xref: /freebsd/usr.bin/cut/cut.c (revision b05785953e555e3dd13a162a8854076212026ca5)
19b50d902SRodney W. Grimes /*
28a16b7a1SPedro F. Giffuni  * SPDX-License-Identifier: BSD-3-Clause
38a16b7a1SPedro F. Giffuni  *
49b50d902SRodney W. Grimes  * Copyright (c) 1989, 1993
59b50d902SRodney W. Grimes  *	The Regents of the University of California.  All rights reserved.
69b50d902SRodney W. Grimes  *
79b50d902SRodney W. Grimes  * This code is derived from software contributed to Berkeley by
89b50d902SRodney W. Grimes  * Adam S. Moskowitz of Menlo Consulting and Marciano Pitargue.
99b50d902SRodney W. Grimes  *
109b50d902SRodney W. Grimes  * Redistribution and use in source and binary forms, with or without
119b50d902SRodney W. Grimes  * modification, are permitted provided that the following conditions
129b50d902SRodney W. Grimes  * are met:
139b50d902SRodney W. Grimes  * 1. Redistributions of source code must retain the above copyright
149b50d902SRodney W. Grimes  *    notice, this list of conditions and the following disclaimer.
159b50d902SRodney W. Grimes  * 2. Redistributions in binary form must reproduce the above copyright
169b50d902SRodney W. Grimes  *    notice, this list of conditions and the following disclaimer in the
179b50d902SRodney W. Grimes  *    documentation and/or other materials provided with the distribution.
18fbbd9655SWarner Losh  * 3. Neither the name of the University nor the names of its contributors
199b50d902SRodney W. Grimes  *    may be used to endorse or promote products derived from this software
209b50d902SRodney W. Grimes  *    without specific prior written permission.
219b50d902SRodney W. Grimes  *
229b50d902SRodney W. Grimes  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
239b50d902SRodney W. Grimes  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
249b50d902SRodney W. Grimes  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
259b50d902SRodney W. Grimes  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
269b50d902SRodney W. Grimes  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
279b50d902SRodney W. Grimes  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
289b50d902SRodney W. Grimes  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
299b50d902SRodney W. Grimes  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
309b50d902SRodney W. Grimes  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
319b50d902SRodney W. Grimes  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
329b50d902SRodney W. Grimes  * SUCH DAMAGE.
339b50d902SRodney W. Grimes  */
349b50d902SRodney W. Grimes 
359b50d902SRodney W. Grimes #include <ctype.h>
36812bff99SPhilippe Charnier #include <err.h>
37a5c4bafcSTim J. Robbins #include <errno.h>
389b50d902SRodney W. Grimes #include <limits.h>
39d51c6625SEivind Eklund #include <locale.h>
409b50d902SRodney W. Grimes #include <stdio.h>
