114516Ssam #ifndef lint
2*21391Smckusick static char sccsid[] = "@(#)old.ucb.grep.c	4.5 (Berkeley) 05/30/85";
314516Ssam #endif
46639Smckusick 
56639Smckusick #include <stdio.h>
6*21391Smckusick #include <sys/types.h>
7*21391Smckusick #include <sys/stat.h>
86639Smckusick /*
96639Smckusick  * grep -- print lines matching (or not matching) a pattern
106639Smckusick  */
116639Smckusick 
12*21391Smckusick #define BLKSIZE 8192
136639Smckusick #define	CCHR	2
146639Smckusick #define	CDOT	4
156639Smckusick #define	CCL	6
166639Smckusick #define	NCCL	8
176639Smckusick #define	CDOL	10
186639Smckusick #define	CEOF	11
196639Smckusick 
206639Smckusick #define	CBRC	14
216639Smckusick #define	CLET	15
226639Smckusick #define	STAR	01
236639Smckusick 
246639Smckusick #define	ESIZE	256
256639Smckusick 
266639Smckusick char	expbuf[ESIZE];
276639Smckusick long	lnum;
28*21391Smckusick char	linebuf[BUFSIZ+1];
296639Smckusick int	bflag;
306639Smckusick int	nflag;
316639Smckusick int	cflag;
326639Smckusick int	vflag;
336639Smckusick int	nfile;
346639Smckusick int	iflag;
356639Smckusick int	lflag;
366639Smckusick int	wflag;
376639Smckusick int	sflag;
386639Smckusick int	nsucc;
396639Smckusick int	circf;
406639Smckusick int	blkno;
416639Smckusick long	tln;
426639Smckusick 
436639Smckusick main(argc, argv)
446639Smckusick char **argv;
456639Smckusick {
466639Smckusick 
476639Smckusick 	while (--argc > 0 && (++argv)[0][0]=='-') {
486639Smckusick 		char *cp = argv[0] + 1;
496639Smckusick 		while (*cp) switch (*cp++) {
506639Smckusick 
516639Smckusick 		case 'v':
526639Smckusick 			vflag++;
536639Smckusick 			continue;
546639Smckusick 
556639Smckusick 		case 'b':
566639Smckusick 			bflag++;
576639Smckusick 			continue;
586639Smckusick 
596639Smckusick 		case 'i':
606639Smckusick 		case 'y':	/* -y for compatibility with btl grep */
616639Smckusick 			iflag++;
626639Smckusick 			continue;
636639Smckusick 
646639Smckusick 		case 'l':
656639Smckusick 			lflag++;
666639Smckusick 		case 'c':
676639Smckusick 			cflag++;
686639Smckusick 			continue;
696639Smckusick 
706639Smckusick 		case 'w':
716639Smckusick 			wflag++;
726639Smckusick 			continue;
736639Smckusick 
746639Smckusick 		case 's':
756639Smckusick 			sflag++;
766639Smckusick 			continue;
776639Smckusick 
786639Smckusick 		case 'n':
796639Smckusick 			nflag++;
806639Smckusick 			continue;
816639Smckusick 
826639Smckusick 		case 'e':
836639Smckusick 			--argc;
846639Smckusick 			++argv;
856639Smckusick 			goto out;
866639Smckusick 
876639Smckusick 		default:
886639Smckusick 			fprintf(stderr, "Unknown flag\n");
896639Smckusick 			continue;
906639Smckusick 		}
916639Smckusick 	}
926639Smckusick out:
936639Smckusick 	if (argc<=0)
946639Smckusick 		exit(2);
956639Smckusick 	compile(*argv);
966639Smckusick 	nfile = --argc;
976639Smckusick 	if (argc<=0) {
986639Smckusick 		if (lflag)
996639Smckusick 			exit(1);
1006639Smckusick 		execute(0);
1016639Smckusick 	}
1026639Smckusick 	else while (--argc >= 0) {
1036639Smckusick 		argv++;
1046639Smckusick 		execute(*argv);
1056639Smckusick 	}
1066639Smckusick 	exit(nsucc == 0);
1076639Smckusick }
1086639Smckusick 
1096639Smckusick compile(astr)
1106639Smckusick char *astr;
1116639Smckusick {
1126639Smckusick 	register c;
1136639Smckusick 	register char *ep, *sp;
1146639Smckusick 	char *lastep;
1156639Smckusick 	int cclcnt;
1166639Smckusick 
1176639Smckusick 	ep = expbuf;
1186639Smckusick 	sp = astr;
1196639Smckusick 	if (*sp == '^') {
