1/*-
2 * SPDX-License-Identifier: BSD-2-Clause
3 *
4 * Copyright (c) 2002 Tim J. Robbins.
5 * All rights reserved.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 *    notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 *    notice, this list of conditions and the following disclaimer in the
14 *    documentation and/or other materials provided with the distribution.
15 *
16 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26 * SUCH DAMAGE.
27 */
28
29/*
30 * csplit -- split files based on context
31 *
32 * This utility splits its input into numbered output files by line number
33 * or by a regular expression. Regular expression matches have an optional
34 * offset with them, allowing the split to occur a specified number of
35 * lines before or after the match.
36 *
37 * To handle negative offsets, we stop reading when the match occurs and
38 * store the offset that the file should have been split at, then use
39 * this output file as input until all the "overflowed" lines have been read.
40 * The file is then closed and truncated to the correct length.
41 *
42 * We assume that the output files can be seeked upon (ie. they cannot be
43 * symlinks to named pipes or character devices), but make no such
44 * assumption about the input.
45 */
46
47#include <sys/types.h>
48
49#include <ctype.h>
50#include <err.h>
51#include <errno.h>
52#include <limits.h>
53#include <locale.h>
54#include <regex.h>
55#include <signal.h>
56#include <stdint.h>
57#include <stdio.h>
58#include <stdlib.h>
59#include <string.h>
60#include <unistd.h>
61
62static void	 cleanup(void);
63static void	 do_lineno(const char *);
64static void	 do_rexp(const char *);
65static char	*get_line(void);
66static void	 handlesig(int);
67static FILE	*newfile(void);
68static void	 toomuch(FILE *, long);
69static void	 usage(void);
70
71/*
72 * Command line options
73 */
74static const char *prefix;	/* File name prefix */
75static long	 sufflen;	/* Number of decimal digits for suffix */
76static int	 sflag;		/* Suppress output of file names */
77static int	 kflag;		/* Keep output if error occurs */
78
79/*
80 * Other miscellaneous globals (XXX too many)
81 */
82static long	 lineno;	/* Current line number in input file */
83static long	 reps;		/* Number of repetitions for this pattern */
84static long	 nfiles;	/* Number of files output so far */
85static long	 maxfiles;	/* Maximum number of files we can create */
86static char	 currfile[PATH_MAX]; /* Current output file */
87static const char *infn;	/* Name of the input file */
88static FILE	*infile;	/* Input file handle */
89static FILE	*overfile;	/* Overflow file for toomuch() */
90static off_t	 truncofs;	/* Offset this file should be truncated at */
91static int	 doclean;	/* Should cleanup() remove output? */
92
93int
94main(int argc, char *argv[])
95{
96	struct sigaction sa;
97	long i;
98	int ch;
99	const char *expr;
100	char *ep, *p;
101	FILE *ofp;
102
103	setlocale(LC_ALL, "");
104
105	kflag = sflag = 0;
106	prefix = "xx";
107	sufflen = 2;
108	while ((ch = getopt(argc, argv, "ksf:n:")) > 0) {
109		switch (ch) {
110		case 'f':
111			prefix = optarg;
112			break;
113		case 'k':
114			kflag = 1;
115			break;
116		case 'n':
117			errno = 0;
118			sufflen = strtol(optarg, &ep, 10);
119			if (sufflen <= 0 || *ep != '\0' || errno != 0)
120				errx(1, "%s: bad suffix length", optarg);
121			break;
122		case 's':
123			sflag = 1;
124			break;
125		default:
126			usage();
127			/*NOTREACHED*/
128		}
129	}
130
131	if (sufflen + strlen(prefix) >= PATH_MAX)
132		errx(1, "name too long");
133
134	argc -= optind;
135	argv += optind;
136
137	if ((infn = *argv++) == NULL)
138		usage();
139	if (strcmp(infn, "-") == 0) {
140		infile = stdin;
141		infn = "stdin";
142	} else if ((infile = fopen(infn, "r")) == NULL)
143		err(1, "%s", infn);
144
145	if (!kflag) {
146		doclean = 1;
147		atexit(cleanup);
148		sa.sa_flags = 0;
149		sa.sa_handler = handlesig;
150		sigemptyset(&sa.sa_mask);
151		sigaddset(&sa.sa_mask, SIGHUP);
