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

Contents of /code/trunk/pcre_internal.h

Parent Directory Parent Directory | Revision Log Revision Log


Revision 87 - (show annotations) (download)
Sat Feb 24 21:41:21 2007 UTC (7 years, 8 months ago) by nigel
File MIME type: text/plain
File size: 37990 byte(s)
Load pcre-6.5 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) 1997-2006 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 /* When PCRE is compiled as a C++ library, the subject pointer can be replaced
118 with a custom type. This makes it possible, for example, to allow pcre_exec()
119 to process subject strings that are discontinuous by using a smart pointer
120 class. It must always be possible to inspect all of the subject string in
121 pcre_exec() because of the way it backtracks. Two macros are required in the
122 normal case, for sign-unspecified and unsigned char pointers. The former is
123 used for the external interface and appears in pcre.h, which is why its name
124 must begin with PCRE_. */
125
126 #ifdef CUSTOM_SUBJECT_PTR
127 #define PCRE_SPTR CUSTOM_SUBJECT_PTR
128 #define USPTR CUSTOM_SUBJECT_PTR
129 #else
130 #define PCRE_SPTR const char *
131 #define USPTR const unsigned char *
132 #endif
133
134 /* Include the public PCRE header and the definitions of UCP character property
135 values. */
136
137 #include "pcre.h"
138 #include "ucp.h"
139
140 /* When compiling for use with the Virtual Pascal compiler, these functions
141 need to have their names changed. PCRE must be compiled with the -DVPCOMPAT
142 option on the command line. */
143
144 #ifdef VPCOMPAT
145 #define strncmp(s1,s2,m) _strncmp(s1,s2,m)
146 #define memcpy(d,s,n) _memcpy(d,s,n)
147 #define memmove(d,s,n) _memmove(d,s,n)
148 #define memset(s,c,n) _memset(s,c,n)
149 #else /* VPCOMPAT */
150
151 /* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
152 define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY
153 is set. Otherwise, include an emulating function for those systems that have
154 neither (there some non-Unix environments where this is the case). This assumes
155 that all calls to memmove are moving strings upwards in store, which is the
156 case in PCRE. */
157
158 #if ! HAVE_MEMMOVE
159 #undef memmove /* some systems may have a macro */
160 #if HAVE_BCOPY
161 #define memmove(a, b, c) bcopy(b, a, c)
162 #else /* HAVE_BCOPY */
163 void *
164 pcre_memmove(unsigned char *dest, const unsigned char *src, size_t n)
165 {
166 size_t i;
167 dest += n;
168 src += n;
169 for (i = 0; i < n; ++i) *(--dest) = *(--src);
170 return dest;
171 }
172 #define memmove(a, b, c) pcre_memmove(a, b, c)
173 #endif /* not HAVE_BCOPY */
174 #endif /* not HAVE_MEMMOVE */
175 #endif /* not VPCOMPAT */
176
177
178 /* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored
179 in big-endian order) by default. These are used, for example, to link from the
180 start of a subpattern to its alternatives and its end. The use of 2 bytes per
181 offset limits the size of the compiled regex to around 64K, which is big enough
182 for almost everybody. However, I received a request for an even bigger limit.
183 For this reason, and also to make the code easier to maintain, the storing and
184 loading of offsets from the byte string is now handled by the macros that are
185 defined here.
