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

Contents of /code/trunk/pcre_internal.h

Parent Directory Parent Directory | Revision Log Revision Log


Revision 93 - (show annotations) (download)
Sat Feb 24 21:41:42 2007 UTC (7 years, 6 months ago) by nigel
File MIME type: text/plain
File size: 42748 byte(s)
Load pcre-7.0 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 It turns out that the Mac Debugging.h header also defines the macro DPRINTF, so
60 be absolutely sure we get our version. */
61
62 #undef DPRINTF
63 #ifdef DEBUG
64 #define DPRINTF(p) printf p
65 #else
66 #define DPRINTF(p) /* Nothing */
67 #endif
68
69
70 /* Get the definitions provided by running "configure" */
71
72 #include "config.h"
73
74 /* Standard C headers plus the external interface definition. The only time
75 setjmp and stdarg are used is when NO_RECURSE is set. */
76
77 #include <ctype.h>
78 #include <limits.h>
79 #include <setjmp.h>
80 #include <stdarg.h>
81 #include <stddef.h>
82 #include <stdio.h>
83 #include <stdlib.h>
84 #include <string.h>
85
86 #ifndef PCRE_SPY
87 #define PCRE_DEFINITION /* Win32 __declspec(export) trigger for .dll */
88 #endif
89
90 /* We need to have types that specify unsigned 16-bit and 32-bit integers. We
91 cannot determine these outside the compilation (e.g. by running a program as
92 part of "configure") because PCRE is often cross-compiled for use on other
93 systems. Instead we make use of the maximum sizes that are available at
94 preprocessor time in standard C environments. */
95
96 #if USHRT_MAX == 65535
97 typedef unsigned short pcre_uint16;
98 #elif UINT_MAX == 65535
99 typedef unsigned int pcre_uint16;
100 #else
101 #error Cannot determine a type for 16-bit unsigned integers
102 #endif
103
104 #if UINT_MAX == 4294967295
105 typedef unsigned int pcre_uint32;
106 #elif ULONG_MAX == 4294967295
107 typedef unsigned long int pcre_uint32;
108 #else
109 #error Cannot determine a type for 32-bit unsigned integers
110 #endif
111
112 /* All character handling must be done as unsigned characters. Otherwise there
113 are problems with top-bit-set characters and functions such as isspace().
114 However, we leave the interface to the outside world as char *, because that
115 should make things easier for callers. We define a short type for unsigned char
116 to save lots of typing. I tried "uchar", but it causes problems on Digital
117 Unix, where it is defined in sys/types, so use "uschar" instead. */
118
119 typedef unsigned char uschar;
120
121 /* This is an unsigned int value that no character can ever have. UTF-8
122 characters only go up to 0x7fffffff (though Unicode doesn't go beyond
123 0x0010ffff). */
124
125 #define NOTACHAR 0xffffffff
126
127 /* PCRE is able to support several different kinds of newline (CR, LF, CRLF,
128 and "all" at present). The following macros are used to package up testing for
129 newlines. NLBLOCK, PSSTART, and PSEND are defined in the various modules to
130 indicate in which datablock the parameters exist, and what the start/end of
131 string field names are. */
132
133 #define NLTYPE_FIXED 0 /* Newline is a fixed length string */
134 #define NLTYPE_ANY 1 /* Newline is any Unicode line ending */
135
136 /* This macro checks for a newline at the given position */
137
138 #define IS_NEWLINE(p) \
139 ((NLBLOCK->nltype != NLTYPE_FIXED)? \
140 ((p) < NLBLOCK->PSEND && \
141 _pcre_is_newline((p), NLBLOCK->PSEND, &(NLBLOCK->nllen), utf8) \
142 ) \
143 : \
144 ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
145 (p)[0] == NLBLOCK->nl[0] && \
146 (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1]) \
147 ) \
148 )
149
150 /* This macro checks for a newline immediately preceding the given position */
151
152 #define WAS_NEWLINE(p) \
153 ((NLBLOCK->nltype != NLTYPE_FIXED)? \
154 ((p) > NLBLOCK->PSSTART && \
155 _pcre_was_newline((p), NLBLOCK->PSSTART, &(NLBLOCK->nllen), utf8) \
156 ) \
157 : \
158 ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
159 (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \
160 (NLBLOCK->nllen == 1 || (p)[-NLBLOCK->nllen+1] == NLBLOCK->nl[1]) \
161 ) \
162 )
163
164 /* When PCRE is compiled as a C++ library, the subject pointer can be replaced
165 with a custom type. This makes it possible, for example, to allow pcre_exec()
166 to process subject strings that are discontinuous by using a smart pointer
167 class. It must always be possible to inspect all of the subject string in
168 pcre_exec() because of the way it backtracks. Two macros are required in the
169 normal case, for sign-unspecified and unsigned char pointers. The former is
170 used for the external interface and appears in pcre.h, which is why its name
171 must begin with PCRE_. */
172
173 #ifdef CUSTOM_SUBJECT_PTR
174 #define PCRE_SPTR CUSTOM_SUBJECT_PTR
175 #define USPTR CUSTOM_SUBJECT_PTR
176 #else
177 #define PCRE_SPTR const char *
178 #define USPTR const unsigned char *
179 #endif
180
