1/*-
2 * SPDX-License-Identifier: BSD-3-Clause
3 *
4 * Copyright (c) 1990, 1993
5 *	The Regents of the University of California.  All rights reserved.
6 *
7 * This code is derived from software contributed to Berkeley by
8 * John B. Roll Jr.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 *    notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 *    notice, this list of conditions and the following disclaimer in the
17 *    documentation and/or other materials provided with the distribution.
18 * 3. Neither the name of the University nor the names of its contributors
19 *    may be used to endorse or promote products derived from this software
20 *    without specific prior written permission.
21 *
22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * SUCH DAMAGE.
33 *
34 * $xMach: xargs.c,v 1.6 2002/02/23 05:27:47 tim Exp $
35 */
36
37#if 0
38#ifndef lint
39static const char copyright[] =
40"@(#) Copyright (c) 1990, 1993\n\
41	The Regents of the University of California.  All rights reserved.\n";
42#endif /* not lint */
43
44#ifndef lint
45static char sccsid[] = "@(#)xargs.c	8.1 (Berkeley) 6/6/93";
46#endif /* not lint */
47#endif
48#include <sys/cdefs.h>
49__FBSDID("$FreeBSD$");
50
51#include <sys/types.h>
52#include <sys/wait.h>
53#include <sys/time.h>
54#include <sys/limits.h>
55#include <sys/resource.h>
56#include <err.h>
57#include <errno.h>
58#include <fcntl.h>
59#include <getopt.h>
60#include <langinfo.h>
61#include <locale.h>
62#include <paths.h>
63#include <regex.h>
64#include <stdio.h>
65#include <stdlib.h>
66#include <string.h>
67#include <unistd.h>
68
69#include "pathnames.h"
70
71static void	parse_input(int, char *[]);
72static void	prerun(int, char *[]);
73static int	prompt(void);
74static void	run(char **);
75static void	usage(void);
76void		strnsubst(char **, const char *, const char *, size_t);
77static pid_t	xwait(int block, int *status);
78static void	xexit(const char *, const int);
79static void	waitchildren(const char *, int);
80static void	pids_init(void);
81static int	pids_empty(void);
82static int	pids_full(void);
83static void	pids_add(pid_t pid);
84static int	pids_remove(pid_t pid);
85static int	findslot(pid_t pid);
86static int	findfreeslot(void);
87static void	clearslot(int slot);
88
89static char echo[] = _PATH_ECHO;
90static char **av, **bxp, **ep, **endxp, **xp;
91static char *argp, *bbp, *ebp, *inpline, *p, *replstr;
92static const char *eofstr;
93static int count, insingle, indouble, oflag, pflag, tflag, Rflag, rval, zflag;
94static int cnt, Iflag, jfound, Lflag, Sflag, wasquoted, xflag;
95static int curprocs, maxprocs;
96static pid_t *childpids;
97
98static volatile int childerr;
99
100extern char **environ;
101
102static const char *optstr = "+0E:I:J:L:n:oP:pR:S:s:rtx";
103
104static const struct option long_options[] =
105{
106	{"exit",		no_argument,		NULL,	'x'},
107	{"interactive",		no_argument,		NULL,	'p'},
108	{"max-args",		required_argument,	NULL,	'n'},
109	{"max-chars",		required_argument,	NULL,	's'},
110	{"max-procs",		required_argument,	NULL,	'P'},
111	{"no-run-if-empty",	no_argument,		NULL,	'r'},
112	{"null",		no_argument,		NULL,	'0'},
113	{"verbose",		no_argument,		NULL,	't'},
114
115	{NULL,			no_argument,		NULL,	0},
116};
117
118int
119main(int argc, char *argv[])
120{
121	long arg_max;
122	int ch, Jflag, nargs, nflag, nline;
123	size_t linelen;
124	struct rlimit rl;
125	char *endptr;
126	const char *errstr;
127
128	inpline = replstr = NULL;
129	ep = environ;
130	eofstr = "";
131	Jflag = nflag = 0;
132
133	(void)setlocale(LC_ALL, "");
134
135	/*
136	 * POSIX.2 limits the exec line length to ARG_MAX - 2K.  Running that
137	 * caused some E2BIG errors, so it was changed to ARG_MAX - 4K.  Given
138	 * that the smallest argument is 2 bytes in length, this means that
139	 * the number of arguments is limited to:
140	 *
141	 *	 (ARG_MAX - 4K - LENGTH(utility + arguments)) / 2.
