1
/* $Cambridge: exim/exim-src/src/pcre/pcre_internal.h,v 1.1 2005/06/15 08:57:10 ph10 Exp $ */
3
/*************************************************
4
* Perl-Compatible Regular Expressions *
5
*************************************************/
8
/* PCRE is a library of functions to support regular expressions whose syntax
9
and semantics are as close as possible to those of the Perl 5 language.
11
Written by Philip Hazel
12
Copyright (c) 1997-2005 University of Cambridge
14
-----------------------------------------------------------------------------
15
Redistribution and use in source and binary forms, with or without
16
modification, are permitted provided that the following conditions are met:
18
* Redistributions of source code must retain the above copyright notice,
19
this list of conditions and the following disclaimer.
21
* Redistributions in binary form must reproduce the above copyright
22
notice, this list of conditions and the following disclaimer in the
23
documentation and/or other materials provided with the distribution.
25
* Neither the name of the University of Cambridge nor the names of its
26
contributors may be used to endorse or promote products derived from
27
this software without specific prior written permission.
29
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
30
AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
31
IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
32
ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
33
LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
34
CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
35
SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
36
INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
37
CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
38
ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
39
POSSIBILITY OF SUCH DAMAGE.
40
-----------------------------------------------------------------------------
43
/* This header contains definitions that are shared between the different
44
modules, but which are not relevant to the exported API. This includes some
45
functions whose names all begin with "_pcre_". */
48
/* Define DEBUG to get debugging output on stdout. */
54
/* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef
55
inline, and there are *still* stupid compilers about that don't like indented
56
pre-processor statements, or at least there were when I first wrote this. After
57
all, it had only been about 10 years then... */
60
#define DPRINTF(p) printf p
62
#define DPRINTF(p) /*nothing*/
66
/* Get the definitions provided by running "configure" */
70
/* Standard C headers plus the external interface definition. The only time
71
setjmp and stdarg are used is when NO_RECURSE is set. */
83
#define PCRE_DEFINITION /* Win32 __declspec(export) trigger for .dll */
86
/* We need to have types that specify unsigned 16-bit and 32-bit integers. We
87
cannot determine these outside the compilation (e.g. by running a program as
88
part of "configure") because PCRE is often cross-compiled for use on other
89
systems. Instead we make use of the maximum sizes that are available at
90
preprocessor time in standard C environments. */
92
#if USHRT_MAX == 65535
93
typedef unsigned short pcre_uint16;
94
#elif UINT_MAX == 65535
95
typedef unsigned int pcre_uint16;
97
#error Cannot determine a type for 16-bit unsigned integers
100
#if UINT_MAX == 4294967295
101
typedef unsigned int pcre_uint32;
102
#elif ULONG_MAX == 4294967295
103
typedef unsigned long int pcre_uint32;
105
#error Cannot determine a type for 32-bit unsigned integers
108
/* All character handling must be done as unsigned characters. Otherwise there
109
are problems with top-bit-set characters and functions such as isspace().
