Home | History | Annotate | Line # | Download | only in quiz
rxp.c revision 1.1.1.1
      1      1.1  cgd /*-
      2  1.1.1.1  cgd  * Copyright (c) 1991, 1993
      3  1.1.1.1  cgd  *	The Regents of the University of California.  All rights reserved.
      4      1.1  cgd  *
      5      1.1  cgd  * This code is derived from software contributed to Berkeley by
      6  1.1.1.1  cgd  * Jim R. Oldroyd at The Instruction Set and Keith Gabryelski at
      7  1.1.1.1  cgd  * Commodore Business Machines.
      8      1.1  cgd  *
      9      1.1  cgd  * Redistribution and use in source and binary forms, with or without
     10      1.1  cgd  * modification, are permitted provided that the following conditions
     11      1.1  cgd  * are met:
     12      1.1  cgd  * 1. Redistributions of source code must retain the above copyright
     13      1.1  cgd  *    notice, this list of conditions and the following disclaimer.
     14      1.1  cgd  * 2. Redistributions in binary form must reproduce the above copyright
     15      1.1  cgd  *    notice, this list of conditions and the following disclaimer in the
     16      1.1  cgd  *    documentation and/or other materials provided with the distribution.
     17      1.1  cgd  * 3. All advertising materials mentioning features or use of this software
     18      1.1  cgd  *    must display the following acknowledgement:
     19      1.1  cgd  *	This product includes software developed by the University of
     20      1.1  cgd  *	California, Berkeley and its contributors.
     21      1.1  cgd  * 4. Neither the name of the University nor the names of its contributors
     22      1.1  cgd  *    may be used to endorse or promote products derived from this software
     23      1.1  cgd  *    without specific prior written permission.
     24      1.1  cgd  *
     25      1.1  cgd  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     26      1.1  cgd  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     27      1.1  cgd  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     28      1.1  cgd  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     29      1.1  cgd  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     30      1.1  cgd  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     31      1.1  cgd  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     32      1.1  cgd  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     33      1.1  cgd  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     34      1.1  cgd  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     35      1.1  cgd  * SUCH DAMAGE.
     36      1.1  cgd  */
     37      1.1  cgd 
     38      1.1  cgd #ifndef lint
     39  1.1.1.1  cgd static char sccsid[] = "@(#)rxp.c	8.1 (Berkeley) 5/31/93";
     40      1.1  cgd #endif /* not lint */
     41      1.1  cgd 
     42      1.1  cgd /*
     43      1.1  cgd  * regular expression parser
     44      1.1  cgd  *
     45      1.1  cgd  * external functions and return values are:
     46      1.1  cgd  * rxp_compile(s)
     47      1.1  cgd  *	TRUE	success
     48      1.1  cgd  *	FALSE	parse failure; error message will be in char rxperr[]
     49      1.1  cgd  * metas are:
     50      1.1  cgd  *	{...}	optional pattern, equialent to [...|]
     51      1.1  cgd  *	|	alternate pattern
     52      1.1  cgd  *	[...]	pattern delimiters
     53      1.1  cgd  *
     54      1.1  cgd  * rxp_match(s)
     55      1.1  cgd  *	TRUE	string s matches compiled pattern