142	 *
143	 * We arbitrarily limit the number of arguments to 5000.  This is
144	 * allowed by POSIX.2 as long as the resulting minimum exec line is
145	 * at least LINE_MAX.  Realloc'ing as necessary is possible, but
146	 * probably not worthwhile.
147	 */
148	nargs = 5000;
149	if ((arg_max = sysconf(_SC_ARG_MAX)) == -1)
150		errx(1, "sysconf(_SC_ARG_MAX) failed");
151	nline = arg_max - 4 * 1024;
152	while (*ep != NULL) {
153		/* 1 byte for each '\0' */
154		nline -= strlen(*ep++) + 1 + sizeof(*ep);
155	}
156	maxprocs = 1;
157	while ((ch = getopt_long(argc, argv, optstr, long_options, NULL)) != -1)
158		switch (ch) {
159		case 'E':
160			eofstr = optarg;
161			break;
162		case 'I':
163			Jflag = 0;
164			Iflag = 1;
165			Lflag = 1;
166			replstr = optarg;
167			break;
168		case 'J':
169			Iflag = 0;
170			Jflag = 1;
171			replstr = optarg;
172			break;
173		case 'L':
174			Lflag = strtonum(optarg, 0, INT_MAX, &errstr);
175			if (errstr)
176				errx(1, "-L %s: %s", optarg, errstr);
177			break;
178		case 'n':
179			nflag = 1;
180			nargs = strtonum(optarg, 1, INT_MAX, &errstr);
181			if (errstr)
182				errx(1, "-n %s: %s", optarg, errstr);
183			break;
184		case 'o':
185			oflag = 1;
186			break;
187		case 'P':
188			maxprocs = strtonum(optarg, 0, INT_MAX, &errstr);
189			if (errstr)
190				errx(1, "-P %s: %s", optarg, errstr);
191			if (getrlimit(RLIMIT_NPROC, &rl) != 0)
192				errx(1, "getrlimit failed");
193			if (maxprocs == 0 || maxprocs > rl.rlim_cur)
194				maxprocs = rl.rlim_cur;
195			break;
196		case 'p':
197			pflag = 1;
198			break;
199		case 'R':
200			Rflag = strtol(optarg, &endptr, 10);
201			if (*endptr != '\0')
202				errx(1, "replacements must be a number");
203			break;
204		case 'r':
205			/* GNU compatibility */
206			break;
207		case 'S':
208			Sflag = strtoul(optarg, &endptr, 10);
209			if (*endptr != '\0')
210				errx(1, "replsize must be a number");
211			break;
212		case 's':
213			nline = strtonum(optarg, 0, INT_MAX, &errstr);
214			if (errstr)
215				errx(1, "-s %s: %s", optarg, errstr);
216			break;
217		case 't':
218			tflag = 1;
219			break;
220		case 'x':
221			xflag = 1;
222			break;
223		case '0':
224			zflag = 1;
225			break;
226		case '?':
227		default:
228			usage();
229	}
230	argc -= optind;
231	argv += optind;
232
233	if (!Iflag && Rflag)
234		usage();
235	if (!Iflag && Sflag)
236		usage();
237	if (Iflag && !Rflag)
238		Rflag = 5;
239	if (Iflag && !Sflag)
240		Sflag = 255;
241	if (xflag && !nflag)
242		usage();
243	if (Iflag || Lflag)
244		xflag = 1;
245	if (replstr != NULL && *replstr == '\0')
246		errx(1, "replstr may not be empty");
247
248	pids_init();
249
250	/*
251	 * Allocate pointers for the utility name, the utility arguments,
252	 * the maximum arguments to be read from stdin and the trailing
253	 * NULL.
