Home | History | Annotate | Line # | Download | only in quiz
rxp.c revision 1.2
      1  1.1      cgd /*-
      2  1.1      cgd  * Copyright (c) 1991 The Regents of the University of California.
      3  1.1      cgd  * All rights reserved.
      4  1.1      cgd  *
      5  1.1      cgd  * This code is derived from software contributed to Berkeley by
      6  1.1      cgd  * Jim R. Oldroyd at The Instruction Set.
      7  1.1      cgd  *
      8  1.1      cgd  * Redistribution and use in source and binary forms, with or without
      9  1.1      cgd  * modification, are permitted provided that the following conditions
     10  1.1      cgd  * are met:
     11  1.1      cgd  * 1. Redistributions of source code must retain the above copyright
     12  1.1      cgd  *    notice, this list of conditions and the following disclaimer.
     13  1.1      cgd  * 2. Redistributions in binary form must reproduce the above copyright
     14  1.1      cgd  *    notice, this list of conditions and the following disclaimer in the
     15  1.1      cgd  *    documentation and/or other materials provided with the distribution.
     16  1.1      cgd  * 3. All advertising materials mentioning features or use of this software
     17  1.1      cgd  *    must display the following acknowledgement:
     18  1.1      cgd  *	This product includes software developed by the University of
     19  1.1      cgd  *	California, Berkeley and its contributors.
     20  1.1      cgd  * 4. Neither the name of the University nor the names of its contributors
     21  1.1      cgd  *    may be used to endorse or promote products derived from this software
     22  1.1      cgd  *    without specific prior written permission.
     23  1.1      cgd  *
     24  1.1      cgd  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     25  1.1      cgd  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     26  1.1      cgd  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     27  1.1      cgd  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     28  1.1      cgd  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     29  1.1      cgd  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     30  1.1      cgd  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     31  1.1      cgd  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     32  1.1      cgd  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     33  1.1      cgd  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     34  1.1      cgd  * SUCH DAMAGE.
     35  1.1      cgd  */
     36  1.1      cgd 
     37  1.1      cgd #ifndef lint
     38  1.1      cgd static char sccsid[] = "@(#)rxp.c	5.1 (Berkeley) 11/10/91";
     39  1.1      cgd #endif /* not lint */
     40  1.1      cgd 
     41  1.1      cgd /*
     42  1.1      cgd  * regular expression parser
     43  1.1      cgd  *
     44  1.1      cgd  * external functions and return values are:
     45  1.1      cgd  * rxp_compile(s)
     46  1.1      cgd  *	TRUE	success
     47  1.1      cgd  *	FALSE	parse failure; error message will be in char rxperr[]
     48  1.1      cgd  * metas are:
     49  1.1      cgd  *	{...}	optional pattern, equialent to [...|]
     50  1.1      cgd  *	|	alternate pattern
     51  1.1      cgd  *	[...]	pattern delimiters
     52  1.1      cgd  *
     53  1.1      cgd  * rxp_match(s)
     54  1.1      cgd  *	TRUE	string s matches compiled pattern
     55  1.1      cgd  *	FALSE	match failure or regexp error
