xref: /dragonfly/contrib/libarchive/tar/cmdline.c (revision 783d47c4)
1 /*-
2  * Copyright (c) 2003-2008 Tim Kientzle
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(S) ``AS IS'' AND ANY EXPRESS OR
15  * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
16  * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
17  * IN NO EVENT SHALL THE AUTHOR(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
18  * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
19  * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
20  * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
21  * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
22  * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF
23  * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
24  */
25 
26 /*
27  * Command line parser for tar.
28  */
29 
30 #include "bsdtar_platform.h"
31 __FBSDID("$FreeBSD$");
32 
33 #ifdef HAVE_ERRNO_H
34 #include <errno.h>
35 #endif
36 #ifdef HAVE_STDLIB_H
37 #include <stdlib.h>
38 #endif
39 #ifdef HAVE_STRING_H
40 #include <string.h>
41 #endif
42 
43 #include "bsdtar.h"
44 #include "err.h"
45 
46 /*
47  * Short options for tar.  Please keep this sorted.
48  */
49 static const char *short_options
50 	= "Bb:C:cf:HhI:JjkLlmnOoPpqrSs:T:tUuvW:wX:xyZz";
51 
52 /*
53  * Long options for tar.  Please keep this list sorted.
54  *
55  * The symbolic names for options that lack a short equivalent are
56  * defined in bsdtar.h.  Also note that so far I've found no need
57  * to support optional arguments to long options.  That would be
58  * a small change to the code below.
59  */
60 
61 static const struct bsdtar_option {
62 	const char *name;
63 	int required;      /* 1 if this option requires an argument. */
64 	int equivalent;    /* Equivalent short option. */
65 } tar_longopts[] = {
66 	{ "absolute-paths",       0, 'P' },
67 	{ "append",               0, 'r' },
68 	{ "block-size",           1, 'b' },
69 	{ "bunzip2",              0, 'j' },
70 	{ "bzip",                 0, 'j' },
71 	{ "bzip2",                0, 'j' },
72 	{ "cd",                   1, 'C' },
73 	{ "check-links",          0, OPTION_CHECK_LINKS },
74 	{ "chroot",               0, OPTION_CHROOT },
75 	{ "compress",             0, 'Z' },
76 	{ "confirmation",         0, 'w' },
77 	{ "create",               0, 'c' },
78 	{ "dereference",	  0, 'L' },
79 	{ "directory",            1, 'C' },
80 	{ "disable-copyfile",	  0, OPTION_DISABLE_COPYFILE },
81 	{ "exclude",              1, OPTION_EXCLUDE },
82 	{ "exclude-from",         1, 'X' },
83 	{ "extract",              0, 'x' },
84 	{ "fast-read",            0, 'q' },
85 	{ "file",                 1, 'f' },
86 	{ "files-from",           1, 'T' },
87 	{ "format",               1, OPTION_FORMAT },
88 	{ "gid",		  1, OPTION_GID },
89 	{ "gname",		  1, OPTION_GNAME },
90 	{ "gunzip",               0, 'z' },
91 	{ "gzip",                 0, 'z' },
92 	{ "help",                 0, OPTION_HELP },
93 	{ "include",              1, OPTION_INCLUDE },
94 	{ "insecure",             0, 'P' },
95 	{ "interactive",          0, 'w' },
96 	{ "keep-newer-files",     0, OPTION_KEEP_NEWER_FILES },
97 	{ "keep-old-files",       0, 'k' },
98 	{ "list",                 0, 't' },
99 	{ "lzip",                 0, OPTION_LZIP },
100 	{ "lzma",                 0, OPTION_LZMA },
101 	{ "modification-time",    0, 'm' },
102 	{ "newer",		  1, OPTION_NEWER_CTIME },
103 	{ "newer-ctime",	  1, OPTION_NEWER_CTIME },
104 	{ "newer-ctime-than",	  1, OPTION_NEWER_CTIME_THAN },
105 	{ "newer-mtime",	  1, OPTION_NEWER_MTIME },
106 	{ "newer-mtime-than",	  1, OPTION_NEWER_MTIME_THAN },
107 	{ "newer-than",		  1, OPTION_NEWER_CTIME_THAN },
108 	{ "no-recursion",         0, 'n' },
109 	{ "no-same-owner",	  0, OPTION_NO_SAME_OWNER },
110 	{ "no-same-permissions",  0, OPTION_NO_SAME_PERMISSIONS },
111 	{ "nodump",               0, OPTION_NODUMP },
112 	{ "norecurse",            0, 'n' },
113 	{ "null",		  0, OPTION_NULL },
114 	{ "numeric-owner",	  0, OPTION_NUMERIC_OWNER },
115 	{ "one-file-system",	  0, OPTION_ONE_FILE_SYSTEM },
116 	{ "options",              1, OPTION_OPTIONS },
117 	{ "posix",		  0, OPTION_POSIX },
118 	{ "preserve-permissions", 0, 'p' },
119 	{ "read-full-blocks",	  0, 'B' },
120 	{ "same-owner",	          0, OPTION_SAME_OWNER },
121 	{ "same-permissions",     0, 'p' },
122 	{ "strip-components",	  1, OPTION_STRIP_COMPONENTS },
123 	{ "to-stdout",            0, 'O' },
124 	{ "totals",		  0, OPTION_TOTALS },
125 	{ "uid",		  1, OPTION_UID },
126 	{ "uname",		  1, OPTION_UNAME },