1206639Smckusick 		circf++;
1216639Smckusick 		sp++;
1226639Smckusick 	}
1236639Smckusick 	if (wflag)
1246639Smckusick 		*ep++ = CBRC;
1256639Smckusick 	for (;;) {
1266639Smckusick 		if (ep >= &expbuf[ESIZE])
1276639Smckusick 			goto cerror;
1286639Smckusick 		if ((c = *sp++) != '*')
1296639Smckusick 			lastep = ep;
1306639Smckusick 		switch (c) {
1316639Smckusick 
1326639Smckusick 		case '\0':
1336639Smckusick 			if (wflag)
1346639Smckusick 				*ep++ = CLET;
1356639Smckusick 			*ep++ = CEOF;
1366639Smckusick 			return;
1376639Smckusick 
1386639Smckusick 		case '.':
1396639Smckusick 			*ep++ = CDOT;
1406639Smckusick 			continue;
1416639Smckusick 
1426639Smckusick 		case '*':
1436639Smckusick 			if (lastep==0)
1446639Smckusick 				goto defchar;
1456639Smckusick 			*lastep |= STAR;
1466639Smckusick 			continue;
1476639Smckusick 
1486639Smckusick 		case '$':
1496639Smckusick 			if (*sp != '\0')
1506639Smckusick 				goto defchar;
1516639Smckusick 			*ep++ = CDOL;
1526639Smckusick 			continue;
1536639Smckusick 
1546639Smckusick 		case '[':
1556639Smckusick 			*ep++ = CCL;
1566639Smckusick 			*ep++ = 0;
1576639Smckusick 			cclcnt = 1;
1586639Smckusick 			if ((c = *sp++) == '^') {
1596639Smckusick 				c = *sp++;
1606639Smckusick 				ep[-2] = NCCL;
1616639Smckusick 			}
1626639Smckusick 			do {
1636639Smckusick 				*ep++ = c;
1646639Smckusick 				cclcnt++;
1656639Smckusick 				if (c=='\0' || ep >= &expbuf[ESIZE])
1666639Smckusick 					goto cerror;
1676639Smckusick 			} while ((c = *sp++) != ']');
1686639Smckusick 			lastep[1] = cclcnt;
1696639Smckusick 			continue;
1706639Smckusick 
1716639Smckusick 		case '\\':
1726639Smckusick 			if ((c = *sp++) == '\0')
1736639Smckusick 				goto cerror;
1746639Smckusick 			if (c == '<') {
1756639Smckusick 				*ep++ = CBRC;
1766639Smckusick 				continue;
1776639Smckusick 			}
1786639Smckusick 			if (c == '>') {
1796639Smckusick 				*ep++ = CLET;
1806639Smckusick 				continue;
1816639Smckusick 			}
1826639Smckusick 		defchar:
1836639Smckusick 		default:
1846639Smckusick 			*ep++ = CCHR;
1856639Smckusick 			*ep++ = c;
1866639Smckusick 		}
1876639Smckusick 	}
1886639Smckusick     cerror:
1896639Smckusick 	fprintf(stderr, "RE error\n");
1906639Smckusick }
1916639Smckusick 
1926639Smckusick same(a, b)
1936639Smckusick 	register int a, b;
1946639Smckusick {
1956639Smckusick 
1966639Smckusick 	return (a == b || iflag && (a ^ b) == ' ' && letter(a) == letter(b));
1976639Smckusick }
1986639Smckusick 
1996639Smckusick letter(c)
2006639Smckusick 	register int c;
2016639Smckusick {
2026639Smckusick 
2036639Smckusick 	if (c >= 'a' && c <= 'z')
2046639Smckusick 		return (c);
2056639Smckusick 	if (c >= 'A' && c <= 'Z')
2066639Smckusick 		return (c + 'a' - 'A');
2076639Smckusick 	return (0);
2086639Smckusick }
2096639Smckusick 
2106639Smckusick execute(file)
2116639Smckusick {
2126639Smckusick 	register char *p1, *p2;
2136639Smckusick 	register c;
2146639Smckusick 	int f;
2156639Smckusick 	char *ebp, *cbp;
216*21391Smckusick 	static char *buf;
217*21391Smckusick 	static int blksize;
218*21391Smckusick 	struct stat stb;
2196639Smckusick 
2206639Smckusick 	if (file) {
2216639Smckusick 		if ((f = open(file, 0)) < 0) {
22212090Smckusick 			perror(file);
2236639Smckusick 		}
2246639Smckusick 	} else
2256639Smckusick 		f = 0;
226*21391Smckusick 	if (buf == NULL) {
227*21391Smckusick 		if (fstat(f, &stb) > 0 && stb.st_blksize > 0)
228*21391Smckusick 			blksize = stb.st_blksize;
