148267Sbostic /*-
2*62029Sbostic  * Copyright (c) 1980, 1993
3*62029Sbostic  *	The Regents of the University of California.  All rights reserved.
448267Sbostic  *
548267Sbostic  * %sccs.include.proprietary.c%
621559Sdist  */
721559Sdist 
814516Ssam #ifndef lint
9*62029Sbostic static char copyright[] =
10*62029Sbostic "@(#) Copyright (c) 1980, 1993\n\
11*62029Sbostic 	The Regents of the University of California.  All rights reserved.\n";
1248267Sbostic #endif /* not lint */
136639Smckusick 
1421559Sdist #ifndef lint
15*62029Sbostic static char sccsid[] = "@(#)old.ucb.grep.c	8.1 (Berkeley) 06/06/93";
1648267Sbostic #endif /* not lint */
1721559Sdist 
186639Smckusick #include <stdio.h>
1921391Smckusick #include <sys/types.h>
2021391Smckusick #include <sys/stat.h>
216639Smckusick /*
226639Smckusick  * grep -- print lines matching (or not matching) a pattern
236639Smckusick  */
246639Smckusick 
2521391Smckusick #define BLKSIZE 8192
266639Smckusick #define	CCHR	2
276639Smckusick #define	CDOT	4
286639Smckusick #define	CCL	6
296639Smckusick #define	NCCL	8
306639Smckusick #define	CDOL	10
316639Smckusick #define	CEOF	11
326639Smckusick 
336639Smckusick #define	CBRC	14
346639Smckusick #define	CLET	15
356639Smckusick #define	STAR	01
366639Smckusick 
376639Smckusick #define	ESIZE	256
386639Smckusick 
396639Smckusick char	expbuf[ESIZE];
406639Smckusick long	lnum;
4121391Smckusick char	linebuf[BUFSIZ+1];
426639Smckusick int	bflag;
436639Smckusick int	nflag;
446639Smckusick int	cflag;
456639Smckusick int	vflag;
466639Smckusick int	nfile;
4732363Sbostic int	hflag;
4832365Sbostic int	oflag;
496639Smckusick int	iflag;
506639Smckusick int	lflag;
516639Smckusick int	wflag;
526639Smckusick int	sflag;
536639Smckusick int	nsucc;
546639Smckusick int	circf;
556639Smckusick int	blkno;
566639Smckusick long	tln;
5726447Slepreau int	retcode = 0;
586639Smckusick 
main(argc,argv)596639Smckusick main(argc, argv)
606639Smckusick char **argv;
616639Smckusick {
626639Smckusick 
636639Smckusick 	while (--argc > 0 && (++argv)[0][0]=='-') {
646639Smckusick 		char *cp = argv[0] + 1;
656639Smckusick 		while (*cp) switch (*cp++) {
666639Smckusick 
676639Smckusick 		case 'v':
686639Smckusick 			vflag++;
696639Smckusick 			continue;
706639Smckusick 
716639Smckusick 		case 'b':
726639Smckusick 			bflag++;
736639Smckusick 			continue;
746639Smckusick 
7532365Sbostic 		case 'o':
7632365Sbostic 			oflag++;
7732365Sbostic 			continue;
7832365Sbostic 
7932363Sbostic 		case 'h':
8032363Sbostic 			hflag++;
8132363Sbostic 			continue;
8232363Sbostic 
836639Smckusick 		case 'i':
846639Smckusick 		case 'y':	/* -y for compatibility with btl grep */
856639Smckusick 			iflag++;
866639Smckusick 			continue;
876639Smckusick 
886639Smckusick 		case 'l':
896639Smckusick 			lflag++;
906639Smckusick 		case 'c':
916639Smckusick 			cflag++;
926639Smckusick 			continue;
936639Smckusick 
946639Smckusick 		case 'w':
956639Smckusick 			wflag++;
966639Smckusick 			continue;
976639Smckusick 