110
However, we leave the interface to the outside world as char *, because that
111
should make things easier for callers. We define a short type for unsigned char
112
to save lots of typing. I tried "uchar", but it causes problems on Digital
113
Unix, where it is defined in sys/types, so use "uschar" instead. */
115
typedef unsigned char uschar;
117
/* Include the public PCRE header */
121
/* Include the (copy of) the public ucp header, changing the external name into
122
a private one. This does no harm, even if we aren't compiling UCP support. */
124
#define ucp_findchar _pcre_ucp_findchar
127
/* When compiling for use with the Virtual Pascal compiler, these functions
128
need to have their names changed. PCRE must be compiled with the -DVPCOMPAT
129
option on the command line. */
132
#define strncmp(s1,s2,m) _strncmp(s1,s2,m)
133
#define memcpy(d,s,n) _memcpy(d,s,n)
134
#define memmove(d,s,n) _memmove(d,s,n)
135
#define memset(s,c,n) _memset(s,c,n)
138
/* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
139
define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY
140
is set. Otherwise, include an emulating function for those systems that have
141
neither (there some non-Unix environments where this is the case). This assumes
142
that all calls to memmove are moving strings upwards in store, which is the
146
#undef memmove /* some systems may have a macro */
148
#define memmove(a, b, c) bcopy(b, a, c)
149
#else /* HAVE_BCOPY */
151
pcre_memmove(unsigned char *dest, const unsigned char *src, size_t n)
156
for (i = 0; i < n; ++i) *(--dest) = *(--src);
158
#define memmove(a, b, c) pcre_memmove(a, b, c)
159
#endif /* not HAVE_BCOPY */
160
#endif /* not HAVE_MEMMOVE */
161
#endif /* not VPCOMPAT */
164
/* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored
165
in big-endian order) by default. These are used, for example, to link from the
166
start of a subpattern to its alternatives and its end. The use of 2 bytes per
167
offset limits the size of the compiled regex to around 64K, which is big enough
168
for almost everybody. However, I received a request for an even bigger limit.
169
For this reason, and also to make the code easier to maintain, the storing and
170
loading of offsets from the byte string is now handled by the macros that are
173
The macros are controlled by the value of LINK_SIZE. This defaults to 2 in
174
the config.h file, but can be overridden by using -D on the command line. This
175
is automated on Unix systems via the "configure" command. */
181
(a[(n)+1] = (d) & 255)
184
(((a)[n] << 8) | (a)[(n)+1])
186
#define MAX_PATTERN_SIZE (1 << 16)
192
(a[n] = (d) >> 16), \
193
(a[(n)+1] = (d) >> 8), \
194
(a[(n)+2] = (d) & 255)
197
(((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2])
199
#define MAX_PATTERN_SIZE (1 << 24)
205
(a[n] = (d) >> 24), \
206
(a[(n)+1] = (d) >> 16), \
207
(a[(n)+2] = (d) >> 8), \
208
(a[(n)+3] = (d) & 255)
211
(((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3])
213
#define MAX_PATTERN_SIZE (1 << 30) /* Keep it positive */
217
#error LINK_SIZE must be either 2, 3, or 4
221
/* Convenience macro defined in terms of the others */
223
#define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE
226
/* PCRE uses some other 2-byte quantities that do not change when the size of
227
offsets changes. There are used for repeat counts and for other things such as
228
capturing parenthesis numbers in back references. */
230
#define PUT2(a,n,d) \
235
(((a)[n] << 8) | (a)[(n)+1])
237
#define PUT2INC(a,n,d) PUT2(a,n,d), a += 2
240
/* When UTF-8 encoding is being used, a character is no longer just a single
241
byte. The macros for character handling generate simple sequences when used in
242
byte-mode, and more complicated ones for UTF-8 characters. */
245
#define GETCHAR(c, eptr) c = *eptr;
246
#define GETCHARTEST(c, eptr) c = *eptr;
247
#define GETCHARINC(c, eptr) c = *eptr++;
248
#define GETCHARINCTEST(c, eptr) c = *eptr++;
249
#define GETCHARLEN(c, eptr, len) c = *eptr;
250
#define BACKCHAR(eptr)
252
#else /* SUPPORT_UTF8 */
254
/* Get the next UTF-8 character, not advancing the pointer. This is called when
255
we know we are in UTF-8 mode. */
257
#define GETCHAR(c, eptr) \
259
if ((c & 0xc0) == 0xc0) \
262
int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
264