229*21391Smckusick 		else
230*21391Smckusick 			blksize = BLKSIZE;
231*21391Smckusick 		buf = (char *)malloc(blksize);
232*21391Smckusick 		if (buf == NULL) {
233*21391Smckusick 			fprintf(stderr, "egrep: no memory for %s\n", file);
234*21391Smckusick 			return;
235*21391Smckusick 		}
236*21391Smckusick 	}
237*21391Smckusick 	ebp = buf;
238*21391Smckusick 	cbp = buf;
2396639Smckusick 	lnum = 0;
2406639Smckusick 	tln = 0;
2416639Smckusick 	blkno = -1;
2426639Smckusick 	for (;;) {
2436639Smckusick 		lnum++;
2446639Smckusick 		if((lnum&0377) == 0)
2456639Smckusick 			fflush(stdout);
2466639Smckusick 		p1 = linebuf;
2476639Smckusick 		p2 = cbp;
2486639Smckusick 		for (;;) {
2496639Smckusick 			if (p2 >= ebp) {
250*21391Smckusick 				if ((c = read(f, buf, blksize)) <= 0) {
2516639Smckusick 					close(f);
2526639Smckusick 					if (cflag) {
2536639Smckusick 						if (lflag) {
2546639Smckusick 							if (tln)
2556639Smckusick 							printf("%s\n", file);
2566639Smckusick 						} else {
2576639Smckusick 							if (nfile > 1)
2586639Smckusick 								printf("%s:", file);
2596639Smckusick 							printf("%ld\n", tln);
2606639Smckusick 						}
2616639Smckusick 					}
2626639Smckusick 					return;
2636639Smckusick 				}
2646639Smckusick 				blkno++;
265*21391Smckusick 				p2 = buf;
266*21391Smckusick 				ebp = buf+c;
2676639Smckusick 			}
2686639Smckusick 			if ((c = *p2++) == '\n')
2696639Smckusick 				break;
2706639Smckusick 			if(c)
271*21391Smckusick 			if (p1 < &linebuf[BUFSIZ-1])
2726639Smckusick 				*p1++ = c;
2736639Smckusick 		}
2746639Smckusick 		*p1++ = 0;
2756639Smckusick 		cbp = p2;
2766639Smckusick 		p1 = linebuf;
2776639Smckusick 		p2 = expbuf;
2786639Smckusick 		if (circf) {
2796639Smckusick 			if (advance(p1, p2))
2806639Smckusick 				goto found;
2816639Smckusick 			goto nfound;
2826639Smckusick 		}
2836639Smckusick 		/* fast check for first character */
2846639Smckusick 		if (*p2==CCHR) {
2856639Smckusick 			c = p2[1];
2866639Smckusick 			do {
2876639Smckusick 				if (*p1!=c && (!iflag || (c ^ *p1) != ' '
2886639Smckusick 					|| letter(c) != letter(*p1)))
2896639Smckusick 					continue;
2906639Smckusick 				if (advance(p1, p2))
2916639Smckusick 					goto found;
2926639Smckusick 			} while (*p1++);
2936639Smckusick 			goto nfound;
2946639Smckusick 		}
2956639Smckusick 		/* regular algorithm */
2966639Smckusick 		do {
2976639Smckusick 			if (advance(p1, p2))
2986639Smckusick 				goto found;
2996639Smckusick 		} while (*p1++);
3006639Smckusick 	nfound:
3016639Smckusick 		if (vflag)
3026639Smckusick 			succeed(file);
3036639Smckusick 		continue;
3046639Smckusick 	found:
3056639Smckusick 		if (vflag==0)
3066639Smckusick 			succeed(file);
3076639Smckusick 	}
3086639Smckusick }
3096639Smckusick 
3106639Smckusick advance(alp, aep)
3116639Smckusick 	char *alp, *aep;
3126639Smckusick {
3136639Smckusick 	register char *lp, *ep, *curlp;
3146639Smckusick 	char *nextep;
3156639Smckusick 
3166639Smckusick 	lp = alp;
3176639Smckusick 	ep = aep;
3186639Smckusick 	for (;;) switch (*ep++) {
3196639Smckusick 
3206639Smckusick 	case CCHR:
3216639Smckusick 		if (!same(*ep, *lp))
3226639Smckusick 			return (0);
3236639Smckusick 		ep++, lp++;
3246639Smckusick 		continue;
3256639Smckusick 
3266639Smckusick 	case CDOT:
3276639Smckusick 		if (*lp++)
3286639Smckusick 			continue;
3296639Smckusick 		return(0);
3306639Smckusick 
3316639Smckusick 	case CDOL:
3326639Smckusick 		if (*lp==0)