419b50d902SRodney W. Grimes #include <stdlib.h>
429b50d902SRodney W. Grimes #include <string.h>
43df3f5d9dSPeter Wemm #include <unistd.h>
44d900c384STim J. Robbins #include <wchar.h>
459b50d902SRodney W. Grimes 
4641b662c5SEd Schouten static int	bflag;
4741b662c5SEd Schouten static int	cflag;
4841b662c5SEd Schouten static wchar_t	dchar;
4941b662c5SEd Schouten static char	dcharmb[MB_LEN_MAX + 1];
5041b662c5SEd Schouten static int	dflag;
5141b662c5SEd Schouten static int	fflag;
5241b662c5SEd Schouten static int	nflag;
5341b662c5SEd Schouten static int	sflag;
5418ba28c8SEitan Adler static int	wflag;
559b50d902SRodney W. Grimes 
5641b662c5SEd Schouten static size_t	autostart, autostop, maxval;
5741b662c5SEd Schouten static char *	positions;
583b90bf79STim J. Robbins 
5941b662c5SEd Schouten static int	b_cut(FILE *, const char *);
6041b662c5SEd Schouten static int	b_n_cut(FILE *, const char *);
6141b662c5SEd Schouten static int	c_cut(FILE *, const char *);
6241b662c5SEd Schouten static int	f_cut(FILE *, const char *);
6341b662c5SEd Schouten static void	get_list(char *);
64bd9dd0c6SAndrew Turner static int	is_delim(wchar_t);
6541b662c5SEd Schouten static void	needpos(size_t);
6699557a79SWill Andrews static void	usage(void);
679b50d902SRodney W. Grimes 
689b50d902SRodney W. Grimes int
69f4ac32deSDavid Malone main(int argc, char *argv[])
709b50d902SRodney W. Grimes {
719b50d902SRodney W. Grimes 	FILE *fp;
72364d0a91STim J. Robbins 	int (*fcn)(FILE *, const char *);
73a6ea32c3STim J. Robbins 	int ch, rval;
74a5c4bafcSTim J. Robbins 	size_t n;
759b50d902SRodney W. Grimes 
76d51c6625SEivind Eklund 	setlocale(LC_ALL, "");
77d51c6625SEivind Eklund 
7893738f50STim J. Robbins 	fcn = NULL;
799b50d902SRodney W. Grimes 	dchar = '\t';			/* default delimiter is \t */
80a5c4bafcSTim J. Robbins 	strcpy(dcharmb, "\t");
819b50d902SRodney W. Grimes 
8218ba28c8SEitan Adler 	while ((ch = getopt(argc, argv, "b:c:d:f:snw")) != -1)
839b50d902SRodney W. Grimes 		switch(ch) {
84d51c6625SEivind Eklund 		case 'b':
85393cf508STim J. Robbins 			get_list(optarg);
86393cf508STim J. Robbins 			bflag = 1;
87393cf508STim J. Robbins 			break;
889b50d902SRodney W. Grimes 		case 'c':
899b50d902SRodney W. Grimes 			get_list(optarg);
909b50d902SRodney W. Grimes 			cflag = 1;
919b50d902SRodney W. Grimes 			break;
929b50d902SRodney W. Grimes 		case 'd':
93a5c4bafcSTim J. Robbins 			n = mbrtowc(&dchar, optarg, MB_LEN_MAX, NULL);
94a5c4bafcSTim J. Robbins 			if (dchar == '\0' || n != strlen(optarg))
95a5c4bafcSTim J. Robbins 				errx(1, "bad delimiter");
96a5c4bafcSTim J. Robbins 			strcpy(dcharmb, optarg);
979b50d902SRodney W. Grimes 			dflag = 1;
989b50d902SRodney W. Grimes 			break;
999b50d902SRodney W. Grimes 		case 'f':
1009b50d902SRodney W. Grimes 			get_list(optarg);
1019b50d902SRodney W. Grimes 			fflag = 1;
1029b50d902SRodney W. Grimes 			break;
1039b50d902SRodney W. Grimes 		case 's':
