meillo@14: /*
meillo@14:  * cut - cut out fields of lines of files
meillo@14:  *
meillo@14:  * Gunnar Ritter, Freiburg i. Br., Germany, December 2002.
meillo@14:  */
meillo@14: /*
meillo@14:  * Copyright (c) 2003 Gunnar Ritter
meillo@14:  *
meillo@14:  * This software is provided 'as-is', without any express or implied
meillo@14:  * warranty. In no event will the authors be held liable for any damages
meillo@14:  * arising from the use of this software.
meillo@14:  *
meillo@14:  * Permission is granted to anyone to use this software for any purpose,
meillo@14:  * including commercial applications, and to alter it and redistribute
meillo@14:  * it freely, subject to the following restrictions:
meillo@14:  *
meillo@14:  * 1. The origin of this software must not be misrepresented; you must not
meillo@14:  *    claim that you wrote the original software. If you use this software
meillo@14:  *    in a product, an acknowledgment in the product documentation would be
meillo@14:  *    appreciated but is not required.
meillo@14:  *
meillo@14:  * 2. Altered source versions must be plainly marked as such, and must not be
meillo@14:  *    misrepresented as being the original software.
meillo@14:  *
meillo@14:  * 3. This notice may not be removed or altered from any source distribution.
meillo@14:  */
meillo@14: 
meillo@14: #if __GNUC__ >= 3 && __GNUC_MINOR__ >= 4 || __GNUC__ >= 4
meillo@14: #define	USED	__attribute__ ((used))
meillo@14: #elif defined __GNUC__
meillo@14: #define	USED	__attribute__ ((unused))
meillo@14: #else
meillo@14: #define	USED
meillo@14: #endif
meillo@14: static const char sccsid[] USED = "@(#)cut.sl	1.20 (gritter) 5/29/05";
meillo@14: 
meillo@14: #include	<sys/types.h>
meillo@14: #include	<sys/stat.h>
meillo@14: #include	<fcntl.h>
meillo@14: #include	<unistd.h>
meillo@14: #include	<stdio.h>
meillo@14: #include	<string.h>
meillo@14: #include	<stdlib.h>
meillo@14: #include	<errno.h>
meillo@14: #include	<libgen.h>
meillo@14: #include	<limits.h>
meillo@14: #include	<wchar.h>
meillo@14: #include	<ctype.h>
meillo@14: #include	<locale.h>
meillo@14: 
meillo@14: #include	"iblok.h"
meillo@14: 
meillo@14: #if defined (__GLIBC__) && defined (_IO_putc_unlocked)
meillo@14: #undef	putc
meillo@14: #define	putc(c, f)	_IO_putc_unlocked(c, f)
meillo@14: #endif
meillo@14: 
meillo@14: struct	range {
meillo@14: 	struct range	*r_nxt;
meillo@14: 	long	r_min;
meillo@14: 	long	r_max;
meillo@14: };
meillo@14: 
meillo@14: static unsigned	errcnt;			/* count of errors */
meillo@14: static int	method;			/* one of b, c, f */
meillo@14: static int	nflag;			/* character boundary bytes */
meillo@14: static int	sflag;			/* suppress lines w/o delimiters */
meillo@14: static char	*progname;		/* argv[0] to main() */
meillo@14: static wchar_t	wcdelim = '\t';		/* delimiter character */
meillo@14: static const char	*mbdelim = "\t";/* delimiter character */
meillo@14: struct range	*fields;		/* range list */
meillo@14: static int	multibyte;		/* multibyte LC_CTYPE */