181 /* Include the public PCRE header and the definitions of UCP character property
182 values. */
183
184 #include "pcre.h"
185 #include "ucp.h"
186
187 /* When compiling for use with the Virtual Pascal compiler, these functions
188 need to have their names changed. PCRE must be compiled with the -DVPCOMPAT
189 option on the command line. */
190
191 #ifdef VPCOMPAT
192 #define strncmp(s1,s2,m) _strncmp(s1,s2,m)
193 #define memcpy(d,s,n) _memcpy(d,s,n)
194 #define memmove(d,s,n) _memmove(d,s,n)
195 #define memset(s,c,n) _memset(s,c,n)
196 #else /* VPCOMPAT */
197
198 /* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
199 define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY
200 is set. Otherwise, include an emulating function for those systems that have
201 neither (there some non-Unix environments where this is the case). This assumes
202 that all calls to memmove are moving strings upwards in store, which is the
203 case in PCRE. */
204
205 #if ! HAVE_MEMMOVE
206 #undef memmove /* some systems may have a macro */
207 #if HAVE_BCOPY
208 #define memmove(a, b, c) bcopy(b, a, c)
209 #else /* HAVE_BCOPY */
210 static void *
211 pcre_memmove(unsigned char *dest, const unsigned char *src, size_t n)
212 {
213 size_t i;
214 dest += n;
215 src += n;
216 for (i = 0; i < n; ++i) *(--dest) = *(--src);
217 return dest;
218 }
219 #define memmove(a, b, c) pcre_memmove(a, b, c)
220 #endif /* not HAVE_BCOPY */
221 #endif /* not HAVE_MEMMOVE */
222 #endif /* not VPCOMPAT */
223
224
225 /* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored
226 in big-endian order) by default. These are used, for example, to link from the
227 start of a subpattern to its alternatives and its end. The use of 2 bytes per
228 offset limits the size of the compiled regex to around 64K, which is big enough
229 for almost everybody. However, I received a request for an even bigger limit.
230 For this reason, and also to make the code easier to maintain, the storing and
231 loading of offsets from the byte string is now handled by the macros that are
232 defined here.
233
234 The macros are controlled by the value of LINK_SIZE. This defaults to 2 in
235 the config.h file, but can be overridden by using -D on the command line. This
236 is automated on Unix systems via the "configure" command. */
237
238 #if LINK_SIZE == 2
239
240 #define PUT(a,n,d) \
241 (a[n] = (d) >> 8), \
242 (a[(n)+1] = (d) & 255)
243
244 #define GET(a,n) \
245 (((a)[n] << 8) | (a)[(n)+1])
246
247 #define MAX_PATTERN_SIZE (1 << 16)
248
249
250 #elif LINK_SIZE == 3
251
252 #define PUT(a,n,d) \
253 (a[n] = (d) >> 16), \
254 (a[(n)+1] = (d) >> 8), \
255 (a[(n)+2] = (d) & 255)
256
257 #define GET(a,n) \
258 (((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2])
259
260 #define MAX_PATTERN_SIZE (1 << 24)
261
262
263 #elif LINK_SIZE == 4
264
265 #define PUT(a,n,d) \
266 (a[n] = (d) >> 24), \
267 (a[(n)+1] = (d) >> 16), \
268 (a[(n)+2] = (d) >> 8), \
269 (a[(n)+3] = (d) & 255)
270
271 #define GET(a,n) \
272 (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3])
273
274 #define MAX_PATTERN_SIZE (1 << 30) /* Keep it positive */
275
276
277 #else
278 #error LINK_SIZE must be either 2, 3, or 4
279 #endif
280
281
282 /* Convenience macro defined in terms of the others */
283
284 #define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE
285
286
287 /* PCRE uses some other 2-byte quantities that do not change when the size of
288 offsets changes. There are used for repeat counts and for other things such as
289 capturing parenthesis numbers in back references. */
290
291 #define PUT2(a,n,d) \
292 a[n] = (d) >> 8; \
293 a[(n)+1] = (d) & 255
294
295 #define GET2(a,n) \
296 (((a)[n] << 8) | (a)[(n)+1])
297
298 #define PUT2INC(a,n,d) PUT2(a,n,d), a += 2
299
300
301 /* When UTF-8 encoding is being used, a character is no longer just a single
302 byte. The macros for character handling generate simple sequences when used in
303 byte-mode, and more complicated ones for UTF-8 characters. */
304
305 #ifndef SUPPORT_UTF8
306 #define GETCHAR(c, eptr) c = *eptr;
307 #define GETCHARTEST(c, eptr) c = *eptr;
308 #define GETCHARINC(c, eptr) c = *eptr++;
309 #define GETCHARINCTEST(c, eptr) c = *eptr++;
310 #define GETCHARLEN(c, eptr, len) c = *eptr;
311 #define BACKCHAR(eptr)
312
313 #else /* SUPPORT_UTF8 */
314
315 /* Get the next UTF-8 character, not advancing the pointer. This is called when
316 we know we are in UTF-8 mode. */
317
318 #define GETCHAR(c, eptr) \
319 c = *eptr; \
320 if (c >= 0xc0) \
321 { \
322 int gcii; \
323 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
324 int gcss = 6*gcaa; \
325 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
326 for (gcii = 1; gcii <= gcaa; gcii++) \
327 { \
328 gcss -= 6; \
329 c |= (eptr[gcii] & 0x3f) << gcss; \
330 } \
331 }
332
333 /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
334 pointer. */
335
336 #define GETCHARTEST(c, eptr) \
337 c = *eptr; \
338 if (utf8 && c >= 0xc0) \
339 { \
340 int gcii; \