186
187 The macros are controlled by the value of LINK_SIZE. This defaults to 2 in
188 the config.h file, but can be overridden by using -D on the command line. This
189 is automated on Unix systems via the "configure" command. */
190
191 #if LINK_SIZE == 2
192
193 #define PUT(a,n,d) \
194 (a[n] = (d) >> 8), \
195 (a[(n)+1] = (d) & 255)
196
197 #define GET(a,n) \
198 (((a)[n] << 8) | (a)[(n)+1])
199
200 #define MAX_PATTERN_SIZE (1 << 16)
201
202
203 #elif LINK_SIZE == 3
204
205 #define PUT(a,n,d) \
206 (a[n] = (d) >> 16), \
207 (a[(n)+1] = (d) >> 8), \
208 (a[(n)+2] = (d) & 255)
209
210 #define GET(a,n) \
211 (((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2])
212
213 #define MAX_PATTERN_SIZE (1 << 24)
214
215
216 #elif LINK_SIZE == 4
217
218 #define PUT(a,n,d) \
219 (a[n] = (d) >> 24), \
220 (a[(n)+1] = (d) >> 16), \
221 (a[(n)+2] = (d) >> 8), \
222 (a[(n)+3] = (d) & 255)
223
224 #define GET(a,n) \
225 (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3])
226
227 #define MAX_PATTERN_SIZE (1 << 30) /* Keep it positive */
228
229
230 #else
231 #error LINK_SIZE must be either 2, 3, or 4
232 #endif
233
234
235 /* Convenience macro defined in terms of the others */
236
237 #define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE
238
239
240 /* PCRE uses some other 2-byte quantities that do not change when the size of
241 offsets changes. There are used for repeat counts and for other things such as
242 capturing parenthesis numbers in back references. */
243
244 #define PUT2(a,n,d) \
245 a[n] = (d) >> 8; \
246 a[(n)+1] = (d) & 255
247
248 #define GET2(a,n) \
249 (((a)[n] << 8) | (a)[(n)+1])
250
251 #define PUT2INC(a,n,d) PUT2(a,n,d), a += 2
252
253
254 /* When UTF-8 encoding is being used, a character is no longer just a single
255 byte. The macros for character handling generate simple sequences when used in
256 byte-mode, and more complicated ones for UTF-8 characters. */
257
258 #ifndef SUPPORT_UTF8
259 #define GETCHAR(c, eptr) c = *eptr;
260 #define GETCHARTEST(c, eptr) c = *eptr;
261 #define GETCHARINC(c, eptr) c = *eptr++;
262 #define GETCHARINCTEST(c, eptr) c = *eptr++;
263 #define GETCHARLEN(c, eptr, len) c = *eptr;
264 #define BACKCHAR(eptr)
265
266 #else /* SUPPORT_UTF8 */
267
268 /* Get the next UTF-8 character, not advancing the pointer. This is called when
269 we know we are in UTF-8 mode. */
270
271 #define GETCHAR(c, eptr) \
272 c = *eptr; \
273 if ((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, testing for UTF-8 mode, and not advancing the
287 pointer. */
288
289 #define GETCHARTEST(c, eptr) \
290 c = *eptr; \
291 if (utf8 && (c & 0xc0) == 0xc0) \
292 { \
293 int gcii; \
294 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
295 int gcss = 6*gcaa; \
296 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
297 for (gcii = 1; gcii <= gcaa; gcii++) \
298 { \
299 gcss -= 6; \
300 c |= (eptr[gcii] & 0x3f) << gcss; \
301 } \
302 }
303
304 /* Get the next UTF-8 character, advancing the pointer. This is called when we
305 know we are in UTF-8 mode. */
306
307 #define GETCHARINC(c, eptr) \
308 c = *eptr++; \
309 if ((c & 0xc0) == 0xc0) \
310 { \
311 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
312 int gcss = 6*gcaa; \
313 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
314 while (gcaa-- > 0) \
315 { \
316 gcss -= 6; \
317 c |= (*eptr++ & 0x3f) << gcss; \
318 } \
319 }
320
321 /* Get the next character, testing for UTF-8 mode, and advancing the pointer */
322
323 #define GETCHARINCTEST(c, eptr) \
324 c = *eptr++; \
325 if (utf8 && (c & 0xc0) == 0xc0) \
326 { \
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 while (gcaa-- > 0) \
331 { \
332 gcss -= 6; \
333 c |= (*eptr++ & 0x3f) << gcss; \
334 } \
335 }
336
337 /* Get the next UTF-8 character, not advancing the pointer, incrementing length
338 if there are extra bytes. This is called when we know we are in UTF-8 mode. */
339
340 #define GETCHARLEN(c, eptr, len) \
341 c = *eptr; \
342 if ((c & 0xc0) == 0xc0) \
343 { \
344 int gcii; \
345 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
346 int gcss = 6*gcaa; \
347 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
348 for (gcii = 1; gcii <= gcaa; gcii++) \
349 { \
350 gcss -= 6; \
351 c |= (eptr[gcii] & 0x3f) << gcss; \
352 } \
353 len += gcaa; \
354 }
355
356 /* If the pointer is not at the start of a character, move it back until
357 it is. Called only in UTF-8 mode. */
358
359 #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--;
360
361 #endif
362
363
364 /* In case there is no definition of offsetof() provided - though any proper
365 Standard C system should have one. */
366
367 #ifndef offsetof
368 #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))
369 #endif
370
371
372 /* These are the public options that can change during matching. */
373
374 #define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)
