1 /* $Cambridge: exim/src/src/pcre/pcre_internal.h,v 1.5 2007/06/26 11:16:54 ph10 Exp $ */
3 /*************************************************
4 * Perl-Compatible Regular Expressions *
5 *************************************************/
8 /* PCRE is a library of functions to support regular expressions whose syntax
9 and semantics are as close as possible to those of the Perl 5 language.
11 Written by Philip Hazel
12 Copyright (c) 1997-2007 University of Cambridge
14 -----------------------------------------------------------------------------
15 Redistribution and use in source and binary forms, with or without
16 modification, are permitted provided that the following conditions are met:
18 * Redistributions of source code must retain the above copyright notice,
19 this list of conditions and the following disclaimer.
21 * Redistributions in binary form must reproduce the above copyright
22 notice, this list of conditions and the following disclaimer in the
23 documentation and/or other materials provided with the distribution.
25 * Neither the name of the University of Cambridge nor the names of its
26 contributors may be used to endorse or promote products derived from
27 this software without specific prior written permission.
29 THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
30 AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
31 IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
32 ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE
33 LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
34 CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
35 SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
36 INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
37 CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
38 ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
39 POSSIBILITY OF SUCH DAMAGE.
40 -----------------------------------------------------------------------------
43 /* This header contains definitions that are shared between the different
44 modules, but which are not relevant to the exported API. This includes some
45 functions whose names all begin with "_pcre_". */
47 #ifndef PCRE_INTERNAL_H
48 #define PCRE_INTERNAL_H
50 /* Define DEBUG to get debugging output on stdout. */
56 /* Use a macro for debugging printing, 'cause that eliminates the use of #ifdef
57 inline, and there are *still* stupid compilers about that don't like indented
58 pre-processor statements, or at least there were when I first wrote this. After
59 all, it had only been about 10 years then...
61 It turns out that the Mac Debugging.h header also defines the macro DPRINTF, so
62 be absolutely sure we get our version. */
66 #define DPRINTF(p) printf p
68 #define DPRINTF(p) /* Nothing */
72 /* Get the definitions provided by running "configure" */
76 /* Standard C headers plus the external interface definition. The only time
77 setjmp and stdarg are used is when NO_RECURSE is set. */
88 /* When compiling a DLL for Windows, the exported symbols have to be declared
89 using some MS magic. I found some useful information on this web page:
90 http://msdn2.microsoft.com/en-us/library/y4h7bcy6(VS.80).aspx. According to the
91 information there, using __declspec(dllexport) without "extern" we have a
92 definition; with "extern" we have a declaration. The settings here override the
93 setting in pcre.h (which is included below); it defines only PCRE_EXP_DECL,
94 which is all that is needed for applications (they just import the symbols). We
97 PCRE_EXP_DECL for declarations
98 PCRE_EXP_DEFN for definitions of exported functions
99 PCRE_EXP_DATA_DEFN for definitions of exported variables
101 The reason for the two DEFN macros is that in non-Windows environments, one
102 does not want to have "extern" before variable definitions because it leads to
103 compiler warnings. So we distinguish between functions and variables. In
104 Windows, the two should always be the same.
106 The reason for wrapping this in #ifndef PCRE_EXP_DECL is so that pcretest,
107 which is an application, but needs to import this file in order to "peek" at
108 internals, can #include pcre.h first to get an application's-eye view.
110 In principle, people compiling for non-Windows, non-Unix-like (i.e. uncommon,
111 special-purpose environments) might want to stick other stuff in front of
112 exported symbols. That's why, in the non-Windows case, we set PCRE_EXP_DEFN and
113 PCRE_EXP_DATA_DEFN only if they are not already set. */
115 #ifndef PCRE_EXP_DECL
118 # define PCRE_EXP_DECL extern __declspec(dllexport)
119 # define PCRE_EXP_DEFN __declspec(dllexport)
120 # define PCRE_EXP_DATA_DEFN __declspec(dllexport)
122 # define PCRE_EXP_DECL extern
123 # define PCRE_EXP_DEFN
124 # define PCRE_EXP_DATA_DEFN
129 # define PCRE_EXP_DECL extern "C"
131 # define PCRE_EXP_DECL extern
133 # ifndef PCRE_EXP_DEFN
134 # define PCRE_EXP_DEFN PCRE_EXP_DECL
136 # ifndef PCRE_EXP_DATA_DEFN
137 # define PCRE_EXP_DATA_DEFN
142 /* We need to have types that specify unsigned 16-bit and 32-bit integers. We
143 cannot determine these outside the compilation (e.g. by running a program as
144 part of "configure") because PCRE is often cross-compiled for use on other
145 systems. Instead we make use of the maximum sizes that are available at
146 preprocessor time in standard C environments. */
148 #if USHRT_MAX == 65535
149 typedef unsigned short pcre_uint16;
150 #elif UINT_MAX == 65535
151 typedef unsigned int pcre_uint16;
153 #error Cannot determine a type for 16-bit unsigned integers
156 #if UINT_MAX == 4294967295
157 typedef unsigned int pcre_uint32;
158 #elif ULONG_MAX == 4294967295
159 typedef unsigned long int pcre_uint32;
161 #error Cannot determine a type for 32-bit unsigned integers
164 /* All character handling must be done as unsigned characters. Otherwise there
165 are problems with top-bit-set characters and functions such as isspace().
