xref: /freebsd/usr.bin/csplit/csplit.c (revision a651f2bc6cb7aa4b3f343bad711cbd08c77ce2e8)
1ea9e961cSTim J. Robbins /*-
2ea9e961cSTim J. Robbins  * Copyright (c) 2002 Tim J. Robbins.
3ea9e961cSTim J. Robbins  * All rights reserved.
4ea9e961cSTim J. Robbins  *
5ea9e961cSTim J. Robbins  * Redistribution and use in source and binary forms, with or without
6ea9e961cSTim J. Robbins  * modification, are permitted provided that the following conditions
7ea9e961cSTim J. Robbins  * are met:
8ea9e961cSTim J. Robbins  * 1. Redistributions of source code must retain the above copyright
9ea9e961cSTim J. Robbins  *    notice, this list of conditions and the following disclaimer.
10ea9e961cSTim J. Robbins  * 2. Redistributions in binary form must reproduce the above copyright
11ea9e961cSTim J. Robbins  *    notice, this list of conditions and the following disclaimer in the
12ea9e961cSTim J. Robbins  *    documentation and/or other materials provided with the distribution.
13ea9e961cSTim J. Robbins  *
14ea9e961cSTim J. Robbins  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15ea9e961cSTim J. Robbins  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16ea9e961cSTim J. Robbins  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17ea9e961cSTim J. Robbins  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18ea9e961cSTim J. Robbins  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19ea9e961cSTim J. Robbins  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20ea9e961cSTim J. Robbins  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21ea9e961cSTim J. Robbins  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22ea9e961cSTim J. Robbins  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23ea9e961cSTim J. Robbins  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24ea9e961cSTim J. Robbins  * SUCH DAMAGE.
25ea9e961cSTim J. Robbins  */
26ea9e961cSTim J. Robbins 
27ea9e961cSTim J. Robbins /*
28ea9e961cSTim J. Robbins  * csplit -- split files based on context
29ea9e961cSTim J. Robbins  *
30ea9e961cSTim J. Robbins  * This utility splits its input into numbered output files by line number
31ea9e961cSTim J. Robbins  * or by a regular expression. Regular expression matches have an optional
32ea9e961cSTim J. Robbins  * offset with them, allowing the split to occur a specified number of
33ea9e961cSTim J. Robbins  * lines before or after the match.
34ea9e961cSTim J. Robbins  *
35ea9e961cSTim J. Robbins  * To handle negative offsets, we stop reading when the match occurs and
36ea9e961cSTim J. Robbins  * store the offset that the file should have been split at, then use
37ea9e961cSTim J. Robbins  * this output file as input until all the "overflowed" lines have been read.
38ea9e961cSTim J. Robbins  * The file is then closed and truncated to the correct length.
39ea9e961cSTim J. Robbins  *
40ea9e961cSTim J. Robbins  * We assume that the output files can be seeked upon (ie. they cannot be
41ea9e961cSTim J. Robbins  * symlinks to named pipes or character devices), but make no such
42ea9e961cSTim J. Robbins  * assumption about the input.
