/[pcre]/code/trunk/pcre_internal.h
ViewVC logotype

Contents of /code/trunk/pcre_internal.h

Parent Directory Parent Directory | Revision Log Revision Log


Revision 85 - (show annotations) (download)
Sat Feb 24 21:41:13 2007 UTC (7 years, 5 months ago) by nigel
File MIME type: text/plain
File size: 36724 byte(s)
Load pcre-6.4 into code/trunk.

1 /*************************************************
2 * Perl-Compatible Regular Expressions *
3 *************************************************/
4
5
6 /* PCRE is a library of functions to support regular expressions whose syntax
7 and semantics are as close as possible to those of the Perl 5 language.
8
9 Written by Philip Hazel
10 Copyright (c) 1997-2005 University of Cambridge
11
12 -----------------------------------------------------------------------------
13 Redistribution and use in source and binary forms, with or without
14 modification, are permitted provided that the following conditions are met:
15
16 * Redistributions of source code must retain the above copyright notice,
17 this list of conditions and the following disclaimer.
18
19 * Redistributions in binary form must reproduce the above copyright
20 notice, this list of conditions and the following disclaimer in the
21 documentation and/or other materials provided with the distribution.
22
23 * Neither the name of the University of Cambridge nor the names of its
24 contributors may be used to endorse or promote products derived from
25 this software without specific prior written permission.
26
27 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
28 AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
29 IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
30 ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
31 LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32 CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33 SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34 INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35 CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36 ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37 POSSIBILITY OF SUCH DAMAGE.
38 -----------------------------------------------------------------------------
39 */
40
41 /* This header contains definitions that are shared between the different
42 modules, but which are not relevant to the exported API. This includes some
43 functions whose names all begin with "_pcre_". */
44
45 #ifndef PCRE_INTERNAL_H
46 #define PCRE_INTERNAL_H
47
48 /* Define DEBUG to get debugging output on stdout. */
49
50 #if 0
51 #define DEBUG
52 #endif
53
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... */
58
59 #ifdef DEBUG
60 #define DPRINTF(p) printf p
61 #else
62 #define DPRINTF(p) /*nothing*/
63 #endif
64
65
66 /* Get the definitions provided by running "configure" */
67
68 #include "config.h"
69
70 /* Standard C headers plus the external interface definition. The only time
71 setjmp and stdarg are used is when NO_RECURSE is set. */
72
73 #include <ctype.h>
74 #include <limits.h>
75 #include <setjmp.h>
76 #include <stdarg.h>
77 #include <stddef.h>
78 #include <stdio.h>
79 #include <stdlib.h>
80 #include <string.h>
81
82 #ifndef PCRE_SPY
83 #define PCRE_DEFINITION /* Win32 __declspec(export) trigger for .dll */
84 #endif
85
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. */
91
92 #if USHRT_MAX == 65535
93 typedef unsigned short pcre_uint16;
94 #elif UINT_MAX == 65535
95 typedef unsigned int pcre_uint16;
96 #else
97 #error Cannot determine a type for 16-bit unsigned integers
98 #endif
99
100 #if UINT_MAX == 4294967295
101 typedef unsigned int pcre_uint32;
102 #elif ULONG_MAX == 4294967295
103 typedef unsigned long int pcre_uint32;
104 #else
105 #error Cannot determine a type for 32-bit unsigned integers
106 #endif
107
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. */
114
115 typedef unsigned char uschar;
116
117 /* Include the public PCRE header and the definitions of UCP character property
118 values. */
119
120 #include "pcre.h"
121 #include "ucp.h"
122
123 /* When compiling for use with the Virtual Pascal compiler, these functions
124 need to have their names changed. PCRE must be compiled with the -DVPCOMPAT
125 option on the command line. */
126
127 #ifdef VPCOMPAT
128 #define strncmp(s1,s2,m) _strncmp(s1,s2,m)
129 #define memcpy(d,s,n) _memcpy(d,s,n)
130 #define memmove(d,s,n) _memmove(d,s,n)
131 #define memset(s,c,n) _memset(s,c,n)
132 #else /* VPCOMPAT */
133
134 /* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
135 define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY
136 is set. Otherwise, include an emulating function for those systems that have
137 neither (there some non-Unix environments where this is the case). This assumes
138 that all calls to memmove are moving strings upwards in store, which is the
139 case in PCRE. */
140
141 #if ! HAVE_MEMMOVE
142 #undef memmove /* some systems may have a macro */
143 #if HAVE_BCOPY
144 #define memmove(a, b, c) bcopy(b, a, c)
145 #else /* HAVE_BCOPY */
146 void *
147 pcre_memmove(unsigned char *dest, const unsigned char *src, size_t n)
148 {
149 int i;
150 dest += n;
151 src += n;
152 for (i = 0; i < n; ++i) *(--dest) = *(--src);
153 }
154 #define memmove(a, b, c) pcre_memmove(a, b, c)
155 #endif /* not HAVE_BCOPY */
156 #endif /* not HAVE_MEMMOVE */
157 #endif /* not VPCOMPAT */
158
159
160 /* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored
161 in big-endian order) by default. These are used, for example, to link from the
162 start of a subpattern to its alternatives and its end. The use of 2 bytes per
163 offset limits the size of the compiled regex to around 64K, which is big enough
164 for almost everybody. However, I received a request for an even bigger limit.