166 However, we leave the interface to the outside world as char *, because that
167 should make things easier for callers. We define a short type for unsigned char
168 to save lots of typing. I tried "uchar", but it causes problems on Digital
169 Unix, where it is defined in sys/types, so use "uschar" instead. */
171 typedef unsigned char uschar;
173 /* This is an unsigned int value that no character can ever have. UTF-8
174 characters only go up to 0x7fffffff (though Unicode doesn't go beyond
177 #define NOTACHAR 0xffffffff
179 /* PCRE is able to support several different kinds of newline (CR, LF, CRLF,
180 "any" and "anycrlf" at present). The following macros are used to package up
181 testing for newlines. NLBLOCK, PSSTART, and PSEND are defined in the various
182 modules to indicate in which datablock the parameters exist, and what the
183 start/end of string field names are. */
185 #define NLTYPE_FIXED 0 /* Newline is a fixed length string */
186 #define NLTYPE_ANY 1 /* Newline is any Unicode line ending */
187 #define NLTYPE_ANYCRLF 2 /* Newline is CR, LF, or CRLF */
189 /* This macro checks for a newline at the given position */
191 #define IS_NEWLINE(p) \
192 ((NLBLOCK->nltype != NLTYPE_FIXED)? \
193 ((p) < NLBLOCK->PSEND && \
194 _pcre_is_newline((p), NLBLOCK->nltype, NLBLOCK->PSEND, &(NLBLOCK->nllen),\
197 ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
198 (p)[0] == NLBLOCK->nl[0] && \
199 (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1]) \
203 /* This macro checks for a newline immediately preceding the given position */
205 #define WAS_NEWLINE(p) \
206 ((NLBLOCK->nltype != NLTYPE_FIXED)? \
207 ((p) > NLBLOCK->PSSTART && \
208 _pcre_was_newline((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \
209 &(NLBLOCK->nllen), utf8)) \
211 ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
212 (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \
213 (NLBLOCK->nllen == 1 || (p)[-NLBLOCK->nllen+1] == NLBLOCK->nl[1]) \
217 /* When PCRE is compiled as a C++ library, the subject pointer can be replaced
218 with a custom type. This makes it possible, for example, to allow pcre_exec()
219 to process subject strings that are discontinuous by using a smart pointer
220 class. It must always be possible to inspect all of the subject string in
221 pcre_exec() because of the way it backtracks. Two macros are required in the
222 normal case, for sign-unspecified and unsigned char pointers. The former is
223 used for the external interface and appears in pcre.h, which is why its name
224 must begin with PCRE_. */
226 #ifdef CUSTOM_SUBJECT_PTR
227 #define PCRE_SPTR CUSTOM_SUBJECT_PTR
228 #define USPTR CUSTOM_SUBJECT_PTR
230 #define PCRE_SPTR const char *
231 #define USPTR const unsigned char *
236 /* Include the public PCRE header and the definitions of UCP character property
242 /* When compiling for use with the Virtual Pascal compiler, these functions
243 need to have their names changed. PCRE must be compiled with the -DVPCOMPAT
244 option on the command line. */
247 #define strlen(s) _strlen(s)
248 #define strncmp(s1,s2,m) _strncmp(s1,s2,m)
249 #define memcmp(s,c,n) _memcmp(s,c,n)
250 #define memcpy(d,s,n) _memcpy(d,s,n)
251 #define memmove(d,s,n) _memmove(d,s,n)
252 #define memset(s,c,n) _memset(s,c,n)
255 /* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
256 define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY
257 is set. Otherwise, include an emulating function for those systems that have
258 neither (there some non-Unix environments where this is the case). */
261 #undef memmove /* some systems may have a macro */
263 #define memmove(a, b, c) bcopy(b, a, c)
264 #else /* HAVE_BCOPY */
266 pcre_memmove(void *d, const void *s, size_t n)
269 unsigned char *dest = (unsigned char *)d;
270 const unsigned char *src = (const unsigned char *)s;
275 for (i = 0; i < n; ++i) *(--dest) = *(--src);
280 for (i = 0; i < n; ++i) *dest++ = *src++;
281 return (void *)(dest - n);
284 #define memmove(a, b, c) pcre_memmove(a, b, c)
285 #endif /* not HAVE_BCOPY */
286 #endif /* not HAVE_MEMMOVE */
287 #endif /* not VPCOMPAT */
290 /* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored
291 in big-endian order) by default. These are used, for example, to link from the
292 start of a subpattern to its alternatives and its end. The use of 2 bytes per
293 offset limits the size of the compiled regex to around 64K, which is big enough
294 for almost everybody. However, I received a request for an even bigger limit.