     56  1.1      cgd  *
     57  1.1      cgd  * rxp_expand()
     58  1.1      cgd  *	char *	reverse-engineered regular expression string
     59  1.1      cgd  *	NULL	regexp error
     60  1.1      cgd  */
     61  1.1      cgd 
     62  1.1      cgd #include <stdio.h>
     63  1.1      cgd #include <ctype.h>
     64  1.1      cgd #include "quiz.h"
     65  1.1      cgd 					/* regexp tokens,	arg */
     66  1.1      cgd #define	LIT	(-1)			/* literal character,	char */
     67  1.1      cgd #define	SOT	(-2)			/* start text anchor,	- */
     68  1.1      cgd #define	EOT	(-3)			/* end text anchor,	- */
     69  1.1      cgd #define	GRP_S	(-4)			/* start alternate grp,	ptr_to_end */
     70  1.1      cgd #define	GRP_E	(-5)			/* end group,		- */
     71  1.1      cgd #define	ALT_S	(-6)			/* alternate starts,	ptr_to_next */
     72  1.1      cgd #define	ALT_E	(-7)			/* alternate ends,	- */
     73  1.1      cgd #define	END	(-8)			/* end of regexp,	- */
     74  1.1      cgd 
     75  1.1      cgd typedef short Rxp_t;			/* type for regexp tokens */
     76  1.1      cgd 
     77  1.1      cgd static Rxp_t rxpbuf[RXP_LINE_SZ];	/* compiled regular expression buffer */
     78  1.1      cgd char rxperr[128];			/* parser error message */
     79  1.1      cgd 
     80  1.2  mycroft static int	 rxp__compile __P((char *, int));
     81  1.2  mycroft static char	*rxp__expand __P((int));
     82  1.2  mycroft static int	 rxp__match __P((char *, int, Rxp_t *, Rxp_t *, char *));
     83  1.1      cgd 
     84  1.1      cgd int
     85  1.1      cgd rxp_compile(s)
     86  1.1      cgd 	register char *	s;
     87  1.1      cgd {
     88  1.1      cgd 	return (rxp__compile(s, TRUE));
     89  1.1      cgd }
     90  1.1      cgd 
     91  1.1      cgd static int
     92  1.1      cgd rxp__compile(s, first)
     93  1.1      cgd 	register char *s;
     94  1.1      cgd 	int first;
     95  1.1      cgd {
     96  1.1      cgd 	static Rxp_t *rp;
     97  1.1      cgd 	static char *sp;
     98  1.1      cgd 	Rxp_t *grp_ptr;
     99  1.1      cgd 	Rxp_t *alt_ptr;
    100  1.1      cgd 	int esc, err;
    101  1.1      cgd 
    102  1.1      cgd 	esc = 0;
    103  1.1      cgd 	if (first) {
    104  1.1      cgd 		rp = rxpbuf;
    105  1.1      cgd 		sp = s;
    106  1.1      cgd 		*rp++ = SOT;	/* auto-anchor: pat is really ^pat$ */
    107  1.1      cgd 		*rp++ = GRP_S;	/* auto-group: ^pat$ is really ^[pat]$ */
    108  1.1      cgd 		*rp++ = 0;
    109  1.1      cgd 	}
    110  1.1      cgd 	*rp++ = ALT_S;
    111  1.1      cgd 	alt_ptr = rp;
    112  1.1      cgd 	*rp++ = 0;
    113  1.1      cgd 	for (; *sp; ++sp) {
    114  1.1      cgd 		if (rp - rxpbuf >= RXP_LINE_SZ - 4) {
    115  1.1      cgd 			(void)snprintf(rxperr, sizeof(rxperr),
    116  1.1      cgd 			    "regular expression too long %s", s);
    117  1.1      cgd 			return (FALSE);
    118  1.1      cgd 		}
    119  1.1      cgd 		if (*sp == ':' && !esc)
    120  1.1      cgd 			break;
    121  1.1      cgd 		if (esc) {
    122  1.1      cgd 			*rp++ = LIT;
    123  1.1      cgd 			*rp++ = *sp;
    124  1.1      cgd 			esc = 0;
    125  1.1      cgd 		}
    126  1.1      cgd 		else switch (*sp) {
    127  1.1      cgd 		case '\\':
    128  1.1      cgd 			esc = 1;
    129  1.1      cgd 			break;
    130  1.1      cgd 		case '{':
    131  1.1      cgd 		case '[':
    132  1.1      cgd 			*rp++ = GRP_S;
    133  1.1      cgd 			grp_ptr = rp;
