <string.h>
114    
115    /* Valgrind (memcheck) support */
116    
117    #ifdef SUPPORT_VALGRIND
118    #include <valgrind/memcheck.h>
119    #endif
120    
121  /* When compiling a DLL for Windows, the exported symbols have to be declared  /* When compiling a DLL for Windows, the exported symbols have to be declared
122  using some MS magic. I found some useful information on this web page:  using some MS magic. I found some useful information on this web page:
123  http://msdn2.microsoft.com/en-us/library/y4h7bcy6(VS.80).aspx. According to the  http://msdn2.microsoft.com/en-us/library/y4h7bcy6(VS.80).aspx. According to the
# Line 132  PCRE_EXP_DATA_DEFN only if they are not Line 171  PCRE_EXP_DATA_DEFN only if they are not
171  #  endif  #  endif
172  #endif  #endif
173    
174  /* We need to have types that specify unsigned 16-bit and 32-bit integers. We  /* When compiling with the MSVC compiler, it is sometimes necessary to include
175    a "calling convention" before exported function names. (This is secondhand
176    information; I know nothing about MSVC myself). For example, something like
177    
178      void __cdecl function(....)
179    
180    might be needed. In order so make this easy, all the exported functions have
181    PCRE_CALL_CONVENTION just before their names. It is rarely needed; if not
182    set, we ensure here that it has no effect. */
183    
184    #ifndef PCRE_CALL_CONVENTION
185    #define PCRE_CALL_CONVENTION
186    #endif
187    
188    /* We need to have types that specify unsigned 8, 16 and 32-bit integers. We
189  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
190  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
191  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
192  preprocessor time in standard C environments. */  preprocessor time in standard C environments. */
193    
194    typedef unsigned char pcre_uint8;
195    
196  #if USHRT_MAX == 65535  #if USHRT_MAX == 65535
197    typedef unsigned short pcre_uint16;  typedef unsigned short pcre_uint16;
198    typedef short pcre_int16;  typedef short pcre_int16;
199    #define PCRE_UINT16_MAX USHRT_MAX
200    #define PCRE_INT16_MAX SHRT_MAX
201  #elif UINT_MAX == 65535  #elif UINT_MAX == 65535
202    typedef unsigned int pcre_uint16;  typedef unsigned int pcre_uint16;
203    typedef int pcre_int16;  typedef int pcre_int16;
204    #define PCRE_UINT16_MAX UINT_MAX
205    #define PCRE_INT16_MAX INT_MAX
206  #else  #else
207    #error Cannot determine a type for 16-bit unsigned integers  #error Cannot determine a type for 16-bit integers
208    #endif
209    
210    #if UINT_MAX == 4294967295U
211    typedef unsigned int pcre_uint32;
212    typedef int pcre_int32;
213    #define PCRE_UINT32_MAX UINT_MAX
214    #define PCRE_INT32_MAX INT_MAX
215    #elif ULONG_MAX == 4294967295UL
216    typedef unsigned long int pcre_uint32;
217    typedef long int pcre_int32;
218    #define PCRE_UINT32_MAX ULONG_MAX
219    #define PCRE_INT32_MAX LONG_MAX
220    #else
221    #error Cannot determine a type for 32-bit integers
222    #endif
223    
224    /* When checking for integer overflow in pcre_compile(), we need to handle
225    large integers. If a 64-bit integer type is available, we can use that.
226    Otherwise we have to cast to double, which of course requires floating point
227    arithmetic. Handle this by defining a macro for the appropriate type. If
228    stdint.h is available, include it; it may define INT64_MAX. Systems that do not
229    have stdint.h (e.g. Solaris) may have inttypes.h. The macro int64_t may be set
230    by "configure". */
231    
232    #if defined HAVE_STDINT_H
233    #include <stdint.h>
234    #elif defined HAVE_INTTYPES_H
235    #include <inttypes.h>
236  #endif  #endif
237    
238  #if UINT_MAX == 4294967295  #if defined INT64_MAX || defined int64_t
239    typedef unsigned int pcre_uint32;  #define INT64_OR_DOUBLE int64_t
   typedef int pcre_int32;  
 #elif ULONG_MAX == 4294967295  
   typedef unsigned long int pcre_uint32;  
   typedef long int pcre_int32;  
240  #else  #else
241    #error Cannot determine a type for 32-bit unsigned integers  #define INT64_OR_DOUBLE double
242  #endif  #endif
243    
244  /* All character handling must be done as unsigned characters. Otherwise there  /* All character handling must be done as unsigned characters. Otherwise there
245  are problems with top-bit-set characters and functions such as isspace().  are problems with top-bit-set characters and functions such as isspace().
246  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 *,
247  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
248  to save lots of typing. I tried "uchar", but it causes problems on Digital  called pcre_uchar.
249  Unix, where it is defined in sys/types, so use "uschar" instead. */  
250    The IN_UCHARS macro multiply its argument with the byte size of the current
251    pcre_uchar type. Useful for memcpy and such operations, whose require the
252    byte size of their input/output buffers.
253    
254    The MAX_255 macro checks whether its pcre_uchar input is less than 256.
255    
256    The TABLE_GET macro is designed for accessing elements of tables whose contain
257    exactly 256 items. When the character is able to contain more than 256
258    items, some check is needed before accessing these tables.
259    */
260    
261    #if defined COMPILE_PCRE8
262    
263  typedef unsigned char uschar;  typedef unsigned char pcre_uchar;
264    #define IN_UCHARS(x) (x)
265    #define MAX_255(c) 1
266    #define TABLE_GET(c, table, default) ((table)[c])
267    
268    #elif defined COMPILE_PCRE16
269    
270    #if USHRT_MAX != 65535
271    /* This is a warning message. Change PCRE_UCHAR16 to a 16 bit data type in
272    pcre.h(.in) and disable (comment out) this message. */
273    #error Warning: PCRE_UCHAR16 is not a 16 bit data type.
274    #endif
275    
276    typedef pcre_uint16 pcre_uchar;
277    #define UCHAR_SHIFT (1)
278    #define IN_UCHARS(x) ((x) * 2)
279    #define MAX_255(c) ((c) <= 255u)
280    #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
281    
282    #elif defined COMPILE_PCRE32
283    
284    typedef pcre_uint32 pcre_uchar;
285    #define UCHAR_SHIFT (2)
286    #define IN_UCHARS(x) ((x) * 4)
287    #define MAX_255(c) ((c) <= 255u)
288    #define TABLE_GET(c, table, default) (MAX_255(c)? ((table)[c]):(default))
289    
290    #else
291    #error Unsupported compiling mode
292    #endif /* COMPILE_PCRE[8|16|32] */
293    
294  /* 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
295  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 188  start/end of string field names are. */ Line 312  start/end of string field names are. */
312  #define IS_NEWLINE(p) \  #define IS_NEWLINE(p) \
313    ((NLBLOCK->nltype != NLTYPE_FIXED)? \    ((NLBLOCK->nltype != NLTYPE_FIXED)? \
314      ((p) < NLBLOCK->PSEND && \      ((p) < NLBLOCK->PSEND && \
315       _pcre_is_newline((p), NLBLOCK->nltype, NLBLOCK->PSEND, &(NLBLOCK->nllen),\       PRIV(is_newline)((p), NLBLOCK->nltype, NLBLOCK->PSEND, \
316         utf8)) \         &(NLBLOCK->nllen), utf)) \
317      : \      : \
318      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \      ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
319       (p)[0] == NLBLOCK->nl[0] && \       UCHAR21TEST(p) == NLBLOCK->nl[0] && \
320       (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || UCHAR21TEST(p+1) == NLBLOCK->nl[1])       \
321      ) \      ) \
322    )    )
323    
# Line 202  start/end of string field names are. */ Line 326  start/end of string field names are. */
326  #define WAS_NEWLINE(p) \  #define WAS_NEWLINE(p) \
327    ((NLBLOCK->nltype != NLTYPE_FIXED)? \    ((NLBLOCK->nltype != NLTYPE_FIXED)? \
328      ((p) > NLBLOCK->PSSTART && \      ((p) > NLBLOCK->PSSTART && \
329       _pcre_was_newline((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \       PRIV(was_newline)((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \
330         &(NLBLOCK->nllen), utf8)) \         &(NLBLOCK->nllen), utf)) \
331      : \      : \
332      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \      ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
333       (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \       UCHAR21TEST(p - NLBLOCK->nllen) == NLBLOCK->nl[0] &&              \
334       (NLBLOCK->nllen == 1 || (p)[-NLBLOCK->nllen+1] == NLBLOCK->nl[1]) \       (NLBLOCK->nllen == 1 || UCHAR21TEST(p - NLBLOCK->nllen + 1) == NLBLOCK->nl[1]) \
335      ) \      ) \
336    )    )
337    
# Line 221  used for the external interface and appe Line 345  used for the external interface and appe
345  must begin with PCRE_. */  must begin with PCRE_. */
346    
347  #ifdef CUSTOM_SUBJECT_PTR  #ifdef CUSTOM_SUBJECT_PTR
348  #define PCRE_SPTR CUSTOM_SUBJECT_PTR  #define PCRE_PUCHAR CUSTOM_SUBJECT_PTR
 #define USPTR CUSTOM_SUBJECT_PTR  
349  #else  #else
350  #define PCRE_SPTR const char *  #define PCRE_PUCHAR const pcre_uchar *
 #define USPTR const unsigned char *  
351  #endif  #endif
352    
   