43ea9e961cSTim J. Robbins  */
44ea9e961cSTim J. Robbins 
45ea9e961cSTim J. Robbins #include <sys/cdefs.h>
46ea9e961cSTim J. Robbins __FBSDID("$FreeBSD$");
47ea9e961cSTim J. Robbins 
48ea9e961cSTim J. Robbins #include <sys/types.h>
49ea9e961cSTim J. Robbins 
50ea9e961cSTim J. Robbins #include <ctype.h>
51ea9e961cSTim J. Robbins #include <err.h>
52ea9e961cSTim J. Robbins #include <errno.h>
53ea9e961cSTim J. Robbins #include <limits.h>
549bd2f93aSTim J. Robbins #include <locale.h>
55ea9e961cSTim J. Robbins #include <regex.h>
56ea9e961cSTim J. Robbins #include <signal.h>
57ea9e961cSTim J. Robbins #include <stdint.h>
58ea9e961cSTim J. Robbins #include <stdio.h>
59ea9e961cSTim J. Robbins #include <stdlib.h>
60ea9e961cSTim J. Robbins #include <string.h>
61ea9e961cSTim J. Robbins #include <unistd.h>
62ea9e961cSTim J. Robbins 
63b168c3c8SEd Schouten static void	 cleanup(void);
64b168c3c8SEd Schouten static void	 do_lineno(const char *);
65b168c3c8SEd Schouten static void	 do_rexp(const char *);
66*a651f2bcSBaptiste Daroussin static char	*get_line(void);
67b168c3c8SEd Schouten static void	 handlesig(int);
68b168c3c8SEd Schouten static FILE	*newfile(void);
69b168c3c8SEd Schouten static void	 toomuch(FILE *, long);
70b168c3c8SEd Schouten static void	 usage(void);
71ea9e961cSTim J. Robbins 
72ea9e961cSTim J. Robbins /*
73ea9e961cSTim J. Robbins  * Command line options
74ea9e961cSTim J. Robbins  */
75b168c3c8SEd Schouten static const char *prefix;	/* File name prefix */
76b168c3c8SEd Schouten static long	 sufflen;	/* Number of decimal digits for suffix */
77b168c3c8SEd Schouten static int	 sflag;		/* Suppress output of file names */
78b168c3c8SEd Schouten static int	 kflag;		/* Keep output if error occurs */
79ea9e961cSTim J. Robbins 
80ea9e961cSTim J. Robbins /*
81ea9e961cSTim J. Robbins  * Other miscellaneous globals (XXX too many)
82ea9e961cSTim J. Robbins  */
83b168c3c8SEd Schouten static long	 lineno;	/* Current line number in input file */
84b168c3c8SEd Schouten static long	 reps;		/* Number of repetitions for this pattern */
85b168c3c8SEd Schouten static long	 nfiles;	/* Number of files output so far */
86b168c3c8SEd Schouten static long	 maxfiles;	/* Maximum number of files we can create */
87b168c3c8SEd Schouten static char	 currfile[PATH_MAX]; /* Current output file */
88b168c3c8SEd Schouten static const char *infn;	/* Name of the input file */
89b168c3c8SEd Schouten static FILE	*infile;	/* Input file handle */
90b168c3c8SEd Schouten static FILE	*overfile;	/* Overflow file for toomuch() */
91b168c3c8SEd Schouten static off_t	 truncofs;	/* Offset this file should be truncated at */
92b168c3c8SEd Schouten static int	 doclean;	/* Should cleanup() remove output? */
93ea9e961cSTim J. Robbins 
94ea9e961cSTim J. Robbins int
95ea9e961cSTim J. Robbins main(int argc, char *argv[])
96ea9e961cSTim J. Robbins {
975b3df62eSTim J. Robbins 	struct sigaction sa;
98ea9e961cSTim J. Robbins 	long i;
99ea9e961cSTim J. Robbins 	int ch;
100ea9e961cSTim J. Robbins 	const char *expr;
101ea9e961cSTim J. Robbins 	char *ep, *p;
102ea9e961cSTim J. Robbins 	FILE *ofp;
103ea9e961cSTim J. Robbins 
1049bd2f93aSTim J. Robbins 	setlocale(LC_ALL, "");