meillo@14: 
meillo@14: #define	next(wc, s)	(multibyte ? mbtowc(&(wc), s, MB_LEN_MAX) :\
meillo@14: 				((wc) = *(s) & 0377, (wc) != 0))
meillo@14: 
meillo@14: void *
meillo@14: lrealloc(void *vp, size_t nbytes)
meillo@14: {
meillo@14: 	void	*p;
meillo@14: 
meillo@14: 	if ((p = realloc(vp, nbytes)) == NULL) {
meillo@14: 		write(2, "line too long\n", 14);
meillo@14: 		exit(076);
meillo@14: 	}
meillo@14: 	return p;
meillo@14: }
meillo@14: 
meillo@14: void *
meillo@14: smalloc(size_t nbytes)
meillo@14: {
meillo@14: 	void	*p;
meillo@14: 
meillo@14: 	if ((p = malloc(nbytes)) == NULL) {
meillo@14: 		write(2, "no memory\n", 11);
meillo@14: 		exit(077);
meillo@14: 	}
meillo@14: 	return p;
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: error(const char *s)
meillo@14: {
meillo@14: 	fprintf(stderr, "%s: ERROR: %s\n", progname, s);
meillo@14: 	exit(2);
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: usage(void)
meillo@14: {
meillo@14: 	error("Usage: cut [-s] [-d<char>] {-c<list> | -f<list>} file ...");
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: badlist(void)
meillo@14: {
meillo@14: 	error(method == 'b' ? "bad list for b/c/f option" : 
meillo@14: 			"bad list for c/f option");
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: setdelim(const char *s)
meillo@14: {
meillo@14: 	int	n;
meillo@14: 
meillo@14: 	if ((n = next(wcdelim, s)) < 0 || (n > 0 && s[n] != '\0'))
meillo@14: 		error("no delimiter");
meillo@14: 	mbdelim = s;
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: addrange(long m, long n)
meillo@14: {
meillo@14: 	struct range	*rp, *rq;
meillo@14: 
meillo@14: 	rp = smalloc(sizeof *rp);
meillo@14: 	rp->r_nxt = NULL;
meillo@14: 	rp->r_min = m;
meillo@14: 	rp->r_max = n ? n : m;
meillo@14: 	if (fields) {
meillo@14: 		for (rq = fields; rq->r_nxt; rq = rq->r_nxt);
meillo@14: 		rq->r_nxt = rp;
meillo@14: 	} else
meillo@14: 		fields = rp;
meillo@14: }
meillo@14: 
meillo@14: static int
meillo@14: have(long i)
meillo@14: {
meillo@14: 	struct range	*rp;
meillo@14: 
meillo@14: 	for (rp = fields; rp; rp = rp->r_nxt)
meillo@14: 		if (i >= rp->r_min && i <= rp->r_max)
meillo@14: 			return 1;
meillo@14: 	return 0;
meillo@14: }
meillo@14: 
meillo@14: #define	mnreset()	m = 0, n = 0, lp = &m
meillo@14: 
meillo@14: static void
meillo@14: setlist(const char *s)
meillo@14: {
meillo@14: 	char	*cbuf, *cp;
meillo@14: 	long	m, n;
meillo@14: 	long	*lp;
meillo@14: 
meillo@14: 	fields = NULL;
meillo@14: 	cbuf = smalloc(strlen(s) + 1);
meillo@14: 	mnreset();
meillo@14: 	for (;;) {
meillo@14: 		if (*s == '-') {
meillo@14: 			if (m == 0)
meillo@14: 				m = 1;
meillo@14: 			n = LONG_MAX;
meillo@14: 			lp = &n;
meillo@14: 			s++;
meillo@14: 		} else if (*s == ',' || *s == ' ' || *s == '\t' || *s == '\0') {
meillo@14: 			if (m)
meillo@14: 				addrange(m, n);
meillo@14: 			mnreset();
meillo@14: 			if (*s == '\0')
meillo@14: 				break;
meillo@14: 			s++;
meillo@14: 		} else if (isdigit(*s & 0377)) {