    134  1.1      cgd 			*rp++ = 0;
    135  1.1      cgd 			sp++;
    136  1.1      cgd 			if ((err = rxp__compile(s, FALSE)) != TRUE)
    137  1.1      cgd 				return (err);
    138  1.1      cgd 			*rp++ = GRP_E;
    139  1.1      cgd 			*grp_ptr = rp - rxpbuf;
    140  1.1      cgd 			break;
    141  1.1      cgd 		case '}':
    142  1.1      cgd 		case ']':
    143  1.1      cgd 		case '|':
    144  1.1      cgd 			*rp++ = ALT_E;
    145  1.1      cgd 			*alt_ptr = rp - rxpbuf;
    146  1.1      cgd 			if (*sp != ']') {
    147  1.1      cgd 				*rp++ = ALT_S;
    148  1.1      cgd 				alt_ptr = rp;
    149  1.1      cgd 				*rp++ = 0;
    150  1.1      cgd 			}
    151  1.1      cgd 			if (*sp != '|') {
    152  1.1      cgd 				if (*sp != ']') {
    153  1.1      cgd 					*rp++ = ALT_E;
    154  1.1      cgd 					*alt_ptr = rp - rxpbuf;
    155  1.1      cgd 				}
    156  1.1      cgd 				if (first) {
    157  1.1      cgd 					(void)snprintf(rxperr, sizeof(rxperr),
    158  1.1      cgd 					    "unmatched alternator in regexp %s",
    159  1.1      cgd 					     s);
    160  1.1      cgd 					return (FALSE);
    161  1.1      cgd 				}
    162  1.1      cgd 				return (TRUE);
    163  1.1      cgd 			}
    164  1.1      cgd 			break;
    165  1.1      cgd 		default:
    166  1.1      cgd 			*rp++ = LIT;
    167  1.1      cgd 			*rp++ = *sp;
    168  1.1      cgd 			esc = 0;
    169  1.1      cgd 			break;
    170  1.1      cgd 		}
    171  1.1      cgd 	}
    172  1.1      cgd 	if (!first) {
    173  1.1      cgd 		(void)snprintf(rxperr, sizeof(rxperr),
    174  1.1      cgd 		    "unmatched alternator in regexp %s", s);
    175  1.1      cgd 		return (FALSE);
    176  1.1      cgd 	}
    177  1.1      cgd 	*rp++ = ALT_E;
    178  1.1      cgd 	*alt_ptr = rp - rxpbuf;
    179  1.1      cgd 	*rp++ = GRP_E;
    180  1.1      cgd 	*(rxpbuf + 2) = rp - rxpbuf;
    181  1.1      cgd 	*rp++ = EOT;
    182  1.1      cgd 	*rp = END;
    183  1.1      cgd 	return (TRUE);
    184  1.1      cgd }
    185  1.1      cgd 
    186  1.1      cgd /*
    187  1.1      cgd  * match string against compiled regular expression
    188  1.1      cgd  */
    189  1.1      cgd int
    190  1.1      cgd rxp_match(s)
    191  1.1      cgd 	register char *	s;
    192  1.1      cgd {
    193  1.1      cgd 	return (rxp__match(s, TRUE, NULL, NULL, NULL));
    194  1.1      cgd }
    195  1.1      cgd 
    196  1.1      cgd static int
    197  1.1      cgd rxp__match(s, first, j_succ, j_fail, sp_fail)
    198  1.1      cgd 	char *s;
    199  1.1      cgd 	int first;
    200  1.1      cgd 	Rxp_t *j_succ;		/* jump here on successful alt match */
    201  1.1      cgd 	Rxp_t *j_fail;		/* jump here on failed match */
    202  1.1      cgd 	char *sp_fail;		/* reset sp to here on failed match */
    203  1.1      cgd {
    204  1.1      cgd 	static Rxp_t *rp;
    205  1.1      cgd 	static char *sp;
    206  1.1      cgd 	register int ch;
    207  1.1      cgd 	Rxp_t *grp_end;
    208  1.1      cgd 	int err;
    209  1.1      cgd 
    210  1.1      cgd 	if (first) {
    211  1.1      cgd 		rp = rxpbuf;
    212  1.1      cgd 		sp = s;
    213  1.1      cgd 	}
    214  1.1      cgd 	while (rp < rxpbuf + RXP_LINE_SZ && *rp != END)
    215  1.1      cgd 		switch(*rp) {
    216  1.1      cgd 		case LIT:
    217  1.1      cgd 			rp++;
    218  1.1      cgd 			ch = isascii(*rp) && isupper(*rp) ? tolower(*rp) : *rp;
