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

Contents of /code/trunk/pcre_internal.h

Parent Directory Parent Directory | Revision Log Revision Log


Revision 117 - (show annotations) (download)
Fri Mar 9 15:59:06 2007 UTC (7 years, 6 months ago) by ph10
File MIME type: text/plain
File size: 42903 byte(s)
Update copyright years to 2007.

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) 1997-2007 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 strlen(s) _strlen(s)
193 #define strncmp(s1,s2,m) _strncmp(s1,s2,m)
194 #define memcmp(s,c,n) _memcmp(s,c,n)
195 #define memcpy(d,s,n) _memcpy(d,s,n)
196 #define memmove(d,s,n) _memmove(d,s,n)
197 #define memset(s,c,n) _memset(s,c,n)
198 #else /* VPCOMPAT */
199
200 /* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
201 define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY
202 is set. Otherwise, include an emulating function for those systems that have
203 neither (there some non-Unix environments where this is the case). This assumes
204 that all calls to memmove are moving strings upwards in store, which is the
205 case in PCRE. */
206
207 #if ! HAVE_MEMMOVE
208 #undef memmove /* some systems may have a macro */
209 #if HAVE_BCOPY
210 #define memmove(a, b, c) bcopy(b, a, c)
211 #else /* HAVE_BCOPY */
212 static void *
213 pcre_memmove(void *d, const void *s, size_t n)
214 {
215 size_t i;
216 unsigned char *dest = (unsigned char *)d;
217 const unsigned char *src = (const unsigned char *)s;
218 dest += n;
219 src += n;
220 for (i = 0; i < n; ++i) *(--dest) = *(--src);
221 return (void *)dest;
222 }
223 #define memmove(a, b, c) pcre_memmove(a, b, c)
224 #endif /* not HAVE_BCOPY */
225 #endif /* not HAVE_MEMMOVE */
226 #endif /* not VPCOMPAT */
227
228
229 /* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored
230 in big-endian order) by default. These are used, for example, to link from the
231 start of a subpattern to its alternatives and its end. The use of 2 bytes per
232 offset limits the size of the compiled regex to around 64K, which is big enough
233 for almost everybody. However, I received a request for an even bigger limit.
234 For this reason, and also to make the code easier to maintain, the storing and
235 loading of offsets from the byte string is now handled by the macros that are
236 defined here.
237
238 The macros are controlled by the value of LINK_SIZE. This defaults to 2 in
239 the config.h file, but can be overridden by using -D on the command line. This
240 is automated on Unix systems via the "configure" command. */
241
242 #if LINK_SIZE == 2
243
244 #define PUT(a,n,d) \
245 (a[n] = (d) >> 8), \
246 (a[(n)+1] = (d) & 255)
247
248 #define GET(a,n) \
249 (((a)[n] << 8) | (a)[(n)+1])
250
251 #define MAX_PATTERN_SIZE (1 << 16)
252
253
254 #elif LINK_SIZE == 3
255
256 #define PUT(a,n,d) \
257 (a[n] = (d) >> 16), \
258 (a[(n)+1] = (d) >> 8), \
259 (a[(n)+2] = (d) & 255)
260
261 #define GET(a,n) \
262 (((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2])
263
264 #define MAX_PATTERN_SIZE (1 << 24)
265
266
267 #elif LINK_SIZE == 4
268
269 #define PUT(a,n,d) \
270 (a[n] = (d) >> 24), \
271 (a[(n)+1] = (d) >> 16), \
272 (a[(n)+2] = (d) >> 8), \
273 (a[(n)+3] = (d) & 255)
274
275 #define GET(a,n) \
276 (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3])
277
278 #define MAX_PATTERN_SIZE (1 << 30) /* Keep it positive */
279
280
281 #else
282 #error LINK_SIZE must be either 2, 3, or 4
283 #endif
284
285
286 /* Convenience macro defined in terms of the others */
287
288 #define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE
289
290
291 /* PCRE uses some other 2-byte quantities that do not change when the size of
292 offsets changes. There are used for repeat counts and for other things such as
293 capturing parenthesis numbers in back references. */
294
295 #define PUT2(a,n,d) \
296 a[n] = (d) >> 8; \
297 a[(n)+1] = (d) & 255
298
299 #define GET2(a,n) \
300 (((a)[n] << 8) | (a)[(n)+1])
301
302 #define PUT2INC(a,n,d) PUT2(a,n,d), a += 2
303
304
305 /* When UTF-8 encoding is being used, a character is no longer just a single
306 byte. The macros for character handling generate simple sequences when used in