295 For this reason, and also to make the code easier to maintain, the storing and
296 loading of offsets from the byte string is now handled by the macros that are
299 The macros are controlled by the value of LINK_SIZE. This defaults to 2 in
300 the config.h file, but can be overridden by using -D on the command line. This
301 is automated on Unix systems via the "configure" command. */
307 (a[(n)+1] = (d) & 255)
310 (((a)[n] << 8) | (a)[(n)+1])
312 #define MAX_PATTERN_SIZE (1 << 16)
318 (a[n] = (d) >> 16), \
319 (a[(n)+1] = (d) >> 8), \
320 (a[(n)+2] = (d) & 255)
323 (((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2])
325 #define MAX_PATTERN_SIZE (1 << 24)
331 (a[n] = (d) >> 24), \
332 (a[(n)+1] = (d) >> 16), \
333 (a[(n)+2] = (d) >> 8), \
334 (a[(n)+3] = (d) & 255)
337 (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3])
339 #define MAX_PATTERN_SIZE (1 << 30) /* Keep it positive */
343 #error LINK_SIZE must be either 2, 3, or 4
347 /* Convenience macro defined in terms of the others */
349 #define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE
352 /* PCRE uses some other 2-byte quantities that do not change when the size of
353 offsets changes. There are used for repeat counts and for other things such as
354 capturing parenthesis numbers in back references. */
356 #define PUT2(a,n,d) \
361 (((a)[n] << 8) | (a)[(n)+1])
363 #define PUT2INC(a,n,d) PUT2(a,n,d), a += 2
366 /* When UTF-8 encoding is being used, a character is no longer just a single
367 byte. The macros for character handling generate simple sequences when used in
368 byte-mode, and more complicated ones for UTF-8 characters. */
371 #define GETCHAR(c, eptr) c = *eptr;
372 #define GETCHARTEST(c, eptr) c = *eptr;
373 #define GETCHARINC(c, eptr) c = *eptr++;
374 #define GETCHARINCTEST(c, eptr) c = *eptr++;
375 #define GETCHARLEN(c, eptr, len) c = *eptr;
376 #define BACKCHAR(eptr)
378 #else /* SUPPORT_UTF8 */
380 /* Get the next UTF-8 character, not advancing the pointer. This is called when
381 we know we are in UTF-8 mode. */
383 #define GETCHAR(c, eptr) \
388 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
390 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
391 for (gcii = 1; gcii <= gcaa; gcii++) \
394 c |= (eptr[gcii] & 0x3f) << gcss; \
398 /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
401 #define GETCHARTEST(c, eptr) \
403 if (utf8 && c >= 0xc0) \
406 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
408 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
409 for (gcii = 1; gcii <= gcaa; gcii++) \
412 c |= (eptr[gcii] & 0x3f) << gcss; \
416 /* Get the next UTF-8 character, advancing the pointer. This is called when we
417 know we are in UTF-8 mode. */
419 #define GETCHARINC(c, eptr) \
423 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
425 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
429 c |= (*eptr++ & 0x3f) << gcss; \
433 /* Get the next character, testing for UTF-8 mode, and advancing the pointer */
435 #define GETCHARINCTEST(c, eptr) \
437 if (utf8 && c >= 0xc0) \
439 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
441 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
445 c |= (*eptr++ & 0x3f) << gcss; \
449 /* Get the next UTF-8 character, not advancing the pointer, incrementing length
450 if there are extra bytes. This is called when we know we are in UTF-8 mode. */
452 #define GETCHARLEN(c, eptr, len) \
457 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
459 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
460 for (gcii = 1; gcii <= gcaa; gcii++) \
463 c |= (eptr[gcii] & 0x3f) << gcss; \
468 /* If the pointer is not at the start of a character, move it back until
469 it is. Called only in UTF-8 mode. */
471 #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--;
476 /* In case there is no definition of offsetof() provided - though any proper
477 Standard C system should have one. */
480 #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))
484 /* These are the public options that can change during matching. */
486 #define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)
488 /* Private options flags start at the most significant end of the four bytes.
