1/* vi: set sw=4 ts=4: */
2/*
3 * Mini xargs implementation for busybox
4 *
5 * (C) 2002,2003 by Vladimir Oleynik <dzo@simtreas.ru>
6 *
7 * Special thanks
8 * - Mark Whitley and Glenn McGrath for stimulus to rewrite :)
9 * - Mike Rendell <michael@cs.mun.ca>
10 * and David MacKenzie <djm@gnu.ai.mit.edu>.
11 *
12 * Licensed under the GPL v2 or later, see the file LICENSE in this tarball.
13 *
14 * xargs is described in the Single Unix Specification v3 at
15 * http://www.opengroup.org/onlinepubs/007904975/utilities/xargs.html
16 */
17
18//applet:IF_XARGS(APPLET_NOEXEC(xargs, xargs, _BB_DIR_USR_BIN, _BB_SUID_DROP, xargs))
19
20//kbuild:lib-$(CONFIG_XARGS) += xargs.o
21
22//config:config XARGS
23//config:	bool "xargs"
24//config:	default y
25//config:	help
26//config:	  xargs is used to execute a specified command for
27//config:	  every item from standard input.
28//config:
29//config:config FEATURE_XARGS_SUPPORT_CONFIRMATION
30//config:	bool "Enable -p: prompt and confirmation"
31//config:	default y
32//config:	depends on XARGS
33//config:	help
34//config:	  Support -p: prompt the user whether to run each command
35//config:	  line and read a line from the terminal.
36//config:
37//config:config FEATURE_XARGS_SUPPORT_QUOTES
38//config:	bool "Enable single and double quotes and backslash"
39//config:	default y
40//config:	depends on XARGS
41//config:	help
42//config:	  Support quoting in the input.
43//config:
44//config:config FEATURE_XARGS_SUPPORT_TERMOPT
45//config:	bool "Enable -x: exit if -s or -n is exceeded"
46//config:	default y
47//config:	depends on XARGS
48//config:	help
49//config:	  Support -x: exit if the command size (see the -s or -n option)
50//config:	  is exceeded.
51//config:
52//config:config FEATURE_XARGS_SUPPORT_ZERO_TERM
53//config:	bool "Enable -0: NUL-terminated input"
54//config:	default y
55//config:	depends on XARGS
56//config:	help
57//config:	  Support -0: input items are terminated by a NUL character
58//config:	  instead of whitespace, and the quotes and backslash
59//config:	  are not special.
60
61#include "libbb.h"
62
63/* This is a NOEXEC applet. Be very careful! */
64
65
66//#define dbg_msg(...) bb_error_msg(__VA_ARGS__)
67#define dbg_msg(...) ((void)0)
68
69
70#ifdef TEST
71# ifndef ENABLE_FEATURE_XARGS_SUPPORT_CONFIRMATION
72#  define ENABLE_FEATURE_XARGS_SUPPORT_CONFIRMATION 1
73# endif
74# ifndef ENABLE_FEATURE_XARGS_SUPPORT_QUOTES
75#  define ENABLE_FEATURE_XARGS_SUPPORT_QUOTES 1
76# endif
77# ifndef ENABLE_FEATURE_XARGS_SUPPORT_TERMOPT
78#  define ENABLE_FEATURE_XARGS_SUPPORT_TERMOPT 1
79# endif
80# ifndef ENABLE_FEATURE_XARGS_SUPPORT_ZERO_TERM
81#  define ENABLE_FEATURE_XARGS_SUPPORT_ZERO_TERM 1
82# endif
83#endif
84
85
86struct globals {
87	char **args;
88	const char *eof_str;
89	int idx;
90} FIX_ALIASING;
91#define G (*(struct globals*)&bb_common_bufsiz1)
92#define INIT_G() do { } while (0)
93
94
95/*
96 * This function has special algorithm.
97 * Don't use fork and include to main!