307 byte-mode, and more complicated ones for UTF-8 characters. */
308
309 #ifndef SUPPORT_UTF8
310 #define GETCHAR(c, eptr) c = *eptr;
311 #define GETCHARTEST(c, eptr) c = *eptr;
312 #define GETCHARINC(c, eptr) c = *eptr++;
313 #define GETCHARINCTEST(c, eptr) c = *eptr++;
314 #define GETCHARLEN(c, eptr, len) c = *eptr;
315 #define BACKCHAR(eptr)
316
317 #else /* SUPPORT_UTF8 */
318
319 /* Get the next UTF-8 character, not advancing the pointer. This is called when
320 we know we are in UTF-8 mode. */
321
322 #define GETCHAR(c, eptr) \
323 c = *eptr; \
324 if (c >= 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 }
336
337 /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
338 pointer. */
339
340 #define GETCHARTEST(c, eptr) \
341 c = *eptr; \
342 if (utf8 && c >= 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 }
354
355 /* Get the next UTF-8 character, advancing the pointer. This is called when we
356 know we are in UTF-8 mode. */
357
358 #define GETCHARINC(c, eptr) \
359 c = *eptr++; \
360 if (c >= 0xc0) \
361 { \
362 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
363 int gcss = 6*gcaa; \
364 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
365 while (gcaa-- > 0) \
366 { \
367 gcss -= 6; \
368 c |= (*eptr++ & 0x3f) << gcss; \
369 } \
370 }
371
372 /* Get the next character, testing for UTF-8 mode, and advancing the pointer */
373
374 #define GETCHARINCTEST(c, eptr) \
375 c = *eptr++; \
376 if (utf8 && c >= 0xc0) \
377 { \
378 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
379 int gcss = 6*gcaa; \
380 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
381 while (gcaa-- > 0) \
382 { \
383 gcss -= 6; \
384 c |= (*eptr++ & 0x3f) << gcss; \
385 } \
386 }
387
388 /* Get the next UTF-8 character, not advancing the pointer, incrementing length
389 if there are extra bytes. This is called when we know we are in UTF-8 mode. */
390
391 #define GETCHARLEN(c, eptr, len) \
392 c = *eptr; \
393 if (c >= 0xc0) \
394 { \
395 int gcii; \
396 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
397 int gcss = 6*gcaa; \
398 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
399 for (gcii = 1; gcii <= gcaa; gcii++) \
400 { \
401 gcss -= 6; \
402 c |= (eptr[gcii] & 0x3f) << gcss; \
403 } \
404 len += gcaa; \
405 }
406
407 /* If the pointer is not at the start of a character, move it back until
408 it is. Called only in UTF-8 mode. */
409
410 #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--;
411
412 #endif
413
414
415 /* In case there is no definition of offsetof() provided - though any proper
416 Standard C system should have one. */
417
418 #ifndef offsetof
419 #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))
420 #endif
421
422
423 /* These are the public options that can change during matching. */
424
425 #define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)
426
427 /* Private options flags start at the most significant end of the four bytes.
428 The public options defined in pcre.h start at the least significant end. Make
429 sure they don't overlap! The bits are getting a bit scarce now -- when we run
430 out, there is a dummy word in the structure that could be used for the private
431 bits. */
432
433 #define PCRE_NOPARTIAL 0x80000000 /* can't use partial with this regex */
434 #define PCRE_FIRSTSET 0x40000000 /* first_byte is set */
435 #define PCRE_REQCHSET 0x20000000 /* req_byte is set */
436 #define PCRE_STARTLINE 0x10000000 /* start after \n for multiline */
437 #define PCRE_JCHANGED 0x08000000 /* j option changes within regex */
438
439 /* Options for the "extra" block produced by pcre_study(). */
440
441 #define PCRE_STUDY_MAPPED 0x01 /* a map of starting chars exists */
442
443 /* Masks for identifying the public options that are permitted at compile
444 time, run time, or study time, respectively. */
445
446 #define PCRE_NEWLINE_BITS (PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|PCRE_NEWLINE_ANY)
447
448 #define PUBLIC_OPTIONS \
449 (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
450 PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
451 PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
