1 /* $Cambridge: exim/src/src/pcre/pcre_internal.h,v 1.7 2007/11/12 13:02:19 nm4 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 /* Standard C headers plus the external interface definition. The only time
73 setjmp and stdarg are used is when NO_RECURSE is set. */
84 /* When compiling a DLL for Windows, the exported symbols have to be declared
85 using some MS magic. I found some useful information on this web page:
86 http://msdn2.microsoft.com/en-us/library/y4h7bcy6(VS.80).aspx. According to the
87 information there, using __declspec(dllexport) without "extern" we have a
88 definition; with "extern" we have a declaration. The settings here override the
89 setting in pcre.h (which is included below); it defines only PCRE_EXP_DECL,
90 which is all that is needed for applications (they just import the symbols). We
93 PCRE_EXP_DECL for declarations
94 PCRE_EXP_DEFN for definitions of exported functions
95 PCRE_EXP_DATA_DEFN for definitions of exported variables
97 The reason for the two DEFN macros is that in non-Windows environments, one
98 does not want to have "extern" before variable definitions because it leads to
99 compiler warnings. So we distinguish between functions and variables. In
100 Windows, the two should always be the same.
102 The reason for wrapping this in #ifndef PCRE_EXP_DECL is so that pcretest,
103 which is an application, but needs to import this file in order to "peek" at
104 internals, can #include pcre.h first to get an application's-eye view.
106 In principle, people compiling for non-Windows, non-Unix-like (i.e. uncommon,
107 special-purpose environments) might want to stick other stuff in front of
108 exported symbols. That's why, in the non-Windows case, we set PCRE_EXP_DEFN and
109 PCRE_EXP_DATA_DEFN only if they are not already set. */
111 #ifndef PCRE_EXP_DECL
114 # define PCRE_EXP_DECL extern __declspec(dllexport)
115 # define PCRE_EXP_DEFN __declspec(dllexport)
116 # define PCRE_EXP_DATA_DEFN __declspec(dllexport)
118 # define PCRE_EXP_DECL extern
119 # define PCRE_EXP_DEFN
120 # define PCRE_EXP_DATA_DEFN
124 # define PCRE_EXP_DECL extern "C"
126 # define PCRE_EXP_DECL extern
128 # ifndef PCRE_EXP_DEFN
129 # define PCRE_EXP_DEFN PCRE_EXP_DECL
131 # ifndef PCRE_EXP_DATA_DEFN
132 # define PCRE_EXP_DATA_DEFN
137 /* We need to have types that specify unsigned 16-bit and 32-bit integers. We
138 cannot determine these outside the compilation (e.g. by running a program as
139 part of "configure") because PCRE is often cross-compiled for use on other
140 systems. Instead we make use of the maximum sizes that are available at
141 preprocessor time in standard C environments. */
143 #if USHRT_MAX == 65535
144 typedef unsigned short pcre_uint16;
145 #elif UINT_MAX == 65535
146 typedef unsigned int pcre_uint16;
148 #error Cannot determine a type for 16-bit unsigned integers
151 #if UINT_MAX == 4294967295
152 typedef unsigned int pcre_uint32;
153 #elif ULONG_MAX == 4294967295
154 typedef unsigned long int pcre_uint32;
156 #error Cannot determine a type for 32-bit unsigned integers
159 /* All character handling must be done as unsigned characters. Otherwise there
160 are problems with top-bit-set characters and functions such as isspace().
161 However, we leave the interface to the outside world as char *, because that
162 should make things easier for callers. We define a short type for unsigned char
163 to save lots of typing. I tried "uchar", but it causes problems on Digital
164 Unix, where it is defined in sys/types, so use "uschar" instead. */
166 typedef unsigned char uschar;
168 /* This is an unsigned int value that no character can ever have. UTF-8
169 characters only go up to 0x7fffffff (though Unicode doesn't go beyond
172 #define NOTACHAR 0xffffffff
174 /* PCRE is able to support several different kinds of newline (CR, LF, CRLF,
175 "any" and "anycrlf" at present). The following macros are used to package up
176 testing for newlines. NLBLOCK, PSSTART, and PSEND are defined in the various
177 modules to indicate in which datablock the parameters exist, and what the
178 start/end of string field names are. */
180 #define NLTYPE_FIXED 0 /* Newline is a fixed length string */
181 #define NLTYPE_ANY 1 /* Newline is any Unicode line ending */
182 #define NLTYPE_ANYCRLF 2 /* Newline is CR, LF, or CRLF */
184 /* This macro checks for a newline at the given position */
186 #define IS_NEWLINE(p) \
187 ((NLBLOCK->nltype != NLTYPE_FIXED)? \
188 ((p) < NLBLOCK->PSEND && \
189 _pcre_is_newline((p), NLBLOCK->nltype, NLBLOCK->PSEND, &(NLBLOCK->nllen),\
192 ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