489 The public options defined in pcre.h start at the least significant end. Make
490 sure they don't overlap! The bits are getting a bit scarce now -- when we run
491 out, there is a dummy word in the structure that could be used for the private
494 #define PCRE_NOPARTIAL 0x80000000 /* can't use partial with this regex */
495 #define PCRE_FIRSTSET 0x40000000 /* first_byte is set */
496 #define PCRE_REQCHSET 0x20000000 /* req_byte is set */
497 #define PCRE_STARTLINE 0x10000000 /* start after \n for multiline */
498 #define PCRE_JCHANGED 0x08000000 /* j option changes within regex */
500 /* Options for the "extra" block produced by pcre_study(). */
502 #define PCRE_STUDY_MAPPED 0x01 /* a map of starting chars exists */
504 /* Masks for identifying the public options that are permitted at compile
505 time, run time, or study time, respectively. */
507 #define PCRE_NEWLINE_BITS (PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|PCRE_NEWLINE_ANY| \
508 PCRE_NEWLINE_ANYCRLF)
510 #define PUBLIC_OPTIONS \
511 (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
512 PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
513 PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
514 PCRE_DUPNAMES|PCRE_NEWLINE_BITS)
516 #define PUBLIC_EXEC_OPTIONS \
517 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
518 PCRE_PARTIAL|PCRE_NEWLINE_BITS)
520 #define PUBLIC_DFA_EXEC_OPTIONS \
521 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
522 PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART|PCRE_NEWLINE_BITS)
524 #define PUBLIC_STUDY_OPTIONS 0 /* None defined */
526 /* Magic number to provide a small check against being handed junk. Also used
527 to detect whether a pattern was compiled on a host of different endianness. */
529 #define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */
531 /* Negative values for the firstchar and reqchar variables */
533 #define REQ_UNSET (-2)
534 #define REQ_NONE (-1)
536 /* The maximum remaining length of subject we are prepared to search for a
539 #define REQ_BYTE_MAX 1000
541 /* Flags added to firstbyte or reqbyte; a "non-literal" item is either a
542 variable-length repeat, or a anything other than literal characters. */
544 #define REQ_CASELESS 0x0100 /* indicates caselessness */
545 #define REQ_VARY 0x0200 /* reqbyte followed non-literal item */
547 /* Miscellaneous definitions */
554 /* Escape items that are just an encoding of a particular data value. */
572 /* We can't officially use ESC_t because it is a POSIX reserved identifier
573 (presumably because of all the others like size_t). */
579 /* Codes for different types of Unicode property */
581 #define PT_ANY 0 /* Any property - matches all chars */
582 #define PT_LAMP 1 /* L& - the union of Lu, Ll, Lt */
583 #define PT_GC 2 /* General characteristic (e.g. L) */
584 #define PT_PC 3 /* Particular characteristic (e.g. Lu) */
585 #define PT_SC 4 /* Script (e.g. Han) */
587 /* Flag bits and data types for the extended class (OP_XCLASS) for classes that
588 contain UTF-8 characters with values greater than 255. */
590 #define XCL_NOT 0x01 /* Flag: this is a negative class */
591 #define XCL_MAP 0x02 /* Flag: a 32-byte map is present */
593 #define XCL_END 0 /* Marks end of individual items */
594 #define XCL_SINGLE 1 /* Single item (one multibyte char) follows */
595 #define XCL_RANGE 2 /* A range (two multibyte chars) follows */
596 #define XCL_PROP 3 /* Unicode property (2-byte property code follows) */
597 #define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */
599 /* These are escaped items that aren't just an encoding of a particular data
600 value such as \n. They must have non-zero values, as check_escape() returns
601 their negation. Also, they must appear in the same order as in the opcode
602 definitions below, up to ESC_z. There's a dummy for OP_ANY because it
603 corresponds to "." rather than an escape sequence. The final one must be
604 ESC_REF as subsequent values are used for backreferences (\1, \2, \3, etc).
605 There are two tests in the code for an escape greater than ESC_b and less than
606 ESC_Z to detect the types that may be repeated. These are the types that
607 consume characters. If any new escapes are put in between that don't consume a
608 character, that code will have to change. */
610 enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s,
611 ESC_W, ESC_w, ESC_dum1, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H, ESC_h,
612 ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z, ESC_E, ESC_Q, ESC_k, ESC_REF };
615 /* Opcode table: OP_BRA must be last, as all values >= it are used for brackets
616 that extract substrings. Starting from 1 (i.e. after OP_END), the values up to
617 OP_EOD must correspond in order to the list of escapes immediately above.
619 To keep stored, compiled patterns compatible, new opcodes should be added
620 immediately before OP_BRA, where (since release 7.0) a gap is left for this
623 *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions
624 that follow must also be updated to match. There is also a table called
625 "coptable" in pcre_dfa_exec.c that must be updated. */