c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
265
for (gcii = 1; gcii <= gcaa; gcii++) \
268
c |= (eptr[gcii] & 0x3f) << gcss; \
272
/* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
275
#define GETCHARTEST(c, eptr) \
277
if (utf8 && (c & 0xc0) == 0xc0) \
280
int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
282
c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
283
for (gcii = 1; gcii <= gcaa; gcii++) \
286
c |= (eptr[gcii] & 0x3f) << gcss; \
290
/* Get the next UTF-8 character, advancing the pointer. This is called when we
291
know we are in UTF-8 mode. */
293
#define GETCHARINC(c, eptr) \
295
if ((c & 0xc0) == 0xc0) \
297
int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
299
c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
303
c |= (*eptr++ & 0x3f) << gcss; \
307
/* Get the next character, testing for UTF-8 mode, and advancing the pointer */
309
#define GETCHARINCTEST(c, eptr) \
311
if (utf8 && (c & 0xc0) == 0xc0) \
313
int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
315
c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
319
c |= (*eptr++ & 0x3f) << gcss; \
323
/* Get the next UTF-8 character, not advancing the pointer, incrementing length
324
if there are extra bytes. This is called when we know we are in UTF-8 mode. */
326
#define GETCHARLEN(c, eptr, len) \
328
if ((c & 0xc0) == 0xc0) \
331
int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
333
c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
334
for (gcii = 1; gcii <= gcaa; gcii++) \
337
c |= (eptr[gcii] & 0x3f) << gcss; \
342
/* If the pointer is not at the start of a character, move it back until
343
it is. Called only in UTF-8 mode. */
345
#define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--;
350
/* In case there is no definition of offsetof() provided - though any proper
351
Standard C system should have one. */
354
#define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))
358
/* These are the public options that can change during matching. */
360
#define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)
362
/* Private options flags start at the most significant end of the four bytes,
363
but skip the top bit so we can use ints for convenience without getting tangled
364
with negative values. The public options defined in pcre.h start at the least
365
significant end. Make sure they don't overlap! */
367
#define PCRE_FIRSTSET 0x40000000 /* first_byte is set */
368
#define PCRE_REQCHSET 0x20000000 /* req_byte is set */
369
#define PCRE_STARTLINE 0x10000000 /* start after \n for multiline */
370
#define PCRE_ICHANGED 0x08000000 /* i option changes within regex */
371
#define PCRE_NOPARTIAL 0x04000000 /* can't use partial with this regex */
373
/* Options for the "extra" block produced by pcre_study(). */
375
#define PCRE_STUDY_MAPPED 0x01 /* a map of starting chars exists */
377
/* Masks for identifying the public options that are permitted at compile
378
time, run time, or study time, respectively. */
380
#define PUBLIC_OPTIONS \
381
(PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
382
PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
383
PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE)
385
#define PUBLIC_EXEC_OPTIONS \
386
(PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
389
#define PUBLIC_DFA_EXEC_OPTIONS \
390
(PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
391
PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART)
393
#define PUBLIC_STUDY_OPTIONS 0 /* None defined */
395
/* Magic number to provide a small check against being handed junk. Also used
396
to detect whether a pattern was compiled on a host of different endianness. */
398
#define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */
400
/* Negative values for the firstchar and reqchar variables */
402
#define REQ_UNSET (-2)
403
#define REQ_NONE (-1)
405
/* The maximum remaining length of subject we are prepared to search for a
408
#define REQ_BYTE_MAX 1000
410
/* Flags added to firstbyte or reqbyte; a "non-literal" item is either a
411
variable-length repeat, or a anything other than literal characters. */
413
#define REQ_CASELESS 0x0100 /* indicates caselessness */
414
#define REQ_VARY 0x0200 /* reqbyte followed non-literal item */
416
/* Miscellaneous definitions */
423
