1 /*************************************************
2 * Perl-Compatible Regular Expressions *
3 *************************************************/
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.
9 Written by Philip Hazel
10 Copyright (c) 1997-2008 University of Cambridge
12 -----------------------------------------------------------------------------
13 Redistribution and use in source and binary forms, with or without
14 modification, are permitted provided that the following conditions are met:
16 * Redistributions of source code must retain the above copyright notice,
17 this list of conditions and the following disclaimer.
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.
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.
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 -----------------------------------------------------------------------------
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_". */
45 #ifndef PCRE_INTERNAL_H
46 #define PCRE_INTERNAL_H
48 /* Define DEBUG to get debugging output on stdout. */
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...
59 It turns out that the Mac Debugging.h header also defines the macro DPRINTF, so
60 be absolutely sure we get our version. */
64 #define DPRINTF(p) printf p
66 #define DPRINTF(p) /* Nothing */
70 /* Standard C headers plus the external interface definition. The only time
71 setjmp and stdarg are used is when NO_RECURSE is set. */
82 /* When compiling a DLL for Windows, the exported symbols have to be declared
83 using some MS magic. I found some useful information on this web page:
84 http://msdn2.microsoft.com/en-us/library/y4h7bcy6(VS.80).aspx. According to the
85 information there, using __declspec(dllexport) without "extern" we have a
86 definition; with "extern" we have a declaration. The settings here override the
87 setting in pcre.h (which is included below); it defines only PCRE_EXP_DECL,
88 which is all that is needed for applications (they just import the symbols). We
91 PCRE_EXP_DECL for declarations
92 PCRE_EXP_DEFN for definitions of exported functions
93 PCRE_EXP_DATA_DEFN for definitions of exported variables
95 The reason for the two DEFN macros is that in non-Windows environments, one
96 does not want to have "extern" before variable definitions because it leads to
97 compiler warnings. So we distinguish between functions and variables. In
98 Windows, the two should always be the same.
100 The reason for wrapping this in #ifndef PCRE_EXP_DECL is so that pcretest,
101 which is an application, but needs to import this file in order to "peek" at
102 internals, can #include pcre.h first to get an application's-eye view.
104 In principle, people compiling for non-Windows, non-Unix-like (i.e. uncommon,
105 special-purpose environments) might want to stick other stuff in front of
106 exported symbols. That's why, in the non-Windows case, we set PCRE_EXP_DEFN and
107 PCRE_EXP_DATA_DEFN only if they are not already set. */
109 #ifndef PCRE_EXP_DECL
112 # define PCRE_EXP_DECL extern __declspec(dllexport)
113 # define PCRE_EXP_DEFN __declspec(dllexport)
114 # define PCRE_EXP_DATA_DEFN __declspec(dllexport)
116 # define PCRE_EXP_DECL extern
117 # define PCRE_EXP_DEFN
118 # define PCRE_EXP_DATA_DEFN
122 # define PCRE_EXP_DECL extern "C"
124 # define PCRE_EXP_DECL extern
126 # ifndef PCRE_EXP_DEFN
127 # define PCRE_EXP_DEFN PCRE_EXP_DECL
129 # ifndef PCRE_EXP_DATA_DEFN
130 # define PCRE_EXP_DATA_DEFN
135 /* We need to have types that specify unsigned 16-bit and 32-bit integers. We
136 cannot determine these outside the compilation (e.g. by running a program as
137 part of "configure") because PCRE is often cross-compiled for use on other
138 systems. Instead we make use of the maximum sizes that are available at
139 preprocessor time in standard C environments. */
141 #if USHRT_MAX == 65535
142 typedef unsigned short pcre_uint16;
143 #elif UINT_MAX == 65535
144 typedef unsigned int pcre_uint16;
146 #error Cannot determine a type for 16-bit unsigned integers
149 #if UINT_MAX == 4294967295
150 typedef unsigned int pcre_uint32;
151 #elif ULONG_MAX == 4294967295
152 typedef unsigned long int pcre_uint32;
154 #error Cannot determine a type for 32-bit unsigned integers
157 /* All character handling must be done as unsigned characters. Otherwise there
158 are problems with top-bit-set characters and functions such as isspace().