193 (p)[0] == NLBLOCK->nl[0] && \
194 (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1]) \
198 /* This macro checks for a newline immediately preceding the given position */
200 #define WAS_NEWLINE(p) \
201 ((NLBLOCK->nltype != NLTYPE_FIXED)? \
202 ((p) > NLBLOCK->PSSTART && \
203 _pcre_was_newline((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \
204 &(NLBLOCK->nllen), utf8)) \
206 ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
207 (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \
208 (NLBLOCK->nllen == 1 || (p)[-NLBLOCK->nllen+1] == NLBLOCK->nl[1]) \
212 /* When PCRE is compiled as a C++ library, the subject pointer can be replaced
213 with a custom type. This makes it possible, for example, to allow pcre_exec()
214 to process subject strings that are discontinuous by using a smart pointer
215 class. It must always be possible to inspect all of the subject string in
216 pcre_exec() because of the way it backtracks. Two macros are required in the
217 normal case, for sign-unspecified and unsigned char pointers. The former is
218 used for the external interface and appears in pcre.h, which is why its name
219 must begin with PCRE_. */
221 #ifdef CUSTOM_SUBJECT_PTR
222 #define PCRE_SPTR CUSTOM_SUBJECT_PTR
223 #define USPTR CUSTOM_SUBJECT_PTR
225 #define PCRE_SPTR const char *
226 #define USPTR const unsigned char *
231 /* Include the public PCRE header and the definitions of UCP character property
237 /* When compiling for use with the Virtual Pascal compiler, these functions
238 need to have their names changed. PCRE must be compiled with the -DVPCOMPAT
239 option on the command line. */
242 #define strlen(s) _strlen(s)
243 #define strncmp(s1,s2,m) _strncmp(s1,s2,m)
244 #define memcmp(s,c,n) _memcmp(s,c,n)
245 #define memcpy(d,s,n) _memcpy(d,s,n)
246 #define memmove(d,s,n) _memmove(d,s,n)
247 #define memset(s,c,n) _memset(s,c,n)
250 /* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
251 define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY
252 is set. Otherwise, include an emulating function for those systems that have
253 neither (there some non-Unix environments where this is the case). */
256 #undef memmove /* some systems may have a macro */
258 #define memmove(a, b, c) bcopy(b, a, c)
259 #else /* HAVE_BCOPY */
261 pcre_memmove(void *d, const void *s, size_t n)
264 unsigned char *dest = (unsigned char *)d;
265 const unsigned char *src = (const unsigned char *)s;
270 for (i = 0; i < n; ++i) *(--dest) = *(--src);
275 for (i = 0; i < n; ++i) *dest++ = *src++;
276 return (void *)(dest - n);
279 #define memmove(a, b, c) pcre_memmove(a, b, c)
280 #endif /* not HAVE_BCOPY */
281 #endif /* not HAVE_MEMMOVE */
282 #endif /* not VPCOMPAT */
285 /* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored
286 in big-endian order) by default. These are used, for example, to link from the
287 start of a subpattern to its alternatives and its end. The use of 2 bytes per
288 offset limits the size of the compiled regex to around 64K, which is big enough
289 for almost everybody. However, I received a request for an even bigger limit.
290 For this reason, and also to make the code easier to maintain, the storing and
291 loading of offsets from the byte string is now handled by the macros that are
294 The macros are controlled by the value of LINK_SIZE. This defaults to 2 in
295 the config.h file, but can be overridden by using -D on the command line. This
296 is automated on Unix systems via the "configure" command. */
302 (a[(n)+1] = (d) & 255)
305 (((a)[n] << 8) | (a)[(n)+1])
307 #define MAX_PATTERN_SIZE (1 << 16)
313 (a[n] = (d) >> 16), \
314 (a[(n)+1] = (d) >> 8), \
315 (a[(n)+2] = (d) & 255)
318 (((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2])
320 #define MAX_PATTERN_SIZE (1 << 24)
326 (a[n] = (d) >> 24), \
327 (a[(n)+1] = (d) >> 16), \
328 (a[(n)+2] = (d) >> 8), \
329 (a[(n)+3] = (d) & 255)
332 (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3])
334 #define MAX_PATTERN_SIZE (1 << 30) /* Keep it positive */
338 #error LINK_SIZE must be either 2, 3, or 4
342 /* Convenience macro defined in terms of the others */
344 #define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE
347 /* PCRE uses some other 2-byte quantities that do not change when the size of
348 offsets changes. There are used for repeat counts and for other things such as
349 capturing parenthesis numbers in back references. */
351 #define PUT2(a,n,d) \
356 (((a)[n] << 8) | (a)[(n)+1])
358 #define PUT2INC(a,n,d) PUT2(a,n,d), a += 2
361 /* When UTF-8 encoding is being used, a character is no longer just a single