    219  1.1      cgd 			if (ch != *sp++) {
    220  1.1      cgd 				rp = j_fail;
    221  1.1      cgd 				sp = sp_fail;
    222  1.1      cgd 				return (TRUE);
    223  1.1      cgd 			}
    224  1.1      cgd 			rp++;
    225  1.1      cgd 			break;
    226  1.1      cgd 		case SOT:
    227  1.1      cgd 			if (sp != s)
    228  1.1      cgd 				return (FALSE);
    229  1.1      cgd 			rp++;
    230  1.1      cgd 			break;
    231  1.1      cgd 		case EOT:
    232  1.1      cgd 			if (*sp != 0)
    233  1.1      cgd 				return (FALSE);
    234  1.1      cgd 			rp++;
    235  1.1      cgd 			break;
    236  1.1      cgd 		case GRP_S:
    237  1.1      cgd 			rp++;
    238  1.1      cgd 			grp_end = rxpbuf + *rp++;
    239  1.1      cgd 			break;
    240  1.1      cgd 		case ALT_S:
    241  1.1      cgd 			rp++;
    242  1.1      cgd 			if ((err = rxp__match(sp,
    243  1.1      cgd 			    FALSE, grp_end, rxpbuf + *rp++, sp)) != TRUE)
    244  1.1      cgd 				return (err);
    245  1.1      cgd 			break;
    246  1.1      cgd 		case ALT_E:
    247  1.1      cgd 			rp = j_succ;
    248  1.1      cgd 			return (TRUE);
    249  1.1      cgd 		case GRP_E:
    250  1.1      cgd 		default:
    251  1.1      cgd 			return (FALSE);
    252  1.1      cgd 		}
    253  1.1      cgd 	return (*rp != END ? FALSE : TRUE);
    254  1.1      cgd }
    255  1.1      cgd 
    256  1.1      cgd /*
    257  1.1      cgd  * Reverse engineer the regular expression, by picking first of all alternates.
    258  1.1      cgd  */
    259  1.1      cgd char *
    260  1.1      cgd rxp_expand()
    261  1.1      cgd {
    262  1.1      cgd 	return (rxp__expand(TRUE));
    263  1.1      cgd }
    264  1.1      cgd 
    265  1.1      cgd static char *
    266  1.1      cgd rxp__expand(first)
    267  1.1      cgd 	int first;
    268  1.1      cgd {
    269  1.1      cgd 	static char buf[RXP_LINE_SZ/2];
    270  1.1      cgd 	static Rxp_t *rp;
    271  1.1      cgd 	static char *bp;
    272  1.1      cgd 	Rxp_t *grp_ptr;
    273  1.1      cgd 	char *err;
    274  1.1      cgd 
    275  1.1      cgd 	if (first) {
    276  1.1      cgd 		rp = rxpbuf;
    277  1.1      cgd 		bp = buf;
    278  1.1      cgd 	}
    279  1.1      cgd 	while (rp < rxpbuf + RXP_LINE_SZ && *rp != END)
    280  1.1      cgd 		switch(*rp) {
    281  1.1      cgd 		case LIT:
    282  1.1      cgd 			rp++;
    283  1.1      cgd 			*bp++ = *rp++;
    284  1.1      cgd 			break;
    285  1.1      cgd 		case GRP_S:
    286  1.1      cgd 			rp++;
    287  1.1      cgd 			grp_ptr = rxpbuf + *rp;
    288  1.1      cgd 			rp++;
    289  1.1      cgd 			if ((err = rxp__expand(FALSE)) == NULL)
    290  1.1      cgd 				return (err);
    291  1.1      cgd 			rp = grp_ptr;
    292  1.1      cgd 			break;
    293  1.1      cgd 		case ALT_E:
    294  1.1      cgd 			return (buf);
    295  1.1      cgd 		case ALT_S:
    296  1.1      cgd 			rp++;
    297  1.1      cgd 			/* FALLTHROUGH */
    298  1.1      cgd 		case SOT:
    299  1.1      cgd 		case EOT:
    300  1.1      cgd 		case GRP_E:
    301  1.1      cgd 			rp++;
    302  1.1      cgd 			break;
    303  1.1      cgd 		default:
    304  1.1      cgd 			return (NULL);
    305  1.1      cgd 		}
    306  1.1      cgd 	if (first) {
    307  1.1      cgd 		if (*rp != END)
    308  1.1      cgd 			return (NULL);
    309  1.1      cgd 		*bp = '\0';
    310  1.1      cgd 	}
    311  1.1      cgd 	return (buf);
    312  1.1      cgd }
    313