159 However, we leave the interface to the outside world as char *, because that
160 should make things easier for callers. We define a short type for unsigned char
161 to save lots of typing. I tried "uchar", but it causes problems on Digital
162 Unix, where it is defined in sys/types, so use "uschar" instead. */
164 typedef unsigned char uschar;
166 /* This is an unsigned int value that no character can ever have. UTF-8
167 characters only go up to 0x7fffffff (though Unicode doesn't go beyond
170 #define NOTACHAR 0xffffffff
172 /* PCRE is able to support several different kinds of newline (CR, LF, CRLF,
173 "any" and "anycrlf" at present). The following macros are used to package up
174 testing for newlines. NLBLOCK, PSSTART, and PSEND are defined in the various
175 modules to indicate in which datablock the parameters exist, and what the
176 start/end of string field names are. */
178 #define NLTYPE_FIXED 0 /* Newline is a fixed length string */
179 #define NLTYPE_ANY 1 /* Newline is any Unicode line ending */
180 #define NLTYPE_ANYCRLF 2 /* Newline is CR, LF, or CRLF */
182 /* This macro checks for a newline at the given position */
184 #define IS_NEWLINE(p) \
185 ((NLBLOCK->nltype != NLTYPE_FIXED)? \
186 ((p) < NLBLOCK->PSEND && \
187 _pcre_is_newline((p), NLBLOCK->nltype, NLBLOCK->PSEND, &(NLBLOCK->nllen),\
190 ((p) <= NLBLOCK->PSEND - NLBLOCK->nllen && \
191 (p)[0] == NLBLOCK->nl[0] && \
192 (NLBLOCK->nllen == 1 || (p)[1] == NLBLOCK->nl[1]) \
196 /* This macro checks for a newline immediately preceding the given position */
198 #define WAS_NEWLINE(p) \
199 ((NLBLOCK->nltype != NLTYPE_FIXED)? \
200 ((p) > NLBLOCK->PSSTART && \
201 _pcre_was_newline((p), NLBLOCK->nltype, NLBLOCK->PSSTART, \
202 &(NLBLOCK->nllen), utf8)) \
204 ((p) >= NLBLOCK->PSSTART + NLBLOCK->nllen && \
205 (p)[-NLBLOCK->nllen] == NLBLOCK->nl[0] && \
206 (NLBLOCK->nllen == 1 || (p)[-NLBLOCK->nllen+1] == NLBLOCK->nl[1]) \
210 /* When PCRE is compiled as a C++ library, the subject pointer can be replaced
211 with a custom type. This makes it possible, for example, to allow pcre_exec()
212 to process subject strings that are discontinuous by using a smart pointer
213 class. It must always be possible to inspect all of the subject string in
214 pcre_exec() because of the way it backtracks. Two macros are required in the
215 normal case, for sign-unspecified and unsigned char pointers. The former is
216 used for the external interface and appears in pcre.h, which is why its name
217 must begin with PCRE_. */
219 #ifdef CUSTOM_SUBJECT_PTR
220 #define PCRE_SPTR CUSTOM_SUBJECT_PTR
221 #define USPTR CUSTOM_SUBJECT_PTR
223 #define PCRE_SPTR const char *
224 #define USPTR const unsigned char *
229 /* Include the public PCRE header and the definitions of UCP character property
235 /* When compiling for use with the Virtual Pascal compiler, these functions
236 need to have their names changed. PCRE must be compiled with the -DVPCOMPAT
237 option on the command line. */
240 #define strlen(s) _strlen(s)
241 #define strncmp(s1,s2,m) _strncmp(s1,s2,m)
242 #define memcmp(s,c,n) _memcmp(s,c,n)
243 #define memcpy(d,s,n) _memcpy(d,s,n)
244 #define memmove(d,s,n) _memmove(d,s,n)
245 #define memset(s,c,n) _memset(s,c,n)
248 /* To cope with SunOS4 and other systems that lack memmove() but have bcopy(),
249 define a macro for memmove() if HAVE_MEMMOVE is false, provided that HAVE_BCOPY
250 is set. Otherwise, include an emulating function for those systems that have
251 neither (there some non-Unix environments where this is the case). */
254 #undef memmove /* some systems may have a macro */
256 #define memmove(a, b, c) bcopy(b, a, c)
257 #else /* HAVE_BCOPY */
259 pcre_memmove(void *d, const void *s, size_t n)
262 unsigned char *dest = (unsigned char *)d;
263 const unsigned char *src = (const unsigned char *)s;
268 for (i = 0; i < n; ++i) *(--dest) = *(--src);
273 for (i = 0; i < n; ++i) *dest++ = *src++;
274 return (void *)(dest - n);
277 #define memmove(a, b, c) pcre_memmove(a, b, c)
278 #endif /* not HAVE_BCOPY */
279 #endif /* not HAVE_MEMMOVE */
280 #endif /* not VPCOMPAT */
283 /* PCRE keeps offsets in its compiled code as 2-byte quantities (always stored
284 in big-endian order) by default. These are used, for example, to link from the
285 start of a subpattern to its alternatives and its end. The use of 2 bytes per
286 offset limits the size of the compiled regex to around 64K, which is big enough
287 for almost everybody. However, I received a request for an even bigger limit.
288 For this reason, and also to make the code easier to maintain, the storing and
289 loading of offsets from the byte string is now handled by the macros that are
292 The macros are controlled by the value of LINK_SIZE. This defaults to 2 in
293 the config.h file, but can be overridden by using -D on the command line. This
294 is automated on Unix systems via the "configure" command. */
300 (a[(n)+1] = (d) & 255)
303 (((a)[n] << 8) | (a)[(n)+1])
305 #define MAX_PATTERN_SIZE (1 << 16)
311 (a[n] = (d) >> 16), \