1049b50d902SRodney W. Grimes 			sflag = 1;
1059b50d902SRodney W. Grimes 			break;
106d51c6625SEivind Eklund 		case 'n':
107393cf508STim J. Robbins 			nflag = 1;
108d51c6625SEivind Eklund 			break;
10918ba28c8SEitan Adler 		case 'w':
11018ba28c8SEitan Adler 			wflag = 1;
11118ba28c8SEitan Adler 			break;
1129b50d902SRodney W. Grimes 		case '?':
1139b50d902SRodney W. Grimes 		default:
1149b50d902SRodney W. Grimes 			usage();
1159b50d902SRodney W. Grimes 		}
1169b50d902SRodney W. Grimes 	argc -= optind;
1179b50d902SRodney W. Grimes 	argv += optind;
1189b50d902SRodney W. Grimes 
1199b50d902SRodney W. Grimes 	if (fflag) {
12018ba28c8SEitan Adler 		if (bflag || cflag || nflag || (wflag && dflag))
1219b50d902SRodney W. Grimes 			usage();
12218ba28c8SEitan Adler 	} else if (!(bflag || cflag) || dflag || sflag || wflag)
1239b50d902SRodney W. Grimes 		usage();
124393cf508STim J. Robbins 	else if (!bflag && nflag)
125393cf508STim J. Robbins 		usage();
126393cf508STim J. Robbins 
127364d0a91STim J. Robbins 	if (fflag)
128364d0a91STim J. Robbins 		fcn = f_cut;
129364d0a91STim J. Robbins 	else if (cflag)
130364d0a91STim J. Robbins 		fcn = MB_CUR_MAX > 1 ? c_cut : b_cut;
131364d0a91STim J. Robbins 	else if (bflag)
132364d0a91STim J. Robbins 		fcn = nflag && MB_CUR_MAX > 1 ? b_n_cut : b_cut;
1339b50d902SRodney W. Grimes 
134a6ea32c3STim J. Robbins 	rval = 0;
1359b50d902SRodney W. Grimes 	if (*argv)
1369b50d902SRodney W. Grimes 		for (; *argv; ++argv) {
137204c78a1STim J. Robbins 			if (strcmp(*argv, "-") == 0)
138364d0a91STim J. Robbins 				rval |= fcn(stdin, "stdin");
139204c78a1STim J. Robbins 			else {
140a6ea32c3STim J. Robbins 				if (!(fp = fopen(*argv, "r"))) {
141a6ea32c3STim J. Robbins 					warn("%s", *argv);
142a6ea32c3STim J. Robbins 					rval = 1;
143a6ea32c3STim J. Robbins 					continue;
144a6ea32c3STim J. Robbins 				}
1459b50d902SRodney W. Grimes 				fcn(fp, *argv);
1469b50d902SRodney W. Grimes 				(void)fclose(fp);
1479b50d902SRodney W. Grimes 			}
148204c78a1STim J. Robbins 		}
1499b50d902SRodney W. Grimes 	else
150364d0a91STim J. Robbins 		rval = fcn(stdin, "stdin");
151a6ea32c3STim J. Robbins 	exit(rval);
1529b50d902SRodney W. Grimes }
1539b50d902SRodney W. Grimes 
15441b662c5SEd Schouten static void
155f4ac32deSDavid Malone get_list(char *list)
1569b50d902SRodney W. Grimes {
157d8a3fbd5SWill Andrews 	size_t setautostart, start, stop;
1582c39ae65SEivind Eklund 	char *pos;
1599b50d902SRodney W. Grimes 	char *p;
1609b50d902SRodney W. Grimes 
1619b50d902SRodney W. Grimes 	/*
1629b50d902SRodney W. Grimes 	 * set a byte in the positions array to indicate if a field or
1639b50d902SRodney W. Grimes 	 * column is to be selected; use +1, it's 1-based, not 0-based.
1640dcb7b75STim J. Robbins 	 * Numbers and number ranges may be overlapping, repeated, and in
165d334b286SKevin Lo 	 * any order. We handle "-3-5" although there's no real reason to.