628 OP_END, /* 0 End of pattern */
630 /* Values corresponding to backslashed metacharacters */
632 OP_SOD, /* 1 Start of data: \A */
633 OP_SOM, /* 2 Start of match (subject + offset): \G */
634 OP_SET_SOM, /* 3 Set start of match (\K) */
635 OP_NOT_WORD_BOUNDARY, /* 4 \B */
636 OP_WORD_BOUNDARY, /* 5 \b */
637 OP_NOT_DIGIT, /* 6 \D */
639 OP_NOT_WHITESPACE, /* 8 \S */
640 OP_WHITESPACE, /* 9 \s */
641 OP_NOT_WORDCHAR, /* 10 \W */
642 OP_WORDCHAR, /* 11 \w */
643 OP_ANY, /* 12 Match any character */
644 OP_ANYBYTE, /* 13 Match any byte (\C); different to OP_ANY for UTF-8 */
645 OP_NOTPROP, /* 14 \P (not Unicode property) */
646 OP_PROP, /* 15 \p (Unicode property) */
647 OP_ANYNL, /* 16 \R (any newline sequence) */
648 OP_NOT_HSPACE, /* 17 \H (not horizontal whitespace) */
649 OP_HSPACE, /* 18 \h (horizontal whitespace) */
650 OP_NOT_VSPACE, /* 19 \V (not vertical whitespace) */
651 OP_VSPACE, /* 20 \v (vertical whitespace) */
652 OP_EXTUNI, /* 21 \X (extended Unicode sequence */
653 OP_EODN, /* 22 End of data or \n at end of data: \Z. */
654 OP_EOD, /* 23 End of data: \z */
656 OP_OPT, /* 24 Set runtime options */
657 OP_CIRC, /* 25 Start of line - varies with multiline switch */
658 OP_DOLL, /* 26 End of line - varies with multiline switch */
659 OP_CHAR, /* 27 Match one character, casefully */
660 OP_CHARNC, /* 28 Match one character, caselessly */
661 OP_NOT, /* 29 Match one character, not the following one */
663 OP_STAR, /* 30 The maximizing and minimizing versions of */
664 OP_MINSTAR, /* 31 these six opcodes must come in pairs, with */
665 OP_PLUS, /* 32 the minimizing one second. */
666 OP_MINPLUS, /* 33 This first set applies to single characters.*/
668 OP_MINQUERY, /* 35 */
670 OP_UPTO, /* 36 From 0 to n matches */
672 OP_EXACT, /* 38 Exactly n matches */
674 OP_POSSTAR, /* 39 Possessified star */
675 OP_POSPLUS, /* 40 Possessified plus */
676 OP_POSQUERY, /* 41 Posesssified query */
677 OP_POSUPTO, /* 42 Possessified upto */
679 OP_NOTSTAR, /* 43 The maximizing and minimizing versions of */
680 OP_NOTMINSTAR, /* 44 these six opcodes must come in pairs, with */
681 OP_NOTPLUS, /* 45 the minimizing one second. They must be in */
682 OP_NOTMINPLUS, /* 46 exactly the same order as those above. */
683 OP_NOTQUERY, /* 47 This set applies to "not" single characters. */
684 OP_NOTMINQUERY, /* 48 */
686 OP_NOTUPTO, /* 49 From 0 to n matches */
687 OP_NOTMINUPTO, /* 50 */
688 OP_NOTEXACT, /* 51 Exactly n matches */
690 OP_NOTPOSSTAR, /* 52 Possessified versions */
691 OP_NOTPOSPLUS, /* 53 */
692 OP_NOTPOSQUERY, /* 54 */
693 OP_NOTPOSUPTO, /* 55 */
695 OP_TYPESTAR, /* 56 The maximizing and minimizing versions of */
696 OP_TYPEMINSTAR, /* 57 these six opcodes must come in pairs, with */
697 OP_TYPEPLUS, /* 58 the minimizing one second. These codes must */
698 OP_TYPEMINPLUS, /* 59 be in exactly the same order as those above. */
699 OP_TYPEQUERY, /* 60 This set applies to character types such as \d */
700 OP_TYPEMINQUERY, /* 61 */
702 OP_TYPEUPTO, /* 62 From 0 to n matches */
703 OP_TYPEMINUPTO, /* 63 */
704 OP_TYPEEXACT, /* 64 Exactly n matches */
706 OP_TYPEPOSSTAR, /* 65 Possessified versions */
707 OP_TYPEPOSPLUS, /* 66 */
708 OP_TYPEPOSQUERY, /* 67 */
709 OP_TYPEPOSUPTO, /* 68 */
711 OP_CRSTAR, /* 69 The maximizing and minimizing versions of */
712 OP_CRMINSTAR, /* 70 all these opcodes must come in pairs, with */
713 OP_CRPLUS, /* 71 the minimizing one second. These codes must */
714 OP_CRMINPLUS, /* 72 be in exactly the same order as those above. */
715 OP_CRQUERY, /* 73 These are for character classes and back refs */
716 OP_CRMINQUERY, /* 74 */
717 OP_CRRANGE, /* 75 These are different to the three sets above. */
718 OP_CRMINRANGE, /* 76 */
720 OP_CLASS, /* 77 Match a character class, chars < 256 only */
721 OP_NCLASS, /* 78 Same, but the bitmap was created from a negative
722 class - the difference is relevant only when a UTF-8
723 character > 255 is encountered. */
725 OP_XCLASS, /* 79 Extended class for handling UTF-8 chars within the
726 class. This does both positive and negative. */
728 OP_REF, /* 80 Match a back reference */
729 OP_RECURSE, /* 81 Match a numbered subpattern (possibly recursive) */
730 OP_CALLOUT, /* 82 Call out to external function if provided */
732 OP_ALT, /* 83 Start of alternation */
733 OP_KET, /* 84 End of group that doesn't have an unbounded repeat */
734 OP_KETRMAX, /* 85 These two must remain together and in this */
735 OP_KETRMIN, /* 86 order. They are for groups the repeat for ever. */
737 /* The assertions must come before BRA, CBRA, ONCE, and COND.*/
739 OP_ASSERT, /* 87 Positive lookahead */
740 OP_ASSERT_NOT, /* 88 Negative lookahead */
741 OP_ASSERTBACK, /* 89 Positive lookbehind */
742 OP_ASSERTBACK_NOT, /* 90 Negative lookbehind */
743 OP_REVERSE, /* 91 Move pointer back - used in lookbehind assertions */
745 /* ONCE, BRA, CBRA, and COND must come after the assertions, with ONCE first,
746 as there's a test for >= ONCE for a subpattern that isn't an assertion. */
748 OP_ONCE, /* 92 Atomic group */
749 OP_BRA, /* 83 Start of non-capturing bracket */
750 OP_CBRA, /* 94 Start of capturing bracket */