452 PCRE_DUPNAMES|PCRE_NEWLINE_BITS)
453
454 #define PUBLIC_EXEC_OPTIONS \
455 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
456 PCRE_PARTIAL|PCRE_NEWLINE_BITS)
457
458 #define PUBLIC_DFA_EXEC_OPTIONS \
459 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
460 PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART|PCRE_NEWLINE_BITS)
461
462 #define PUBLIC_STUDY_OPTIONS 0 /* None defined */
463
464 /* Magic number to provide a small check against being handed junk. Also used
465 to detect whether a pattern was compiled on a host of different endianness. */
466
467 #define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */
468
469 /* Negative values for the firstchar and reqchar variables */
470
471 #define REQ_UNSET (-2)
472 #define REQ_NONE (-1)
473
474 /* The maximum remaining length of subject we are prepared to search for a
475 req_byte match. */
476
477 #define REQ_BYTE_MAX 1000
478
479 /* Flags added to firstbyte or reqbyte; a "non-literal" item is either a
480 variable-length repeat, or a anything other than literal characters. */
481
482 #define REQ_CASELESS 0x0100 /* indicates caselessness */
483 #define REQ_VARY 0x0200 /* reqbyte followed non-literal item */
484
485 /* Miscellaneous definitions */
486
487 typedef int BOOL;
488
489 #define FALSE 0
490 #define TRUE 1
491
492 /* Escape items that are just an encoding of a particular data value. */
493
494 #ifndef ESC_e
495 #define ESC_e 27
496 #endif
497
498 #ifndef ESC_f
499 #define ESC_f '\f'
500 #endif
501
502 #ifndef ESC_n
503 #define ESC_n '\n'
504 #endif
505
506 #ifndef ESC_r
507 #define ESC_r '\r'
508 #endif
509
510 /* We can't officially use ESC_t because it is a POSIX reserved identifier
511 (presumably because of all the others like size_t). */
512
513 #ifndef ESC_tee
514 #define ESC_tee '\t'
515 #endif
516
517 /* Codes for different types of Unicode property */
518
519 #define PT_ANY 0 /* Any property - matches all chars */
520 #define PT_LAMP 1 /* L& - the union of Lu, Ll, Lt */
521 #define PT_GC 2 /* General characteristic (e.g. L) */
522 #define PT_PC 3 /* Particular characteristic (e.g. Lu) */
523 #define PT_SC 4 /* Script (e.g. Han) */
524
525 /* Flag bits and data types for the extended class (OP_XCLASS) for classes that
526 contain UTF-8 characters with values greater than 255. */
527
528 #define XCL_NOT 0x01 /* Flag: this is a negative class */
529 #define XCL_MAP 0x02 /* Flag: a 32-byte map is present */
530
531 #define XCL_END 0 /* Marks end of individual items */
532 #define XCL_SINGLE 1 /* Single item (one multibyte char) follows */
533 #define XCL_RANGE 2 /* A range (two multibyte chars) follows */
534 #define XCL_PROP 3 /* Unicode property (2-byte property code follows) */
535 #define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */
536
537 /* These are escaped items that aren't just an encoding of a particular data
538 value such as \n. They must have non-zero values, as check_escape() returns
539 their negation. Also, they must appear in the same order as in the opcode
540 definitions below, up to ESC_z. There's a dummy for OP_ANY because it
541 corresponds to "." rather than an escape sequence. The final one must be
542 ESC_REF as subsequent values are used for backreferences (\1, \2, \3, etc).
543 There are two tests in the code for an escape greater than ESC_b and less than
544 ESC_Z to detect the types that may be repeated. These are the types that
545 consume characters. If any new escapes are put in between that don't consume a
546 character, that code will have to change. */
547
548 enum { ESC_A = 1, ESC_G, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s, ESC_W,
549 ESC_w, ESC_dum1, ESC_C, ESC_P, ESC_p, ESC_R, ESC_X, ESC_Z, ESC_z,
550 ESC_E, ESC_Q, ESC_k, ESC_REF };
551
552
553 /* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
554 that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
555 OP_EOD must correspond in order to the list of escapes immediately above.
556
557 To keep stored, compiled patterns compatible, new opcodes should be added
558 immediately before OP_BRA, where (since release 7.0) a gap is left for this
559 purpose.