362 byte. The macros for character handling generate simple sequences when used in
363 byte-mode, and more complicated ones for UTF-8 characters. BACKCHAR should
364 never be called in byte mode. To make sure it can never even appear when UTF-8
365 support is omitted, we don't even define it. */
368 #define GETCHAR(c, eptr) c = *eptr;
369 #define GETCHARTEST(c, eptr) c = *eptr;
370 #define GETCHARINC(c, eptr) c = *eptr++;
371 #define GETCHARINCTEST(c, eptr) c = *eptr++;
372 #define GETCHARLEN(c, eptr, len) c = *eptr;
373 /* #define BACKCHAR(eptr) */
375 #else /* SUPPORT_UTF8 */
377 /* Get the next UTF-8 character, not advancing the pointer. This is called when
378 we know we are in UTF-8 mode. */
380 #define GETCHAR(c, eptr) \
385 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
387 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
388 for (gcii = 1; gcii <= gcaa; gcii++) \
391 c |= (eptr[gcii] & 0x3f) << gcss; \
395 /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
398 #define GETCHARTEST(c, eptr) \
400 if (utf8 && c >= 0xc0) \
403 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
405 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
406 for (gcii = 1; gcii <= gcaa; gcii++) \
409 c |= (eptr[gcii] & 0x3f) << gcss; \
413 /* Get the next UTF-8 character, advancing the pointer. This is called when we
414 know we are in UTF-8 mode. */
416 #define GETCHARINC(c, eptr) \
420 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
422 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
426 c |= (*eptr++ & 0x3f) << gcss; \
430 /* Get the next character, testing for UTF-8 mode, and advancing the pointer */
432 #define GETCHARINCTEST(c, eptr) \
434 if (utf8 && c >= 0xc0) \
436 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
438 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
442 c |= (*eptr++ & 0x3f) << gcss; \
446 /* Get the next UTF-8 character, not advancing the pointer, incrementing length
447 if there are extra bytes. This is called when we know we are in UTF-8 mode. */
449 #define GETCHARLEN(c, eptr, len) \
454 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
456 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
457 for (gcii = 1; gcii <= gcaa; gcii++) \
460 c |= (eptr[gcii] & 0x3f) << gcss; \
465 /* If the pointer is not at the start of a character, move it back until
466 it is. This is called only in UTF-8 mode - we don't put a test within the macro
467 because almost all calls are already within a block of UTF-8 only code. */
469 #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--
474 /* In case there is no definition of offsetof() provided - though any proper
475 Standard C system should have one. */
478 #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))
482 /* These are the public options that can change during matching. */
484 #define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)
486 /* Private flags containing information about the compiled regex. They used to
487 live at the top end of the options word, but that got almost full, so now they
488 are in a 16-bit flags word. */
490 #define PCRE_NOPARTIAL 0x0001 /* can't use partial with this regex */
491 #define PCRE_FIRSTSET 0x0002 /* first_byte is set */
492 #define PCRE_REQCHSET 0x0004 /* req_byte is set */
493 #define PCRE_STARTLINE 0x0008 /* start after \n for multiline */
494 #define PCRE_JCHANGED 0x0010 /* j option used in regex */
495 #define PCRE_HASCRORLF 0x0020 /* explicit \r or \n in pattern */
497 /* Options for the "extra" block produced by pcre_study(). */
499 #define PCRE_STUDY_MAPPED 0x01 /* a map of starting chars exists */
501 /* Masks for identifying the public options that are permitted at compile
502 time, run time, or study time, respectively. */
504 #define PCRE_NEWLINE_BITS (PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|PCRE_NEWLINE_ANY| \
505 PCRE_NEWLINE_ANYCRLF)
507 #define PUBLIC_OPTIONS \
508 (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
509 PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
510 PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
511 PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)
513 #define PUBLIC_EXEC_OPTIONS \
514 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
515 PCRE_PARTIAL|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)
517 #define PUBLIC_DFA_EXEC_OPTIONS \
518 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
519 PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART|PCRE_NEWLINE_BITS| \
520 PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)
522 #define PUBLIC_STUDY_OPTIONS 0 /* None defined */