1669b50d902SRodney W. Grimes 	 */
1672c39ae65SEivind Eklund 	for (; (p = strsep(&list, ", \t")) != NULL;) {
1689b50d902SRodney W. Grimes 		setautostart = start = stop = 0;
1699b50d902SRodney W. Grimes 		if (*p == '-') {
1709b50d902SRodney W. Grimes 			++p;
1719b50d902SRodney W. Grimes 			setautostart = 1;
1729b50d902SRodney W. Grimes 		}
1732c39ae65SEivind Eklund 		if (isdigit((unsigned char)*p)) {
1749b50d902SRodney W. Grimes 			start = stop = strtol(p, &p, 10);
1759b50d902SRodney W. Grimes 			if (setautostart && start > autostart)
1769b50d902SRodney W. Grimes 				autostart = start;
1779b50d902SRodney W. Grimes 		}
1789b50d902SRodney W. Grimes 		if (*p == '-') {
1792c39ae65SEivind Eklund 			if (isdigit((unsigned char)p[1]))
1809b50d902SRodney W. Grimes 				stop = strtol(p + 1, &p, 10);
1819b50d902SRodney W. Grimes 			if (*p == '-') {
1829b50d902SRodney W. Grimes 				++p;
1839b50d902SRodney W. Grimes 				if (!autostop || autostop > stop)
1849b50d902SRodney W. Grimes 					autostop = stop;
1859b50d902SRodney W. Grimes 			}
1869b50d902SRodney W. Grimes 		}
1879b50d902SRodney W. Grimes 		if (*p)
188d334b286SKevin Lo 			errx(1, "[-bcf] list: illegal list value");
1899b50d902SRodney W. Grimes 		if (!stop || !start)
190d334b286SKevin Lo 			errx(1, "[-bcf] list: values may not include zero");
191a8522a9bSTim J. Robbins 		if (maxval < stop) {
1929b50d902SRodney W. Grimes 			maxval = stop;
193a8522a9bSTim J. Robbins 			needpos(maxval + 1);
194a8522a9bSTim J. Robbins 		}
1959b50d902SRodney W. Grimes 		for (pos = positions + start; start++ <= stop; *pos++ = 1);
1969b50d902SRodney W. Grimes 	}
1979b50d902SRodney W. Grimes 
1989b50d902SRodney W. Grimes 	/* overlapping ranges */
199a8522a9bSTim J. Robbins 	if (autostop && maxval > autostop) {
2009b50d902SRodney W. Grimes 		maxval = autostop;
201a8522a9bSTim J. Robbins 		needpos(maxval + 1);
202a8522a9bSTim J. Robbins 	}
2039b50d902SRodney W. Grimes 
2046872fd3cSEitan Adler 	/* reversed range with autostart */
2056872fd3cSEitan Adler 	if (maxval < autostart) {
2066872fd3cSEitan Adler 		maxval = autostart;
2076872fd3cSEitan Adler 		needpos(maxval + 1);
2086872fd3cSEitan Adler 	}
2096872fd3cSEitan Adler 
2109b50d902SRodney W. Grimes 	/* set autostart */
2119b50d902SRodney W. Grimes 	if (autostart)
2129b50d902SRodney W. Grimes 		memset(positions + 1, '1', autostart);
2139b50d902SRodney W. Grimes }
2149b50d902SRodney W. Grimes 
21541b662c5SEd Schouten static void
216a8522a9bSTim J. Robbins needpos(size_t n)
217a8522a9bSTim J. Robbins {
218a8522a9bSTim J. Robbins 	static size_t npos;
219f457179aSTim J. Robbins 	size_t oldnpos;
220a8522a9bSTim J. Robbins 
221a8522a9bSTim J. Robbins 	/* Grow the positions array to at least the specified size. */
222a8522a9bSTim J. Robbins 	if (n > npos) {
223f457179aSTim J. Robbins 		oldnpos = npos;
224a8522a9bSTim J. Robbins 		if (npos == 0)
225a8522a9bSTim J. Robbins 			npos = n;
226a8522a9bSTim J. Robbins 		while (n > npos)
227a8522a9bSTim J. Robbins 			npos *= 2;
228a8522a9bSTim J. Robbins 		if ((positions = realloc(positions, npos)) == NULL)
229a8522a9bSTim J. Robbins 			err(1, "realloc");
230f457179aSTim J. Robbins 		memset((char *)positions + oldnpos, 0, npos - oldnpos);