1059bd2f93aSTim J. Robbins 
106ea9e961cSTim J. Robbins 	kflag = sflag = 0;
107ea9e961cSTim J. Robbins 	prefix = "xx";
108ea9e961cSTim J. Robbins 	sufflen = 2;
109ea9e961cSTim J. Robbins 	while ((ch = getopt(argc, argv, "ksf:n:")) > 0) {
110ea9e961cSTim J. Robbins 		switch (ch) {
111ea9e961cSTim J. Robbins 		case 'f':
112ea9e961cSTim J. Robbins 			prefix = optarg;
113ea9e961cSTim J. Robbins 			break;
114ea9e961cSTim J. Robbins 		case 'k':
115ea9e961cSTim J. Robbins 			kflag = 1;
116ea9e961cSTim J. Robbins 			break;
117ea9e961cSTim J. Robbins 		case 'n':
118ea9e961cSTim J. Robbins 			errno = 0;
119ea9e961cSTim J. Robbins 			sufflen = strtol(optarg, &ep, 10);
120ea9e961cSTim J. Robbins 			if (sufflen <= 0 || *ep != '\0' || errno != 0)
121ea9e961cSTim J. Robbins 				errx(1, "%s: bad suffix length", optarg);
122ea9e961cSTim J. Robbins 			break;
123ea9e961cSTim J. Robbins 		case 's':
124ea9e961cSTim J. Robbins 			sflag = 1;
125ea9e961cSTim J. Robbins 			break;
126ea9e961cSTim J. Robbins 		default:
127ea9e961cSTim J. Robbins 			usage();
128ea9e961cSTim J. Robbins 			/*NOTREACHED*/
129ea9e961cSTim J. Robbins 		}
130ea9e961cSTim J. Robbins 	}
131ea9e961cSTim J. Robbins 
132ea9e961cSTim J. Robbins 	if (sufflen + strlen(prefix) >= PATH_MAX)
133ea9e961cSTim J. Robbins 		errx(1, "name too long");
134ea9e961cSTim J. Robbins 
135ea9e961cSTim J. Robbins 	argc -= optind;
136ea9e961cSTim J. Robbins 	argv += optind;
137ea9e961cSTim J. Robbins 
138ea9e961cSTim J. Robbins 	if ((infn = *argv++) == NULL)
139ea9e961cSTim J. Robbins 		usage();
140ea9e961cSTim J. Robbins 	if (strcmp(infn, "-") == 0) {
141ea9e961cSTim J. Robbins 		infile = stdin;
142ea9e961cSTim J. Robbins 		infn = "stdin";
143ea9e961cSTim J. Robbins 	} else if ((infile = fopen(infn, "r")) == NULL)
144ea9e961cSTim J. Robbins 		err(1, "%s", infn);
145ea9e961cSTim J. Robbins 
146ea9e961cSTim J. Robbins 	if (!kflag) {
147ea9e961cSTim J. Robbins 		doclean = 1;
148ea9e961cSTim J. Robbins 		atexit(cleanup);
1495b3df62eSTim J. Robbins 		sa.sa_flags = 0;
1505b3df62eSTim J. Robbins 		sa.sa_handler = handlesig;
1515b3df62eSTim J. Robbins 		sigemptyset(&sa.sa_mask);
1525b3df62eSTim J. Robbins 		sigaddset(&sa.sa_mask, SIGHUP);
1535b3df62eSTim J. Robbins 		sigaddset(&sa.sa_mask, SIGINT);
1545b3df62eSTim J. Robbins 		sigaddset(&sa.sa_mask, SIGTERM);
1555b3df62eSTim J. Robbins 		sigaction(SIGHUP, &sa, NULL);
1565b3df62eSTim J. Robbins 		sigaction(SIGINT, &sa, NULL);
1575b3df62eSTim J. Robbins 		sigaction(SIGTERM, &sa, NULL);
158ea9e961cSTim J. Robbins 	}
159ea9e961cSTim J. Robbins 
160ea9e961cSTim J. Robbins 	lineno = 0;
161ea9e961cSTim J. Robbins 	nfiles = 0;
162ea9e961cSTim J. Robbins 	truncofs = 0;
163ea9e961cSTim J. Robbins 	overfile = NULL;
164ea9e961cSTim J. Robbins 
165ea9e961cSTim J. Robbins 	/* Ensure 10^sufflen < LONG_MAX. */
166ea9e961cSTim J. Robbins 	for (maxfiles = 1, i = 0; i < sufflen; i++) {
167ea9e961cSTim J. Robbins 		if (maxfiles > LONG_MAX / 10)
168ea9e961cSTim J. Robbins 			errx(1, "%ld: suffix too long (limit %ld)",
169ea9e961cSTim J. Robbins 			    sufflen, i);
170ea9e961cSTim J. Robbins 		maxfiles *= 10;
171ea9e961cSTim J. Robbins 	}