751 OP_COND, /* 95 Conditional group */
753 /* These three must follow the previous three, in the same order. There's a
754 check for >= SBRA to distinguish the two sets. */
756 OP_SBRA, /* 96 Start of non-capturing bracket, check empty */
757 OP_SCBRA, /* 97 Start of capturing bracket, check empty */
758 OP_SCOND, /* 98 Conditional group, check empty */
760 OP_CREF, /* 99 Used to hold a capture number as condition */
761 OP_RREF, /* 100 Used to hold a recursion number as condition */
762 OP_DEF, /* 101 The DEFINE condition */
764 OP_BRAZERO, /* 102 These two must remain together and in this */
765 OP_BRAMINZERO /* 103 order. */
769 /* This macro defines textual names for all the opcodes. These are used only
770 for debugging. The macro is referenced only in pcre_printint.c. */
772 #define OP_NAME_LIST \
773 "End", "\\A", "\\G", "\\K", "\\B", "\\b", "\\D", "\\d", \
774 "\\S", "\\s", "\\W", "\\w", "Any", "Anybyte", \
775 "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v", \
776 "extuni", "\\Z", "\\z", \
777 "Opt", "^", "$", "char", "charnc", "not", \
778 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
779 "*+","++", "?+", "{", \
780 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
781 "*+","++", "?+", "{", \
782 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
783 "*+","++", "?+", "{", \
784 "*", "*?", "+", "+?", "?", "??", "{", "{", \
785 "class", "nclass", "xclass", "Ref", "Recurse", "Callout", \
786 "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \
787 "AssertB", "AssertB not", "Reverse", \
788 "Once", "Bra 0", "Bra", "Cond", "SBra 0", "SBra", "SCond", \
789 "Cond ref", "Cond rec", "Cond def", "Brazero", "Braminzero"
792 /* This macro defines the length of fixed length operations in the compiled
793 regex. The lengths are used when searching for specific things, and also in the
794 debugging printing of a compiled regex. We use a macro so that it can be
795 defined close to the definitions of the opcodes themselves.
797 As things have been extended, some of these are no longer fixed lenths, but are
798 minima instead. For example, the length of a single-character repeat may vary
799 in UTF-8 mode. The code that uses this table must know about such things. */
803 1, 1, 1, 1, 1, /* \A, \G, \K, \B, \b */ \
804 1, 1, 1, 1, 1, 1, /* \D, \d, \S, \s, \W, \w */ \
805 1, 1, /* Any, Anybyte */ \
806 3, 3, 1, /* NOTPROP, PROP, EXTUNI */ \
807 1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */ \
808 1, 1, 2, 1, 1, /* \Z, \z, Opt, ^, $ */ \
809 2, /* Char - the minimum length */ \
810 2, /* Charnc - the minimum length */ \
812 /* Positive single-char repeats ** These are */ \
813 2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \
814 4, 4, 4, /* upto, minupto, exact ** UTF-8 mode */ \
815 2, 2, 2, 4, /* *+, ++, ?+, upto+ */ \
816 /* Negative single-char repeats - only for chars < 256 */ \
817 2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \
818 4, 4, 4, /* NOT upto, minupto, exact */ \
819 2, 2, 2, 4, /* Possessive *, +, ?, upto */ \
820 /* Positive type repeats */ \
821 2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \
822 4, 4, 4, /* Type upto, minupto, exact */ \
823 2, 2, 2, 4, /* Possessive *+, ++, ?+, upto+ */ \
824 /* Character class & ref repeats */ \
825 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \
826 5, 5, /* CRRANGE, CRMINRANGE */ \
829 0, /* XCLASS - variable length */ \
831 1+LINK_SIZE, /* RECURSE */ \
832 2+2*LINK_SIZE, /* CALLOUT */ \
833 1+LINK_SIZE, /* Alt */ \
834 1+LINK_SIZE, /* Ket */ \
835 1+LINK_SIZE, /* KetRmax */ \
836 1+LINK_SIZE, /* KetRmin */ \
837 1+LINK_SIZE, /* Assert */ \
838 1+LINK_SIZE, /* Assert not */ \
839 1+LINK_SIZE, /* Assert behind */ \
840 1+LINK_SIZE, /* Assert behind not */ \
841 1+LINK_SIZE, /* Reverse */ \
842 1+LINK_SIZE, /* ONCE */ \
843 1+LINK_SIZE, /* BRA */ \
844 3+LINK_SIZE, /* CBRA */ \
845 1+LINK_SIZE, /* COND */ \
846 1+LINK_SIZE, /* SBRA */ \
847 3+LINK_SIZE, /* SCBRA */ \
848 1+LINK_SIZE, /* SCOND */ \
852 1, 1, /* BRAZERO, BRAMINZERO */ \
855 /* A magic value for OP_RREF to indicate the "any recursion" condition. */
857 #define RREF_ANY 0xffff
859 /* Error code numbers. They are given names so that they can more easily be
862 enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9,
863 ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19,
864 ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29,
865 ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
866 ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
867 ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58 };
869 /* The real format of the start of the pcre block; the index of names and the
870 code vector run on as long as necessary after the end. We store an explicit
871 offset to the name table so that if a regex is compiled on one host, saved, and
872 then run on another where the size of pointers is different, all might still
873 be well. For the case of compiled-on-4 and run-on-8, we include an extra
874 pointer that is always NULL. For future-proofing, a few dummy fields were
875 originally included - even though you can never get this planning right - but
876 there is only one left now.