/* Escape items that are just an encoding of a particular data value. Note that
424
ESC_n is defined as yet another macro, which is set in config.h to either \n
425
(the default) or \r (which some people want). */
436
#define ESC_n NEWLINE
443
/* We can't officially use ESC_t because it is a POSIX reserved identifier
444
(presumably because of all the others like size_t). */
450
/* These are escaped items that aren't just an encoding of a particular data
451
value such as \n. They must have non-zero values, as check_escape() returns
452
their negation. Also, they must appear in the same order as in the opcode
453
definitions below, up to ESC_z. There's a dummy for OP_ANY because it
454
corresponds to "." rather than an escape sequence. The final one must be
455
ESC_REF as subsequent values are used for \1, \2, \3, etc. There is are two
456
tests in the code for an escape greater than ESC_b and less than ESC_Z to
457
detect the types that may be repeated. These are the types that consume
458
characters. If any new escapes are put in between that don't consume a
459
character, that code will have to change. */
461
enum { ESC_A = 1, ESC_G, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W,
462
ESC_w, ESC_dum1, ESC_C, ESC_P, ESC_p, ESC_X, ESC_Z, ESC_z, ESC_E,
465
/* Flag bits and data types for the extended class (OP_XCLASS) for classes that
466
contain UTF-8 characters with values greater than 255. */
468
#define XCL_NOT 0x01 /* Flag: this is a negative class */
469
#define XCL_MAP 0x02 /* Flag: a 32-byte map is present */
471
#define XCL_END 0 /* Marks end of individual items */
472
#define XCL_SINGLE 1 /* Single item (one multibyte char) follows */
473
#define XCL_RANGE 2 /* A range (two multibyte chars) follows */
474
#define XCL_PROP 3 /* Unicode property (one property code) follows */
475
#define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */
478
/* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
479
that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
480
OP_EOD must correspond in order to the list of escapes immediately above.
481
Note that whenever this list is updated, the two macro definitions that follow
482
must also be updated to match. */
485
OP_END, /* 0 End of pattern */
487
/* Values corresponding to backslashed metacharacters */
489
OP_SOD, /* 1 Start of data: \A */
490
OP_SOM, /* 2 Start of match (subject + offset): \G */
491
OP_NOT_WORD_BOUNDARY, /* 3 \B */
492
OP_WORD_BOUNDARY, /* 4 \b */
493
OP_NOT_DIGIT, /* 5 \D */
495
OP_NOT_WHITESPACE, /* 7 \S */
496
OP_WHITESPACE, /* 8 \s */
497
OP_NOT_WORDCHAR, /* 9 \W */
498
OP_WORDCHAR, /* 10 \w */
499
OP_ANY, /* 11 Match any character */
500
OP_ANYBYTE, /* 12 Match any byte (\C); different to OP_ANY for UTF-8 */
501
OP_NOTPROP, /* 13 \P (not Unicode property) */
502
OP_PROP, /* 14 \p (Unicode property) */
503
OP_EXTUNI, /* 15 \X (extended Unicode sequence */
504
OP_EODN, /* 16 End of data or \n at end of data: \Z. */
505
OP_EOD, /* 17 End of data: \z */
507
OP_OPT, /* 18 Set runtime options */
508
OP_CIRC, /* 19 Start of line - varies with multiline switch */
509
OP_DOLL, /* 20 End of line - varies with multiline switch */
510
OP_CHAR, /* 21 Match one character, casefully */
511
OP_CHARNC, /* 22 Match one character, caselessly */
512
OP_NOT, /* 23 Match anything but the following char */
514
OP_STAR, /* 24 The maximizing and minimizing versions of */
515
OP_MINSTAR, /* 25 all these opcodes must come in pairs, with */
516
OP_PLUS, /* 26 the minimizing one second. */
517
OP_MINPLUS, /* 27 This first set applies to single characters */
519
OP_MINQUERY, /* 29 */
520
OP_UPTO, /* 30 From 0 to n matches */
522
OP_EXACT, /* 32 Exactly n matches */
524
OP_NOTSTAR, /* 33 The maximizing and minimizing versions of */
525
OP_NOTMINSTAR, /* 34 all these opcodes must come in pairs, with */
526
OP_NOTPLUS, /* 35 the minimizing one second. */
527
OP_NOTMINPLUS, /* 36 This set applies to "not" single characters */
528
OP_NOTQUERY, /* 37 */
529
OP_NOTMINQUERY, /* 38 */
530
OP_NOTUPTO, /* 39 From 0 to n matches */
531
OP_NOTMINUPTO, /* 40 */
532
OP_NOTEXACT, /* 41 Exactly n matches */
534
OP_TYPESTAR, /* 42 The maximizing and minimizing versions of */
535
OP_TYPEMINSTAR, /* 43 all these opcodes must come in pairs, with */
536
OP_TYPEPLUS, /* 44 the minimizing one second. These codes must */
