sl@0: #ifndef _REGEX_H_
sl@0: #define	_REGEX_H_	/* never again */
sl@0: /*
sl@0:  * regular expressions
sl@0:  *
sl@0:  * Copyright (c) 1998, 1999 Henry Spencer.  All rights reserved.
sl@0:  * 
sl@0:  * Development of this software was funded, in part, by Cray Research Inc.,
sl@0:  * UUNET Communications Services Inc., Sun Microsystems Inc., and Scriptics
sl@0:  * Corporation, none of whom are responsible for the results.  The author
sl@0:  * thanks all of them. 
sl@0:  * 
sl@0:  * Redistribution and use in source and binary forms -- with or without
sl@0:  * modification -- are permitted for any purpose, provided that
sl@0:  * redistributions in source form retain this entire copyright notice and
sl@0:  * indicate the origin and nature of any modifications.
sl@0:  *
sl@0:  * I'd appreciate being given credit for this package in the documentation
sl@0:  * of software which uses it, but that is not a requirement.
sl@0:  * 
sl@0:  * THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES,
sl@0:  * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY
sl@0:  * AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL
sl@0:  * HENRY SPENCER BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
sl@0:  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
sl@0:  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
sl@0:  * OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
sl@0:  * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
sl@0:  * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
sl@0:  * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
sl@0:  *
sl@0:  *
sl@0:  *
sl@0:  * Prototypes etc. marked with "^" within comments get gathered up (and
sl@0:  * possibly edited) by the regfwd program and inserted near the bottom of
sl@0:  * this file.
sl@0:  *
sl@0:  * We offer the option of declaring one wide-character version of the
sl@0:  * RE functions as well as the char versions.  To do that, define
sl@0:  * __REG_WIDE_T to the type of wide characters (unfortunately, there
sl@0:  * is no consensus that wchar_t is suitable) and __REG_WIDE_COMPILE and
sl@0:  * __REG_WIDE_EXEC to the names to be used for the compile and execute
sl@0:  * functions (suggestion:  re_Xcomp and re_Xexec, where X is a letter
sl@0:  * suggestive of the wide type, e.g. re_ucomp and re_uexec for Unicode).
sl@0:  * For cranky old compilers, it may be necessary to do something like:
sl@0:  * #define	__REG_WIDE_COMPILE(a,b,c,d)	re_Xcomp(a,b,c,d)
sl@0:  * #define	__REG_WIDE_EXEC(a,b,c,d,e,f,g)	re_Xexec(a,b,c,d,e,f,g)
sl@0:  * rather than just #defining the names as parameterless macros.
sl@0:  *
sl@0:  * For some specialized purposes, it may be desirable to suppress the
sl@0:  * declarations of the "front end" functions, regcomp() and regexec(),
sl@0:  * or of the char versions of the compile and execute functions.  To
sl@0:  * suppress the front-end functions, define __REG_NOFRONT.  To suppress
sl@0:  * the char versions, define __REG_NOCHAR.
sl@0:  *
sl@0:  * The right place to do those defines (and some others you may want, see
sl@0:  * below) would be <sys/types.h>.  If you don't have control of that file,
sl@0:  * the right place to add your own defines to this file is marked below.
sl@0:  * This is normally done automatically, by the makefile and regmkhdr, based
sl@0:  * on the contents of regcustom.h.
sl@0:  */
sl@0: 
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * voodoo for C++
sl@0:  */
sl@0: #ifdef __cplusplus
sl@0: extern "C" {
sl@0: #endif
sl@0: 
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * Add your own defines, if needed, here.
sl@0:  */
sl@0: 
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * Location where a chunk of regcustom.h is automatically spliced into
sl@0:  * this file (working from its prototype, regproto.h).