254	 */
255	linelen = 1 + argc + nargs + 1;
256	if ((av = bxp = malloc(linelen * sizeof(char *))) == NULL)
257		errx(1, "malloc failed");
258
259	/*
260	 * Use the user's name for the utility as argv[0], just like the
261	 * shell.  Echo is the default.  Set up pointers for the user's
262	 * arguments.
263	 */
264	if (*argv == NULL)
265		cnt = strlen(*bxp++ = echo);
266	else {
267		do {
268			if (Jflag && strcmp(*argv, replstr) == 0) {
269				char **avj;
270				jfound = 1;
271				argv++;
272				for (avj = argv; *avj; avj++)
273					cnt += strlen(*avj) + 1;
274				break;
275			}
276			cnt += strlen(*bxp++ = *argv) + 1;
277		} while (*++argv != NULL);
278	}
279
280	/*
281	 * Set up begin/end/traversing pointers into the array.  The -n
282	 * count doesn't include the trailing NULL pointer, so the malloc
283	 * added in an extra slot.
284	 */
285	endxp = (xp = bxp) + nargs;
286
287	/*
288	 * Allocate buffer space for the arguments read from stdin and the
289	 * trailing NULL.  Buffer space is defined as the default or specified
290	 * space, minus the length of the utility name and arguments.  Set up
291	 * begin/end/traversing pointers into the array.  The -s count does
292	 * include the trailing NULL, so the malloc didn't add in an extra
293	 * slot.
294	 */
295	nline -= cnt;
296	if (nline <= 0)
297		errx(1, "insufficient space for command");
298
299	if ((bbp = malloc((size_t)(nline + 1))) == NULL)
300		errx(1, "malloc failed");
301	ebp = (argp = p = bbp) + nline - 1;
302	for (;;)
303		parse_input(argc, argv);
304}
305
306static void
307parse_input(int argc, char *argv[])
308{
309	int ch, foundeof;
310	char **avj;
311
312	foundeof = 0;
313
314	switch (ch = getchar()) {
315	case EOF:
316		/* No arguments since last exec. */
317		if (p == bbp)
318			xexit(*av, rval);
319		goto arg1;
320	case ' ':
321	case '\t':
322		/* Quotes escape tabs and spaces. */
323		if (insingle || indouble || zflag)
324			goto addch;
325		goto arg2;
326	case '\0':
327		if (zflag) {
328			/*
329			 * Increment 'count', so that nulls will be treated
330			 * as end-of-line, as well as end-of-argument.  This
331			 * is needed so -0 works properly with -I and -L.
332			 */
333			count++;
334			goto arg2;
335		}
336		goto addch;
337	case '\n':
338		if (zflag)
339			goto addch;
340		count++;	    /* Indicate end-of-line (used by -L) */
341
342		/* Quotes do not escape newlines. */
343arg1:		if (insingle || indouble) {
344			warnx("unterminated quote");
345			xexit(*av, 1);
346		}
347arg2:
348		foundeof = *eofstr != '\0' &&
349		    strncmp(argp, eofstr, p - argp) == 0;
350
351		/* Do not make empty args unless they are quoted */
352		if ((argp != p || wasquoted) && !foundeof) {
353			*p++ = '\0';
354			*xp++ = argp;
355			if (Iflag) {
356				size_t curlen;
357
358				if (inpline == NULL)
359					curlen = 0;
360				else {
361					/*
362					 * If this string is not zero
363					 * length, append a space for
364					 * separation before the next
365					 * argument.
366					 */
367					if ((curlen = strlen(inpline)))
368						strcat(inpline, " ");
369				}
370				curlen++;
371				/*
372				 * Allocate enough to hold what we will
373				 * be holding in a second, and to append
374				 * a space next time through, if we have
375				 * to.