375
376 /* Private options flags start at the most significant end of the four bytes,
377 but skip the top bit so we can use ints for convenience without getting tangled
378 with negative values. The public options defined in pcre.h start at the least
379 significant end. Make sure they don't overlap! */
380
381 #define PCRE_FIRSTSET 0x40000000 /* first_byte is set */
382 #define PCRE_REQCHSET 0x20000000 /* req_byte is set */
383 #define PCRE_STARTLINE 0x10000000 /* start after \n for multiline */
384 #define PCRE_ICHANGED 0x08000000 /* i option changes within regex */
385 #define PCRE_NOPARTIAL 0x04000000 /* can't use partial with this regex */
386
387 /* Options for the "extra" block produced by pcre_study(). */
388
389 #define PCRE_STUDY_MAPPED 0x01 /* a map of starting chars exists */
390
391 /* Masks for identifying the public options that are permitted at compile
392 time, run time, or study time, respectively. */
393
394 #define PUBLIC_OPTIONS \
395 (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
396 PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
397 PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE)
398
399 #define PUBLIC_EXEC_OPTIONS \
400 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
401 PCRE_PARTIAL)
402
403 #define PUBLIC_DFA_EXEC_OPTIONS \
404 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
405 PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART)
406
407 #define PUBLIC_STUDY_OPTIONS 0 /* None defined */
408
409 /* Magic number to provide a small check against being handed junk. Also used
410 to detect whether a pattern was compiled on a host of different endianness. */
411
412 #define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */
413
414 /* Negative values for the firstchar and reqchar variables */
415
416 #define REQ_UNSET (-2)
417 #define REQ_NONE (-1)
418
419 /* The maximum remaining length of subject we are prepared to search for a
420 req_byte match. */
421
422 #define REQ_BYTE_MAX 1000
423
424 /* Flags added to firstbyte or reqbyte; a "non-literal" item is either a
425 variable-length repeat, or a anything other than literal characters. */
426
427 #define REQ_CASELESS 0x0100 /* indicates caselessness */
428 #define REQ_VARY 0x0200 /* reqbyte followed non-literal item */
429
430 /* Miscellaneous definitions */
431
432 typedef int BOOL;
433
434 #define FALSE 0
435 #define TRUE 1
436
437 /* Escape items that are just an encoding of a particular data value. Note that
438 ESC_n is defined as yet another macro, which is set in config.h to either \n
439 (the default) or \r (which some people want). */
440
441 #ifndef ESC_e
442 #define ESC_e 27
443 #endif
444
445 #ifndef ESC_f
446 #define ESC_f '\f'
447 #endif
448
449 #ifndef ESC_n
450 #define ESC_n NEWLINE
451 #endif
452
453 #ifndef ESC_r
454 #define ESC_r '\r'
455 #endif
456
457 /* We can't officially use ESC_t because it is a POSIX reserved identifier
458 (presumably because of all the others like size_t). */
459
460 #ifndef ESC_tee
461 #define ESC_tee '\t'
462 #endif
463
464 /* Codes for different types of Unicode property */
465
466 #define PT_ANY 0 /* Any property - matches all chars */
467 #define PT_LAMP 1 /* L& - the union of Lu, Ll, Lt */
468 #define PT_GC 2 /* General characteristic (e.g. L) */
469 #define PT_PC 3 /* Particular characteristic (e.g. Lu) */
470 #define PT_SC 4 /* Script (e.g. Han) */
471
472 /* Flag bits and data types for the extended class (OP_XCLASS) for classes that
473 contain UTF-8 characters with values greater than 255. */
474
475 #define XCL_NOT 0x01 /* Flag: this is a negative class */
476 #define XCL_MAP 0x02 /* Flag: a 32-byte map is present */
477
478 #define XCL_END 0 /* Marks end of individual items */
479 #define XCL_SINGLE 1 /* Single item (one multibyte char) follows */
480 #define XCL_RANGE 2 /* A range (two multibyte chars) follows */
481 #define XCL_PROP 3 /* Unicode property (2-byte property code follows) */
482 #define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */
483
484 /* These are escaped items that aren't just an encoding of a particular data
485 value such as \n. They must have non-zero values, as check_escape() returns
486 their negation. Also, they must appear in the same order as in the opcode
487 definitions below, up to ESC_z. There's a dummy for OP_ANY because it
488 corresponds to "." rather than an escape sequence. The final one must be
489 ESC_REF as subsequent values are used for \1, \2, \3, etc. There is are two
490 tests in the code for an escape greater than ESC_b and less than ESC_Z to
491 detect the types that may be repeated. These are the types that consume
492 characters. If any new escapes are put in between that don't consume a
493 character, that code will have to change. */
494
495 enum { ESC_A = 1, ESC_G, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W,