879 Because people can now save and re-use compiled patterns, any additions to this
880 structure should be made at the end, and something earlier (e.g. a new
881 flag in the options or one of the dummy fields) should indicate that the new
882 fields are present. Currently PCRE always sets the dummy fields to zero.
886 typedef struct real_pcre {
887 pcre_uint32 magic_number;
888 pcre_uint32 size; /* Total that was malloced */
890 pcre_uint32 dummy1; /* For future use, maybe */
892 pcre_uint16 top_bracket;
893 pcre_uint16 top_backref;
894 pcre_uint16 first_byte;
895 pcre_uint16 req_byte;
896 pcre_uint16 name_table_offset; /* Offset to name table that follows */
897 pcre_uint16 name_entry_size; /* Size of any name items */
898 pcre_uint16 name_count; /* Number of name items */
899 pcre_uint16 ref_count; /* Reference count */
901 const unsigned char *tables; /* Pointer to tables or NULL for std */
902 const unsigned char *nullpad; /* NULL padding */
905 /* The format of the block used to store data from pcre_study(). The same
906 remark (see NOTE above) about extending this structure applies. */
908 typedef struct pcre_study_data {
909 pcre_uint32 size; /* Total that was malloced */
911 uschar start_bits[32];
914 /* Structure for passing "static" information around between the functions
915 doing the compiling, so that they are thread-safe. */
917 typedef struct compile_data {
918 const uschar *lcc; /* Points to lower casing table */
919 const uschar *fcc; /* Points to case-flipping table */
920 const uschar *cbits; /* Points to character type table */
921 const uschar *ctypes; /* Points to table of type maps */
922 const uschar *start_workspace;/* The start of working space */
923 const uschar *start_code; /* The start of the compiled code */
924 const uschar *start_pattern; /* The start of the pattern */
925 const uschar *end_pattern; /* The end of the pattern */
926 uschar *hwm; /* High watermark of workspace */
927 uschar *name_table; /* The name/number table */
928 int names_found; /* Number of entries so far */
929 int name_entry_size; /* Size of each entry */
930 int bracount; /* Count of capturing parens */
931 int top_backref; /* Maximum back reference */
932 unsigned int backref_map; /* Bitmap of low back refs */
933 int external_options; /* External (initial) options */
934 int req_varyopt; /* "After variable item" flag for reqbyte */
935 BOOL nopartial; /* Set TRUE if partial won't work */
936 int nltype; /* Newline type */
937 int nllen; /* Newline string length */
938 uschar nl[4]; /* Newline string when fixed length */
941 /* Structure for maintaining a chain of pointers to the currently incomplete
942 branches, for testing for left recursion. */
944 typedef struct branch_chain {
945 struct branch_chain *outer;
949 /* Structure for items in a linked list that represents an explicit recursive
950 call within the pattern. */
952 typedef struct recursion_info {
953 struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
954 int group_num; /* Number of group that was called */
955 const uschar *after_call; /* "Return value": points after the call in the expr */
956 USPTR save_start; /* Old value of mstart */
957 int *offset_save; /* Pointer to start of saved offsets */
958 int saved_max; /* Number of saved offsets */
961 /* Structure for building a chain of data for holding the values of the subject
962 pointer at the start of each subpattern, so as to detect when an empty string
963 has been matched by a subpattern - to break infinite loops. */
965 typedef struct eptrblock {
966 struct eptrblock *epb_prev;
967 USPTR epb_saved_eptr;
971 /* Structure for passing "static" information around between the functions
972 doing traditional NFA matching, so that they are thread-safe. */
974 typedef struct match_data {
975 unsigned long int match_call_count; /* As it says */
976 unsigned long int match_limit; /* As it says */
977 unsigned long int match_limit_recursion; /* As it says */
978 int *offset_vector; /* Offset vector */
979 int offset_end; /* One past the end */
980 int offset_max; /* The maximum usable for return data */
981 int nltype; /* Newline type */
982 int nllen; /* Newline string length */
983 uschar nl[4]; /* Newline string when fixed */
984 const uschar *lcc; /* Points to lower casing table */
985 const uschar *ctypes; /* Points to table of type maps */
986 BOOL offset_overflow; /* Set if too many extractions */