537
OP_TYPEMINPLUS, /* 45 be in exactly the same order as those above. */
538
OP_TYPEQUERY, /* 46 This set applies to character types such as \d */
539
OP_TYPEMINQUERY, /* 47 */
540
OP_TYPEUPTO, /* 48 From 0 to n matches */
541
OP_TYPEMINUPTO, /* 49 */
542
OP_TYPEEXACT, /* 50 Exactly n matches */
544
OP_CRSTAR, /* 51 The maximizing and minimizing versions of */
545
OP_CRMINSTAR, /* 52 all these opcodes must come in pairs, with */
546
OP_CRPLUS, /* 53 the minimizing one second. These codes must */
547
OP_CRMINPLUS, /* 54 be in exactly the same order as those above. */
548
OP_CRQUERY, /* 55 These are for character classes and back refs */
549
OP_CRMINQUERY, /* 56 */
550
OP_CRRANGE, /* 57 These are different to the three sets above. */
551
OP_CRMINRANGE, /* 58 */
553
OP_CLASS, /* 59 Match a character class, chars < 256 only */
554
OP_NCLASS, /* 60 Same, but the bitmap was created from a negative
555
class - the difference is relevant only when a UTF-8
556
character > 255 is encountered. */
558
OP_XCLASS, /* 61 Extended class for handling UTF-8 chars within the
559
class. This does both positive and negative. */
561
OP_REF, /* 62 Match a back reference */
562
OP_RECURSE, /* 63 Match a numbered subpattern (possibly recursive) */
563
OP_CALLOUT, /* 64 Call out to external function if provided */
565
OP_ALT, /* 65 Start of alternation */
566
OP_KET, /* 66 End of group that doesn't have an unbounded repeat */
567
OP_KETRMAX, /* 67 These two must remain together and in this */
568
OP_KETRMIN, /* 68 order. They are for groups the repeat for ever. */
570
/* The assertions must come before ONCE and COND */
572
OP_ASSERT, /* 69 Positive lookahead */
573
OP_ASSERT_NOT, /* 70 Negative lookahead */
574
OP_ASSERTBACK, /* 71 Positive lookbehind */
575
OP_ASSERTBACK_NOT, /* 72 Negative lookbehind */
576
OP_REVERSE, /* 73 Move pointer back - used in lookbehind assertions */
578
/* ONCE and COND must come after the assertions, with ONCE first, as there's
579
a test for >= ONCE for a subpattern that isn't an assertion. */
581
OP_ONCE, /* 74 Once matched, don't back up into the subpattern */
582
OP_COND, /* 75 Conditional group */
583
OP_CREF, /* 76 Used to hold an extraction string number (cond ref) */
585
OP_BRAZERO, /* 77 These two must remain together and in this */
586
OP_BRAMINZERO, /* 78 order. */
588
OP_BRANUMBER, /* 79 Used for extracting brackets whose number is greater
589
than can fit into an opcode. */
591
OP_BRA /* 80 This and greater values are used for brackets that
592
extract substrings up to EXTRACT_BASIC_MAX. After
593
that, use is made of OP_BRANUMBER. */
596
/* WARNING WARNING WARNING: There is an implicit assumption in pcre.c and
597
study.c that all opcodes are less than 128 in value. This makes handling UTF-8
598
character sequences easier. */
600
/* The highest extraction number before we have to start using additional
601
bytes. (Originally PCRE didn't have support for extraction counts highter than
602
this number.) The value is limited by the number of opcodes left after OP_BRA,
603
i.e. 255 - OP_BRA. We actually set it a bit lower to leave room for additional
606
#define EXTRACT_BASIC_MAX 100
609
/* This macro defines textual names for all the opcodes. These are used only
610
for debugging. The macro is referenced only in pcre_printint.c. */
612
#define OP_NAME_LIST \
613
"End", "\\A", "\\G", "\\B", "\\b", "\\D", "\\d", \
614
"\\S", "\\s", "\\W", "\\w", "Any", "Anybyte", \
615
"notprop", "prop", "extuni", \
617
"Opt", "^", "$", "char", "charnc", "not", \
618
"*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
619
"*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
620
"*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
621
"*", "*?", "+", "+?", "?", "??", "{", "{", \
622
"class", "nclass", "xclass", "Ref", "Recurse", "Callout", \
623
"Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \
624
"AssertB", "AssertB not", "Reverse", "Once", "Cond", "Cond ref",\
625
"Brazero", "Braminzero", "Branumber", "Bra"
628
/* This macro defines the length of fixed length operations in the compiled
629
regex. The lengths are used when searching for specific things, and also in the
630
debugging printing of a compiled regex. We use a macro so that it can be
631
defined close to the definitions of the opcodes themselves.
