1/*-
2 * Copyright (c) 1990, 1993
3 *	The Regents of the University of California.  All rights reserved.
4 *
5 * This code is derived from software contributed to Berkeley by
6 * John B. Roll Jr.
7 *
8 * Redistribution and use in source and binary forms, with or without
9 * modification, are permitted provided that the following conditions
10 * are met:
11 * 1. Redistributions of source code must retain the above copyright
12 *    notice, this list of conditions and the following disclaimer.
13 * 2. Redistributions in binary form must reproduce the above copyright
14 *    notice, this list of conditions and the following disclaimer in the
15 *    documentation and/or other materials provided with the distribution.
16 * 4. Neither the name of the University nor the names of its contributors
17 *    may be used to endorse or promote products derived from this software
18 *    without specific prior written permission.
19 *
20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE.
31 *
32 * $xMach: xargs.c,v 1.6 2002/02/23 05:27:47 tim Exp $
33 */
34
35#if 0
36#ifndef lint
37static const char copyright[] =
38"@(#) Copyright (c) 1990, 1993\n\
39	The Regents of the University of California.  All rights reserved.\n";
40#endif /* not lint */
41
42#ifndef lint
43static char sccsid[] = "@(#)xargs.c	8.1 (Berkeley) 6/6/93";
44#endif /* not lint */
45#endif
46#include <sys/cdefs.h>
47__FBSDID("$FreeBSD$");
48
49#include <sys/param.h>
50#include <sys/wait.h>
51
52#include <err.h>
53#include <errno.h>
54#include <fcntl.h>
55#include <langinfo.h>
56#include <locale.h>
57#include <paths.h>
58#include <regex.h>
59#include <stdio.h>
60#include <stdlib.h>
61#include <string.h>
62#include <unistd.h>
63
64#include "pathnames.h"
65
66static void	parse_input(int, char *[]);
67static void	prerun(int, char *[]);
68static int	prompt(void);
69static void	run(char **);
70static void	usage(void);
71void		strnsubst(char **, const char *, const char *, size_t);
72static pid_t	xwait(int block, int *status);
73static void	xexit(const char *, const int);
74static void	waitchildren(const char *, int);
75static void	pids_init(void);
76static int	pids_empty(void);
77static int	pids_full(void);
78static void	pids_add(pid_t pid);
79static int	pids_remove(pid_t pid);
80static int	findslot(pid_t pid);
81static int	findfreeslot(void);
82static void	clearslot(int slot);
83
84static char echo[] = _PATH_ECHO;
85static char **av, **bxp, **ep, **endxp, **xp;
86static char *argp, *bbp, *ebp, *inpline, *p, *replstr;
87static const char *eofstr;
88static int count, insingle, indouble, oflag, pflag, tflag, Rflag, rval, zflag;
89static int cnt, Iflag, jfound, Lflag, Sflag, wasquoted, xflag;
90static int curprocs, maxprocs;
91static pid_t *childpids;
92
93static volatile int childerr;
94
95extern char **environ;
96
97int
98main(int argc, char *argv[])
99{
100	long arg_max;
101	int ch, Jflag, nargs, nflag, nline;
102	size_t linelen;
103	char *endptr;
104
105	inpline = replstr = NULL;
106	ep = environ;
107	eofstr = "";
108	Jflag = nflag = 0;
109
110	(void)setlocale(LC_ALL, "");
111
112	/*
113	 * POSIX.2 limits the exec line length to ARG_MAX - 2K.  Running that
114	 * caused some E2BIG errors, so it was changed to ARG_MAX - 4K.  Given
115	 * that the smallest argument is 2 bytes in length, this means that
116	 * the number of arguments is limited to:
117	 *
118	 *	 (ARG_MAX - 4K - LENGTH(utility + arguments)) / 2.
119	 *
120	 * We arbitrarily limit the number of arguments to 5000.  This is
121	 * allowed by POSIX.2 as long as the resulting minimum exec line is
122	 * at least LINE_MAX.  Realloc'ing as necessary is possible, but
123	 * probably not worthwhile.