312 (a[(n)+1] = (d) >> 8), \
313 (a[(n)+2] = (d) & 255)
316 (((a)[n] << 16) | ((a)[(n)+1] << 8) | (a)[(n)+2])
318 #define MAX_PATTERN_SIZE (1 << 24)
324 (a[n] = (d) >> 24), \
325 (a[(n)+1] = (d) >> 16), \
326 (a[(n)+2] = (d) >> 8), \
327 (a[(n)+3] = (d) & 255)
330 (((a)[n] << 24) | ((a)[(n)+1] << 16) | ((a)[(n)+2] << 8) | (a)[(n)+3])
332 #define MAX_PATTERN_SIZE (1 << 30) /* Keep it positive */
336 #error LINK_SIZE must be either 2, 3, or 4
340 /* Convenience macro defined in terms of the others */
342 #define PUTINC(a,n,d) PUT(a,n,d), a += LINK_SIZE
345 /* PCRE uses some other 2-byte quantities that do not change when the size of
346 offsets changes. There are used for repeat counts and for other things such as
347 capturing parenthesis numbers in back references. */
349 #define PUT2(a,n,d) \
354 (((a)[n] << 8) | (a)[(n)+1])
356 #define PUT2INC(a,n,d) PUT2(a,n,d), a += 2
359 /* When UTF-8 encoding is being used, a character is no longer just a single
360 byte. The macros for character handling generate simple sequences when used in
361 byte-mode, and more complicated ones for UTF-8 characters. BACKCHAR should
362 never be called in byte mode. To make sure it can never even appear when UTF-8
363 support is omitted, we don't even define it. */
366 #define NEXTCHAR(p) p++;
367 #define GETCHAR(c, eptr) c = *eptr;
368 #define GETCHARTEST(c, eptr) c = *eptr;
369 #define GETCHARINC(c, eptr) c = *eptr++;
370 #define GETCHARINCTEST(c, eptr) c = *eptr++;
371 #define GETCHARLEN(c, eptr, len) c = *eptr;
372 /* #define BACKCHAR(eptr) */
374 #else /* SUPPORT_UTF8 */
376 /* Advance a character pointer one byte in non-UTF-8 mode and by one character
379 #define NEXTCHAR(p) \
381 if (utf8) { while((*p & 0xc0) == 0x80) p++; }
383 /* Get the next UTF-8 character, not advancing the pointer. This is called when
384 we know we are in UTF-8 mode. */
386 #define GETCHAR(c, eptr) \
391 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
393 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
394 for (gcii = 1; gcii <= gcaa; gcii++) \
397 c |= (eptr[gcii] & 0x3f) << gcss; \
401 /* Get the next UTF-8 character, testing for UTF-8 mode, and not advancing the
404 #define GETCHARTEST(c, eptr) \
406 if (utf8 && c >= 0xc0) \
409 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
411 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
412 for (gcii = 1; gcii <= gcaa; gcii++) \
415 c |= (eptr[gcii] & 0x3f) << gcss; \
419 /* Get the next UTF-8 character, advancing the pointer. This is called when we
420 know we are in UTF-8 mode. */
422 #define GETCHARINC(c, eptr) \
426 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
428 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
432 c |= (*eptr++ & 0x3f) << gcss; \
436 /* Get the next character, testing for UTF-8 mode, and advancing the pointer */
438 #define GETCHARINCTEST(c, eptr) \
440 if (utf8 && c >= 0xc0) \
442 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
444 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
448 c |= (*eptr++ & 0x3f) << gcss; \
452 /* Get the next UTF-8 character, not advancing the pointer, incrementing length
453 if there are extra bytes. This is called when we know we are in UTF-8 mode. */
455 #define GETCHARLEN(c, eptr, len) \
460 int gcaa = _pcre_utf8_table4[c & 0x3f]; /* Number of additional bytes */ \
462 c = (c & _pcre_utf8_table3[gcaa]) << gcss; \
463 for (gcii = 1; gcii <= gcaa; gcii++) \
466 c |= (eptr[gcii] & 0x3f) << gcss; \
471 /* If the pointer is not at the start of a character, move it back until
472 it is. This is called only in UTF-8 mode - we don't put a test within the macro
473 because almost all calls are already within a block of UTF-8 only code. */
475 #define BACKCHAR(eptr) while((*eptr & 0xc0) == 0x80) eptr--
480 /* In case there is no definition of offsetof() provided - though any proper
481 Standard C system should have one. */
484 #define offsetof(p_type,field) ((size_t)&(((p_type *)0)->field))
488 /* These are the public options that can change during matching. */
490 #define PCRE_IMS (PCRE_CASELESS|PCRE_MULTILINE|PCRE_DOTALL)
492 /* Private flags containing information about the compiled regex. They used to
493 live at the top end of the options word, but that got almost full, so now they
494 are in a 16-bit flags word. */
496 #define PCRE_NOPARTIAL 0x0001 /* can't use partial with this regex */
497 #define PCRE_FIRSTSET 0x0002 /* first_byte is set */
498 #define PCRE_REQCHSET 0x0004 /* req_byte is set */
499 #define PCRE_STARTLINE 0x0008 /* start after \n for multiline */
500 #define PCRE_JCHANGED 0x0010 /* j option used in regex */
501 #define PCRE_HASCRORLF 0x0020 /* explicit \r or \n in pattern */
503 /* Options for the "extra" block produced by pcre_study(). */
505 #define PCRE_STUDY_MAPPED 0x01 /* a map of starting chars exists */
507 /* Masks for identifying the public options that are permitted at compile
508 time, run time, or study time, respectively. */
510 #define PCRE_NEWLINE_BITS (PCRE_NEWLINE_CR|PCRE_NEWLINE_LF|PCRE_NEWLINE_ANY| \
