1/*-
2 * Copyright (c) 2002 Tim J. Robbins.
3 * All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 *    notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 *    notice, this list of conditions and the following disclaimer in the
12 *    documentation and/or other materials provided with the distribution.
13 *
14 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
16 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
17 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
18 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
19 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
20 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
21 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
22 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
23 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
24 * SUCH DAMAGE.
25 */
26
27/*
28 * csplit -- split files based on context
29 *
30 * This utility splits its input into numbered output files by line number
31 * or by a regular expression. Regular expression matches have an optional
32 * offset with them, allowing the split to occur a specified number of
33 * lines before or after the match.
34 *
35 * To handle negative offsets, we stop reading when the match occurs and
36 * store the offset that the file should have been split at, then use
37 * this output file as input until all the "overflowed" lines have been read.
38 * The file is then closed and truncated to the correct length.
39 *
40 * We assume that the output files can be seeked upon (ie. they cannot be
41 * symlinks to named pipes or character devices), but make no such
42 * assumption about the input.
43 */
44
45#include <sys/cdefs.h>
46__FBSDID("$FreeBSD: src/usr.bin/csplit/csplit.c,v 1.9 2004/03/22 11:15:03 tjr Exp $");
47
48#include <sys/types.h>
49
50#include <ctype.h>
51#include <err.h>
52#include <errno.h>
53#include <limits.h>
54#include <locale.h>
55#include <regex.h>
56#include <signal.h>
57#include <stdint.h>
58#include <stdio.h>
59#include <stdlib.h>
60#include <string.h>
61#include <unistd.h>
62
63void	 cleanup(void);
64void	 do_lineno(const char *);
65void	 do_rexp(const char *);
66char	*csplit_getline(void);
67void	 handlesig(int);
68FILE	*newfile(void);
69void	 toomuch(FILE *, long);
70void	 usage(void);
71
72/*
73 * Command line options
74 */
75const char *prefix;		/* File name prefix */
76long	 sufflen;		/* Number of decimal digits for suffix */
77int	 sflag;			/* Suppress output of file names */
78int	 kflag;			/* Keep output if error occurs */
79
80/*
81 * Other miscellaneous globals (XXX too many)
82 */
83long	 lineno;		/* Current line number in input file */
84long	 reps;			/* Number of repetitions for this pattern */
85long	 nfiles;		/* Number of files output so far */
86long	 maxfiles;		/* Maximum number of files we can create */
87char	 currfile[PATH_MAX];	/* Current output file */
88const char *infn;		/* Name of the input file */
89FILE	*infile;		/* Input file handle */
90FILE	*overfile;		/* Overflow file for toomuch() */
91off_t	 truncofs;		/* Offset this file should be truncated at */
92int	 doclean;		/* Should cleanup() remove output? */
93
94int
95main(int argc, char *argv[])
96{
97	struct sigaction sa;
98	long i;
99	int ch;
100	const char *expr;
101	char *ep, *p;
102	FILE *ofp;
103
104	setlocale(LC_ALL, "");
105
106	kflag = sflag = 0;
107	prefix = "xx";
108	sufflen = 2;
109	while ((ch = getopt(argc, argv, "ksf:n:")) > 0) {
110		switch (ch) {
111		case 'f':
112			prefix = optarg;
113			break;
114		case 'k':
115			kflag = 1;
116			break;
117		case 'n':
118			errno = 0;
119			sufflen = strtol(optarg, &ep, 10);
120			if (sufflen <= 0 || *ep != '\0' || errno != 0)
121				errx(1, "%s: bad suffix length", optarg);
122			break;
123		case 's':
124			sflag = 1;
125			break;
126		default:
127			usage();
128			/*NOTREACHED*/
129		}
130	}
131
132	if (sufflen + strlen(prefix) >= PATH_MAX)
133		errx(1, "name too long");
134
135	argc -= optind;
136	argv += optind;
137
138	if ((infn = *argv++) == NULL)
139		usage();
140	if (strcmp(infn, "-") == 0) {
141		infile = stdin;
142		infn = "stdin";
143	} else if ((infile = fopen(infn, "r")) == NULL)
144		err(1, "%s", infn);