meillo@14: 			cp = cbuf;
meillo@14: 			do
meillo@14: 				*cp++ = *s++;
meillo@14: 			while (isdigit(*s & 0377));
meillo@14: 			*cp = '\0';
meillo@14: 			*lp = strtol(cbuf, NULL, 10);
meillo@14: 		} else
meillo@14: 			badlist();
meillo@14: 	}
meillo@14: 	if (fields == NULL)
meillo@14: 		error("no fields");
meillo@14: 	free(cbuf);
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: cutb(struct iblok *ip)
meillo@14: {
meillo@14: 	int	c, i;
meillo@14: 
meillo@14: 	i = 1;
meillo@14: 	while ((c = ib_get(ip)) != EOF) {
meillo@14: 		if (c == '\n') {
meillo@14: 			i = 1;
meillo@14: 			putc(c, stdout);
meillo@14: 		} else if (have(i++))
meillo@14: 			putc(c, stdout);
meillo@14: 	}
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: cutbn(struct iblok *ip)
meillo@14: {
meillo@14: 	char	*cp;
meillo@14: 	int	i, m, n;
meillo@14: 	wint_t	wc;
meillo@14: 
meillo@14: 	i = 1;
meillo@14: 	while ((cp = ib_getw(ip, &wc, &n)) != NULL) {
meillo@14: 		if (wc == '\n') {
meillo@14: 			i = 1;
meillo@14: 			putc('\n', stdout);
meillo@14: 		} else {
meillo@14: 			if (have(i + n - 1))
meillo@14: 				for (m = 0; m < n; m++)
meillo@14: 					putc(cp[m], stdout);
meillo@14: 			i += n;
meillo@14: 		}
meillo@14: 	}
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: cutc(struct iblok *ip)
meillo@14: {
meillo@14: 	char	*cp;
meillo@14: 	int	i, n, m;
meillo@14: 	wint_t	wc;
meillo@14: 
meillo@14: 	i = 1;
meillo@14: 	while ((cp = ib_getw(ip, &wc, &n)) != NULL) {
meillo@14: 		if (wc == '\n') {
meillo@14: 			i = 1;
meillo@14: 			putc('\n', stdout);
meillo@14: 		} else if (wc != WEOF && have(i++)) {
meillo@14: 			for (m = 0; m < n; m++)
meillo@14: 				putc(cp[m], stdout);
meillo@14: 		}
meillo@14: 	}
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: cutf(struct iblok *ip)
meillo@14: {
meillo@14: 	static char	*line;
meillo@14: 	static size_t	linesize;
meillo@14: 	char	*cp, *lp, *lq;
meillo@14: 	int	c, i, n, m, gotcha;
meillo@14: 	char	b;
meillo@14: 	wint_t	wc;
meillo@14: 	const int	incr = 128;
meillo@14: 
meillo@14: 	if (linesize == 0)
meillo@14: 		line = smalloc(linesize = incr);
meillo@14: 	lp = line;
meillo@14: 	gotcha = 0;
meillo@14: 	i = 1;
meillo@14: 	do {
meillo@14: 		if (multibyte)
meillo@14: 			cp = ib_getw(ip, &wc, &n);
meillo@14: 		else {
meillo@14: 			if ((c = ib_get(ip)) != EOF) {
meillo@14: 				wc = c;
meillo@14: 				b = (char)c;
meillo@14: 				cp = &b;
meillo@14: 			} else {
meillo@14: 				wc = WEOF;
meillo@14: 				cp = NULL;
meillo@14: 			}
meillo@14: 			n = 1;
meillo@14: 		}
meillo@14: 		if (cp == NULL || wc == '\n' || wc == wcdelim) {
meillo@14: 			if (have(i) && (!sflag || gotcha || wc == wcdelim) ||
meillo@14: 					(!sflag && i == 1 &&
meillo@14: 						(cp == NULL || wc == '\n'))) {
meillo@14: 				if (gotcha)
meillo@14: 					for (m = 0; mbdelim[m]; m++)
meillo@14: 						putc(mbdelim[m], stdout);
meillo@14: 				for (lq = line; lq < lp; lq++)
meillo@14: 					putc(*lq, stdout);
meillo@14: 				gotcha = 1;