165 For this reason, and also to make the code easier to maintain, the storing and
166 loading of offsets from the byte string is now handled by the macros that are
167 defined here.
168
169 The macros are controlled by the value of LINK_SIZE. This defaults to 2 in
170 the config.h file, but can be overridden by using -D on the command line. This
171 is automated on Unix systems via the "configure" command. */
172
173 #if LINK_SIZE == 2
174
175 #define PUT(a,n,d) \
176 (a[n] = (d) >> 8), \
177 (a[(n)+1] = (d) & 255)
178
179 #define GET(a,n) \
180 (((a)[n] << 8) | (a)[(n)+1])
181
182 #define MAX_PATTERN_SIZE (1 << 16)
183
184
185 #elif LINK_SIZE == 3
186
187 #define PUT(a,n,d) \
188 (a[n] = (d) >> 16), \
189 (a[(n)+1] = (d) >> 8), \
190 (a[(n)+2] = (d) & 255)
191
192 #define GET(a,n) \
193 (((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2])
194
195 #define MAX_PATTERN_SIZE (1 << 24)
196
197
198 #elif LINK_SIZE == 4
199
200 #define PUT(a,n,d) \
201 (a[n] = (d) >> 24), \
202 (a[(n)+1] = (d) >> 16), \
203 (a[(n)+2] = (d) >> 8), \
204 (a[(n)+3] = (d) & 255)
205
206 #define GET(a,n) \
207 (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3])
208
209 #define MAX_PATTERN_SIZE (1 << 30) /* Keep it positive */
210
211
212 #else
213 #error LINK_SIZE must be either 2, 3, or 4
214 #endif
215
216
217 /* Convenience macro defined in terms of the others */
218
219 #define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE
220
221
222 /* PCRE uses some other 2-byte quantities that do not change when the size of
223 offsets changes. There are used for repeat counts and for other things such as
224 capturing parenthesis numbers in back references. */
225
226 #define PUT2(a,n,d) \
227 a[n] = (d) >> 8; \
228 a[(n)+1] = (d) & 255
229
230 #define GET2(a,n) \
231 (((a)[n] << 8) | (a)[(n)+1])
232
233 #define PUT2INC(a,n,d) PUT2(a,n,d), a += 2
234
235
236 /* When UTF-8 encoding is being used, a character is no longer just a single
237 byte. The macros for character handling generate simple sequences when used in
238 byte-mode, and more complicated ones for UTF-8 characters. */
239
240 #ifndef SUPPORT_UTF8
241 #define GETCHAR(c, eptr) c = *eptr;
242 #define GETCHARTEST(c, eptr) c = *eptr;
243 #define GETCHARINC(c, eptr) c = *eptr++;
244 #define GETCHARINCTEST(c, eptr) c = *eptr++;
245 #define GETCHARLEN(c, eptr, len) c = *eptr;
246 #define BACKCHAR(eptr)
247
248 #else /* SUPPORT_UTF8 */
249
250 /* Get the next UTF-8 character, not advancing the pointer. This is called when
251 we know we are in UTF-8 mode. */
252
253 #define GETCHAR(c, eptr) \
254 c = *eptr; \
255 if ((c & 0xc0) == 0xc0) \
256 { \
257 int gcii; \
258 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
259 int gcss = 6*gcaa; \
260 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
261 for (gcii = 1; gcii <= gcaa; gcii++) \
262 { \
263 gcss -= 6; \
264 c |= (eptr[gcii] & 0x3f) << gcss; \
265 } \
266 }
267
268 /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
269 pointer. */
270
271 #define GETCHARTEST(c, eptr) \
272 c = *eptr; \
273 if (utf8 && (c & 0xc0) == 0xc0) \
274 { \
275 int gcii; \
276 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
277 int gcss = 6*gcaa; \
278 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
279 for (gcii = 1; gcii <= gcaa; gcii++) \
280 { \
281 gcss -= 6; \
282 c |= (eptr[gcii] & 0x3f) << gcss; \
283 } \
284 }
285
286 /* Get the next UTF-8 character, advancing the pointer. This is called when we
287 know we are in UTF-8 mode. */
288
289 #define GETCHARINC(c, eptr) \
290 c = *eptr++; \
291 if ((c & 0xc0) == 0xc0) \
292 { \
293 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
294 int gcss = 6*gcaa; \
295 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
296 while (gcaa-- > 0) \
297 { \
298 gcss -= 6; \
299 c |= (*eptr++ & 0x3f) << gcss; \
300 } \
301 }
302
303 /* Get the next character, testing for UTF-8 mode, and advancing the pointer */
304
305 #define GETCHARINCTEST(c, eptr) \
306 c = *eptr++; \
307 if (utf8 && (c & 0xc0) == 0xc0) \