124	 */
125	nargs = 5000;
126	if ((arg_max = sysconf(_SC_ARG_MAX)) == -1)
127		errx(1, "sysconf(_SC_ARG_MAX) failed");
128	nline = arg_max - 4 * 1024;
129	while (*ep != NULL) {
130		/* 1 byte for each '\0' */
131		nline -= strlen(*ep++) + 1 + sizeof(*ep);
132	}
133	maxprocs = 1;
134	while ((ch = getopt(argc, argv, "0E:I:J:L:n:oP:pR:S:s:rtx")) != -1)
135		switch (ch) {
136		case 'E':
137			eofstr = optarg;
138			break;
139		case 'I':
140			Jflag = 0;
141			Iflag = 1;
142			Lflag = 1;
143			replstr = optarg;
144			break;
145		case 'J':
146			Iflag = 0;
147			Jflag = 1;
148			replstr = optarg;
149			break;
150		case 'L':
151			Lflag = atoi(optarg);
152			break;
153		case 'n':
154			nflag = 1;
155			if ((nargs = atoi(optarg)) <= 0)
156				errx(1, "illegal argument count");
157			break;
158		case 'o':
159			oflag = 1;
160			break;
161		case 'P':
162			if ((maxprocs = atoi(optarg)) <= 0)
163				errx(1, "max. processes must be >0");
164			break;
165		case 'p':
166			pflag = 1;
167			break;
168		case 'R':
169			Rflag = strtol(optarg, &endptr, 10);
170			if (*endptr != '\0')
171				errx(1, "replacements must be a number");
172			break;
173		case 'r':
174			/* GNU compatibility */
175			break;
176		case 'S':
177			Sflag = strtoul(optarg, &endptr, 10);
178			if (*endptr != '\0')
179				errx(1, "replsize must be a number");
180			break;
181		case 's':
182			nline = atoi(optarg);
183			break;
184		case 't':
185			tflag = 1;
186			break;
187		case 'x':
188			xflag = 1;
189			break;
190		case '0':
191			zflag = 1;
192			break;
193		case '?':
194		default:
195			usage();
196	}
197	argc -= optind;
198	argv += optind;
199
200	if (!Iflag && Rflag)
201		usage();
202	if (!Iflag && Sflag)
203		usage();
204	if (Iflag && !Rflag)
205		Rflag = 5;
206	if (Iflag && !Sflag)
207		Sflag = 255;
208	if (xflag && !nflag)
209		usage();
210	if (Iflag || Lflag)
211		xflag = 1;
212	if (replstr != NULL && *replstr == '\0')
213		errx(1, "replstr may not be empty");
214
215	pids_init();
216
217	/*
218	 * Allocate pointers for the utility name, the utility arguments,
219	 * the maximum arguments to be read from stdin and the trailing
220	 * NULL.
221	 */
222	linelen = 1 + argc + nargs + 1;
223	if ((av = bxp = malloc(linelen * sizeof(char **))) == NULL)
224		errx(1, "malloc failed");
225
226	/*
227	 * Use the user's name for the utility as argv[0], just like the
228	 * shell.  Echo is the default.  Set up pointers for the user's
229	 * arguments.
230	 */
231	if (*argv == NULL)
232		cnt = strlen(*bxp++ = echo);
233	else {
234		do {
235			if (Jflag && strcmp(*argv, replstr) == 0) {
236				char **avj;
237				jfound = 1;
238				argv++;
239				for (avj = argv; *avj; avj++)
240					cnt += strlen(*avj) + 1;
241				break;
242			}
243			cnt += strlen(*bxp++ = *argv) + 1;
244		} while (*++argv != NULL);
245	}
246
247	/*
248	 * Set up begin/end/traversing pointers into the array.  The -n
249	 * count doesn't include the trailing NULL pointer, so the malloc
250	 * added in an extra slot.
251	 */
252	endxp = (xp = bxp) + nargs;
253
254	/*
255	 * Allocate buffer space for the arguments read from stdin and the
256	 * trailing NULL.  Buffer space is defined as the default or specified
257	 * space, minus the length of the utility name and arguments.  Set up
258	 * begin/end/traversing pointers into the array.  The -s count does
259	 * include the trailing NULL, so the malloc didn't add in an extra
260	 * slot.
261	 */
262	nline -= cnt;
263	if (nline <= 0)
264		errx(1, "insufficient space for command");
265
266	if ((bbp = malloc((size_t)(nline + 1))) == NULL)
267		errx(1, "malloc failed");
268	ebp = (argp = p = bbp) + nline - 1;
269	for (;;)
270		parse_input(argc, argv);
271}
272
273static void
274parse_input(int argc, char *argv[])
275{
276	int ch, foundeof;
277	char **avj;
278
279	foundeof = 0;
280
281	switch (ch = getchar()) {
282	case EOF:
283		/* No arguments since last exec. */
284		if (p == bbp)
285			xexit(*av, rval);
286		goto arg1;
287	case ' ':
288	case '\t':
289		/* Quotes escape tabs and spaces. */
290		if (insingle || indouble || zflag)
291			goto addch;
292		goto arg2;
293	case '\0':
294		if (zflag) {
295			/*
296			 * Increment 'count', so that nulls will be treated
297			 * as end-of-line, as well as end-of-argument.  This
298			 * is needed so -0 works properly with -I and -L.
