meillo@14: /*
meillo@14:  * Copyright (c) 1989, 1993
meillo@14:  *	The Regents of the University of California.  All rights reserved.
meillo@14:  *
meillo@14:  * This code is derived from software contributed to Berkeley by
meillo@14:  * Adam S. Moskowitz of Menlo Consulting and Marciano Pitargue.
meillo@14:  *
meillo@14:  * Redistribution and use in source and binary forms, with or without
meillo@14:  * modification, are permitted provided that the following conditions
meillo@14:  * are met:
meillo@14:  * 1. Redistributions of source code must retain the above copyright
meillo@14:  *    notice, this list of conditions and the following disclaimer.
meillo@14:  * 2. Redistributions in binary form must reproduce the above copyright
meillo@14:  *    notice, this list of conditions and the following disclaimer in the
meillo@14:  *    documentation and/or other materials provided with the distribution.
meillo@14:  * 4. Neither the name of the University nor the names of its contributors
meillo@14:  *    may be used to endorse or promote products derived from this software
meillo@14:  *    without specific prior written permission.
meillo@14:  *
meillo@14:  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
meillo@14:  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
meillo@14:  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
meillo@14:  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
meillo@14:  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
meillo@14:  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
meillo@14:  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
meillo@14:  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
meillo@14:  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
meillo@14:  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
meillo@14:  * SUCH DAMAGE.
meillo@14:  */
meillo@14: 
meillo@14: #ifndef lint
meillo@14: static const char copyright[] =
meillo@14: "@(#) Copyright (c) 1989, 1993\n\
meillo@14: 	The Regents of the University of California.  All rights reserved.\n";
meillo@14: static const char sccsid[] = "@(#)cut.c	8.3 (Berkeley) 5/4/95";
meillo@14: #endif /* not lint */
meillo@14: #include <sys/cdefs.h>
meillo@14: __FBSDID("$FreeBSD$");
meillo@14: 
meillo@14: #include <ctype.h>
meillo@14: #include <err.h>
meillo@14: #include <errno.h>
meillo@14: #include <limits.h>
meillo@14: #include <locale.h>
meillo@14: #include <stdio.h>
meillo@14: #include <stdlib.h>
meillo@14: #include <string.h>
meillo@14: #include <unistd.h>
meillo@14: #include <wchar.h>
meillo@14: 
meillo@14: static int	bflag;
meillo@14: static int	cflag;
meillo@14: static wchar_t	dchar;
meillo@14: static char	dcharmb[MB_LEN_MAX + 1];
meillo@14: static int	dflag;
meillo@14: static int	fflag;
meillo@14: static int	nflag;
meillo@14: static int	sflag;
meillo@14: static int	wflag;
meillo@14: 
meillo@14: static size_t	autostart, autostop, maxval;
meillo@14: static char *	positions;
meillo@14: 
meillo@14: static int	b_cut(FILE *, const char *);
meillo@14: static int	b_n_cut(FILE *, const char *);
meillo@14: static int	c_cut(FILE *, const char *);