341 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
342 int gcss = 6*gcaa; \
343 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
344 for (gcii = 1; gcii <= gcaa; gcii++) \
345 { \
346 gcss -= 6; \
347 c |= (eptr[gcii] & 0x3f) << gcss; \
348 } \
349 }
350
351 /* Get the next UTF-8 character, advancing the pointer. This is called when we
352 know we are in UTF-8 mode. */
353
354 #define GETCHARINC(c, eptr) \
355 c = *eptr++; \
356 if (c >= 0xc0) \
357 { \
358 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
359 int gcss = 6*gcaa; \
360 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
361 while (gcaa-- > 0) \
362 { \
363 gcss -= 6; \
364 c |= (*eptr++ & 0x3f) << gcss; \
365 } \
366 }
367
368 /* Get the next character, testing for UTF-8 mode, and advancing the pointer */
369
370 #define GETCHARINCTEST(c, eptr) \
371 c = *eptr++; \
372 if (utf8 && c >= 0xc0) \
373 { \
374 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
375 int gcss = 6*gcaa; \
376 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
377 while (gcaa-- > 0) \
378 { \
379 gcss -= 6; \
380 c |= (*eptr++ & 0x3f) << gcss; \
381 } \
382 }
383
384 /* Get the next UTF-8 character, not advancing the pointer, incrementing length
385 if there are extra bytes. This is called when we know we are in UTF-8 mode. */
386
387 #define GETCHARLEN(c, eptr, len) \
388 c = *eptr; \
389 if (c >= 0xc0) \
390 { \
391 int gcii; \
392 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
393 int gcss = 6*gcaa; \
394 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
395 for (gcii = 1; gcii <= gcaa; gcii++) \
396 { \
397 gcss -= 6; \
398 c |= (eptr[gcii] & 0x3f) << gcss; \
399 } \
400 len += gcaa; \
401 }
402
403 /* If the pointer is not at the start of a character, move it back until
404 it is. Called only in UTF-8 mode. */
405
406 #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--;
407
408 #endif
409
410
411 /* In case there is no definition of offsetof() provided - though any proper
412 Standard C system should have one. */
413
414 #ifndef offsetof
415 #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))
416 #endif
417
418
419 /* These are the public options that can change during matching. */
420
421 #define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)
422
423 /* Private options flags start at the most significant end of the four bytes.
424 The public options defined in pcre.h start at the least significant end. Make
425 sure they don't overlap! The bits are getting a bit scarce now -- when we run
426 out, there is a dummy word in the structure that could be used for the private
427 bits. */
428
429 #define PCRE_NOPARTIAL 0x80000000 /* can't use partial with this regex */
430 #define PCRE_FIRSTSET 0x40000000 /* first_byte is set */
431 #define PCRE_REQCHSET 0x20000000 /* req_byte is set */
432 #define PCRE_STARTLINE 0x10000000 /* start after \n for multiline */
433 #define PCRE_JCHANGED 0x08000000 /* j option changes within regex */
434
435 /* Options for the "extra" block produced by pcre_study(). */
436
437 #define PCRE_STUDY_MAPPED 0x01 /* a map of starting chars exists */
438
439 /* Masks for identifying the public options that are permitted at compile
440 time, run time, or study time, respectively. */
441
442 #define PCRE_NEWLINE_BITS (PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|PCRE_NEWLINE_ANY)
443
444 #define PUBLIC_OPTIONS \
445 (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
446 PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
447 PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
448 PCRE_DUPNAMES|PCRE_NEWLINE_BITS)
449
450 #define PUBLIC_EXEC_OPTIONS \
451 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
452 PCRE_PARTIAL|PCRE_NEWLINE_BITS)
453
454 #define PUBLIC_DFA_EXEC_OPTIONS \
455 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
456 PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART|PCRE_NEWLINE_BITS)
457
458 #define PUBLIC_STUDY_OPTIONS 0 /* None defined */
459
460 /* Magic number to provide a small check against being handed junk. Also used
461 to detect whether a pattern was compiled on a host of different endianness. */
462
463 #define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */
464
465 /* Negative values for the firstchar and reqchar variables */
466
467 #define REQ_UNSET (-2)
468 #define REQ_NONE (-1)
469
470 /* The maximum remaining length of subject we are prepared to search for a
471 req_byte match. */
472
473 #define REQ_BYTE_MAX 1000
474
475 /* Flags added to firstbyte or reqbyte; a "non-literal" item is either a
476 variable-length repeat, or a anything other than literal characters. */
477
478 #define REQ_CASELESS 0x0100 /* indicates caselessness */
479 #define REQ_VARY 0x0200 /* reqbyte followed non-literal item */
480
481 /* Miscellaneous definitions */
482
483 typedef int BOOL;
484
485 #define FALSE 0
486 #define TRUE 1
487
488 /* Escape items that are just an encoding of a particular data value. */
489
490 #ifndef ESC_e