299			 */
300			count++;
301			goto arg2;
302		}
303		goto addch;
304	case '\n':
305		if (zflag)
306			goto addch;
307		count++;	    /* Indicate end-of-line (used by -L) */
308
309		/* Quotes do not escape newlines. */
310arg1:		if (insingle || indouble) {
311			warnx("unterminated quote");
312			xexit(*av, 1);
313		}
314arg2:
315		foundeof = *eofstr != '\0' &&
316		    strncmp(argp, eofstr, p - argp) == 0;
317
318		/* Do not make empty args unless they are quoted */
319		if ((argp != p || wasquoted) && !foundeof) {
320			*p++ = '\0';
321			*xp++ = argp;
322			if (Iflag) {
323				size_t curlen;
324
325				if (inpline == NULL)
326					curlen = 0;
327				else {
328					/*
329					 * If this string is not zero
330					 * length, append a space for
331					 * separation before the next
332					 * argument.
333					 */
334					if ((curlen = strlen(inpline)))
335						strcat(inpline, " ");
336				}
337				curlen++;
338				/*
339				 * Allocate enough to hold what we will
340				 * be holding in a second, and to append
341				 * a space next time through, if we have
342				 * to.
343				 */
344				inpline = realloc(inpline, curlen + 2 +
345				    strlen(argp));
346				if (inpline == NULL) {
347					warnx("realloc failed");
348					xexit(*av, 1);
349				}
350				if (curlen == 1)
351					strcpy(inpline, argp);
352				else
353					strcat(inpline, argp);
354			}
355		}
356
357		/*
358		 * If max'd out on args or buffer, or reached EOF,
359		 * run the command.  If xflag and max'd out on buffer
360		 * but not on args, object.  Having reached the limit
361		 * of input lines, as specified by -L is the same as
362		 * maxing out on arguments.
363		 */
364		if (xp == endxp || p > ebp || ch == EOF ||
365		    (Lflag <= count && xflag) || foundeof) {
366			if (xflag && xp != endxp && p > ebp) {
367				warnx("insufficient space for arguments");
368				xexit(*av, 1);
369			}
370			if (jfound) {
371				for (avj = argv; *avj; avj++)
372					*xp++ = *avj;
373			}
374			prerun(argc, av);
375			if (ch == EOF || foundeof)
376				xexit(*av, rval);
377			p = bbp;
378			xp = bxp;
379			count = 0;
380		}
381		argp = p;
382		wasquoted = 0;
383		break;
384	case '\'':
385		if (indouble || zflag)
386			goto addch;
387		insingle = !insingle;
388		wasquoted = 1;
389		break;
390	case '"':
391		if (insingle || zflag)
392			goto addch;
393		indouble = !indouble;
394		wasquoted = 1;
395		break;
396	case '\\':
397		if (zflag)
398			goto addch;
399		/* Backslash escapes anything, is escaped by quotes. */
400		if (!insingle && !indouble && (ch = getchar()) == EOF) {
401			warnx("backslash at EOF");
402			xexit(*av, 1);
403		}
404		/* FALLTHROUGH */
405	default:
406addch:		if (p < ebp) {
407			*p++ = ch;
408			break;
409		}
410
411		/* If only one argument, not enough buffer space. */
412		if (bxp == xp) {
413			warnx("insufficient space for argument");
414			xexit(*av, 1);
415		}
416		/* Didn't hit argument limit, so if xflag object. */
417		if (xflag) {
418			warnx("insufficient space for arguments");
419			xexit(*av, 1);
420		}
421
422		if (jfound) {
423			for (avj = argv; *avj; avj++)
424				*xp++ = *avj;
425		}
426		prerun(argc, av);
427		xp = bxp;
428		cnt = ebp - argp;
429		memcpy(bbp, argp, (size_t)cnt);
430		p = (argp = bbp) + cnt;
431		*p++ = ch;
432		break;
433	}
434}
435
436/*
437 * Do things necessary before run()'ing, such as -I substitution,
438 * and then call run().