511 PCRE_NEWLINE_ANYCRLF)
513 #define PUBLIC_OPTIONS \
514 (PCRE_CASELESS|PCRE_EXTENDED|PCRE_ANCHORED|PCRE_MULTILINE| \
515 PCRE_DOTALL|PCRE_DOLLAR_ENDONLY|PCRE_EXTRA|PCRE_UNGREEDY|PCRE_UTF8| \
516 PCRE_NO_AUTO_CAPTURE|PCRE_NO_UTF8_CHECK|PCRE_AUTO_CALLOUT|PCRE_FIRSTLINE| \
517 PCRE_DUPNAMES|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE| \
518 PCRE_JAVASCRIPT_COMPAT)
520 #define PUBLIC_EXEC_OPTIONS \
521 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
522 PCRE_PARTIAL|PCRE_NEWLINE_BITS|PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)
524 #define PUBLIC_DFA_EXEC_OPTIONS \
525 (PCRE_ANCHORED|PCRE_NOTBOL|PCRE_NOTEOL|PCRE_NOTEMPTY|PCRE_NO_UTF8_CHECK| \
526 PCRE_PARTIAL|PCRE_DFA_SHORTEST|PCRE_DFA_RESTART|PCRE_NEWLINE_BITS| \
527 PCRE_BSR_ANYCRLF|PCRE_BSR_UNICODE)
529 #define PUBLIC_STUDY_OPTIONS 0 /* None defined */
531 /* Magic number to provide a small check against being handed junk. Also used
532 to detect whether a pattern was compiled on a host of different endianness. */
534 #define MAGIC_NUMBER 0x50435245UL /* 'PCRE' */
536 /* Negative values for the firstchar and reqchar variables */
538 #define REQ_UNSET (-2)
539 #define REQ_NONE (-1)
541 /* The maximum remaining length of subject we are prepared to search for a
544 #define REQ_BYTE_MAX 1000
546 /* Flags added to firstbyte or reqbyte; a "non-literal" item is either a
547 variable-length repeat, or a anything other than literal characters. */
549 #define REQ_CASELESS 0x0100 /* indicates caselessness */
550 #define REQ_VARY 0x0200 /* reqbyte followed non-literal item */
552 /* Miscellaneous definitions */
559 /* Escape items that are just an encoding of a particular data value. */
577 /* We can't officially use ESC_t because it is a POSIX reserved identifier
578 (presumably because of all the others like size_t). */
584 /* Codes for different types of Unicode property */
586 #define PT_ANY 0 /* Any property - matches all chars */
587 #define PT_LAMP 1 /* L& - the union of Lu, Ll, Lt */
588 #define PT_GC 2 /* General characteristic (e.g. L) */
589 #define PT_PC 3 /* Particular characteristic (e.g. Lu) */
590 #define PT_SC 4 /* Script (e.g. Han) */
592 /* Flag bits and data types for the extended class (OP_XCLASS) for classes that
593 contain UTF-8 characters with values greater than 255. */
595 #define XCL_NOT 0x01 /* Flag: this is a negative class */
596 #define XCL_MAP 0x02 /* Flag: a 32-byte map is present */
598 #define XCL_END 0 /* Marks end of individual items */
599 #define XCL_SINGLE 1 /* Single item (one multibyte char) follows */
600 #define XCL_RANGE 2 /* A range (two multibyte chars) follows */
601 #define XCL_PROP 3 /* Unicode property (2-byte property code follows) */
602 #define XCL_NOTPROP 4 /* Unicode inverted property (ditto) */
604 /* These are escaped items that aren't just an encoding of a particular data
605 value such as \n. They must have non-zero values, as check_escape() returns
606 their negation. Also, they must appear in the same order as in the opcode
607 definitions below, up to ESC_z. There's a dummy for OP_ANY because it
608 corresponds to "." rather than an escape sequence, and another for OP_ALLANY
609 (which is used for [^] in JavaScript compatibility mode).
611 The final escape must be ESC_REF as subsequent values are used for
612 backreferences (\1, \2, \3, etc). There are two tests in the code for an escape
613 greater than ESC_b and less than ESC_Z to detect the types that may be
614 repeated. These are the types that consume characters. If any new escapes are
615 put in between that don't consume a character, that code will have to change.
618 enum { ESC_A = 1, ESC_G, ESC_K, ESC_B, ESC_b, ESC_D, ESC_d, ESC_S, ESC_s,
619 ESC_W, ESC_w, ESC_dum1, ESC_dum2, ESC_C, ESC_P, ESC_p, ESC_R, ESC_H,
620 ESC_h, ESC_V, ESC_v, ESC_X, ESC_Z, ESC_z, ESC_E, ESC_Q, ESC_g, ESC_k,
624 /* Opcode table: Starting from 1 (i.e. after OP_END), the values up to
625 OP_EOD must correspond in order to the list of escapes immediately above.
627 *** NOTE NOTE NOTE *** Whenever this list is updated, the two macro definitions
628 that follow must also be updated to match. There is also a table called
629 "coptable" in pcre_dfa_exec.c that must be updated. */
632 OP_END, /* 0 End of pattern */
634 /* Values corresponding to backslashed metacharacters */
636 OP_SOD, /* 1 Start of data: \A */
637 OP_SOM, /* 2 Start of match (subject + offset): \G */
638 OP_SET_SOM, /* 3 Set start of match (\K) */
639 OP_NOT_WORD_BOUNDARY, /* 4 \B */
640 OP_WORD_BOUNDARY, /* 5 \b */
641 OP_NOT_DIGIT, /* 6 \D */
643 OP_NOT_WHITESPACE, /* 8 \S */
644 OP_WHITESPACE, /* 9 \s */
645 OP_NOT_WORDCHAR, /* 10 \W */
646 OP_WORDCHAR, /* 11 \w */
647 OP_ANY, /* 12 Match any character (subject to DOTALL) */
648 OP_ALLANY, /* 13 Match any character (not subject to DOTALL) */
649 OP_ANYBYTE, /* 14 Match any byte (\C); different to OP_ANY for UTF-8 */
650 OP_NOTPROP, /* 15 \P (not Unicode property) */
651 OP_PROP, /* 16 \p (Unicode property) */
652 OP_ANYNL, /* 17 \R (any newline sequence) */