308 { \
309 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
310 int gcss = 6*gcaa; \
311 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
312 while (gcaa-- > 0) \
313 { \
314 gcss -= 6; \
315 c |= (*eptr++ & 0x3f) << gcss; \
316 } \
317 }
318
319 /* Get the next UTF-8 character, not advancing the pointer, incrementing length
320 if there are extra bytes. This is called when we know we are in UTF-8 mode. */
321
322 #define GETCHARLEN(c, eptr, len) \
323 c = *eptr; \
324 if ((c & 0xc0) == 0xc0) \
325 { \
326 int gcii; \
327 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
328 int gcss = 6*gcaa; \
329 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
330 for (gcii = 1; gcii <= gcaa; gcii++) \
331 { \
332 gcss -= 6; \
333 c |= (eptr[gcii] & 0x3f) << gcss; \
334 } \
335 len += gcaa; \
336 }
337
338 /* If the pointer is not at the start of a character, move it back until
339 it is. Called only in UTF-8 mode. */
340
341 #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--;
342
343 #endif
344
345
346 /* In case there is no definition of offsetof() provided - though any proper
347 Standard C system should have one. */
348
349 #ifndef offsetof
350 #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))
351 #endif
352
353
354 /* These are the public options that can change during matching. */
355
356 #define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)
357
358 /* Private options flags start at the most significant end of the four bytes,
359 but skip the top bit so we can use ints for convenience without getting tangled
360 with negative values. The public options defined in pcre.h start at the least
361 significant end. Make sure they don't overlap! */
362
363 #define PCRE_FIRSTSET 0x40000000 /* first_byte is set */
364 #define PCRE_REQCHSET 0x20000000 /* req_byte is set */
365 #define PCRE_STARTLINE 0x10000000 /* start after \n for multiline */
366 #define PCRE_ICHANGED 0x08000000 /* i option changes within regex */
367 #define PCRE_NOPARTIAL 0x04000000 /* can't use partial with this regex */
368
369 /* Options for the "extra" block produced by pcre_study(). */
370
371 #define PCRE_STUDY_MAPPED 0x01 /* a map of starting chars exists */
372
373 /* Masks for identifying the public options that are permitted at compile
374 time, run time, or study time, respectively. */
375
376 #define PUBLIC_OPTIONS \
377 (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
378 PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
379 PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE)
380
381 #define PUBLIC_EXEC_OPTIONS \
382 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
383 PCRE_PARTIAL)
384
385 #define PUBLIC_DFA_EXEC_OPTIONS \
386 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
387 PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART)
388
389 #define PUBLIC_STUDY_OPTIONS 0 /* None defined */
390
391 /* Magic number to provide a small check against being handed junk. Also used
392 to detect whether a pattern was compiled on a host of different endianness. */
393
394 #define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */
395
396 /* Negative values for the firstchar and reqchar variables */
397
398 #define REQ_UNSET (-2)
399 #define REQ_NONE (-1)
400
401 /* The maximum remaining length of subject we are prepared to search for a
402 req_byte match. */
403
404 #define REQ_BYTE_MAX 1000
405
406 /* Flags added to firstbyte or reqbyte; a "non-literal" item is either a
407 variable-length repeat, or a anything other than literal characters. */
408
409 #define REQ_CASELESS 0x0100 /* indicates caselessness */
410 #define REQ_VARY 0x0200 /* reqbyte followed non-literal item */
411
412 /* Miscellaneous definitions */
413
414 typedef int BOOL;
415
416 #define FALSE 0
417 #define TRUE 1
418
419 /* Escape items that are just an encoding of a particular data value. Note that
420 ESC_n is defined as yet another macro, which is set in config.h to either \n
421 (the default) or \r (which some people want). */