   
353  /* Include the public PCRE header and the definitions of UCP character property  /* Include the public PCRE header and the definitions of UCP character property
354  values. */  values. */
355    
356  #include "pcre.h"  #include "pcre.h"
357  #include "ucp.h"  #include "ucp.h"
358    
359    #ifdef COMPILE_PCRE32
360    /* Assert that the public PCRE_UCHAR32 is a 32-bit type */
361    typedef int __assert_pcre_uchar32_size[sizeof(PCRE_UCHAR32) == 4 ? 1 : -1];
362    #endif
363    
364  /* When compiling for use with the Virtual Pascal compiler, these functions  /* When compiling for use with the Virtual Pascal compiler, these functions
365  need to have their names changed. PCRE must be compiled with the -DVPCOMPAT  need to have their names changed. PCRE must be compiled with the -DVPCOMPAT
366  option on the command line. */  option on the command line. */
# Line 297  The macros are controlled by the value o Line 422  The macros are controlled by the value o
422  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
423  is automated on Unix systems via the "configure" command. */  is automated on Unix systems via the "configure" command. */
424    
425    #if defined COMPILE_PCRE8
426    
427  #if LINK_SIZE == 2  #if LINK_SIZE == 2
428    
429  #define PUT(a,n,d)   \  #define PUT(a,n,d)   \
# Line 333  is automated on Unix systems via the "co Line 460  is automated on Unix systems via the "co
460  #define GET(a,n) \  #define GET(a,n) \
461    (((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])
462    
463  #define MAX_PATTERN_SIZE (1 << 30)   /* Keep it positive */  /* Keep it positive */
464    #define MAX_PATTERN_SIZE (1 << 30)
465    
466    #else
467    #error LINK_SIZE must be either 2, 3, or 4
468    #endif
469    
470    #elif defined COMPILE_PCRE16
471    
472    #if LINK_SIZE == 2
473    
474    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
475    #undef LINK_SIZE
476    #define LINK_SIZE 1
477    
478    #define PUT(a,n,d)   \
479      (a[n] = (d))
480    
481    #define GET(a,n) \
482      (a[n])
483    
484    #define MAX_PATTERN_SIZE (1 << 16)
485    
486    #elif LINK_SIZE == 3 || LINK_SIZE == 4
487    
488    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
489    #undef LINK_SIZE
490    #define LINK_SIZE 2
491    
492    #define PUT(a,n,d)   \
493      (a[n] = (d) >> 16), \
494      (a[(n)+1] = (d) & 65535)
495    
496    #define GET(a,n) \
497      (((a)[n] << 16) | (a)[(n)+1])
498    
499    /* Keep it positive */
500    #define MAX_PATTERN_SIZE (1 << 30)
501    
502  #else  #else
503  #error LINK_SIZE must be either 2, 3, or 4  #error LINK_SIZE must be either 2, 3, or 4
504  #endif  #endif
505    
506    #elif defined COMPILE_PCRE32
507    
508    /* Only supported LINK_SIZE is 4 */
509    /* Redefine LINK_SIZE as a multiple of sizeof(pcre_uchar) */
510    #undef LINK_SIZE
511    #define LINK_SIZE 1
512    
513    #define PUT(a,n,d)   \
514      (a[n] = (d))
515    
516    #define GET(a,n) \
517      (a[n])
518    
519    /* Keep it positive */
520    #define MAX_PATTERN_SIZE (1 << 30)
521    
522    #else
523    #error Unsupported compiling mode
524    #endif /* COMPILE_PCRE[8|16|32] */
525    
526  /* Convenience macro defined in terms of the others */  /* Convenience macro defined in terms of the others */
527    
# Line 350  is automated on Unix systems via the "co Line 532  is automated on Unix systems via the "co
532  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
533  capturing parenthesis numbers in back references. */  capturing parenthesis numbers in back references. */
534    
535    #if defined COMPILE_PCRE8
536    
537    #define IMM2_SIZE 2
538    
539  #define PUT2(a,n,d)   \  #define PUT2(a,n,d)   \
540    a[n] = (d) >> 8; \    a[n] = (d) >> 8; \
541    a[(n)+1] = (d) & 255    a[(n)+1] = (d) & 255
542    
543    /* For reasons that I do not understand, the expression in this GET2 macro is
544    treated by gcc as a signed expression, even when a is declared as unsigned. It
545    seems that any kind of arithmetic results in a signed value. */
546    
547  #define GET2(a,n) \  #define GET2(a,n) \
548    (((a)[n] << 8) | (a)[(n)+1])    (unsigned int)(((a)[n] << 8) | (a)[(n)+1])
549    
550    #elif defined COMPILE_PCRE16
551    
552  #define PUT2INC(a,n,d)  PUT2(a,n,d), a += 2  #define IMM2_SIZE 1
553    
554    #define PUT2(a,n,d)   \
555       a[n] = d
556    
557    #define GET2(a,n) \
558       a[n]
559    
560    #elif defined COMPILE_PCRE32
561    
562    #define IMM2_SIZE 1
563    
564    #define PUT2(a,n,d)   \
565       a[n] = d
566    
567    #define GET2(a,n) \
568       a[n]
569    
570  /* When UTF-8 encoding is being used, a character is no longer just a single  #else
571  byte. The macros for character handling generate simple sequences when used in  #error Unsupported compiling mode
572  byte-mode, and more complicated ones for UTF-8 characters. BACKCHAR should  #endif /* COMPILE_PCRE[8|16|32] */
573  never be called in byte mode. To make sure it can never even appear when UTF-8  
574  support is omitted, we don't even define it. */  #define PUT2INC(a,n,d)  PUT2(a,n,d), a += IMM2_SIZE
575    
576    /* The maximum length of a MARK name is currently one data unit; it may be
577    changed in future to be a fixed number of bytes or to depend on LINK_SIZE. */
578    
579  #ifndef SUPPORT_UTF8  #if defined COMPILE_PCRE16 || defined COMPILE_PCRE32
580  #define NEXTCHAR(p) p++;  #define MAX_MARK ((1u << 16) - 1)
581    #else
582    #define MAX_MARK ((1u << 8) - 1)
583    #endif
584    
585    /* There is a proposed future special "UTF-21" mode, in which only the lowest
586    21 bits of a 32-bit character are interpreted as UTF, with the remaining 11
587    high-order bits available to the application for other uses. In preparation for
588    the future implementation of this mode, there are macros that load a data item
589    and, if in this special mode, mask it to 21 bits. These macros all have names
590    starting with UCHAR21. In all other modes, including the normal 32-bit
591    library, the macros all have the same simple definitions. When the new mode is
592    implemented, it is expected that these definitions will be varied appropriately
593    using #ifdef when compiling the library that supports the special mode. */
594    
595    #define UCHAR21(eptr)        (*(eptr))
596    #define UCHAR21TEST(eptr)    (*(eptr))
597    #define UCHAR21INC(eptr)     (*(eptr)++)
598    #define UCHAR21INCTEST(eptr) (*(eptr)++)
599    
600    /* When UTF encoding is being used, a character is no longer just a single
601    byte in 8-bit mode or a single short in 16-bit mode. The macros for character
602    handling generate simple sequences when used in the basic mode, and more
603    complicated ones for UTF characters. GETCHARLENTEST and other macros are not
604    used when UTF is not supported. To make sure they can never even appear when
605    UTF support is omitted, we don't even define them. */
606    
607    #ifndef SUPPORT_UTF
608    
609    /* #define MAX_VALUE_FOR_SINGLE_CHAR */
610    /* #define HAS_EXTRALEN(c) */
611    /* #define GET_EXTRALEN(c) */
612    /* #define NOT_FIRSTCHAR(c) */
613  #define GETCHAR(c, eptr) c = *eptr;  #define GETCHAR(c, eptr) c = *eptr;
614  #define GETCHARTEST(c, eptr) c = *eptr;  #define GETCHARTEST(c, eptr) c = *eptr;
615  #define GETCHARINC(c, eptr) c = *eptr++;  #define GETCHARINC(c, eptr) c = *eptr++;
616  #define GETCHARINCTEST(c, eptr) c = *eptr++;  #define GETCHARINCTEST(c, eptr) c = *eptr++;
617  #define GETCHARLEN(c, eptr, len) c = *eptr;  #define GETCHARLEN(c, eptr, len) c = *eptr;
618    /* #define GETCHARLENTEST(c, eptr, len) */
619  /* #define BACKCHAR(eptr) */  /* #define BACKCHAR(eptr) */
620    /* #define FORWARDCHAR(eptr) */
621    /* #define ACROSSCHAR(condition, eptr, action) */
622    
623  #else   /* SUPPORT_UTF8 */  #else   /* SUPPORT_UTF */
624    
625  /* Advance a character pointer one byte in non-UTF-8 mode and by one character  /* Tests whether the code point needs extra characters to decode. */
 in UTF-8 mode. */  
626    
627  #define NEXTCHAR(p) \  #define HASUTF8EXTRALEN(c) ((c) >= 0xc0)
   p++; \  
   if (utf8) { while((*p & 0xc0) == 0x80) p++; }  
628    
629  /* Get the next UTF-8 character, not advancing the pointer. This is called when  /* Base macro to pick up the remaining bytes of a UTF-8 character, not
630  we know we are in UTF-8 mode. */  advancing the pointer. */
631    
632  #define GETCHAR(c, eptr) \  #define GETUTF8(c, eptr) \
633    c = *eptr; \      { \
634    if (c >= 0xc0) \      if ((c & 0x20) == 0) \
635          c = ((c & 0x1f) << 6) | (eptr[1] & 0x3f); \
636        else if ((c & 0x10) == 0) \
637          c = ((c & 0x0f) << 12) | ((eptr[1] & 0x3f) << 6) | (eptr[2] & 0x3f); \
638        else if ((c & 0x08) == 0) \
639          c = ((c & 0x07) << 18) | ((eptr[1] & 0x3f) << 12) | \
640          ((eptr[2] & 0x3f) << 6) | (eptr[3] & 0x3f); \
641        else if ((c & 0x04) == 0) \
642          c = ((c & 0x03) << 24) | ((eptr[1] & 0x3f) << 18) | \
643              ((eptr[2] & 0x3f) << 12) | ((eptr[3] & 0x3f) << 6) | \
644              (eptr[4] & 0x3f); \
645        else \
646          c = ((c & 0x01) << 30) | ((eptr[1] & 0x3f) << 24) | \
647              ((eptr[2] & 0x3f) << 18) | ((eptr[3] & 0x3f) << 12) | \
648              ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \
649        }
650    
651    /* Base macro to pick up the remaining bytes of a UTF-8 character, advancing
652    the pointer. */
653    
654    #define GETUTF8INC(c, eptr) \
655      { \      { \
656      int gcii; \      if ((c & 0x20) == 0) \
657      int gcaa = _pcre_utf8_table4[c & 0x3f];  /* Number of additional bytes */ \        c = ((c & 0x1f) << 6) | (*eptr++ & 0x3f); \
658      int gcss = 6*gcaa; \      else if ((c & 0x10) == 0) \
659      c = (c & _pcre_utf8_table3[gcaa]) << gcss; \        { \
660      for (gcii = 1; gcii <= gcaa; gcii++) \        c = ((c & 0x0f) << 12) | ((*eptr & 0x3f) << 6) | (eptr[1] & 0x3f); \
661          eptr += 2; \
662          } \
663        else if ((c & 0x08) == 0) \
664          { \
665          c = ((c & 0x07) << 18) | ((*eptr & 0x3f) << 12) | \
666              ((eptr[1] & 0x3f) << 6) | (eptr[2] & 0x3f); \
667          eptr += 3; \
668          } \
669        else if ((c & 0x04) == 0) \
670        { \        { \
671        gcss -= 6; \        c = ((c & 0x03) << 24) | ((*eptr & 0x3f) << 18) | \
672        c |= (eptr[gcii] & 0x3f) << gcss; \            ((eptr[1] & 0x3f) << 12) | ((eptr[2] & 0x3f) << 6) | \
673              (eptr[3] & 0x3f); \
674          eptr += 4; \
675          } \
676        else \
677          { \
678          c = ((c & 0x01) << 30) | ((*eptr & 0x3f) << 24) | \
679              ((eptr[1] & 0x3f) << 18) | ((eptr[2] & 0x3f) << 12) | \
680              ((eptr[3] & 0x3f) << 6) | (eptr[4] & 0x3f); \
681          eptr += 5; \
682        } \        } \
683      }      }
684    
685    #if defined COMPILE_PCRE8
686    
687    /* These macros were originally written in the form of loops that used data
688    from the tables whose names start with PRIV(utf8_table). They were rewritten by
689    a user so as not to use loops, because in some environments this gives a
690    significant performance advantage, and it seems never to do any harm. */
691    
692    /* Tells the biggest code point which can be encoded as a single character. */
693    
694    #define MAX_VALUE_FOR_SINGLE_CHAR 127
695    
696    /* Tests whether the code point needs extra characters to decode. */
697    
698    #define HAS_EXTRALEN(c) ((c) >= 0xc0)
699    
700    /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.
701    Otherwise it has an undefined behaviour. */
702    
703    #define GET_EXTRALEN(c) (PRIV(utf8_table4)[(c) & 0x3f])
704    
705    /* Returns TRUE, if the given character is not the first character
706    of a UTF sequence. */
707    
708    #define NOT_FIRSTCHAR(c) (((c) & 0xc0) == 0x80)
709    
710    /* Get the next UTF-8 character, not advancing the pointer. This is called when
711    we know we are in UTF-8 mode. */
712    
713    #define GETCHAR(c, eptr) \
714      c = *eptr; \
715      if (c >= 0xc0) GETUTF8(c, eptr);
716    
717  /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the  /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
718  pointer. */  pointer. */
719    
720  #define GETCHARTEST(c, eptr) \  #define GETCHARTEST(c, eptr) \
721    c = *eptr; \    c = *eptr; \
722    if (utf8 && c >= 0xc0) \    if (utf && c >= 0xc0) GETUTF8(c, eptr);
     { \  
     int gcii; \  
     int gcaa = _pcre_utf8_table4[c & 0x3f];  /* Number of additional bytes */ \  
     int gcss = 6*gcaa; \  
     c = (c & _pcre_utf8_table3[gcaa]) << gcss; \  
     for (gcii = 1; gcii <= gcaa; gcii++) \  
       { \  
       gcss -= 6; \  
       c |= (eptr[gcii] & 0x3f) << gcss; \  
       } \  
     }  
723    
724  /* Get the next UTF-8 character, advancing the pointer. This is called when we  /* Get the next UTF-8 character, advancing the pointer. This is called when we
725  know we are in UTF-8 mode. */  know we are in UTF-8 mode. */
726    
727  #define GETCHARINC(c, eptr) \  #define GETCHARINC(c, eptr) \
728    c = *eptr++; \    c = *eptr++; \
729    if (c >= 0xc0) \    if (c >= 0xc0) GETUTF8INC(c, eptr);
     { \  
     int gcaa = _pcre_utf8_table4[c & 0x3f];  /* Number of additional bytes */ \  
     int gcss = 6*gcaa; \  
     c = (c & _pcre_utf8_table3[gcaa]) << gcss; \  
     while (gcaa-- > 0) \  
       { \  
       gcss -= 6; \  
       c |= (*eptr++ & 0x3f) << gcss; \  
       } \  
     }  
730    
731  /* Get the next character, testing for UTF-8 mode, and advancing the pointer */  /* Get the next character, testing for UTF-8 mode, and advancing the pointer.
732    This is called when we don't know if we are in UTF-8 mode. */
733    
734  #define GETCHARINCTEST(c, eptr) \  #define GETCHARINCTEST(c, eptr) \
735    c = *eptr++; \    c = *eptr++; \
736    if (utf8 && c >= 0xc0) \    if (utf && c >= 0xc0) GETUTF8INC(c, eptr);
737    
738    /* Base macro to pick up the remaining bytes of a UTF-8 character, not
739    advancing the pointer, incrementing the length. */
740    
741    #define GETUTF8LEN(c, eptr, len) \
742      { \      { \
743      int gcaa = _pcre_utf8_table4[c & 0x3f];  /* Number of additional bytes */ \      if ((c & 0x20) == 0) \
744      int gcss = 6*gcaa; \        { \
745      c = (c & _pcre_utf8_table3[gcaa]) << gcss; \        c = ((c & 0x1f) << 6) | (eptr[1] & 0x3f); \
746      while (gcaa-- > 0) \        len++; \
747          } \
748        else if ((c & 0x10)  == 0) \
749        { \        { \
750        gcss -= 6; \        c = ((c & 0x0f) << 12) | ((eptr[1] & 0x3f) << 6) | (eptr[2] & 0x3f); \
751        c |= (*eptr++ & 0x3f) << gcss; \        len += 2; \
752          } \
753        else if ((c & 0x08)  == 0) \
754          {\
755          c = ((c & 0x07) << 18) | ((eptr[1] & 0x3f) << 12) | \
756              ((eptr[2] & 0x3f) << 6) | (eptr[3] & 0x3f); \
757          len += 3; \
758          } \
759        else if ((c & 0x04)  == 0) \
760          { \
761          c = ((c & 0x03) << 24) | ((eptr[1] & 0x3f) << 18) | \
762              ((eptr[2] & 0x3f) << 12) | ((eptr[3] & 0x3f) << 6) | \
763              (eptr[4] & 0x3f); \
764          len += 4; \
765          } \
766        else \
767          {\
768          c = ((c & 0x01) << 30) | ((eptr[1] & 0x3f) << 24) | \
769              ((eptr[2] & 0x3f) << 18) | ((eptr[3] & 0x3f) << 12) | \
770              ((eptr[4] & 0x3f) << 6) | (eptr[5] & 0x3f); \
771          len += 5; \
772        } \        } \
773      }      }
774    
# Line 458  if there are extra bytes. This is called Line 777  if there are extra bytes. This is called
777    
778  #define GETCHARLEN(c, eptr, len) \  #define GETCHARLEN(c, eptr, len) \
779    c = *eptr; \    c = *eptr; \
780    if (c >= 0xc0) \    if (c >= 0xc0) GETUTF8LEN(c, eptr, len);
781      { \  
782      int gcii; \  /* Get the next UTF-8 character, testing for UTF-8 mode, not advancing the
783      int gcaa = _pcre_utf8_table4[c & 0x3f];  /* Number of additional bytes */ \  pointer, incrementing length if there are extra bytes. This is called when we
784      int gcss = 6*gcaa; \  do not know if we are in UTF-8 mode. */
785      c = (c & _pcre_utf8_table3[gcaa]) << gcss; \  
786      for (gcii = 1; gcii <= gcaa; gcii++) \  #define GETCHARLENTEST(c, eptr, len) \
787        { \    c = *eptr; \
788        gcss -= 6; \    if (utf && c >= 0xc0) GETUTF8LEN(c, eptr, len);
       c |= (eptr[gcii] & 0x3f) << gcss; \  
       } \  
     len += gcaa; \  
     }  
789    
790  /* If the pointer is not at the start of a character, move it back until  /* If the pointer is not at the start of a character, move it back until
791  it is. This is called only in UTF-8 mode - we don't put a test within the macro  it is. This is called only in UTF-8 mode - we don't put a test within the macro
# Line 478  because almost all calls are already wit Line 793  because almost all calls are already wit
793    
794  #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--  #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--
795    
796  #endif  /* Same as above, just in the other direction. */
797    #define FORWARDCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr++
798    
799    /* Same as above, but it allows a fully customizable form. */
800    #define ACROSSCHAR(condition, eptr, action) \
801      while((condition) && ((eptr) & 0xc0) == 0x80) action
802    
803    #elif defined COMPILE_PCRE16
804    
805    /* Tells the biggest code point which can be encoded as a single character. */
806    
807  /* In case there is no definition of offsetof() provided - though any proper  #define MAX_VALUE_FOR_SINGLE_CHAR 65535
 Standard C system should have one. */  
808    
809  #ifndef offsetof  /* Tests whether the code point needs extra characters to decode. */
810  #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))  
811    #define HAS_EXTRALEN(c) (((c) & 0xfc00) == 0xd800)
812    
813    /* Returns with the additional number of characters if IS_MULTICHAR(c) is TRUE.
814    Otherwise it has an undefined behaviour. */
815    
816    #define GET_EXTRALEN(c) 1
817    
818    /* Returns TRUE, if the given character is not the first character
819    of a UTF sequence. */
820    
821    #define NOT_FIRSTCHAR(c) (((c) & 0xfc00) == 0xdc00)
822    
823    /* Base macro to pick up the low surrogate of a UTF-16 character, not
824    advancing the pointer. */
825    
826    #define GETUTF16(c, eptr) \
827       { c = (((c & 0x3ff) << 10) | (eptr[1] & 0x3ff)) + 0x10000; }
828    
829    /* Get the next UTF-16 character, not advancing the pointer. This is called when
830    we know we are in UTF-16 mode. */
831    
832    #define GETCHAR(c, eptr) \
833      c = *eptr; \
834      if ((c & 0xfc00) == 0xd800) GETUTF16(c, eptr);
835    
836    /* Get the next UTF-16 character, testing for UTF-16 mode, and not advancing the
837    pointer. */
838    
839    #define GETCHARTEST(c, eptr) \
840      c = *eptr; \
841      if (utf && (c & 0xfc00) == 0xd800) GETUTF16(c, eptr);
842    
843    /* Base macro to pick up the low surrogate of a UTF-16 character, advancing
844    the pointer. */
845    
846    #define GETUTF16INC(c, eptr) \
847       { c = (((c & 0x3ff) << 10) | (*eptr++ & 0x3ff)) + 0x10000; }
848    
849    /* Get the next UTF-16 character, advancing the pointer. This is called when we
850    know we are in UTF-16 mode. */
851    
852    #define GETCHARINC(c, eptr) \
853      c = *eptr++; \
854      if ((c & 0xfc00) == 0xd800) GETUTF16INC(c, eptr);
855    
856    /* Get the next character, testing for UTF-16 mode, and advancing the pointer.
857    This is called when we don't know if we are in UTF-16 mode. */
858    
859    #define GETCHARINCTEST(c, eptr) \
860      c = *eptr++; \
861      if (utf && (c & 0xfc00) == 0xd800) GETUTF16INC(c, eptr);
862    
863    /* Base macro to pick up the low surrogate of a UTF-16 character, not
864    advancing the pointer, incrementing the length. */
865    
866    #define GETUTF16LEN(c, eptr, len) \
867       { c = (((c & 0x3ff) << 10) | (eptr[1] & 0x3ff)) + 0x10000; len++; }
868    
869    /* Get the next UTF-16 character, not advancing the pointer, incrementing
870    length if there is a low surrogate. This is called when we know we are in
871    UTF-16 mode. */
872    
873    #define GETCHARLEN(c, eptr, len) \
874      c = *eptr; \
875      if ((c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);
876    
877    /* Get the next UTF-816character, testing for UTF-16 mode, not advancing the
878    pointer, incrementing length if there is a low surrogate. This is called when
879    we do not know if we are in UTF-16 mode. */
880    
881    #define GETCHARLENTEST(c, eptr, len) \
882      c = *eptr; \
883      if (utf && (c & 0xfc00) == 0xd800) GETUTF16LEN(c, eptr, len);
884    
885    /* If the pointer is not at the start of a character, move it back until
886    it is. This is called only in UTF-16 mode - we don't put a test within the
887    macro because almost all calls are already within a block of UTF-16 only
888    code. */
889    
890    #define BACKCHAR(eptr) if ((*eptr & 0xfc00) == 0xdc00) eptr--
891    
892    /* Same as above, just in the other direction. */
893    #define FORWARDCHAR(eptr) if ((*eptr & 0xfc00) == 0xdc00) eptr++
894    
895    /* Same as above, but it allows a fully customizable form. */
896    #define ACROSSCHAR(condition, eptr, action) \
897      if ((condition) && ((eptr) & 0xfc00) == 0xdc00) action
898    
899    #elif defined COMPILE_PCRE32
900    
901    /* These are trivial for the 32-bit library, since all UTF-32 characters fit
902    into one pcre_uchar unit. */
903    #define MAX_VALUE_FOR_SINGLE_CHAR (0x10ffffu)
904    #define HAS_EXTRALEN(c) (0)
905    #define GET_EXTRALEN(c) (0)
906    #define NOT_FIRSTCHAR(c) (0)
907    
908    /* Get the next UTF-32 character, not advancing the pointer. This is called when
909    we know we are in UTF-32 mode. */
910    
911    #define GETCHAR(c, eptr) \
912      c = *(eptr);
913    
914    /* Get the next UTF-32 character, testing for UTF-32 mode, and not advancing the
915    pointer. */
916    
917    #define GETCHARTEST(c, eptr) \
918      c = *(eptr);
919    
920    /* Get the next UTF-32 character, advancing the pointer. This is called when we
921    know we are in UTF-32 mode. */
922    
923    #define GETCHARINC(c, eptr) \
924      c = *((eptr)++);
925    
926    /* Get the next character, testing for UTF-32 mode, and advancing the pointer.
927    This is called when we don't know if we are in UTF-32 mode. */
928    
929    #define GETCHARINCTEST(c, eptr) \
930      c = *((eptr)++);
931    
932    /* Get the next UTF-32 character, not advancing the pointer, not incrementing
933    length (since all UTF-32 is of length 1). This is called when we know we are in
934    UTF-32 mode. */
935    
936    #define GETCHARLEN(c, eptr, len) \
937      GETCHAR(c, eptr)
938    
939    /* Get the next UTF-32character, testing for UTF-32 mode, not advancing the
940    pointer, not incrementing the length (since all UTF-32 is of length 1).
941    This is called when we do not know if we are in UTF-32 mode. */
942    
943    #define GETCHARLENTEST(c, eptr, len) \
944      GETCHARTEST(c, eptr)
945    
946    /* If the pointer is not at the start of a character, move it back until
947    it is. This is called only in UTF-32 mode - we don't put a test within the
948    macro because almost all calls are already within a block of UTF-32 only
949    code.
950    These are all no-ops since all UTF-32 characters fit into one pcre_uchar. */
951    
952    #define BACKCHAR(eptr) do { } while (0)
953    
954    /* Same as above, just in the other direction. */
955    #define FORWARDCHAR(eptr) do { } while (0)
956    
957    /* Same as above, but it allows a fully customizable form. */
958    #define ACROSSCHAR(condition, eptr, action) do { } while (0)
959    
960    #else
961    #error Unsupported compiling mode
962    #endif /* COMPILE_PCRE[8|16|32] */
963    
964    #endif  /* SUPPORT_UTF */
965    
966    /* Tests for Unicode horizontal and vertical whitespace characters must check a
967    number of different values. Using a switch statement for this generates the
968    fastest code (no loop, no memory access), and there are several places in the
969    interpreter code where this happens. In order to ensure that all the case lists
970    remain in step, we use macros so that there is only one place where the lists
971    are defined.
972    
973    These values are also required as lists in pcre_compile.c when processing \h,
974    \H, \v and \V in a character class. The lists are defined in pcre_tables.c, but
975    macros that define the values are here so that all the definitions are
976    together. The lists must be in ascending character order, terminated by
977    NOTACHAR (which is 0xffffffff).
978    
979    Any changes should ensure that the various macros are kept in step with each
980    other. NOTE: The values also appear in pcre_jit_compile.c. */
981    
982    /* ------ ASCII/Unicode environments ------ */
983    
984    #ifndef EBCDIC
985    
986    #define HSPACE_LIST \
987      CHAR_HT, CHAR_SPACE, CHAR_NBSP, \
988      0x1680, 0x180e, 0x2000, 0x2001, 0x2002, 0x2003, 0x2004, 0x2005, \
989      0x2006, 0x2007, 0x2008, 0x2009, 0x200A, 0x202f, 0x205f, 0x3000, \
990      NOTACHAR
991    
992    #define HSPACE_MULTIBYTE_CASES \
993      case 0x1680:  /* OGHAM SPACE MARK */ \
994      case 0x180e:  /* MONGOLIAN VOWEL SEPARATOR */ \
995      case 0x2000:  /* EN QUAD */ \
996      case 0x2001:  /* EM QUAD */ \
997      case 0x2002:  /* EN SPACE */ \
998      case 0x2003:  /* EM SPACE */ \
999      case 0x2004:  /* THREE-PER-EM SPACE */ \
1000      case 0x2005:  /* FOUR-PER-EM SPACE */ \
1001      case 0x2006:  /* SIX-PER-EM SPACE */ \
1002      case 0x2007:  /* FIGURE SPACE */ \
1003      case 0x2008:  /* PUNCTUATION SPACE */ \
1004      case 0x2009:  /* THIN SPACE */ \
1005      case 0x200A:  /* HAIR SPACE */ \
1006      case 0x202f:  /* NARROW NO-BREAK SPACE */ \
1007      case 0x205f:  /* MEDIUM MATHEMATICAL SPACE */ \
1008      case 0x3000   /* IDEOGRAPHIC SPACE */
1009    
1010    #define HSPACE_BYTE_CASES \
1011      case CHAR_HT: \
1012      case CHAR_SPACE: \
1013      case CHAR_NBSP
1014    
1015    #define HSPACE_CASES \
1016      HSPACE_BYTE_CASES: \
1017      HSPACE_MULTIBYTE_CASES
1018    
1019    #define VSPACE_LIST \
1020      CHAR_LF, CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, 0x2028, 0x2029, NOTACHAR
1021    
1022    #define VSPACE_MULTIBYTE_CASES \
1023      case 0x2028:    /* LINE SEPARATOR */ \
1024      case 0x2029     /* PARAGRAPH SEPARATOR */
1025    
1026    #define VSPACE_BYTE_CASES \
1027      case CHAR_LF: \
1028      case CHAR_VT: \
1029      case CHAR_FF: \
1030      case CHAR_CR: \
1031      case CHAR_NEL
1032    
1033    #define VSPACE_CASES \
1034      VSPACE_BYTE_CASES: \
1035      VSPACE_MULTIBYTE_CASES
1036    
1037    /* ------ EBCDIC environments ------ */
1038    
1039    #else
1040    #define HSPACE_LIST CHAR_HT, CHAR_SPACE, CHAR_NBSP, NOTACHAR
1041    
1042    #define HSPACE_BYTE_CASES \
1043      case CHAR_HT: \
1044      case CHAR_SPACE: \
1045      case CHAR_NBSP
1046    
1047    #define HSPACE_CASES HSPACE_BYTE_CASES
1048    
1049    #ifdef EBCDIC_NL25
1050    #define VSPACE_LIST \
1051      CHAR_VT, CHAR_FF, CHAR_CR, CHAR_NEL, CHAR_LF, NOTACHAR
1052    #else
1053    #define VSPACE_LIST \
1054      CHAR_VT, CHAR_FF, CHAR_CR, CHAR_LF, CHAR_NEL, NOTACHAR
1055  #endif  #endif
1056    
1057    #define VSPACE_BYTE_CASES \
1058      case CHAR_LF: \
1059      case CHAR_VT: \
1060      case CHAR_FF: \
1061      case CHAR_CR: \
1062      case CHAR_NEL
1063    
1064  /* These are the public options that can change during matching. */  #define VSPACE_CASES VSPACE_BYTE_CASES
1065    #endif  /* EBCDIC */
1066    
1067  #define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)  /* ------ End of whitespace macros ------ */
1068    
 /* Private flags containing information about the compiled regex. They used to  
 live at the top end of the options word, but that got almost full, so now they  
 are in a 16-bit flags word. */  
1069    
 #define PCRE_NOPARTIAL     0x0001  /* can't use partial with this regex */  
 #define PCRE_FIRSTSET      0x0002  /* first_byte is set */  
 #define PCRE_REQCHSET      0x0004  /* req_byte is set */  
 #define PCRE_STARTLINE     0x0008  /* start after \n for multiline */  
 #define PCRE_JCHANGED      0x0010  /* j option used in regex */  
 #define PCRE_HASCRORLF     0x0020  /* explicit \r or \n in pattern */  
1070    
1071  /* Options for the "extra" block produced by pcre_study(). */  /* Private flags containing information about the compiled regex. They used to
1072    live at the top end of the options word, but that got almost full, so they were
1073    moved to a 16-bit flags word - which got almost full, so now they are in a
1074    32-bit flags word. From release 8.00, PCRE_NOPARTIAL is unused, as the
1075    restrictions on partial matching have been lifted. It remains for backwards
1076    compatibility. */
1077    
1078    #define PCRE_MODE8         0x00000001  /* compiled in 8 bit mode */
1079    #define PCRE_MODE16        0x00000002  /* compiled in 16 bit mode */
1080    #define PCRE_MODE32        0x00000004  /* compiled in 32 bit mode */
1081    #define PCRE_FIRSTSET      0x00000010  /* first_char is set */
1082    #define PCRE_FCH_CASELESS  0x00000020  /* caseless first char */
1083    #define PCRE_REQCHSET      0x00000040  /* req_byte is set */
1084    #define PCRE_RCH_CASELESS  0x00000080  /* caseless requested char */
1085    #define PCRE_STARTLINE     0x00000100  /* start after \n for multiline */
1086    #define PCRE_NOPARTIAL     0x00000200  /* can't use partial with this regex */
1087    #define PCRE_JCHANGED      0x00000400  /* j option used in regex */
1088    #define PCRE_HASCRORLF     0x00000800  /* explicit \r or \n in pattern */
1089    #define PCRE_HASTHEN       0x00001000  /* pattern contains (*THEN) */
1090    #define PCRE_MLSET         0x00002000  /* match limit set by regex */
1091    #define PCRE_RLSET         0x00004000  /* recursion limit set by regex */
1092    #define PCRE_MATCH_EMPTY   0x00008000  /* pattern can match empty string */
1093    
1094    #if defined COMPILE_PCRE8
1095    #define PCRE_MODE          PCRE_MODE8
1096    #elif defined COMPILE_PCRE16
1097    #define PCRE_MODE          PCRE_MODE16
1098    #elif defined COMPILE_PCRE32
1099    #define PCRE_MODE          PCRE_MODE32
1100    #endif
1101    #define PCRE_MODE_MASK     (PCRE_MODE8 | PCRE_MODE16 | PCRE_MODE32)
1102    
1103    /* Flags for the "extra" block produced by pcre_study(). */
1104    
1105  #define PCRE_STUDY_MAPPED   0x01     /* a map of starting chars exists */  #define PCRE_STUDY_MAPPED  0x0001  /* a map of starting chars exists */
1106    #define PCRE_STUDY_MINLEN  0x0002  /* a minimum length field exists */
1107    
1108  /* Masks for identifying the public options that are permitted at compile  /* Masks for identifying the public options that are permitted at compile
1109  time, run time, or study time, respectively. */  time, run time, or study time, respectively. */
# Line 514  time, run time, or study time, respectiv Line 1111  time, run time, or study time, respectiv
1111  #define PCRE_NEWLINE_BITS (PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|PCRE_NEWLINE_ANY| \  #define PCRE_NEWLINE_BITS (PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|PCRE_NEWLINE_ANY| \
1112                             PCRE_NEWLINE_ANYCRLF)                             PCRE_NEWLINE_ANYCRLF)
1113    
1114  #define PUBLIC_OPTIONS \  #define PUBLIC_COMPILE_OPTIONS \
1115    (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \    (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
1116     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \     PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
1117     PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \     PCRE_NO_AUTO_CAPTURE|PCRE_NO_AUTO_POSSESS| \
1118       PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
1119     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \     PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \
1120     PCRE_JAVASCRIPT_COMPAT)     PCRE_JAVASCRIPT_COMPAT|PCRE_UCP|PCRE_NO_START_OPTIMIZE|PCRE_NEVER_UTF)
1121    
1122  #define PUBLIC_EXEC_OPTIONS \  #define PUBLIC_EXEC_OPTIONS \
1123    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \
1124     PCRE_PARTIAL|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)     PCRE_NO_UTF8_CHECK|PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT|PCRE_NEWLINE_BITS| \
1125       PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE|PCRE_NO_START_OPTIMIZE)
1126    
1127  #define PUBLIC_DFA_EXEC_OPTIONS \  #define PUBLIC_DFA_EXEC_OPTIONS \
1128    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \    (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NOTEMPTY_ATSTART| \
1129     PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART|PCRE_NEWLINE_BITS| \     PCRE_NO_UTF8_CHECK|PCRE_PARTIAL_HARD|PCRE_PARTIAL_SOFT|PCRE_DFA_SHORTEST| \
1130     PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)     PCRE_DFA_RESTART|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \
1131       PCRE_NO_START_OPTIMIZE)
1132  #define PUBLIC_STUDY_OPTIONS 0   /* None defined */  
1133    #define PUBLIC_STUDY_OPTIONS \
1134       (PCRE_STUDY_JIT_COMPILE|PCRE_STUDY_JIT_PARTIAL_SOFT_COMPILE| \
1135        PCRE_STUDY_JIT_PARTIAL_HARD_COMPILE|PCRE_STUDY_EXTRA_NEEDED)
1136    
1137    #define PUBLIC_JIT_EXEC_OPTIONS \
1138       (PCRE_NO_UTF8_CHECK|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|\
1139        PCRE_NOTEMPTY_ATSTART|PCRE_PARTIAL_SOFT|PCRE_PARTIAL_HARD)
1140    
1141  /* 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. */  
1142    
1143  #define MAGIC_NUMBER  0x50435245UL   /* 'PCRE' */  #define MAGIC_NUMBER  0x50435245UL   /* 'PCRE' */
1144    
1145  /* Negative values for the firstchar and reqchar variables */  /* This variable is used to detect a loaded regular expression
1146    in different endianness. */
1147    
1148  #define REQ_UNSET (-2)  #define REVERSED_MAGIC_NUMBER  0x45524350UL   /* 'ERCP' */
 #define REQ_NONE  (-1)  
1149    
1150  /* The maximum remaining length of subject we are prepared to search for a  /* The maximum remaining length of subject we are prepared to search for a
1151  req_byte match. */  req_byte match. */
1152    
1153  #define REQ_BYTE_MAX 1000  #define REQ_BYTE_MAX 1000
1154    
1155  /* Flags added to firstbyte or reqbyte; a "non-literal" item is either a  /* Miscellaneous definitions. The #ifndef is to pacify compiler warnings in
1156  variable-length repeat, or a anything other than literal characters. */  environments where these macros are defined elsewhere. Unfortunately, there
1157    is no way to do the same for the typedef. */
 #define REQ_CASELESS 0x0100    /* indicates caselessness */  
 #define REQ_VARY     0x0200    /* reqbyte followed non-literal item */  
   