524 /* Magic number to provide a small check against being handed junk. Also used
525 to detect whether a pattern was compiled on a host of different endianness. */
527 #define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */
529 /* Negative values for the firstchar and reqchar variables */
531 #define REQ_UNSET (-2)
532 #define REQ_NONE (-1)
534 /* The maximum remaining length of subject we are prepared to search for a
537 #define REQ_BYTE_MAX 1000
539 /* Flags added to firstbyte or reqbyte; a "non-literal" item is either a
540 variable-length repeat, or a anything other than literal characters. */
542 #define REQ_CASELESS 0x0100 /* indicates caselessness */
543 #define REQ_VARY 0x0200 /* reqbyte followed non-literal item */
545 /* Miscellaneous definitions */
552 /* Escape items that are just an encoding of a particular data value. */
570 /* We can't officially use ESC_t because it is a POSIX reserved identifier
571 (presumably because of all the others like size_t). */
577 /* Codes for different types of Unicode property */
579 #define PT_ANY 0 /* Any property - matches all chars */
580 #define PT_LAMP 1 /* L& - the union of Lu, Ll, Lt */
581 #define PT_GC 2 /* General characteristic (e.g. L) */
582 #define PT_PC 3 /* Particular characteristic (e.g. Lu) */
583 #define PT_SC 4 /* Script (e.g. Han) */
585 /* Flag bits and data types for the extended class (OP_XCLASS) for classes that
586 contain UTF-8 characters with values greater than 255. */
588 #define XCL_NOT 0x01 /* Flag: this is a negative class */
589 #define XCL_MAP 0x02 /* Flag: a 32-byte map is present */
591 #define XCL_END 0 /* Marks end of individual items */
592 #define XCL_SINGLE 1 /* Single item (one multibyte char) follows */
593 #define XCL_RANGE 2 /* A range (two multibyte chars) follows */
594 #define XCL_PROP 3 /* Unicode property (2-byte property code follows) */
595 #define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */
597 /* These are escaped items that aren't just an encoding of a particular data
598 value such as \n. They must have non-zero values, as check_escape() returns
599 their negation. Also, they must appear in the same order as in the opcode
600 definitions below, up to ESC_z. There's a dummy for OP_ANY because it
601 corresponds to "." rather than an escape sequence. The final one must be
602 ESC_REF as subsequent values are used for backreferences (\1, \2, \3, etc).
603 There are two tests in the code for an escape greater than ESC_b and less than
604 ESC_Z to detect the types that may be repeated. These are the types that
605 consume characters. If any new escapes are put in between that don't consume a
606 character, that code will have to change. */
608 enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s,
609 ESC_W, ESC_w, ESC_dum1, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H, ESC_h,
610 ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z, ESC_E, ESC_Q, ESC_k, ESC_REF };
613 /* Opcode table: Starting from 1 (i.e. after OP_END), the values up to
614 OP_EOD must correspond in order to the list of escapes immediately above.
616 *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions
617 that follow must also be updated to match. There is also a table called
618 "coptable" in pcre_dfa_exec.c that must be updated. */
621 OP_END, /* 0 End of pattern */
623 /* Values corresponding to backslashed metacharacters */
625 OP_SOD, /* 1 Start of data: \A */
626 OP_SOM, /* 2 Start of match (subject + offset): \G */
627 OP_SET_SOM, /* 3 Set start of match (\K) */
628 OP_NOT_WORD_BOUNDARY, /* 4 \B */
629 OP_WORD_BOUNDARY, /* 5 \b */
630 OP_NOT_DIGIT, /* 6 \D */
632 OP_NOT_WHITESPACE, /* 8 \S */
633 OP_WHITESPACE, /* 9 \s */
634 OP_NOT_WORDCHAR, /* 10 \W */
635 OP_WORDCHAR, /* 11 \w */
636 OP_ANY, /* 12 Match any character */
637 OP_ANYBYTE, /* 13 Match any byte (\C); different to OP_ANY for UTF-8 */
638 OP_NOTPROP, /* 14 \P (not Unicode property) */
639 OP_PROP, /* 15 \p (Unicode property) */
640 OP_ANYNL, /* 16 \R (any newline sequence) */
641 OP_NOT_HSPACE, /* 17 \H (not horizontal whitespace) */
642 OP_HSPACE, /* 18 \h (horizontal whitespace) */
643 OP_NOT_VSPACE, /* 19 \V (not vertical whitespace) */
644 OP_VSPACE, /* 20 \v (vertical whitespace) */
645 OP_EXTUNI, /* 21 \X (extended Unicode sequence */
646 OP_EODN, /* 22 End of data or \n at end of data: \Z. */
647 OP_EOD, /* 23 End of data: \z */
649 OP_OPT, /* 24 Set runtime options */
650 OP_CIRC, /* 25 Start of line - varies with multiline switch */
651 OP_DOLL, /* 26 End of line - varies with multiline switch */
652 OP_CHAR, /* 27 Match one character, casefully */