422
423 #ifndef ESC_e
424 #define ESC_e 27
425 #endif
426
427 #ifndef ESC_f
428 #define ESC_f '\f'
429 #endif
430
431 #ifndef ESC_n
432 #define ESC_n NEWLINE
433 #endif
434
435 #ifndef ESC_r
436 #define ESC_r '\r'
437 #endif
438
439 /* We can't officially use ESC_t because it is a POSIX reserved identifier
440 (presumably because of all the others like size_t). */
441
442 #ifndef ESC_tee
443 #define ESC_tee '\t'
444 #endif
445
446 /* These are escaped items that aren't just an encoding of a particular data
447 value such as \n. They must have non-zero values, as check_escape() returns
448 their negation. Also, they must appear in the same order as in the opcode
449 definitions below, up to ESC_z. There's a dummy for OP_ANY because it
450 corresponds to "." rather than an escape sequence. The final one must be
451 ESC_REF as subsequent values are used for \1, \2, \3, etc. There is are two
452 tests in the code for an escape greater than ESC_b and less than ESC_Z to
453 detect the types that may be repeated. These are the types that consume
454 characters. If any new escapes are put in between that don't consume a
455 character, that code will have to change. */
456
457 enum { ESC_A = 1, ESC_G, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W,
458 ESC_w, ESC_dum1, ESC_C, ESC_P, ESC_p, ESC_X, ESC_Z, ESC_z, ESC_E,
459 ESC_Q, ESC_REF };
460
461 /* Flag bits and data types for the extended class (OP_XCLASS) for classes that
462 contain UTF-8 characters with values greater than 255. */
463
464 #define XCL_NOT 0x01 /* Flag: this is a negative class */
465 #define XCL_MAP 0x02 /* Flag: a 32-byte map is present */
466
467 #define XCL_END 0 /* Marks end of individual items */
468 #define XCL_SINGLE 1 /* Single item (one multibyte char) follows */
469 #define XCL_RANGE 2 /* A range (two multibyte chars) follows */
470 #define XCL_PROP 3 /* Unicode property (one property code) follows */
471 #define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */
472
473
474 /* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
475 that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
476 OP_EOD must correspond in order to the list of escapes immediately above.
477 Note that whenever this list is updated, the two macro definitions that follow
478 must also be updated to match. */
479
480 enum {
481 OP_END, /* 0 End of pattern */
482
483 /* Values corresponding to backslashed metacharacters */
484
485 OP_SOD, /* 1 Start of data: \A */
486 OP_SOM, /* 2 Start of match (subject + offset): \G */
487 OP_NOT_WORD_BOUNDARY, /* 3 \B */
488 OP_WORD_BOUNDARY, /* 4 \b */
489 OP_NOT_DIGIT, /* 5 \D */
490 OP_DIGIT, /* 6 \d */
491 OP_NOT_WHITESPACE, /* 7 \S */
492 OP_WHITESPACE, /* 8 \s */
493 OP_NOT_WORDCHAR, /* 9 \W */
494 OP_WORDCHAR, /* 10 \w */
495 OP_ANY, /* 11 Match any character */
496 OP_ANYBYTE, /* 12 Match any byte (\C); different to OP_ANY for UTF-8 */
497 OP_NOTPROP, /* 13 \P (not Unicode property) */
498 OP_PROP, /* 14 \p (Unicode property) */
499 OP_EXTUNI, /* 15 \X (extended Unicode sequence */
500 OP_EODN, /* 16 End of data or \n at end of data: \Z. */
501 OP_EOD, /* 17 End of data: \z */
502
503 OP_OPT, /* 18 Set runtime options */
504 OP_CIRC, /* 19 Start of line - varies with multiline switch */
505 OP_DOLL, /* 20 End of line - varies with multiline switch */
506 OP_CHAR, /* 21 Match one character, casefully */
507 OP_CHARNC, /* 22 Match one character, caselessly */
508 OP_NOT, /* 23 Match anything but the following char */
509
510 OP_STAR, /* 24 The maximizing and minimizing versions of */
511 OP_MINSTAR, /* 25 all these opcodes must come in pairs, with */
512 OP_PLUS, /* 26 the minimizing one second. */
513 OP_MINPLUS, /* 27 This first set applies to single characters */
514 OP_QUERY, /* 28 */
515 OP_MINQUERY, /* 29 */
516 OP_UPTO, /* 30 From 0 to n matches */
517 OP_MINUPTO, /* 31 */
518 OP_EXACT, /* 32 Exactly n matches */
519
520 OP_NOTSTAR, /* 33 The maximizing and minimizing versions of */