 /* Miscellaneous definitions */  
1158    
1159  typedef int BOOL;  typedef int BOOL;
1160    
1161    #ifndef FALSE
1162  #define FALSE   0  #define FALSE   0
1163  #define TRUE    1  #define TRUE    1
1164    #endif
1165    
1166    /* If PCRE is to support UTF-8 on EBCDIC platforms, we cannot use normal
1167    character constants like '*' because the compiler would emit their EBCDIC code,
1168    which is different from their ASCII/UTF-8 code. Instead we define macros for
1169    the characters so that they always use the ASCII/UTF-8 code when UTF-8 support
1170    is enabled. When UTF-8 support is not enabled, the definitions use character
1171    literals. Both character and string versions of each character are needed, and
1172    there are some longer strings as well.
1173    
1174    This means that, on EBCDIC platforms, the PCRE library can handle either
1175    EBCDIC, or UTF-8, but not both. To support both in the same compiled library
1176    would need different lookups depending on whether PCRE_UTF8 was set or not.
1177    This would make it impossible to use characters in switch/case statements,
1178    which would reduce performance. For a theoretical use (which nobody has asked
1179    for) in a minority area (EBCDIC platforms), this is not sensible. Any
1180    application that did need both could compile two versions of the library, using
1181    macros to give the functions distinct names. */
1182    
1183    #ifndef SUPPORT_UTF
1184    
1185    /* UTF-8 support is not enabled; use the platform-dependent character literals
1186    so that PCRE works in both ASCII and EBCDIC environments, but only in non-UTF
1187    mode. Newline characters are problematic in EBCDIC. Though it has CR and LF
1188    characters, a common practice has been to use its NL (0x15) character as the
1189    line terminator in C-like processing environments. However, sometimes the LF
1190    (0x25) character is used instead, according to this Unicode document:
1191    
1192    http://unicode.org/standard/reports/tr13/tr13-5.html
1193    
1194    PCRE defaults EBCDIC NL to 0x15, but has a build-time option to select 0x25
1195    instead. Whichever is *not* chosen is defined as NEL.
1196    
1197    In both ASCII and EBCDIC environments, CHAR_NL and CHAR_LF are synonyms for the
1198    same code point. */
1199    
1200    #ifdef EBCDIC
1201    
1202    #ifndef EBCDIC_NL25
1203    #define CHAR_NL                     '\x15'
1204    #define CHAR_NEL                    '\x25'
1205    #define STR_NL                      "\x15"
1206    #define STR_NEL                     "\x25"
1207    #else
1208    #define CHAR_NL                     '\x25'
1209    #define CHAR_NEL                    '\x15'
1210    #define STR_NL                      "\x25"
1211    #define STR_NEL                     "\x15"
1212    #endif
1213    
1214    #define CHAR_LF                     CHAR_NL
1215    #define STR_LF                      STR_NL
1216    
1217    #define CHAR_ESC                    '\047'
1218    #define CHAR_DEL                    '\007'
1219    #define CHAR_NBSP                   '\x41'
1220    #define STR_ESC                     "\047"
1221    #define STR_DEL                     "\007"
1222    
1223    #else  /* Not EBCDIC */
1224    
1225    /* In ASCII/Unicode, linefeed is '\n' and we equate this to NL for
1226    compatibility. NEL is the Unicode newline character; make sure it is
1227    a positive value. */
1228    
1229    #define CHAR_LF                     '\n'
1230    #define CHAR_NL                     CHAR_LF
1231    #define CHAR_NEL                    ((unsigned char)'\x85')
1232    #define CHAR_ESC                    '\033'
1233    #define CHAR_DEL                    '\177'
1234    #define CHAR_NBSP                   ((unsigned char)'\xa0')
1235    
1236    #define STR_LF                      "\n"
1237    #define STR_NL                      STR_LF
1238    #define STR_NEL                     "\x85"
1239    #define STR_ESC                     "\033"
1240    #define STR_DEL                     "\177"
1241    
1242    #endif  /* EBCDIC */
1243    
1244    /* The remaining definitions work in both environments. */
1245    
1246    #define CHAR_NULL                   '\0'
1247    #define CHAR_HT                     '\t'
1248    #define CHAR_VT                     '\v'
1249    #define CHAR_FF                     '\f'
1250    #define CHAR_CR                     '\r'
1251    #define CHAR_BS                     '\b'
1252    #define CHAR_BEL                    '\a'
1253    
1254    #define CHAR_SPACE                  ' '
1255    #define CHAR_EXCLAMATION_MARK       '!'
1256    #define CHAR_QUOTATION_MARK         '"'
1257    #define CHAR_NUMBER_SIGN            '#'
1258    #define CHAR_DOLLAR_SIGN            '$'
1259    #define CHAR_PERCENT_SIGN           '%'
1260    #define CHAR_AMPERSAND              '&'
1261    #define CHAR_APOSTROPHE             '\''
1262    #define CHAR_LEFT_PARENTHESIS       '('
1263    #define CHAR_RIGHT_PARENTHESIS      ')'
1264    #define CHAR_ASTERISK               '*'
1265    #define CHAR_PLUS                   '+'
1266    #define CHAR_COMMA                  ','
1267    #define CHAR_MINUS                  '-'
1268    #define CHAR_DOT                    '.'
1269    #define CHAR_SLASH                  '/'
1270    #define CHAR_0                      '0'
1271    #define CHAR_1                      '1'
1272    #define CHAR_2                      '2'
1273    #define CHAR_3                      '3'
1274    #define CHAR_4                      '4'
1275    #define CHAR_5                      '5'
1276    #define CHAR_6                      '6'
1277    #define CHAR_7                      '7'
1278    #define CHAR_8                      '8'
1279    #define CHAR_9                      '9'
1280    #define CHAR_COLON                  ':'
1281    #define CHAR_SEMICOLON              ';'
1282    #define CHAR_LESS_THAN_SIGN         '<'
1283    #define CHAR_EQUALS_SIGN            '='
1284    #define CHAR_GREATER_THAN_SIGN      '>'
1285    #define CHAR_QUESTION_MARK          '?'
1286    #define CHAR_COMMERCIAL_AT          '@'
1287    #define CHAR_A                      'A'
1288    #define CHAR_B                      'B'
1289    #define CHAR_C                      'C'
1290    #define CHAR_D                      'D'
1291    #define CHAR_E                      'E'
1292    #define CHAR_F                      'F'
1293    #define CHAR_G                      'G'
1294    #define CHAR_H                      'H'
1295    #define CHAR_I                      'I'
1296    #define CHAR_J                      'J'
1297    #define CHAR_K                      'K'
1298    #define CHAR_L                      'L'
1299    #define CHAR_M                      'M'
1300    #define CHAR_N                      'N'
1301    #define CHAR_O                      'O'
1302    #define CHAR_P                      'P'
1303    #define CHAR_Q                      'Q'
1304    #define CHAR_R                      'R'
1305    #define CHAR_S                      'S'
1306    #define CHAR_T                      'T'
1307    #define CHAR_U                      'U'
1308    #define CHAR_V                      'V'
1309    #define CHAR_W                      'W'
1310    #define CHAR_X                      'X'
1311    #define CHAR_Y                      'Y'
1312    #define CHAR_Z                      'Z'
1313    #define CHAR_LEFT_SQUARE_BRACKET    '['
1314    #define CHAR_BACKSLASH              '\\'
1315    #define CHAR_RIGHT_SQUARE_BRACKET   ']'
1316    #define CHAR_CIRCUMFLEX_ACCENT      '^'
1317    #define CHAR_UNDERSCORE             '_'
1318    #define CHAR_GRAVE_ACCENT           '`'
1319    #define CHAR_a                      'a'
1320    #define CHAR_b                      'b'
1321    #define CHAR_c                      'c'
1322    #define CHAR_d                      'd'
1323    #define CHAR_e                      'e'
1324    #define CHAR_f                      'f'
1325    #define CHAR_g                      'g'
1326    #define CHAR_h                      'h'
1327    #define CHAR_i                      'i'
1328    #define CHAR_j                      'j'
1329    #define CHAR_k                      'k'
1330    #define CHAR_l                      'l'
1331    #define CHAR_m                      'm'
1332    #define CHAR_n                      'n'
1333    #define CHAR_o                      'o'
1334    #define CHAR_p                      'p'
1335    #define CHAR_q                      'q'
1336    #define CHAR_r                      'r'
1337    #define CHAR_s                      's'
1338    #define CHAR_t                      't'
1339    #define CHAR_u                      'u'
1340    #define CHAR_v                      'v'
1341    #define CHAR_w                      'w'
1342    #define CHAR_x                      'x'
1343    #define CHAR_y                      'y'
1344    #define CHAR_z                      'z'
1345    #define CHAR_LEFT_CURLY_BRACKET     '{'
1346    #define CHAR_VERTICAL_LINE          '|'
1347    #define CHAR_RIGHT_CURLY_BRACKET    '}'
1348    #define CHAR_TILDE                  '~'
1349    
1350    #define STR_HT                      "\t"
1351    #define STR_VT                      "\v"
1352    #define STR_FF                      "\f"
1353    #define STR_CR                      "\r"
1354    #define STR_BS                      "\b"
1355    #define STR_BEL                     "\a"
1356    
1357    #define STR_SPACE                   " "
1358    #define STR_EXCLAMATION_MARK        "!"
1359    #define STR_QUOTATION_MARK          "\""
1360    #define STR_NUMBER_SIGN             "#"
1361    #define STR_DOLLAR_SIGN             "$"
1362    #define STR_PERCENT_SIGN            "%"
1363    #define STR_AMPERSAND               "&"
1364    #define STR_APOSTROPHE              "'"
1365    #define STR_LEFT_PARENTHESIS        "("
1366    #define STR_RIGHT_PARENTHESIS       ")"
1367    #define STR_ASTERISK                "*"
1368    #define STR_PLUS                    "+"
1369    #define STR_COMMA                   ","
1370    #define STR_MINUS                   "-"
1371    #define STR_DOT                     "."
1372    #define STR_SLASH                   "/"
1373    #define STR_0                       "0"
1374    #define STR_1                       "1"
1375    #define STR_2                       "2"
1376    #define STR_3                       "3"
1377    #define STR_4                       "4"
1378    #define STR_5                       "5"
1379    #define STR_6                       "6"
1380    #define STR_7                       "7"
1381    #define STR_8                       "8"
1382    #define STR_9                       "9"
1383    #define STR_COLON                   ":"
1384    #define STR_SEMICOLON               ";"
1385    #define STR_LESS_THAN_SIGN          "<"
1386    #define STR_EQUALS_SIGN             "="
1387    #define STR_GREATER_THAN_SIGN       ">"
1388    #define STR_QUESTION_MARK           "?"
1389    #define STR_COMMERCIAL_AT           "@"
1390    #define STR_A                       "A"
1391    #define STR_B                       "B"
1392    #define STR_C                       "C"
1393    #define STR_D                       "D"
1394    #define STR_E                       "E"
1395    #define STR_F                       "F"
1396    #define STR_G                       "G"
1397    #define STR_H                       "H"
1398    #define STR_I                       "I"
1399    #define STR_J                       "J"
1400    #define STR_K                       "K"
1401    #define STR_L                       "L"
1402    #define STR_M                       "M"
1403    #define STR_N                       "N"
1404    #define STR_O                       "O"
1405    #define STR_P                       "P"
1406    #define STR_Q                       "Q"
1407    #define STR_R                       "R"
1408    #define STR_S                       "S"
1409    #define STR_T                       "T"
1410    #define STR_U                       "U"
1411    #define STR_V                       "V"
1412    #define STR_W                       "W"
1413    #define STR_X                       "X"
1414    #define STR_Y                       "Y"
1415    #define STR_Z                       "Z"
1416    #define STR_LEFT_SQUARE_BRACKET     "["
1417    #define STR_BACKSLASH               "\\"
1418    #define STR_RIGHT_SQUARE_BRACKET    "]"
1419    #define STR_CIRCUMFLEX_ACCENT       "^"
1420    #define STR_UNDERSCORE              "_"
1421    #define STR_GRAVE_ACCENT            "`"
1422    #define STR_a                       "a"
1423    #define STR_b                       "b"
1424    #define STR_c                       "c"
1425    #define STR_d                       "d"
1426    #define STR_e                       "e"
1427    #define STR_f                       "f"
1428    #define STR_g                       "g"
1429    #define STR_h                       "h"
1430    #define STR_i                       "i"
1431    #define STR_j                       "j"
1432    #define STR_k                       "k"
1433    #define STR_l                       "l"
1434    #define STR_m                       "m"
1435    #define STR_n                       "n"
1436    #define STR_o                       "o"
1437    #define STR_p                       "p"
1438    #define STR_q                       "q"
1439    #define STR_r                       "r"
1440    #define STR_s                       "s"
1441    #define STR_t                       "t"
1442    #define STR_u                       "u"
1443    #define STR_v                       "v"
1444    #define STR_w                       "w"
1445    #define STR_x                       "x"
1446    #define STR_y                       "y"
1447    #define STR_z                       "z"
1448    #define STR_LEFT_CURLY_BRACKET      "{"
1449    #define STR_VERTICAL_LINE           "|"
1450    #define STR_RIGHT_CURLY_BRACKET     "}"
1451    #define STR_TILDE                   "~"
1452    
1453    #define STRING_ACCEPT0              "ACCEPT\0"
1454    #define STRING_COMMIT0              "COMMIT\0"
1455    #define STRING_F0                   "F\0"
1456    #define STRING_FAIL0                "FAIL\0"
1457    #define STRING_MARK0                "MARK\0"
1458    #define STRING_PRUNE0               "PRUNE\0"
1459    #define STRING_SKIP0                "SKIP\0"
1460    #define STRING_THEN                 "THEN"
1461    
1462    #define STRING_alpha0               "alpha\0"
1463    #define STRING_lower0               "lower\0"
1464    #define STRING_upper0               "upper\0"
1465    #define STRING_alnum0               "alnum\0"
1466    #define STRING_ascii0               "ascii\0"
1467    #define STRING_blank0               "blank\0"
1468    #define STRING_cntrl0               "cntrl\0"
1469    #define STRING_digit0               "digit\0"
1470    #define STRING_graph0               "graph\0"
1471    #define STRING_print0               "print\0"
1472    #define STRING_punct0               "punct\0"
1473    #define STRING_space0               "space\0"
1474    #define STRING_word0                "word\0"
1475    #define STRING_xdigit               "xdigit"
1476    
1477    #define STRING_DEFINE               "DEFINE"
1478    #define STRING_WEIRD_STARTWORD      "[:<:]]"
1479    #define STRING_WEIRD_ENDWORD        "[:>:]]"
1480    
1481    #define STRING_CR_RIGHTPAR              "CR)"
1482    #define STRING_LF_RIGHTPAR              "LF)"
1483    #define STRING_CRLF_RIGHTPAR            "CRLF)"
1484    #define STRING_ANY_RIGHTPAR             "ANY)"
1485    #define STRING_ANYCRLF_RIGHTPAR         "ANYCRLF)"
1486    #define STRING_BSR_ANYCRLF_RIGHTPAR     "BSR_ANYCRLF)"
1487    #define STRING_BSR_UNICODE_RIGHTPAR     "BSR_UNICODE)"
1488    #define STRING_UTF8_RIGHTPAR            "UTF8)"
1489    #define STRING_UTF16_RIGHTPAR           "UTF16)"
1490    #define STRING_UTF32_RIGHTPAR           "UTF32)"
1491    #define STRING_UTF_RIGHTPAR             "UTF)"
1492    #define STRING_UCP_RIGHTPAR             "UCP)"
1493    #define STRING_NO_AUTO_POSSESS_RIGHTPAR "NO_AUTO_POSSESS)"
1494    #define STRING_NO_START_OPT_RIGHTPAR    "NO_START_OPT)"
1495    #define STRING_LIMIT_MATCH_EQ           "LIMIT_MATCH="
1496    #define STRING_LIMIT_RECURSION_EQ       "LIMIT_RECURSION="
1497    
1498    #else  /* SUPPORT_UTF */
1499    
1500    /* UTF-8 support is enabled; always use UTF-8 (=ASCII) character codes. This
1501    works in both modes non-EBCDIC platforms, and on EBCDIC platforms in UTF-8 mode
1502    only. */
1503    
1504    #define CHAR_HT                     '\011'
1505    #define CHAR_VT                     '\013'
1506    #define CHAR_FF                     '\014'
1507    #define CHAR_CR                     '\015'
1508    #define CHAR_LF                     '\012'
1509    #define CHAR_NL                     CHAR_LF
1510    #define CHAR_NEL                    ((unsigned char)'\x85')
1511    #define CHAR_BS                     '\010'
1512    #define CHAR_BEL                    '\007'
1513    #define CHAR_ESC                    '\033'
1514    #define CHAR_DEL                    '\177'
1515    
1516    #define CHAR_NULL                   '\0'
1517    #define CHAR_SPACE                  '\040'
1518    #define CHAR_EXCLAMATION_MARK       '\041'
1519    #define CHAR_QUOTATION_MARK         '\042'
1520    #define CHAR_NUMBER_SIGN            '\043'
1521    #define CHAR_DOLLAR_SIGN            '\044'
1522    #define CHAR_PERCENT_SIGN           '\045'
1523    #define CHAR_AMPERSAND              '\046'
1524    #define CHAR_APOSTROPHE             '\047'
1525    #define CHAR_LEFT_PARENTHESIS       '\050'
1526    #define CHAR_RIGHT_PARENTHESIS      '\051'
1527    #define CHAR_ASTERISK               '\052'
1528    #define CHAR_PLUS                   '\053'
1529    #define CHAR_COMMA                  '\054'
1530    #define CHAR_MINUS                  '\055'
1531    #define CHAR_DOT                    '\056'
1532    #define CHAR_SLASH                  '\057'
1533    #define CHAR_0                      '\060'
1534    #define CHAR_1                      '\061'
1535    #define CHAR_2                      '\062'
1536    #define CHAR_3                      '\063'
1537    #define CHAR_4                      '\064'
1538    #define CHAR_5                      '\065'
1539    #define CHAR_6                      '\066'
1540    #define CHAR_7                      '\067'
1541    #define CHAR_8                      '\070'
1542    #define CHAR_9                      '\071'
1543    #define CHAR_COLON                  '\072'
1544    #define CHAR_SEMICOLON              '\073'
1545    #define CHAR_LESS_THAN_SIGN         '\074'
1546    #define CHAR_EQUALS_SIGN            '\075'
1547    #define CHAR_GREATER_THAN_SIGN      '\076'
1548    #define CHAR_QUESTION_MARK          '\077'
1549    #define CHAR_COMMERCIAL_AT          '\100'
1550    #define CHAR_A                      '\101'
1551    #define CHAR_B                      '\102'
1552    #define CHAR_C                      '\103'
1553    #define CHAR_D                      '\104'
1554    #define CHAR_E                      '\105'
1555    #define CHAR_F                      '\106'
1556    #define CHAR_G                      '\107'
1557    #define CHAR_H                      '\110'
1558    #define CHAR_I                      '\111'
1559    #define CHAR_J                      '\112'
1560    #define CHAR_K                      '\113'
1561    #define CHAR_L                      '\114'
1562    #define CHAR_M                      '\115'
1563    #define CHAR_N                      '\116'
1564    #define CHAR_O                      '\117'
1565    #define CHAR_P                      '\120'
1566    #define CHAR_Q                      '\121'
1567    #define CHAR_R                      '\122'
1568    #define CHAR_S                      '\123'
1569    #define CHAR_T                      '\124'
1570    #define CHAR_U                      '\125'
1571    #define CHAR_V                      '\126'
1572    #define CHAR_W                      '\127'
1573    #define CHAR_X                      '\130'
1574    #define CHAR_Y                      '\131'
1575    #define CHAR_Z                      '\132'
1576    #define CHAR_LEFT_SQUARE_BRACKET    '\133'
1577    #define CHAR_BACKSLASH              '\134'
1578    #define CHAR_RIGHT_SQUARE_BRACKET   '\135'
1579    #define CHAR_CIRCUMFLEX_ACCENT      '\136'
1580    #define CHAR_UNDERSCORE             '\137'
1581    #define CHAR_GRAVE_ACCENT           '\140'
1582    #define CHAR_a                      '\141'
1583    #define CHAR_b                      '\142'
1584    #define CHAR_c                      '\143'
1585    #define CHAR_d                      '\144'
1586    #define CHAR_e                      '\145'
1587    #define CHAR_f                      '\146'
1588    #define CHAR_g                      '\147'
1589    #define CHAR_h                      '\150'
1590    #define CHAR_i                      '\151'
1591    #define CHAR_j                      '\152'
1592    #define CHAR_k                      '\153'