560
561 *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions
562 that follow must also be updated to match. There is also a table called
563 "coptable" in pcre_dfa_exec.c that must be updated. */
564
565 enum {
566 OP_END, /* 0 End of pattern */
567
568 /* Values corresponding to backslashed metacharacters */
569
570 OP_SOD, /* 1 Start of data: \A */
571 OP_SOM, /* 2 Start of match (subject + offset): \G */
572 OP_NOT_WORD_BOUNDARY, /* 3 \B */
573 OP_WORD_BOUNDARY, /* 4 \b */
574 OP_NOT_DIGIT, /* 5 \D */
575 OP_DIGIT, /* 6 \d */
576 OP_NOT_WHITESPACE, /* 7 \S */
577 OP_WHITESPACE, /* 8 \s */
578 OP_NOT_WORDCHAR, /* 9 \W */
579 OP_WORDCHAR, /* 10 \w */
580 OP_ANY, /* 11 Match any character */
581 OP_ANYBYTE, /* 12 Match any byte (\C); different to OP_ANY for UTF-8 */
582 OP_NOTPROP, /* 13 \P (not Unicode property) */
583 OP_PROP, /* 14 \p (Unicode property) */
584 OP_ANYNL, /* 15 \R (any newline sequence) */
585 OP_EXTUNI, /* 16 \X (extended Unicode sequence */
586 OP_EODN, /* 17 End of data or \n at end of data: \Z. */
587 OP_EOD, /* 18 End of data: \z */
588
589 OP_OPT, /* 19 Set runtime options */
590 OP_CIRC, /* 20 Start of line - varies with multiline switch */
591 OP_DOLL, /* 21 End of line - varies with multiline switch */
592 OP_CHAR, /* 22 Match one character, casefully */
593 OP_CHARNC, /* 23 Match one character, caselessly */
594 OP_NOT, /* 24 Match one character, not the following one */
595
596 OP_STAR, /* 25 The maximizing and minimizing versions of */
597 OP_MINSTAR, /* 26 these six opcodes must come in pairs, with */
598 OP_PLUS, /* 27 the minimizing one second. */
599 OP_MINPLUS, /* 28 This first set applies to single characters.*/
600 OP_QUERY, /* 29 */
601 OP_MINQUERY, /* 30 */
602
603 OP_UPTO, /* 31 From 0 to n matches */
604 OP_MINUPTO, /* 32 */
605 OP_EXACT, /* 33 Exactly n matches */
606
607 OP_POSSTAR, /* 34 Possessified star */
608 OP_POSPLUS, /* 35 Possessified plus */
609 OP_POSQUERY, /* 36 Posesssified query */
610 OP_POSUPTO, /* 37 Possessified upto */
611
612 OP_NOTSTAR, /* 38 The maximizing and minimizing versions of */
613 OP_NOTMINSTAR, /* 39 these six opcodes must come in pairs, with */
614 OP_NOTPLUS, /* 40 the minimizing one second. They must be in */
615 OP_NOTMINPLUS, /* 41 exactly the same order as those above. */
616 OP_NOTQUERY, /* 42 This set applies to "not" single characters. */
617 OP_NOTMINQUERY, /* 43 */
618
619 OP_NOTUPTO, /* 44 From 0 to n matches */
620 OP_NOTMINUPTO, /* 45 */
621 OP_NOTEXACT, /* 46 Exactly n matches */
622
623 OP_NOTPOSSTAR, /* 47 Possessified versions */
624 OP_NOTPOSPLUS, /* 48 */
625 OP_NOTPOSQUERY, /* 49 */
626 OP_NOTPOSUPTO, /* 50 */
627
628 OP_TYPESTAR, /* 51 The maximizing and minimizing versions of */
629 OP_TYPEMINSTAR, /* 52 these six opcodes must come in pairs, with */
630 OP_TYPEPLUS, /* 53 the minimizing one second. These codes must */
631 OP_TYPEMINPLUS, /* 54 be in exactly the same order as those above. */
632 OP_TYPEQUERY, /* 55 This set applies to character types such as \d */
633 OP_TYPEMINQUERY, /* 56 */
634
635 OP_TYPEUPTO, /* 57 From 0 to n matches */
636 OP_TYPEMINUPTO, /* 58 */
637 OP_TYPEEXACT, /* 59 Exactly n matches */
638
639 OP_TYPEPOSSTAR, /* 60 Possessified versions */
640 OP_TYPEPOSPLUS, /* 61 */
641 OP_TYPEPOSQUERY, /* 62 */
642 OP_TYPEPOSUPTO, /* 63 */
643
644 OP_CRSTAR, /* 64 The maximizing and minimizing versions of */
645 OP_CRMINSTAR, /* 65 all these opcodes must come in pairs, with */
646 OP_CRPLUS, /* 66 the minimizing one second. These codes must */
647 OP_CRMINPLUS, /* 67 be in exactly the same order as those above. */