3336639Smckusick 			continue;
3346639Smckusick 		return(0);
3356639Smckusick 
3366639Smckusick 	case CEOF:
3376639Smckusick 		return(1);
3386639Smckusick 
3396639Smckusick 	case CCL:
3406639Smckusick 		if (cclass(ep, *lp++, 1)) {
3416639Smckusick 			ep += *ep;
3426639Smckusick 			continue;
3436639Smckusick 		}
3446639Smckusick 		return(0);
3456639Smckusick 
3466639Smckusick 	case NCCL:
3476639Smckusick 		if (cclass(ep, *lp++, 0)) {
3486639Smckusick 			ep += *ep;
3496639Smckusick 			continue;
3506639Smckusick 		}
3516639Smckusick 		return(0);
3526639Smckusick 
3536639Smckusick 	case CDOT|STAR:
3546639Smckusick 		curlp = lp;
3556639Smckusick 		while (*lp++);
3566639Smckusick 		goto star;
3576639Smckusick 
3586639Smckusick 	case CCHR|STAR:
3596639Smckusick 		curlp = lp;
3606639Smckusick 		while (same(*lp, *ep))
3616639Smckusick 			lp++;
3626639Smckusick 		lp++;
3636639Smckusick 		ep++;
3646639Smckusick 		goto star;
3656639Smckusick 
3666639Smckusick 	case CCL|STAR:
3676639Smckusick 	case NCCL|STAR:
3686639Smckusick 		curlp = lp;
3696639Smckusick 		while (cclass(ep, *lp++, ep[-1]==(CCL|STAR)));
3706639Smckusick 		ep += *ep;
3716639Smckusick 		goto star;
3726639Smckusick 
3736639Smckusick 	star:
3746639Smckusick 		do {
3756639Smckusick 			lp--;
3766639Smckusick 			if (advance(lp, ep))
3776639Smckusick 				return(1);
3786639Smckusick 		} while (lp > curlp);
3796639Smckusick 		return(0);
3806639Smckusick 
3816639Smckusick 	case CBRC:
3826639Smckusick 		if (lp == expbuf)
3836639Smckusick 			continue;
3846639Smckusick #define	uletter(c)	(letter(c) || c == '_')
3856639Smckusick 		if ( ( uletter(*lp) || digit ( * lp ) )  && !uletter(lp[-1]) && !digit(lp[-1]))
3866639Smckusick 			continue;
3876639Smckusick 		return (0);
3886639Smckusick 
3896639Smckusick 	case CLET:
3906639Smckusick 		if (!uletter(*lp) && !digit(*lp))
3916639Smckusick 			continue;
3926639Smckusick 		return (0);
3936639Smckusick 
3946639Smckusick 	default:
3956639Smckusick 		fprintf(stderr, "RE botch\n");
3966639Smckusick 	}
3976639Smckusick }
3986639Smckusick 
3996639Smckusick cclass(aset, ac, af)
4006639Smckusick 	char *aset;
4016639Smckusick {
4026639Smckusick 	register char *set, c;
4036639Smckusick 	register n;
4046639Smckusick 
4056639Smckusick 	set = aset;
4066639Smckusick 	if ((c = ac) == 0)
4076639Smckusick 		return(0);
4086639Smckusick 	n = *set++;
4096639Smckusick 	while (--n)
4106639Smckusick 		if (n > 2 && set[1] == '-') {
4116639Smckusick 			if (c >= (set[0] & 0177) && c <= (set[2] & 0177))
4126639Smckusick 				return (af);
4136639Smckusick 			set += 3;
4146639Smckusick 			n -= 2;
4156639Smckusick 		} else
4166639Smckusick 			if ((*set++ & 0177) == c)
4176639Smckusick 				return(af);
4186639Smckusick 	return(!af);
4196639Smckusick }
4206639Smckusick 
4216639Smckusick succeed(f)
4226639Smckusick {
4236639Smckusick 	nsucc = 1;
4246639Smckusick 	if (sflag)
4256639Smckusick 		return;
4266639Smckusick 	if (cflag) {
4276639Smckusick 		tln++;
4286639Smckusick 		return;
4296639Smckusick 	}
4306639Smckusick 	if (nfile > 1)
4316639Smckusick 		printf("%s:", f);
4326639Smckusick 	if (bflag)
4336639Smckusick 		printf("%d:", blkno);
4346639Smckusick 	if (nflag)
4356639Smckusick 		printf("%ld:", lnum);
4366639Smckusick 	printf("%s\n", linebuf);
4376639Smckusick }
4386639Smckusick 
4396639Smckusick digit(c)
4406639Smckusick 	char c;
4416639Smckusick {
4426639Smckusick 	return (c>='0' && c<='9');
4436639Smckusick }
444