986639Smckusick 		case 's':
996639Smckusick 			sflag++;
1006639Smckusick 			continue;
1016639Smckusick 
1026639Smckusick 		case 'n':
1036639Smckusick 			nflag++;
1046639Smckusick 			continue;
1056639Smckusick 
1066639Smckusick 		case 'e':
1076639Smckusick 			--argc;
1086639Smckusick 			++argv;
1096639Smckusick 			goto out;
1106639Smckusick 
1116639Smckusick 		default:
11226447Slepreau 			fprintf(stderr, "grep: unknown flag\n");
1136639Smckusick 			continue;
1146639Smckusick 		}
1156639Smckusick 	}
1166639Smckusick out:
1176639Smckusick 	if (argc<=0)
1186639Smckusick 		exit(2);
1196639Smckusick 	compile(*argv);
1206639Smckusick 	nfile = --argc;
1216639Smckusick 	if (argc<=0) {
1226639Smckusick 		if (lflag)
1236639Smckusick 			exit(1);
1246639Smckusick 		execute(0);
1256639Smckusick 	}
1266639Smckusick 	else while (--argc >= 0) {
1276639Smckusick 		argv++;
1286639Smckusick 		execute(*argv);
1296639Smckusick 	}
13026447Slepreau 	exit(retcode != 0 ? retcode : nsucc == 0);
1316639Smckusick }
1326639Smckusick 
compile(astr)1336639Smckusick compile(astr)
1346639Smckusick char *astr;
1356639Smckusick {
1366639Smckusick 	register c;
1376639Smckusick 	register char *ep, *sp;
1386639Smckusick 	char *lastep;
1396639Smckusick 	int cclcnt;
1406639Smckusick 
1416639Smckusick 	ep = expbuf;
1426639Smckusick 	sp = astr;
1436639Smckusick 	if (*sp == '^') {
1446639Smckusick 		circf++;
1456639Smckusick 		sp++;
1466639Smckusick 	}
1476639Smckusick 	if (wflag)
1486639Smckusick 		*ep++ = CBRC;
1496639Smckusick 	for (;;) {
1506639Smckusick 		if (ep >= &expbuf[ESIZE])
1516639Smckusick 			goto cerror;
1526639Smckusick 		if ((c = *sp++) != '*')
1536639Smckusick 			lastep = ep;
1546639Smckusick 		switch (c) {
1556639Smckusick 
1566639Smckusick 		case '\0':
1576639Smckusick 			if (wflag)
1586639Smckusick 				*ep++ = CLET;
1596639Smckusick 			*ep++ = CEOF;
1606639Smckusick 			return;
1616639Smckusick 
1626639Smckusick 		case '.':
1636639Smckusick 			*ep++ = CDOT;
1646639Smckusick 			continue;
1656639Smckusick 
1666639Smckusick 		case '*':
1676639Smckusick 			if (lastep==0)
1686639Smckusick 				goto defchar;
1696639Smckusick 			*lastep |= STAR;
1706639Smckusick 			continue;
1716639Smckusick 
1726639Smckusick 		case '$':
1736639Smckusick 			if (*sp != '\0')
1746639Smckusick 				goto defchar;
1756639Smckusick 			*ep++ = CDOL;
1766639Smckusick 			continue;
1776639Smckusick 
1786639Smckusick 		case '[':
1796639Smckusick 			*ep++ = CCL;
1806639Smckusick 			*ep++ = 0;
1816639Smckusick 			cclcnt = 1;
1826639Smckusick 			if ((c = *sp++) == '^') {
1836639Smckusick 				c = *sp++;
1846639Smckusick 				ep[-2] = NCCL;
1856639Smckusick 			}
1866639Smckusick 			do {
1876639Smckusick 				*ep++ = c;
1886639Smckusick 				cclcnt++;
1896639Smckusick 				if (c=='\0' || ep >= &expbuf[ESIZE])
1906639Smckusick 					goto cerror;
1916639Smckusick 			} while ((c = *sp++) != ']');
1926639Smckusick 			lastep[1] = cclcnt;
1936639Smckusick 			continue;
1946639Smckusick 