653 OP_NOT_HSPACE, /* 18 \H (not horizontal whitespace) */
654 OP_HSPACE, /* 19 \h (horizontal whitespace) */
655 OP_NOT_VSPACE, /* 20 \V (not vertical whitespace) */
656 OP_VSPACE, /* 21 \v (vertical whitespace) */
657 OP_EXTUNI, /* 22 \X (extended Unicode sequence */
658 OP_EODN, /* 23 End of data or \n at end of data: \Z. */
659 OP_EOD, /* 24 End of data: \z */
661 OP_OPT, /* 25 Set runtime options */
662 OP_CIRC, /* 26 Start of line - varies with multiline switch */
663 OP_DOLL, /* 27 End of line - varies with multiline switch */
664 OP_CHAR, /* 28 Match one character, casefully */
665 OP_CHARNC, /* 29 Match one character, caselessly */
666 OP_NOT, /* 30 Match one character, not the following one */
668 OP_STAR, /* 31 The maximizing and minimizing versions of */
669 OP_MINSTAR, /* 32 these six opcodes must come in pairs, with */
670 OP_PLUS, /* 33 the minimizing one second. */
671 OP_MINPLUS, /* 34 This first set applies to single characters.*/
673 OP_MINQUERY, /* 36 */
675 OP_UPTO, /* 37 From 0 to n matches */
677 OP_EXACT, /* 39 Exactly n matches */
679 OP_POSSTAR, /* 40 Possessified star */
680 OP_POSPLUS, /* 41 Possessified plus */
681 OP_POSQUERY, /* 42 Posesssified query */
682 OP_POSUPTO, /* 43 Possessified upto */
684 OP_NOTSTAR, /* 44 The maximizing and minimizing versions of */
685 OP_NOTMINSTAR, /* 45 these six opcodes must come in pairs, with */
686 OP_NOTPLUS, /* 46 the minimizing one second. They must be in */
687 OP_NOTMINPLUS, /* 47 exactly the same order as those above. */
688 OP_NOTQUERY, /* 48 This set applies to "not" single characters. */
689 OP_NOTMINQUERY, /* 49 */
691 OP_NOTUPTO, /* 50 From 0 to n matches */
692 OP_NOTMINUPTO, /* 51 */
693 OP_NOTEXACT, /* 52 Exactly n matches */
695 OP_NOTPOSSTAR, /* 53 Possessified versions */
696 OP_NOTPOSPLUS, /* 54 */
697 OP_NOTPOSQUERY, /* 55 */
698 OP_NOTPOSUPTO, /* 56 */
700 OP_TYPESTAR, /* 57 The maximizing and minimizing versions of */
701 OP_TYPEMINSTAR, /* 58 these six opcodes must come in pairs, with */
702 OP_TYPEPLUS, /* 59 the minimizing one second. These codes must */
703 OP_TYPEMINPLUS, /* 60 be in exactly the same order as those above. */
704 OP_TYPEQUERY, /* 61 This set applies to character types such as \d */
705 OP_TYPEMINQUERY, /* 62 */
707 OP_TYPEUPTO, /* 63 From 0 to n matches */
708 OP_TYPEMINUPTO, /* 64 */
709 OP_TYPEEXACT, /* 65 Exactly n matches */
711 OP_TYPEPOSSTAR, /* 66 Possessified versions */
712 OP_TYPEPOSPLUS, /* 67 */
713 OP_TYPEPOSQUERY, /* 68 */
714 OP_TYPEPOSUPTO, /* 69 */
716 OP_CRSTAR, /* 70 The maximizing and minimizing versions of */
717 OP_CRMINSTAR, /* 71 all these opcodes must come in pairs, with */
718 OP_CRPLUS, /* 72 the minimizing one second. These codes must */
719 OP_CRMINPLUS, /* 73 be in exactly the same order as those above. */
720 OP_CRQUERY, /* 74 These are for character classes and back refs */
721 OP_CRMINQUERY, /* 75 */
722 OP_CRRANGE, /* 76 These are different to the three sets above. */
723 OP_CRMINRANGE, /* 77 */
725 OP_CLASS, /* 78 Match a character class, chars < 256 only */
726 OP_NCLASS, /* 79 Same, but the bitmap was created from a negative
727 class - the difference is relevant only when a UTF-8
728 character > 255 is encountered. */
730 OP_XCLASS, /* 80 Extended class for handling UTF-8 chars within the
731 class. This does both positive and negative. */
733 OP_REF, /* 81 Match a back reference */
734 OP_RECURSE, /* 82 Match a numbered subpattern (possibly recursive) */
735 OP_CALLOUT, /* 83 Call out to external function if provided */
737 OP_ALT, /* 84 Start of alternation */
738 OP_KET, /* 85 End of group that doesn't have an unbounded repeat */
739 OP_KETRMAX, /* 86 These two must remain together and in this */
740 OP_KETRMIN, /* 87 order. They are for groups the repeat for ever. */
742 /* The assertions must come before BRA, CBRA, ONCE, and COND.*/
744 OP_ASSERT, /* 88 Positive lookahead */
745 OP_ASSERT_NOT, /* 89 Negative lookahead */
746 OP_ASSERTBACK, /* 90 Positive lookbehind */
747 OP_ASSERTBACK_NOT, /* 91 Negative lookbehind */
748 OP_REVERSE, /* 92 Move pointer back - used in lookbehind assertions */
750 /* ONCE, BRA, CBRA, and COND must come after the assertions, with ONCE first,
751 as there's a test for >= ONCE for a subpattern that isn't an assertion. */
753 OP_ONCE, /* 93 Atomic group */
754 OP_BRA, /* 94 Start of non-capturing bracket */
755 OP_CBRA, /* 95 Start of capturing bracket */
756 OP_COND, /* 96 Conditional group */
758 /* These three must follow the previous three, in the same order. There's a
759 check for >= SBRA to distinguish the two sets. */
761 OP_SBRA, /* 97 Start of non-capturing bracket, check empty */
762 OP_SCBRA, /* 98 Start of capturing bracket, check empty */
763 OP_SCOND, /* 99 Conditional group, check empty */
765 OP_CREF, /* 100 Used to hold a capture number as condition */
766 OP_RREF, /* 101 Used to hold a recursion number as condition */
767 OP_DEF, /* 102 The DEFINE condition */
769 OP_BRAZERO, /* 103 These two must remain together and in this */