633
As things have been extended, some of these are no longer fixed lenths, but are
634
minima instead. For example, the length of a single-character repeat may vary
635
in UTF-8 mode. The code that uses this table must know about such things. */
639
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* \A, \G, \B, \B, \D, \d, \S, \s, \W, \w */ \
640
1, 1, /* Any, Anybyte */ \
641
2, 2, 1, /* NOTPROP, PROP, EXTUNI */ \
642
1, 1, 2, 1, 1, /* \Z, \z, Opt, ^, $ */ \
643
2, /* Char - the minimum length */ \
644
2, /* Charnc - the minimum length */ \
646
/* Positive single-char repeats ** These are */ \
647
2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \
648
4, 4, 4, /* upto, minupto, exact ** UTF-8 mode */ \
649
/* Negative single-char repeats - only for chars < 256 */ \
650
2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \
651
4, 4, 4, /* NOT upto, minupto, exact */ \
652
/* Positive type repeats */ \
653
2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \
654
4, 4, 4, /* Type upto, minupto, exact */ \
655
/* Character class & ref repeats */ \
656
1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \
657
5, 5, /* CRRANGE, CRMINRANGE */ \
660
0, /* XCLASS - variable length */ \
662
1+LINK_SIZE, /* RECURSE */ \
663
2+2*LINK_SIZE, /* CALLOUT */ \
664
1+LINK_SIZE, /* Alt */ \
665
1+LINK_SIZE, /* Ket */ \
666
1+LINK_SIZE, /* KetRmax */ \
667
1+LINK_SIZE, /* KetRmin */ \
668
1+LINK_SIZE, /* Assert */ \
669
1+LINK_SIZE, /* Assert not */ \
670
1+LINK_SIZE, /* Assert behind */ \
671
1+LINK_SIZE, /* Assert behind not */ \
672
1+LINK_SIZE, /* Reverse */ \
673
1+LINK_SIZE, /* Once */ \
674
1+LINK_SIZE, /* COND */ \
676
1, 1, /* BRAZERO, BRAMINZERO */ \
678
1+LINK_SIZE /* BRA */ \
681
/* A magic value for OP_CREF to indicate the "in recursion" condition. */
683
#define CREF_RECURSE 0xffff
685
/* Error code numbers. They are given names so that they can more easily be
688
enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9,
689
ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19,
690
ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29,
691
ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
692
ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47 };
694
/* The real format of the start of the pcre block; the index of names and the
695
code vector run on as long as necessary after the end. We store an explicit
696
offset to the name table so that if a regex is compiled on one host, saved, and
697
then run on another where the size of pointers is different, all might still
698
be well. For the case of compiled-on-4 and run-on-8, we include an extra
699
pointer that is always NULL. For future-proofing, a few dummy fields were
700
originally included - even though you can never get this planning right - but
701
there is only one left now.
704
Because people can now save and re-use compiled patterns, any additions to this
705
structure should be made at the end, and something earlier (e.g. a new
706
flag in the options or one of the dummy fields) should indicate that the new
707
fields are present. Currently PCRE always sets the dummy fields to zero.