648 OP_CRQUERY, /* 68 These are for character classes and back refs */
649 OP_CRMINQUERY, /* 69 */
650 OP_CRRANGE, /* 70 These are different to the three sets above. */
651 OP_CRMINRANGE, /* 71 */
652
653 OP_CLASS, /* 72 Match a character class, chars < 256 only */
654 OP_NCLASS, /* 73 Same, but the bitmap was created from a negative
655 class - the difference is relevant only when a UTF-8
656 character > 255 is encountered. */
657
658 OP_XCLASS, /* 74 Extended class for handling UTF-8 chars within the
659 class. This does both positive and negative. */
660
661 OP_REF, /* 75 Match a back reference */
662 OP_RECURSE, /* 76 Match a numbered subpattern (possibly recursive) */
663 OP_CALLOUT, /* 77 Call out to external function if provided */
664
665 OP_ALT, /* 78 Start of alternation */
666 OP_KET, /* 79 End of group that doesn't have an unbounded repeat */
667 OP_KETRMAX, /* 80 These two must remain together and in this */
668 OP_KETRMIN, /* 81 order. They are for groups the repeat for ever. */
669
670 /* The assertions must come before BRA, CBRA, ONCE, and COND.*/
671
672 OP_ASSERT, /* 82 Positive lookahead */
673 OP_ASSERT_NOT, /* 83 Negative lookahead */
674 OP_ASSERTBACK, /* 84 Positive lookbehind */
675 OP_ASSERTBACK_NOT, /* 85 Negative lookbehind */
676 OP_REVERSE, /* 86 Move pointer back - used in lookbehind assertions */
677
678 /* ONCE, BRA, CBRA, and COND must come after the assertions, with ONCE first,
679 as there's a test for >= ONCE for a subpattern that isn't an assertion. */
680
681 OP_ONCE, /* 87 Atomic group */
682 OP_BRA, /* 88 Start of non-capturing bracket */
683 OP_CBRA, /* 89 Start of capturing bracket */
684 OP_COND, /* 90 Conditional group */
685
686 /* These three must follow the previous three, in the same order. There's a
687 check for >= SBRA to distinguish the two sets. */
688
689 OP_SBRA, /* 91 Start of non-capturing bracket, check empty */
690 OP_SCBRA, /* 92 Start of capturing bracket, check empty */
691 OP_SCOND, /* 93 Conditional group, check empty */
692
693 OP_CREF, /* 94 Used to hold a capture number as condition */
694 OP_RREF, /* 95 Used to hold a recursion number as condition */
695 OP_DEF, /* 96 The DEFINE condition */
696
697 OP_BRAZERO, /* 97 These two must remain together and in this */
698 OP_BRAMINZERO /* 98 order. */
699 };
700
701
702 /* This macro defines textual names for all the opcodes. These are used only
703 for debugging. The macro is referenced only in pcre_printint.c. */
704
705 #define OP_NAME_LIST \
706 "End", "\\A", "\\G", "\\B", "\\b", "\\D", "\\d", \
707 "\\S", "\\s", "\\W", "\\w", "Any", "Anybyte", \
708 "notprop", "prop", "anynl", "extuni", \
709 "\\Z", "\\z", \
710 "Opt", "^", "$", "char", "charnc", "not", \
711 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
712 "*+","++", "?+", "{", \
713 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
714 "*+","++", "?+", "{", \
715 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
716 "*+","++", "?+", "{", \
717 "*", "*?", "+", "+?", "?", "??", "{", "{", \
718 "class", "nclass", "xclass", "Ref", "Recurse", "Callout", \
719 "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \
720 "AssertB", "AssertB not", "Reverse", \
721 "Once", "Bra 0", "Bra", "Cond", "SBra 0", "SBra", "SCond", \
722 "Cond ref", "Cond rec", "Cond def", "Brazero", "Braminzero"
723
724
725 /* This macro defines the length of fixed length operations in the compiled
726 regex. The lengths are used when searching for specific things, and also in the
727 debugging printing of a compiled regex. We use a macro so that it can be
728 defined close to the definitions of the opcodes themselves.