meillo@14: 			}
meillo@14: 			if (wc == '\n') {
meillo@14: 				if (gotcha)
meillo@14: 					putc('\n', stdout);
meillo@14: 				i = 1;
meillo@14: 				gotcha = 0;
meillo@14: 			} else
meillo@14: 				i++;
meillo@14: 			lp = line;
meillo@14: 		} else {
meillo@14: 			for (m = 0; m < n; m++) {
meillo@14: 				if (lp >= &line[linesize]) {
meillo@14: 					size_t	diff = lp - line;
meillo@14: 					line = lrealloc(line, linesize += incr);
meillo@14: 					lp = &line[diff];
meillo@14: 				}
meillo@14: 				*lp++ = cp[m];
meillo@14: 			}
meillo@14: 		}
meillo@14: 	} while (cp != NULL);
meillo@14: }
meillo@14: 
meillo@14: static int
meillo@14: fdcut(int fd)
meillo@14: {
meillo@14: 	struct iblok	*ip;
meillo@14: 
meillo@14: 	ip = ib_alloc(fd, 0);
meillo@14: 	switch (method) {
meillo@14: 	case 'b':
meillo@14: 		if (nflag && multibyte)
meillo@14: 			cutbn(ip);
meillo@14: 		else
meillo@14: 			cutb(ip);
meillo@14: 		break;
meillo@14: 	case 'c':
meillo@14: 		if (multibyte)
meillo@14: 			cutc(ip);
meillo@14: 		else
meillo@14: 			cutb(ip);
meillo@14: 		break;
meillo@14: 	case 'f':
meillo@14: 		cutf(ip);
meillo@14: 		break;
meillo@14: 	}
meillo@14: 	ib_free(ip);
meillo@14: 	return 0;
meillo@14: }
meillo@14: 
meillo@14: static int
meillo@14: fncut(const char *fn)
meillo@14: {
meillo@14: 	int	fd, res;
meillo@14: 
meillo@14: 	if (fn[0] == '-' && fn[1] == '\0')
meillo@14: 		fd = 0;
meillo@14: 	else if ((fd = open(fn, O_RDONLY)) < 0) {
meillo@14: 		fprintf(stderr, "%s: WARNING: cannot open %s\n", progname, fn);
meillo@14: 		return 1;
meillo@14: 	}
meillo@14: 	res = fdcut(fd);
meillo@14: 	if (fd)
meillo@14: 		close(fd);
meillo@14: 	return res;
meillo@14: }
meillo@14: 
meillo@14: int
meillo@14: main(int argc, char **argv)
meillo@14: {
meillo@14: 	const char	optstring[] = "b:c:d:f:ns";
meillo@14: 	int	i;
meillo@14: 
meillo@14: 	progname = basename(argv[0]);
meillo@14: 	setlocale(LC_CTYPE, "");
meillo@14: 	multibyte = MB_CUR_MAX > 1;
meillo@14: #ifdef	__GLIBC__
meillo@14: 	putenv("POSIXLY_CORRECT=1");
meillo@14: #endif
meillo@14: 	while ((i = getopt(argc, argv, optstring)) != EOF) {
meillo@14: 		switch (i) {
meillo@14: 		case 'b':
meillo@14: 		case 'c':
meillo@14: 		case 'f':
meillo@14: 			if (method && method != i)
meillo@14: 				usage();
meillo@14: 			method = i;
meillo@14: 			setlist(optarg);
meillo@14: 			break;
meillo@14: 		case 'd':
meillo@14: 			setdelim(optarg);
meillo@14: 			break;
meillo@14: 		case 'n':
meillo@14: 			nflag = 1;
meillo@14: 			break;
meillo@14: 		case 's':
meillo@14: 			sflag = 1;
meillo@14: 			break;
meillo@14: 		default:
meillo@14: 			usage();
meillo@14: 		}
meillo@14: 	}
meillo@14: 	/*if ((sflag && method != 'f') || (nflag && method != 'b'))
meillo@14: 		usage();*/
meillo@14: 	if (method == 0)
meillo@14: 		badlist();
meillo@14: 	if (argv[optind]) {
meillo@14: 		for (i = optind; argv[i]; i++)
meillo@14: 			errcnt |= fncut(argv[i]);
meillo@14: 	} else
meillo@14: 		errcnt |= fdcut(0);
meillo@14: 	return errcnt;
meillo@14: }