sl@0:  */
sl@0: /* --- begin --- */
sl@0: /* ensure certain things don't sneak in from system headers */
sl@0: #ifdef __REG_WIDE_T
sl@0: #undef __REG_WIDE_T
sl@0: #endif
sl@0: #ifdef __REG_WIDE_COMPILE
sl@0: #undef __REG_WIDE_COMPILE
sl@0: #endif
sl@0: #ifdef __REG_WIDE_EXEC
sl@0: #undef __REG_WIDE_EXEC
sl@0: #endif
sl@0: #ifdef __REG_REGOFF_T
sl@0: #undef __REG_REGOFF_T
sl@0: #endif
sl@0: #ifdef __REG_VOID_T
sl@0: #undef __REG_VOID_T
sl@0: #endif
sl@0: #ifdef __REG_CONST
sl@0: #undef __REG_CONST
sl@0: #endif
sl@0: #ifdef __REG_NOFRONT
sl@0: #undef __REG_NOFRONT
sl@0: #endif
sl@0: #ifdef __REG_NOCHAR
sl@0: #undef __REG_NOCHAR
sl@0: #endif
sl@0: /* interface types */
sl@0: #define	__REG_WIDE_T	Tcl_UniChar
sl@0: #define	__REG_REGOFF_T	long	/* not really right, but good enough... */
sl@0: #define	__REG_VOID_T	VOID
sl@0: #define	__REG_CONST	CONST
sl@0: /* names and declarations */
sl@0: #define	__REG_WIDE_COMPILE	TclReComp
sl@0: #define	__REG_WIDE_EXEC		TclReExec
sl@0: #define	__REG_NOFRONT		/* don't want regcomp() and regexec() */
sl@0: #define	__REG_NOCHAR		/* or the char versions */
sl@0: #define	regfree		TclReFree
sl@0: #define	regerror	TclReError
sl@0: /* --- end --- */
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * interface types etc.
sl@0:  */
sl@0: 
sl@0: /*
sl@0:  * regoff_t has to be large enough to hold either off_t or ssize_t,
sl@0:  * and must be signed; it's only a guess that long is suitable, so we
sl@0:  * offer <sys/types.h> an override.
sl@0:  */
sl@0: #ifdef __REG_REGOFF_T
sl@0: typedef __REG_REGOFF_T regoff_t;
sl@0: #else
sl@0: typedef long regoff_t;
sl@0: #endif
sl@0: 
sl@0: /*
sl@0:  * For benefit of old compilers, we offer <sys/types.h> the option of
sl@0:  * overriding the `void' type used to declare nonexistent return types.
sl@0:  */
sl@0: #ifdef __REG_VOID_T
sl@0: typedef __REG_VOID_T re_void;
sl@0: #else
sl@0: typedef void re_void;
sl@0: #endif
sl@0: 
sl@0: /*
sl@0:  * Also for benefit of old compilers, <sys/types.h> can supply a macro
sl@0:  * which expands to a substitute for `const'.
sl@0:  */
sl@0: #ifndef __REG_CONST
sl@0: #define	__REG_CONST	const
sl@0: #endif
sl@0: 
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * other interface types
sl@0:  */
sl@0: 
sl@0: /* the biggie, a compiled RE (or rather, a front end to same) */
sl@0: typedef struct {
sl@0: 	int re_magic;		/* magic number */
sl@0: 	size_t re_nsub;		/* number of subexpressions */
sl@0: 	long re_info;		/* information about RE */
sl@0: #		define	REG_UBACKREF		000001
sl@0: #		define	REG_ULOOKAHEAD		000002
sl@0: #		define	REG_UBOUNDS		000004
sl@0: #		define	REG_UBRACES		000010
sl@0: #		define	REG_UBSALNUM		000020
sl@0: #		define	REG_UPBOTCH		000040
sl@0: #		define	REG_UBBS		000100
sl@0: #		define	REG_UNONPOSIX		000200
sl@0: #		define	REG_UUNSPEC		000400
sl@0: #		define	REG_UUNPORT		001000
sl@0: #		define	REG_ULOCALE		002000
sl@0: #		define	REG_UEMPTYMATCH		004000
sl@0: #		define	REG_UIMPOSSIBLE		010000
sl@0: #		define	REG_USHORTEST		020000
sl@0: 	int re_csize;		/* sizeof(character) */
sl@0: 	char *re_endp;		/* backward compatibility kludge */
sl@0: 	/* the rest is opaque pointers to hidden innards */
sl@0: 	char *re_guts;		/* `char *' is more portable than `void *' */
sl@0: 	char *re_fns;