491 #define ESC_e 27
492 #endif
493
494 #ifndef ESC_f
495 #define ESC_f '\f'
496 #endif
497
498 #ifndef ESC_n
499 #define ESC_n '\n'
500 #endif
501
502 #ifndef ESC_r
503 #define ESC_r '\r'
504 #endif
505
506 /* We can't officially use ESC_t because it is a POSIX reserved identifier
507 (presumably because of all the others like size_t). */
508
509 #ifndef ESC_tee
510 #define ESC_tee '\t'
511 #endif
512
513 /* Codes for different types of Unicode property */
514
515 #define PT_ANY 0 /* Any property - matches all chars */
516 #define PT_LAMP 1 /* L& - the union of Lu, Ll, Lt */
517 #define PT_GC 2 /* General characteristic (e.g. L) */
518 #define PT_PC 3 /* Particular characteristic (e.g. Lu) */
519 #define PT_SC 4 /* Script (e.g. Han) */
520
521 /* Flag bits and data types for the extended class (OP_XCLASS) for classes that
522 contain UTF-8 characters with values greater than 255. */
523
524 #define XCL_NOT 0x01 /* Flag: this is a negative class */
525 #define XCL_MAP 0x02 /* Flag: a 32-byte map is present */
526
527 #define XCL_END 0 /* Marks end of individual items */
528 #define XCL_SINGLE 1 /* Single item (one multibyte char) follows */
529 #define XCL_RANGE 2 /* A range (two multibyte chars) follows */
530 #define XCL_PROP 3 /* Unicode property (2-byte property code follows) */
531 #define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */
532
533 /* These are escaped items that aren't just an encoding of a particular data
534 value such as \n. They must have non-zero values, as check_escape() returns
535 their negation. Also, they must appear in the same order as in the opcode
536 definitions below, up to ESC_z. There's a dummy for OP_ANY because it
537 corresponds to "." rather than an escape sequence. The final one must be
538 ESC_REF as subsequent values are used for backreferences (\1, \2, \3, etc).
539 There are two tests in the code for an escape greater than ESC_b and less than
540 ESC_Z to detect the types that may be repeated. These are the types that
541 consume characters. If any new escapes are put in between that don't consume a
542 character, that code will have to change. */
543
544 enum { ESC_A = 1, ESC_G, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W,
545 ESC_w, ESC_dum1, ESC_C, ESC_P, ESC_p, ESC_R, ESC_X, ESC_Z, ESC_z,
546 ESC_E, ESC_Q, ESC_k, ESC_REF };
547
548
549 /* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
550 that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
551 OP_EOD must correspond in order to the list of escapes immediately above.
552
553 To keep stored, compiled patterns compatible, new opcodes should be added
554 immediately before OP_BRA, where (since release 7.0) a gap is left for this
555 purpose.
556
557 *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions
558 that follow must also be updated to match. There is also a table called
559 "coptable" in pcre_dfa_exec.c that must be updated. */
560
561 enum {
562 OP_END, /* 0 End of pattern */
563
564 /* Values corresponding to backslashed metacharacters */
565
566 OP_SOD, /* 1 Start of data: \A */
567 OP_SOM, /* 2 Start of match (subject + offset): \G */
568 OP_NOT_WORD_BOUNDARY, /* 3 \B */
569 OP_WORD_BOUNDARY, /* 4 \b */
570 OP_NOT_DIGIT, /* 5 \D */
571 OP_DIGIT, /* 6 \d */
572 OP_NOT_WHITESPACE, /* 7 \S */
573 OP_WHITESPACE, /* 8 \s */
574 OP_NOT_WORDCHAR, /* 9 \W */
575 OP_WORDCHAR, /* 10 \w */
576 OP_ANY, /* 11 Match any character */
577 OP_ANYBYTE, /* 12 Match any byte (\C); different to OP_ANY for UTF-8 */
578 OP_NOTPROP, /* 13 \P (not Unicode property) */
579 OP_PROP, /* 14 \p (Unicode property) */
580 OP_ANYNL, /* 15 \R (any newline sequence) */
581 OP_EXTUNI, /* 16 \X (extended Unicode sequence */
582 OP_EODN, /* 17 End of data or \n at end of data: \Z. */
583 OP_EOD, /* 18 End of data: \z */
584
585 OP_OPT, /* 19 Set runtime options */
586 OP_CIRC, /* 20 Start of line - varies with multiline switch */
587 OP_DOLL, /* 21 End of line - varies with multiline switch */
588 OP_CHAR, /* 22 Match one character, casefully */
589 OP_CHARNC, /* 23 Match one character, caselessly */
590 OP_NOT, /* 24 Match one character, not the following one */
591
592 OP_STAR, /* 25 The maximizing and minimizing versions of */
593 OP_MINSTAR, /* 26 these six opcodes must come in pairs, with */
594 OP_PLUS, /* 27 the minimizing one second. */
595 OP_MINPLUS, /* 28 This first set applies to single characters.*/
596 OP_QUERY, /* 29 */
597 OP_MINQUERY, /* 30 */
598
599 OP_UPTO, /* 31 From 0 to n matches */
600 OP_MINUPTO, /* 32 */
601 OP_EXACT, /* 33 Exactly n matches */
602
603 OP_POSSTAR, /* 34 Possessified star */
604 OP_POSPLUS, /* 35 Possessified plus */
605 OP_POSQUERY, /* 36 Posesssified query */
606 OP_POSUPTO, /* 37 Possessified upto */
607
608 OP_NOTSTAR, /* 38 The maximizing and minimizing versions of */
609 OP_NOTMINSTAR, /* 39 these six opcodes must come in pairs, with */
610 OP_NOTPLUS, /* 40 the minimizing one second. They must be in */