439 */
440static void
441prerun(int argc, char *argv[])
442{
443	char **tmp, **tmp2, **avj;
444	int repls;
445
446	repls = Rflag;
447
448	if (argc == 0 || repls == 0) {
449		*xp = NULL;
450		run(argv);
451		return;
452	}
453
454	avj = argv;
455
456	/*
457	 * Allocate memory to hold the argument list, and
458	 * a NULL at the tail.
459	 */
460	tmp = malloc((argc + 1) * sizeof(char**));
461	if (tmp == NULL) {
462		warnx("malloc failed");
463		xexit(*argv, 1);
464	}
465	tmp2 = tmp;
466
467	/*
468	 * Save the first argument and iterate over it, we
469	 * cannot do strnsubst() to it.
470	 */
471	if ((*tmp++ = strdup(*avj++)) == NULL) {
472		warnx("strdup failed");
473		xexit(*argv, 1);
474	}
475
476	/*
477	 * For each argument to utility, if we have not used up
478	 * the number of replacements we are allowed to do, and
479	 * if the argument contains at least one occurrence of
480	 * replstr, call strnsubst(), else just save the string.
481	 * Iterations over elements of avj and tmp are done
482	 * where appropriate.
483	 */
484	while (--argc) {
485		*tmp = *avj++;
486		if (repls && strstr(*tmp, replstr) != NULL) {
487			strnsubst(tmp++, replstr, inpline, (size_t)Sflag);
488			if (repls > 0)
489				repls--;
490		} else {
491			if ((*tmp = strdup(*tmp)) == NULL) {
492				warnx("strdup failed");
493				xexit(*argv, 1);
494			}
495			tmp++;
496		}
497	}
498
499	/*
500	 * Run it.
501	 */
502	*tmp = NULL;
503	run(tmp2);
504
505	/*
506	 * Walk from the tail to the head, free along the way.
507	 */
508	for (; tmp2 != tmp; tmp--)
509		free(*tmp);
510	/*
511	 * Now free the list itself.
512	 */
513	free(tmp2);
514
515	/*
516	 * Free the input line buffer, if we have one.
517	 */
518	if (inpline != NULL) {
519		free(inpline);
520		inpline = NULL;
521	}
522}
523
524static void
525run(char **argv)
526{
527	pid_t pid;
528	int fd;
529	char **avec;
530
531	/*
532	 * If the user wants to be notified of each command before it is
533	 * executed, notify them.  If they want the notification to be
534	 * followed by a prompt, then prompt them.
535	 */
536	if (tflag || pflag) {
537		(void)fprintf(stderr, "%s", *argv);
538		for (avec = argv + 1; *avec != NULL; ++avec)
539			(void)fprintf(stderr, " %s", *avec);
540		/*
541		 * If the user has asked to be prompted, do so.
542		 */
543		if (pflag)
544			/*
545			 * If they asked not to exec, return without execution
546			 * but if they asked to, go to the execution.  If we
547			 * could not open their tty, break the switch and drop
548			 * back to -t behaviour.
549			 */
550			switch (prompt()) {
551			case 0:
552				return;
553			case 1:
554				goto exec;
555			case 2:
556				break;
557			}
558		(void)fprintf(stderr, "\n");
559		(void)fflush(stderr);
560	}
561exec:
562	childerr = 0;
563	switch (pid = vfork()) {
564	case -1:
565		warn("vfork");
566		xexit(*argv, 1);
567	case 0:
568		if (oflag) {
569			if ((fd = open(_PATH_TTY, O_RDONLY)) == -1)
570				err(1, "can't open /dev/tty");
571		} else {
572			fd = open(_PATH_DEVNULL, O_RDONLY);
573		}
574		if (fd > STDIN_FILENO) {
575			if (dup2(fd, STDIN_FILENO) != 0)
576				err(1, "can't dup2 to stdin");
577			close(fd);
578		}
579		execvp(argv[0], argv);
580		childerr = errno;
581		_exit(1);
582	}
583	pids_add(pid);
584	waitchildren(*argv, 0);
585}
586
587/*
588 * Wait for a tracked child to exit and return its pid and exit status.
589 *
590 * Ignores (discards) all untracked child processes.
591 * Returns -1 and sets errno to ECHILD if no tracked children exist.
592 * If block is set, waits indefinitely for a child process to exit.
593 * If block is not set and no children have exited, returns 0 immediately.