1593    #define CHAR_l                      '\154'
1594    #define CHAR_m                      '\155'
1595    #define CHAR_n                      '\156'
1596    #define CHAR_o                      '\157'
1597    #define CHAR_p                      '\160'
1598    #define CHAR_q                      '\161'
1599    #define CHAR_r                      '\162'
1600    #define CHAR_s                      '\163'
1601    #define CHAR_t                      '\164'
1602    #define CHAR_u                      '\165'
1603    #define CHAR_v                      '\166'
1604    #define CHAR_w                      '\167'
1605    #define CHAR_x                      '\170'
1606    #define CHAR_y                      '\171'
1607    #define CHAR_z                      '\172'
1608    #define CHAR_LEFT_CURLY_BRACKET     '\173'
1609    #define CHAR_VERTICAL_LINE          '\174'
1610    #define CHAR_RIGHT_CURLY_BRACKET    '\175'
1611    #define CHAR_TILDE                  '\176'
1612    #define CHAR_NBSP                   ((unsigned char)'\xa0')
1613    
1614    #define STR_HT                      "\011"
1615    #define STR_VT                      "\013"
1616    #define STR_FF                      "\014"
1617    #define STR_CR                      "\015"
1618    #define STR_NL                      "\012"
1619    #define STR_BS                      "\010"
1620    #define STR_BEL                     "\007"
1621    #define STR_ESC                     "\033"
1622    #define STR_DEL                     "\177"
1623    
1624    #define STR_SPACE                   "\040"
1625    #define STR_EXCLAMATION_MARK        "\041"
1626    #define STR_QUOTATION_MARK          "\042"
1627    #define STR_NUMBER_SIGN             "\043"
1628    #define STR_DOLLAR_SIGN             "\044"
1629    #define STR_PERCENT_SIGN            "\045"
1630    #define STR_AMPERSAND               "\046"
1631    #define STR_APOSTROPHE              "\047"
1632    #define STR_LEFT_PARENTHESIS        "\050"
1633    #define STR_RIGHT_PARENTHESIS       "\051"
1634    #define STR_ASTERISK                "\052"
1635    #define STR_PLUS                    "\053"
1636    #define STR_COMMA                   "\054"
1637    #define STR_MINUS                   "\055"
1638    #define STR_DOT                     "\056"
1639    #define STR_SLASH                   "\057"
1640    #define STR_0                       "\060"
1641    #define STR_1                       "\061"
1642    #define STR_2                       "\062"
1643    #define STR_3                       "\063"
1644    #define STR_4                       "\064"
1645    #define STR_5                       "\065"
1646    #define STR_6                       "\066"
1647    #define STR_7                       "\067"
1648    #define STR_8                       "\070"
1649    #define STR_9                       "\071"
1650    #define STR_COLON                   "\072"
1651    #define STR_SEMICOLON               "\073"
1652    #define STR_LESS_THAN_SIGN          "\074"
1653    #define STR_EQUALS_SIGN             "\075"
1654    #define STR_GREATER_THAN_SIGN       "\076"
1655    #define STR_QUESTION_MARK           "\077"
1656    #define STR_COMMERCIAL_AT           "\100"
1657    #define STR_A                       "\101"
1658    #define STR_B                       "\102"
1659    #define STR_C                       "\103"
1660    #define STR_D                       "\104"
1661    #define STR_E                       "\105"
1662    #define STR_F                       "\106"
1663    #define STR_G                       "\107"
1664    #define STR_H                       "\110"
1665    #define STR_I                       "\111"
1666    #define STR_J                       "\112"
1667    #define STR_K                       "\113"
1668    #define STR_L                       "\114"
1669    #define STR_M                       "\115"
1670    #define STR_N                       "\116"
1671    #define STR_O                       "\117"
1672    #define STR_P                       "\120"
1673    #define STR_Q                       "\121"
1674    #define STR_R                       "\122"
1675    #define STR_S                       "\123"
1676    #define STR_T                       "\124"
1677    #define STR_U                       "\125"
1678    #define STR_V                       "\126"
1679    #define STR_W                       "\127"
1680    #define STR_X                       "\130"
1681    #define STR_Y                       "\131"
1682    #define STR_Z                       "\132"
1683    #define STR_LEFT_SQUARE_BRACKET     "\133"
1684    #define STR_BACKSLASH               "\134"
1685    #define STR_RIGHT_SQUARE_BRACKET    "\135"
1686    #define STR_CIRCUMFLEX_ACCENT       "\136"
1687    #define STR_UNDERSCORE              "\137"
1688    #define STR_GRAVE_ACCENT            "\140"
1689    #define STR_a                       "\141"
1690    #define STR_b                       "\142"
1691    #define STR_c                       "\143"
1692    #define STR_d                       "\144"
1693    #define STR_e                       "\145"
1694    #define STR_f                       "\146"
1695    #define STR_g                       "\147"
1696    #define STR_h                       "\150"
1697    #define STR_i                       "\151"
1698    #define STR_j                       "\152"
1699    #define STR_k                       "\153"
1700    #define STR_l                       "\154"
1701    #define STR_m                       "\155"
1702    #define STR_n                       "\156"
1703    #define STR_o                       "\157"
1704    #define STR_p                       "\160"
1705    #define STR_q                       "\161"
1706    #define STR_r                       "\162"
1707    #define STR_s                       "\163"
1708    #define STR_t                       "\164"
1709    #define STR_u                       "\165"
1710    #define STR_v                       "\166"
1711    #define STR_w                       "\167"
1712    #define STR_x                       "\170"
1713    #define STR_y                       "\171"
1714    #define STR_z                       "\172"
1715    #define STR_LEFT_CURLY_BRACKET      "\173"
1716    #define STR_VERTICAL_LINE           "\174"
1717    #define STR_RIGHT_CURLY_BRACKET     "\175"
1718    #define STR_TILDE                   "\176"
1719    
1720    #define STRING_ACCEPT0              STR_A STR_C STR_C STR_E STR_P STR_T "\0"
1721    #define STRING_COMMIT0              STR_C STR_O STR_M STR_M STR_I STR_T "\0"
1722    #define STRING_F0                   STR_F "\0"
1723    #define STRING_FAIL0                STR_F STR_A STR_I STR_L "\0"
1724    #define STRING_MARK0                STR_M STR_A STR_R STR_K "\0"
1725    #define STRING_PRUNE0               STR_P STR_R STR_U STR_N STR_E "\0"
1726    #define STRING_SKIP0                STR_S STR_K STR_I STR_P "\0"
1727    #define STRING_THEN                 STR_T STR_H STR_E STR_N
1728    
1729    #define STRING_alpha0               STR_a STR_l STR_p STR_h STR_a "\0"
1730    #define STRING_lower0               STR_l STR_o STR_w STR_e STR_r "\0"
1731    #define STRING_upper0               STR_u STR_p STR_p STR_e STR_r "\0"
1732    #define STRING_alnum0               STR_a STR_l STR_n STR_u STR_m "\0"
1733    #define STRING_ascii0               STR_a STR_s STR_c STR_i STR_i "\0"
1734    #define STRING_blank0               STR_b STR_l STR_a STR_n STR_k "\0"
1735    #define STRING_cntrl0               STR_c STR_n STR_t STR_r STR_l "\0"
1736    #define STRING_digit0               STR_d STR_i STR_g STR_i STR_t "\0"
1737    #define STRING_graph0               STR_g STR_r STR_a STR_p STR_h "\0"
1738    #define STRING_print0               STR_p STR_r STR_i STR_n STR_t "\0"
1739    #define STRING_punct0               STR_p STR_u STR_n STR_c STR_t "\0"
1740    #define STRING_space0               STR_s STR_p STR_a STR_c STR_e "\0"
1741    #define STRING_word0                STR_w STR_o STR_r STR_d       "\0"
1742    #define STRING_xdigit               STR_x STR_d STR_i STR_g STR_i STR_t
1743    
1744    #define STRING_DEFINE               STR_D STR_E STR_F STR_I STR_N STR_E
1745    #define STRING_WEIRD_STARTWORD      STR_LEFT_SQUARE_BRACKET STR_COLON STR_LESS_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
1746    #define STRING_WEIRD_ENDWORD        STR_LEFT_SQUARE_BRACKET STR_COLON STR_GREATER_THAN_SIGN STR_COLON STR_RIGHT_SQUARE_BRACKET STR_RIGHT_SQUARE_BRACKET
1747    
1748    #define STRING_CR_RIGHTPAR              STR_C STR_R STR_RIGHT_PARENTHESIS
1749    #define STRING_LF_RIGHTPAR              STR_L STR_F STR_RIGHT_PARENTHESIS
1750    #define STRING_CRLF_RIGHTPAR            STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
1751    #define STRING_ANY_RIGHTPAR             STR_A STR_N STR_Y STR_RIGHT_PARENTHESIS
1752    #define STRING_ANYCRLF_RIGHTPAR         STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
1753    #define STRING_BSR_ANYCRLF_RIGHTPAR     STR_B STR_S STR_R STR_UNDERSCORE STR_A STR_N STR_Y STR_C STR_R STR_L STR_F STR_RIGHT_PARENTHESIS
1754    #define STRING_BSR_UNICODE_RIGHTPAR     STR_B STR_S STR_R STR_UNDERSCORE STR_U STR_N STR_I STR_C STR_O STR_D STR_E STR_RIGHT_PARENTHESIS
1755    #define STRING_UTF8_RIGHTPAR            STR_U STR_T STR_F STR_8 STR_RIGHT_PARENTHESIS
1756    #define STRING_UTF16_RIGHTPAR           STR_U STR_T STR_F STR_1 STR_6 STR_RIGHT_PARENTHESIS
1757    #define STRING_UTF32_RIGHTPAR           STR_U STR_T STR_F STR_3 STR_2 STR_RIGHT_PARENTHESIS
1758    #define STRING_UTF_RIGHTPAR             STR_U STR_T STR_F STR_RIGHT_PARENTHESIS
1759    #define STRING_UCP_RIGHTPAR             STR_U STR_C STR_P STR_RIGHT_PARENTHESIS
1760    #define STRING_NO_AUTO_POSSESS_RIGHTPAR STR_N STR_O STR_UNDERSCORE STR_A STR_U STR_T STR_O STR_UNDERSCORE STR_P STR_O STR_S STR_S STR_E STR_S STR_S STR_RIGHT_PARENTHESIS
1761    #define STRING_NO_START_OPT_RIGHTPAR    STR_N STR_O STR_UNDERSCORE STR_S STR_T STR_A STR_R STR_T STR_UNDERSCORE STR_O STR_P STR_T STR_RIGHT_PARENTHESIS
1762    #define STRING_LIMIT_MATCH_EQ           STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_M STR_A STR_T STR_C STR_H STR_EQUALS_SIGN
1763    #define STRING_LIMIT_RECURSION_EQ       STR_L STR_I STR_M STR_I STR_T STR_UNDERSCORE STR_R STR_E STR_C STR_U STR_R STR_S STR_I STR_O STR_N STR_EQUALS_SIGN
1764    
1765    #endif  /* SUPPORT_UTF */
1766    
1767  /* Escape items that are just an encoding of a particular data value. */  /* Escape items that are just an encoding of a particular data value. */
1768    
1769    #ifndef ESC_a
1770    #define ESC_a CHAR_BEL
1771    #endif
1772    
1773  #ifndef ESC_e  #ifndef ESC_e
1774  #define ESC_e 27  #define ESC_e CHAR_ESC
1775  #endif  #endif
1776    
1777  #ifndef ESC_f  #ifndef ESC_f
1778  #define ESC_f '\f'  #define ESC_f CHAR_FF
1779  #endif  #endif
1780    
1781  #ifndef ESC_n  #ifndef ESC_n
1782  #define ESC_n '\n'  #define ESC_n CHAR_LF
1783  #endif  #endif
1784    
1785  #ifndef ESC_r  #ifndef ESC_r
1786  #define ESC_r '\r'  #define ESC_r CHAR_CR
1787  #endif  #endif
1788    
1789  /* We can't officially use ESC_t because it is a POSIX reserved identifier  /* We can't officially use ESC_t because it is a POSIX reserved identifier
1790  (presumably because of all the others like size_t). */  (presumably because of all the others like size_t). */
1791    
1792  #ifndef ESC_tee  #ifndef ESC_tee
1793  #define ESC_tee '\t'  #define ESC_tee CHAR_HT
1794  #endif  #endif
1795    
1796  /* Codes for different types of Unicode property */  /* Codes for different types of Unicode property */
1797    
1798  #define PT_ANY        0    /* Any property - matches all chars */  #define PT_ANY        0    /* Any property - matches all chars */
1799  #define PT_LAMP       1    /* L& - the union of Lu, Ll, Lt */  #define PT_LAMP       1    /* L& - the union of Lu, Ll, Lt */
1800  #define PT_GC         2    /* General characteristic (e.g. L) */  #define PT_GC         2    /* Specified general characteristic (e.g. L) */
1801  #define PT_PC         3    /* Particular characteristic (e.g. Lu) */  #define PT_PC         3    /* Specified particular characteristic (e.g. Lu) */
1802  #define PT_SC         4    /* Script (e.g. Han) */  #define PT_SC         4    /* Script (e.g. Han) */
1803    #define PT_ALNUM      5    /* Alphanumeric - the union of L and N */
1804    #define PT_SPACE      6    /* Perl space - Z plus 9,10,12,13 */
1805    #define PT_PXSPACE    7    /* POSIX space - Z plus 9,10,11,12,13 */
1806    #define PT_WORD       8    /* Word - L plus N plus underscore */
1807    #define PT_CLIST      9    /* Pseudo-property: match character list */
1808    #define PT_UCNC      10    /* Universal Character nameable character */
1809    #define PT_TABSIZE   11    /* Size of square table for autopossessify tests */
1810    
1811    /* The following special properties are used only in XCLASS items, when POSIX
1812    classes are specified and PCRE_UCP is set - in other words, for Unicode
1813    handling of these classes. They are not available via the \p or \P escapes like
1814    those in the above list, and so they do not take part in the autopossessifying
1815    table. */
1816    
1817    #define PT_PXGRAPH   11    /* [:graph:] - characters that mark the paper */
1818    #define PT_PXPRINT   12    /* [:print:] - [:graph:] plus non-control spaces */
1819    #define PT_PXPUNCT   13    /* [:punct:] - punctuation characters */
1820    
1821  /* Flag bits and data types for the extended class (OP_XCLASS) for classes that  /* Flag bits and data types for the extended class (OP_XCLASS) for classes that
1822  contain UTF-8 characters with values greater than 255. */  contain characters with values greater than 255. */
1823    
1824  #define XCL_NOT    0x01    /* Flag: this is a negative class */  #define XCL_NOT       0x01    /* Flag: this is a negative class */
1825  #define XCL_MAP    0x02    /* Flag: a 32-byte map is present */  #define XCL_MAP       0x02    /* Flag: a 32-byte map is present */
1826    #define XCL_HASPROP   0x04    /* Flag: property checks are present. */
1827    
1828  #define XCL_END       0    /* Marks end of individual items */  #define XCL_END       0    /* Marks end of individual items */
1829  #define XCL_SINGLE    1    /* Single item (one multibyte char) follows */  #define XCL_SINGLE    1    /* Single item (one multibyte char) follows */
# Line 606  contain UTF-8 characters with values gre Line 1832  contain UTF-8 characters with values gre
1832  #define XCL_NOTPROP   4    /* Unicode inverted property (ditto) */  #define XCL_NOTPROP   4    /* Unicode inverted property (ditto) */
1833    
1834  /* These are escaped items that aren't just an encoding of a particular data  /* These are escaped items that aren't just an encoding of a particular data
1835  value such as \n. They must have non-zero values, as check_escape() returns  value such as \n. They must have non-zero values, as check_escape() returns 0
1836  their negation. Also, they must appear in the same order as in the opcode  for a data character.  Also, they must appear in the same order as in the
1837  definitions below, up to ESC_z. There's a dummy for OP_ANY because it  opcode definitions below, up to ESC_z. There's a dummy for OP_ALLANY because it
1838  corresponds to "." rather than an escape sequence, and another for OP_ALLANY  corresponds to "." in DOTALL mode rather than an escape sequence. It is also
1839  (which is used for [^] in JavaScript compatibility mode).  used for [^] in JavaScript compatibility mode, and for \C in non-utf mode. In
1840    non-DOTALL mode, "." behaves like \N.
1841    
1842    The special values ESC_DU, ESC_du, etc. are used instead of ESC_D, ESC_d, etc.
1843    when PCRE_UCP is set and replacement of \d etc by \p sequences is required.
1844    They must be contiguous, and remain in order so that the replacements can be
1845    looked up from a table.
1846    
1847  The final escape must be ESC_REF as subsequent values are used for  Negative numbers are used to encode a backreference (\1, \2, \3, etc.) in
1848  backreferences (\1, \2, \3, etc). There are two tests in the code for an escape  check_escape(). There are two tests in the code for an escape
1849  greater than ESC_b and less than ESC_Z to detect the types that may be  greater than ESC_b and less than ESC_Z to detect the types that may be
1850  repeated. These are the types that consume characters. If any new escapes are  repeated. These are the types that consume characters. If any new escapes are
1851  put in between that don't consume a character, that code will have to change.  put in between that don't consume a character, that code will have to change.
1852  */  */
1853    
1854  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s,  enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s,
1855         ESC_W, ESC_w, ESC_dum1, ESC_dum2, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H,         ESC_W, ESC_w, ESC_N, ESC_dum, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H,
1856         ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z, ESC_E, ESC_Q, ESC_g, ESC_k,         ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z,
1857         ESC_REF };         ESC_E, ESC_Q, ESC_g, ESC_k,
1858           ESC_DU, ESC_du, ESC_SU, ESC_su, ESC_WU, ESC_wu };
1859    
1860    
1861    /********************** Opcode definitions ******************/
1862    
1863    /****** NOTE NOTE NOTE ******
1864    
1865    Starting from 1 (i.e. after OP_END), the values up to OP_EOD must correspond in
1866    order to the list of escapes immediately above. Furthermore, values up to
1867    OP_DOLLM must not be changed without adjusting the table called autoposstab in
1868    pcre_compile.c
1869    
1870    Whenever this list is updated, the two macro definitions that follow must be
1871    updated to match. The possessification table called "opcode_possessify" in
1872    pcre_compile.c must also be updated, and also the tables called "coptable"
1873    and "poptable" in pcre_dfa_exec.c.
1874    
1875  /* Opcode table: Starting from 1 (i.e. after OP_END), the values up to  ****** NOTE NOTE NOTE ******/
 OP_EOD must correspond in order to the list of escapes immediately above.  
1876    
1877  *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions  
1878  that follow must also be updated to match. There is also a table called  /* The values between FIRST_AUTOTAB_OP and LAST_AUTOTAB_RIGHT_OP, inclusive,
1879  "coptable" in pcre_dfa_exec.c that must be updated. */  are used in a table for deciding whether a repeated character type can be
1880    auto-possessified. */
1881    
1882    #define FIRST_AUTOTAB_OP       OP_NOT_DIGIT
1883    #define LAST_AUTOTAB_LEFT_OP   OP_EXTUNI
1884    #define LAST_AUTOTAB_RIGHT_OP  OP_DOLLM
1885    
1886  enum {  enum {
1887    OP_END,            /* 0 End of pattern */    OP_END,            /* 0 End of pattern */
# Line 648  enum { Line 1899  enum {
1899    OP_WHITESPACE,         /*  9 \s */    OP_WHITESPACE,         /*  9 \s */
1900    OP_NOT_WORDCHAR,       /* 10 \W */    OP_NOT_WORDCHAR,       /* 10 \W */
1901    OP_WORDCHAR,           /* 11 \w */    OP_WORDCHAR,           /* 11 \w */
1902    OP_ANY,            /* 12 Match any character (subject to DOTALL) */  
1903    OP_ALLANY,         /* 13 Match any character (not subject to DOTALL) */    OP_ANY,            /* 12 Match any character except newline (\N) */
1904      OP_ALLANY,         /* 13 Match any character */
1905    OP_ANYBYTE,        /* 14 Match any byte (\C); different to OP_ANY for UTF-8 */    OP_ANYBYTE,        /* 14 Match any byte (\C); different to OP_ANY for UTF-8 */
1906    OP_NOTPROP,        /* 15 \P (not Unicode property) */    OP_NOTPROP,        /* 15 \P (not Unicode property) */
1907    OP_PROP,           /* 16 \p (Unicode property) */    OP_PROP,           /* 16 \p (Unicode property) */
# Line 659  enum { Line 1911  enum {
1911    OP_NOT_VSPACE,     /* 20 \V (not vertical whitespace) */    OP_NOT_VSPACE,     /* 20 \V (not vertical whitespace) */
1912    OP_VSPACE,         /* 21 \v (vertical whitespace) */    OP_VSPACE,         /* 21 \v (vertical whitespace) */
1913    OP_EXTUNI,         /* 22 \X (extended Unicode sequence */    OP_EXTUNI,         /* 22 \X (extended Unicode sequence */
1914    OP_EODN,           /* 23 End of data or \n at end of data: \Z. */    OP_EODN,           /* 23 End of data or \n at end of data (\Z) */
1915    OP_EOD,            /* 24 End of data: \z */    OP_EOD,            /* 24 End of data (\z) */
1916    
1917    OP_OPT,            /* 25 Set runtime options */    /* Line end assertions */
   OP_CIRC,           /* 26 Start of line - varies with multiline switch */  
   OP_DOLL,           /* 27 End of line - varies with multiline switch */  
   OP_CHAR,           /* 28 Match one character, casefully */  
   OP_CHARNC,         /* 29 Match one character, caselessly */  
   OP_NOT,            /* 30 Match one character, not the following one */  
   