376				 */
377				inpline = realloc(inpline, curlen + 2 +
378				    strlen(argp));
379				if (inpline == NULL) {
380					warnx("realloc failed");
381					xexit(*av, 1);
382				}
383				if (curlen == 1)
384					strcpy(inpline, argp);
385				else
386					strcat(inpline, argp);
387			}
388		}
389
390		/*
391		 * If max'd out on args or buffer, or reached EOF,
392		 * run the command.  If xflag and max'd out on buffer
393		 * but not on args, object.  Having reached the limit
394		 * of input lines, as specified by -L is the same as
395		 * maxing out on arguments.
396		 */
397		if (xp == endxp || p > ebp || ch == EOF ||
398		    (Lflag <= count && xflag) || foundeof) {
399			if (xflag && xp != endxp && p > ebp) {
400				warnx("insufficient space for arguments");
401				xexit(*av, 1);
402			}
403			if (jfound) {
404				for (avj = argv; *avj; avj++)
405					*xp++ = *avj;
406			}
407			prerun(argc, av);
408			if (ch == EOF || foundeof)
409				xexit(*av, rval);
410			p = bbp;
411			xp = bxp;
412			count = 0;
413		}
414		argp = p;
415		wasquoted = 0;
416		break;
417	case '\'':
418		if (indouble || zflag)
419			goto addch;
420		insingle = !insingle;
421		wasquoted = 1;
422		break;
423	case '"':
424		if (insingle || zflag)
425			goto addch;
426		indouble = !indouble;
427		wasquoted = 1;
428		break;
429	case '\\':
430		if (zflag)
431			goto addch;
432		/* Backslash escapes anything, is escaped by quotes. */
433		if (!insingle && !indouble && (ch = getchar()) == EOF) {
434			warnx("backslash at EOF");
435			xexit(*av, 1);
436		}
437		/* FALLTHROUGH */
438	default:
439addch:		if (p < ebp) {
440			*p++ = ch;
441			break;
442		}
443
444		/* If only one argument, not enough buffer space. */
445		if (bxp == xp) {
446			warnx("insufficient space for argument");
447			xexit(*av, 1);
448		}
449		/* Didn't hit argument limit, so if xflag object. */
450		if (xflag) {
451			warnx("insufficient space for arguments");
452			xexit(*av, 1);
453		}
454
455		if (jfound) {
456			for (avj = argv; *avj; avj++)
457				*xp++ = *avj;
458		}
459		prerun(argc, av);
460		xp = bxp;
461		cnt = ebp - argp;
462		memcpy(bbp, argp, (size_t)cnt);
463		p = (argp = bbp) + cnt;
464		*p++ = ch;
465		break;
466	}
467}
468
469/*
470 * Do things necessary before run()'ing, such as -I substitution,
471 * and then call run().
472 */
473static void
474prerun(int argc, char *argv[])
475{
476	char **tmp, **tmp2, **avj;
477	int repls;
478
479	repls = Rflag;
480
481	if (argc == 0 || repls == 0) {
482		*xp = NULL;
483		run(argv);
484		return;
485	}
486
487	avj = argv;
488
489	/*
490	 * Allocate memory to hold the argument list, and
491	 * a NULL at the tail.
492	 */
493	tmp = malloc((argc + 1) * sizeof(char *));
494	if (tmp == NULL) {
495		warnx("malloc failed");
496		xexit(*argv, 1);
497	}
498	tmp2 = tmp;
499
500	/*
501	 * Save the first argument and iterate over it, we
502	 * cannot do strnsubst() to it.
503	 */
504	if ((*tmp++ = strdup(*avj++)) == NULL) {
505		warnx("strdup failed");
506		xexit(*argv, 1);
507	}
508
509	/*
510	 * For each argument to utility, if we have not used up
511	 * the number of replacements we are allowed to do, and
512	 * if the argument contains at least one occurrence of
513	 * replstr, call strnsubst(), else just save the string.