653 OP_CHARNC, /* 28 Match one character, caselessly */
654 OP_NOT, /* 29 Match one character, not the following one */
656 OP_STAR, /* 30 The maximizing and minimizing versions of */
657 OP_MINSTAR, /* 31 these six opcodes must come in pairs, with */
658 OP_PLUS, /* 32 the minimizing one second. */
659 OP_MINPLUS, /* 33 This first set applies to single characters.*/
661 OP_MINQUERY, /* 35 */
663 OP_UPTO, /* 36 From 0 to n matches */
665 OP_EXACT, /* 38 Exactly n matches */
667 OP_POSSTAR, /* 39 Possessified star */
668 OP_POSPLUS, /* 40 Possessified plus */
669 OP_POSQUERY, /* 41 Posesssified query */
670 OP_POSUPTO, /* 42 Possessified upto */
672 OP_NOTSTAR, /* 43 The maximizing and minimizing versions of */
673 OP_NOTMINSTAR, /* 44 these six opcodes must come in pairs, with */
674 OP_NOTPLUS, /* 45 the minimizing one second. They must be in */
675 OP_NOTMINPLUS, /* 46 exactly the same order as those above. */
676 OP_NOTQUERY, /* 47 This set applies to "not" single characters. */
677 OP_NOTMINQUERY, /* 48 */
679 OP_NOTUPTO, /* 49 From 0 to n matches */
680 OP_NOTMINUPTO, /* 50 */
681 OP_NOTEXACT, /* 51 Exactly n matches */
683 OP_NOTPOSSTAR, /* 52 Possessified versions */
684 OP_NOTPOSPLUS, /* 53 */
685 OP_NOTPOSQUERY, /* 54 */
686 OP_NOTPOSUPTO, /* 55 */
688 OP_TYPESTAR, /* 56 The maximizing and minimizing versions of */
689 OP_TYPEMINSTAR, /* 57 these six opcodes must come in pairs, with */
690 OP_TYPEPLUS, /* 58 the minimizing one second. These codes must */
691 OP_TYPEMINPLUS, /* 59 be in exactly the same order as those above. */
692 OP_TYPEQUERY, /* 60 This set applies to character types such as \d */
693 OP_TYPEMINQUERY, /* 61 */
695 OP_TYPEUPTO, /* 62 From 0 to n matches */
696 OP_TYPEMINUPTO, /* 63 */
697 OP_TYPEEXACT, /* 64 Exactly n matches */
699 OP_TYPEPOSSTAR, /* 65 Possessified versions */
700 OP_TYPEPOSPLUS, /* 66 */
701 OP_TYPEPOSQUERY, /* 67 */
702 OP_TYPEPOSUPTO, /* 68 */
704 OP_CRSTAR, /* 69 The maximizing and minimizing versions of */
705 OP_CRMINSTAR, /* 70 all these opcodes must come in pairs, with */
706 OP_CRPLUS, /* 71 the minimizing one second. These codes must */
707 OP_CRMINPLUS, /* 72 be in exactly the same order as those above. */
708 OP_CRQUERY, /* 73 These are for character classes and back refs */
709 OP_CRMINQUERY, /* 74 */
710 OP_CRRANGE, /* 75 These are different to the three sets above. */
711 OP_CRMINRANGE, /* 76 */
713 OP_CLASS, /* 77 Match a character class, chars < 256 only */
714 OP_NCLASS, /* 78 Same, but the bitmap was created from a negative
715 class - the difference is relevant only when a UTF-8
716 character > 255 is encountered. */
718 OP_XCLASS, /* 79 Extended class for handling UTF-8 chars within the
719 class. This does both positive and negative. */
721 OP_REF, /* 80 Match a back reference */
722 OP_RECURSE, /* 81 Match a numbered subpattern (possibly recursive) */
723 OP_CALLOUT, /* 82 Call out to external function if provided */
725 OP_ALT, /* 83 Start of alternation */
726 OP_KET, /* 84 End of group that doesn't have an unbounded repeat */
727 OP_KETRMAX, /* 85 These two must remain together and in this */
728 OP_KETRMIN, /* 86 order. They are for groups the repeat for ever. */
730 /* The assertions must come before BRA, CBRA, ONCE, and COND.*/
732 OP_ASSERT, /* 87 Positive lookahead */
733 OP_ASSERT_NOT, /* 88 Negative lookahead */
734 OP_ASSERTBACK, /* 89 Positive lookbehind */
735 OP_ASSERTBACK_NOT, /* 90 Negative lookbehind */
736 OP_REVERSE, /* 91 Move pointer back - used in lookbehind assertions */
738 /* ONCE, BRA, CBRA, and COND must come after the assertions, with ONCE first,
739 as there's a test for >= ONCE for a subpattern that isn't an assertion. */
741 OP_ONCE, /* 92 Atomic group */
742 OP_BRA, /* 93 Start of non-capturing bracket */
743 OP_CBRA, /* 94 Start of capturing bracket */
744 OP_COND, /* 95 Conditional group */
746 /* These three must follow the previous three, in the same order. There's a
747 check for >= SBRA to distinguish the two sets. */
749 OP_SBRA, /* 96 Start of non-capturing bracket, check empty */
750 OP_SCBRA, /* 97 Start of capturing bracket, check empty */
751 OP_SCOND, /* 98 Conditional group, check empty */
753 OP_CREF, /* 99 Used to hold a capture number as condition */
754 OP_RREF, /* 100 Used to hold a recursion number as condition */
755 OP_DEF, /* 101 The DEFINE condition */
757 OP_BRAZERO, /* 102 These two must remain together and in this */
758 OP_BRAMINZERO, /* 103 order. */
760 /* These are backtracking control verbs */
767 /* These are forced failure and success verbs */
774 /* This macro defines textual names for all the opcodes. These are used only