   OP_STAR,           /* 31 The maximizing and minimizing versions of */  
   OP_MINSTAR,        /* 32 these six opcodes must come in pairs, with */  
   OP_PLUS,           /* 33 the minimizing one second. */  
   OP_MINPLUS,        /* 34 This first set applies to single characters.*/  
   OP_QUERY,          /* 35 */  
   OP_MINQUERY,       /* 36 */  
   
   OP_UPTO,           /* 37 From 0 to n matches */  
   OP_MINUPTO,        /* 38 */  
   OP_EXACT,          /* 39 Exactly n matches */  
   
   OP_POSSTAR,        /* 40 Possessified star */  
   OP_POSPLUS,        /* 41 Possessified plus */  
   OP_POSQUERY,       /* 42 Posesssified query */  
   OP_POSUPTO,        /* 43 Possessified upto */  
   
   OP_NOTSTAR,        /* 44 The maximizing and minimizing versions of */  
   OP_NOTMINSTAR,     /* 45 these six opcodes must come in pairs, with */  
   OP_NOTPLUS,        /* 46 the minimizing one second. They must be in */  
   OP_NOTMINPLUS,     /* 47 exactly the same order as those above. */  
   OP_NOTQUERY,       /* 48 This set applies to "not" single characters. */  
   OP_NOTMINQUERY,    /* 49 */  
   
   OP_NOTUPTO,        /* 50 From 0 to n matches */  
   OP_NOTMINUPTO,     /* 51 */  
   OP_NOTEXACT,       /* 52 Exactly n matches */  
   
   OP_NOTPOSSTAR,     /* 53 Possessified versions */  
   OP_NOTPOSPLUS,     /* 54 */  
   OP_NOTPOSQUERY,    /* 55 */  
   OP_NOTPOSUPTO,     /* 56 */  
   
   OP_TYPESTAR,       /* 57 The maximizing and minimizing versions of */  
   OP_TYPEMINSTAR,    /* 58 these six opcodes must come in pairs, with */  
   OP_TYPEPLUS,       /* 59 the minimizing one second. These codes must */  
   OP_TYPEMINPLUS,    /* 60 be in exactly the same order as those above. */  
   OP_TYPEQUERY,      /* 61 This set applies to character types such as \d */  
   OP_TYPEMINQUERY,   /* 62 */  
   
   OP_TYPEUPTO,       /* 63 From 0 to n matches */  
   OP_TYPEMINUPTO,    /* 64 */  
   OP_TYPEEXACT,      /* 65 Exactly n matches */  
   
   OP_TYPEPOSSTAR,    /* 66 Possessified versions */  
   OP_TYPEPOSPLUS,    /* 67 */  
   OP_TYPEPOSQUERY,   /* 68 */  
   OP_TYPEPOSUPTO,    /* 69 */  
   
   OP_CRSTAR,         /* 70 The maximizing and minimizing versions of */  
   OP_CRMINSTAR,      /* 71 all these opcodes must come in pairs, with */  
   OP_CRPLUS,         /* 72 the minimizing one second. These codes must */  
   OP_CRMINPLUS,      /* 73 be in exactly the same order as those above. */  
   OP_CRQUERY,        /* 74 These are for character classes and back refs */  
   OP_CRMINQUERY,     /* 75 */  
   OP_CRRANGE,        /* 76 These are different to the three sets above. */  
   OP_CRMINRANGE,     /* 77 */  
   
   OP_CLASS,          /* 78 Match a character class, chars < 256 only */  
   OP_NCLASS,         /* 79 Same, but the bitmap was created from a negative  
                            class - the difference is relevant only when a UTF-8  
                            character > 255 is encountered. */  
   
   OP_XCLASS,         /* 80 Extended class for handling UTF-8 chars within the  
                            class. This does both positive and negative. */  
   
   OP_REF,            /* 81 Match a back reference */  
   OP_RECURSE,        /* 82 Match a numbered subpattern (possibly recursive) */  
   OP_CALLOUT,        /* 83 Call out to external function if provided */  
   
   OP_ALT,            /* 84 Start of alternation */  
   OP_KET,            /* 85 End of group that doesn't have an unbounded repeat */  
   OP_KETRMAX,        /* 86 These two must remain together and in this */  
   OP_KETRMIN,        /* 87 order. They are for groups the repeat for ever. */  
   
   /* The assertions must come before BRA, CBRA, ONCE, and COND.*/  
   
   OP_ASSERT,         /* 88 Positive lookahead */  
   OP_ASSERT_NOT,     /* 89 Negative lookahead */  
   OP_ASSERTBACK,     /* 90 Positive lookbehind */  
   OP_ASSERTBACK_NOT, /* 91 Negative lookbehind */  
   OP_REVERSE,        /* 92 Move pointer back - used in lookbehind assertions */  
   
   /* ONCE, BRA, CBRA, and COND must come after the assertions, with ONCE first,  
   as there's a test for >= ONCE for a subpattern that isn't an assertion. */  
   