     56      1.1  cgd  *	FALSE	match failure or regexp error
     57      1.1  cgd  *
     58      1.1  cgd  * rxp_expand()
     59      1.1  cgd  *	char *	reverse-engineered regular expression string
     60      1.1  cgd  *	NULL	regexp error
     61      1.1  cgd  */
     62      1.1  cgd 
     63      1.1  cgd #include <stdio.h>
     64      1.1  cgd #include <ctype.h>
     65      1.1  cgd #include "quiz.h"
     66      1.1  cgd 					/* regexp tokens,	arg */
     67      1.1  cgd #define	LIT	(-1)			/* literal character,	char */
     68      1.1  cgd #define	SOT	(-2)			/* start text anchor,	- */
     69      1.1  cgd #define	EOT	(-3)			/* end text anchor,	- */
     70      1.1  cgd #define	GRP_S	(-4)			/* start alternate grp,	ptr_to_end */
     71      1.1  cgd #define	GRP_E	(-5)			/* end group,		- */
     72      1.1  cgd #define	ALT_S	(-6)			/* alternate starts,	ptr_to_next */
     73      1.1  cgd #define	ALT_E	(-7)			/* alternate ends,	- */
     74      1.1  cgd #define	END	(-8)			/* end of regexp,	- */
     75      1.1  cgd 
     76      1.1  cgd typedef short Rxp_t;			/* type for regexp tokens */
     77      1.1  cgd 
     78      1.1  cgd static Rxp_t rxpbuf[RXP_LINE_SZ];	/* compiled regular expression buffer */
     79      1.1  cgd char rxperr[128];			/* parser error message */
     80      1.1  cgd 
     81  1.1.1.1  cgd static int	 rxp__compile __P((char *, int));
     82  1.1.1.1  cgd static char	*rxp__expand __P((int));
     83  1.1.1.1  cgd static int	 rxp__match __P((char *, int, Rxp_t *, Rxp_t *, char *));
     84      1.1  cgd 
     85      1.1  cgd int
     86      1.1  cgd rxp_compile(s)
     87      1.1  cgd 	register char *	s;
     88      1.1  cgd {
     89      1.1  cgd 	return (rxp__compile(s, TRUE));
     90      1.1  cgd }
     91      1.1  cgd 
     92      1.1  cgd static int
     93      1.1  cgd rxp__compile(s, first)
     94      1.1  cgd 	register char *s;
     95      1.1  cgd 	int first;
     96      1.1  cgd {
     97      1.1  cgd 	static Rxp_t *rp;
     98      1.1  cgd 	static char *sp;
     99      1.1  cgd 	Rxp_t *grp_ptr;
    100      1.1  cgd 	Rxp_t *alt_ptr;
    101      1.1  cgd 	int esc, err;
    102      1.1  cgd 
    103      1.1  cgd 	esc = 0;
    104      1.1  cgd 	if (first) {
    105      1.1  cgd 		rp = rxpbuf;
    106      1.1  cgd 		sp = s;
    107      1.1  cgd 		*rp++ = SOT;	/* auto-anchor: pat is really ^pat$ */
    108      1.1  cgd 		*rp++ = GRP_S;	/* auto-group: ^pat$ is really ^[pat]$ */
    109      1.1  cgd 		*rp++ = 0;
    110      1.1  cgd 	}
    111      1.1  cgd 	*rp++ = ALT_S;
    112      1.1  cgd 	alt_ptr = rp;
    113      1.1  cgd 	*rp++ = 0;
    114      1.1  cgd 	for (; *sp; ++sp) {
    115      1.1  cgd 		if (rp - rxpbuf >= RXP_LINE_SZ - 4) {
    116      1.1  cgd 			(void)snprintf(rxperr, sizeof(rxperr),
    117      1.1  cgd 			    "regular expression too long %s", s);
    118      1.1  cgd 			return (FALSE);
    119      1.1  cgd 		}
    120      1.1  cgd 		if (*sp == ':' && !esc)
    121      1.1  cgd 			break;
    122      1.1  cgd 		if (esc) {
    123      1.1  cgd 			*rp++ = LIT;
    124      1.1  cgd 			*rp++ = *sp;
    125      1.1  cgd 			esc = 0;
    126      1.1  cgd 		}
    127      1.1  cgd 		else switch (*sp) {
    128      1.1  cgd 		case '\\':
    129      1.1  cgd 			esc = 1;
    130      1.1  cgd 			break;
    131      1.1  cgd 		case '{':
    132      1.1  cgd 		case '[':
    133      1.1  cgd 			*rp++ = GRP_S;