770 OP_BRAMINZERO, /* 104 order. */
772 /* These are backtracking control verbs */
779 /* These are forced failure and success verbs */
784 /* This is used to skip a subpattern with a {0} quantifier */
786 OP_SKIPZERO /* 111 */
790 /* This macro defines textual names for all the opcodes. These are used only
791 for debugging. The macro is referenced only in pcre_printint.c. */
793 #define OP_NAME_LIST \
794 "End", "\\A", "\\G", "\\K", "\\B", "\\b", "\\D", "\\d", \
795 "\\S", "\\s", "\\W", "\\w", "Any", "AllAny", "Anybyte", \
796 "notprop", "prop", "\\R", "\\H", "\\h", "\\V", "\\v", \
797 "extuni", "\\Z", "\\z", \
798 "Opt", "^", "$", "char", "charnc", "not", \
799 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
800 "*+","++", "?+", "{", \
801 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
802 "*+","++", "?+", "{", \
803 "*", "*?", "+", "+?", "?", "??", "{", "{", "{", \
804 "*+","++", "?+", "{", \
805 "*", "*?", "+", "+?", "?", "??", "{", "{", \
806 "class", "nclass", "xclass", "Ref", "Recurse", "Callout", \
807 "Alt", "Ket", "KetRmax", "KetRmin", "Assert", "Assert not", \
808 "AssertB", "AssertB not", "Reverse", \
809 "Once", "Bra", "CBra", "Cond", "SBra", "SCBra", "SCond", \
810 "Cond ref", "Cond rec", "Cond def", "Brazero", "Braminzero", \
811 "*PRUNE", "*SKIP", "*THEN", "*COMMIT", "*FAIL", "*ACCEPT", \
815 /* This macro defines the length of fixed length operations in the compiled
816 regex. The lengths are used when searching for specific things, and also in the
817 debugging printing of a compiled regex. We use a macro so that it can be
818 defined close to the definitions of the opcodes themselves.
820 As things have been extended, some of these are no longer fixed lenths, but are
821 minima instead. For example, the length of a single-character repeat may vary
822 in UTF-8 mode. The code that uses this table must know about such things. */
826 1, 1, 1, 1, 1, /* \A, \G, \K, \B, \b */ \
827 1, 1, 1, 1, 1, 1, /* \D, \d, \S, \s, \W, \w */ \
828 1, 1, 1, /* Any, AllAny, Anybyte */ \
829 3, 3, 1, /* NOTPROP, PROP, EXTUNI */ \
830 1, 1, 1, 1, 1, /* \R, \H, \h, \V, \v */ \
831 1, 1, 2, 1, 1, /* \Z, \z, Opt, ^, $ */ \
832 2, /* Char - the minimum length */ \
833 2, /* Charnc - the minimum length */ \
835 /* Positive single-char repeats ** These are */ \
836 2, 2, 2, 2, 2, 2, /* *, *?, +, +?, ?, ?? ** minima in */ \
837 4, 4, 4, /* upto, minupto, exact ** UTF-8 mode */ \
838 2, 2, 2, 4, /* *+, ++, ?+, upto+ */ \
839 /* Negative single-char repeats - only for chars < 256 */ \
840 2, 2, 2, 2, 2, 2, /* NOT *, *?, +, +?, ?, ?? */ \
841 4, 4, 4, /* NOT upto, minupto, exact */ \
842 2, 2, 2, 4, /* Possessive *, +, ?, upto */ \
843 /* Positive type repeats */ \
844 2, 2, 2, 2, 2, 2, /* Type *, *?, +, +?, ?, ?? */ \
845 4, 4, 4, /* Type upto, minupto, exact */ \
846 2, 2, 2, 4, /* Possessive *+, ++, ?+, upto+ */ \
847 /* Character class & ref repeats */ \
848 1, 1, 1, 1, 1, 1, /* *, *?, +, +?, ?, ?? */ \
849 5, 5, /* CRRANGE, CRMINRANGE */ \
852 0, /* XCLASS - variable length */ \
854 1+LINK_SIZE, /* RECURSE */ \
855 2+2*LINK_SIZE, /* CALLOUT */ \
856 1+LINK_SIZE, /* Alt */ \
857 1+LINK_SIZE, /* Ket */ \
858 1+LINK_SIZE, /* KetRmax */ \
859 1+LINK_SIZE, /* KetRmin */ \
860 1+LINK_SIZE, /* Assert */ \
861 1+LINK_SIZE, /* Assert not */ \
862 1+LINK_SIZE, /* Assert behind */ \
863 1+LINK_SIZE, /* Assert behind not */ \
864 1+LINK_SIZE, /* Reverse */ \
865 1+LINK_SIZE, /* ONCE */ \
866 1+LINK_SIZE, /* BRA */ \
867 3+LINK_SIZE, /* CBRA */ \
868 1+LINK_SIZE, /* COND */ \
869 1+LINK_SIZE, /* SBRA */ \
870 3+LINK_SIZE, /* SCBRA */ \
871 1+LINK_SIZE, /* SCOND */ \
875 1, 1, /* BRAZERO, BRAMINZERO */ \
876 1, 1, 1, 1, /* PRUNE, SKIP, THEN, COMMIT, */ \
877 1, 1, 1 /* FAIL, ACCEPT, SKIPZERO */
880 /* A magic value for OP_RREF to indicate the "any recursion" condition. */
882 #define RREF_ANY 0xffff
884 /* Error code numbers. They are given names so that they can more easily be
887 enum { ERR0, ERR1, ERR2, ERR3, ERR4, ERR5, ERR6, ERR7, ERR8, ERR9,
888 ERR10, ERR11, ERR12, ERR13, ERR14, ERR15, ERR16, ERR17, ERR18, ERR19,
889 ERR20, ERR21, ERR22, ERR23, ERR24, ERR25, ERR26, ERR27, ERR28, ERR29,
890 ERR30, ERR31, ERR32, ERR33, ERR34, ERR35, ERR36, ERR37, ERR38, ERR39,
891 ERR40, ERR41, ERR42, ERR43, ERR44, ERR45, ERR46, ERR47, ERR48, ERR49,
892 ERR50, ERR51, ERR52, ERR53, ERR54, ERR55, ERR56, ERR57, ERR58, ERR59,
893 ERR60, ERR61, ERR62, ERR63, ERR64 };
895 /* The real format of the start of the pcre block; the index of names and the
896 code vector run on as long as necessary after the end. We store an explicit
897 offset to the name table so that if a regex is compiled on one host, saved, and
898 then run on another where the size of pointers is different, all might still
899 be well. For the case of compiled-on-4 and run-on-8, we include an extra
900 pointer that is always NULL. For future-proofing, a few dummy fields were
901 originally included - even though you can never get this planning right - but
902 there is only one left now.