611 OP_NOTMINPLUS, /* 41 exactly the same order as those above. */
612 OP_NOTQUERY, /* 42 This set applies to "not" single characters. */
613 OP_NOTMINQUERY, /* 43 */
614
615 OP_NOTUPTO, /* 44 From 0 to n matches */
616 OP_NOTMINUPTO, /* 45 */
617 OP_NOTEXACT, /* 46 Exactly n matches */
618
619 OP_NOTPOSSTAR, /* 47 Possessified versions */
620 OP_NOTPOSPLUS, /* 48 */
621 OP_NOTPOSQUERY, /* 49 */
622 OP_NOTPOSUPTO, /* 50 */
623
624 OP_TYPESTAR, /* 51 The maximizing and minimizing versions of */
625 OP_TYPEMINSTAR, /* 52 these six opcodes must come in pairs, with */
626 OP_TYPEPLUS, /* 53 the minimizing one second. These codes must */
627 OP_TYPEMINPLUS, /* 54 be in exactly the same order as those above. */
628 OP_TYPEQUERY, /* 55 This set applies to character types such as \d */
629 OP_TYPEMINQUERY, /* 56 */
630
631 OP_TYPEUPTO, /* 57 From 0 to n matches */
632 OP_TYPEMINUPTO, /* 58 */
633 OP_TYPEEXACT, /* 59 Exactly n matches */
634
635 OP_TYPEPOSSTAR, /* 60 Possessified versions */
636 OP_TYPEPOSPLUS, /* 61 */
637 OP_TYPEPOSQUERY, /* 62 */
638 OP_TYPEPOSUPTO, /* 63 */
639
640 OP_CRSTAR, /* 64 The maximizing and minimizing versions of */
641 OP_CRMINSTAR, /* 65 all these opcodes must come in pairs, with */
642 OP_CRPLUS, /* 66 the minimizing one second. These codes must */
643 OP_CRMINPLUS, /* 67 be in exactly the same order as those above. */
644 OP_CRQUERY, /* 68 These are for character classes and back refs */
645 OP_CRMINQUERY, /* 69 */
646 OP_CRRANGE, /* 70 These are different to the three sets above. */
647 OP_CRMINRANGE, /* 71 */
648
649 OP_CLASS, /* 72 Match a character class, chars < 256 only */
650 OP_NCLASS, /* 73 Same, but the bitmap was created from a negative
651 class - the difference is relevant only when a UTF-8
652 character > 255 is encountered. */
653
654 OP_XCLASS, /* 74 Extended class for handling UTF-8 chars within the
655 class. This does both positive and negative. */
656
657 OP_REF, /* 75 Match a back reference */
658 OP_RECURSE, /* 76 Match a numbered subpattern (possibly recursive) */
659 OP_CALLOUT, /* 77 Call out to external function if provided */
660
661 OP_ALT, /* 78 Start of alternation */
662 OP_KET, /* 79 End of group that doesn't have an unbounded repeat */
663 OP_KETRMAX, /* 80 These two must remain together and in this */
664 OP_KETRMIN, /* 81 order. They are for groups the repeat for ever. */
665
666 /* The assertions must come before BRA, CBRA, ONCE, and COND.*/
667
668 OP_ASSERT, /* 82 Positive lookahead */
669 OP_ASSERT_NOT, /* 83 Negative lookahead */
670 OP_ASSERTBACK, /* 84 Positive lookbehind */
671 OP_ASSERTBACK_NOT, /* 85 Negative lookbehind */
672 OP_REVERSE, /* 86 Move pointer back - used in lookbehind assertions */
673
674 /* ONCE, BRA, CBRA, and COND must come after the assertions, with ONCE first,
675 as there's a test for >= ONCE for a subpattern that isn't an assertion. */
676
677 OP_ONCE, /* 87 Atomic group */
678 OP_BRA, /* 88 Start of non-capturing bracket */
679 OP_CBRA, /* 89 Start of capturing bracket */
680 OP_COND, /* 90 Conditional group */
681
682 /* These three must follow the previous three, in the same order. There's a
683 check for >= SBRA to distinguish the two sets. */
684
685 OP_SBRA, /* 91 Start of non-capturing bracket, check empty */
686 OP_SCBRA, /* 92 Start of capturing bracket, check empty */
687 OP_SCOND, /* 93 Conditional group, check empty */
688
689 OP_CREF, /* 94 Used to hold a capture number as condition */
690 OP_RREF, /* 95 Used to hold a recursion number as condition */
691 OP_DEF, /* 96 The DEFINE condition */
692
693 OP_BRAZERO, /* 97 These two must remain together and in this */
694 OP_BRAMINZERO /* 98 order. */
695 };
696
697
698 /* This macro defines textual names for all the opcodes. These are used only
699 for debugging. The macro is referenced only in pcre_printint.c. */
700
701 #define OP_NAME_LIST \
702 "End", "\\A", "\\G", "\\B", "\\b", "\\D", "\\d", \
703 "\\S", "\\s", "\\W", "\\w", "Any", "Anybyte", \
704 "notprop", "prop", "anynl", "extuni", \
705 "\\Z", "\\z", \
706 "Opt", "^", "$", "char", "charnc", "not", \
707 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
708 "*+","++", "?+", "{", \
709 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
710 "*+","++", "?+", "{", \
711 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
712 "*+","++", "?+", "{", \
713 "*", "*?", "+", "+?", "?", "??", "{", "{", \
714 "class", "nclass", "xclass", "Ref", "Recurse", "Callout", \
715 "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \
716 "AssertB", "AssertB not", "Reverse", \
717 "Once", "Bra 0", "Bra", "Cond", "SBra 0", "SBra", "SCond", \
718 "Cond ref", "Cond rec", "Cond def", "Brazero", "Braminzero"
719
720
721 /* This macro defines the length of fixed length operations in the compiled
722 regex. The lengths are used when searching for specific things, and also in the
723 debugging printing of a compiled regex. We use a macro so that it can be
724 defined close to the definitions of the opcodes themselves.
