1*21559Sdist /*
2*21559Sdist  * Copyright (c) 1980 Regents of the University of California.
3*21559Sdist  * All rights reserved.  The Berkeley software License Agreement
4*21559Sdist  * specifies the terms and conditions for redistribution.
5*21559Sdist  */
6*21559Sdist 
714516Ssam #ifndef lint
8*21559Sdist char copyright[] =
9*21559Sdist "@(#) Copyright (c) 1980 Regents of the University of California.\n\
10*21559Sdist  All rights reserved.\n";
11*21559Sdist #endif not lint
126639Smckusick 
13*21559Sdist #ifndef lint
14*21559Sdist static char sccsid[] = "@(#)old.ucb.grep.c	5.1 (Berkeley) 05/31/85";
15*21559Sdist #endif not lint
16*21559Sdist 
176639Smckusick #include <stdio.h>
1821391Smckusick #include <sys/types.h>
1921391Smckusick #include <sys/stat.h>
206639Smckusick /*
216639Smckusick  * grep -- print lines matching (or not matching) a pattern
226639Smckusick  */
236639Smckusick 
2421391Smckusick #define BLKSIZE 8192
256639Smckusick #define	CCHR	2
266639Smckusick #define	CDOT	4
276639Smckusick #define	CCL	6
286639Smckusick #define	NCCL	8
296639Smckusick #define	CDOL	10
306639Smckusick #define	CEOF	11
316639Smckusick 
326639Smckusick #define	CBRC	14
336639Smckusick #define	CLET	15
346639Smckusick #define	STAR	01
356639Smckusick 
366639Smckusick #define	ESIZE	256
376639Smckusick 
386639Smckusick char	expbuf[ESIZE];
396639Smckusick long	lnum;
4021391Smckusick char	linebuf[BUFSIZ+1];
416639Smckusick int	bflag;
426639Smckusick int	nflag;
436639Smckusick int	cflag;
446639Smckusick int	vflag;
456639Smckusick int	nfile;
466639Smckusick int	iflag;
476639Smckusick int	lflag;
486639Smckusick int	wflag;
496639Smckusick int	sflag;
506639Smckusick int	nsucc;
516639Smckusick int	circf;
526639Smckusick int	blkno;
536639Smckusick long	tln;
546639Smckusick 
556639Smckusick main(argc, argv)
566639Smckusick char **argv;
576639Smckusick {
586639Smckusick 
596639Smckusick 	while (--argc > 0 && (++argv)[0][0]=='-') {
606639Smckusick 		char *cp = argv[0] + 1;
616639Smckusick 		while (*cp) switch (*cp++) {
626639Smckusick 
636639Smckusick 		case 'v':
646639Smckusick 			vflag++;
656639Smckusick 			continue;
666639Smckusick 
676639Smckusick 		case 'b':
686639Smckusick 			bflag++;
696639Smckusick 			continue;
706639Smckusick 
716639Smckusick 		case 'i':
726639Smckusick 		case 'y':	/* -y for compatibility with btl grep */
736639Smckusick 			iflag++;
746639Smckusick 			continue;
756639Smckusick 
766639Smckusick 		case 'l':
776639Smckusick 			lflag++;
786639Smckusick 		case 'c':
796639Smckusick 			cflag++;
806639Smckusick 			continue;
816639Smckusick 
826639Smckusick 		case 'w':
836639Smckusick 			wflag++;
846639Smckusick 			continue;
856639Smckusick 
866639Smckusick 		case 's':
876639Smckusick 			sflag++;
886639Smckusick 			continue;
896639Smckusick 
906639Smckusick 		case 'n':
916639Smckusick 			nflag++;
926639Smckusick 			continue;