98 */
99static int xargs_exec(void)
100{
101	int status;
102
103	status = spawn_and_wait(G.args);
104	if (status < 0) {
105		bb_simple_perror_msg(G.args[0]);
106		return errno == ENOENT ? 127 : 126;
107	}
108	if (status == 255) {
109		bb_error_msg("%s: exited with status 255; aborting", G.args[0]);
110		return 124;
111	}
112	if (status >= 0x180) {
113		bb_error_msg("%s: terminated by signal %d",
114			G.args[0], status - 0x180);
115		return 125;
116	}
117	if (status)
118		return 123;
119	return 0;
120}
121
122/* In POSIX/C locale isspace is only these chars: "\t\n\v\f\r" and space.
123 * "\t\n\v\f\r" happen to have ASCII codes 9,10,11,12,13.
124 */
125#define ISSPACE(a) ({ unsigned char xargs__isspace = (a) - 9; xargs__isspace == (' ' - 9) || xargs__isspace <= (13 - 9); })
126
127static void store_param(char *s)
128{
129	/* Grow by 256 elements at once */
130	if (!(G.idx & 0xff)) { /* G.idx == N*256 */
131		/* Enlarge, make G.args[(N+1)*256 - 1] last valid idx */
132		G.args = xrealloc(G.args, sizeof(G.args[0]) * (G.idx + 0x100));
133	}
134	G.args[G.idx++] = s;
135}
136
137/* process[0]_stdin:
138 * Read characters into buf[n_max_chars+1], and when parameter delimiter
139 * is seen, store the address of a new parameter to args[].
140 * If reading discovers that last chars do not form the complete
141 * parameter, the pointer to the first such "tail character" is returned.
142 * (buf has extra byte at the end to accomodate terminating NUL
143 * of "tail characters" string).
144 * Otherwise, the returned pointer points to NUL byte.
145 * On entry, buf[] may contain some "seed chars" which are to become
146 * the beginning of the first parameter.
147 */
148
149#if ENABLE_FEATURE_XARGS_SUPPORT_QUOTES
150static char* FAST_FUNC process_stdin(int n_max_chars, int n_max_arg, char *buf)
151{
152#define NORM      0
153#define QUOTE     1
154#define BACKSLASH 2
155#define SPACE     4
156	char q = '\0';             /* quote char */
157	char state = NORM;
158	char *s = buf;             /* start of the word */
159	char *p = s + strlen(buf); /* end of the word */
160
161	buf += n_max_chars;        /* past buffer's end */
162
163	/* "goto ret" is used instead of "break" to make control flow
164	 * more obvious: */
165
166	while (1) {
167		int c = getchar();
168		if (c == EOF) {
169			if (p != s)
170				goto close_word;
171			goto ret;
172		}
173		if (state == BACKSLASH) {
174			state = NORM;
175			goto set;
176		}
177		if (state == QUOTE) {
178			if (c != q)
179				goto set;
180			q = '\0';
181			state = NORM;
182		} else { /* if (state == NORM) */
183			if (ISSPACE(c)) {
184				if (p != s) {
185 close_word:
186					state = SPACE;
187					c = '\0';
188					goto set;
189				}
190			} else {
191				if (c == '\\') {
192					state = BACKSLASH;
193				} else if (c == '\'' || c == '"') {
194					q = c;
195					state = QUOTE;
196				} else {
197 set:
198					*p++ = c;
199				}
200			}
201		}
202		if (state == SPACE) {   /* word's delimiter or EOF detected */
203			if (q) {
204				bb_error_msg_and_die("unmatched %s quote",
205					q == '\'' ? "single" : "double");
206			}
207			/* A full word is loaded */
208			if (G.eof_str) {
209				if (strcmp(s, G.eof_str) == 0) {
210					while (getchar() != EOF)
211						continue;
212					p = s;
213					goto ret;
214				}
215			}
216			store_param(s);