725
726 As things have been extended, some of these are no longer fixed lenths, but are
727 minima instead. For example, the length of a single-character repeat may vary
728 in UTF-8 mode. The code that uses this table must know about such things. */
729
730 #define OP_LENGTHS \
731 1, /* End */ \
732 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* \A, \G, \B, \B, \D, \d, \S, \s, \W, \w */ \
733 1, 1, /* Any, Anybyte */ \
734 3, 3, 1, 1, /* NOTPROP, PROP, EXTUNI, ANYNL */ \
735 1, 1, 2, 1, 1, /* \Z, \z, Opt, ^, $ */ \
736 2, /* Char - the minimum length */ \
737 2, /* Charnc - the minimum length */ \
738 2, /* not */ \
739 /* Positive single-char repeats ** These are */ \
740 2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \
741 4, 4, 4, /* upto, minupto, exact ** UTF-8 mode */ \
742 2, 2, 2, 4, /* *+, ++, ?+, upto+ */ \
743 /* Negative single-char repeats - only for chars < 256 */ \
744 2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \
745 4, 4, 4, /* NOT upto, minupto, exact */ \
746 2, 2, 2, 4, /* Possessive *, +, ?, upto */ \
747 /* Positive type repeats */ \
748 2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \
749 4, 4, 4, /* Type upto, minupto, exact */ \
750 2, 2, 2, 4, /* Possessive *+, ++, ?+, upto+ */ \
751 /* Character class & ref repeats */ \
752 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \
753 5, 5, /* CRRANGE, CRMINRANGE */ \
754 33, /* CLASS */ \
755 33, /* NCLASS */ \
756 0, /* XCLASS - variable length */ \
757 3, /* REF */ \
758 1+LINK_SIZE, /* RECURSE */ \
759 2+2*LINK_SIZE, /* CALLOUT */ \
760 1+LINK_SIZE, /* Alt */ \
761 1+LINK_SIZE, /* Ket */ \
762 1+LINK_SIZE, /* KetRmax */ \
763 1+LINK_SIZE, /* KetRmin */ \
764 1+LINK_SIZE, /* Assert */ \
765 1+LINK_SIZE, /* Assert not */ \
766 1+LINK_SIZE, /* Assert behind */ \
767 1+LINK_SIZE, /* Assert behind not */ \
768 1+LINK_SIZE, /* Reverse */ \
769 1+LINK_SIZE, /* ONCE */ \
770 1+LINK_SIZE, /* BRA */ \
771 3+LINK_SIZE, /* CBRA */ \
772 1+LINK_SIZE, /* COND */ \
773 1+LINK_SIZE, /* SBRA */ \
774 3+LINK_SIZE, /* SCBRA */ \
775 1+LINK_SIZE, /* SCOND */ \
776 3, /* CREF */ \
777 3, /* RREF */ \
778 1, /* DEF */ \
779 1, 1, /* BRAZERO, BRAMINZERO */ \
780
781
782 /* A magic value for OP_RREF to indicate the "any recursion" condition. */
783
784 #define RREF_ANY 0xffff
785
786 /* Error code numbers. They are given names so that they can more easily be
787 tracked. */
788
789 enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9,
790 ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19,
791 ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29,
792 ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
793 ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
794 ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57 };
795
796 /* The real format of the start of the pcre block; the index of names and the
797 code vector run on as long as necessary after the end. We store an explicit
798 offset to the name table so that if a regex is compiled on one host, saved, and
799 then run on another where the size of pointers is different, all might still
800 be well. For the case of compiled-on-4 and run-on-8, we include an extra
801 pointer that is always NULL. For future-proofing, a few dummy fields were
802 originally included - even though you can never get this planning right - but
803 there is only one left now.