231a8522a9bSTim J. Robbins 	}
232a8522a9bSTim J. Robbins }
233a8522a9bSTim J. Robbins 
23441b662c5SEd Schouten static int
235f0b4606fSTim J. Robbins b_cut(FILE *fp, const char *fname __unused)
236364d0a91STim J. Robbins {
237364d0a91STim J. Robbins 	int ch, col;
238364d0a91STim J. Robbins 	char *pos;
239364d0a91STim J. Robbins 
240364d0a91STim J. Robbins 	ch = 0;
241364d0a91STim J. Robbins 	for (;;) {
242364d0a91STim J. Robbins 		pos = positions + 1;
243364d0a91STim J. Robbins 		for (col = maxval; col; --col) {
244364d0a91STim J. Robbins 			if ((ch = getc(fp)) == EOF)
245364d0a91STim J. Robbins 				return (0);
246364d0a91STim J. Robbins 			if (ch == '\n')
247364d0a91STim J. Robbins 				break;
248364d0a91STim J. Robbins 			if (*pos++)
249364d0a91STim J. Robbins 				(void)putchar(ch);
250364d0a91STim J. Robbins 		}
251364d0a91STim J. Robbins 		if (ch != '\n') {
252364d0a91STim J. Robbins 			if (autostop)
253364d0a91STim J. Robbins 				while ((ch = getc(fp)) != EOF && ch != '\n')
254364d0a91STim J. Robbins 					(void)putchar(ch);
255364d0a91STim J. Robbins 			else
256364d0a91STim J. Robbins 				while ((ch = getc(fp)) != EOF && ch != '\n');
257364d0a91STim J. Robbins 		}
258364d0a91STim J. Robbins 		(void)putchar('\n');
259364d0a91STim J. Robbins 	}
260364d0a91STim J. Robbins 	return (0);
261364d0a91STim J. Robbins }
262364d0a91STim J. Robbins 
263393cf508STim J. Robbins /*
264393cf508STim J. Robbins  * Cut based on byte positions, taking care not to split multibyte characters.
265393cf508STim J. Robbins  * Although this function also handles the case where -n is not specified,
266364d0a91STim J. Robbins  * b_cut() ought to be much faster.
267393cf508STim J. Robbins  */
26841b662c5SEd Schouten static int
269f4ac32deSDavid Malone b_n_cut(FILE *fp, const char *fname)
270393cf508STim J. Robbins {
271*b0578595SMartin Tournoij 	size_t col, i, bufsize = 0;
272*b0578595SMartin Tournoij 	ssize_t lbuflen;
273*b0578595SMartin Tournoij 	char *lbuf = NULL;
274393cf508STim J. Robbins 	int canwrite, clen, warned;
275d900c384STim J. Robbins 	mbstate_t mbs;
276393cf508STim J. Robbins 
277d900c384STim J. Robbins 	memset(&mbs, 0, sizeof(mbs));
278393cf508STim J. Robbins 	warned = 0;
279*b0578595SMartin Tournoij 	while ((lbuflen = getline(&lbuf, &bufsize, fp)) >= 0) {
280393cf508STim J. Robbins 		for (col = 0; lbuflen > 0; col += clen) {
281d900c384STim J. Robbins 			if ((clen = mbrlen(lbuf, lbuflen, &mbs)) < 0) {
282393cf508STim J. Robbins 				if (!warned) {
283393cf508STim J. Robbins 					warn("%s", fname);
284393cf508STim J. Robbins 					warned = 1;
285393cf508STim J. Robbins 				}
286d900c384STim J. Robbins 				memset(&mbs, 0, sizeof(mbs));
287393cf508STim J. Robbins 				clen = 1;
288393cf508STim J. Robbins 			}
289393cf508STim J. Robbins 			if (clen == 0 || *lbuf == '\n')
290393cf508STim J. Robbins 				break;
291393cf508STim J. Robbins 			if (col < maxval && !positions[1 + col]) {
292393cf508STim J. Robbins 				/*
293393cf508STim J. Robbins 				 * Print the character if (1) after an initial
294393cf508STim J. Robbins 				 * segment of un-selected bytes, the rest of
295393cf508STim J. Robbins 				 * it is selected, and (2) the last byte is
296393cf508STim J. Robbins 				 * selected.