145
146	if (!kflag) {
147		doclean = 1;
148		atexit(cleanup);
149		sa.sa_flags = 0;
150		sa.sa_handler = handlesig;
151		sigemptyset(&sa.sa_mask);
152		sigaddset(&sa.sa_mask, SIGHUP);
153		sigaddset(&sa.sa_mask, SIGINT);
154		sigaddset(&sa.sa_mask, SIGTERM);
155		sigaction(SIGHUP, &sa, NULL);
156		sigaction(SIGINT, &sa, NULL);
157		sigaction(SIGTERM, &sa, NULL);
158	}
159
160	lineno = 0;
161	nfiles = 0;
162	truncofs = 0;
163	overfile = NULL;
164
165	/* Ensure 10^sufflen < LONG_MAX. */
166	for (maxfiles = 1, i = 0; i < sufflen; i++) {
167		if (maxfiles > LONG_MAX / 10)
168			errx(1, "%ld: suffix too long (limit %ld)",
169			    sufflen, i);
170		maxfiles *= 10;
171	}
172
173	/* Create files based on supplied patterns. */
174	while (nfiles < maxfiles - 1 && (expr = *argv++) != NULL) {
175		/* Look ahead & see if this pattern has any repetitions. */
176		if (*argv != NULL && **argv == '{') {
177			errno = 0;
178			reps = strtol(*argv + 1, &ep, 10);
179			if (reps < 0 || *ep != '}' || errno != 0)
180				errx(1, "%s: bad repetition count", *argv + 1);
181			argv++;
182		} else
183			reps = 0;
184
185		if (*expr == '/' || *expr == '%') {
186			do
187				do_rexp(expr);
188			while (reps-- != 0 && nfiles < maxfiles - 1);
189		} else if (isdigit((unsigned char)*expr))
190			do_lineno(expr);
191		else
192			errx(1, "%s: unrecognised pattern", expr);
193	}
194
195	/* Copy the rest into a new file. */
196	if (!feof(infile)) {
197		ofp = newfile();
198		while ((p = csplit_getline()) != NULL && fputs(p, ofp) != EOF)
199			;
200		if (!sflag)
201			printf("%jd\n", (intmax_t)ftello(ofp));
202		if (fclose(ofp) != 0)
203			err(1, "%s", currfile);
204	}
205
206	toomuch(NULL, 0);
207	doclean = 0;
208
209	return (0);
210}
211
212void
213usage(void)
214{
215
216	fprintf(stderr,
217"usage: csplit [-ks] [-f prefix] [-n number] file args ...\n");
218	exit(1);
219}
220
221void
222handlesig(int sig __unused)
223{
224	const char msg[] = "csplit: caught signal, cleaning up\n";
225
226	write(STDERR_FILENO, msg, sizeof(msg) - 1);
227	cleanup();
228	_exit(2);
229}
230
231/* Create a new output file. */
232FILE *
233newfile(void)
234{
235	FILE *fp;
236
237	if ((size_t)snprintf(currfile, sizeof(currfile), "%s%0*ld", prefix,
238	    (int)sufflen, nfiles) >= sizeof(currfile))
239		errc(1, ENAMETOOLONG, NULL);
240	if ((fp = fopen(currfile, "w+")) == NULL)
241		err(1, "%s", currfile);
242	nfiles++;
243
244	return (fp);
245}
246
247/* Remove partial output, called before exiting. */
248void
249cleanup(void)
250{
251	char fnbuf[PATH_MAX];
252	long i;
253
254	if (!doclean)
255		return;
256
257	/*
258	 * NOTE: One cannot portably assume to be able to call snprintf()
259	 * from inside a signal handler. It does, however, appear to be safe
260	 * to do on FreeBSD. The solution to this problem is worse than the
261	 * problem itself.
262	 */
263
264	for (i = 0; i < nfiles; i++) {
265		snprintf(fnbuf, sizeof(fnbuf), "%s%0*ld", prefix,
266		    (int)sufflen, i);
267		unlink(fnbuf);
268	}
269}
270
271/* Read a line from the input into a static buffer. */
272char *
273csplit_getline(void)
274{
275	static char lbuf[LINE_MAX];
276	FILE *src;
277
278	src = overfile != NULL ? overfile : infile;
279
280again: if (fgets(lbuf, sizeof(lbuf), src) == NULL) {
281		if (src == overfile) {
282			src = infile;
283			goto again;
284		}
285		return (NULL);
286	}
287	if (ferror(src))
288		err(1, "%s", infn);
289	lineno++;
290
291	return (lbuf);
292}
293
294/* Conceptually rewind the input (as obtained by csplit_getline()) back `n' lines. */
295void
296toomuch(FILE *ofp, long n)
297{
298	char buf[BUFSIZ];
299	size_t i, nread;
300
301	if (overfile != NULL) {
302		/*
303		 * Truncate the previous file we overflowed into back to
304		 * the correct length, close it.
305		 */
306		if (fflush(overfile) != 0)
307			err(1, "overflow");
308		if (ftruncate(fileno(overfile), truncofs) != 0)
309			err(1, "overflow");
310		if (fclose(overfile) != 0)
311			err(1, "overflow");
312		overfile = NULL;
313	}
314
315	if (n == 0)
316		/* Just tidying up */
317		return;
318
319	lineno -= n;
320
321	/*
322	 * Wind the overflow file backwards to `n' lines before the
323	 * current one.