   OP_ONCE,           /* 93 Atomic group */  
   OP_BRA,            /* 94 Start of non-capturing bracket */  
   OP_CBRA,           /* 95 Start of capturing bracket */  
   OP_COND,           /* 96 Conditional group */  
1918    
1919    /* These three must follow the previous three, in the same order. There's a    OP_DOLL,           /* 25 End of line - not multiline */
1920    check for >= SBRA to distinguish the two sets. */    OP_DOLLM,          /* 26 End of line - multiline */
1921      OP_CIRC,           /* 27 Start of line - not multiline */
1922      OP_CIRCM,          /* 28 Start of line - multiline */
1923    
1924      /* Single characters; caseful must precede the caseless ones */
1925    
1926      OP_CHAR,           /* 29 Match one character, casefully */
1927      OP_CHARI,          /* 30 Match one character, caselessly */
1928      OP_NOT,            /* 31 Match one character, not the given one, casefully */
1929      OP_NOTI,           /* 32 Match one character, not the given one, caselessly */
1930    
1931      /* The following sets of 13 opcodes must always be kept in step because
1932      the offset from the first one is used to generate the others. */
1933    
1934      /* Repeated characters; caseful must precede the caseless ones */
1935    
1936      OP_STAR,           /* 33 The maximizing and minimizing versions of */
1937      OP_MINSTAR,        /* 34 these six opcodes must come in pairs, with */
1938      OP_PLUS,           /* 35 the minimizing one second. */
1939      OP_MINPLUS,        /* 36 */
1940      OP_QUERY,          /* 37 */
1941      OP_MINQUERY,       /* 38 */
1942    
1943      OP_UPTO,           /* 39 From 0 to n matches of one character, caseful*/
1944      OP_MINUPTO,        /* 40 */
1945      OP_EXACT,          /* 41 Exactly n matches */
1946    
1947      OP_POSSTAR,        /* 42 Possessified star, caseful */
1948      OP_POSPLUS,        /* 43 Possessified plus, caseful */
1949      OP_POSQUERY,       /* 44 Posesssified query, caseful */
1950      OP_POSUPTO,        /* 45 Possessified upto, caseful */
1951    
1952      /* Repeated characters; caseless must follow the caseful ones */
1953    
1954      OP_STARI,          /* 46 */
1955      OP_MINSTARI,       /* 47 */
1956      OP_PLUSI,          /* 48 */
1957      OP_MINPLUSI,       /* 49 */
1958      OP_QUERYI,         /* 50 */
1959      OP_MINQUERYI,      /* 51 */
1960    
1961      OP_UPTOI,          /* 52 From 0 to n matches of one character, caseless */
1962      OP_MINUPTOI,       /* 53 */
1963      OP_EXACTI,         /* 54 */
1964    
1965      OP_POSSTARI,       /* 55 Possessified star, caseless */
1966      OP_POSPLUSI,       /* 56 Possessified plus, caseless */
1967      OP_POSQUERYI,      /* 57 Posesssified query, caseless */
1968      OP_POSUPTOI,       /* 58 Possessified upto, caseless */
1969    
1970      /* The negated ones must follow the non-negated ones, and match them */
1971      /* Negated repeated character, caseful; must precede the caseless ones */
1972    
1973      OP_NOTSTAR,        /* 59 The maximizing and minimizing versions of */
1974      OP_NOTMINSTAR,     /* 60 these six opcodes must come in pairs, with */
1975      OP_NOTPLUS,        /* 61 the minimizing one second. They must be in */
1976      OP_NOTMINPLUS,     /* 62 exactly the same order as those above. */
1977      OP_NOTQUERY,       /* 63 */
1978      OP_NOTMINQUERY,    /* 64 */
1979    
1980      OP_NOTUPTO,        /* 65 From 0 to n matches, caseful */
1981      OP_NOTMINUPTO,     /* 66 */
1982      OP_NOTEXACT,       /* 67 Exactly n matches */
1983    
1984      OP_NOTPOSSTAR,     /* 68 Possessified versions, caseful */
1985      OP_NOTPOSPLUS,     /* 69 */
1986      OP_NOTPOSQUERY,    /* 70 */
1987      OP_NOTPOSUPTO,     /* 71 */
1988    
1989      /* Negated repeated character, caseless; must follow the caseful ones */
1990    
1991      OP_NOTSTARI,       /* 72 */
1992      OP_NOTMINSTARI,    /* 73 */
1993      OP_NOTPLUSI,       /* 74 */
1994      OP_NOTMINPLUSI,    /* 75 */
1995      OP_NOTQUERYI,      /* 76 */
1996      OP_NOTMINQUERYI,   /* 77 */
1997    
1998      OP_NOTUPTOI,       /* 78 From 0 to n matches, caseless */
1999      OP_NOTMINUPTOI,    /* 79 */
2000      OP_NOTEXACTI,      /* 80 Exactly n matches */
2001    
2002      OP_NOTPOSSTARI,    /* 81 Possessified versions, caseless */
2003      OP_NOTPOSPLUSI,    /* 82 */
2004      OP_NOTPOSQUERYI,   /* 83 */
2005      OP_NOTPOSUPTOI,    /* 84 */
2006    
2007      /* Character types */
2008    
2009      OP_TYPESTAR,       /* 85 The maximizing and minimizing versions of */
2010      OP_TYPEMINSTAR,    /* 86 these six opcodes must come in pairs, with */
2011      OP_TYPEPLUS,       /* 87 the minimizing one second. These codes must */
2012      OP_TYPEMINPLUS,    /* 88 be in exactly the same order as those above. */
2013      OP_TYPEQUERY,      /* 89 */
2014      OP_TYPEMINQUERY,   /* 90 */
2015    
2016      OP_TYPEUPTO,       /* 91 From 0 to n matches */
2017      OP_TYPEMINUPTO,    /* 92 */
2018      OP_TYPEEXACT,      /* 93 Exactly n matches */
2019    
2020      OP_TYPEPOSSTAR,    /* 94 Possessified versions */
2021      OP_TYPEPOSPLUS,    /* 95 */
2022      OP_TYPEPOSQUERY,   /* 96 */
2023      OP_TYPEPOSUPTO,    /* 97 */
2024    
2025      /* These are used for character classes and back references; only the
2026      first six are the same as the sets above. */
2027    
2028      OP_CRSTAR,         /* 98 The maximizing and minimizing versions of */
2029      OP_CRMINSTAR,      /* 99 all these opcodes must come in pairs, with */
2030      OP_CRPLUS,         /* 100 the minimizing one second. These codes must */
2031      OP_CRMINPLUS,      /* 101 be in exactly the same order as those above. */
2032      OP_CRQUERY,        /* 102 */
2033      OP_CRMINQUERY,     /* 103 */
2034    
2035      OP_CRRANGE,        /* 104 These are different to the three sets above. */
2036      OP_CRMINRANGE,     /* 105 */
2037    
2038      OP_CRPOSSTAR,      /* 106 Possessified versions */
2039      OP_CRPOSPLUS,      /* 107 */
2040      OP_CRPOSQUERY,     /* 108 */
2041      OP_CRPOSRANGE,     /* 109 */
2042    
2043      /* End of quantifier opcodes */
2044    
2045      OP_CLASS,          /* 110 Match a character class, chars < 256 only */
2046      OP_NCLASS,         /* 111 Same, but the bitmap was created from a negative
2047                                  class - the difference is relevant only when a
2048                                  character > 255 is encountered. */
2049      OP_XCLASS,         /* 112 Extended class for handling > 255 chars within the
2050                                  class. This does both positive and negative. */
2051      OP_REF,            /* 113 Match a back reference, casefully */
2052      OP_REFI,           /* 114 Match a back reference, caselessly */
2053      OP_DNREF,          /* 115 Match a duplicate name backref, casefully */
2054      OP_DNREFI,         /* 116 Match a duplicate name backref, caselessly */
2055      OP_RECURSE,        /* 117 Match a numbered subpattern (possibly recursive) */
2056      OP_CALLOUT,        /* 118 Call out to external function if provided */
2057    
2058      OP_ALT,            /* 119 Start of alternation */
2059      OP_KET,            /* 120 End of group that doesn't have an unbounded repeat */
2060      OP_KETRMAX,        /* 121 These two must remain together and in this */
2061      OP_KETRMIN,        /* 122 order. They are for groups the repeat for ever. */
2062      OP_KETRPOS,        /* 123 Possessive unlimited repeat. */
2063    
2064      /* The assertions must come before BRA, CBRA, ONCE, and COND, and the four
2065      asserts must remain in order. */
2066    
2067      OP_REVERSE,        /* 124 Move pointer back - used in lookbehind assertions */
2068      OP_ASSERT,         /* 125 Positive lookahead */
2069      OP_ASSERT_NOT,     /* 126 Negative lookahead */
2070      OP_ASSERTBACK,     /* 127 Positive lookbehind */
2071      OP_ASSERTBACK_NOT, /* 128 Negative lookbehind */
2072    
2073      /* ONCE, ONCE_NC, BRA, BRAPOS, CBRA, CBRAPOS, and COND must come immediately
2074      after the assertions, with ONCE first, as there's a test for >= ONCE for a
2075      subpattern that isn't an assertion. The POS versions must immediately follow
2076      the non-POS versions in each case. */
2077    
2078      OP_ONCE,           /* 129 Atomic group, contains captures */
2079      OP_ONCE_NC,        /* 130 Atomic group containing no captures */
2080      OP_BRA,            /* 131 Start of non-capturing bracket */
2081      OP_BRAPOS,         /* 132 Ditto, with unlimited, possessive repeat */
2082      OP_CBRA,           /* 133 Start of capturing bracket */
2083      OP_CBRAPOS,        /* 134 Ditto, with unlimited, possessive repeat */
2084      OP_COND,           /* 135 Conditional group */
2085    
2086    OP_SBRA,           /* 97 Start of non-capturing bracket, check empty  */    /* These five must follow the previous five, in the same order. There's a
2087    OP_SCBRA,          /* 98 Start of capturing bracket, check empty */    check for >= SBRA to distinguish the two sets. */
   OP_SCOND,          /* 99 Conditional group, check empty */  
   