172ea9e961cSTim J. Robbins 
173ea9e961cSTim J. Robbins 	/* Create files based on supplied patterns. */
174ea9e961cSTim J. Robbins 	while (nfiles < maxfiles - 1 && (expr = *argv++) != NULL) {
175ea9e961cSTim J. Robbins 		/* Look ahead & see if this pattern has any repetitions. */
176ea9e961cSTim J. Robbins 		if (*argv != NULL && **argv == '{') {
177ea9e961cSTim J. Robbins 			errno = 0;
178ea9e961cSTim J. Robbins 			reps = strtol(*argv + 1, &ep, 10);
179ea9e961cSTim J. Robbins 			if (reps < 0 || *ep != '}' || errno != 0)
180ea9e961cSTim J. Robbins 				errx(1, "%s: bad repetition count", *argv + 1);
181ea9e961cSTim J. Robbins 			argv++;
182ea9e961cSTim J. Robbins 		} else
183ea9e961cSTim J. Robbins 			reps = 0;
184ea9e961cSTim J. Robbins 
185ea9e961cSTim J. Robbins 		if (*expr == '/' || *expr == '%') {
186ea9e961cSTim J. Robbins 			do
187ea9e961cSTim J. Robbins 				do_rexp(expr);
188ea9e961cSTim J. Robbins 			while (reps-- != 0 && nfiles < maxfiles - 1);
189ea9e961cSTim J. Robbins 		} else if (isdigit((unsigned char)*expr))
190ea9e961cSTim J. Robbins 			do_lineno(expr);
191ea9e961cSTim J. Robbins 		else
192ea9e961cSTim J. Robbins 			errx(1, "%s: unrecognised pattern", expr);
193ea9e961cSTim J. Robbins 	}
194ea9e961cSTim J. Robbins 
195ea9e961cSTim J. Robbins 	/* Copy the rest into a new file. */
196ea9e961cSTim J. Robbins 	if (!feof(infile)) {
197ea9e961cSTim J. Robbins 		ofp = newfile();
198*a651f2bcSBaptiste Daroussin 		while ((p = get_line()) != NULL && fputs(p, ofp) == 0)
199ea9e961cSTim J. Robbins 			;
200ea9e961cSTim J. Robbins 		if (!sflag)
201ea9e961cSTim J. Robbins 			printf("%jd\n", (intmax_t)ftello(ofp));
202ea9e961cSTim J. Robbins 		if (fclose(ofp) != 0)
203ea9e961cSTim J. Robbins 			err(1, "%s", currfile);
204ea9e961cSTim J. Robbins 	}
205ea9e961cSTim J. Robbins 
206ea9e961cSTim J. Robbins 	toomuch(NULL, 0);
207ea9e961cSTim J. Robbins 	doclean = 0;
208ea9e961cSTim J. Robbins 
209ea9e961cSTim J. Robbins 	return (0);
210ea9e961cSTim J. Robbins }
211ea9e961cSTim J. Robbins 
212b168c3c8SEd Schouten static void
213ea9e961cSTim J. Robbins usage(void)
214ea9e961cSTim J. Robbins {
215ea9e961cSTim J. Robbins 
216ea9e961cSTim J. Robbins 	fprintf(stderr,
217bc025955STim J. Robbins "usage: csplit [-ks] [-f prefix] [-n number] file args ...\n");
218ea9e961cSTim J. Robbins 	exit(1);
219ea9e961cSTim J. Robbins }
220ea9e961cSTim J. Robbins 
221b168c3c8SEd Schouten static void
222ea9e961cSTim J. Robbins handlesig(int sig __unused)
223ea9e961cSTim J. Robbins {
224ea9e961cSTim J. Robbins 	const char msg[] = "csplit: caught signal, cleaning up\n";
225ea9e961cSTim J. Robbins 
226ea9e961cSTim J. Robbins 	write(STDERR_FILENO, msg, sizeof(msg) - 1);
227ea9e961cSTim J. Robbins 	cleanup();
228ea9e961cSTim J. Robbins 	_exit(2);
229ea9e961cSTim J. Robbins }
230ea9e961cSTim J. Robbins 
231ea9e961cSTim J. Robbins /* Create a new output file. */
232b168c3c8SEd Schouten static FILE *
233ea9e961cSTim J. Robbins newfile(void)
234ea9e961cSTim J. Robbins {
235ea9e961cSTim J. Robbins 	FILE *fp;
236ea9e961cSTim J. Robbins 
2378015edeeSDavid Malone 	if ((size_t)snprintf(currfile, sizeof(currfile), "%s%0*ld", prefix,