1956639Smckusick 		case '\\':
1966639Smckusick 			if ((c = *sp++) == '\0')
1976639Smckusick 				goto cerror;
1986639Smckusick 			if (c == '<') {
1996639Smckusick 				*ep++ = CBRC;
2006639Smckusick 				continue;
2016639Smckusick 			}
2026639Smckusick 			if (c == '>') {
2036639Smckusick 				*ep++ = CLET;
2046639Smckusick 				continue;
2056639Smckusick 			}
2066639Smckusick 		defchar:
2076639Smckusick 		default:
2086639Smckusick 			*ep++ = CCHR;
2096639Smckusick 			*ep++ = c;
2106639Smckusick 		}
2116639Smckusick 	}
2126639Smckusick     cerror:
21326447Slepreau 	fprintf(stderr, "grep: RE error\n");
21426447Slepreau 	exit(2);
2156639Smckusick }
2166639Smckusick 
same(a,b)2176639Smckusick same(a, b)
2186639Smckusick 	register int a, b;
2196639Smckusick {
2206639Smckusick 
2216639Smckusick 	return (a == b || iflag && (a ^ b) == ' ' && letter(a) == letter(b));
2226639Smckusick }
2236639Smckusick 
letter(c)2246639Smckusick letter(c)
2256639Smckusick 	register int c;
2266639Smckusick {
2276639Smckusick 
2286639Smckusick 	if (c >= 'a' && c <= 'z')
2296639Smckusick 		return (c);
2306639Smckusick 	if (c >= 'A' && c <= 'Z')
2316639Smckusick 		return (c + 'a' - 'A');
2326639Smckusick 	return (0);
2336639Smckusick }
2346639Smckusick 
execute(file)2356639Smckusick execute(file)
23646829Sbostic 	char *file;
2376639Smckusick {
2386639Smckusick 	register char *p1, *p2;
2396639Smckusick 	register c;
2406639Smckusick 	int f;
2416639Smckusick 	char *ebp, *cbp;
24221391Smckusick 	static char *buf;
24321391Smckusick 	static int blksize;
24421391Smckusick 	struct stat stb;
2456639Smckusick 
2466639Smckusick 	if (file) {
2476639Smckusick 		if ((f = open(file, 0)) < 0) {
24812090Smckusick 			perror(file);
24926447Slepreau 			retcode = 2;
2506639Smckusick 		}
2516639Smckusick 	} else
2526639Smckusick 		f = 0;
25321391Smckusick 	if (buf == NULL) {
25421391Smckusick 		if (fstat(f, &stb) > 0 && stb.st_blksize > 0)
25521391Smckusick 			blksize = stb.st_blksize;
25621391Smckusick 		else
25721391Smckusick 			blksize = BLKSIZE;
25821391Smckusick 		buf = (char *)malloc(blksize);
25921391Smckusick 		if (buf == NULL) {
26026447Slepreau 			fprintf(stderr, "grep: no memory for %s\n", file);
26126447Slepreau 			retcode = 2;
26221391Smckusick 			return;
26321391Smckusick 		}
26421391Smckusick 	}
26521391Smckusick 	ebp = buf;
26621391Smckusick 	cbp = buf;
2676639Smckusick 	lnum = 0;
2686639Smckusick 	tln = 0;
2696639Smckusick 	blkno = -1;
2706639Smckusick 	for (;;) {
2716639Smckusick 		lnum++;
2726639Smckusick 		if((lnum&0377) == 0)
2736639Smckusick 			fflush(stdout);
2746639Smckusick 		p1 = linebuf;
2756639Smckusick 		p2 = cbp;
2766639Smckusick 		for (;;) {
2776639Smckusick 			if (p2 >= ebp) {
27821391Smckusick 				if ((c = read(f, buf, blksize)) <= 0) {
2796639Smckusick 					close(f);
2806639Smckusick 					if (cflag) {
2816639Smckusick 						if (lflag) {
2826639Smckusick 							if (tln)
2836639Smckusick 							printf("%s\n", file);
2846639Smckusick 						} else {