    134      1.1  cgd 			grp_ptr = rp;
    135      1.1  cgd 			*rp++ = 0;
    136      1.1  cgd 			sp++;
    137      1.1  cgd 			if ((err = rxp__compile(s, FALSE)) != TRUE)
    138      1.1  cgd 				return (err);
    139      1.1  cgd 			*rp++ = GRP_E;
    140      1.1  cgd 			*grp_ptr = rp - rxpbuf;
    141      1.1  cgd 			break;
    142      1.1  cgd 		case '}':
    143      1.1  cgd 		case ']':
    144      1.1  cgd 		case '|':
    145      1.1  cgd 			*rp++ = ALT_E;
    146      1.1  cgd 			*alt_ptr = rp - rxpbuf;
    147      1.1  cgd 			if (*sp != ']') {
    148      1.1  cgd 				*rp++ = ALT_S;
    149      1.1  cgd 				alt_ptr = rp;
    150      1.1  cgd 				*rp++ = 0;
    151      1.1  cgd 			}
    152      1.1  cgd 			if (*sp != '|') {
    153      1.1  cgd 				if (*sp != ']') {
    154      1.1  cgd 					*rp++ = ALT_E;
    155      1.1  cgd 					*alt_ptr = rp - rxpbuf;
    156      1.1  cgd 				}
    157      1.1  cgd 				if (first) {
    158      1.1  cgd 					(void)snprintf(rxperr, sizeof(rxperr),
    159      1.1  cgd 					    "unmatched alternator in regexp %s",
    160      1.1  cgd 					     s);
    161      1.1  cgd 					return (FALSE);
    162      1.1  cgd 				}
    163      1.1  cgd 				return (TRUE);
    164      1.1  cgd 			}
    165      1.1  cgd 			break;
    166      1.1  cgd 		default:
    167      1.1  cgd 			*rp++ = LIT;
    168      1.1  cgd 			*rp++ = *sp;
    169      1.1  cgd 			esc = 0;
    170      1.1  cgd 			break;
    171      1.1  cgd 		}
    172      1.1  cgd 	}
    173      1.1  cgd 	if (!first) {
    174      1.1  cgd 		(void)snprintf(rxperr, sizeof(rxperr),
    175      1.1  cgd 		    "unmatched alternator in regexp %s", s);
    176      1.1  cgd 		return (FALSE);
    177      1.1  cgd 	}
    178      1.1  cgd 	*rp++ = ALT_E;
    179      1.1  cgd 	*alt_ptr = rp - rxpbuf;
    180      1.1  cgd 	*rp++ = GRP_E;
    181      1.1  cgd 	*(rxpbuf + 2) = rp - rxpbuf;
    182      1.1  cgd 	*rp++ = EOT;
    183      1.1  cgd 	*rp = END;
    184      1.1  cgd 	return (TRUE);
    185      1.1  cgd }
    186      1.1  cgd 
    187      1.1  cgd /*
    188      1.1  cgd  * match string against compiled regular expression
    189      1.1  cgd  */
    190      1.1  cgd int
    191      1.1  cgd rxp_match(s)
    192      1.1  cgd 	register char *	s;
    193      1.1  cgd {
    194      1.1  cgd 	return (rxp__match(s, TRUE, NULL, NULL, NULL));
    195      1.1  cgd }
    196      1.1  cgd 
    197      1.1  cgd static int
    198      1.1  cgd rxp__match(s, first, j_succ, j_fail, sp_fail)
    199      1.1  cgd 	char *s;
    200      1.1  cgd 	int first;
    201      1.1  cgd 	Rxp_t *j_succ;		/* jump here on successful alt match */
    202      1.1  cgd 	Rxp_t *j_fail;		/* jump here on failed match */
    203      1.1  cgd 	char *sp_fail;		/* reset sp to here on failed match */
    204      1.1  cgd {
    205      1.1  cgd 	static Rxp_t *rp;
    206      1.1  cgd 	static char *sp;
    207      1.1  cgd 	register int ch;
    208      1.1  cgd 	Rxp_t *grp_end;
    209      1.1  cgd 	int err;
    210      1.1  cgd 
    211      1.1  cgd 	if (first) {
    212      1.1  cgd 		rp = rxpbuf;
    213      1.1  cgd 		sp = s;
    214      1.1  cgd 	}
    215      1.1  cgd 	while (rp < rxpbuf + RXP_LINE_SZ && *rp != END)
    216      1.1  cgd 		switch(*rp) {
    217      1.1  cgd 		case LIT:
    218      1.1  cgd 			rp++;
    219      1.1  cgd 			ch = isascii(*rp) && isupper(*rp) ? tolower(*rp) : *rp;