804
805 NOTE NOTE NOTE:
806 Because people can now save and re-use compiled patterns, any additions to this
807 structure should be made at the end, and something earlier (e.g. a new
808 flag in the options or one of the dummy fields) should indicate that the new
809 fields are present. Currently PCRE always sets the dummy fields to zero.
810 NOTE NOTE NOTE:
811 */
812
813 typedef struct real_pcre {
814 pcre_uint32 magic_number;
815 pcre_uint32 size; /* Total that was malloced */
816 pcre_uint32 options;
817 pcre_uint32 dummy1; /* For future use, maybe */
818
819 pcre_uint16 top_bracket;
820 pcre_uint16 top_backref;
821 pcre_uint16 first_byte;
822 pcre_uint16 req_byte;
823 pcre_uint16 name_table_offset; /* Offset to name table that follows */
824 pcre_uint16 name_entry_size; /* Size of any name items */
825 pcre_uint16 name_count; /* Number of name items */
826 pcre_uint16 ref_count; /* Reference count */
827
828 const unsigned char *tables; /* Pointer to tables or NULL for std */
829 const unsigned char *nullpad; /* NULL padding */
830 } real_pcre;
831
832 /* The format of the block used to store data from pcre_study(). The same
833 remark (see NOTE above) about extending this structure applies. */
834
835 typedef struct pcre_study_data {
836 pcre_uint32 size; /* Total that was malloced */
837 pcre_uint32 options;
838 uschar start_bits[32];
839 } pcre_study_data;
840
841 /* Structure for passing "static" information around between the functions
842 doing the compiling, so that they are thread-safe. */
843
844 typedef struct compile_data {
845 const uschar *lcc; /* Points to lower casing table */
846 const uschar *fcc; /* Points to case-flipping table */
847 const uschar *cbits; /* Points to character type table */
848 const uschar *ctypes; /* Points to table of type maps */
849 const uschar *start_workspace;/* The start of working space */
850 const uschar *start_code; /* The start of the compiled code */
851 const uschar *start_pattern; /* The start of the pattern */
852 const uschar *end_pattern; /* The end of the pattern */
853 uschar *hwm; /* High watermark of workspace */
854 uschar *name_table; /* The name/number table */
855 int names_found; /* Number of entries so far */
856 int name_entry_size; /* Size of each entry */
857 int bracount; /* Count of capturing parens */
858 int top_backref; /* Maximum back reference */
859 unsigned int backref_map; /* Bitmap of low back refs */
860 int external_options; /* External (initial) options */
861 int req_varyopt; /* "After variable item" flag for reqbyte */
862 BOOL nopartial; /* Set TRUE if partial won't work */
863 int nltype; /* Newline type */
864 int nllen; /* Newline string length */
865 uschar nl[4]; /* Newline string when fixed length */
866 } compile_data;
867
868 /* Structure for maintaining a chain of pointers to the currently incomplete
869 branches, for testing for left recursion. */
870
871 typedef struct branch_chain {
872 struct branch_chain *outer;
873 uschar *current;
874 } branch_chain;
875
876 /* Structure for items in a linked list that represents an explicit recursive
877 call within the pattern. */
878
879 typedef struct recursion_info {
880 struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
881 int group_num; /* Number of group that was called */
882 const uschar *after_call; /* "Return value": points after the call in the expr */
883 USPTR save_start; /* Old value of md->start_match */
884 int *offset_save; /* Pointer to start of saved offsets */
885 int saved_max; /* Number of saved offsets */
886 } recursion_info;
887
888 /* When compiling in a mode that doesn't use recursive calls to match(),
889 a structure is used to remember local variables on the heap. It is defined in
890 pcre_exec.c, close to the match() function, so that it is easy to keep it in
891 step with any changes of local variable. However, the pointer to the current
892 frame must be saved in some "static" place over a longjmp(). We declare the
893 structure here so that we can put a pointer in the match_data structure. NOTE:
894 This isn't used for a "normal" compilation of pcre. */
895
896 struct heapframe;
897
898 /* Structure for building a chain of data for holding the values of the subject
899 pointer at the start of each subpattern, so as to detect when an empty string
900 has been matched by a subpattern - to break infinite loops. */
901
902 typedef struct eptrblock {
903 struct eptrblock *epb_prev;
904 USPTR epb_saved_eptr;
905 } eptrblock;
906
907
908 /* Structure for passing "static" information around between the functions
909 doing traditional NFA matching, so that they are thread-safe. */
910
911 typedef struct match_data {
912 unsigned long int match_call_count; /* As it says */
913 unsigned long int match_limit; /* As it says */
914 unsigned long int match_limit_recursion; /* As it says */
915 int *offset_vector; /* Offset vector */
916 int offset_end; /* One past the end */
917 int offset_max; /* The maximum usable for return data */
918 int nltype; /* Newline type */
919 int nllen; /* Newline string length */