2856639Smckusick 							if (nfile > 1)
2866639Smckusick 								printf("%s:", file);
2876639Smckusick 							printf("%ld\n", tln);
2886639Smckusick 						}
2896639Smckusick 					}
2906639Smckusick 					return;
2916639Smckusick 				}
2926639Smckusick 				blkno++;
29321391Smckusick 				p2 = buf;
29421391Smckusick 				ebp = buf+c;
2956639Smckusick 			}
2966639Smckusick 			if ((c = *p2++) == '\n')
2976639Smckusick 				break;
2986639Smckusick 			if(c)
29921391Smckusick 			if (p1 < &linebuf[BUFSIZ-1])
3006639Smckusick 				*p1++ = c;
3016639Smckusick 		}
3026639Smckusick 		*p1++ = 0;
3036639Smckusick 		cbp = p2;
3046639Smckusick 		p1 = linebuf;
3056639Smckusick 		p2 = expbuf;
3066639Smckusick 		if (circf) {
3076639Smckusick 			if (advance(p1, p2))
3086639Smckusick 				goto found;
3096639Smckusick 			goto nfound;
3106639Smckusick 		}
3116639Smckusick 		/* fast check for first character */
3126639Smckusick 		if (*p2==CCHR) {
3136639Smckusick 			c = p2[1];
3146639Smckusick 			do {
3156639Smckusick 				if (*p1!=c && (!iflag || (c ^ *p1) != ' '
3166639Smckusick 					|| letter(c) != letter(*p1)))
3176639Smckusick 					continue;
3186639Smckusick 				if (advance(p1, p2))
3196639Smckusick 					goto found;
3206639Smckusick 			} while (*p1++);
3216639Smckusick 			goto nfound;
3226639Smckusick 		}
3236639Smckusick 		/* regular algorithm */
3246639Smckusick 		do {
3256639Smckusick 			if (advance(p1, p2))
3266639Smckusick 				goto found;
3276639Smckusick 		} while (*p1++);
3286639Smckusick 	nfound:
3296639Smckusick 		if (vflag)
3306639Smckusick 			succeed(file);
3316639Smckusick 		continue;
3326639Smckusick 	found:
3336639Smckusick 		if (vflag==0)
3346639Smckusick 			succeed(file);
3356639Smckusick 	}
3366639Smckusick }
3376639Smckusick 
advance(alp,aep)3386639Smckusick advance(alp, aep)
3396639Smckusick 	char *alp, *aep;
3406639Smckusick {
3416639Smckusick 	register char *lp, *ep, *curlp;
3426639Smckusick 	char *nextep;
3436639Smckusick 
3446639Smckusick 	lp = alp;
3456639Smckusick 	ep = aep;
3466639Smckusick 	for (;;) switch (*ep++) {
3476639Smckusick 
3486639Smckusick 	case CCHR:
3496639Smckusick 		if (!same(*ep, *lp))
3506639Smckusick 			return (0);
3516639Smckusick 		ep++, lp++;
3526639Smckusick 		continue;
3536639Smckusick 
3546639Smckusick 	case CDOT:
3556639Smckusick 		if (*lp++)
3566639Smckusick 			continue;
3576639Smckusick 		return(0);
3586639Smckusick 
3596639Smckusick 	case CDOL:
3606639Smckusick 		if (*lp==0)
3616639Smckusick 			continue;
3626639Smckusick 		return(0);
3636639Smckusick 
3646639Smckusick 	case CEOF:
3656639Smckusick 		return(1);
3666639Smckusick 
3676639Smckusick 	case CCL:
3686639Smckusick 		if (cclass(ep, *lp++, 1)) {
3696639Smckusick 			ep += *ep;
3706639Smckusick 			continue;
3716639Smckusick 		}
3726639Smckusick 		return(0);
3736639Smckusick 
3746639Smckusick 	case NCCL:
3756639Smckusick 		if (cclass(ep, *lp++, 0)) {
3766639Smckusick 			ep += *ep;
3776639Smckusick 			continue;
3786639Smckusick 		}
3796639Smckusick 		return(0);