905 Because people can now save and re-use compiled patterns, any additions to this
906 structure should be made at the end, and something earlier (e.g. a new
907 flag in the options or one of the dummy fields) should indicate that the new
908 fields are present. Currently PCRE always sets the dummy fields to zero.
912 typedef struct real_pcre {
913 pcre_uint32 magic_number;
914 pcre_uint32 size; /* Total that was malloced */
915 pcre_uint32 options; /* Public options */
916 pcre_uint16 flags; /* Private flags */
917 pcre_uint16 dummy1; /* For future use */
918 pcre_uint16 top_bracket;
919 pcre_uint16 top_backref;
920 pcre_uint16 first_byte;
921 pcre_uint16 req_byte;
922 pcre_uint16 name_table_offset; /* Offset to name table that follows */
923 pcre_uint16 name_entry_size; /* Size of any name items */
924 pcre_uint16 name_count; /* Number of name items */
925 pcre_uint16 ref_count; /* Reference count */
927 const unsigned char *tables; /* Pointer to tables or NULL for std */
928 const unsigned char *nullpad; /* NULL padding */
931 /* The format of the block used to store data from pcre_study(). The same
932 remark (see NOTE above) about extending this structure applies. */
934 typedef struct pcre_study_data {
935 pcre_uint32 size; /* Total that was malloced */
937 uschar start_bits[32];
940 /* Structure for passing "static" information around between the functions
941 doing the compiling, so that they are thread-safe. */
943 typedef struct compile_data {
944 const uschar *lcc; /* Points to lower casing table */
945 const uschar *fcc; /* Points to case-flipping table */
946 const uschar *cbits; /* Points to character type table */
947 const uschar *ctypes; /* Points to table of type maps */
948 const uschar *start_workspace;/* The start of working space */
949 const uschar *start_code; /* The start of the compiled code */
950 const uschar *start_pattern; /* The start of the pattern */
951 const uschar *end_pattern; /* The end of the pattern */
952 uschar *hwm; /* High watermark of workspace */
953 uschar *name_table; /* The name/number table */
954 int names_found; /* Number of entries so far */
955 int name_entry_size; /* Size of each entry */
956 int bracount; /* Count of capturing parens as we compile */
957 int final_bracount; /* Saved value after first pass */
958 int top_backref; /* Maximum back reference */
959 unsigned int backref_map; /* Bitmap of low back refs */
960 int external_options; /* External (initial) options */
961 int external_flags; /* External flag bits to be set */
962 int req_varyopt; /* "After variable item" flag for reqbyte */
963 BOOL had_accept; /* (*ACCEPT) encountered */
964 int nltype; /* Newline type */
965 int nllen; /* Newline string length */
966 uschar nl[4]; /* Newline string when fixed length */
969 /* Structure for maintaining a chain of pointers to the currently incomplete
970 branches, for testing for left recursion. */
972 typedef struct branch_chain {
973 struct branch_chain *outer;
977 /* Structure for items in a linked list that represents an explicit recursive
978 call within the pattern. */
980 typedef struct recursion_info {
981 struct recursion_info *prevrec; /* Previous recursion record (or NULL) */
982 int group_num; /* Number of group that was called */
983 const uschar *after_call; /* "Return value": points after the call in the expr */
984 USPTR save_start; /* Old value of mstart */
985 int *offset_save; /* Pointer to start of saved offsets */
986 int saved_max; /* Number of saved offsets */
989 /* Structure for building a chain of data for holding the values of the subject
990 pointer at the start of each subpattern, so as to detect when an empty string
991 has been matched by a subpattern - to break infinite loops. */
993 typedef struct eptrblock {
994 struct eptrblock *epb_prev;
995 USPTR epb_saved_eptr;
999 /* Structure for passing "static" information around between the functions
1000 doing traditional NFA matching, so that they are thread-safe. */
1002 typedef struct match_data {
1003 unsigned long int match_call_count; /* As it says */
1004 unsigned long int match_limit; /* As it says */
1005 unsigned long int match_limit_recursion; /* As it says */
1006 int *offset_vector; /* Offset vector */
1007 int offset_end; /* One past the end */
1008 int offset_max; /* The maximum usable for return data */
1009 int nltype; /* Newline type */
1010 int nllen; /* Newline string length */
1011 uschar nl[4]; /* Newline string when fixed */
1012 const uschar *lcc; /* Points to lower casing table */
1013 const uschar *ctypes; /* Points to table of type maps */
1014 BOOL offset_overflow; /* Set if too many extractions */
1015 BOOL notbol; /* NOTBOL flag */
1016 BOOL noteol; /* NOTEOL flag */