775 for debugging. The macro is referenced only in pcre_printint.c. */
777 #define OP_NAME_LIST \
778 "End", "\\A", "\\G", "\\K", "\\B", "\\b", "\\D", "\\d", \
779 "\\S", "\\s", "\\W", "\\w", "Any", "Anybyte", \
780 "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v", \
781 "extuni", "\\Z", "\\z", \
782 "Opt", "^", "$", "char", "charnc", "not", \
783 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
784 "*+","++", "?+", "{", \
785 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
786 "*+","++", "?+", "{", \
787 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
788 "*+","++", "?+", "{", \
789 "*", "*?", "+", "+?", "?", "??", "{", "{", \
790 "class", "nclass", "xclass", "Ref", "Recurse", "Callout", \
791 "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \
792 "AssertB", "AssertB not", "Reverse", \
793 "Once", "Bra", "CBra", "Cond", "SBra", "SCBra", "SCond", \
794 "Cond ref", "Cond rec", "Cond def", "Brazero", "Braminzero", \
795 "*PRUNE", "*SKIP", "*THEN", "*COMMIT", "*FAIL", "*ACCEPT"
798 /* This macro defines the length of fixed length operations in the compiled
799 regex. The lengths are used when searching for specific things, and also in the
800 debugging printing of a compiled regex. We use a macro so that it can be
801 defined close to the definitions of the opcodes themselves.
803 As things have been extended, some of these are no longer fixed lenths, but are
804 minima instead. For example, the length of a single-character repeat may vary
805 in UTF-8 mode. The code that uses this table must know about such things. */
809 1, 1, 1, 1, 1, /* \A, \G, \K, \B, \b */ \
810 1, 1, 1, 1, 1, 1, /* \D, \d, \S, \s, \W, \w */ \
811 1, 1, /* Any, Anybyte */ \
812 3, 3, 1, /* NOTPROP, PROP, EXTUNI */ \
813 1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */ \
814 1, 1, 2, 1, 1, /* \Z, \z, Opt, ^, $ */ \
815 2, /* Char - the minimum length */ \
816 2, /* Charnc - the minimum length */ \
818 /* Positive single-char repeats ** These are */ \
819 2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \
820 4, 4, 4, /* upto, minupto, exact ** UTF-8 mode */ \
821 2, 2, 2, 4, /* *+, ++, ?+, upto+ */ \
822 /* Negative single-char repeats - only for chars < 256 */ \
823 2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \
824 4, 4, 4, /* NOT upto, minupto, exact */ \
825 2, 2, 2, 4, /* Possessive *, +, ?, upto */ \
826 /* Positive type repeats */ \
827 2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \
828 4, 4, 4, /* Type upto, minupto, exact */ \
829 2, 2, 2, 4, /* Possessive *+, ++, ?+, upto+ */ \
830 /* Character class & ref repeats */ \
831 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \
832 5, 5, /* CRRANGE, CRMINRANGE */ \
835 0, /* XCLASS - variable length */ \
837 1+LINK_SIZE, /* RECURSE */ \
838 2+2*LINK_SIZE, /* CALLOUT */ \
839 1+LINK_SIZE, /* Alt */ \
840 1+LINK_SIZE, /* Ket */ \
841 1+LINK_SIZE, /* KetRmax */ \
842 1+LINK_SIZE, /* KetRmin */ \
843 1+LINK_SIZE, /* Assert */ \
844 1+LINK_SIZE, /* Assert not */ \
845 1+LINK_SIZE, /* Assert behind */ \
846 1+LINK_SIZE, /* Assert behind not */ \
847 1+LINK_SIZE, /* Reverse */ \
848 1+LINK_SIZE, /* ONCE */ \
849 1+LINK_SIZE, /* BRA */ \
850 3+LINK_SIZE, /* CBRA */ \
851 1+LINK_SIZE, /* COND */ \
852 1+LINK_SIZE, /* SBRA */ \
853 3+LINK_SIZE, /* SCBRA */ \
854 1+LINK_SIZE, /* SCOND */ \
858 1, 1, /* BRAZERO, BRAMINZERO */ \
859 1, 1, 1, 1, /* PRUNE, SKIP, THEN, COMMIT, */ \
860 1, 1 /* FAIL, ACCEPT */
863 /* A magic value for OP_RREF to indicate the "any recursion" condition. */
865 #define RREF_ANY 0xffff
867 /* Error code numbers. They are given names so that they can more easily be
870 enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9,
871 ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19,
872 ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29,
873 ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
874 ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
875 ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
878 /* The real format of the start of the pcre block; the index of names and the
879 code vector run on as long as necessary after the end. We store an explicit
880 offset to the name table so that if a regex is compiled on one host, saved, and
881 then run on another where the size of pointers is different, all might still
882 be well. For the case of compiled-on-4 and run-on-8, we include an extra
883 pointer that is always NULL. For future-proofing, a few dummy fields were
884 originally included - even though you can never get this planning right - but
885 there is only one left now.