521 OP_NOTMINSTAR, /* 34 all these opcodes must come in pairs, with */
522 OP_NOTPLUS, /* 35 the minimizing one second. */
523 OP_NOTMINPLUS, /* 36 This set applies to "not" single characters */
524 OP_NOTQUERY, /* 37 */
525 OP_NOTMINQUERY, /* 38 */
526 OP_NOTUPTO, /* 39 From 0 to n matches */
527 OP_NOTMINUPTO, /* 40 */
528 OP_NOTEXACT, /* 41 Exactly n matches */
529
530 OP_TYPESTAR, /* 42 The maximizing and minimizing versions of */
531 OP_TYPEMINSTAR, /* 43 all these opcodes must come in pairs, with */
532 OP_TYPEPLUS, /* 44 the minimizing one second. These codes must */
533 OP_TYPEMINPLUS, /* 45 be in exactly the same order as those above. */
534 OP_TYPEQUERY, /* 46 This set applies to character types such as \d */
535 OP_TYPEMINQUERY, /* 47 */
536 OP_TYPEUPTO, /* 48 From 0 to n matches */
537 OP_TYPEMINUPTO, /* 49 */
538 OP_TYPEEXACT, /* 50 Exactly n matches */
539
540 OP_CRSTAR, /* 51 The maximizing and minimizing versions of */
541 OP_CRMINSTAR, /* 52 all these opcodes must come in pairs, with */
542 OP_CRPLUS, /* 53 the minimizing one second. These codes must */
543 OP_CRMINPLUS, /* 54 be in exactly the same order as those above. */
544 OP_CRQUERY, /* 55 These are for character classes and back refs */
545 OP_CRMINQUERY, /* 56 */
546 OP_CRRANGE, /* 57 These are different to the three sets above. */
547 OP_CRMINRANGE, /* 58 */
548
549 OP_CLASS, /* 59 Match a character class, chars < 256 only */
550 OP_NCLASS, /* 60 Same, but the bitmap was created from a negative
551 class - the difference is relevant only when a UTF-8
552 character > 255 is encountered. */
553
554 OP_XCLASS, /* 61 Extended class for handling UTF-8 chars within the
555 class. This does both positive and negative. */
556
557 OP_REF, /* 62 Match a back reference */
558 OP_RECURSE, /* 63 Match a numbered subpattern (possibly recursive) */
559 OP_CALLOUT, /* 64 Call out to external function if provided */
560
561 OP_ALT, /* 65 Start of alternation */
562 OP_KET, /* 66 End of group that doesn't have an unbounded repeat */
563 OP_KETRMAX, /* 67 These two must remain together and in this */
564 OP_KETRMIN, /* 68 order. They are for groups the repeat for ever. */
565
566 /* The assertions must come before ONCE and COND */
567
568 OP_ASSERT, /* 69 Positive lookahead */
569 OP_ASSERT_NOT, /* 70 Negative lookahead */
570 OP_ASSERTBACK, /* 71 Positive lookbehind */
571 OP_ASSERTBACK_NOT, /* 72 Negative lookbehind */
572 OP_REVERSE, /* 73 Move pointer back - used in lookbehind assertions */
573
574 /* ONCE and COND must come after the assertions, with ONCE first, as there's
575 a test for >= ONCE for a subpattern that isn't an assertion. */
576
577 OP_ONCE, /* 74 Once matched, don't back up into the subpattern */
578 OP_COND, /* 75 Conditional group */
579 OP_CREF, /* 76 Used to hold an extraction string number (cond ref) */
580
581 OP_BRAZERO, /* 77 These two must remain together and in this */
582 OP_BRAMINZERO, /* 78 order. */
583
584 OP_BRANUMBER, /* 79 Used for extracting brackets whose number is greater
585 than can fit into an opcode. */
586
587 OP_BRA /* 80 This and greater values are used for brackets that
588 extract substrings up to EXTRACT_BASIC_MAX. After
589 that, use is made of OP_BRANUMBER. */
590 };
591
592 /* WARNING WARNING WARNING: There is an implicit assumption in pcre.c and
593 study.c that all opcodes are less than 128 in value. This makes handling UTF-8
594 character sequences easier. */
595
596 /* The highest extraction number before we have to start using additional
597 bytes. (Originally PCRE didn't have support for extraction counts highter than
598 this number.) The value is limited by the number of opcodes left after OP_BRA,
599 i.e. 255 - OP_BRA. We actually set it a bit lower to leave room for additional
600 opcodes. */
601
602 #define EXTRACT_BASIC_MAX 100
603
604
605 /* This macro defines textual names for all the opcodes. These are used only
606 for debugging. The macro is referenced only in pcre_printint.c. */
607
608 #define OP_NAME_LIST \