1017 BOOL utf8; /* UTF8 flag */
1018 BOOL jscript_compat; /* JAVASCRIPT_COMPAT flag */
1019 BOOL endonly; /* Dollar not before final \n */
1020 BOOL notempty; /* Empty string match not wanted */
1021 BOOL partial; /* PARTIAL flag */
1022 BOOL hitend; /* Hit the end of the subject at some point */
1023 BOOL bsr_anycrlf; /* \R is just any CRLF, not full Unicode */
1024 const uschar *start_code; /* For use when recursing */
1025 USPTR start_subject; /* Start of the subject string */
1026 USPTR end_subject; /* End of the subject string */
1027 USPTR start_match_ptr; /* Start of matched string */
1028 USPTR end_match_ptr; /* Subject position at end match */
1029 int end_offset_top; /* Highwater mark at end of match */
1030 int capture_last; /* Most recent capture number */
1031 int start_offset; /* The start offset value */
1032 eptrblock *eptrchain; /* Chain of eptrblocks for tail recursions */
1033 int eptrn; /* Next free eptrblock */
1034 recursion_info *recursive; /* Linked list of recursion data */
1035 void *callout_data; /* To pass back to callouts */
1038 /* A similar structure is used for the same purpose by the DFA matching
1041 typedef struct dfa_match_data {
1042 const uschar *start_code; /* Start of the compiled pattern */
1043 const uschar *start_subject; /* Start of the subject string */
1044 const uschar *end_subject; /* End of subject string */
1045 const uschar *tables; /* Character tables */
1046 int moptions; /* Match options */
1047 int poptions; /* Pattern options */
1048 int nltype; /* Newline type */
1049 int nllen; /* Newline string length */
1050 uschar nl[4]; /* Newline string when fixed */
1051 void *callout_data; /* To pass back to callouts */
1054 /* Bit definitions for entries in the pcre_ctypes table. */
1056 #define ctype_space 0x01
1057 #define ctype_letter 0x02
1058 #define ctype_digit 0x04
1059 #define ctype_xdigit 0x08
1060 #define ctype_word 0x10 /* alphanumeric or '_' */
1061 #define ctype_meta 0x80 /* regexp meta char or zero (end pattern) */
1063 /* Offsets for the bitmap tables in pcre_cbits. Each table contains a set
1064 of bits for a class map. Some classes are built by combining these tables. */
1066 #define cbit_space 0 /* [:space:] or \s */
1067 #define cbit_xdigit 32 /* [:xdigit:] */
1068 #define cbit_digit 64 /* [:digit:] or \d */
1069 #define cbit_upper 96 /* [:upper:] */
1070 #define cbit_lower 128 /* [:lower:] */
1071 #define cbit_word 160 /* [:word:] or \w */
1072 #define cbit_graph 192 /* [:graph:] */
1073 #define cbit_print 224 /* [:print:] */
1074 #define cbit_punct 256 /* [:punct:] */
1075 #define cbit_cntrl 288 /* [:cntrl:] */
1076 #define cbit_length 320 /* Length of the cbits table */
1078 /* Offsets of the various tables from the base tables pointer, and
1081 #define lcc_offset 0
1082 #define fcc_offset 256
1083 #define cbits_offset 512
1084 #define ctypes_offset (cbits_offset + cbit_length)
1085 #define tables_length (ctypes_offset + 256)
1087 /* Layout of the UCP type table that translates property names into types and
1088 codes. Each entry used to point directly to a name, but to reduce the number of
1089 relocations in shared libraries, it now has an offset into a single string
1093 pcre_uint16 name_offset;
1099 /* Internal shared data tables. These are tables that are used by more than one
1100 of the exported public functions. They have to be "external" in the C sense,
1101 but are not part of the PCRE public API. The data for these tables is in the
1102 pcre_tables.c module. */
1104 extern const int _pcre_utf8_table1[];
1105 extern const int _pcre_utf8_table2[];
1106 extern const int _pcre_utf8_table3[];
1107 extern const uschar _pcre_utf8_table4[];
1109 extern const int _pcre_utf8_table1_size;
1111 extern const char _pcre_utt_names[];
1112 extern const ucp_type_table _pcre_utt[];
1113 extern const int _pcre_utt_size;
1115 extern const uschar _pcre_default_tables[];
1117 extern const uschar _pcre_OP_lengths[];
1120 /* Internal shared functions. These are functions that are used by more than
1121 one of the exported public functions. They have to be "external" in the C
1122 sense, but are not part of the PCRE public API. */
1124 extern BOOL _pcre_is_newline(const uschar *, int, const uschar *,
1126 extern int _pcre_ord2utf8(int, uschar *);
1127 extern real_pcre *_pcre_try_flipped(const real_pcre *, real_pcre *,
1128 const pcre_study_data *, pcre_study_data *);
1129 extern int _pcre_ucp_findprop(const unsigned int, int *, int *);
1130 extern unsigned int _pcre_ucp_othercase(const unsigned int);
1131 extern int _pcre_valid_utf8(const uschar *, int);
1132 extern BOOL _pcre_was_newline(const uschar *, int, const uschar *,
1134 extern BOOL _pcre_xclass(int, const uschar *);
1138 /* End of pcre_internal.h */