987 BOOL notbol; /* NOTBOL flag */
988 BOOL noteol; /* NOTEOL flag */
989 BOOL utf8; /* UTF8 flag */
990 BOOL endonly; /* Dollar not before final \n */
991 BOOL notempty; /* Empty string match not wanted */
992 BOOL partial; /* PARTIAL flag */
993 BOOL hitend; /* Hit the end of the subject at some point */
994 const uschar *start_code; /* For use when recursing */
995 USPTR start_subject; /* Start of the subject string */
996 USPTR end_subject; /* End of the subject string */
997 USPTR start_match_ptr; /* Start of matched string */
998 USPTR end_match_ptr; /* Subject position at end match */
999 int end_offset_top; /* Highwater mark at end of match */
1000 int capture_last; /* Most recent capture number */
1001 int start_offset; /* The start offset value */
1002 eptrblock *eptrchain; /* Chain of eptrblocks for tail recursions */
1003 int eptrn; /* Next free eptrblock */
1004 recursion_info *recursive; /* Linked list of recursion data */
1005 void *callout_data; /* To pass back to callouts */
1008 /* A similar structure is used for the same purpose by the DFA matching
1011 typedef struct dfa_match_data {
1012 const uschar *start_code; /* Start of the compiled pattern */
1013 const uschar *start_subject; /* Start of the subject string */
1014 const uschar *end_subject; /* End of subject string */
1015 const uschar *tables; /* Character tables */
1016 int moptions; /* Match options */
1017 int poptions; /* Pattern options */
1018 int nltype; /* Newline type */
1019 int nllen; /* Newline string length */
1020 uschar nl[4]; /* Newline string when fixed */
1021 void *callout_data; /* To pass back to callouts */
1024 /* Bit definitions for entries in the pcre_ctypes table. */
1026 #define ctype_space 0x01
1027 #define ctype_letter 0x02
1028 #define ctype_digit 0x04
1029 #define ctype_xdigit 0x08
1030 #define ctype_word 0x10 /* alphameric or '_' */
1031 #define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
1033 /* Offsets for the bitmap tables in pcre_cbits. Each table contains a set
1034 of bits for a class map. Some classes are built by combining these tables. */
1036 #define cbit_space 0 /* [:space:] or \s */
1037 #define cbit_xdigit 32 /* [:xdigit:] */
1038 #define cbit_digit 64 /* [:digit:] or \d */
1039 #define cbit_upper 96 /* [:upper:] */
1040 #define cbit_lower 128 /* [:lower:] */
1041 #define cbit_word 160 /* [:word:] or \w */
1042 #define cbit_graph 192 /* [:graph:] */
1043 #define cbit_print 224 /* [:print:] */
1044 #define cbit_punct 256 /* [:punct:] */
1045 #define cbit_cntrl 288 /* [:cntrl:] */
1046 #define cbit_length 320 /* Length of the cbits table */
1048 /* Offsets of the various tables from the base tables pointer, and
1051 #define lcc_offset 0
1052 #define fcc_offset 256
1053 #define cbits_offset 512
1054 #define ctypes_offset (cbits_offset + cbit_length)
1055 #define tables_length (ctypes_offset + 256)
1057 /* Layout of the UCP type table that translates property names into types and
1067 /* Internal shared data tables. These are tables that are used by more than one
1068 of the exported public functions. They have to be "external" in the C sense,
1069 but are not part of the PCRE public API. The data for these tables is in the
1070 pcre_tables.c module. */
1072 extern const int _pcre_utf8_table1[];
1073 extern const int _pcre_utf8_table2[];
1074 extern const int _pcre_utf8_table3[];
1075 extern const uschar _pcre_utf8_table4[];
1077 extern const int _pcre_utf8_table1_size;
1079 extern const ucp_type_table _pcre_utt[];
1080 extern const int _pcre_utt_size;
1082 extern const uschar _pcre_default_tables[];
1084 extern const uschar _pcre_OP_lengths[];
1087 /* Internal shared functions. These are functions that are used by more than
1088 one of the exported public functions. They have to be "external" in the C
1089 sense, but are not part of the PCRE public API. */
1091 extern BOOL _pcre_is_newline(const uschar *, int, const uschar *,
1093 extern int _pcre_ord2utf8(int, uschar *);
1094 extern real_pcre *_pcre_try_flipped(const real_pcre *, real_pcre *,
1095 const pcre_study_data *, pcre_study_data *);
1096 extern int _pcre_ucp_findprop(const unsigned int, int *, int *);
1097 extern unsigned int _pcre_ucp_othercase(const unsigned int);
1098 extern int _pcre_valid_utf8(const uschar *, int);
1099 extern BOOL _pcre_was_newline(const uschar *, int, const uschar *,
1101 extern BOOL _pcre_xclass(int, const uschar *);
1105 /* End of pcre_internal.h */