217			dbg_msg("args[]:'%s'", s);
218			s = p;
219			n_max_arg--;
220			if (n_max_arg == 0) {
221				goto ret;
222			}
223			state = NORM;
224		}
225		if (p == buf) {
226			goto ret;
227		}
228	}
229 ret:
230	*p = '\0';
231	/* store_param(NULL) - caller will do it */
232	dbg_msg("return:'%s'", s);
233	return s;
234}
235#else
236/* The variant does not support single quotes, double quotes or backslash */
237static char* FAST_FUNC process_stdin(int n_max_chars, int n_max_arg, char *buf)
238{
239	char *s = buf;             /* start of the word */
240	char *p = s + strlen(buf); /* end of the word */
241
242	buf += n_max_chars;        /* past buffer's end */
243
244	while (1) {
245		int c = getchar();
246		if (c == EOF) {
247			if (p == s)
248				goto ret;
249		}
250		if (c == EOF || ISSPACE(c)) {
251			if (p == s)
252				continue;
253			c = EOF;
254		}
255		*p++ = (c == EOF ? '\0' : c);
256		if (c == EOF) { /* word's delimiter or EOF detected */
257			/* A full word is loaded */
258			if (G.eof_str) {
259				if (strcmp(s, G.eof_str) == 0) {
260					while (getchar() != EOF)
261						continue;
262					p = s;
263					goto ret;
264				}
265			}
266			store_param(s);
267			dbg_msg("args[]:'%s'", s);
268			s = p;
269			n_max_arg--;
270			if (n_max_arg == 0) {
271				goto ret;
272			}
273		}
274		if (p == buf) {
275			goto ret;
276		}
277	}
278 ret:
279	*p = '\0';
280	/* store_param(NULL) - caller will do it */
281	dbg_msg("return:'%s'", s);
282	return s;
283}
284#endif /* FEATURE_XARGS_SUPPORT_QUOTES */
285
286#if ENABLE_FEATURE_XARGS_SUPPORT_ZERO_TERM
287static char* FAST_FUNC process0_stdin(int n_max_chars, int n_max_arg, char *buf)
288{
289	char *s = buf;             /* start of the word */
290	char *p = s + strlen(buf); /* end of the word */
291
292	buf += n_max_chars;        /* past buffer's end */
293
294	while (1) {
295		int c = getchar();
296		if (c == EOF) {
297			if (p == s)
298				goto ret;
299			c = '\0';
300		}
301		*p++ = c;
302		if (c == '\0') {   /* word's delimiter or EOF detected */
303			/* A full word is loaded */
304			store_param(s);
305			dbg_msg("args[]:'%s'", s);
306			s = p;
307			n_max_arg--;
308			if (n_max_arg == 0) {
309				goto ret;
310			}
311		}
312		if (p == buf) {
313			goto ret;
314		}
315	}
316 ret:
317	*p = '\0';
318	/* store_param(NULL) - caller will do it */
319	dbg_msg("return:'%s'", s);
320	return s;
321}
322#endif /* FEATURE_XARGS_SUPPORT_ZERO_TERM */
323
324#if ENABLE_FEATURE_XARGS_SUPPORT_CONFIRMATION
325/* Prompt the user for a response, and
326   if the user responds affirmatively, return true;
327   otherwise, return false. Uses "/dev/tty", not stdin. */
328static int xargs_ask_confirmation(void)
329{
330	FILE *tty_stream;
331	int c, savec;
332
333	tty_stream = xfopen_for_read(CURRENT_TTY);
334	fputs(" ?...", stderr);
335	fflush_all();
336	c = savec = getc(tty_stream);
337	while (c != EOF && c != '\n')
338		c = getc(tty_stream);
339	fclose(tty_stream);
340	return (savec == 'y' || savec == 'Y');
341}
342#else
343# define xargs_ask_confirmation() 1
344#endif
345
346//usage:#define xargs_trivial_usage
347//usage:       "[OPTIONS] [PROG ARGS]"
348//usage:#define xargs_full_usage "\n\n"
349//usage:       "Run PROG on every item given by stdin\n"
350//usage:     "\nOptions:"
351//usage:	IF_FEATURE_XARGS_SUPPORT_CONFIRMATION(
352//usage:     "\n	-p	Ask user whether to run each command"