609 "End", "\\A", "\\G", "\\B", "\\b", "\\D", "\\d", \
610 "\\S", "\\s", "\\W", "\\w", "Any", "Anybyte", \
611 "notprop", "prop", "extuni", \
612 "\\Z", "\\z", \
613 "Opt", "^", "$", "char", "charnc", "not", \
614 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
615 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
616 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
617 "*", "*?", "+", "+?", "?", "??", "{", "{", \
618 "class", "nclass", "xclass", "Ref", "Recurse", "Callout", \
619 "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \
620 "AssertB", "AssertB not", "Reverse", "Once", "Cond", "Cond ref",\
621 "Brazero", "Braminzero", "Branumber", "Bra"
622
623
624 /* This macro defines the length of fixed length operations in the compiled
625 regex. The lengths are used when searching for specific things, and also in the
626 debugging printing of a compiled regex. We use a macro so that it can be
627 defined close to the definitions of the opcodes themselves.
628
629 As things have been extended, some of these are no longer fixed lenths, but are
630 minima instead. For example, the length of a single-character repeat may vary
631 in UTF-8 mode. The code that uses this table must know about such things. */
632
633 #define OP_LENGTHS \
634 1, /* End */ \
635 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* \A, \G, \B, \B, \D, \d, \S, \s, \W, \w */ \
636 1, 1, /* Any, Anybyte */ \
637 2, 2, 1, /* NOTPROP, PROP, EXTUNI */ \
638 1, 1, 2, 1, 1, /* \Z, \z, Opt, ^, $ */ \
639 2, /* Char - the minimum length */ \
640 2, /* Charnc - the minimum length */ \
641 2, /* not */ \
642 /* Positive single-char repeats ** These are */ \
643 2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \
644 4, 4, 4, /* upto, minupto, exact ** UTF-8 mode */ \
645 /* Negative single-char repeats - only for chars < 256 */ \
646 2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \
647 4, 4, 4, /* NOT upto, minupto, exact */ \
648 /* Positive type repeats */ \
649 2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \
650 4, 4, 4, /* Type upto, minupto, exact */ \
651 /* Character class & ref repeats */ \
652 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \
653 5, 5, /* CRRANGE, CRMINRANGE */ \
654 33, /* CLASS */ \
655 33, /* NCLASS */ \
656 0, /* XCLASS - variable length */ \
657 3, /* REF */ \
658 1+LINK_SIZE, /* RECURSE */ \
659 2+2*LINK_SIZE, /* CALLOUT */ \
660 1+LINK_SIZE, /* Alt */ \
661 1+LINK_SIZE, /* Ket */ \
662 1+LINK_SIZE, /* KetRmax */ \
663 1+LINK_SIZE, /* KetRmin */ \
664 1+LINK_SIZE, /* Assert */ \
665 1+LINK_SIZE, /* Assert not */ \
666 1+LINK_SIZE, /* Assert behind */ \
667 1+LINK_SIZE, /* Assert behind not */ \
668 1+LINK_SIZE, /* Reverse */ \
669 1+LINK_SIZE, /* Once */ \
670 1+LINK_SIZE, /* COND */ \
671 3, /* CREF */ \
672 1, 1, /* BRAZERO, BRAMINZERO */ \
673 3, /* BRANUMBER */ \
674 1+LINK_SIZE /* BRA */ \
675
676
677 /* A magic value for OP_CREF to indicate the "in recursion" condition. */
678
679 #define CREF_RECURSE 0xffff
680
681 /* Error code numbers. They are given names so that they can more easily be
682 tracked. */
683
684 enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9,
685 ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19,
686 ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29,
687 ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
688 ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47 };
689
690 /* The real format of the start of the pcre block; the index of names and the
691 code vector run on as long as necessary after the end. We store an explicit
692 offset to the name table so that if a regex is compiled on one host, saved, and
693 then run on another where the size of pointers is different, all might still
694 be well. For the case of compiled-on-4 and run-on-8, we include an extra
695 pointer that is always NULL. For future-proofing, a few dummy fields were
696 originally included - even though you can never get this planning right - but
697 there is only one left now.