496 ESC_w, ESC_dum1, ESC_C, ESC_P, ESC_p, ESC_X, ESC_Z, ESC_z, ESC_E,
497 ESC_Q, ESC_REF };
498
499 /* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
500 that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
501 OP_EOD must correspond in order to the list of escapes immediately above.
502 Note that whenever this list is updated, the two macro definitions that follow
503 must also be updated to match. */
504
505 enum {
506 OP_END, /* 0 End of pattern */
507
508 /* Values corresponding to backslashed metacharacters */
509
510 OP_SOD, /* 1 Start of data: \A */
511 OP_SOM, /* 2 Start of match (subject + offset): \G */
512 OP_NOT_WORD_BOUNDARY, /* 3 \B */
513 OP_WORD_BOUNDARY, /* 4 \b */
514 OP_NOT_DIGIT, /* 5 \D */
515 OP_DIGIT, /* 6 \d */
516 OP_NOT_WHITESPACE, /* 7 \S */
517 OP_WHITESPACE, /* 8 \s */
518 OP_NOT_WORDCHAR, /* 9 \W */
519 OP_WORDCHAR, /* 10 \w */
520 OP_ANY, /* 11 Match any character */
521 OP_ANYBYTE, /* 12 Match any byte (\C); different to OP_ANY for UTF-8 */
522 OP_NOTPROP, /* 13 \P (not Unicode property) */
523 OP_PROP, /* 14 \p (Unicode property) */
524 OP_EXTUNI, /* 15 \X (extended Unicode sequence */
525 OP_EODN, /* 16 End of data or \n at end of data: \Z. */
526 OP_EOD, /* 17 End of data: \z */
527
528 OP_OPT, /* 18 Set runtime options */
529 OP_CIRC, /* 19 Start of line - varies with multiline switch */
530 OP_DOLL, /* 20 End of line - varies with multiline switch */
531 OP_CHAR, /* 21 Match one character, casefully */
532 OP_CHARNC, /* 22 Match one character, caselessly */
533 OP_NOT, /* 23 Match anything but the following char */
534
535 OP_STAR, /* 24 The maximizing and minimizing versions of */
536 OP_MINSTAR, /* 25 all these opcodes must come in pairs, with */
537 OP_PLUS, /* 26 the minimizing one second. */
538 OP_MINPLUS, /* 27 This first set applies to single characters */
539 OP_QUERY, /* 28 */
540 OP_MINQUERY, /* 29 */
541 OP_UPTO, /* 30 From 0 to n matches */
542 OP_MINUPTO, /* 31 */
543 OP_EXACT, /* 32 Exactly n matches */
544
545 OP_NOTSTAR, /* 33 The maximizing and minimizing versions of */
546 OP_NOTMINSTAR, /* 34 all these opcodes must come in pairs, with */
547 OP_NOTPLUS, /* 35 the minimizing one second. */
548 OP_NOTMINPLUS, /* 36 This set applies to "not" single characters */
549 OP_NOTQUERY, /* 37 */
550 OP_NOTMINQUERY, /* 38 */
551 OP_NOTUPTO, /* 39 From 0 to n matches */
552 OP_NOTMINUPTO, /* 40 */
553 OP_NOTEXACT, /* 41 Exactly n matches */
554
555 OP_TYPESTAR, /* 42 The maximizing and minimizing versions of */
556 OP_TYPEMINSTAR, /* 43 all these opcodes must come in pairs, with */
557 OP_TYPEPLUS, /* 44 the minimizing one second. These codes must */
558 OP_TYPEMINPLUS, /* 45 be in exactly the same order as those above. */
559 OP_TYPEQUERY, /* 46 This set applies to character types such as \d */
560 OP_TYPEMINQUERY, /* 47 */
561 OP_TYPEUPTO, /* 48 From 0 to n matches */
562 OP_TYPEMINUPTO, /* 49 */
563 OP_TYPEEXACT, /* 50 Exactly n matches */
564
565 OP_CRSTAR, /* 51 The maximizing and minimizing versions of */
566 OP_CRMINSTAR, /* 52 all these opcodes must come in pairs, with */
567 OP_CRPLUS, /* 53 the minimizing one second. These codes must */
568 OP_CRMINPLUS, /* 54 be in exactly the same order as those above. */
569 OP_CRQUERY, /* 55 These are for character classes and back refs */
570 OP_CRMINQUERY, /* 56 */
571 OP_CRRANGE, /* 57 These are different to the three sets above. */
572 OP_CRMINRANGE, /* 58 */
573
574 OP_CLASS, /* 59 Match a character class, chars < 256 only */
575 OP_NCLASS, /* 60 Same, but the bitmap was created from a negative