353//usage:	)
354//usage:     "\n	-r	Don't run command if input is empty"
355//usage:	IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(
356//usage:     "\n	-0	Input is separated by NUL characters"
357//usage:	)
358//usage:     "\n	-t	Print the command on stderr before execution"
359//usage:     "\n	-e[STR]	STR stops input processing"
360//usage:     "\n	-n N	Pass no more than N args to PROG"
361//usage:     "\n	-s N	Pass command line of no more than N bytes"
362//usage:	IF_FEATURE_XARGS_SUPPORT_TERMOPT(
363//usage:     "\n	-x	Exit if size is exceeded"
364//usage:	)
365//usage:#define xargs_example_usage
366//usage:       "$ ls | xargs gzip\n"
367//usage:       "$ find . -name '*.c' -print | xargs rm\n"
368
369/* Correct regardless of combination of CONFIG_xxx */
370enum {
371	OPTBIT_VERBOSE = 0,
372	OPTBIT_NO_EMPTY,
373	OPTBIT_UPTO_NUMBER,
374	OPTBIT_UPTO_SIZE,
375	OPTBIT_EOF_STRING,
376	OPTBIT_EOF_STRING1,
377	IF_FEATURE_XARGS_SUPPORT_CONFIRMATION(OPTBIT_INTERACTIVE,)
378	IF_FEATURE_XARGS_SUPPORT_TERMOPT(     OPTBIT_TERMINATE  ,)
379	IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(   OPTBIT_ZEROTERM   ,)
380
381	OPT_VERBOSE     = 1 << OPTBIT_VERBOSE    ,
382	OPT_NO_EMPTY    = 1 << OPTBIT_NO_EMPTY   ,
383	OPT_UPTO_NUMBER = 1 << OPTBIT_UPTO_NUMBER,
384	OPT_UPTO_SIZE   = 1 << OPTBIT_UPTO_SIZE  ,
385	OPT_EOF_STRING  = 1 << OPTBIT_EOF_STRING , /* GNU: -e[<param>] */
386	OPT_EOF_STRING1 = 1 << OPTBIT_EOF_STRING1, /* SUS: -E<param> */
387	OPT_INTERACTIVE = IF_FEATURE_XARGS_SUPPORT_CONFIRMATION((1 << OPTBIT_INTERACTIVE)) + 0,
388	OPT_TERMINATE   = IF_FEATURE_XARGS_SUPPORT_TERMOPT(     (1 << OPTBIT_TERMINATE  )) + 0,
389	OPT_ZEROTERM    = IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(   (1 << OPTBIT_ZEROTERM   )) + 0,
390};
391#define OPTION_STR "+trn:s:e::E:" \
392	IF_FEATURE_XARGS_SUPPORT_CONFIRMATION("p") \
393	IF_FEATURE_XARGS_SUPPORT_TERMOPT(     "x") \
394	IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(   "0")
395
396int xargs_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
397int xargs_main(int argc, char **argv)
398{
399	int i;
400	int child_error = 0;
401	char *max_args;
402	char *max_chars;
403	char *buf;
404	unsigned opt;
405	int n_max_chars;
406	int n_max_arg;
407#if ENABLE_FEATURE_XARGS_SUPPORT_ZERO_TERM
408	char* FAST_FUNC (*read_args)(int, int, char*) = process_stdin;
409#else
410#define read_args process_stdin
411#endif
412
413	INIT_G();
414
415	G.eof_str = NULL;
416	opt = getopt32(argv, OPTION_STR, &max_args, &max_chars, &G.eof_str, &G.eof_str);
417
418	/* -E ""? You may wonder why not just omit -E?
419	 * This is used for portability:
420	 * old xargs was using "_" as default for -E / -e */
421	if ((opt & OPT_EOF_STRING1) && G.eof_str[0] == '\0')
422		G.eof_str = NULL;
423
424	if (opt & OPT_ZEROTERM)
425		IF_FEATURE_XARGS_SUPPORT_ZERO_TERM(read_args = process0_stdin);
426
427	argv += optind;
428	argc -= optind;
429	if (!argv[0]) {
430		/* default behavior is to echo all the filenames */
431		*--argv = (char*)"echo";
432		argc++;
433	}
434
435	/* -s NUM default. fileutils-4.4.2 uses 128k, but I heasitate
436	 * to use such a big value - first need to change code to use
437	 * growable buffer instead of fixed one.
438	 */
439	n_max_chars = 32 * 1024;
440	/* Make smaller if system does not allow our default value.