698
699 NOTE NOTE NOTE:
700 Because people can now save and re-use compiled patterns, any additions to this
701 structure should be made at the end, and something earlier (e.g. a new
702 flag in the options or one of the dummy fields) should indicate that the new
703 fields are present. Currently PCRE always sets the dummy fields to zero.
704 NOTE NOTE NOTE:
705 */
706
707 typedef struct real_pcre {
708 pcre_uint32 magic_number;
709 pcre_uint32 size; /* Total that was malloced */
710 pcre_uint32 options;
711 pcre_uint32 dummy1; /* For future use, maybe */
712
713 pcre_uint16 top_bracket;
714 pcre_uint16 top_backref;
715 pcre_uint16 first_byte;
716 pcre_uint16 req_byte;
717 pcre_uint16 name_table_offset; /* Offset to name table that follows */
718 pcre_uint16 name_entry_size; /* Size of any name items */
719 pcre_uint16 name_count; /* Number of name items */
720 pcre_uint16 ref_count; /* Reference count */
721
722 const unsigned char *tables; /* Pointer to tables or NULL for std */
723 const unsigned char *nullpad; /* NULL padding */
724 } real_pcre;
725
726 /* The format of the block used to store data from pcre_study(). The same
727 remark (see NOTE above) about extending this structure applies. */
728
729 typedef struct pcre_study_data {
730 pcre_uint32 size; /* Total that was malloced */
731 pcre_uint32 options;
732 uschar start_bits[32];
733 } pcre_study_data;
734
735 /* Structure for passing "static" information around between the functions
736 doing the compiling, so that they are thread-safe. */
737
738 typedef struct compile_data {
739 const uschar *lcc; /* Points to lower casing table */
740 const uschar *fcc; /* Points to case-flipping table */
741 const uschar *cbits; /* Points to character type table */
742 const uschar *ctypes; /* Points to table of type maps */
743 const uschar *start_code; /* The start of the compiled code */
744 const uschar *start_pattern; /* The start of the pattern */
745 uschar *name_table; /* The name/number table */
746 int names_found; /* Number of entries so far */
747 int name_entry_size; /* Size of each entry */
748 int top_backref; /* Maximum back reference */
749 unsigned int backref_map; /* Bitmap of low back refs */
750 int req_varyopt; /* "After variable item" flag for reqbyte */
751 BOOL nopartial; /* Set TRUE if partial won't work */
752 } compile_data;
753
754 /* Structure for maintaining a chain of pointers to the currently incomplete
755 branches, for testing for left recursion. */
756
757 typedef struct branch_chain {
758 struct branch_chain *outer;
759 uschar *current;
760 } branch_chain;
761
762 /* Structure for items in a linked list that represents an explicit recursive
763 call within the pattern. */
764
765 typedef struct recursion_info {
766 struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
767 int group_num; /* Number of group that was called */
768 const uschar *after_call; /* "Return value": points after the call in the expr */
769 const uschar *save_start; /* Old value of md->start_match */
770 int *offset_save; /* Pointer to start of saved offsets */
771 int saved_max; /* Number of saved offsets */
772 } recursion_info;
773
774 /* When compiling in a mode that doesn't use recursive calls to match(),
775 a structure is used to remember local variables on the heap. It is defined in
776 pcre.c, close to the match() function, so that it is easy to keep it in step
777 with any changes of local variable. However, the pointer to the current frame
778 must be saved in some "static" place over a longjmp(). We declare the
779 structure here so that we can put a pointer in the match_data structure.