729
730 As things have been extended, some of these are no longer fixed lenths, but are
731 minima instead. For example, the length of a single-character repeat may vary
732 in UTF-8 mode. The code that uses this table must know about such things. */
733
734 #define OP_LENGTHS \
735 1, /* End */ \
736 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, /* \A, \G, \B, \B, \D, \d, \S, \s, \W, \w */ \
737 1, 1, /* Any, Anybyte */ \
738 3, 3, 1, 1, /* NOTPROP, PROP, EXTUNI, ANYNL */ \
739 1, 1, 2, 1, 1, /* \Z, \z, Opt, ^, $ */ \
740 2, /* Char - the minimum length */ \
741 2, /* Charnc - the minimum length */ \
742 2, /* not */ \
743 /* Positive single-char repeats ** These are */ \
744 2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \
745 4, 4, 4, /* upto, minupto, exact ** UTF-8 mode */ \
746 2, 2, 2, 4, /* *+, ++, ?+, upto+ */ \
747 /* Negative single-char repeats - only for chars < 256 */ \
748 2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \
749 4, 4, 4, /* NOT upto, minupto, exact */ \
750 2, 2, 2, 4, /* Possessive *, +, ?, upto */ \
751 /* Positive type repeats */ \
752 2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \
753 4, 4, 4, /* Type upto, minupto, exact */ \
754 2, 2, 2, 4, /* Possessive *+, ++, ?+, upto+ */ \
755 /* Character class & ref repeats */ \
756 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \
757 5, 5, /* CRRANGE, CRMINRANGE */ \
758 33, /* CLASS */ \
759 33, /* NCLASS */ \
760 0, /* XCLASS - variable length */ \
761 3, /* REF */ \
762 1+LINK_SIZE, /* RECURSE */ \
763 2+2*LINK_SIZE, /* CALLOUT */ \
764 1+LINK_SIZE, /* Alt */ \
765 1+LINK_SIZE, /* Ket */ \
766 1+LINK_SIZE, /* KetRmax */ \
767 1+LINK_SIZE, /* KetRmin */ \
768 1+LINK_SIZE, /* Assert */ \
769 1+LINK_SIZE, /* Assert not */ \
770 1+LINK_SIZE, /* Assert behind */ \
771 1+LINK_SIZE, /* Assert behind not */ \
772 1+LINK_SIZE, /* Reverse */ \
773 1+LINK_SIZE, /* ONCE */ \
774 1+LINK_SIZE, /* BRA */ \
775 3+LINK_SIZE, /* CBRA */ \
776 1+LINK_SIZE, /* COND */ \
777 1+LINK_SIZE, /* SBRA */ \
778 3+LINK_SIZE, /* SCBRA */ \
779 1+LINK_SIZE, /* SCOND */ \
780 3, /* CREF */ \
781 3, /* RREF */ \
782 1, /* DEF */ \
783 1, 1, /* BRAZERO, BRAMINZERO */ \
784
785
786 /* A magic value for OP_RREF to indicate the "any recursion" condition. */
787
788 #define RREF_ANY 0xffff
789
790 /* Error code numbers. They are given names so that they can more easily be
791 tracked. */
792
793 enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9,
794 ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19,
795 ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29,
796 ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
797 ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
798 ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57 };
799
800 /* The real format of the start of the pcre block; the index of names and the
801 code vector run on as long as necessary after the end. We store an explicit
802 offset to the name table so that if a regex is compiled on one host, saved, and
803 then run on another where the size of pointers is different, all might still
804 be well. For the case of compiled-on-4 and run-on-8, we include an extra
805 pointer that is always NULL. For future-proofing, a few dummy fields were
806 originally included - even though you can never get this planning right - but
807 there is only one left now.
808
809 NOTE NOTE NOTE:
810 Because people can now save and re-use compiled patterns, any additions to this
811 structure should be made at the end, and something earlier (e.g. a new
812 flag in the options or one of the dummy fields) should indicate that the new
813 fields are present. Currently PCRE always sets the dummy fields to zero.