936639Smckusick 
946639Smckusick 		case 'e':
956639Smckusick 			--argc;
966639Smckusick 			++argv;
976639Smckusick 			goto out;
986639Smckusick 
996639Smckusick 		default:
1006639Smckusick 			fprintf(stderr, "Unknown flag\n");
1016639Smckusick 			continue;
1026639Smckusick 		}
1036639Smckusick 	}
1046639Smckusick out:
1056639Smckusick 	if (argc<=0)
1066639Smckusick 		exit(2);
1076639Smckusick 	compile(*argv);
1086639Smckusick 	nfile = --argc;
1096639Smckusick 	if (argc<=0) {
1106639Smckusick 		if (lflag)
1116639Smckusick 			exit(1);
1126639Smckusick 		execute(0);
1136639Smckusick 	}
1146639Smckusick 	else while (--argc >= 0) {
1156639Smckusick 		argv++;
1166639Smckusick 		execute(*argv);
1176639Smckusick 	}
1186639Smckusick 	exit(nsucc == 0);
1196639Smckusick }
1206639Smckusick 
1216639Smckusick compile(astr)
1226639Smckusick char *astr;
1236639Smckusick {
1246639Smckusick 	register c;
1256639Smckusick 	register char *ep, *sp;
1266639Smckusick 	char *lastep;
1276639Smckusick 	int cclcnt;
1286639Smckusick 
1296639Smckusick 	ep = expbuf;
1306639Smckusick 	sp = astr;
1316639Smckusick 	if (*sp == '^') {
1326639Smckusick 		circf++;
1336639Smckusick 		sp++;
1346639Smckusick 	}
1356639Smckusick 	if (wflag)
1366639Smckusick 		*ep++ = CBRC;
1376639Smckusick 	for (;;) {
1386639Smckusick 		if (ep >= &expbuf[ESIZE])
1396639Smckusick 			goto cerror;
1406639Smckusick 		if ((c = *sp++) != '*')
1416639Smckusick 			lastep = ep;
1426639Smckusick 		switch (c) {
1436639Smckusick 
1446639Smckusick 		case '\0':
1456639Smckusick 			if (wflag)
1466639Smckusick 				*ep++ = CLET;
1476639Smckusick 			*ep++ = CEOF;
1486639Smckusick 			return;
1496639Smckusick 
1506639Smckusick 		case '.':
1516639Smckusick 			*ep++ = CDOT;
1526639Smckusick 			continue;
1536639Smckusick 
1546639Smckusick 		case '*':
1556639Smckusick 			if (lastep==0)
1566639Smckusick 				goto defchar;
1576639Smckusick 			*lastep |= STAR;
1586639Smckusick 			continue;
1596639Smckusick 
1606639Smckusick 		case '$':
1616639Smckusick 			if (*sp != '\0')
1626639Smckusick 				goto defchar;
1636639Smckusick 			*ep++ = CDOL;
1646639Smckusick 			continue;
1656639Smckusick 
1666639Smckusick 		case '[':
1676639Smckusick 			*ep++ = CCL;
1686639Smckusick 			*ep++ = 0;
1696639Smckusick 			cclcnt = 1;
1706639Smckusick 			if ((c = *sp++) == '^') {
1716639Smckusick 				c = *sp++;
1726639Smckusick 				ep[-2] = NCCL;
1736639Smckusick 			}
1746639Smckusick 			do {
1756639Smckusick 				*ep++ = c;
1766639Smckusick 				cclcnt++;
1776639Smckusick 				if (c=='\0' || ep >= &expbuf[ESIZE])
1786639Smckusick 					goto cerror;
1796639Smckusick 			} while ((c = *sp++) != ']');
1806639Smckusick 			lastep[1] = cclcnt;
1816639Smckusick 			continue;
1826639Smckusick 
1836639Smckusick 		case '\\':
1846639Smckusick 			if ((c = *sp++) == '\0')
1856639Smckusick 				goto cerror;