780 NOTE: This isn't used for a "normal" compilation of pcre. */
781
782 struct heapframe;
783
784 /* Structure for passing "static" information around between the functions
785 doing traditional NFA matching, so that they are thread-safe. */
786
787 typedef struct match_data {
788 unsigned long int match_call_count; /* As it says */
789 unsigned long int match_limit;/* As it says */
790 int *offset_vector; /* Offset vector */
791 int offset_end; /* One past the end */
792 int offset_max; /* The maximum usable for return data */
793 const uschar *lcc; /* Points to lower casing table */
794 const uschar *ctypes; /* Points to table of type maps */
795 BOOL offset_overflow; /* Set if too many extractions */
796 BOOL notbol; /* NOTBOL flag */
797 BOOL noteol; /* NOTEOL flag */
798 BOOL utf8; /* UTF8 flag */
799 BOOL endonly; /* Dollar not before final \n */
800 BOOL notempty; /* Empty string match not wanted */
801 BOOL partial; /* PARTIAL flag */
802 BOOL hitend; /* Hit the end of the subject at some point */
803 const uschar *start_code; /* For use when recursing */
804 const uschar *start_subject; /* Start of the subject string */
805 const uschar *end_subject; /* End of the subject string */
806 const uschar *start_match; /* Start of this match attempt */
807 const uschar *end_match_ptr; /* Subject position at end match */
808 int end_offset_top; /* Highwater mark at end of match */
809 int capture_last; /* Most recent capture number */
810 int start_offset; /* The start offset value */
811 recursion_info *recursive; /* Linked list of recursion data */
812 void *callout_data; /* To pass back to callouts */
813 struct heapframe *thisframe; /* Used only when compiling for no recursion */
814 } match_data;
815
816 /* A similar structure is used for the same purpose by the DFA matching
817 functions. */
818
819 typedef struct dfa_match_data {
820 const uschar *start_code; /* Start of the compiled pattern */
821 const uschar *start_subject; /* Start of the subject string */
822 const uschar *end_subject; /* End of subject string */
823 const uschar *tables; /* Character tables */
824 int moptions; /* Match options */
825 int poptions; /* Pattern options */
826 void *callout_data; /* To pass back to callouts */
827 } dfa_match_data;
828
829 /* Bit definitions for entries in the pcre_ctypes table. */
830
831 #define ctype_space 0x01
832 #define ctype_letter 0x02
833 #define ctype_digit 0x04
834 #define ctype_xdigit 0x08
835 #define ctype_word 0x10 /* alphameric or '_' */
836 #define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
837
838 /* Offsets for the bitmap tables in pcre_cbits. Each table contains a set
839 of bits for a class map. Some classes are built by combining these tables. */
840
841 #define cbit_space 0 /* [:space:] or \s */
842 #define cbit_xdigit 32 /* [:xdigit:] */
843 #define cbit_digit 64 /* [:digit:] or \d */
844 #define cbit_upper 96 /* [:upper:] */
845 #define cbit_lower 128 /* [:lower:] */
846 #define cbit_word 160 /* [:word:] or \w */
847 #define cbit_graph 192 /* [:graph:] */
848 #define cbit_print 224 /* [:print:] */
849 #define cbit_punct 256 /* [:punct:] */
850 #define cbit_cntrl 288 /* [:cntrl:] */
851 #define cbit_length 320 /* Length of the cbits table */
852
853 /* Offsets of the various tables from the base tables pointer, and
854 total length. */
855
856 #define lcc_offset 0
857 #define fcc_offset 256
858 #define cbits_offset 512
859 #define ctypes_offset (cbits_offset + cbit_length)
860 #define tables_length (ctypes_offset + 256)
861
862 /* Layout of the UCP type table that translates property names into codes for
863 pcre_ucp_findchar(). */
864
865 typedef struct {
866 const char *name;
867 int value;
868 } ucp_type_table;
869
870
871 /* Internal shared data tables. These are tables that are used by more than one
872 of the exported public functions. They have to be "external" in the C sense,
873 but are not part of the PCRE public API. The data for these tables is in the
874 pcre_tables.c module. */
875
876 extern const int _pcre_utf8_table1[];
877 extern const int _pcre_utf8_table2[];
878 extern const int _pcre_utf8_table3[];
879 extern const uschar _pcre_utf8_table4[];
880
881 extern const int _pcre_utf8_table1_size;
882
883 extern const ucp_type_table _pcre_utt[];
884 extern const int _pcre_utt_size;
885
886 extern const uschar _pcre_default_tables[];
887
888 extern const uschar _pcre_OP_lengths[];
889
890
891 /* Internal shared functions. These are functions that are used by more than
892 one of the exported public functions. They have to be "external" in the C
893 sense, but are not part of the PCRE public API. */
894
895 extern int _pcre_ord2utf8(int, uschar *);
896 extern real_pcre * _pcre_try_flipped(const real_pcre *, real_pcre *,
897 const pcre_study_data *, pcre_study_data *);
898 extern int _pcre_ucp_findchar(const int, int *, int *);
899 extern int _pcre_valid_utf8(const uschar *, int);
900 extern BOOL _pcre_xclass(int, const uschar *);
901
902 #endif
903
904 /* End of pcre_internal.h */

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12