888 Because people can now save and re-use compiled patterns, any additions to this
889 structure should be made at the end, and something earlier (e.g. a new
890 flag in the options or one of the dummy fields) should indicate that the new
891 fields are present. Currently PCRE always sets the dummy fields to zero.
895 typedef struct real_pcre {
896 pcre_uint32 magic_number;
897 pcre_uint32 size; /* Total that was malloced */
898 pcre_uint32 options; /* Public options */
899 pcre_uint16 flags; /* Private flags */
900 pcre_uint16 dummy1; /* For future use */
901 pcre_uint16 top_bracket;
902 pcre_uint16 top_backref;
903 pcre_uint16 first_byte;
904 pcre_uint16 req_byte;
905 pcre_uint16 name_table_offset; /* Offset to name table that follows */
906 pcre_uint16 name_entry_size; /* Size of any name items */
907 pcre_uint16 name_count; /* Number of name items */
908 pcre_uint16 ref_count; /* Reference count */
910 const unsigned char *tables; /* Pointer to tables or NULL for std */
911 const unsigned char *nullpad; /* NULL padding */
914 /* The format of the block used to store data from pcre_study(). The same
915 remark (see NOTE above) about extending this structure applies. */
917 typedef struct pcre_study_data {
918 pcre_uint32 size; /* Total that was malloced */
920 uschar start_bits[32];
923 /* Structure for passing "static" information around between the functions
924 doing the compiling, so that they are thread-safe. */
926 typedef struct compile_data {
927 const uschar *lcc; /* Points to lower casing table */
928 const uschar *fcc; /* Points to case-flipping table */
929 const uschar *cbits; /* Points to character type table */
930 const uschar *ctypes; /* Points to table of type maps */
931 const uschar *start_workspace;/* The start of working space */
932 const uschar *start_code; /* The start of the compiled code */
933 const uschar *start_pattern; /* The start of the pattern */
934 const uschar *end_pattern; /* The end of the pattern */
935 uschar *hwm; /* High watermark of workspace */
936 uschar *name_table; /* The name/number table */
937 int names_found; /* Number of entries so far */
938 int name_entry_size; /* Size of each entry */
939 int bracount; /* Count of capturing parens */
940 int top_backref; /* Maximum back reference */
941 unsigned int backref_map; /* Bitmap of low back refs */
942 int external_options; /* External (initial) options */
943 int external_flags; /* External flag bits to be set */
944 int req_varyopt; /* "After variable item" flag for reqbyte */
945 BOOL had_accept; /* (*ACCEPT) encountered */
946 int nltype; /* Newline type */
947 int nllen; /* Newline string length */
948 uschar nl[4]; /* Newline string when fixed length */
951 /* Structure for maintaining a chain of pointers to the currently incomplete
952 branches, for testing for left recursion. */
954 typedef struct branch_chain {
955 struct branch_chain *outer;
959 /* Structure for items in a linked list that represents an explicit recursive
960 call within the pattern. */
962 typedef struct recursion_info {
963 struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
964 int group_num; /* Number of group that was called */
965 const uschar *after_call; /* "Return value": points after the call in the expr */
966 USPTR save_start; /* Old value of mstart */
967 int *offset_save; /* Pointer to start of saved offsets */
968 int saved_max; /* Number of saved offsets */
971 /* Structure for building a chain of data for holding the values of the subject
972 pointer at the start of each subpattern, so as to detect when an empty string
973 has been matched by a subpattern - to break infinite loops. */
975 typedef struct eptrblock {
976 struct eptrblock *epb_prev;
977 USPTR epb_saved_eptr;
981 /* Structure for passing "static" information around between the functions
982 doing traditional NFA matching, so that they are thread-safe. */
984 typedef struct match_data {
985 unsigned long int match_call_count; /* As it says */
986 unsigned long int match_limit; /* As it says */
987 unsigned long int match_limit_recursion; /* As it says */
988 int *offset_vector; /* Offset vector */
989 int offset_end; /* One past the end */
990 int offset_max; /* The maximum usable for return data */
991 int nltype; /* Newline type */
992 int nllen; /* Newline string length */
993 uschar nl[4]; /* Newline string when fixed */
994 const uschar *lcc; /* Points to lower casing table */
995 const uschar *ctypes; /* Points to table of type maps */
996 BOOL offset_overflow; /* Set if too many extractions */
997 BOOL notbol; /* NOTBOL flag */
998 BOOL noteol; /* NOTEOL flag */
999 BOOL utf8; /* UTF8 flag */