23823e7704bSTim J. Robbins 	    (int)sufflen, nfiles) >= sizeof(currfile))
23923e7704bSTim J. Robbins 		errc(1, ENAMETOOLONG, NULL);
240ea9e961cSTim J. Robbins 	if ((fp = fopen(currfile, "w+")) == NULL)
241ea9e961cSTim J. Robbins 		err(1, "%s", currfile);
242ea9e961cSTim J. Robbins 	nfiles++;
243ea9e961cSTim J. Robbins 
244ea9e961cSTim J. Robbins 	return (fp);
245ea9e961cSTim J. Robbins }
246ea9e961cSTim J. Robbins 
247ea9e961cSTim J. Robbins /* Remove partial output, called before exiting. */
248b168c3c8SEd Schouten static void
249ea9e961cSTim J. Robbins cleanup(void)
250ea9e961cSTim J. Robbins {
251ea9e961cSTim J. Robbins 	char fnbuf[PATH_MAX];
252ea9e961cSTim J. Robbins 	long i;
253ea9e961cSTim J. Robbins 
254ea9e961cSTim J. Robbins 	if (!doclean)
255ea9e961cSTim J. Robbins 		return;
256ea9e961cSTim J. Robbins 
257ea9e961cSTim J. Robbins 	/*
258ea9e961cSTim J. Robbins 	 * NOTE: One cannot portably assume to be able to call snprintf()
259ea9e961cSTim J. Robbins 	 * from inside a signal handler. It does, however, appear to be safe
260ea9e961cSTim J. Robbins 	 * to do on FreeBSD. The solution to this problem is worse than the
261ea9e961cSTim J. Robbins 	 * problem itself.
262ea9e961cSTim J. Robbins 	 */
263ea9e961cSTim J. Robbins 
264ea9e961cSTim J. Robbins 	for (i = 0; i < nfiles; i++) {
265ea9e961cSTim J. Robbins 		snprintf(fnbuf, sizeof(fnbuf), "%s%0*ld", prefix,
266ea9e961cSTim J. Robbins 		    (int)sufflen, i);
267ea9e961cSTim J. Robbins 		unlink(fnbuf);
268ea9e961cSTim J. Robbins 	}
269ea9e961cSTim J. Robbins }
270ea9e961cSTim J. Robbins 
271ea9e961cSTim J. Robbins /* Read a line from the input into a static buffer. */
272b168c3c8SEd Schouten static char *
273*a651f2bcSBaptiste Daroussin get_line(void)
274ea9e961cSTim J. Robbins {
275ea9e961cSTim J. Robbins 	static char lbuf[LINE_MAX];
276ea9e961cSTim J. Robbins 	FILE *src;
277ea9e961cSTim J. Robbins 
278ea9e961cSTim J. Robbins 	src = overfile != NULL ? overfile : infile;
279ea9e961cSTim J. Robbins 
280ea9e961cSTim J. Robbins again: if (fgets(lbuf, sizeof(lbuf), src) == NULL) {
281ea9e961cSTim J. Robbins 		if (src == overfile) {
282ea9e961cSTim J. Robbins 			src = infile;
283ea9e961cSTim J. Robbins 			goto again;
284ea9e961cSTim J. Robbins 		}
285ea9e961cSTim J. Robbins 		return (NULL);
286ea9e961cSTim J. Robbins 	}
287ea9e961cSTim J. Robbins 	if (ferror(src))
288ea9e961cSTim J. Robbins 		err(1, "%s", infn);
289ea9e961cSTim J. Robbins 	lineno++;
290ea9e961cSTim J. Robbins 
291ea9e961cSTim J. Robbins 	return (lbuf);
292ea9e961cSTim J. Robbins }
293ea9e961cSTim J. Robbins 
294*a651f2bcSBaptiste Daroussin /* Conceptually rewind the input (as obtained by get_line()) back `n' lines. */
295b168c3c8SEd Schouten static void
296ea9e961cSTim J. Robbins toomuch(FILE *ofp, long n)
297ea9e961cSTim J. Robbins {
298ea9e961cSTim J. Robbins 	char buf[BUFSIZ];
299ea9e961cSTim J. Robbins 	size_t i, nread;
300ea9e961cSTim J. Robbins 
301ea9e961cSTim J. Robbins 	if (overfile != NULL) {
302ea9e961cSTim J. Robbins 		/*
303ea9e961cSTim J. Robbins 		 * Truncate the previous file we overflowed into back to
304ea9e961cSTim J. Robbins 		 * the correct length, close it.