meillo@14: static int	f_cut(FILE *, const char *);
meillo@14: static void	get_list(char *);
meillo@14: static int	is_delim(wchar_t);
meillo@14: static void	needpos(size_t);
meillo@14: static void	usage(void);
meillo@14: 
meillo@14: int
meillo@14: main(int argc, char *argv[])
meillo@14: {
meillo@14: 	FILE *fp;
meillo@14: 	int (*fcn)(FILE *, const char *);
meillo@14: 	int ch, rval;
meillo@14: 	size_t n;
meillo@14: 
meillo@14: 	setlocale(LC_ALL, "");
meillo@14: 
meillo@14: 	fcn = NULL;
meillo@14: 	dchar = '\t';			/* default delimiter is \t */
meillo@14: 	strcpy(dcharmb, "\t");
meillo@14: 
meillo@14: 	while ((ch = getopt(argc, argv, "b:c:d:f:snw")) != -1)
meillo@14: 		switch(ch) {
meillo@14: 		case 'b':
meillo@14: 			get_list(optarg);
meillo@14: 			bflag = 1;
meillo@14: 			break;
meillo@14: 		case 'c':
meillo@14: 			get_list(optarg);
meillo@14: 			cflag = 1;
meillo@14: 			break;
meillo@14: 		case 'd':
meillo@14: 			n = mbrtowc(&dchar, optarg, MB_LEN_MAX, NULL);
meillo@14: 			if (dchar == '\0' || n != strlen(optarg))
meillo@14: 				errx(1, "bad delimiter");
meillo@14: 			strcpy(dcharmb, optarg);
meillo@14: 			dflag = 1;
meillo@14: 			break;
meillo@14: 		case 'f':
meillo@14: 			get_list(optarg);
meillo@14: 			fflag = 1;
meillo@14: 			break;
meillo@14: 		case 's':
meillo@14: 			sflag = 1;
meillo@14: 			break;
meillo@14: 		case 'n':
meillo@14: 			nflag = 1;
meillo@14: 			break;
meillo@14: 		case 'w':
meillo@14: 			wflag = 1;
meillo@14: 			break;
meillo@14: 		case '?':
meillo@14: 		default:
meillo@14: 			usage();
meillo@14: 		}
meillo@14: 	argc -= optind;
meillo@14: 	argv += optind;
meillo@14: 
meillo@14: 	if (fflag) {
meillo@14: 		if (bflag || cflag || nflag || (wflag && dflag))
meillo@14: 			usage();
meillo@14: 	} else if (!(bflag || cflag) || dflag || sflag || wflag)
meillo@14: 		usage();
meillo@14: 	else if (!bflag && nflag)
meillo@14: 		usage();
meillo@14: 
meillo@14: 	if (fflag)
meillo@14: 		fcn = f_cut;
meillo@14: 	else if (cflag)
meillo@14: 		fcn = MB_CUR_MAX > 1 ? c_cut : b_cut;
meillo@14: 	else if (bflag)
meillo@14: 		fcn = nflag && MB_CUR_MAX > 1 ? b_n_cut : b_cut;
meillo@14: 
meillo@14: 	rval = 0;
meillo@14: 	if (*argv)
meillo@14: 		for (; *argv; ++argv) {
meillo@14: 			if (strcmp(*argv, "-") == 0)
meillo@14: 				rval |= fcn(stdin, "stdin");
meillo@14: 			else {
meillo@14: 				if (!(fp = fopen(*argv, "r"))) {
meillo@14: 					warn("%s", *argv);
meillo@14: 					rval = 1;
meillo@14: 					continue;
meillo@14: 				}
meillo@14: 				fcn(fp, *argv);
meillo@14: 				(void)fclose(fp);
meillo@14: 			}
meillo@14: 		}
meillo@14: 	else
meillo@14: 		rval = fcn(stdin, "stdin");
meillo@14: 	exit(rval);
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: get_list(char *list)
meillo@14: {
meillo@14: 	size_t setautostart, start, stop;
meillo@14: 	char *pos;
meillo@14: 	char *p;
meillo@14: 
meillo@14: 	/*
meillo@14: 	 * set a byte in the positions array to indicate if a field or
meillo@14: 	 * column is to be selected; use +1, it's 1-based, not 0-based.
meillo@14: 	 * Numbers and number ranges may be overlapping, repeated, and in
meillo@14: 	 * any order. We handle "-3-5" although there's no real reason to.