576 class - the difference is relevant only when a UTF-8
577 character > 255 is encountered. */
578
579 OP_XCLASS, /* 61 Extended class for handling UTF-8 chars within the
580 class. This does both positive and negative. */
581
582 OP_REF, /* 62 Match a back reference */
583 OP_RECURSE, /* 63 Match a numbered subpattern (possibly recursive) */
584 OP_CALLOUT, /* 64 Call out to external function if provided */
585
586 OP_ALT, /* 65 Start of alternation */
587 OP_KET, /* 66 End of group that doesn't have an unbounded repeat */
588 OP_KETRMAX, /* 67 These two must remain together and in this */
589 OP_KETRMIN, /* 68 order. They are for groups the repeat for ever. */
590
591 /* The assertions must come before ONCE and COND */
592
593 OP_ASSERT, /* 69 Positive lookahead */
594 OP_ASSERT_NOT, /* 70 Negative lookahead */
595 OP_ASSERTBACK, /* 71 Positive lookbehind */
596 OP_ASSERTBACK_NOT, /* 72 Negative lookbehind */
597 OP_REVERSE, /* 73 Move pointer back - used in lookbehind assertions */
598
599 /* ONCE and COND must come after the assertions, with ONCE first, as there's
600 a test for >= ONCE for a subpattern that isn't an assertion. */
601
602 OP_ONCE, /* 74 Once matched, don't back up into the subpattern */
603 OP_COND, /* 75 Conditional group */
604 OP_CREF, /* 76 Used to hold an extraction string number (cond ref) */
605
606 OP_BRAZERO, /* 77 These two must remain together and in this */
607 OP_BRAMINZERO, /* 78 order. */
608
609 OP_BRANUMBER, /* 79 Used for extracting brackets whose number is greater
610 than can fit into an opcode. */
611
612 OP_BRA /* 80 This and greater values are used for brackets that
613 extract substrings up to EXTRACT_BASIC_MAX. After
614 that, use is made of OP_BRANUMBER. */
615 };
616
617 /* WARNING WARNING WARNING: There is an implicit assumption in pcre.c and
618 study.c that all opcodes are less than 128 in value. This makes handling UTF-8
619 character sequences easier. */
620
621 /* The highest extraction number before we have to start using additional
622 bytes. (Originally PCRE didn't have support for extraction counts highter than
623 this number.) The value is limited by the number of opcodes left after OP_BRA,
624 i.e. 255 - OP_BRA. We actually set it a bit lower to leave room for additional
625 opcodes. */
626
627 #define EXTRACT_BASIC_MAX 100
628
629
630 /* This macro defines textual names for all the opcodes. These are used only
631 for debugging. The macro is referenced only in pcre_printint.c. */
632
633 #define OP_NAME_LIST \
634 "End", "\\A", "\\G", "\\B", "\\b", "\\D", "\\d", \
635 "\\S", "\\s", "\\W", "\\w", "Any", "Anybyte", \
636 "notprop", "prop", "extuni", \
637 "\\Z", "\\z", \
638 "Opt", "^", "$", "char", "charnc", "not", \
639 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
640 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
641 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
642 "*", "*?", "+", "+?", "?", "??", "{", "{", \
643 "class", "nclass", "xclass", "Ref", "Recurse", "Callout", \
644 "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \
645 "AssertB", "AssertB not", "Reverse", "Once", "Cond", "Cond ref",\
646 "Brazero", "Braminzero", "Branumber", "Bra"
647
648
649 /* This macro defines the length of fixed length operations in the compiled
650 regex. The lengths are used when searching for specific things, and also in the
651 debugging printing of a compiled regex. We use a macro so that it can be
652 defined close to the definitions of the opcodes themselves.