711
typedef struct real_pcre {
712
pcre_uint32 magic_number;
713
pcre_uint32 size; /* Total that was malloced */
715
pcre_uint32 dummy1; /* For future use, maybe */
717
pcre_uint16 top_bracket;
718
pcre_uint16 top_backref;
719
pcre_uint16 first_byte;
720
pcre_uint16 req_byte;
721
pcre_uint16 name_table_offset; /* Offset to name table that follows */
722
pcre_uint16 name_entry_size; /* Size of any name items */
723
pcre_uint16 name_count; /* Number of name items */
724
pcre_uint16 ref_count; /* Reference count */
726
const unsigned char *tables; /* Pointer to tables or NULL for std */
727
const unsigned char *nullpad; /* NULL padding */
730
/* The format of the block used to store data from pcre_study(). The same
731
remark (see NOTE above) about extending this structure applies. */
733
typedef struct pcre_study_data {
734
pcre_uint32 size; /* Total that was malloced */
736
uschar start_bits[32];
739
/* Structure for passing "static" information around between the functions
740
doing the compiling, so that they are thread-safe. */
742
typedef struct compile_data {
743
const uschar *lcc; /* Points to lower casing table */
744
const uschar *fcc; /* Points to case-flipping table */
745
const uschar *cbits; /* Points to character type table */
746
const uschar *ctypes; /* Points to table of type maps */
747
const uschar *start_code; /* The start of the compiled code */
748
const uschar *start_pattern; /* The start of the pattern */
749
uschar *name_table; /* The name/number table */
750
int names_found; /* Number of entries so far */
751
int name_entry_size; /* Size of each entry */
752
int top_backref; /* Maximum back reference */
753
unsigned int backref_map; /* Bitmap of low back refs */
754
int req_varyopt; /* "After variable item" flag for reqbyte */
755
BOOL nopartial; /* Set TRUE if partial won't work */
758
/* Structure for maintaining a chain of pointers to the currently incomplete
759
branches, for testing for left recursion. */
761
typedef struct branch_chain {
762
struct branch_chain *outer;
766
/* Structure for items in a linked list that represents an explicit recursive
767
call within the pattern. */
769
typedef struct recursion_info {
770
struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
771
int group_num; /* Number of group that was called */
772
const uschar *after_call; /* "Return value": points after the call in the expr */
773
const uschar *save_start; /* Old value of md->start_match */
774
int *offset_save; /* Pointer to start of saved offsets */
775
int saved_max; /* Number of saved offsets */
778
/* When compiling in a mode that doesn't use recursive calls to match(),
779
a structure is used to remember local variables on the heap. It is defined in
780
pcre.c, close to the match() function, so that it is easy to keep it in step
781
with any changes of local variable. However, the pointer to the current frame
782
must be saved in some "static" place over a longjmp(). We declare the
783
structure here so that we can put a pointer in the match_data structure.