    220      1.1  cgd 			if (ch != *sp++) {
    221      1.1  cgd 				rp = j_fail;
    222      1.1  cgd 				sp = sp_fail;
    223      1.1  cgd 				return (TRUE);
    224      1.1  cgd 			}
    225      1.1  cgd 			rp++;
    226      1.1  cgd 			break;
    227      1.1  cgd 		case SOT:
    228      1.1  cgd 			if (sp != s)
    229      1.1  cgd 				return (FALSE);
    230      1.1  cgd 			rp++;
    231      1.1  cgd 			break;
    232      1.1  cgd 		case EOT:
    233      1.1  cgd 			if (*sp != 0)
    234      1.1  cgd 				return (FALSE);
    235      1.1  cgd 			rp++;
    236      1.1  cgd 			break;
    237      1.1  cgd 		case GRP_S:
    238      1.1  cgd 			rp++;
    239      1.1  cgd 			grp_end = rxpbuf + *rp++;
    240      1.1  cgd 			break;
    241      1.1  cgd 		case ALT_S:
    242      1.1  cgd 			rp++;
    243      1.1  cgd 			if ((err = rxp__match(sp,
    244      1.1  cgd 			    FALSE, grp_end, rxpbuf + *rp++, sp)) != TRUE)
    245      1.1  cgd 				return (err);
    246      1.1  cgd 			break;
    247      1.1  cgd 		case ALT_E:
    248      1.1  cgd 			rp = j_succ;
    249      1.1  cgd 			return (TRUE);
    250      1.1  cgd 		case GRP_E:
    251      1.1  cgd 		default:
    252      1.1  cgd 			return (FALSE);
    253      1.1  cgd 		}
    254      1.1  cgd 	return (*rp != END ? FALSE : TRUE);
    255      1.1  cgd }
    256      1.1  cgd 
    257      1.1  cgd /*
    258      1.1  cgd  * Reverse engineer the regular expression, by picking first of all alternates.
    259      1.1  cgd  */
    260      1.1  cgd char *
    261      1.1  cgd rxp_expand()
    262      1.1  cgd {
    263      1.1  cgd 	return (rxp__expand(TRUE));
    264      1.1  cgd }
    265      1.1  cgd 
    266      1.1  cgd static char *
    267      1.1  cgd rxp__expand(first)
    268      1.1  cgd 	int first;
    269      1.1  cgd {
    270      1.1  cgd 	static char buf[RXP_LINE_SZ/2];
    271      1.1  cgd 	static Rxp_t *rp;
    272      1.1  cgd 	static char *bp;
    273      1.1  cgd 	Rxp_t *grp_ptr;
    274      1.1  cgd 	char *err;
    275      1.1  cgd 
    276      1.1  cgd 	if (first) {
    277      1.1  cgd 		rp = rxpbuf;
    278      1.1  cgd 		bp = buf;
    279      1.1  cgd 	}
    280      1.1  cgd 	while (rp < rxpbuf + RXP_LINE_SZ && *rp != END)
    281      1.1  cgd 		switch(*rp) {
    282      1.1  cgd 		case LIT:
    283      1.1  cgd 			rp++;
    284      1.1  cgd 			*bp++ = *rp++;
    285      1.1  cgd 			break;
    286      1.1  cgd 		case GRP_S:
    287      1.1  cgd 			rp++;
    288      1.1  cgd 			grp_ptr = rxpbuf + *rp;
    289      1.1  cgd 			rp++;
    290      1.1  cgd 			if ((err = rxp__expand(FALSE)) == NULL)
    291      1.1  cgd 				return (err);
    292      1.1  cgd 			rp = grp_ptr;
    293      1.1  cgd 			break;
    294      1.1  cgd 		case ALT_E:
    295      1.1  cgd 			return (buf);
    296      1.1  cgd 		case ALT_S:
    297      1.1  cgd 			rp++;
    298      1.1  cgd 			/* FALLTHROUGH */
    299      1.1  cgd 		case SOT:
    300      1.1  cgd 		case EOT:
    301      1.1  cgd 		case GRP_E:
    302      1.1  cgd 			rp++;
    303      1.1  cgd 			break;
    304      1.1  cgd 		default:
    305      1.1  cgd 			return (NULL);
    306      1.1  cgd 		}
    307      1.1  cgd 	if (first) {
    308      1.1  cgd 		if (*rp != END)
    309      1.1  cgd 			return (NULL);
    310      1.1  cgd 		*bp = '\0';
    311      1.1  cgd 	}
    312      1.1  cgd 	return (buf);
    313      1.1  cgd }
    314