152		sigaddset(&sa.sa_mask, SIGINT);
153		sigaddset(&sa.sa_mask, SIGTERM);
154		sigaction(SIGHUP, &sa, NULL);
155		sigaction(SIGINT, &sa, NULL);
156		sigaction(SIGTERM, &sa, NULL);
157	}
158
159	lineno = 0;
160	nfiles = 0;
161	truncofs = 0;
162	overfile = NULL;
163
164	/* Ensure 10^sufflen < LONG_MAX. */
165	for (maxfiles = 1, i = 0; i < sufflen; i++) {
166		if (maxfiles > LONG_MAX / 10)
167			errx(1, "%ld: suffix too long (limit %ld)",
168			    sufflen, i);
169		maxfiles *= 10;
170	}
171
172	/* Create files based on supplied patterns. */
173	while (nfiles < maxfiles - 1 && (expr = *argv++) != NULL) {
174		/* Look ahead & see if this pattern has any repetitions. */
175		if (*argv != NULL && **argv == '{') {
176			errno = 0;
177			reps = strtol(*argv + 1, &ep, 10);
178			if (reps < 0 || *ep != '}' || errno != 0)
179				errx(1, "%s: bad repetition count", *argv + 1);
180			argv++;
181		} else
182			reps = 0;
183
184		if (*expr == '/' || *expr == '%') {
185			do
186				do_rexp(expr);
187			while (reps-- != 0 && nfiles < maxfiles - 1);
188		} else if (isdigit((unsigned char)*expr))
189			do_lineno(expr);
190		else
191			errx(1, "%s: unrecognised pattern", expr);
192	}
193
194	/* Copy the rest into a new file. */
195	if (!feof(infile)) {
196		ofp = newfile();
197		while ((p = get_line()) != NULL && fputs(p, ofp) != EOF)
198			;
199		if (!sflag)
200			printf("%jd\n", (intmax_t)ftello(ofp));
201		if (fclose(ofp) != 0)
202			err(1, "%s", currfile);
203	}
204
205	toomuch(NULL, 0);
206	doclean = 0;
207
208	return (0);
209}
210
211static void
212usage(void)
213{
214
215	fprintf(stderr,
216"usage: csplit [-ks] [-f prefix] [-n number] file args ...\n");
217	exit(1);
218}
219
220static void
221handlesig(int sig __unused)
222{
223	const char msg[] = "csplit: caught signal, cleaning up\n";
224
225	write(STDERR_FILENO, msg, sizeof(msg) - 1);
226	cleanup();
227	_exit(2);
228}
229
230/* Create a new output file. */
231static FILE *
232newfile(void)
233{
234	FILE *fp;
235
236	if ((size_t)snprintf(currfile, sizeof(currfile), "%s%0*ld", prefix,
237	    (int)sufflen, nfiles) >= sizeof(currfile))
238		errc(1, ENAMETOOLONG, NULL);
239	if ((fp = fopen(currfile, "w+")) == NULL)
240		err(1, "%s", currfile);
241	nfiles++;
242
243	return (fp);
244}
245
246/* Remove partial output, called before exiting. */
247static void
248cleanup(void)
249{
250	char fnbuf[PATH_MAX];
251	long i;
252
253	if (!doclean)
254		return;
255
256	/*
257	 * NOTE: One cannot portably assume to be able to call snprintf()
258	 * from inside a signal handler. It does, however, appear to be safe
259	 * to do on FreeBSD. The solution to this problem is worse than the
260	 * problem itself.
261	 */
262
263	for (i = 0; i < nfiles; i++) {
264		snprintf(fnbuf, sizeof(fnbuf), "%s%0*ld", prefix,
265		    (int)sufflen, i);
266		unlink(fnbuf);
267	}
268}
269
270/* Read a line from the input into a static buffer. */
271static char *
272get_line(void)
273{
274	static char lbuf[LINE_MAX];
275	FILE *src;
276
277	src = overfile != NULL ? overfile : infile;
278
279again: if (fgets(lbuf, sizeof(lbuf), src) == NULL) {
280		if (src == overfile) {
281			src = infile;
282			goto again;
283		}
284		return (NULL);
285	}
286	if (ferror(src))
287		err(1, "%s", infn);
288	lineno++;
289
290	return (lbuf);
291}
292
293/* Conceptually rewind the input (as obtained by get_line()) back `n' lines. */
294static void
295toomuch(FILE *ofp, long n)
296{
297	char buf[BUFSIZ];
298	size_t i, nread;
299
300	if (overfile != NULL) {
301		/*
302		 * Truncate the previous file we overflowed into back to
303		 * the correct length, close it.
304		 */
305		if (fflush(overfile) != 0)
306			err(1, "overflow");
307		if (ftruncate(fileno(overfile), truncofs) != 0)
308			err(1, "overflow");
309		if (fclose(overfile) != 0)
310			err(1, "overflow");
311		overfile = NULL;
312	}
313
314	if (n == 0)
315		/* Just tidying up */
316		return;
317
318	lineno -= n;
319
320	/*
321	 * Wind the overflow file backwards to `n' lines before the
322	 * current one.