814 NOTE NOTE NOTE:
815 */
816
817 typedef struct real_pcre {
818 pcre_uint32 magic_number;
819 pcre_uint32 size; /* Total that was malloced */
820 pcre_uint32 options;
821 pcre_uint32 dummy1; /* For future use, maybe */
822
823 pcre_uint16 top_bracket;
824 pcre_uint16 top_backref;
825 pcre_uint16 first_byte;
826 pcre_uint16 req_byte;
827 pcre_uint16 name_table_offset; /* Offset to name table that follows */
828 pcre_uint16 name_entry_size; /* Size of any name items */
829 pcre_uint16 name_count; /* Number of name items */
830 pcre_uint16 ref_count; /* Reference count */
831
832 const unsigned char *tables; /* Pointer to tables or NULL for std */
833 const unsigned char *nullpad; /* NULL padding */
834 } real_pcre;
835
836 /* The format of the block used to store data from pcre_study(). The same
837 remark (see NOTE above) about extending this structure applies. */
838
839 typedef struct pcre_study_data {
840 pcre_uint32 size; /* Total that was malloced */
841 pcre_uint32 options;
842 uschar start_bits[32];
843 } pcre_study_data;
844
845 /* Structure for passing "static" information around between the functions
846 doing the compiling, so that they are thread-safe. */
847
848 typedef struct compile_data {
849 const uschar *lcc; /* Points to lower casing table */
850 const uschar *fcc; /* Points to case-flipping table */
851 const uschar *cbits; /* Points to character type table */
852 const uschar *ctypes; /* Points to table of type maps */
853 const uschar *start_workspace;/* The start of working space */
854 const uschar *start_code; /* The start of the compiled code */
855 const uschar *start_pattern; /* The start of the pattern */
856 const uschar *end_pattern; /* The end of the pattern */
857 uschar *hwm; /* High watermark of workspace */
858 uschar *name_table; /* The name/number table */
859 int names_found; /* Number of entries so far */
860 int name_entry_size; /* Size of each entry */
861 int bracount; /* Count of capturing parens */
862 int top_backref; /* Maximum back reference */
863 unsigned int backref_map; /* Bitmap of low back refs */
864 int external_options; /* External (initial) options */
865 int req_varyopt; /* "After variable item" flag for reqbyte */
866 BOOL nopartial; /* Set TRUE if partial won't work */
867 int nltype; /* Newline type */
868 int nllen; /* Newline string length */
869 uschar nl[4]; /* Newline string when fixed length */
870 } compile_data;
871
872 /* Structure for maintaining a chain of pointers to the currently incomplete
873 branches, for testing for left recursion. */
874
875 typedef struct branch_chain {
876 struct branch_chain *outer;
877 uschar *current;
878 } branch_chain;
879
880 /* Structure for items in a linked list that represents an explicit recursive
881 call within the pattern. */
882
883 typedef struct recursion_info {
884 struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
885 int group_num; /* Number of group that was called */
886 const uschar *after_call; /* "Return value": points after the call in the expr */
887 USPTR save_start; /* Old value of md->start_match */
888 int *offset_save; /* Pointer to start of saved offsets */
889 int saved_max; /* Number of saved offsets */
890 } recursion_info;
891
892 /* When compiling in a mode that doesn't use recursive calls to match(),
893 a structure is used to remember local variables on the heap. It is defined in
894 pcre_exec.c, close to the match() function, so that it is easy to keep it in
895 step with any changes of local variable. However, the pointer to the current
896 frame must be saved in some "static" place over a longjmp(). We declare the
897 structure here so that we can put a pointer in the match_data structure. NOTE:
898 This isn't used for a "normal" compilation of pcre. */
899
900 struct heapframe;
901
902 /* Structure for building a chain of data for holding the values of the subject
903 pointer at the start of each subpattern, so as to detect when an empty string
904 has been matched by a subpattern - to break infinite loops. */
905
906 typedef struct eptrblock {
907 struct eptrblock *epb_prev;
908 USPTR epb_saved_eptr;
909 } eptrblock;
910
911
912 /* Structure for passing "static" information around between the functions
913 doing traditional NFA matching, so that they are thread-safe. */
914
915 typedef struct match_data {
916 unsigned long int match_call_count; /* As it says */
917 unsigned long int match_limit; /* As it says */
918 unsigned long int match_limit_recursion; /* As it says */
919 int *offset_vector; /* Offset vector */
920 int offset_end; /* One past the end */
921 int offset_max; /* The maximum usable for return data */
922 int nltype; /* Newline type */
923 int nllen; /* Newline string length */
924 uschar nl[4]; /* Newline string when fixed */
925 const uschar *lcc; /* Points to lower casing table */
926 const uschar *ctypes; /* Points to table of type maps */
927 BOOL offset_overflow; /* Set if too many extractions */