   OP_CREF,           /* 100 Used to hold a capture number as condition */  
   OP_RREF,           /* 101 Used to hold a recursion number as condition */  
   OP_DEF,            /* 102 The DEFINE condition */  
2088    
2089    OP_BRAZERO,        /* 103 These two must remain together and in this */    OP_SBRA,           /* 136 Start of non-capturing bracket, check empty  */
2090    OP_BRAMINZERO,     /* 104 order. */    OP_SBRAPOS,        /* 137 Ditto, with unlimited, possessive repeat */
2091      OP_SCBRA,          /* 138 Start of capturing bracket, check empty */
2092      OP_SCBRAPOS,       /* 139 Ditto, with unlimited, possessive repeat */
2093      OP_SCOND,          /* 140 Conditional group, check empty */
2094    
2095      /* The next two pairs must (respectively) be kept together. */
2096    
2097      OP_CREF,           /* 141 Used to hold a capture number as condition */
2098      OP_DNCREF,         /* 142 Used to point to duplicate names as a condition */
2099      OP_RREF,           /* 143 Used to hold a recursion number as condition */
2100      OP_DNRREF,         /* 144 Used to point to duplicate names as a condition */
2101      OP_DEF,            /* 145 The DEFINE condition */
2102    
2103      OP_BRAZERO,        /* 146 These two must remain together and in this */
2104      OP_BRAMINZERO,     /* 147 order. */
2105      OP_BRAPOSZERO,     /* 148 */
2106    
2107    /* These are backtracking control verbs */    /* These are backtracking control verbs */
2108    
2109    OP_PRUNE,          /* 105 */    OP_MARK,           /* 149 always has an argument */
2110    OP_SKIP,           /* 106 */    OP_PRUNE,          /* 150 */
2111    OP_THEN,           /* 107 */    OP_PRUNE_ARG,      /* 151 same, but with argument */
2112    OP_COMMIT,         /* 108 */    OP_SKIP,           /* 152 */
2113      OP_SKIP_ARG,       /* 153 same, but with argument */
2114      OP_THEN,           /* 154 */
2115      OP_THEN_ARG,       /* 155 same, but with argument */
2116      OP_COMMIT,         /* 156 */
2117    
2118    /* These are forced failure and success verbs */    /* These are forced failure and success verbs */
2119    
2120    OP_FAIL,           /* 109 */    OP_FAIL,           /* 157 */
2121    OP_ACCEPT,         /* 110 */    OP_ACCEPT,         /* 158 */
2122      OP_ASSERT_ACCEPT,  /* 159 Used inside assertions */
2123      OP_CLOSE,          /* 160 Used before OP_ACCEPT to close open captures */
2124    
2125    /* This is used to skip a subpattern with a {0} quantifier */    /* This is used to skip a subpattern with a {0} quantifier */
2126    
2127    OP_SKIPZERO        /* 111 */    OP_SKIPZERO,       /* 161 */
2128    
2129      /* This is not an opcode, but is used to check that tables indexed by opcode
2130      are the correct length, in order to catch updating errors - there have been
2131      some in the past. */
2132    
2133      OP_TABLE_LENGTH
2134  };  };
2135    
2136    /* *** NOTE NOTE NOTE *** Whenever the list above is updated, the two macro
2137    definitions that follow must also be updated to match. There are also tables
2138    called "opcode_possessify" in pcre_compile.c and "coptable" and "poptable" in
2139    pcre_dfa_exec.c that must be updated. */
2140    
2141    
2142  /* This macro defines textual names for all the opcodes. These are used only  /* This macro defines textual names for all the opcodes. These are used only
2143  for debugging. The macro is referenced only in pcre_printint.c. */  for debugging, and some of them are only partial names. The macro is referenced
2144    only in pcre_printint.c, which fills out the full names in many cases (and in
2145    some cases doesn't actually use these names at all). */
2146    
2147  #define OP_NAME_LIST \  #define OP_NAME_LIST \
2148    "End", "\\A", "\\G", "\\K", "\\B", "\\b", "\\D", "\\d",         \    "End", "\\A", "\\G", "\\K", "\\B", "\\b", "\\D", "\\d",         \
2149    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \    "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte",         \
2150    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \    "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v",           \
2151    "extuni",  "\\Z", "\\z",                                        \    "extuni",  "\\Z", "\\z",                                        \
2152    "Opt", "^", "$", "char", "charnc", "not",                       \    "$", "$", "^", "^", "char", "chari", "not", "noti",             \
2153    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \    "*", "*?", "+", "+?", "?", "??",                                \
2154      "{", "{", "{",                                                  \
2155    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
2156    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \    "*", "*?", "+", "+?", "?", "??",                                \
2157      "{", "{", "{",                                                  \
2158      "*+","++", "?+", "{",                                           \
2159      "*", "*?", "+", "+?", "?", "??",                                \
2160      "{", "{", "{",                                                  \
2161      "*+","++", "?+", "{",                                           \
2162      "*", "*?", "+", "+?", "?", "??",                                \
2163      "{", "{", "{",                                                  \
2164    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
2165    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \    "*", "*?", "+", "+?", "?", "??", "{", "{", "{",                 \
2166    "*+","++", "?+", "{",                                           \    "*+","++", "?+", "{",                                           \
2167    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \    "*", "*?", "+", "+?", "?", "??", "{", "{",                      \
2168    "class", "nclass", "xclass", "Ref", "Recurse", "Callout",       \    "*+","++", "?+", "{",                                           \
2169    "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not",     \    "class", "nclass", "xclass", "Ref", "Refi", "DnRef", "DnRefi",  \
2170    "AssertB", "AssertB not", "Reverse",                            \    "Recurse", "Callout",                                           \
2171    "Once", "Bra", "CBra", "Cond", "SBra", "SCBra", "SCond",        \    "Alt", "Ket", "KetRmax", "KetRmin", "KetRpos",                  \
2172    "Cond ref", "Cond rec", "Cond def", "Brazero", "Braminzero",    \    "Reverse", "Assert", "Assert not", "AssertB", "AssertB not",    \
2173    "*PRUNE", "*SKIP", "*THEN", "*COMMIT", "*FAIL", "*ACCEPT",      \    "Once", "Once_NC",                                              \
2174    "Skip zero"    "Bra", "BraPos", "CBra", "CBraPos",                             \
2175      "Cond",                                                         \
2176      "SBra", "SBraPos", "SCBra", "SCBraPos",                         \
2177      "SCond",                                                        \
2178      "Cond ref", "Cond dnref", "Cond rec", "Cond dnrec", "Cond def", \
2179      "Brazero", "Braminzero", "Braposzero",                          \
2180      "*MARK", "*PRUNE", "*PRUNE", "*SKIP", "*SKIP",                  \
2181      "*THEN", "*THEN", "*COMMIT", "*FAIL",                           \
2182      "*ACCEPT", "*ASSERT_ACCEPT",                                    \
2183      "Close", "Skip zero"
2184    
2185    
2186  /* This macro defines the length of fixed length operations in the compiled  /* This macro defines the length of fixed length operations in the compiled
# Line 830  in UTF-8 mode. The code that uses this t Line 2197  in UTF-8 mode. The code that uses this t
2197    1, 1, 1, 1, 1,                 /* \A, \G, \K, \B, \b                     */ \    1, 1, 1, 1, 1,                 /* \A, \G, \K, \B, \b                     */ \
2198    1, 1, 1, 1, 1, 1,              /* \D, \d, \S, \s, \W, \w                 */ \    1, 1, 1, 1, 1, 1,              /* \D, \d, \S, \s, \W, \w                 */ \
2199    1, 1, 1,                       /* Any, AllAny, Anybyte                   */ \    1, 1, 1,                       /* Any, AllAny, Anybyte                   */ \
2200    3, 3, 1,                       /* NOTPROP, PROP, EXTUNI                  */ \    3, 3,                          /* \P, \p                                 */ \
2201    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \    1, 1, 1, 1, 1,                 /* \R, \H, \h, \V, \v                     */ \
2202    1, 1, 2, 1, 1,                 /* \Z, \z, Opt, ^, $                      */ \    1,                             /* \X                                     */ \
2203      1, 1, 1, 1, 1, 1,              /* \Z, \z, $, $M ^, ^M                    */ \
2204    2,                             /* Char  - the minimum length             */ \    2,                             /* Char  - the minimum length             */ \
2205    2,                             /* Charnc  - the minimum length           */ \    2,                             /* Chari  - the minimum length            */ \
2206    2,                             /* not                                    */ \    2,                             /* not                                    */ \
2207    /* Positive single-char repeats                            ** These are  */ \    2,                             /* noti                                   */ \
2208    2, 2, 2, 2, 2, 2,              /* *, *?, +, +?, ?, ??      ** minima in  */ \    /* Positive single-char repeats                             ** These are */ \
2209    4, 4, 4,                       /* upto, minupto, exact     ** UTF-8 mode */ \    2, 2, 2, 2, 2, 2,              /* *, *?, +, +?, ?, ??       ** minima in */ \
2210    2, 2, 2, 4,                    /* *+, ++, ?+, upto+                      */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* upto, minupto             ** mode      */ \
2211      2+IMM2_SIZE,                   /* exact                                  */ \
2212      2, 2, 2, 2+IMM2_SIZE,          /* *+, ++, ?+, upto+                      */ \
2213      2, 2, 2, 2, 2, 2,              /* *I, *?I, +I, +?I, ?I, ??I ** UTF-8     */ \
2214      2+IMM2_SIZE, 2+IMM2_SIZE,      /* upto I, minupto I                      */ \
2215      2+IMM2_SIZE,                   /* exact I                                */ \
2216      2, 2, 2, 2+IMM2_SIZE,          /* *+I, ++I, ?+I, upto+I                  */ \
2217    /* Negative single-char repeats - only for chars < 256                   */ \    /* Negative single-char repeats - only for chars < 256                   */ \
2218    2, 2, 2, 2, 2, 2,              /* NOT *, *?, +, +?, ?, ??                */ \    2, 2, 2, 2, 2, 2,              /* NOT *, *?, +, +?, ?, ??                */ \
2219    4, 4, 4,                       /* NOT upto, minupto, exact               */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* NOT upto, minupto                      */ \
2220    2, 2, 2, 4,                    /* Possessive *, +, ?, upto               */ \    2+IMM2_SIZE,                   /* NOT exact                              */ \
2221      2, 2, 2, 2+IMM2_SIZE,          /* Possessive NOT *, +, ?, upto           */ \
2222      2, 2, 2, 2, 2, 2,              /* NOT *I, *?I, +I, +?I, ?I, ??I          */ \
2223      2+IMM2_SIZE, 2+IMM2_SIZE,      /* NOT upto I, minupto I                  */ \
2224      2+IMM2_SIZE,                   /* NOT exact I                            */ \
2225      2, 2, 2, 2+IMM2_SIZE,          /* Possessive NOT *I, +I, ?I, upto I      */ \
2226    /* Positive type repeats                                                 */ \    /* Positive type repeats                                                 */ \
2227    2, 2, 2, 2, 2, 2,              /* Type *, *?, +, +?, ?, ??               */ \    2, 2, 2, 2, 2, 2,              /* Type *, *?, +, +?, ?, ??               */ \
2228    4, 4, 4,                       /* Type upto, minupto, exact              */ \    2+IMM2_SIZE, 2+IMM2_SIZE,      /* Type upto, minupto                     */ \
2229    2, 2, 2, 4,                    /* Possessive *+, ++, ?+, upto+           */ \    2+IMM2_SIZE,                   /* Type exact                             */ \
2230      2, 2, 2, 2+IMM2_SIZE,          /* Possessive *+, ++, ?+, upto+           */ \
2231    /* Character class & ref repeats                                         */ \    /* Character class & ref repeats                                         */ \
2232    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \    1, 1, 1, 1, 1, 1,              /* *, *?, +, +?, ?, ??                    */ \
2233    5, 5,                          /* CRRANGE, CRMINRANGE                    */ \    1+2*IMM2_SIZE, 1+2*IMM2_SIZE,  /* CRRANGE, CRMINRANGE                    */ \
2234   33,                             /* CLASS                                  */ \    1, 1, 1, 1+2*IMM2_SIZE,        /* Possessive *+, ++, ?+, CRPOSRANGE      */ \
2235   33,                             /* NCLASS                                 */ \    1+(32/sizeof(pcre_uchar)),     /* CLASS                                  */ \
2236      1+(32/sizeof(pcre_uchar)),     /* NCLASS                                 */ \
2237    0,                             /* XCLASS - variable length               */ \    0,                             /* XCLASS - variable length               */ \
2238    3,                             /* REF                                    */ \    1+IMM2_SIZE,                   /* REF                                    */ \
2239      1+IMM2_SIZE,                   /* REFI                                   */ \
2240      1+2*IMM2_SIZE,                 /* DNREF                                  */ \
2241      1+2*IMM2_SIZE,                 /* DNREFI                                 */ \
2242    1+LINK_SIZE,                   /* RECURSE                                */ \    1+LINK_SIZE,                   /* RECURSE                                */ \
2243    2+2*LINK_SIZE,                 /* CALLOUT                                */ \    2+2*LINK_SIZE,                 /* CALLOUT                                */ \
2244    1+LINK_SIZE,                   /* Alt                                    */ \    1+LINK_SIZE,                   /* Alt                                    */ \
2245    1+LINK_SIZE,                   /* Ket                                    */ \    1+LINK_SIZE,                   /* Ket                                    */ \
2246    1+LINK_SIZE,                   /* KetRmax                                */ \    1+LINK_SIZE,                   /* KetRmax                                */ \
2247    1+LINK_SIZE,                   /* KetRmin                                */ \    1+LINK_SIZE,                   /* KetRmin                                */ \
2248      1+LINK_SIZE,                   /* KetRpos                                */ \
2249      1+LINK_SIZE,                   /* Reverse                                */ \
2250    1+LINK_SIZE,                   /* Assert                                 */ \    1+LINK_SIZE,                   /* Assert                                 */ \
2251    1+LINK_SIZE,                   /* Assert not                             */ \    1+LINK_SIZE,                   /* Assert not                             */ \
2252    1+LINK_SIZE,                   /* Assert behind                          */ \    1+LINK_SIZE,                   /* Assert behind                          */ \
2253    1+LINK_SIZE,                   /* Assert behind not                      */ \    1+LINK_SIZE,                   /* Assert behind not                      */ \
   1+LINK_SIZE,                   /* Reverse                                */ \  
2254    1+LINK_SIZE,                   /* ONCE                                   */ \    1+LINK_SIZE,                   /* ONCE                                   */ \
2255      1+LINK_SIZE,                   /* ONCE_NC                                */ \
2256    1+LINK_SIZE,                   /* BRA                                    */ \    1+LINK_SIZE,                   /* BRA                                    */ \
2257    3+LINK_SIZE,                   /* CBRA                                   */ \    1+LINK_SIZE,                   /* BRAPOS                                 */ \
2258      1+LINK_SIZE+IMM2_SIZE,         /* CBRA                                   */ \
2259      1+LINK_SIZE+IMM2_SIZE,         /* CBRAPOS                                */ \
2260    1+LINK_SIZE,                   /* COND                                   */ \    1+LINK_SIZE,                   /* COND                                   */ \
2261    1+LINK_SIZE,                   /* SBRA                                   */ \    1+LINK_SIZE,                   /* SBRA                                   */ \
2262    3+LINK_SIZE,                   /* SCBRA                                  */ \    1+LINK_SIZE,                   /* SBRAPOS                                */ \
2263      1+LINK_SIZE+IMM2_SIZE,         /* SCBRA                                  */ \
2264      1+LINK_SIZE+IMM2_SIZE,         /* SCBRAPOS                               */ \
2265    1+LINK_SIZE,                   /* SCOND                                  */ \    1+LINK_SIZE,                   /* SCOND                                  */ \
2266    3,                             /* CREF                                   */ \    1+IMM2_SIZE, 1+2*IMM2_SIZE,    /* CREF, DNCREF                           */ \
2267    3,                             /* RREF                                   */ \    1+IMM2_SIZE, 1+2*IMM2_SIZE,    /* RREF, DNRREF                           */ \
2268    1,                             /* DEF                                    */ \    1,                             /* DEF                                    */ \
2269    1, 1,                          /* BRAZERO, BRAMINZERO                    */ \    1, 1, 1,                       /* BRAZERO, BRAMINZERO, BRAPOSZERO        */ \
2270    1, 1, 1, 1,                    /* PRUNE, SKIP, THEN, COMMIT,             */ \    3, 1, 3,                       /* MARK, PRUNE, PRUNE_ARG                 */ \
2271    1, 1, 1                        /* FAIL, ACCEPT, SKIPZERO                 */    1, 3,                          /* SKIP, SKIP_ARG                         */ \
2272      1, 3,                          /* THEN, THEN_ARG                         */ \
2273      1, 1, 1, 1,                    /* COMMIT, FAIL, ACCEPT, ASSERT_ACCEPT    */ \
2274      1+IMM2_SIZE, 1                 /* CLOSE, SKIPZERO                        */
2275    
2276  /* A magic value for OP_RREF to indicate the "any recursion" condition. */  /* A magic value for OP_RREF to indicate the "any recursion" condition. */
2277    
2278  #define RREF_ANY  0xffff  #define RREF_ANY  0xffff
2279    
2280  /* Error code numbers. They are given names so that they can more easily be  /* Compile time error code numbers. They are given names so that they can more
2281  tracked. */  easily be tracked. When a new number is added, the table called eint in
2282    pcreposix.c must be updated. */
2283    
2284  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,  ERR5,  ERR6,  ERR7,  ERR8,  ERR9,  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,  ERR5,  ERR6,  ERR7,  ERR8,  ERR9,
2285         ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19,         ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19,
# Line 894  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4, Line 2287  enum { ERR0,  ERR1,  ERR2,  ERR3,  ERR4,
2287         ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,         ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
2288         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,         ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
2289         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,         ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
2290         ERR60, ERR61, ERR62, ERR63, ERR64 };         ERR60, ERR61, ERR62, ERR63, ERR64, ERR65, ERR66, ERR67, ERR68, ERR69,
2291           ERR70, ERR71, ERR72, ERR73, ERR74, ERR75, ERR76, ERR77, ERR78, ERR79,
2292           ERR80, ERR81, ERR82, ERR83, ERR84, ERR85, ERR86, ERR87, ERRCOUNT };
2293    
2294    /* JIT compiling modes. The function list is indexed by them. */
2295    
2296    enum { JIT_COMPILE, JIT_PARTIAL_SOFT_COMPILE, JIT_PARTIAL_HARD_COMPILE,
2297           JIT_NUMBER_OF_COMPILE_MODES };
2298    
2299  /* 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
2300  code vector run on as long as necessary after the end. We store an explicit  code vector run on as long as necessary after the end. We store an explicit
2301  offset to the name table so that if a regex is compiled on one host, saved, and  offset to the name table so that if a regex is compiled on one host, saved, and
2302  then run on another where the size of pointers is different, all might still  then run on another where the size of pointers is different, all might still
2303  be well. For the case of compiled-on-4 and run-on-8, we include an extra  be well.
2304  pointer that is always NULL. For future-proofing, a few dummy fields were  
2305  originally included - even though you can never get this planning right - but  The size of the structure must be a multiple of 8 bytes. For the case of
2306  there is only one left now.  compiled-on-4 and run-on-8, we include an extra pointer that is always NULL so
2307    that there are an even number of pointers which therefore are a multiple of 8
2308  NOTE NOTE NOTE:  bytes.
2309  Because people can now save and re-use compiled patterns, any additions to this  
2310  structure should be made at the end, and something earlier (e.g. a new  It is necessary to fork the struct for the 32 bit library, since it needs to
2311  flag in the options or one of the dummy fields) should indicate that the new  use pcre_uint32 for first_char and req_char. We can't put an ifdef inside the
2312  fields are present. Currently PCRE always sets the dummy fields to zero.  typedef because pcretest needs access to the struct of the 8-, 16- and 32-bit
2313  NOTE NOTE NOTE:  variants.
2314    
2315    *** WARNING ***
2316    When new fields are added to these structures, remember to adjust the code in
2317    pcre_byte_order.c that is concerned with swapping the byte order of the fields
2318    when a compiled regex is reloaded on a host with different endianness.
2319    *** WARNING ***
2320    There is also similar byte-flipping code in pcretest.c, which is used for
2321    testing the byte-flipping features. It must also be kept in step.
2322    *** WARNING ***
2323  */  */
2324    
2325  typedef struct real_pcre {  typedef struct real_pcre8_or_16 {
2326      pcre_uint32 magic_number;
2327      pcre_uint32 size;               /* Total that was malloced */
2328      pcre_uint32 options;            /* Public options */
2329      pcre_uint32 flags;              /* Private flags */
2330      pcre_uint32 limit_match;        /* Limit set from regex */
2331      pcre_uint32 limit_recursion;    /* Limit set from regex */
2332      pcre_uint16 first_char;         /* Starting character */
2333      pcre_uint16 req_char;           /* This character must be seen */
2334      pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2335      pcre_uint16 top_bracket;        /* Highest numbered group */
2336      pcre_uint16 top_backref;        /* Highest numbered back reference */
2337      pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2338      pcre_uint16 name_entry_size;    /* Size of any name items */
2339      pcre_uint16 name_count;         /* Number of name items */
2340      pcre_uint16 ref_count;          /* Reference count */
2341      pcre_uint16 dummy1;             /* To ensure size is a multiple of 8 */
2342      pcre_uint16 dummy2;             /* To ensure size is a multiple of 8 */
2343      pcre_uint16 dummy3;             /* To ensure size is a multiple of 8 */
2344      const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2345      void             *nullpad;      /* NULL padding */
2346    } real_pcre8_or_16;
2347    
2348    typedef struct real_pcre8_or_16 real_pcre;
2349    typedef struct real_pcre8_or_16 real_pcre16;
2350    
2351    typedef struct real_pcre32 {
2352    pcre_uint32 magic_number;    pcre_uint32 magic_number;
2353    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2354    pcre_uint32 options;            /* Public options */    pcre_uint32 options;            /* Public options */
2355    pcre_uint16 flags;              /* Private flags */    pcre_uint32 flags;              /* Private flags */
2356    pcre_uint16 dummy1;             /* For future use */    pcre_uint32 limit_match;        /* Limit set from regex */
2357    pcre_uint16 top_bracket;    pcre_uint32 limit_recursion;    /* Limit set from regex */
2358    pcre_uint16 top_backref;    pcre_uint32 first_char;         /* Starting character */
2359    pcre_uint16 first_byte;    pcre_uint32 req_char;           /* This character must be seen */
2360    pcre_uint16 req_byte;    pcre_uint16 max_lookbehind;     /* Longest lookbehind (characters) */
2361      pcre_uint16 top_bracket;        /* Highest numbered group */
2362      pcre_uint16 top_backref;        /* Highest numbered back reference */
2363    pcre_uint16 name_table_offset;  /* Offset to name table that follows */    pcre_uint16 name_table_offset;  /* Offset to name table that follows */
2364    pcre_uint16 name_entry_size;    /* Size of any name items */    pcre_uint16 name_entry_size;    /* Size of any name items */
2365    pcre_uint16 name_count;         /* Number of name items */    pcre_uint16 name_count;         /* Number of name items */
2366    pcre_uint16 ref_count;          /* Reference count */    pcre_uint16 ref_count;          /* Reference count */
2367      pcre_uint16 dummy;              /* To ensure size is a multiple of 8 */
2368      const pcre_uint8 *tables;       /* Pointer to tables or NULL for std */
2369      void             *nullpad;      /* NULL padding */
2370    } real_pcre32;
2371    
2372    #if defined COMPILE_PCRE8
2373    #define REAL_PCRE real_pcre
2374    #elif defined COMPILE_PCRE16
2375    #define REAL_PCRE real_pcre16
2376    #elif defined COMPILE_PCRE32
2377    #define REAL_PCRE real_pcre32
2378    #endif
2379    
2380    const unsigned char *tables;    /* Pointer to tables or NULL for std */  /* Assert that the size of REAL_PCRE is divisible by 8 */
2381    const unsigned char *nullpad;   /* NULL padding */  typedef int __assert_real_pcre_size_divisible_8[(sizeof(REAL_PCRE) % 8) == 0 ? 1 : -1];
2382  } real_pcre;  
2383    /* Needed in pcretest to access some fields in the real_pcre* structures
2384     * directly. They're unified for 8/16/32 bits since the structs only differ
2385     * after these fields; if that ever changes, need to fork those defines into