1000 BOOL endonly; /* Dollar not before final \n */
1001 BOOL notempty; /* Empty string match not wanted */
1002 BOOL partial; /* PARTIAL flag */
1003 BOOL hitend; /* Hit the end of the subject at some point */
1004 BOOL bsr_anycrlf; /* \R is just any CRLF, not full Unicode */
1005 const uschar *start_code; /* For use when recursing */
1006 USPTR start_subject; /* Start of the subject string */
1007 USPTR end_subject; /* End of the subject string */
1008 USPTR start_match_ptr; /* Start of matched string */
1009 USPTR end_match_ptr; /* Subject position at end match */
1010 int end_offset_top; /* Highwater mark at end of match */
1011 int capture_last; /* Most recent capture number */
1012 int start_offset; /* The start offset value */
1013 eptrblock *eptrchain; /* Chain of eptrblocks for tail recursions */
1014 int eptrn; /* Next free eptrblock */
1015 recursion_info *recursive; /* Linked list of recursion data */
1016 void *callout_data; /* To pass back to callouts */
1019 /* A similar structure is used for the same purpose by the DFA matching
1022 typedef struct dfa_match_data {
1023 const uschar *start_code; /* Start of the compiled pattern */
1024 const uschar *start_subject; /* Start of the subject string */
1025 const uschar *end_subject; /* End of subject string */
1026 const uschar *tables; /* Character tables */
1027 int moptions; /* Match options */
1028 int poptions; /* Pattern options */
1029 int nltype; /* Newline type */
1030 int nllen; /* Newline string length */
1031 uschar nl[4]; /* Newline string when fixed */
1032 void *callout_data; /* To pass back to callouts */
1035 /* Bit definitions for entries in the pcre_ctypes table. */
1037 #define ctype_space 0x01
1038 #define ctype_letter 0x02
1039 #define ctype_digit 0x04
1040 #define ctype_xdigit 0x08
1041 #define ctype_word 0x10 /* alphameric or '_' */
1042 #define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
1044 /* Offsets for the bitmap tables in pcre_cbits. Each table contains a set
1045 of bits for a class map. Some classes are built by combining these tables. */
1047 #define cbit_space 0 /* [:space:] or \s */
1048 #define cbit_xdigit 32 /* [:xdigit:] */
1049 #define cbit_digit 64 /* [:digit:] or \d */
1050 #define cbit_upper 96 /* [:upper:] */
1051 #define cbit_lower 128 /* [:lower:] */
1052 #define cbit_word 160 /* [:word:] or \w */
1053 #define cbit_graph 192 /* [:graph:] */
1054 #define cbit_print 224 /* [:print:] */
1055 #define cbit_punct 256 /* [:punct:] */
1056 #define cbit_cntrl 288 /* [:cntrl:] */
1057 #define cbit_length 320 /* Length of the cbits table */
1059 /* Offsets of the various tables from the base tables pointer, and
1062 #define lcc_offset 0
1063 #define fcc_offset 256
1064 #define cbits_offset 512
1065 #define ctypes_offset (cbits_offset + cbit_length)
1066 #define tables_length (ctypes_offset + 256)
1068 /* Layout of the UCP type table that translates property names into types and
1069 codes. Each entry used to point directly to a name, but to reduce the number of
1070 relocations in shared libraries, it now has an offset into a single string
1074 pcre_uint16 name_offset;
1080 /* Internal shared data tables. These are tables that are used by more than one
1081 of the exported public functions. They have to be "external" in the C sense,
1082 but are not part of the PCRE public API. The data for these tables is in the
1083 pcre_tables.c module. */
1085 extern const int _pcre_utf8_table1[];
1086 extern const int _pcre_utf8_table2[];
1087 extern const int _pcre_utf8_table3[];
1088 extern const uschar _pcre_utf8_table4[];
1090 extern const int _pcre_utf8_table1_size;
1092 extern const char _pcre_utt_names[];
1093 extern const ucp_type_table _pcre_utt[];
1094 extern const int _pcre_utt_size;
1096 extern const uschar _pcre_default_tables[];
1098 extern const uschar _pcre_OP_lengths[];
1101 /* Internal shared functions. These are functions that are used by more than
1102 one of the exported public functions. They have to be "external" in the C
1103 sense, but are not part of the PCRE public API. */
1105 extern BOOL _pcre_is_newline(const uschar *, int, const uschar *,
1107 extern int _pcre_ord2utf8(int, uschar *);
1108 extern real_pcre *_pcre_try_flipped(const real_pcre *, real_pcre *,
1109 const pcre_study_data *, pcre_study_data *);
1110 extern int _pcre_ucp_findprop(const unsigned int, int *, int *);
1111 extern unsigned int _pcre_ucp_othercase(const unsigned int);
1112 extern int _pcre_valid_utf8(const uschar *, int);
1113 extern BOOL _pcre_was_newline(const uschar *, int, const uschar *,
1115 extern BOOL _pcre_xclass(int, const uschar *);
1119 /* End of pcre_internal.h */