514	 * Iterations over elements of avj and tmp are done
515	 * where appropriate.
516	 */
517	while (--argc) {
518		*tmp = *avj++;
519		if (repls && strstr(*tmp, replstr) != NULL) {
520			strnsubst(tmp++, replstr, inpline, (size_t)Sflag);
521			if (repls > 0)
522				repls--;
523		} else {
524			if ((*tmp = strdup(*tmp)) == NULL) {
525				warnx("strdup failed");
526				xexit(*argv, 1);
527			}
528			tmp++;
529		}
530	}
531
532	/*
533	 * Run it.
534	 */
535	*tmp = NULL;
536	run(tmp2);
537
538	/*
539	 * Walk from the tail to the head, free along the way.
540	 */
541	for (; tmp2 != tmp; tmp--)
542		free(*tmp);
543	/*
544	 * Now free the list itself.
545	 */
546	free(tmp2);
547
548	/*
549	 * Free the input line buffer, if we have one.
550	 */
551	if (inpline != NULL) {
552		free(inpline);
553		inpline = NULL;
554	}
555}
556
557static void
558run(char **argv)
559{
560	pid_t pid;
561	int fd;
562	char **avec;
563
564	/*
565	 * If the user wants to be notified of each command before it is
566	 * executed, notify them.  If they want the notification to be
567	 * followed by a prompt, then prompt them.
568	 */
569	if (tflag || pflag) {
570		(void)fprintf(stderr, "%s", *argv);
571		for (avec = argv + 1; *avec != NULL; ++avec)
572			(void)fprintf(stderr, " %s", *avec);
573		/*
574		 * If the user has asked to be prompted, do so.
575		 */
576		if (pflag)
577			/*
578			 * If they asked not to exec, return without execution
579			 * but if they asked to, go to the execution.  If we
580			 * could not open their tty, break the switch and drop
581			 * back to -t behaviour.
582			 */
583			switch (prompt()) {
584			case 0:
585				return;
586			case 1:
587				goto exec;
588			case 2:
589				break;
590			}
591		(void)fprintf(stderr, "\n");
592		(void)fflush(stderr);
593	}
594exec:
595	childerr = 0;
596	switch (pid = vfork()) {
597	case -1:
598		warn("vfork");
599		xexit(*argv, 1);
600	case 0:
601		if (oflag) {
602			if ((fd = open(_PATH_TTY, O_RDONLY)) == -1)
603				err(1, "can't open /dev/tty");
604		} else {
605			fd = open(_PATH_DEVNULL, O_RDONLY);
606		}
607		if (fd > STDIN_FILENO) {
608			if (dup2(fd, STDIN_FILENO) != 0)
609				err(1, "can't dup2 to stdin");
610			close(fd);
611		}
612		execvp(argv[0], argv);
613		childerr = errno;
614		_exit(1);
615	}
616	pids_add(pid);
617	waitchildren(*argv, 0);
618}
619
620/*
621 * Wait for a tracked child to exit and return its pid and exit status.
622 *
623 * Ignores (discards) all untracked child processes.
624 * Returns -1 and sets errno to ECHILD if no tracked children exist.
625 * If block is set, waits indefinitely for a child process to exit.
626 * If block is not set and no children have exited, returns 0 immediately.
627 */
628static pid_t
629xwait(int block, int *status) {
630	pid_t pid;
631
632	if (pids_empty()) {
633		errno = ECHILD;
634		return (-1);
635	}
636
637	while ((pid = waitpid(-1, status, block ? 0 : WNOHANG)) > 0)
638		if (pids_remove(pid))
639			break;
640
641	return (pid);
642}
643
644static void
645xexit(const char *name, const int exit_code) {
646	waitchildren(name, 1);
647	exit(exit_code);
648}
649
650static void
651waitchildren(const char *name, int waitall)
652{
653	pid_t pid;
654	int status;
655	int cause_exit = 0;
656
657	while ((pid = xwait(waitall || pids_full(), &status)) > 0) {
658		/*
659		 * If we couldn't invoke the utility or if utility exited
660		 * because of a signal or with a value of 255, warn (per
661		 * POSIX), and then wait until all other children have
662		 * exited before exiting 1-125. POSIX requires us to stop
663		 * reading if child exits because of a signal or with 255,
664		 * but it does not require us to exit immediately; waiting
665		 * is preferable to orphaning.