1866639Smckusick 			if (c == '<') {
1876639Smckusick 				*ep++ = CBRC;
1886639Smckusick 				continue;
1896639Smckusick 			}
1906639Smckusick 			if (c == '>') {
1916639Smckusick 				*ep++ = CLET;
1926639Smckusick 				continue;
1936639Smckusick 			}
1946639Smckusick 		defchar:
1956639Smckusick 		default:
1966639Smckusick 			*ep++ = CCHR;
1976639Smckusick 			*ep++ = c;
1986639Smckusick 		}
1996639Smckusick 	}
2006639Smckusick     cerror:
2016639Smckusick 	fprintf(stderr, "RE error\n");
2026639Smckusick }
2036639Smckusick 
2046639Smckusick same(a, b)
2056639Smckusick 	register int a, b;
2066639Smckusick {
2076639Smckusick 
2086639Smckusick 	return (a == b || iflag && (a ^ b) == ' ' && letter(a) == letter(b));
2096639Smckusick }
2106639Smckusick 
2116639Smckusick letter(c)
2126639Smckusick 	register int c;
2136639Smckusick {
2146639Smckusick 
2156639Smckusick 	if (c >= 'a' && c <= 'z')
2166639Smckusick 		return (c);
2176639Smckusick 	if (c >= 'A' && c <= 'Z')
2186639Smckusick 		return (c + 'a' - 'A');
2196639Smckusick 	return (0);
2206639Smckusick }
2216639Smckusick 
2226639Smckusick execute(file)
2236639Smckusick {
2246639Smckusick 	register char *p1, *p2;
2256639Smckusick 	register c;
2266639Smckusick 	int f;
2276639Smckusick 	char *ebp, *cbp;
22821391Smckusick 	static char *buf;
22921391Smckusick 	static int blksize;
23021391Smckusick 	struct stat stb;
2316639Smckusick 
2326639Smckusick 	if (file) {
2336639Smckusick 		if ((f = open(file, 0)) < 0) {
23412090Smckusick 			perror(file);
2356639Smckusick 		}
2366639Smckusick 	} else
2376639Smckusick 		f = 0;
23821391Smckusick 	if (buf == NULL) {
23921391Smckusick 		if (fstat(f, &stb) > 0 && stb.st_blksize > 0)
24021391Smckusick 			blksize = stb.st_blksize;
24121391Smckusick 		else
24221391Smckusick 			blksize = BLKSIZE;
24321391Smckusick 		buf = (char *)malloc(blksize);
24421391Smckusick 		if (buf == NULL) {
24521391Smckusick 			fprintf(stderr, "egrep: no memory for %s\n", file);
24621391Smckusick 			return;
24721391Smckusick 		}
24821391Smckusick 	}
24921391Smckusick 	ebp = buf;
25021391Smckusick 	cbp = buf;
2516639Smckusick 	lnum = 0;
2526639Smckusick 	tln = 0;
2536639Smckusick 	blkno = -1;
2546639Smckusick 	for (;;) {
2556639Smckusick 		lnum++;
2566639Smckusick 		if((lnum&0377) == 0)
2576639Smckusick 			fflush(stdout);
2586639Smckusick 		p1 = linebuf;
2596639Smckusick 		p2 = cbp;
2606639Smckusick 		for (;;) {
2616639Smckusick 			if (p2 >= ebp) {
26221391Smckusick 				if ((c = read(f, buf, blksize)) <= 0) {
2636639Smckusick 					close(f);
2646639Smckusick 					if (cflag) {
2656639Smckusick 						if (lflag) {
2666639Smckusick 							if (tln)
2676639Smckusick 							printf("%s\n", file);
2686639Smckusick 						} else {
2696639Smckusick 							if (nfile > 1)
2706639Smckusick 								printf("%s:", file);
2716639Smckusick 							printf("%ld\n", tln);
2726639Smckusick 						}