meillo@14: 	 */
meillo@14: 	for (; (p = strsep(&list, ", \t")) != NULL;) {
meillo@14: 		setautostart = start = stop = 0;
meillo@14: 		if (*p == '-') {
meillo@14: 			++p;
meillo@14: 			setautostart = 1;
meillo@14: 		}
meillo@14: 		if (isdigit((unsigned char)*p)) {
meillo@14: 			start = stop = strtol(p, &p, 10);
meillo@14: 			if (setautostart && start > autostart)
meillo@14: 				autostart = start;
meillo@14: 		}
meillo@14: 		if (*p == '-') {
meillo@14: 			if (isdigit((unsigned char)p[1]))
meillo@14: 				stop = strtol(p + 1, &p, 10);
meillo@14: 			if (*p == '-') {
meillo@14: 				++p;
meillo@14: 				if (!autostop || autostop > stop)
meillo@14: 					autostop = stop;
meillo@14: 			}
meillo@14: 		}
meillo@14: 		if (*p)
meillo@14: 			errx(1, "[-bcf] list: illegal list value");
meillo@14: 		if (!stop || !start)
meillo@14: 			errx(1, "[-bcf] list: values may not include zero");
meillo@14: 		if (maxval < stop) {
meillo@14: 			maxval = stop;
meillo@14: 			needpos(maxval + 1);
meillo@14: 		}
meillo@14: 		for (pos = positions + start; start++ <= stop; *pos++ = 1);
meillo@14: 	}
meillo@14: 
meillo@14: 	/* overlapping ranges */
meillo@14: 	if (autostop && maxval > autostop) {
meillo@14: 		maxval = autostop;
meillo@14: 		needpos(maxval + 1);
meillo@14: 	}
meillo@14: 
meillo@14: 	/* set autostart */
meillo@14: 	if (autostart)
meillo@14: 		memset(positions + 1, '1', autostart);
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: needpos(size_t n)
meillo@14: {
meillo@14: 	static size_t npos;
meillo@14: 	size_t oldnpos;
meillo@14: 
meillo@14: 	/* Grow the positions array to at least the specified size. */
meillo@14: 	if (n > npos) {
meillo@14: 		oldnpos = npos;
meillo@14: 		if (npos == 0)
meillo@14: 			npos = n;
meillo@14: 		while (n > npos)
meillo@14: 			npos *= 2;
meillo@14: 		if ((positions = realloc(positions, npos)) == NULL)
meillo@14: 			err(1, "realloc");
meillo@14: 		memset((char *)positions + oldnpos, 0, npos - oldnpos);
meillo@14: 	}
meillo@14: }
meillo@14: 
meillo@14: static int
meillo@14: b_cut(FILE *fp, const char *fname __unused)
meillo@14: {
meillo@14: 	int ch, col;
meillo@14: 	char *pos;
meillo@14: 
meillo@14: 	ch = 0;
meillo@14: 	for (;;) {
meillo@14: 		pos = positions + 1;
meillo@14: 		for (col = maxval; col; --col) {
meillo@14: 			if ((ch = getc(fp)) == EOF)
meillo@14: 				return (0);
meillo@14: 			if (ch == '\n')
meillo@14: 				break;
meillo@14: 			if (*pos++)
meillo@14: 				(void)putchar(ch);
meillo@14: 		}
meillo@14: 		if (ch != '\n') {
meillo@14: 			if (autostop)
meillo@14: 				while ((ch = getc(fp)) != EOF && ch != '\n')
meillo@14: 					(void)putchar(ch);
meillo@14: 			else
meillo@14: 				while ((ch = getc(fp)) != EOF && ch != '\n');
meillo@14: 		}
meillo@14: 		(void)putchar('\n');
meillo@14: 	}
meillo@14: 	return (0);
meillo@14: }
meillo@14: 
meillo@14: /*
meillo@14:  * Cut based on byte positions, taking care not to split multibyte characters.
meillo@14:  * Although this function also handles the case where -n is not specified,
meillo@14:  * b_cut() ought to be much faster.
meillo@14:  */
meillo@14: static int
meillo@14: b_n_cut(FILE *fp, const char *fname)
meillo@14: {
meillo@14: 	size_t col, i, lbuflen;
meillo@14: 	char *lbuf;
meillo@14: 	int canwrite, clen, warned;
meillo@14: 	mbstate_t mbs;
meillo@14: 
meillo@14: 	memset(&mbs, 0, sizeof(mbs));
meillo@14: 	warned = 0;
meillo@14: 	while ((lbuf = fgetln(fp, &lbuflen)) != NULL) {
meillo@14: 		for (col = 0; lbuflen > 0; col += clen) {
meillo@14: 			if ((clen = mbrlen(lbuf, lbuflen, &mbs)) < 0) {
meillo@14: 				if (!warned) {
meillo@14: 					warn("%s", fname);
meillo@14: 					warned = 1;
meillo@14: 				}
meillo@14: 				memset(&mbs, 0, sizeof(mbs));
meillo@14: 				clen = 1;
meillo@14: 			}
meillo@14: 			if (clen == 0 || *lbuf == '\n')
meillo@14: 				break;
meillo@14: 			if (col < maxval && !positions[1 + col]) {
meillo@14: 				/*
meillo@14: 				 * Print the character if (1) after an initial
meillo@14: 				 * segment of un-selected bytes, the rest of
meillo@14: 				 * it is selected, and (2) the last byte is
meillo@14: 				 * selected.