653
654 As things have been extended, some of these are no longer fixed lenths, but are
655 minima instead. For example, the length of a single-character repeat may vary
656 in UTF-8 mode. The code that uses this table must know about such things. */
657
658 #define OP_LENGTHS \
659 1, /* End */ \
660 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* \A, \G, \B, \B, \D, \d, \S, \s, \W, \w */ \
661 1, 1, /* Any, Anybyte */ \
662 3, 3, 1, /* NOTPROP, PROP, EXTUNI */ \
663 1, 1, 2, 1, 1, /* \Z, \z, Opt, ^, $ */ \
664 2, /* Char - the minimum length */ \
665 2, /* Charnc - the minimum length */ \
666 2, /* not */ \
667 /* Positive single-char repeats ** These are */ \
668 2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \
669 4, 4, 4, /* upto, minupto, exact ** UTF-8 mode */ \
670 /* Negative single-char repeats - only for chars < 256 */ \
671 2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \
672 4, 4, 4, /* NOT upto, minupto, exact */ \
673 /* Positive type repeats */ \
674 2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \
675 4, 4, 4, /* Type upto, minupto, exact */ \
676 /* Character class & ref repeats */ \
677 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \
678 5, 5, /* CRRANGE, CRMINRANGE */ \
679 33, /* CLASS */ \
680 33, /* NCLASS */ \
681 0, /* XCLASS - variable length */ \
682 3, /* REF */ \
683 1+LINK_SIZE, /* RECURSE */ \
684 2+2*LINK_SIZE, /* CALLOUT */ \
685 1+LINK_SIZE, /* Alt */ \
686 1+LINK_SIZE, /* Ket */ \
687 1+LINK_SIZE, /* KetRmax */ \
688 1+LINK_SIZE, /* KetRmin */ \
689 1+LINK_SIZE, /* Assert */ \
690 1+LINK_SIZE, /* Assert not */ \
691 1+LINK_SIZE, /* Assert behind */ \
692 1+LINK_SIZE, /* Assert behind not */ \
693 1+LINK_SIZE, /* Reverse */ \
694 1+LINK_SIZE, /* Once */ \
695 1+LINK_SIZE, /* COND */ \
696 3, /* CREF */ \
697 1, 1, /* BRAZERO, BRAMINZERO */ \
698 3, /* BRANUMBER */ \
699 1+LINK_SIZE /* BRA */ \
700
701
702 /* A magic value for OP_CREF to indicate the "in recursion" condition. */
703
704 #define CREF_RECURSE 0xffff
705
706 /* Error code numbers. They are given names so that they can more easily be
707 tracked. */
708
709 enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9,
710 ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19,
711 ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29,
712 ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
713 ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47 };
714
715 /* The real format of the start of the pcre block; the index of names and the
716 code vector run on as long as necessary after the end. We store an explicit
717 offset to the name table so that if a regex is compiled on one host, saved, and
718 then run on another where the size of pointers is different, all might still
719 be well. For the case of compiled-on-4 and run-on-8, we include an extra
720 pointer that is always NULL. For future-proofing, a few dummy fields were
721 originally included - even though you can never get this planning right - but
722 there is only one left now.
723
724 NOTE NOTE NOTE:
725 Because people can now save and re-use compiled patterns, any additions to this
726 structure should be made at the end, and something earlier (e.g. a new
727 flag in the options or one of the dummy fields) should indicate that the new
728 fields are present. Currently PCRE always sets the dummy fields to zero.