2736639Smckusick 					}
2746639Smckusick 					return;
2756639Smckusick 				}
2766639Smckusick 				blkno++;
27721391Smckusick 				p2 = buf;
27821391Smckusick 				ebp = buf+c;
2796639Smckusick 			}
2806639Smckusick 			if ((c = *p2++) == '\n')
2816639Smckusick 				break;
2826639Smckusick 			if(c)
28321391Smckusick 			if (p1 < &linebuf[BUFSIZ-1])
2846639Smckusick 				*p1++ = c;
2856639Smckusick 		}
2866639Smckusick 		*p1++ = 0;
2876639Smckusick 		cbp = p2;
2886639Smckusick 		p1 = linebuf;
2896639Smckusick 		p2 = expbuf;
2906639Smckusick 		if (circf) {
2916639Smckusick 			if (advance(p1, p2))
2926639Smckusick 				goto found;
2936639Smckusick 			goto nfound;
2946639Smckusick 		}
2956639Smckusick 		/* fast check for first character */
2966639Smckusick 		if (*p2==CCHR) {
2976639Smckusick 			c = p2[1];
2986639Smckusick 			do {
2996639Smckusick 				if (*p1!=c && (!iflag || (c ^ *p1) != ' '
3006639Smckusick 					|| letter(c) != letter(*p1)))
3016639Smckusick 					continue;
3026639Smckusick 				if (advance(p1, p2))
3036639Smckusick 					goto found;
3046639Smckusick 			} while (*p1++);
3056639Smckusick 			goto nfound;
3066639Smckusick 		}
3076639Smckusick 		/* regular algorithm */
3086639Smckusick 		do {
3096639Smckusick 			if (advance(p1, p2))
3106639Smckusick 				goto found;
3116639Smckusick 		} while (*p1++);
3126639Smckusick 	nfound:
3136639Smckusick 		if (vflag)
3146639Smckusick 			succeed(file);
3156639Smckusick 		continue;
3166639Smckusick 	found:
3176639Smckusick 		if (vflag==0)
3186639Smckusick 			succeed(file);
3196639Smckusick 	}
3206639Smckusick }
3216639Smckusick 
3226639Smckusick advance(alp, aep)
3236639Smckusick 	char *alp, *aep;
3246639Smckusick {
3256639Smckusick 	register char *lp, *ep, *curlp;
3266639Smckusick 	char *nextep;
3276639Smckusick 
3286639Smckusick 	lp = alp;
3296639Smckusick 	ep = aep;
3306639Smckusick 	for (;;) switch (*ep++) {
3316639Smckusick 
3326639Smckusick 	case CCHR:
3336639Smckusick 		if (!same(*ep, *lp))
3346639Smckusick 			return (0);
3356639Smckusick 		ep++, lp++;
3366639Smckusick 		continue;
3376639Smckusick 
3386639Smckusick 	case CDOT:
3396639Smckusick 		if (*lp++)
3406639Smckusick 			continue;
3416639Smckusick 		return(0);
3426639Smckusick 
3436639Smckusick 	case CDOL:
3446639Smckusick 		if (*lp==0)
3456639Smckusick 			continue;
3466639Smckusick 		return(0);
3476639Smckusick 
3486639Smckusick 	case CEOF:
3496639Smckusick 		return(1);
3506639Smckusick 
3516639Smckusick 	case CCL:
3526639Smckusick 		if (cclass(ep, *lp++, 1)) {
3536639Smckusick 			ep += *ep;
3546639Smckusick 			continue;
3556639Smckusick 		}
3566639Smckusick 		return(0);
3576639Smckusick 
3586639Smckusick 	case NCCL:
3596639Smckusick 		if (cclass(ep, *lp++, 0)) {
3606639Smckusick 			ep += *ep;
3616639Smckusick 			continue;
3626639Smckusick 		}
3636639Smckusick 		return(0);
3646639Smckusick 