324	 */
325	do {
326		if (ftello(ofp) < (off_t)sizeof(buf))
327			rewind(ofp);
328		else
329			fseeko(ofp, -(off_t)sizeof(buf), SEEK_CUR);
330		if (ferror(ofp))
331			errx(1, "%s: can't seek", currfile);
332		if ((nread = fread(buf, 1, sizeof(buf), ofp)) == 0)
333			errx(1, "can't read overflowed output");
334		if (fseeko(ofp, -(off_t)nread, SEEK_CUR) != 0)
335			err(1, "%s", currfile);
336		for (i = 1; i <= nread; i++)
337			if (buf[nread - i] == '\n' && n-- == 0)
338				break;
339		if (ftello(ofp) == 0)
340			break;
341	} while (n > 0);
342	if (fseeko(ofp, nread - i + 1, SEEK_CUR) != 0)
343		err(1, "%s", currfile);
344
345	/*
346	 * csplit_getline() will read from here. Next call will truncate to
347	 * truncofs in this file.
348	 */
349	overfile = ofp;
350	truncofs = ftello(overfile);
351}
352
353/* Handle splits for /regexp/ and %regexp% patterns. */
354void
355do_rexp(const char *expr)
356{
357	regex_t cre;
358	intmax_t nwritten;
359	long ofs;
360	int first;
361	char *ecopy, *ep, *p, *pofs, *re;
362	FILE *ofp;
363
364	if ((ecopy = strdup(expr)) == NULL)
365		err(1, "strdup");
366
367	re = ecopy + 1;
368	if ((pofs = strrchr(ecopy, *expr)) == NULL || pofs[-1] == '\\')
369		errx(1, "%s: missing trailing %c", expr, *expr);
370	*pofs++ = '\0';
371
372	if (*pofs != '\0') {
373		errno = 0;
374		ofs = strtol(pofs, &ep, 10);
375		if (*ep != '\0' || errno != 0)
376			errx(1, "%s: bad offset", pofs);
377	} else
378		ofs = 0;
379
380	if (regcomp(&cre, re, REG_BASIC|REG_NOSUB) != 0)
381		errx(1, "%s: bad regular expression", re);
382
383	if (*expr == '/')
384		/* /regexp/: Save results to a file. */
385		ofp = newfile();
386	else {
387		/* %regexp%: Make a temporary file for overflow. */
388		if ((ofp = tmpfile()) == NULL)
389			err(1, "tmpfile");
390	}
391
392	/* Read and output lines until we get a match. */
393	first = 1;
394	while ((p = csplit_getline()) != NULL) {
395		if (fputs(p, ofp) == EOF)
396			break;
397		if (!first && regexec(&cre, p, 0, NULL, 0) == 0)
398			break;
399		first = 0;
400	}
401
402	if (p == NULL)
403		errx(1, "%s: no match", re);
404
405	if (ofs <= 0) {
406		/*
407		 * Negative (or zero) offset: throw back any lines we should
408		 * not have read yet.
409		  */
410		if (p != NULL) {
411			toomuch(ofp, -ofs + 1);
412			nwritten = (intmax_t)truncofs;
413		} else
414			nwritten = (intmax_t)ftello(ofp);
415	} else {
416		/*
417		 * Positive offset: copy the requested number of lines
418		 * after the match.
419		 */
420		while (--ofs > 0 && (p = csplit_getline()) != NULL)
421			fputs(p, ofp);
422		toomuch(NULL, 0);
423		nwritten = (intmax_t)ftello(ofp);
424		if (fclose(ofp) != 0)
425			err(1, "%s", currfile);
426	}
427
428	if (!sflag && *expr == '/')
429		printf("%jd\n", nwritten);
430
431	regfree(&cre);
432	free(ecopy);
433}
434
435/* Handle splits based on line number. */
436void
437do_lineno(const char *expr)
438{
439	long lastline, tgtline;
440	char *ep, *p;
441	FILE *ofp;
442
443	errno = 0;
444	tgtline = strtol(expr, &ep, 10);
445	if (tgtline <= 0 || errno != 0 || *ep != '\0')
446		errx(1, "%s: bad line number", expr);
447	lastline = tgtline;
448	if (lastline <= lineno)
449		errx(1, "%s: can't go backwards", expr);
450
451	while (nfiles < maxfiles - 1) {
452		ofp = newfile();
453		while (lineno + 1 != lastline) {
454			if ((p = csplit_getline()) == NULL)
455				errx(1, "%ld: out of range", lastline);
456			if (fputs(p, ofp) == EOF)
457				break;
458		}
459		if (!sflag)
460			printf("%jd\n", (intmax_t)ftello(ofp));
461		if (fclose(ofp) != 0)
462			err(1, "%s", currfile);
463		if (reps-- == 0)
464			break;
465		lastline += tgtline;
466	}
467}
468