323	 */
324	do {
325		if (ftello(ofp) < (off_t)sizeof(buf))
326			rewind(ofp);
327		else
328			fseeko(ofp, -(off_t)sizeof(buf), SEEK_CUR);
329		if (ferror(ofp))
330			errx(1, "%s: can't seek", currfile);
331		if ((nread = fread(buf, 1, sizeof(buf), ofp)) == 0)
332			errx(1, "can't read overflowed output");
333		if (fseeko(ofp, -(off_t)nread, SEEK_CUR) != 0)
334			err(1, "%s", currfile);
335		for (i = 1; i <= nread; i++)
336			if (buf[nread - i] == '\n' && n-- == 0)
337				break;
338		if (ftello(ofp) == 0)
339			break;
340	} while (n > 0);
341	if (fseeko(ofp, nread - i + 1, SEEK_CUR) != 0)
342		err(1, "%s", currfile);
343
344	/*
345	 * get_line() will read from here. Next call will truncate to
346	 * truncofs in this file.
347	 */
348	overfile = ofp;
349	truncofs = ftello(overfile);
350}
351
352/* Handle splits for /regexp/ and %regexp% patterns. */
353static void
354do_rexp(const char *expr)
355{
356	regex_t cre;
357	intmax_t nwritten;
358	long ofs;
359	int first;
360	char *ecopy, *ep, *p, *pofs, *re;
361	FILE *ofp;
362
363	if ((ecopy = strdup(expr)) == NULL)
364		err(1, "strdup");
365
366	re = ecopy + 1;
367	if ((pofs = strrchr(ecopy, *expr)) == NULL || pofs[-1] == '\\')
368		errx(1, "%s: missing trailing %c", expr, *expr);
369	*pofs++ = '\0';
370
371	if (*pofs != '\0') {
372		errno = 0;
373		ofs = strtol(pofs, &ep, 10);
374		if (*ep != '\0' || errno != 0)
375			errx(1, "%s: bad offset", pofs);
376	} else
377		ofs = 0;
378
379	if (regcomp(&cre, re, REG_BASIC|REG_NOSUB) != 0)
380		errx(1, "%s: bad regular expression", re);
381
382	if (*expr == '/')
383		/* /regexp/: Save results to a file. */
384		ofp = newfile();
385	else {
386		/* %regexp%: Make a temporary file for overflow. */
387		if ((ofp = tmpfile()) == NULL)
388			err(1, "tmpfile");
389	}
390
391	/* Read and output lines until we get a match. */
392	first = 1;
393	while ((p = get_line()) != NULL) {
394		if (fputs(p, ofp) == EOF)
395			break;
396		if (!first && regexec(&cre, p, 0, NULL, 0) == 0)
397			break;
398		first = 0;
399	}
400
401	if (p == NULL) {
402		toomuch(NULL, 0);
403		errx(1, "%s: no match", re);
404	}
405
406	if (ofs <= 0) {
407		/*
408		 * Negative (or zero) offset: throw back any lines we should
409		 * not have read yet.
410		  */
411		if (p != NULL) {
412			toomuch(ofp, -ofs + 1);
413			nwritten = (intmax_t)truncofs;
414		} else
415			nwritten = (intmax_t)ftello(ofp);
416	} else {
417		/*
418		 * Positive offset: copy the requested number of lines
419		 * after the match.
420		 */
421		while (--ofs > 0 && (p = get_line()) != NULL)
422			fputs(p, ofp);
423		toomuch(NULL, 0);
424		nwritten = (intmax_t)ftello(ofp);
425		if (fclose(ofp) != 0)
426			err(1, "%s", currfile);
427	}
428
429	if (!sflag && *expr == '/')
430		printf("%jd\n", nwritten);
431
432	regfree(&cre);
433	free(ecopy);
434}
435
436/* Handle splits based on line number. */
437static void
438do_lineno(const char *expr)
439{
440	long lastline, tgtline;
441	char *ep, *p;
442	FILE *ofp;
443
444	errno = 0;
445	tgtline = strtol(expr, &ep, 10);
446	if (tgtline <= 0 || errno != 0 || *ep != '\0')
447		errx(1, "%s: bad line number", expr);
448	lastline = tgtline;
449	if (lastline <= lineno)
450		errx(1, "%s: can't go backwards", expr);
451
452	while (nfiles < maxfiles - 1) {
453		ofp = newfile();
454		while (lineno + 1 != lastline) {
455			if ((p = get_line()) == NULL)
456				errx(1, "%ld: out of range", lastline);
457			if (fputs(p, ofp) == EOF)
458				break;
459		}
460		if (!sflag)
461			printf("%jd\n", (intmax_t)ftello(ofp));
462		if (fclose(ofp) != 0)
463			err(1, "%s", currfile);
464		if (reps-- == 0)
465			break;
466		lastline += tgtline;
467	}
468}
469