sl@0: } regex_t;
sl@0: 
sl@0: /* result reporting (may acquire more fields later) */
sl@0: typedef struct {
sl@0: 	regoff_t rm_so;		/* start of substring */
sl@0: 	regoff_t rm_eo;		/* end of substring */
sl@0: } regmatch_t;
sl@0: 
sl@0: /* supplementary control and reporting */
sl@0: typedef struct {
sl@0: 	regmatch_t rm_extend;	/* see REG_EXPECT */
sl@0: } rm_detail_t;
sl@0: 
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * compilation
sl@0:  ^ #ifndef __REG_NOCHAR
sl@0:  ^ int re_comp(regex_t *, __REG_CONST char *, size_t, int);
sl@0:  ^ #endif
sl@0:  ^ #ifndef __REG_NOFRONT
sl@0:  ^ int regcomp(regex_t *, __REG_CONST char *, int);
sl@0:  ^ #endif
sl@0:  ^ #ifdef __REG_WIDE_T
sl@0:  ^ int __REG_WIDE_COMPILE(regex_t *, __REG_CONST __REG_WIDE_T *, size_t, int);
sl@0:  ^ #endif
sl@0:  */
sl@0: #define	REG_BASIC	000000	/* BREs (convenience) */
sl@0: #define	REG_EXTENDED	000001	/* EREs */
sl@0: #define	REG_ADVF	000002	/* advanced features in EREs */
sl@0: #define	REG_ADVANCED	000003	/* AREs (which are also EREs) */
sl@0: #define	REG_QUOTE	000004	/* no special characters, none */
sl@0: #define	REG_NOSPEC	REG_QUOTE	/* historical synonym */
sl@0: #define	REG_ICASE	000010	/* ignore case */
sl@0: #define	REG_NOSUB	000020	/* don't care about subexpressions */
sl@0: #define	REG_EXPANDED	000040	/* expanded format, white space & comments */
sl@0: #define	REG_NLSTOP	000100	/* \n doesn't match . or [^ ] */
sl@0: #define	REG_NLANCH	000200	/* ^ matches after \n, $ before */
sl@0: #define	REG_NEWLINE	000300	/* newlines are line terminators */
sl@0: #define	REG_PEND	000400	/* ugh -- backward-compatibility hack */
sl@0: #define	REG_EXPECT	001000	/* report details on partial/limited matches */
sl@0: #define	REG_BOSONLY	002000	/* temporary kludge for BOS-only matches */
sl@0: #define	REG_DUMP	004000	/* none of your business :-) */
sl@0: #define	REG_FAKE	010000	/* none of your business :-) */
sl@0: #define	REG_PROGRESS	020000	/* none of your business :-) */
sl@0: 
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * execution
sl@0:  ^ #ifndef __REG_NOCHAR
sl@0:  ^ int re_exec(regex_t *, __REG_CONST char *, size_t,
sl@0:  ^				rm_detail_t *, size_t, regmatch_t [], int);
sl@0:  ^ #endif
sl@0:  ^ #ifndef __REG_NOFRONT
sl@0:  ^ int regexec(regex_t *, __REG_CONST char *, size_t, regmatch_t [], int);
sl@0:  ^ #endif
sl@0:  ^ #ifdef __REG_WIDE_T
sl@0:  ^ int __REG_WIDE_EXEC(regex_t *, __REG_CONST __REG_WIDE_T *, size_t,
sl@0:  ^				rm_detail_t *, size_t, regmatch_t [], int);
sl@0:  ^ #endif
sl@0:  */
sl@0: #define	REG_NOTBOL	0001	/* BOS is not BOL */
sl@0: #define	REG_NOTEOL	0002	/* EOS is not EOL */
sl@0: #define	REG_STARTEND	0004	/* backward compatibility kludge */
sl@0: #define	REG_FTRACE	0010	/* none of your business */
sl@0: #define	REG_MTRACE	0020	/* none of your business */
sl@0: #define	REG_SMALL	0040	/* none of your business */
sl@0: 
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * misc generics (may be more functions here eventually)
sl@0:  ^ re_void regfree(regex_t *);
sl@0:  */
sl@0: 
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * error reporting
sl@0:  * Be careful if modifying the list of error codes -- the table used by
sl@0:  * regerror() is generated automatically from this file!