2386     * 8/16 and 32 bit versions. */
2387    #define REAL_PCRE_MAGIC(re)     (((REAL_PCRE*)re)->magic_number)
2388    #define REAL_PCRE_SIZE(re)      (((REAL_PCRE*)re)->size)
2389    #define REAL_PCRE_OPTIONS(re)   (((REAL_PCRE*)re)->options)
2390    #define REAL_PCRE_FLAGS(re)     (((REAL_PCRE*)re)->flags)
2391    
2392  /* 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
2393  remark (see NOTE above) about extending this structure applies. */  remark (see NOTE above) about extending this structure applies. */
2394    
2395  typedef struct pcre_study_data {  typedef struct pcre_study_data {
2396    pcre_uint32 size;               /* Total that was malloced */    pcre_uint32 size;               /* Total that was malloced */
2397    pcre_uint32 options;    pcre_uint32 flags;              /* Private flags */
2398    uschar start_bits[32];    pcre_uint8 start_bits[32];      /* Starting char bits */
2399      pcre_uint32 minlength;          /* Minimum subject length */
2400  } pcre_study_data;  } pcre_study_data;
2401    
2402    /* Structure for building a chain of open capturing subpatterns during
2403    compiling, so that instructions to close them can be compiled when (*ACCEPT) is
2404    encountered. This is also used to identify subpatterns that contain recursive
2405    back references to themselves, so that they can be made atomic. */
2406    
2407    typedef struct open_capitem {
2408      struct open_capitem *next;    /* Chain link */
2409      pcre_uint16 number;           /* Capture number */
2410      pcre_uint16 flag;             /* Set TRUE if recursive back ref */
2411    } open_capitem;
2412    
2413    /* Structure for building a list of named groups during the first pass of
2414    compiling. */
2415    
2416    typedef struct named_group {
2417      const pcre_uchar  *name;          /* Points to the name in the pattern */
2418      int                length;        /* Length of the name */
2419      pcre_uint32        number;        /* Group number */
2420    } named_group;
2421    
2422  /* Structure for passing "static" information around between the functions  /* Structure for passing "static" information around between the functions
2423  doing the compiling, so that they are thread-safe. */  doing the compiling, so that they are thread-safe. */
2424    
2425  typedef struct compile_data {  typedef struct compile_data {
2426    const uschar *lcc;            /* Points to lower casing table */    const pcre_uint8 *lcc;            /* Points to lower casing table */
2427    const uschar *fcc;            /* Points to case-flipping table */    const pcre_uint8 *fcc;            /* Points to case-flipping table */
2428    const uschar *cbits;          /* Points to character type table */    const pcre_uint8 *cbits;          /* Points to character type table */
2429    const uschar *ctypes;         /* Points to table of type maps */    const pcre_uint8 *ctypes;         /* Points to table of type maps */
2430    const uschar *start_workspace;/* The start of working space */    const pcre_uchar *start_workspace;/* The start of working space */
2431    const uschar *start_code;     /* The start of the compiled code */    const pcre_uchar *start_code;     /* The start of the compiled code */
2432    const uschar *start_pattern;  /* The start of the pattern */    const pcre_uchar *start_pattern;  /* The start of the pattern */
2433    const uschar *end_pattern;    /* The end of the pattern */    const pcre_uchar *end_pattern;    /* The end of the pattern */
2434    uschar *hwm;                  /* High watermark of workspace */    pcre_uchar *hwm;                  /* High watermark of workspace */
2435    uschar *name_table;           /* The name/number table */    open_capitem *open_caps;          /* Chain of open capture items */
2436    int  names_found;             /* Number of entries so far */    named_group *named_groups;        /* Points to vector in pre-compile */
2437    int  name_entry_size;         /* Size of each entry */    pcre_uchar *name_table;           /* The name/number table */
2438    int  bracount;                /* Count of capturing parens as we compile */    int  names_found;                 /* Number of entries so far */
2439    int  final_bracount;          /* Saved value after first pass */    int  name_entry_size;             /* Size of each entry */
2440    int  top_backref;             /* Maximum back reference */    int  named_group_list_size;       /* Number of entries in the list */
2441    unsigned int backref_map;     /* Bitmap of low back refs */    int  workspace_size;              /* Size of workspace */
2442    int  external_options;        /* External (initial) options */    unsigned int bracount;            /* Count of capturing parens as we compile */
2443    int  external_flags;          /* External flag bits to be set */    int  final_bracount;              /* Saved value after first pass */
2444    int  req_varyopt;             /* "After variable item" flag for reqbyte */    int  max_lookbehind;              /* Maximum lookbehind (characters) */
2445    BOOL had_accept;              /* (*ACCEPT) encountered */    int  top_backref;                 /* Maximum back reference */
2446    int  nltype;                  /* Newline type */    unsigned int backref_map;         /* Bitmap of low back refs */
2447    int  nllen;                   /* Newline string length */    unsigned int namedrefcount;       /* Number of backreferences by name */
2448    uschar nl[4];                 /* Newline string when fixed length */    int  parens_depth;                /* Depth of nested parentheses */
2449      int  assert_depth;                /* Depth of nested assertions */
2450      pcre_uint32 external_options;     /* External (initial) options */
2451      pcre_uint32 external_flags;       /* External flag bits to be set */
2452      int  req_varyopt;                 /* "After variable item" flag for reqbyte */
2453      BOOL had_accept;                  /* (*ACCEPT) encountered */
2454      BOOL had_pruneorskip;             /* (*PRUNE) or (*SKIP) encountered */
2455      BOOL check_lookbehind;            /* Lookbehinds need later checking */
2456      BOOL dupnames;                    /* Duplicate names exist */
2457      BOOL dupgroups;                   /* Duplicate groups exist: (?| found */
2458      BOOL iscondassert;                /* Next assert is a condition */
2459      int  nltype;                      /* Newline type */
2460      int  nllen;                       /* Newline string length */
2461      pcre_uchar nl[4];                 /* Newline string when fixed length */
2462  } compile_data;  } compile_data;
2463    
2464  /* Structure for maintaining a chain of pointers to the currently incomplete  /* Structure for maintaining a chain of pointers to the currently incomplete
2465  branches, for testing for left recursion. */  branches, for testing for left recursion while compiling. */
2466    
2467  typedef struct branch_chain {  typedef struct branch_chain {
2468    struct branch_chain *outer;    struct branch_chain *outer;
2469    uschar *current;    pcre_uchar *current_branch;
2470  } branch_chain;  } branch_chain;
2471    
2472    /* Structure for mutual recursion detection. */
2473    
2474    typedef struct recurse_check {
2475      struct recurse_check *prev;
2476      const pcre_uchar *group;
2477    } recurse_check;
2478    
2479  /* Structure for items in a linked list that represents an explicit recursive  /* Structure for items in a linked list that represents an explicit recursive
2480  call within the pattern. */  call within the pattern; used by pcre_exec(). */
2481    
2482  typedef struct recursion_info {  typedef struct recursion_info {
2483    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */    struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
2484    int group_num;                /* Number of group that was called */    unsigned int group_num;         /* Number of group that was called */
2485    const uschar *after_call;     /* "Return value": points after the call in the expr */    int *offset_save;               /* Pointer to start of saved offsets */
2486    USPTR save_start;             /* Old value of mstart */    int saved_max;                  /* Number of saved offsets */
2487    int *offset_save;             /* Pointer to start of saved offsets */    int saved_capture_last;         /* Last capture number */
2488    int saved_max;                /* Number of saved offsets */    PCRE_PUCHAR subject_position;   /* Position at start of recursion */
2489  } recursion_info;  } recursion_info;
2490    
2491    /* A similar structure for pcre_dfa_exec(). */
2492    
2493    typedef struct dfa_recursion_info {
2494      struct dfa_recursion_info *prevrec;
2495      int group_num;
2496      PCRE_PUCHAR subject_position;
2497    } dfa_recursion_info;
2498    
2499  /* 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
2500  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
2501  has been matched by a subpattern - to break infinite loops. */  has been matched by a subpattern - to break infinite loops; used by
2502    pcre_exec(). */
2503    
2504  typedef struct eptrblock {  typedef struct eptrblock {
2505    struct eptrblock *epb_prev;    struct eptrblock *epb_prev;
2506    USPTR epb_saved_eptr;    PCRE_PUCHAR epb_saved_eptr;
2507  } eptrblock;  } eptrblock;
2508    
2509    
# Line 1007  typedef struct match_data { Line 2514  typedef struct match_data {
2514    unsigned long int match_call_count;      /* As it says */    unsigned long int match_call_count;      /* As it says */
2515    unsigned long int match_limit;           /* As it says */    unsigned long int match_limit;           /* As it says */
2516    unsigned long int match_limit_recursion; /* As it says */    unsigned long int match_limit_recursion; /* As it says */
2517    int   *offset_vector;         /* Offset vector */    int   *offset_vector;           /* Offset vector */
2518    int    offset_end;            /* One past the end */    int    offset_end;              /* One past the end */
2519    int    offset_max;            /* The maximum usable for return data */    int    offset_max;              /* The maximum usable for return data */
2520    int    nltype;                /* Newline type */    int    nltype;                  /* Newline type */
2521    int    nllen;                 /* Newline string length */    int    nllen;                   /* Newline string length */
2522    uschar nl[4];                 /* Newline string when fixed */    int    name_count;              /* Number of names in name table */
2523    const uschar *lcc;            /* Points to lower casing table */    int    name_entry_size;         /* Size of entry in names table */
2524    const uschar *ctypes;         /* Points to table of type maps */    unsigned int skip_arg_count;    /* For counting SKIP_ARGs */
2525    BOOL   offset_overflow;       /* Set if too many extractions */    unsigned int ignore_skip_arg;   /* For re-run when SKIP arg name not found */
2526    BOOL   notbol;                /* NOTBOL flag */    pcre_uchar *name_table;         /* Table of names */
2527    BOOL   noteol;                /* NOTEOL flag */    pcre_uchar nl[4];               /* Newline string when fixed */
2528    BOOL   utf8;                  /* UTF8 flag */    const  pcre_uint8 *lcc;         /* Points to lower casing table */
2529    BOOL   jscript_compat;        /* JAVASCRIPT_COMPAT flag */    const  pcre_uint8 *fcc;         /* Points to case-flipping table */
2530    BOOL   endonly;               /* Dollar not before final \n */    const  pcre_uint8 *ctypes;      /* Points to table of type maps */
2531    BOOL   notempty;              /* Empty string match not wanted */    BOOL   notbol;                  /* NOTBOL flag */
2532    BOOL   partial;               /* PARTIAL flag */    BOOL   noteol;                  /* NOTEOL flag */
2533    BOOL   hitend;                /* Hit the end of the subject at some point */    BOOL   utf;                     /* UTF-8 / UTF-16 flag */
2534    BOOL   bsr_anycrlf;           /* \R is just any CRLF, not full Unicode */    BOOL   jscript_compat;          /* JAVASCRIPT_COMPAT flag */
2535    const uschar *start_code;     /* For use when recursing */    BOOL   use_ucp;                 /* PCRE_UCP flag */
2536    USPTR  start_subject;         /* Start of the subject string */    BOOL   endonly;                 /* Dollar not before final \n */
2537    USPTR  end_subject;           /* End of the subject string */    BOOL   notempty;                /* Empty string match not wanted */
2538    USPTR  start_match_ptr;       /* Start of matched string */    BOOL   notempty_atstart;        /* Empty string match at start not wanted */
2539    USPTR  end_match_ptr;         /* Subject position at end match */    BOOL   hitend;                  /* Hit the end of the subject at some point */
2540    int    end_offset_top;        /* Highwater mark at end of match */    BOOL   bsr_anycrlf;             /* \R is just any CRLF, not full Unicode */
2541    int    capture_last;          /* Most recent capture number */    BOOL   hasthen;                 /* Pattern contains (*THEN) */
2542    int    start_offset;          /* The start offset value */    const  pcre_uchar *start_code;  /* For use when recursing */
2543    eptrblock *eptrchain;         /* Chain of eptrblocks for tail recursions */    PCRE_PUCHAR start_subject;      /* Start of the subject string */
2544    int    eptrn;                 /* Next free eptrblock */    PCRE_PUCHAR end_subject;        /* End of the subject string */
2545    recursion_info *recursive;    /* Linked list of recursion data */    PCRE_PUCHAR start_match_ptr;    /* Start of matched string */
2546    void  *callout_data;          /* To pass back to callouts */    PCRE_PUCHAR end_match_ptr;      /* Subject position at end match */
2547      PCRE_PUCHAR start_used_ptr;     /* Earliest consulted character */
2548      int    partial;                 /* PARTIAL options */
2549      int    end_offset_top;          /* Highwater mark at end of match */
2550      pcre_int32 capture_last;        /* Most recent capture number + overflow flag */
2551      int    start_offset;            /* The start offset value */
2552      int    match_function_type;     /* Set for certain special calls of MATCH() */
2553      eptrblock *eptrchain;           /* Chain of eptrblocks for tail recursions */
2554      int    eptrn;                   /* Next free eptrblock */
2555      recursion_info *recursive;      /* Linked list of recursion data */
2556      void  *callout_data;            /* To pass back to callouts */
2557      const  pcre_uchar *mark;        /* Mark pointer to pass back on success */
2558      const  pcre_uchar *nomatch_mark;/* Mark pointer to pass back on failure */
2559      const  pcre_uchar *once_target; /* Where to back up to for atomic groups */
2560    #ifdef NO_RECURSE
2561      void  *match_frames_base;       /* For remembering malloc'd frames */
2562    #endif
2563  } match_data;  } match_data;
2564    
2565  /* 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
2566  functions. */  functions. */
2567    
2568  typedef struct dfa_match_data {  typedef struct dfa_match_data {
2569    const uschar *start_code;     /* Start of the compiled pattern */    const pcre_uchar *start_code;     /* Start of the compiled pattern */
2570    const uschar *start_subject;  /* Start of the subject string */    const pcre_uchar *start_subject ; /* Start of the subject string */
2571    const uschar *end_subject;    /* End of subject string */    const pcre_uchar *end_subject;    /* End of subject string */
2572    const uschar *tables;         /* Character tables */    const pcre_uchar *start_used_ptr; /* Earliest consulted character */
2573    int   moptions;               /* Match options */    const pcre_uint8 *tables;         /* Character tables */
2574    int   poptions;               /* Pattern options */    int   start_offset;               /* The start offset value */
2575    int    nltype;                /* Newline type */    int   moptions;                   /* Match options */
2576    int    nllen;                 /* Newline string length */    int   poptions;                   /* Pattern options */
2577    uschar nl[4];                 /* Newline string when fixed */    int   nltype;                     /* Newline type */
2578    void  *callout_data;          /* To pass back to callouts */    int   nllen;                      /* Newline string length */
2579      pcre_uchar nl[4];                 /* Newline string when fixed */
2580      void *callout_data;               /* To pass back to callouts */
2581      dfa_recursion_info *recursive;    /* Linked list of recursion data */
2582  } dfa_match_data;  } dfa_match_data;
2583    
2584  /* Bit definitions for entries in the pcre_ctypes table. */  /* Bit definitions for entries in the pcre_ctypes table. */
# Line 1088  total length. */ Line 2614  total length. */
2614  #define ctypes_offset (cbits_offset + cbit_length)  #define ctypes_offset (cbits_offset + cbit_length)
2615  #define tables_length (ctypes_offset + 256)  #define tables_length (ctypes_offset + 256)
2616    
2617    /* Internal function and data prefixes. */
2618    
2619    #if defined COMPILE_PCRE8
2620    #ifndef PUBL
2621    #define PUBL(name) pcre_##name
2622    #endif
2623    #ifndef PRIV
2624    #define PRIV(name) _pcre_##name
2625    #endif
2626    #elif defined COMPILE_PCRE16
2627    #ifndef PUBL
2628    #define PUBL(name) pcre16_##name
2629    #endif
2630    #ifndef PRIV
2631    #define PRIV(name) _pcre16_##name
2632    #endif
2633    #elif defined COMPILE_PCRE32
2634    #ifndef PUBL
2635    #define PUBL(name) pcre32_##name
2636    #endif
2637    #ifndef PRIV
2638    #define PRIV(name) _pcre32_##name
2639    #endif
2640    #else
2641    #error Unsupported compiling mode
2642    #endif /* COMPILE_PCRE[8|16|32] */
2643    
2644  /* 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
2645  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
2646  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 1105  of the exported public functions. They h Line 2658  of the exported public functions. They h
2658  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
2659  pcre_tables.c module. */  pcre_tables.c module. */
2660    
2661  extern const int    _pcre_utf8_table1[];  #ifdef COMPILE_PCRE8
2662  extern const int    _pcre_utf8_table2[];  extern const int            PRIV(utf8_table1)[];
2663  extern const int    _pcre_utf8_table3[];  extern const int            PRIV(utf8_table1_size);
2664  extern const uschar _pcre_utf8_table4[];  extern const int            PRIV(utf8_table2)[];
2665    extern const int            PRIV(utf8_table3)[];
2666  extern const int    _pcre_utf8_table1_size;  extern const pcre_uint8     PRIV(utf8_table4)[];
2667    #endif /* COMPILE_PCRE8 */
2668    
2669    extern const char           PRIV(utt_names)[];
2670    extern const ucp_type_table PRIV(utt)[];
2671    extern const int            PRIV(utt_size);
2672    
2673  extern const char   _pcre_utt_names[];  extern const pcre_uint8     PRIV(OP_lengths)[];
2674  extern const ucp_type_table _pcre_utt[];  extern const pcre_uint8     PRIV(default_tables)[];
 extern const int _pcre_utt_size;  
2675    
2676  extern const uschar _pcre_default_tables[];  extern const pcre_uint32    PRIV(hspace_list)[];
2677    extern const pcre_uint32    PRIV(vspace_list)[];
 extern const uschar _pcre_OP_lengths[];  
2678    
2679    
2680  /* Internal shared functions. These are functions that are used by more than  /* Internal shared functions. These are functions that are used by more than
2681  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
2682  sense, but are not part of the PCRE public API. */  sense, but are not part of the PCRE public API. */
2683    
2684  extern BOOL         _pcre_is_newline(const uschar *, int, const uschar *,  /* String comparison functions. */
2685                        int *, BOOL);  #if defined COMPILE_PCRE8
 extern int          _pcre_ord2utf8(int, uschar *);  
 extern real_pcre   *_pcre_try_flipped(const real_pcre *, real_pcre *,  
                       const pcre_study_data *, pcre_study_data *);  
 extern int          _pcre_valid_utf8(const uschar *, int);  
 extern BOOL         _pcre_was_newline(const uschar *, int, const uschar *,  
                       int *, BOOL);  
 extern BOOL         _pcre_xclass(int, const uschar *);  
2686    
2687    #define STRCMP_UC_UC(str1, str2) \
2688      strcmp((char *)(str1), (char *)(str2))
2689    #define STRCMP_UC_C8(str1, str2) \
2690      strcmp((char *)(str1), (str2))
2691    #define STRNCMP_UC_UC(str1, str2, num) \
2692      strncmp((char *)(str1), (char *)(str2), (num))
2693    #define STRNCMP_UC_C8(str1, str2, num) \
2694      strncmp((char *)(str1), (str2), (num))
2695    #define STRLEN_UC(str) strlen((const char *)str)
2696    
2697    #elif defined COMPILE_PCRE16 || defined COMPILE_PCRE32
2698    
2699    extern int               PRIV(strcmp_uc_uc)(const pcre_uchar *,
2700                               const pcre_uchar *);
2701    extern int               PRIV(strcmp_uc_c8)(const pcre_uchar *,
2702                               const char *);
2703    extern int               PRIV(strncmp_uc_uc)(const pcre_uchar *,
2704                               const pcre_uchar *, unsigned int num);
2705    extern int               PRIV(strncmp_uc_c8)(const pcre_uchar *,
2706                               const char *, unsigned int num);
2707    extern unsigned int      PRIV(strlen_uc)(const pcre_uchar *str);
2708    
2709    #define STRCMP_UC_UC(str1, str2) \
2710      PRIV(strcmp_uc_uc)((str1), (str2))
2711    #define STRCMP_UC_C8(str1, str2) \
2712      PRIV(strcmp_uc_c8)((str1), (str2))
2713    #define STRNCMP_UC_UC(str1, str2, num) \
2714      PRIV(strncmp_uc_uc)((str1), (str2), (num))
2715    #define STRNCMP_UC_C8(str1, str2, num) \
2716      PRIV(strncmp_uc_c8)((str1), (str2), (num))
2717    #define STRLEN_UC(str) PRIV(strlen_uc)(str)
2718    
2719    #endif /* COMPILE_PCRE[8|16|32] */
2720    
2721    #if defined COMPILE_PCRE8 || defined COMPILE_PCRE16
2722    
2723    #define STRCMP_UC_UC_TEST(str1, str2) STRCMP_UC_UC(str1, str2)
2724    #define STRCMP_UC_C8_TEST(str1, str2) STRCMP_UC_C8(str1, str2)
2725    
2726    #elif defined COMPILE_PCRE32
2727    
2728    extern int               PRIV(strcmp_uc_uc_utf)(const pcre_uchar *,
2729                               const pcre_uchar *);
2730    extern int               PRIV(strcmp_uc_c8_utf)(const pcre_uchar *,
2731                               const char *);
2732    
2733    #define STRCMP_UC_UC_TEST(str1, str2) \
2734      (utf ? PRIV(strcmp_uc_uc_utf)((str1), (str2)) : PRIV(strcmp_uc_uc)((str1), (str2)))
2735    #define STRCMP_UC_C8_TEST(str1, str2) \
2736      (utf ? PRIV(strcmp_uc_c8_utf)((str1), (str2)) : PRIV(strcmp_uc_c8)((str1), (str2)))
2737    
2738    #endif /* COMPILE_PCRE[8|16|32] */
2739    
2740    extern const pcre_uchar *PRIV(find_bracket)(const pcre_uchar *, BOOL, int);
2741    extern BOOL              PRIV(is_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2742                               int *, BOOL);
2743    extern unsigned int      PRIV(ord2utf)(pcre_uint32, pcre_uchar *);
2744    extern int               PRIV(valid_utf)(PCRE_PUCHAR, int, int *);
2745    extern BOOL              PRIV(was_newline)(PCRE_PUCHAR, int, PCRE_PUCHAR,
2746                               int *, BOOL);
2747    extern BOOL              PRIV(xclass)(pcre_uint32, const pcre_uchar *, BOOL);
2748    
2749    #ifdef SUPPORT_JIT
2750    extern void              PRIV(jit_compile)(const REAL_PCRE *,
2751                               PUBL(extra) *, int);
2752    extern int               PRIV(jit_exec)(const PUBL(extra) *,
2753                               const pcre_uchar *, int, int, int, int *, int);
2754    extern void              PRIV(jit_free)(void *);
2755    extern int               PRIV(jit_get_size)(void *);
2756    extern const char*       PRIV(jit_get_target)(void);
2757    #endif
2758    
2759  /* Unicode character database (UCD) */  /* Unicode character database (UCD) */
2760    
2761  typedef struct {  typedef struct {
2762    uschar script;    pcre_uint8 script;     /* ucp_Arabic, etc. */
2763    uschar chartype;    pcre_uint8 chartype;   /* ucp_Cc, etc. (general categories) */
2764    pcre_int16 other_case;    pcre_uint8 gbprop;     /* ucp_gbControl, etc. (grapheme break property) */
2765      pcre_uint8 caseset;    /* offset to multichar other cases or zero */
2766      pcre_int32 other_case; /* offset to other case, or zero if none */
2767  } ucd_record;  } ucd_record;
2768    
2769  extern const ucd_record  ucd_records[];  extern const pcre_uint32 PRIV(ucd_caseless_sets)[];
2770  extern const uschar      ucd_stage1[];  extern const ucd_record  PRIV(ucd_records)[];
2771  extern const pcre_uint16 ucd_stage2[];  extern const pcre_uint8  PRIV(ucd_stage1)[];
2772  extern const int ucp_gentype[];  extern const pcre_uint16 PRIV(ucd_stage2)[];
2773    extern const pcre_uint32 PRIV(ucp_gentype)[];
2774    extern const pcre_uint32 PRIV(ucp_gbtable)[];
2775    #ifdef COMPILE_PCRE32
2776    extern const ucd_record  PRIV(dummy_ucd_record)[];
2777    #endif
2778    #ifdef SUPPORT_JIT
2779    extern const int         PRIV(ucp_typerange)[];
2780    #endif
2781    
2782    #ifdef SUPPORT_UCP
2783  /* UCD access macros */  /* UCD access macros */
2784    
2785  #define UCD_BLOCK_SIZE 128  #define UCD_BLOCK_SIZE 128
2786  #define GET_UCD(ch) (ucd_records + ucd_stage2[ucd_stage1[(ch) / UCD_BLOCK_SIZE] * \  #define REAL_GET_UCD(ch) (PRIV(ucd_records) + \
2787          UCD_BLOCK_SIZE + ch % UCD_BLOCK_SIZE])          PRIV(ucd_stage2)[PRIV(ucd_stage1)[(int)(ch) / UCD_BLOCK_SIZE] * \
2788            UCD_BLOCK_SIZE + (int)(ch) % UCD_BLOCK_SIZE])
2789    
2790    #ifdef COMPILE_PCRE32
2791    #define GET_UCD(ch) ((ch > 0x10ffff)? PRIV(dummy_ucd_record) : REAL_GET_UCD(ch))
2792    #else
2793    #define GET_UCD(ch) REAL_GET_UCD(ch)
2794    #endif
2795    
2796    #define UCD_CHARTYPE(ch)    GET_UCD(ch)->chartype
2797    #define UCD_SCRIPT(ch)      GET_UCD(ch)->script
2798    #define UCD_CATEGORY(ch)    PRIV(ucp_gentype)[UCD_CHARTYPE(ch)]
2799    #define UCD_GRAPHBREAK(ch)  GET_UCD(ch)->gbprop
2800    #define UCD_CASESET(ch)     GET_UCD(ch)->caseset
2801    #define UCD_OTHERCASE(ch)   ((pcre_uint32)((int)ch + (int)(GET_UCD(ch)->other_case)))
2802    
2803  #define UCD_CHARTYPE(ch)  GET_UCD(ch)->chartype  #endif /* SUPPORT_UCP */
 #define UCD_SCRIPT(ch)    GET_UCD(ch)->script  
 #define UCD_CATEGORY(ch)  ucp_gentype[UCD_CHARTYPE(ch)]  
 #define UCD_OTHERCASE(ch) (ch + GET_UCD(ch)->other_case)  
2804    
2805  #endif  #endif
2806    

Legend:
Removed from v.349  
changed lines
  Added in v.1688

  ViewVC Help
Powered by ViewVC 1.1.5