928 BOOL notbol; /* NOTBOL flag */
929 BOOL noteol; /* NOTEOL flag */
930 BOOL utf8; /* UTF8 flag */
931 BOOL endonly; /* Dollar not before final \n */
932 BOOL notempty; /* Empty string match not wanted */
933 BOOL partial; /* PARTIAL flag */
934 BOOL hitend; /* Hit the end of the subject at some point */
935 const uschar *start_code; /* For use when recursing */
936 USPTR start_subject; /* Start of the subject string */
937 USPTR end_subject; /* End of the subject string */
938 USPTR start_match; /* Start of this match attempt */
939 USPTR end_match_ptr; /* Subject position at end match */
940 int end_offset_top; /* Highwater mark at end of match */
941 int capture_last; /* Most recent capture number */
942 int start_offset; /* The start offset value */
943 eptrblock *eptrchain; /* Chain of eptrblocks for tail recursions */
944 int eptrn; /* Next free eptrblock */
945 recursion_info *recursive; /* Linked list of recursion data */
946 void *callout_data; /* To pass back to callouts */
947 struct heapframe *thisframe; /* Used only when compiling for no recursion */
948 } match_data;
949
950 /* A similar structure is used for the same purpose by the DFA matching
951 functions. */
952
953 typedef struct dfa_match_data {
954 const uschar *start_code; /* Start of the compiled pattern */
955 const uschar *start_subject; /* Start of the subject string */
956 const uschar *end_subject; /* End of subject string */
957 const uschar *tables; /* Character tables */
958 int moptions; /* Match options */
959 int poptions; /* Pattern options */
960 int nltype; /* Newline type */
961 int nllen; /* Newline string length */
962 uschar nl[4]; /* Newline string when fixed */
963 void *callout_data; /* To pass back to callouts */
964 } dfa_match_data;
965
966 /* Bit definitions for entries in the pcre_ctypes table. */
967
968 #define ctype_space 0x01
969 #define ctype_letter 0x02
970 #define ctype_digit 0x04
971 #define ctype_xdigit 0x08
972 #define ctype_word 0x10 /* alphameric or '_' */
973 #define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
974
975 /* Offsets for the bitmap tables in pcre_cbits. Each table contains a set
976 of bits for a class map. Some classes are built by combining these tables. */
977
978 #define cbit_space 0 /* [:space:] or \s */
979 #define cbit_xdigit 32 /* [:xdigit:] */
980 #define cbit_digit 64 /* [:digit:] or \d */
981 #define cbit_upper 96 /* [:upper:] */
982 #define cbit_lower 128 /* [:lower:] */
983 #define cbit_word 160 /* [:word:] or \w */
984 #define cbit_graph 192 /* [:graph:] */
985 #define cbit_print 224 /* [:print:] */
986 #define cbit_punct 256 /* [:punct:] */
987 #define cbit_cntrl 288 /* [:cntrl:] */
988 #define cbit_length 320 /* Length of the cbits table */
989
990 /* Offsets of the various tables from the base tables pointer, and
991 total length. */
992
993 #define lcc_offset 0
994 #define fcc_offset 256
995 #define cbits_offset 512
996 #define ctypes_offset (cbits_offset + cbit_length)
997 #define tables_length (ctypes_offset + 256)
998
999 /* Layout of the UCP type table that translates property names into types and
1000 codes. */
1001
1002 typedef struct {
1003 const char *name;
1004 pcre_uint16 type;
1005 pcre_uint16 value;
1006 } ucp_type_table;
1007
1008
1009 /* Internal shared data tables. These are tables that are used by more than one
1010 of the exported public functions. They have to be "external" in the C sense,
1011 but are not part of the PCRE public API. The data for these tables is in the
1012 pcre_tables.c module. */
1013
1014 extern const int _pcre_utf8_table1[];
1015 extern const int _pcre_utf8_table2[];
1016 extern const int _pcre_utf8_table3[];
1017 extern const uschar _pcre_utf8_table4[];
1018
1019 extern const int _pcre_utf8_table1_size;
1020
1021 extern const ucp_type_table _pcre_utt[];
1022 extern const int _pcre_utt_size;
1023
1024 extern const uschar _pcre_default_tables[];
1025
1026 extern const uschar _pcre_OP_lengths[];
1027
1028
1029 /* Internal shared functions. These are functions that are used by more than
1030 one of the exported public functions. They have to be "external" in the C
1031 sense, but are not part of the PCRE public API. */
1032
1033 extern BOOL _pcre_is_newline(const uschar *, const uschar *, int *,
1034 BOOL);
1035 extern int _pcre_ord2utf8(int, uschar *);
1036 extern real_pcre *_pcre_try_flipped(const real_pcre *, real_pcre *,
1037 const pcre_study_data *, pcre_study_data *);
1038 extern int _pcre_ucp_findprop(const unsigned int, int *, int *);
1039 extern unsigned int _pcre_ucp_othercase(const unsigned int);
1040 extern int _pcre_valid_utf8(const uschar *, int);
1041 extern BOOL _pcre_was_newline(const uschar *, const uschar *, int *,
1042 BOOL);
1043 extern BOOL _pcre_xclass(int, const uschar *);
1044
1045 #endif
1046
1047 /* End of pcre_internal.h */

Properties

Name Value
svn:eol-style native
svn:keywords "Author Date Id Revision Url"

webmaster@exim.org
ViewVC Help
Powered by ViewVC 1.1.12