729 NOTE NOTE NOTE:
730 */
731
732 typedef struct real_pcre {
733 pcre_uint32 magic_number;
734 pcre_uint32 size; /* Total that was malloced */
735 pcre_uint32 options;
736 pcre_uint32 dummy1; /* For future use, maybe */
737
738 pcre_uint16 top_bracket;
739 pcre_uint16 top_backref;
740 pcre_uint16 first_byte;
741 pcre_uint16 req_byte;
742 pcre_uint16 name_table_offset; /* Offset to name table that follows */
743 pcre_uint16 name_entry_size; /* Size of any name items */
744 pcre_uint16 name_count; /* Number of name items */
745 pcre_uint16 ref_count; /* Reference count */
746
747 const unsigned char *tables; /* Pointer to tables or NULL for std */
748 const unsigned char *nullpad; /* NULL padding */
749 } real_pcre;
750
751 /* The format of the block used to store data from pcre_study(). The same
752 remark (see NOTE above) about extending this structure applies. */
753
754 typedef struct pcre_study_data {
755 pcre_uint32 size; /* Total that was malloced */
756 pcre_uint32 options;
757 uschar start_bits[32];
758 } pcre_study_data;
759
760 /* Structure for passing "static" information around between the functions
761 doing the compiling, so that they are thread-safe. */
762
763 typedef struct compile_data {
764 const uschar *lcc; /* Points to lower casing table */
765 const uschar *fcc; /* Points to case-flipping table */
766 const uschar *cbits; /* Points to character type table */
767 const uschar *ctypes; /* Points to table of type maps */
768 const uschar *start_code; /* The start of the compiled code */
769 const uschar *start_pattern; /* The start of the pattern */
770 uschar *name_table; /* The name/number table */
771 int names_found; /* Number of entries so far */
772 int name_entry_size; /* Size of each entry */
773 int top_backref; /* Maximum back reference */
774 unsigned int backref_map; /* Bitmap of low back refs */
775 int req_varyopt; /* "After variable item" flag for reqbyte */
776 BOOL nopartial; /* Set TRUE if partial won't work */
777 } compile_data;
778
779 /* Structure for maintaining a chain of pointers to the currently incomplete
780 branches, for testing for left recursion. */
781
782 typedef struct branch_chain {
783 struct branch_chain *outer;
784 uschar *current;
785 } branch_chain;
786
787 /* Structure for items in a linked list that represents an explicit recursive
788 call within the pattern. */
789
790 typedef struct recursion_info {
791 struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
792 int group_num; /* Number of group that was called */
793 const uschar *after_call; /* "Return value": points after the call in the expr */
794 USPTR save_start; /* Old value of md->start_match */
795 int *offset_save; /* Pointer to start of saved offsets */
796 int saved_max; /* Number of saved offsets */
797 } recursion_info;
798
799 /* When compiling in a mode that doesn't use recursive calls to match(),
800 a structure is used to remember local variables on the heap. It is defined in
801 pcre.c, close to the match() function, so that it is easy to keep it in step
802 with any changes of local variable. However, the pointer to the current frame
803 must be saved in some "static" place over a longjmp(). We declare the
804 structure here so that we can put a pointer in the match_data structure.