297393cf508STim J. Robbins 				 */
298393cf508STim J. Robbins 				i = col;
299393cf508STim J. Robbins 				while (i < col + clen && i < maxval &&
300393cf508STim J. Robbins 				    !positions[1 + i])
301393cf508STim J. Robbins 					i++;
302393cf508STim J. Robbins 				canwrite = i < col + clen;
303393cf508STim J. Robbins 				for (; i < col + clen && i < maxval; i++)
304393cf508STim J. Robbins 					canwrite &= positions[1 + i];
305393cf508STim J. Robbins 				if (canwrite)
306393cf508STim J. Robbins 					fwrite(lbuf, 1, clen, stdout);
307393cf508STim J. Robbins 			} else {
308393cf508STim J. Robbins 				/*
309393cf508STim J. Robbins 				 * Print the character if all of it has
310393cf508STim J. Robbins 				 * been selected.
311393cf508STim J. Robbins 				 */
312393cf508STim J. Robbins 				canwrite = 1;
313393cf508STim J. Robbins 				for (i = col; i < col + clen; i++)
314393cf508STim J. Robbins 					if ((i >= maxval && !autostop) ||
315393cf508STim J. Robbins 					    (i < maxval && !positions[1 + i])) {
316393cf508STim J. Robbins 						canwrite = 0;
317393cf508STim J. Robbins 						break;
318393cf508STim J. Robbins 					}
319393cf508STim J. Robbins 				if (canwrite)
320393cf508STim J. Robbins 					fwrite(lbuf, 1, clen, stdout);
321393cf508STim J. Robbins 			}
322393cf508STim J. Robbins 			lbuf += clen;
323393cf508STim J. Robbins 			lbuflen -= clen;
324393cf508STim J. Robbins 		}
325393cf508STim J. Robbins 		if (lbuflen > 0)
326393cf508STim J. Robbins 			putchar('\n');
327393cf508STim J. Robbins 	}
328*b0578595SMartin Tournoij 	free(lbuf);
329364d0a91STim J. Robbins 	return (warned);
330393cf508STim J. Robbins }
331393cf508STim J. Robbins 
33241b662c5SEd Schouten static int
333364d0a91STim J. Robbins c_cut(FILE *fp, const char *fname)
3349b50d902SRodney W. Grimes {
335364d0a91STim J. Robbins 	wint_t ch;
336364d0a91STim J. Robbins 	int col;
3372c39ae65SEivind Eklund 	char *pos;
3389b50d902SRodney W. Grimes 
3392c39ae65SEivind Eklund 	ch = 0;
3409b50d902SRodney W. Grimes 	for (;;) {
3419b50d902SRodney W. Grimes 		pos = positions + 1;
3429b50d902SRodney W. Grimes 		for (col = maxval; col; --col) {
343364d0a91STim J. Robbins 			if ((ch = getwc(fp)) == WEOF)
344364d0a91STim J. Robbins 				goto out;
3459b50d902SRodney W. Grimes 			if (ch == '\n')
3469b50d902SRodney W. Grimes 				break;
3479b50d902SRodney W. Grimes 			if (*pos++)
348364d0a91STim J. Robbins 				(void)putwchar(ch);
3499b50d902SRodney W. Grimes 		}
3502c39ae65SEivind Eklund 		if (ch != '\n') {
3519b50d902SRodney W. Grimes 			if (autostop)
352364d0a91STim J. Robbins 				while ((ch = getwc(fp)) != WEOF && ch != '\n')
353364d0a91STim J. Robbins 					(void)putwchar(ch);
3549b50d902SRodney W. Grimes 			else
355364d0a91STim J. Robbins 				while ((ch = getwc(fp)) != WEOF && ch != '\n');
3562c39ae65SEivind Eklund 		}
357364d0a91STim J. Robbins 		(void)putwchar('\n');
3589b50d902SRodney W. Grimes 	}
359364d0a91STim J. Robbins out:
360364d0a91STim J. Robbins 	if (ferror(fp)) {
361364d0a91STim J. Robbins 		warn("%s", fname);
362364d0a91STim J. Robbins 		return (1);
363364d0a91STim J. Robbins 	}
364364d0a91STim J. Robbins 	return (0);
3659b50d902SRodney W. Grimes }
3669b50d902SRodney W. Grimes 
36741b662c5SEd Schouten static int
368bd9dd0c6SAndrew Turner is_delim(wchar_t ch)
36918ba28c8SEitan Adler {
37018ba28c8SEitan Adler 	if (wflag) {
37118ba28c8SEitan Adler 		if (ch == ' ' || ch == '\t')
37218ba28c8SEitan Adler 			return 1;
37318ba28c8SEitan Adler 	} else {
37418ba28c8SEitan Adler 		if (ch == dchar)
37518ba28c8SEitan Adler 			return 1;
37618ba28c8SEitan Adler 	}
37718ba28c8SEitan Adler 	return 0;
37818ba28c8SEitan Adler }
37918ba28c8SEitan Adler 
38018ba28c8SEitan Adler static int
381a5c4bafcSTim J. Robbins f_cut(FILE *fp, const char *fname)
3829b50d902SRodney W. Grimes {
383a5c4bafcSTim J. Robbins 	wchar_t ch;
384a5c4bafcSTim J. Robbins 	int field, i, isdelim;
385a5c4bafcSTim J. Robbins 	char *pos, *p;
3869b50d902SRodney W. Grimes 	int output;
387*b0578595SMartin Tournoij 	char *lbuf = NULL;
388*b0578595SMartin Tournoij 	size_t clen, bufsize = 0, reallen;
389*b0578595SMartin Tournoij 	ssize_t lbuflen;
3909b50d902SRodney W. Grimes 
391*b0578595SMartin Tournoij 	while ((lbuflen = getline(&lbuf, &bufsize, fp)) >= 0) {
392a2222839STim J. Robbins 		reallen = lbuflen;
3931928e20eSDima Dorfman 		/* Assert EOL has a newline. */
394*b0578595SMartin Tournoij 		if (lbuflen > 0 && *(lbuf + lbuflen - 1) != '\n') {
3951928e20eSDima Dorfman 			/* Can't have > 1 line with no trailing newline. */
396*b0578595SMartin Tournoij 			if ((ssize_t)bufsize < (lbuflen + 1)) {
397*b0578595SMartin Tournoij 				bufsize = lbuflen + 1;
398*b0578595SMartin Tournoij 				lbuf = realloc(lbuf, bufsize);
399*b0578595SMartin Tournoij 			}
400*b0578595SMartin Tournoij 			if (lbuf == NULL)
401*b0578595SMartin Tournoij 				err(1, "realloc");
402*b0578595SMartin Tournoij 			lbuf[lbuflen] = '\n';
403a2222839STim J. Robbins 			reallen++;
4041928e20eSDima Dorfman 		}
405eaf92380SAndrey A. Chernov 		output = 0;
406a5c4bafcSTim J. Robbins 		for (isdelim = 0, p = lbuf;; p += clen) {
407a2222839STim J. Robbins 			clen = mbrtowc(&ch, p, lbuf + reallen - p, NULL);
408a5c4bafcSTim J. Robbins 			if (clen == (size_t)-1 || clen == (size_t)-2) {
409a5c4bafcSTim J. Robbins 				warnc(EILSEQ, "%s", fname);
410*b0578595SMartin Tournoij 				free(lbuf);
411a5c4bafcSTim J. Robbins 				return (1);
412a5c4bafcSTim J. Robbins 			}
413a5c4bafcSTim J. Robbins 			if (clen == 0)
414a5c4bafcSTim J. Robbins 				clen = 1;
4159b50d902SRodney W. Grimes 			/* this should work if newline is delimiter */
41618ba28c8SEitan Adler 			if (is_delim(ch))
4179b50d902SRodney W. Grimes 				isdelim = 1;
4189b50d902SRodney W. Grimes 			if (ch == '\n') {
4199b50d902SRodney W. Grimes 				if (!isdelim && !sflag)
4201928e20eSDima Dorfman 					(void)fwrite(lbuf, lbuflen, 1, stdout);