3656639Smckusick 	case CDOT|STAR:
3666639Smckusick 		curlp = lp;
3676639Smckusick 		while (*lp++);
3686639Smckusick 		goto star;
3696639Smckusick 
3706639Smckusick 	case CCHR|STAR:
3716639Smckusick 		curlp = lp;
3726639Smckusick 		while (same(*lp, *ep))
3736639Smckusick 			lp++;
3746639Smckusick 		lp++;
3756639Smckusick 		ep++;
3766639Smckusick 		goto star;
3776639Smckusick 
3786639Smckusick 	case CCL|STAR:
3796639Smckusick 	case NCCL|STAR:
3806639Smckusick 		curlp = lp;
3816639Smckusick 		while (cclass(ep, *lp++, ep[-1]==(CCL|STAR)));
3826639Smckusick 		ep += *ep;
3836639Smckusick 		goto star;
3846639Smckusick 
3856639Smckusick 	star:
3866639Smckusick 		do {
3876639Smckusick 			lp--;
3886639Smckusick 			if (advance(lp, ep))
3896639Smckusick 				return(1);
3906639Smckusick 		} while (lp > curlp);
3916639Smckusick 		return(0);
3926639Smckusick 
3936639Smckusick 	case CBRC:
3946639Smckusick 		if (lp == expbuf)
3956639Smckusick 			continue;
3966639Smckusick #define	uletter(c)	(letter(c) || c == '_')
3976639Smckusick 		if ( ( uletter(*lp) || digit ( * lp ) )  && !uletter(lp[-1]) && !digit(lp[-1]))
3986639Smckusick 			continue;
3996639Smckusick 		return (0);
4006639Smckusick 
4016639Smckusick 	case CLET:
4026639Smckusick 		if (!uletter(*lp) && !digit(*lp))
4036639Smckusick 			continue;
4046639Smckusick 		return (0);
4056639Smckusick 
4066639Smckusick 	default:
4076639Smckusick 		fprintf(stderr, "RE botch\n");
4086639Smckusick 	}
4096639Smckusick }
4106639Smckusick 
4116639Smckusick cclass(aset, ac, af)
4126639Smckusick 	char *aset;
4136639Smckusick {
4146639Smckusick 	register char *set, c;
4156639Smckusick 	register n;
4166639Smckusick 
4176639Smckusick 	set = aset;
4186639Smckusick 	if ((c = ac) == 0)
4196639Smckusick 		return(0);
4206639Smckusick 	n = *set++;
4216639Smckusick 	while (--n)
4226639Smckusick 		if (n > 2 && set[1] == '-') {
4236639Smckusick 			if (c >= (set[0] & 0177) && c <= (set[2] & 0177))
4246639Smckusick 				return (af);
4256639Smckusick 			set += 3;
4266639Smckusick 			n -= 2;
4276639Smckusick 		} else
4286639Smckusick 			if ((*set++ & 0177) == c)
4296639Smckusick 				return(af);
4306639Smckusick 	return(!af);
4316639Smckusick }
4326639Smckusick 
4336639Smckusick succeed(f)
4346639Smckusick {
4356639Smckusick 	nsucc = 1;
4366639Smckusick 	if (sflag)
4376639Smckusick 		return;
4386639Smckusick 	if (cflag) {
4396639Smckusick 		tln++;
4406639Smckusick 		return;
4416639Smckusick 	}
4426639Smckusick 	if (nfile > 1)
4436639Smckusick 		printf("%s:", f);
4446639Smckusick 	if (bflag)
4456639Smckusick 		printf("%d:", blkno);
4466639Smckusick 	if (nflag)
4476639Smckusick 		printf("%ld:", lnum);
4486639Smckusick 	printf("%s\n", linebuf);
4496639Smckusick }
4506639Smckusick 
4516639Smckusick digit(c)
4526639Smckusick 	char c;
4536639Smckusick {
4546639Smckusick 	return (c>='0' && c<='9');
4556639Smckusick }
456