305ea9e961cSTim J. Robbins 		 */
306ea9e961cSTim J. Robbins 		if (fflush(overfile) != 0)
307ea9e961cSTim J. Robbins 			err(1, "overflow");
308ea9e961cSTim J. Robbins 		if (ftruncate(fileno(overfile), truncofs) != 0)
309ea9e961cSTim J. Robbins 			err(1, "overflow");
310ea9e961cSTim J. Robbins 		if (fclose(overfile) != 0)
311ea9e961cSTim J. Robbins 			err(1, "overflow");
312ea9e961cSTim J. Robbins 		overfile = NULL;
313ea9e961cSTim J. Robbins 	}
314ea9e961cSTim J. Robbins 
315ea9e961cSTim J. Robbins 	if (n == 0)
316ea9e961cSTim J. Robbins 		/* Just tidying up */
317ea9e961cSTim J. Robbins 		return;
318ea9e961cSTim J. Robbins 
319ea9e961cSTim J. Robbins 	lineno -= n;
320ea9e961cSTim J. Robbins 
321ea9e961cSTim J. Robbins 	/*
322ea9e961cSTim J. Robbins 	 * Wind the overflow file backwards to `n' lines before the
323ea9e961cSTim J. Robbins 	 * current one.
324ea9e961cSTim J. Robbins 	 */
325ea9e961cSTim J. Robbins 	do {
326ea9e961cSTim J. Robbins 		if (ftello(ofp) < (off_t)sizeof(buf))
327ea9e961cSTim J. Robbins 			rewind(ofp);
328ea9e961cSTim J. Robbins 		else
3292fa2fea1STim J. Robbins 			fseeko(ofp, -(off_t)sizeof(buf), SEEK_CUR);
330ea9e961cSTim J. Robbins 		if (ferror(ofp))
331ea9e961cSTim J. Robbins 			errx(1, "%s: can't seek", currfile);
332ea9e961cSTim J. Robbins 		if ((nread = fread(buf, 1, sizeof(buf), ofp)) == 0)
333ea9e961cSTim J. Robbins 			errx(1, "can't read overflowed output");
3342fa2fea1STim J. Robbins 		if (fseeko(ofp, -(off_t)nread, SEEK_CUR) != 0)
335ea9e961cSTim J. Robbins 			err(1, "%s", currfile);
336ea9e961cSTim J. Robbins 		for (i = 1; i <= nread; i++)
337ea9e961cSTim J. Robbins 			if (buf[nread - i] == '\n' && n-- == 0)
338ea9e961cSTim J. Robbins 				break;
339248602a1STim J. Robbins 		if (ftello(ofp) == 0)
340248602a1STim J. Robbins 			break;
341ea9e961cSTim J. Robbins 	} while (n > 0);
3422fa2fea1STim J. Robbins 	if (fseeko(ofp, nread - i + 1, SEEK_CUR) != 0)
343ea9e961cSTim J. Robbins 		err(1, "%s", currfile);
344ea9e961cSTim J. Robbins 
345ea9e961cSTim J. Robbins 	/*
346*a651f2bcSBaptiste Daroussin 	 * get_line() will read from here. Next call will truncate to
347ea9e961cSTim J. Robbins 	 * truncofs in this file.