666		 */
667		if (childerr != 0 && cause_exit == 0) {
668			errno = childerr;
669			waitall = 1;
670			cause_exit = errno == ENOENT ? 127 : 126;
671			warn("%s", name);
672		} else if (WIFSIGNALED(status)) {
673			waitall = cause_exit = 1;
674			warnx("%s: terminated with signal %d; aborting",
675			    name, WTERMSIG(status));
676		} else if (WEXITSTATUS(status) == 255) {
677			waitall = cause_exit = 1;
678			warnx("%s: exited with status 255; aborting", name);
679		} else if (WEXITSTATUS(status))
680 			rval = 1;
681	}
682
683 	if (cause_exit)
684		exit(cause_exit);
685	if (pid == -1 && errno != ECHILD)
686		err(1, "waitpid");
687}
688
689#define	NOPID	(0)
690
691static void
692pids_init(void)
693{
694	int i;
695
696	if ((childpids = malloc(maxprocs * sizeof(*childpids))) == NULL)
697		errx(1, "malloc failed");
698
699	for (i = 0; i < maxprocs; i++)
700		clearslot(i);
701}
702
703static int
704pids_empty(void)
705{
706
707	return (curprocs == 0);
708}
709
710static int
711pids_full(void)
712{
713
714	return (curprocs >= maxprocs);
715}
716
717static void
718pids_add(pid_t pid)
719{
720	int slot;
721
722	slot = findfreeslot();
723	childpids[slot] = pid;
724	curprocs++;
725}
726
727static int
728pids_remove(pid_t pid)
729{
730	int slot;
731
732	if ((slot = findslot(pid)) < 0)
733		return (0);
734
735	clearslot(slot);
736	curprocs--;
737	return (1);
738}
739
740static int
741findfreeslot(void)
742{
743	int slot;
744
745	if ((slot = findslot(NOPID)) < 0)
746		errx(1, "internal error: no free pid slot");
747	return (slot);
748}
749
750static int
751findslot(pid_t pid)
752{
753	int slot;
754
755	for (slot = 0; slot < maxprocs; slot++)
756		if (childpids[slot] == pid)
757			return (slot);
758	return (-1);
759}
760
761static void
762clearslot(int slot)
763{
764
765	childpids[slot] = NOPID;
766}
767
768/*
769 * Prompt the user about running a command.
770 */
771static int
772prompt(void)
773{
774	regex_t cre;
775	size_t rsize;
776	int match;
777	char *response;
778	FILE *ttyfp;
779
780	if ((ttyfp = fopen(_PATH_TTY, "r")) == NULL)
781		return (2);	/* Indicate that the TTY failed to open. */
782	(void)fprintf(stderr, "?...");
783	(void)fflush(stderr);
784	if ((response = fgetln(ttyfp, &rsize)) == NULL ||
785	    regcomp(&cre, nl_langinfo(YESEXPR), REG_EXTENDED) != 0) {
786		(void)fclose(ttyfp);
787		return (0);
788	}
789	response[rsize - 1] = '\0';
790	match = regexec(&cre, response, 0, NULL, 0);
791	(void)fclose(ttyfp);
792	regfree(&cre);
793	return (match == 0);
794}
795
796static void
797usage(void)
798{
799
800	fprintf(stderr,
801"usage: xargs [-0opt] [-E eofstr] [-I replstr [-R replacements] [-S replsize]]\n"
802"             [-J replstr] [-L number] [-n number [-x]] [-P maxprocs]\n"
803"             [-s size] [utility [argument ...]]\n");
804	exit(1);
805}
806