meillo@14: 				 */
meillo@14: 				i = col;
meillo@14: 				while (i < col + clen && i < maxval &&
meillo@14: 				    !positions[1 + i])
meillo@14: 					i++;
meillo@14: 				canwrite = i < col + clen;
meillo@14: 				for (; i < col + clen && i < maxval; i++)
meillo@14: 					canwrite &= positions[1 + i];
meillo@14: 				if (canwrite)
meillo@14: 					fwrite(lbuf, 1, clen, stdout);
meillo@14: 			} else {
meillo@14: 				/*
meillo@14: 				 * Print the character if all of it has
meillo@14: 				 * been selected.
meillo@14: 				 */
meillo@14: 				canwrite = 1;
meillo@14: 				for (i = col; i < col + clen; i++)
meillo@14: 					if ((i >= maxval && !autostop) ||
meillo@14: 					    (i < maxval && !positions[1 + i])) {
meillo@14: 						canwrite = 0;
meillo@14: 						break;
meillo@14: 					}
meillo@14: 				if (canwrite)
meillo@14: 					fwrite(lbuf, 1, clen, stdout);
meillo@14: 			}
meillo@14: 			lbuf += clen;
meillo@14: 			lbuflen -= clen;
meillo@14: 		}
meillo@14: 		if (lbuflen > 0)
meillo@14: 			putchar('\n');
meillo@14: 	}
meillo@14: 	return (warned);
meillo@14: }
meillo@14: 
meillo@14: static int
meillo@14: c_cut(FILE *fp, const char *fname)
meillo@14: {
meillo@14: 	wint_t ch;
meillo@14: 	int col;
meillo@14: 	char *pos;
meillo@14: 
meillo@14: 	ch = 0;
meillo@14: 	for (;;) {
meillo@14: 		pos = positions + 1;
meillo@14: 		for (col = maxval; col; --col) {
meillo@14: 			if ((ch = getwc(fp)) == WEOF)
meillo@14: 				goto out;
meillo@14: 			if (ch == '\n')
meillo@14: 				break;
meillo@14: 			if (*pos++)
meillo@14: 				(void)putwchar(ch);
meillo@14: 		}
meillo@14: 		if (ch != '\n') {
meillo@14: 			if (autostop)
meillo@14: 				while ((ch = getwc(fp)) != WEOF && ch != '\n')
meillo@14: 					(void)putwchar(ch);
meillo@14: 			else
meillo@14: 				while ((ch = getwc(fp)) != WEOF && ch != '\n');
meillo@14: 		}
meillo@14: 		(void)putwchar('\n');
meillo@14: 	}
meillo@14: out:
meillo@14: 	if (ferror(fp)) {
meillo@14: 		warn("%s", fname);
meillo@14: 		return (1);
meillo@14: 	}
meillo@14: 	return (0);
meillo@14: }
meillo@14: 
meillo@14: static int
meillo@14: is_delim(wchar_t ch)
meillo@14: {
meillo@14: 	if (wflag) {
meillo@14: 		if (ch == ' ' || ch == '\t')
meillo@14: 			return 1;
meillo@14: 	} else {
meillo@14: 		if (ch == dchar)
meillo@14: 			return 1;
meillo@14: 	}
meillo@14: 	return 0;
meillo@14: }
meillo@14: 
meillo@14: static int
meillo@14: f_cut(FILE *fp, const char *fname)
meillo@14: {
meillo@14: 	wchar_t ch;
meillo@14: 	int field, i, isdelim;
meillo@14: 	char *pos, *p;
meillo@14: 	int output;
meillo@14: 	char *lbuf, *mlbuf;
meillo@14: 	size_t clen, lbuflen, reallen;
meillo@14: 
meillo@14: 	mlbuf = NULL;
meillo@14: 	while ((lbuf = fgetln(fp, &lbuflen)) != NULL) {
meillo@14: 		reallen = lbuflen;
meillo@14: 		/* Assert EOL has a newline. */
meillo@14: 		if (*(lbuf + lbuflen - 1) != '\n') {