594 */
595static pid_t
596xwait(int block, int *status) {
597	pid_t pid;
598
599	if (pids_empty()) {
600		errno = ECHILD;
601		return (-1);
602	}
603
604	while ((pid = waitpid(-1, status, block ? 0 : WNOHANG)) > 0)
605		if (pids_remove(pid))
606			break;
607
608	return (pid);
609}
610
611static void
612xexit(const char *name, const int exit_code) {
613	waitchildren(name, 1);
614	exit(exit_code);
615}
616
617static void
618waitchildren(const char *name, int waitall)
619{
620	pid_t pid;
621	int status;
622	int cause_exit = 0;
623
624	while ((pid = xwait(waitall || pids_full(), &status)) > 0) {
625		/*
626		 * If we couldn't invoke the utility or if utility exited
627		 * because of a signal or with a value of 255, warn (per
628		 * POSIX), and then wait until all other children have
629		 * exited before exiting 1-125. POSIX requires us to stop
630		 * reading if child exits because of a signal or with 255,
631		 * but it does not require us to exit immediately; waiting
632		 * is preferable to orphaning.
633		 */
634		if (childerr != 0 && cause_exit == 0) {
635			errno = childerr;
636			waitall = 1;
637			cause_exit = ENOENT ? 127 : 126;
638			warn("%s", name);
639		} else if (WIFSIGNALED(status)) {
640			waitall = cause_exit = 1;
641			warnx("%s: terminated with signal %d; aborting",
642			    name, WTERMSIG(status));
643		} else if (WEXITSTATUS(status) == 255) {
644			waitall = cause_exit = 1;
645			warnx("%s: exited with status 255; aborting", name);
646		} else if (WEXITSTATUS(status))
647 			rval = 1;
648	}
649
650 	if (cause_exit)
651		exit(cause_exit);
652	if (pid == -1 && errno != ECHILD)
653		err(1, "waitpid");
654}
655
656#define	NOPID	(0)
657
658static void
659pids_init(void)
660{
661	int i;
662
663	if ((childpids = malloc(maxprocs * sizeof(*childpids))) == NULL)
664		errx(1, "malloc failed");
665
666	for (i = 0; i < maxprocs; i++)
667		clearslot(i);
668}
669
670static int
671pids_empty(void)
672{
673
674	return (curprocs == 0);
675}
676
677static int
678pids_full(void)
679{
680
681	return (curprocs >= maxprocs);
682}
683
684static void
685pids_add(pid_t pid)
686{
687	int slot;
688
689	slot = findfreeslot();
690	childpids[slot] = pid;
691	curprocs++;
692}
693
694static int
695pids_remove(pid_t pid)
696{
697	int slot;
698
699	if ((slot = findslot(pid)) < 0)
700		return (0);
701
702	clearslot(slot);
703	curprocs--;
704	return (1);
705}
706
707static int
708findfreeslot(void)
709{
710	int slot;
711
712	if ((slot = findslot(NOPID)) < 0)
713		errx(1, "internal error: no free pid slot");
714	return (slot);
715}
716
717static int
718findslot(pid_t pid)
719{
720	int slot;
721
722	for (slot = 0; slot < maxprocs; slot++)
723		if (childpids[slot] == pid)
724			return (slot);
725	return (-1);
726}
727
728static void
729clearslot(int slot)
730{
731
732	childpids[slot] = NOPID;
733}
734
735/*
736 * Prompt the user about running a command.
737 */
738static int
739prompt(void)
740{
741	regex_t cre;
742	size_t rsize;
743	int match;
744	char *response;
745	FILE *ttyfp;
746
747	if ((ttyfp = fopen(_PATH_TTY, "r")) == NULL)
748		return (2);	/* Indicate that the TTY failed to open. */
749	(void)fprintf(stderr, "?...");
750	(void)fflush(stderr);
751	if ((response = fgetln(ttyfp, &rsize)) == NULL ||
752	    regcomp(&cre, nl_langinfo(YESEXPR), REG_BASIC) != 0) {
753		(void)fclose(ttyfp);
754		return (0);
755	}
756	response[rsize - 1] = '\0';
757	match = regexec(&cre, response, 0, NULL, 0);
758	(void)fclose(ttyfp);
759	regfree(&cre);
760	return (match == 0);
761}
762
763static void
764usage(void)
765{
766
767	fprintf(stderr,
768"usage: xargs [-0opt] [-E eofstr] [-I replstr [-R replacements] [-S replsize]]\n"
769"             [-J replstr] [-L number] [-n number [-x]] [-P maxprocs]\n"
770"             [-s size] [utility [argument ...]]\n");
771	exit(1);
772}
773