4219b50d902SRodney W. Grimes 				break;
4229b50d902SRodney W. Grimes 			}
4239b50d902SRodney W. Grimes 		}
4249b50d902SRodney W. Grimes 		if (!isdelim)
4259b50d902SRodney W. Grimes 			continue;
4269b50d902SRodney W. Grimes 
4279b50d902SRodney W. Grimes 		pos = positions + 1;
4289b50d902SRodney W. Grimes 		for (field = maxval, p = lbuf; field; --field, ++pos) {
429a5c4bafcSTim J. Robbins 			if (*pos && output++)
430a5c4bafcSTim J. Robbins 				for (i = 0; dcharmb[i] != '\0'; i++)
431a5c4bafcSTim J. Robbins 					putchar(dcharmb[i]);
432a5c4bafcSTim J. Robbins 			for (;;) {
433a2222839STim J. Robbins 				clen = mbrtowc(&ch, p, lbuf + reallen - p,
434a5c4bafcSTim J. Robbins 				    NULL);
435a5c4bafcSTim J. Robbins 				if (clen == (size_t)-1 || clen == (size_t)-2) {
436a5c4bafcSTim J. Robbins 					warnc(EILSEQ, "%s", fname);
437*b0578595SMartin Tournoij 					free(lbuf);
438a5c4bafcSTim J. Robbins 					return (1);
439a5c4bafcSTim J. Robbins 				}
440a5c4bafcSTim J. Robbins 				if (clen == 0)
441a5c4bafcSTim J. Robbins 					clen = 1;
442a5c4bafcSTim J. Robbins 				p += clen;
44318ba28c8SEitan Adler 				if (ch == '\n' || is_delim(ch)) {
44418ba28c8SEitan Adler 					/* compress whitespace */
44518ba28c8SEitan Adler 					if (wflag && ch != '\n')
44618ba28c8SEitan Adler 						while (is_delim(*p))
44718ba28c8SEitan Adler 							p++;
448a5c4bafcSTim J. Robbins 					break;
44918ba28c8SEitan Adler 				}
450a5c4bafcSTim J. Robbins 				if (*pos)
451a5c4bafcSTim J. Robbins 					for (i = 0; i < (int)clen; i++)
452a5c4bafcSTim J. Robbins 						putchar(p[i - clen]);
4532c39ae65SEivind Eklund 			}
4549b50d902SRodney W. Grimes 			if (ch == '\n')
4559b50d902SRodney W. Grimes 				break;
4569b50d902SRodney W. Grimes 		}
4572c39ae65SEivind Eklund 		if (ch != '\n') {
4589b50d902SRodney W. Grimes 			if (autostop) {
4599b50d902SRodney W. Grimes 				if (output)
460a5c4bafcSTim J. Robbins 					for (i = 0; dcharmb[i] != '\0'; i++)
461a5c4bafcSTim J. Robbins 						putchar(dcharmb[i]);
4629b50d902SRodney W. Grimes 				for (; (ch = *p) != '\n'; ++p)
4639b50d902SRodney W. Grimes 					(void)putchar(ch);
4649b50d902SRodney W. Grimes 			} else
4659b50d902SRodney W. Grimes 				for (; (ch = *p) != '\n'; ++p);
4662c39ae65SEivind Eklund 		}
4679b50d902SRodney W. Grimes 		(void)putchar('\n');
4689b50d902SRodney W. Grimes 	}
469*b0578595SMartin Tournoij 	free(lbuf);
470364d0a91STim J. Robbins 	return (0);
4719b50d902SRodney W. Grimes }
4729b50d902SRodney W. Grimes 
473812bff99SPhilippe Charnier static void
474f4ac32deSDavid Malone usage(void)
4759b50d902SRodney W. Grimes {
476d51c6625SEivind Eklund 	(void)fprintf(stderr, "%s\n%s\n%s\n",
477d51c6625SEivind Eklund 		"usage: cut -b list [-n] [file ...]",
478d51c6625SEivind Eklund 		"       cut -c list [file ...]",
47918ba28c8SEitan Adler 		"       cut -f list [-s] [-w | -d delim] [file ...]");
4809b50d902SRodney W. Grimes 	exit(1);
4819b50d902SRodney W. Grimes }
482