3806639Smckusick 
3816639Smckusick 	case CDOT|STAR:
3826639Smckusick 		curlp = lp;
3836639Smckusick 		while (*lp++);
3846639Smckusick 		goto star;
3856639Smckusick 
3866639Smckusick 	case CCHR|STAR:
3876639Smckusick 		curlp = lp;
3886639Smckusick 		while (same(*lp, *ep))
3896639Smckusick 			lp++;
3906639Smckusick 		lp++;
3916639Smckusick 		ep++;
3926639Smckusick 		goto star;
3936639Smckusick 
3946639Smckusick 	case CCL|STAR:
3956639Smckusick 	case NCCL|STAR:
3966639Smckusick 		curlp = lp;
3976639Smckusick 		while (cclass(ep, *lp++, ep[-1]==(CCL|STAR)));
3986639Smckusick 		ep += *ep;
3996639Smckusick 		goto star;
4006639Smckusick 
4016639Smckusick 	star:
4026639Smckusick 		do {
4036639Smckusick 			lp--;
4046639Smckusick 			if (advance(lp, ep))
4056639Smckusick 				return(1);
4066639Smckusick 		} while (lp > curlp);
4076639Smckusick 		return(0);
4086639Smckusick 
4096639Smckusick 	case CBRC:
4106639Smckusick 		if (lp == expbuf)
4116639Smckusick 			continue;
4126639Smckusick #define	uletter(c)	(letter(c) || c == '_')
4136639Smckusick 		if ( ( uletter(*lp) || digit ( * lp ) )  && !uletter(lp[-1]) && !digit(lp[-1]))
4146639Smckusick 			continue;
4156639Smckusick 		return (0);
4166639Smckusick 
4176639Smckusick 	case CLET:
4186639Smckusick 		if (!uletter(*lp) && !digit(*lp))
4196639Smckusick 			continue;
4206639Smckusick 		return (0);
4216639Smckusick 
4226639Smckusick 	default:
42326447Slepreau 		fprintf(stderr, "grep: RE botch\n");
42426447Slepreau 		exit(2);
4256639Smckusick 	}
4266639Smckusick }
4276639Smckusick 
cclass(aset,ac,af)4286639Smckusick cclass(aset, ac, af)
4296639Smckusick 	char *aset;
4306639Smckusick {
4316639Smckusick 	register char *set, c;
4326639Smckusick 	register n;
4336639Smckusick 
4346639Smckusick 	set = aset;
4356639Smckusick 	if ((c = ac) == 0)
4366639Smckusick 		return(0);
4376639Smckusick 	n = *set++;
4386639Smckusick 	while (--n)
4396639Smckusick 		if (n > 2 && set[1] == '-') {
4406639Smckusick 			if (c >= (set[0] & 0177) && c <= (set[2] & 0177))
4416639Smckusick 				return (af);
4426639Smckusick 			set += 3;
4436639Smckusick 			n -= 2;
4446639Smckusick 		} else
4456639Smckusick 			if ((*set++ & 0177) == c)
4466639Smckusick 				return(af);
4476639Smckusick 	return(!af);
4486639Smckusick }
4496639Smckusick 
succeed(f)4506639Smckusick succeed(f)
4516639Smckusick {
4526639Smckusick 	nsucc = 1;
4536639Smckusick 	if (sflag)
4546639Smckusick 		return;
4556639Smckusick 	if (cflag) {
4566639Smckusick 		tln++;
4576639Smckusick 		return;
4586639Smckusick 	}
45932365Sbostic 	if (oflag || !hflag && nfile > 1)
4606639Smckusick 		printf("%s:", f);
4616639Smckusick 	if (bflag)
4626639Smckusick 		printf("%d:", blkno);
4636639Smckusick 	if (nflag)
4646639Smckusick 		printf("%ld:", lnum);
4656639Smckusick 	printf("%s\n", linebuf);
4666639Smckusick }
4676639Smckusick 
digit(c)4686639Smckusick digit(c)
4696639Smckusick 	char c;
4706639Smckusick {
4716639Smckusick 	return (c>='0' && c<='9');
4726639Smckusick }
473