441	 * The Open Group Base Specifications Issue 6:
442	 * "The xargs utility shall limit the command line length such that
443	 * when the command line is invoked, the combined argument
444	 * and environment lists (see the exec family of functions
445	 * in the System Interfaces volume of IEEE Std 1003.1-2001)
446	 * shall not exceed {ARG_MAX}-2048 bytes".
447	 */
448	{
449		long arg_max = 0;
450#if defined _SC_ARG_MAX
451		arg_max = sysconf(_SC_ARG_MAX) - 2048;
452#elif defined ARG_MAX
453		arg_max = ARG_MAX - 2048;
454#endif
455		if (arg_max > 0 && n_max_chars > arg_max)
456			n_max_chars = arg_max;
457	}
458	if (opt & OPT_UPTO_SIZE) {
459		n_max_chars = xatou_range(max_chars, 1, INT_MAX);
460	}
461	/* Account for prepended fixed arguments */
462	{
463		size_t n_chars = 0;
464		for (i = 0; argv[i]; i++) {
465			n_chars += strlen(argv[i]) + 1;
466		}
467		n_max_chars -= n_chars;
468	}
469	/* Sanity check */
470	if (n_max_chars <= 0) {
471		bb_error_msg_and_die("can't fit single argument within argument list size limit");
472	}
473
474	buf = xzalloc(n_max_chars + 1);
475
476	n_max_arg = n_max_chars;
477	if (opt & OPT_UPTO_NUMBER) {
478		n_max_arg = xatou_range(max_args, 1, INT_MAX);
479		/* Not necessary, we use growable args[]: */
480		/* if (n_max_arg > n_max_chars) n_max_arg = n_max_chars */
481	}
482
483	/* Allocate pointers for execvp */
484	/* We can statically allocate (argc + n_max_arg + 1) elements
485	 * and do not bother with resizing args[], but on 64-bit machines
486	 * this results in args[] vector which is ~8 times bigger
487	 * than n_max_chars! That is, with n_max_chars == 20k,
488	 * args[] will take 160k (!), which will most likely be
489	 * almost entirely unused.
490	 */
491	/* See store_param() for matching 256-step growth logic */
492	G.args = xmalloc(sizeof(G.args[0]) * ((argc + 0xff) & ~0xff));
493
494	/* Store the command to be executed, part 1 */
495	for (i = 0; argv[i]; i++)
496		G.args[i] = argv[i];
497
498	while (1) {
499		char *rem;
500
501		G.idx = argc;
502		rem = read_args(n_max_chars, n_max_arg, buf);
503		store_param(NULL);
504
505		if (!G.args[argc]) {
506			if (*rem != '\0')
507				bb_error_msg_and_die("argument line too long");
508			if (opt & OPT_NO_EMPTY)
509				break;
510		}
511		opt |= OPT_NO_EMPTY;
512
513		if (opt & (OPT_INTERACTIVE | OPT_VERBOSE)) {
514			const char *fmt = " %s" + 1;
515			char **args = G.args;
516			for (i = 0; args[i]; i++) {
517				fprintf(stderr, fmt, args[i]);
518				fmt = " %s";
519			}
520			if (!(opt & OPT_INTERACTIVE))
521				bb_putchar_stderr('\n');
522		}
523
524		if (!(opt & OPT_INTERACTIVE) || xargs_ask_confirmation()) {
525			child_error = xargs_exec();
526		}
527
528		if (child_error > 0 && child_error != 123) {
529			break;
530		}
531
532		overlapping_strcpy(buf, rem);
533	} /* while */
534
535	if (ENABLE_FEATURE_CLEAN_UP) {
536		free(G.args);
537		free(buf);
538	}
539
540	return child_error;
541}
542
543
544#ifdef TEST
545
546const char *applet_name = "debug stuff usage";
547
548void bb_show_usage(void)
549{
550	fprintf(stderr, "Usage: %s [-p] [-r] [-t] -[x] [-n max_arg] [-s max_chars]\n",
551		applet_name);
552	exit(EXIT_FAILURE);
553}
554
555int main(int argc, char **argv)
556{
557	return xargs_main(argc, argv);
558}
559#endif /* TEST */
560