348ea9e961cSTim J. Robbins 	 */
349ea9e961cSTim J. Robbins 	overfile = ofp;
350ea9e961cSTim J. Robbins 	truncofs = ftello(overfile);
351ea9e961cSTim J. Robbins }
352ea9e961cSTim J. Robbins 
353ea9e961cSTim J. Robbins /* Handle splits for /regexp/ and %regexp% patterns. */
354b168c3c8SEd Schouten static void
355ea9e961cSTim J. Robbins do_rexp(const char *expr)
356ea9e961cSTim J. Robbins {
357ea9e961cSTim J. Robbins 	regex_t cre;
358ea9e961cSTim J. Robbins 	intmax_t nwritten;
359ea9e961cSTim J. Robbins 	long ofs;
360ea9e961cSTim J. Robbins 	int first;
361ea9e961cSTim J. Robbins 	char *ecopy, *ep, *p, *pofs, *re;
362ea9e961cSTim J. Robbins 	FILE *ofp;
363ea9e961cSTim J. Robbins 
364ea9e961cSTim J. Robbins 	if ((ecopy = strdup(expr)) == NULL)
365ea9e961cSTim J. Robbins 		err(1, "strdup");
366ea9e961cSTim J. Robbins 
367ea9e961cSTim J. Robbins 	re = ecopy + 1;
368ea9e961cSTim J. Robbins 	if ((pofs = strrchr(ecopy, *expr)) == NULL || pofs[-1] == '\\')
369ea9e961cSTim J. Robbins 		errx(1, "%s: missing trailing %c", expr, *expr);
370ea9e961cSTim J. Robbins 	*pofs++ = '\0';
371ea9e961cSTim J. Robbins 
372ea9e961cSTim J. Robbins 	if (*pofs != '\0') {
373ea9e961cSTim J. Robbins 		errno = 0;
374ea9e961cSTim J. Robbins 		ofs = strtol(pofs, &ep, 10);
375ea9e961cSTim J. Robbins 		if (*ep != '\0' || errno != 0)
376ea9e961cSTim J. Robbins 			errx(1, "%s: bad offset", pofs);
377ea9e961cSTim J. Robbins 	} else
378ea9e961cSTim J. Robbins 		ofs = 0;
379ea9e961cSTim J. Robbins 
380ea9e961cSTim J. Robbins 	if (regcomp(&cre, re, REG_BASIC|REG_NOSUB) != 0)
381ea9e961cSTim J. Robbins 		errx(1, "%s: bad regular expression", re);
382ea9e961cSTim J. Robbins 
383ea9e961cSTim J. Robbins 	if (*expr == '/')
384ea9e961cSTim J. Robbins 		/* /regexp/: Save results to a file. */
385ea9e961cSTim J. Robbins 		ofp = newfile();
386ea9e961cSTim J. Robbins 	else {
387ea9e961cSTim J. Robbins 		/* %regexp%: Make a temporary file for overflow. */
388ea9e961cSTim J. Robbins 		if ((ofp = tmpfile()) == NULL)
389ea9e961cSTim J. Robbins 			err(1, "tmpfile");
390ea9e961cSTim J. Robbins 	}
391ea9e961cSTim J. Robbins 
392ea9e961cSTim J. Robbins 	/* Read and output lines until we get a match. */
393ea9e961cSTim J. Robbins 	first = 1;
394*a651f2bcSBaptiste Daroussin 	while ((p = get_line()) != NULL) {
395ea9e961cSTim J. Robbins 		if (fputs(p, ofp) != 0)
396ea9e961cSTim J. Robbins 			break;
397ea9e961cSTim J. Robbins 		if (!first && regexec(&cre, p, 0, NULL, 0) == 0)
398ea9e961cSTim J. Robbins 			break;
399ea9e961cSTim J. Robbins 		first = 0;
400ea9e961cSTim J. Robbins 	}
401ea9e961cSTim J. Robbins 
402ea9e961cSTim J. Robbins 	if (p == NULL)
403ea9e961cSTim J. Robbins 		errx(1, "%s: no match", re);
404ea9e961cSTim J. Robbins 
405ea9e961cSTim J. Robbins 	if (ofs <= 0) {
406ea9e961cSTim J. Robbins 		/*
407ea9e961cSTim J. Robbins 		 * Negative (or zero) offset: throw back any lines we should
408ea9e961cSTim J. Robbins 		 * not have read yet.