meillo@14: 			/* Can't have > 1 line with no trailing newline. */
meillo@14: 			mlbuf = malloc(lbuflen + 1);
meillo@14: 			if (mlbuf == NULL)
meillo@14: 				err(1, "malloc");
meillo@14: 			memcpy(mlbuf, lbuf, lbuflen);
meillo@14: 			*(mlbuf + lbuflen) = '\n';
meillo@14: 			lbuf = mlbuf;
meillo@14: 			reallen++;
meillo@14: 		}
meillo@14: 		output = 0;
meillo@14: 		for (isdelim = 0, p = lbuf;; p += clen) {
meillo@14: 			clen = mbrtowc(&ch, p, lbuf + reallen - p, NULL);
meillo@14: 			if (clen == (size_t)-1 || clen == (size_t)-2) {
meillo@14: 				warnc(EILSEQ, "%s", fname);
meillo@14: 				free(mlbuf);
meillo@14: 				return (1);
meillo@14: 			}
meillo@14: 			if (clen == 0)
meillo@14: 				clen = 1;
meillo@14: 			/* this should work if newline is delimiter */
meillo@14: 			if (is_delim(ch))
meillo@14: 				isdelim = 1;
meillo@14: 			if (ch == '\n') {
meillo@14: 				if (!isdelim && !sflag)
meillo@14: 					(void)fwrite(lbuf, lbuflen, 1, stdout);
meillo@14: 				break;
meillo@14: 			}
meillo@14: 		}
meillo@14: 		if (!isdelim)
meillo@14: 			continue;
meillo@14: 
meillo@14: 		pos = positions + 1;
meillo@14: 		for (field = maxval, p = lbuf; field; --field, ++pos) {
meillo@14: 			if (*pos && output++)
meillo@14: 				for (i = 0; dcharmb[i] != '\0'; i++)
meillo@14: 					putchar(dcharmb[i]);
meillo@14: 			for (;;) {
meillo@14: 				clen = mbrtowc(&ch, p, lbuf + reallen - p,
meillo@14: 				    NULL);
meillo@14: 				if (clen == (size_t)-1 || clen == (size_t)-2) {
meillo@14: 					warnc(EILSEQ, "%s", fname);
meillo@14: 					free(mlbuf);
meillo@14: 					return (1);
meillo@14: 				}
meillo@14: 				if (clen == 0)
meillo@14: 					clen = 1;
meillo@14: 				p += clen;
meillo@14: 				if (ch == '\n' || is_delim(ch)) {
meillo@14: 					/* compress whitespace */
meillo@14: 					if (wflag && ch != '\n')
meillo@14: 						while (is_delim(*p))
meillo@14: 							p++;
meillo@14: 					break;
meillo@14: 				}
meillo@14: 				if (*pos)
meillo@14: 					for (i = 0; i < (int)clen; i++)
meillo@14: 						putchar(p[i - clen]);
meillo@14: 			}
meillo@14: 			if (ch == '\n')
meillo@14: 				break;
meillo@14: 		}
meillo@14: 		if (ch != '\n') {
meillo@14: 			if (autostop) {
meillo@14: 				if (output)
meillo@14: 					for (i = 0; dcharmb[i] != '\0'; i++)
meillo@14: 						putchar(dcharmb[i]);
meillo@14: 				for (; (ch = *p) != '\n'; ++p)
meillo@14: 					(void)putchar(ch);
meillo@14: 			} else
meillo@14: 				for (; (ch = *p) != '\n'; ++p);
meillo@14: 		}
meillo@14: 		(void)putchar('\n');
meillo@14: 	}
meillo@14: 	free(mlbuf);
meillo@14: 	return (0);
meillo@14: }
meillo@14: 
meillo@14: static void
meillo@14: usage(void)
meillo@14: {
meillo@14: 	(void)fprintf(stderr, "%s\n%s\n%s\n",
meillo@14: 		"usage: cut -b list [-n] [file ...]",
meillo@14: 		"       cut -c list [file ...]",
meillo@14: 		"       cut -f list [-s] [-w | -d delim] [file ...]");
meillo@14: 	exit(1);
meillo@14: }