784
NOTE: This isn't used for a "normal" compilation of pcre. */
788
/* Structure for passing "static" information around between the functions
789
doing traditional NFA matching, so that they are thread-safe. */
791
typedef struct match_data {
792
unsigned long int match_call_count; /* As it says */
793
unsigned long int match_limit;/* As it says */
794
int *offset_vector; /* Offset vector */
795
int offset_end; /* One past the end */
796
int offset_max; /* The maximum usable for return data */
797
const uschar *lcc; /* Points to lower casing table */
798
const uschar *ctypes; /* Points to table of type maps */
799
BOOL offset_overflow; /* Set if too many extractions */
800
BOOL notbol; /* NOTBOL flag */
801
BOOL noteol; /* NOTEOL flag */
802
BOOL utf8; /* UTF8 flag */
803
BOOL endonly; /* Dollar not before final \n */
804
BOOL notempty; /* Empty string match not wanted */
805
BOOL partial; /* PARTIAL flag */
806
BOOL hitend; /* Hit the end of the subject at some point */
807
const uschar *start_code; /* For use when recursing */
808
const uschar *start_subject; /* Start of the subject string */
809
const uschar *end_subject; /* End of the subject string */
810
const uschar *start_match; /* Start of this match attempt */
811
const uschar *end_match_ptr; /* Subject position at end match */
812
int end_offset_top; /* Highwater mark at end of match */
813
int capture_last; /* Most recent capture number */
814
int start_offset; /* The start offset value */
815
recursion_info *recursive; /* Linked list of recursion data */
816
void *callout_data; /* To pass back to callouts */
817
struct heapframe *thisframe; /* Used only when compiling for no recursion */
820
/* A similar structure is used for the same purpose by the DFA matching
823
typedef struct dfa_match_data {
824
const uschar *start_code; /* Start of the compiled pattern */
825
const uschar *start_subject; /* Start of the subject string */
826
const uschar *end_subject; /* End of subject string */
827
const uschar *tables; /* Character tables */
828
int moptions; /* Match options */
829
int poptions; /* Pattern options */
830
void *callout_data; /* To pass back to callouts */
833
/* Bit definitions for entries in the pcre_ctypes table. */
835
#define ctype_space 0x01
836
#define ctype_letter 0x02
837
#define ctype_digit 0x04
838
#define ctype_xdigit 0x08
839
#define ctype_word 0x10 /* alphameric or '_' */
840
#define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
842
/* Offsets for the bitmap tables in pcre_cbits. Each table contains a set
843
of bits for a class map. Some classes are built by combining these tables. */
845
#define cbit_space 0 /* [:space:] or \s */
846
#define cbit_xdigit 32 /* [:xdigit:] */
847
#define cbit_digit 64 /* [:digit:] or \d */
848
#define cbit_upper 96 /* [:upper:] */
849
#define cbit_lower 128 /* [:lower:] */
850
#define cbit_word 160 /* [:word:] or \w */
851
#define cbit_graph 192 /* [:graph:] */
852
#define cbit_print 224 /* [:print:] */
853
#define cbit_punct 256 /* [:punct:] */
854
#define cbit_cntrl 288 /* [:cntrl:] */
855
#define cbit_length 320 /* Length of the cbits table */
857
/* Offsets of the various tables from the base tables pointer, and
861
#define fcc_offset 256
862
#define cbits_offset 512
863
#define ctypes_offset (cbits_offset + cbit_length)
864
#define tables_length (ctypes_offset + 256)
866
/* Layout of the UCP type table that translates property names into codes for
875
/* Internal shared data tables. These are tables that are used by more than one
876
of the exported public functions. They have to be "external" in the C sense,
877
but are not part of the PCRE public API. The data for these tables is in the
878
pcre_tables.c module. */
880
extern const int _pcre_utf8_table1[];
881
extern const int _pcre_utf8_table2[];
882
extern const int _pcre_utf8_table3[];
883
extern const uschar _pcre_utf8_table4[];
885
extern const int _pcre_utf8_table1_size;
887
extern const ucp_type_table _pcre_utt[];
888
extern const int _pcre_utt_size;
890
extern const uschar _pcre_default_tables[];
892
extern const uschar _pcre_OP_lengths[];
895
/* Internal shared functions. These are functions that are used by more than
896
one of the exported public functions. They have to be "external" in the C
897
sense, but are not part of the PCRE public API. */
899
extern int _pcre_ord2utf8(int, uschar *);
900
extern void _pcre_printint(pcre *, FILE *);
901
extern real_pcre * _pcre_try_flipped(const real_pcre *, real_pcre *,
902
const pcre_study_data *, pcre_study_data *);
903
extern int _pcre_ucp_findchar(const int, int *, int *);
904
extern int _pcre_valid_utf8(const uschar *, int);
905
extern BOOL _pcre_xclass(int, const uschar *);
907
/* End of pcre_internal.h */
added
removed
src/pcre/pcre_internal.h