920 uschar nl[4]; /* Newline string when fixed */
921 const uschar *lcc; /* Points to lower casing table */
922 const uschar *ctypes; /* Points to table of type maps */
923 BOOL offset_overflow; /* Set if too many extractions */
924 BOOL notbol; /* NOTBOL flag */
925 BOOL noteol; /* NOTEOL flag */
926 BOOL utf8; /* UTF8 flag */
927 BOOL endonly; /* Dollar not before final \n */
928 BOOL notempty; /* Empty string match not wanted */
929 BOOL partial; /* PARTIAL flag */
930 BOOL hitend; /* Hit the end of the subject at some point */
931 const uschar *start_code; /* For use when recursing */
932 USPTR start_subject; /* Start of the subject string */
933 USPTR end_subject; /* End of the subject string */
934 USPTR start_match; /* Start of this match attempt */
935 USPTR end_match_ptr; /* Subject position at end match */
936 int end_offset_top; /* Highwater mark at end of match */
937 int capture_last; /* Most recent capture number */
938 int start_offset; /* The start offset value */
939 eptrblock *eptrchain; /* Chain of eptrblocks for tail recursions */
940 int eptrn; /* Next free eptrblock */
941 recursion_info *recursive; /* Linked list of recursion data */
942 void *callout_data; /* To pass back to callouts */
943 struct heapframe *thisframe; /* Used only when compiling for no recursion */
944 } match_data;
945
946 /* A similar structure is used for the same purpose by the DFA matching
947 functions. */
948
949 typedef struct dfa_match_data {
950 const uschar *start_code; /* Start of the compiled pattern */
951 const uschar *start_subject; /* Start of the subject string */
952 const uschar *end_subject; /* End of subject string */
953 const uschar *tables; /* Character tables */
954 int moptions; /* Match options */
955 int poptions; /* Pattern options */
956 int nltype; /* Newline type */
957 int nllen; /* Newline string length */
958 uschar nl[4]; /* Newline string when fixed */
959 void *callout_data; /* To pass back to callouts */
960 } dfa_match_data;
961
962 /* Bit definitions for entries in the pcre_ctypes table. */
963
964 #define ctype_space 0x01
965 #define ctype_letter 0x02
966 #define ctype_digit 0x04
967 #define ctype_xdigit 0x08
968 #define ctype_word 0x10 /* alphameric or '_' */
969 #define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
970
971 /* Offsets for the bitmap tables in pcre_cbits. Each table contains a set
972 of bits for a class map. Some classes are built by combining these tables. */
973
974 #define cbit_space 0 /* [:space:] or \s */
975 #define cbit_xdigit 32 /* [:xdigit:] */
976 #define cbit_digit 64 /* [:digit:] or \d */
977 #define cbit_upper 96 /* [:upper:] */
978 #define cbit_lower 128 /* [:lower:] */
979 #define cbit_word 160 /* [:word:] or \w */
980 #define cbit_graph 192 /* [:graph:] */
981 #define cbit_print 224 /* [:print:] */
982 #define cbit_punct 256 /* [:punct:] */
983 #define cbit_cntrl 288 /* [:cntrl:] */
984 #define cbit_length 320 /* Length of the cbits table */
985
986 /* Offsets of the various tables from the base tables pointer, and
987 total length. */
988
989 #define lcc_offset 0
990 #define fcc_offset 256
991 #define cbits_offset 512
992 #define ctypes_offset (cbits_offset + cbit_length)
993 #define tables_length (ctypes_offset + 256)
994
995 /* Layout of the UCP type table that translates property names into types and
996 codes. */
997
998 typedef struct {
999 const char *name;
1000 pcre_uint16 type;
1001 pcre_uint16 value;
1002 } ucp_type_table;
1003
1004
1005 /* Internal shared data tables. These are tables that are used by more than one
1006 of the exported public functions. They have to be "external" in the C sense,
1007 but are not part of the PCRE public API. The data for these tables is in the
1008 pcre_tables.c module. */
1009
1010 extern const int _pcre_utf8_table1[];
1011 extern const int _pcre_utf8_table2[];
1012 extern const int _pcre_utf8_table3[];
1013 extern const uschar _pcre_utf8_table4[];
1014
1015 extern const int _pcre_utf8_table1_size;
1016
1017 extern const ucp_type_table _pcre_utt[];
1018 extern const int _pcre_utt_size;
1019
1020 extern const uschar _pcre_default_tables[];
1021
1022 extern const uschar _pcre_OP_lengths[];
1023
1024
1025 /* Internal shared functions. These are functions that are used by more than
1026 one of the exported public functions. They have to be "external" in the C
1027 sense, but are not part of the PCRE public API. */
1028
1029 extern BOOL _pcre_is_newline(const uschar *, const uschar *, int *,
1030 BOOL);
1031 extern int _pcre_ord2utf8(int, uschar *);
1032 extern real_pcre *_pcre_try_flipped(const real_pcre *, real_pcre *,
1033 const pcre_study_data *, pcre_study_data *);
1034 extern int _pcre_ucp_findprop(const unsigned int, int *, int *);
1035 extern unsigned int _pcre_ucp_othercase(const unsigned int);
1036 extern int _pcre_valid_utf8(const uschar *, int);
1037 extern BOOL _pcre_was_newline(const uschar *, const uschar *, int *,
1038 BOOL);
1039 extern BOOL _pcre_xclass(int, const uschar *);
1040
1041 #endif
1042
1043 /* End of pcre_internal.h */

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12