sl@0:  *
sl@0:  * Note that there is no wide-char variant of regerror at this time; what
sl@0:  * kind of character is used for error reports is independent of what kind
sl@0:  * is used in matching.
sl@0:  *
sl@0:  ^ extern size_t regerror(int, __REG_CONST regex_t *, char *, size_t);
sl@0:  */
sl@0: #define	REG_OKAY	 0	/* no errors detected */
sl@0: #define	REG_NOMATCH	 1	/* failed to match */
sl@0: #define	REG_BADPAT	 2	/* invalid regexp */
sl@0: #define	REG_ECOLLATE	 3	/* invalid collating element */
sl@0: #define	REG_ECTYPE	 4	/* invalid character class */
sl@0: #define	REG_EESCAPE	 5	/* invalid escape \ sequence */
sl@0: #define	REG_ESUBREG	 6	/* invalid backreference number */
sl@0: #define	REG_EBRACK	 7	/* brackets [] not balanced */
sl@0: #define	REG_EPAREN	 8	/* parentheses () not balanced */
sl@0: #define	REG_EBRACE	 9	/* braces {} not balanced */
sl@0: #define	REG_BADBR	10	/* invalid repetition count(s) */
sl@0: #define	REG_ERANGE	11	/* invalid character range */
sl@0: #define	REG_ESPACE	12	/* out of memory */
sl@0: #define	REG_BADRPT	13	/* quantifier operand invalid */
sl@0: #define	REG_ASSERT	15	/* "can't happen" -- you found a bug */
sl@0: #define	REG_INVARG	16	/* invalid argument to regex function */
sl@0: #define	REG_MIXED	17	/* character widths of regex and string differ */
sl@0: #define	REG_BADOPT	18	/* invalid embedded option */
sl@0: /* two specials for debugging and testing */
sl@0: #define	REG_ATOI	101	/* convert error-code name to number */
sl@0: #define	REG_ITOA	102	/* convert error-code number to name */
sl@0: 
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * the prototypes, as possibly munched by regfwd
sl@0:  */
sl@0: /* =====^!^===== begin forwards =====^!^===== */
sl@0: /* automatically gathered by fwd; do not hand-edit */
sl@0: /* === regproto.h === */
sl@0: #ifndef __REG_NOCHAR
sl@0: int re_comp _ANSI_ARGS_((regex_t *, __REG_CONST char *, size_t, int));
sl@0: #endif
sl@0: #ifndef __REG_NOFRONT
sl@0: int regcomp _ANSI_ARGS_((regex_t *, __REG_CONST char *, int));
sl@0: #endif
sl@0: #ifdef __REG_WIDE_T
sl@0: int __REG_WIDE_COMPILE _ANSI_ARGS_((regex_t *, __REG_CONST __REG_WIDE_T *, size_t, int));
sl@0: #endif
sl@0: #ifndef __REG_NOCHAR
sl@0: int re_exec _ANSI_ARGS_((regex_t *, __REG_CONST char *, size_t, rm_detail_t *, size_t, regmatch_t [], int));
sl@0: #endif
sl@0: #ifndef __REG_NOFRONT
sl@0: int regexec _ANSI_ARGS_((regex_t *, __REG_CONST char *, size_t, regmatch_t [], int));
sl@0: #endif
sl@0: #ifdef __REG_WIDE_T
sl@0: int __REG_WIDE_EXEC _ANSI_ARGS_((regex_t *, __REG_CONST __REG_WIDE_T *, size_t, rm_detail_t *, size_t, regmatch_t [], int));
sl@0: #endif
sl@0: re_void regfree _ANSI_ARGS_((regex_t *));
sl@0: extern size_t regerror _ANSI_ARGS_((int, __REG_CONST regex_t *, char *, size_t));
sl@0: /* automatically gathered by fwd; do not hand-edit */
sl@0: /* =====^!^===== end forwards =====^!^===== */
sl@0: 
sl@0: 
sl@0: 
sl@0: /*
sl@0:  * more C++ voodoo
sl@0:  */
sl@0: #ifdef __cplusplus
sl@0: }
sl@0: #endif
sl@0: 
sl@0: 
sl@0: 
sl@0: #endif