409ea9e961cSTim J. Robbins 		  */
410ea9e961cSTim J. Robbins 		if (p != NULL) {
411ea9e961cSTim J. Robbins 			toomuch(ofp, -ofs + 1);
412ea9e961cSTim J. Robbins 			nwritten = (intmax_t)truncofs;
413ea9e961cSTim J. Robbins 		} else
414ea9e961cSTim J. Robbins 			nwritten = (intmax_t)ftello(ofp);
415ea9e961cSTim J. Robbins 	} else {
416ea9e961cSTim J. Robbins 		/*
417ea9e961cSTim J. Robbins 		 * Positive offset: copy the requested number of lines
418ea9e961cSTim J. Robbins 		 * after the match.
419ea9e961cSTim J. Robbins 		 */
420*a651f2bcSBaptiste Daroussin 		while (--ofs > 0 && (p = get_line()) != NULL)
421ea9e961cSTim J. Robbins 			fputs(p, ofp);
422ea9e961cSTim J. Robbins 		toomuch(NULL, 0);
423ea9e961cSTim J. Robbins 		nwritten = (intmax_t)ftello(ofp);
424ea9e961cSTim J. Robbins 		if (fclose(ofp) != 0)
425ea9e961cSTim J. Robbins 			err(1, "%s", currfile);
426ea9e961cSTim J. Robbins 	}
427ea9e961cSTim J. Robbins 
428ea9e961cSTim J. Robbins 	if (!sflag && *expr == '/')
429ea9e961cSTim J. Robbins 		printf("%jd\n", nwritten);
430ea9e961cSTim J. Robbins 
431ea9e961cSTim J. Robbins 	regfree(&cre);
432ea9e961cSTim J. Robbins 	free(ecopy);
433ea9e961cSTim J. Robbins }
434ea9e961cSTim J. Robbins 
435ea9e961cSTim J. Robbins /* Handle splits based on line number. */
436b168c3c8SEd Schouten static void
437ea9e961cSTim J. Robbins do_lineno(const char *expr)
438ea9e961cSTim J. Robbins {
439ea9e961cSTim J. Robbins 	long lastline, tgtline;
440ea9e961cSTim J. Robbins 	char *ep, *p;
441ea9e961cSTim J. Robbins 	FILE *ofp;
442ea9e961cSTim J. Robbins 
443ea9e961cSTim J. Robbins 	errno = 0;
444ea9e961cSTim J. Robbins 	tgtline = strtol(expr, &ep, 10);
445ea9e961cSTim J. Robbins 	if (tgtline <= 0 || errno != 0 || *ep != '\0')
446ea9e961cSTim J. Robbins 		errx(1, "%s: bad line number", expr);
447ea9e961cSTim J. Robbins 	lastline = tgtline;
448ea9e961cSTim J. Robbins 	if (lastline <= lineno)
449ea9e961cSTim J. Robbins 		errx(1, "%s: can't go backwards", expr);
450ea9e961cSTim J. Robbins 
451ea9e961cSTim J. Robbins 	while (nfiles < maxfiles - 1) {
452ea9e961cSTim J. Robbins 		ofp = newfile();
453ea9e961cSTim J. Robbins 		while (lineno + 1 != lastline) {
454*a651f2bcSBaptiste Daroussin 			if ((p = get_line()) == NULL)
455ea9e961cSTim J. Robbins 				errx(1, "%ld: out of range", lastline);
456ea9e961cSTim J. Robbins 			if (fputs(p, ofp) != 0)
457ea9e961cSTim J. Robbins 				break;
458ea9e961cSTim J. Robbins 		}
459ea9e961cSTim J. Robbins 		if (!sflag)
460ea9e961cSTim J. Robbins 			printf("%jd\n", (intmax_t)ftello(ofp));
461ea9e961cSTim J. Robbins 		if (fclose(ofp) != 0)
462ea9e961cSTim J. Robbins 			err(1, "%s", currfile);
463ea9e961cSTim J. Robbins 		if (reps-- == 0)
464ea9e961cSTim J. Robbins 			break;
465ea9e961cSTim J. Robbins 		lastline += tgtline;
466ea9e961cSTim J. Robbins 	}
467ea9e961cSTim J. Robbins }
468