805 NOTE: This isn't used for a "normal" compilation of pcre. */
806
807 struct heapframe;
808
809 /* Structure for passing "static" information around between the functions
810 doing traditional NFA matching, so that they are thread-safe. */
811
812 typedef struct match_data {
813 unsigned long int match_call_count; /* As it says */
814 unsigned long int match_limit; /* As it says */
815 unsigned long int match_limit_recursion; /* As it says */
816 int *offset_vector; /* Offset vector */
817 int offset_end; /* One past the end */
818 int offset_max; /* The maximum usable for return data */
819 const uschar *lcc; /* Points to lower casing table */
820 const uschar *ctypes; /* Points to table of type maps */
821 BOOL offset_overflow; /* Set if too many extractions */
822 BOOL notbol; /* NOTBOL flag */
823 BOOL noteol; /* NOTEOL flag */
824 BOOL utf8; /* UTF8 flag */
825 BOOL endonly; /* Dollar not before final \n */
826 BOOL notempty; /* Empty string match not wanted */
827 BOOL partial; /* PARTIAL flag */
828 BOOL hitend; /* Hit the end of the subject at some point */
829 const uschar *start_code; /* For use when recursing */
830 USPTR start_subject; /* Start of the subject string */
831 USPTR end_subject; /* End of the subject string */
832 USPTR start_match; /* Start of this match attempt */
833 USPTR end_match_ptr; /* Subject position at end match */
834 int end_offset_top; /* Highwater mark at end of match */
835 int capture_last; /* Most recent capture number */
836 int start_offset; /* The start offset value */
837 recursion_info *recursive; /* Linked list of recursion data */
838 void *callout_data; /* To pass back to callouts */
839 struct heapframe *thisframe; /* Used only when compiling for no recursion */
840 } match_data;
841
842 /* A similar structure is used for the same purpose by the DFA matching
843 functions. */
844
845 typedef struct dfa_match_data {
846 const uschar *start_code; /* Start of the compiled pattern */
847 const uschar *start_subject; /* Start of the subject string */
848 const uschar *end_subject; /* End of subject string */
849 const uschar *tables; /* Character tables */
850 int moptions; /* Match options */
851 int poptions; /* Pattern options */
852 void *callout_data; /* To pass back to callouts */
853 } dfa_match_data;
854
855 /* Bit definitions for entries in the pcre_ctypes table. */
856
857 #define ctype_space 0x01
858 #define ctype_letter 0x02
859 #define ctype_digit 0x04
860 #define ctype_xdigit 0x08
861 #define ctype_word 0x10 /* alphameric or '_' */
862 #define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
863
864 /* Offsets for the bitmap tables in pcre_cbits. Each table contains a set
865 of bits for a class map. Some classes are built by combining these tables. */
866
867 #define cbit_space 0 /* [:space:] or \s */
868 #define cbit_xdigit 32 /* [:xdigit:] */
869 #define cbit_digit 64 /* [:digit:] or \d */
870 #define cbit_upper 96 /* [:upper:] */
871 #define cbit_lower 128 /* [:lower:] */
872 #define cbit_word 160 /* [:word:] or \w */
873 #define cbit_graph 192 /* [:graph:] */
874 #define cbit_print 224 /* [:print:] */
875 #define cbit_punct 256 /* [:punct:] */
876 #define cbit_cntrl 288 /* [:cntrl:] */
877 #define cbit_length 320 /* Length of the cbits table */
878
879 /* Offsets of the various tables from the base tables pointer, and
880 total length. */
881
882 #define lcc_offset 0
883 #define fcc_offset 256
884 #define cbits_offset 512
885 #define ctypes_offset (cbits_offset + cbit_length)
886 #define tables_length (ctypes_offset + 256)
887
888 /* Layout of the UCP type table that translates property names into types and
889 codes. */
890
891 typedef struct {
892 const char *name;
893 pcre_uint16 type;
894 pcre_uint16 value;
895 } ucp_type_table;
896
897
898 /* Internal shared data tables. These are tables that are used by more than one
899 of the exported public functions. They have to be "external" in the C sense,
900 but are not part of the PCRE public API. The data for these tables is in the
901 pcre_tables.c module. */
902
903 extern const int _pcre_utf8_table1[];
904 extern const int _pcre_utf8_table2[];
905 extern const int _pcre_utf8_table3[];
906 extern const uschar _pcre_utf8_table4[];
907
908 extern const int _pcre_utf8_table1_size;
909
910 extern const ucp_type_table _pcre_utt[];
911 extern const int _pcre_utt_size;
912
913 extern const uschar _pcre_default_tables[];
914
915 extern const uschar _pcre_OP_lengths[];
916
917
918 /* Internal shared functions. These are functions that are used by more than
919 one of the exported public functions. They have to be "external" in the C
920 sense, but are not part of the PCRE public API. */
921
922 extern int _pcre_ord2utf8(int, uschar *);
923 extern real_pcre * _pcre_try_flipped(const real_pcre *, real_pcre *,
924 const pcre_study_data *, pcre_study_data *);
925 extern int _pcre_ucp_findprop(const int, int *, int *);
926 extern int _pcre_ucp_othercase(const int);
927 extern int _pcre_valid_utf8(const uschar *, int);
928 extern BOOL _pcre_xclass(int, const uschar *);
929
930 #endif
931
932 /* End of pcre_internal.h */

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12