127 	{ "uncompress",           0, 'Z' },
128 	{ "unlink",		  0, 'U' },
129 	{ "unlink-first",	  0, 'U' },
130 	{ "update",               0, 'u' },
131 	{ "use-compress-program", 1, OPTION_USE_COMPRESS_PROGRAM },
132 	{ "verbose",              0, 'v' },
133 	{ "version",              0, OPTION_VERSION },
134 	{ "xz",                   0, 'J' },
135 	{ NULL, 0, 0 }
136 };
137 
138 /*
139  * This getopt implementation has two key features that common
140  * getopt_long() implementations lack.  Apart from those, it's a
141  * straightforward option parser, considerably simplified by not
142  * needing to support the wealth of exotic getopt_long() features.  It
143  * has, of course, been shamelessly tailored for bsdtar.  (If you're
144  * looking for a generic getopt_long() implementation for your
145  * project, I recommend Gregory Pietsch's public domain getopt_long()
146  * implementation.)  The two additional features are:
147  *
148  * Old-style tar arguments: The original tar implementation treated
149  * the first argument word as a list of single-character option
150  * letters.  All arguments follow as separate words.  For example,
151  *    tar xbf 32 /dev/tape
152  * Here, the "xbf" is three option letters, "32" is the argument for
153  * "b" and "/dev/tape" is the argument for "f".  We support this usage
154  * if the first command-line argument does not begin with '-'.  We
155  * also allow regular short and long options to follow, e.g.,
156  *    tar xbf 32 /dev/tape -P --format=pax
157  *
158  * -W long options: There's an obscure GNU convention (only rarely
159  * supported even there) that allows "-W option=argument" as an
160  * alternative way to support long options.  This was supported in
161  * early bsdtar as a way to access long options on platforms that did
162  * not support getopt_long() and is preserved here for backwards
163  * compatibility.  (Of course, if I'd started with a custom
164  * command-line parser from the beginning, I would have had normal
165  * long option support on every platform so that hack wouldn't have
166  * been necessary.  Oh, well.  Some mistakes you just have to live
167  * with.)
168  *
169  * TODO: We should be able to use this to pull files and intermingled
170  * options (such as -C) from the command line in write mode.  That
171  * will require a little rethinking of the argument handling in
172  * bsdtar.c.
173  *
174  * TODO: If we want to support arbitrary command-line options from -T
175  * input (as GNU tar does), we may need to extend this to handle option
176  * words from sources other than argv/argc.  I'm not really sure if I
177  * like that feature of GNU tar, so it's certainly not a priority.
178  */
179 
180 int
181 bsdtar_getopt(struct bsdtar *bsdtar)
182 {
183 	enum { state_start = 0, state_old_tar, state_next_word,
184 	       state_short, state_long };
185 
186 	const struct bsdtar_option *popt, *match = NULL, *match2 = NULL;
187 	const char *p, *long_prefix = "--";
188 	size_t optlength;
189 	int opt = '?';
190 	int required = 0;
191 
192 	bsdtar->argument = NULL;
193 
194 	/* First time through, initialize everything. */
195 	if (bsdtar->getopt_state == state_start) {
196 		/* Skip program name. */
197 		++bsdtar->argv;
198 		--bsdtar->argc;
199 		if (*bsdtar->argv == NULL)
200 			return (-1);
201 		/* Decide between "new style" and "old style" arguments. */
202 		if (bsdtar->argv[0][0] == '-') {
203 			bsdtar->getopt_state = state_next_word;
204 		} else {
205 			bsdtar->getopt_state = state_old_tar;
206 			bsdtar->getopt_word = *bsdtar->argv++;
207 			--bsdtar->argc;
208 		}
209 	}
210 
211 	/*
212 	 * We're parsing old-style tar arguments
213 	 */
214 	if (bsdtar->getopt_state == state_old_tar) {
215 		/* Get the next option character. */
216 		opt = *bsdtar->getopt_word++;
217 		if (opt == '\0') {
218 			/* New-style args can follow old-style. */
219 			bsdtar->getopt_state = state_next_word;
220 		} else {
221 			/* See if it takes an argument. */
222 			p = strchr(short_options, opt);
223 			if (p == NULL)
224 				return ('?');
225 			if (p[1] == ':') {
226 				bsdtar->argument = *bsdtar->argv;
227 				if (bsdtar->argument == NULL) {
228 					lafe_warnc(0,
229 					    "Option %c requires an argument",
230 					    opt);
231 					return ('?');
232 				}
233 				++bsdtar->argv;
234 				--bsdtar->argc;
235 			}
236 		}
237 	}
238 
239 	/*
240 	 * We're ready to look at the next word in argv.
241 	 */
242 	if (bsdtar->getopt_state == state_next_word) {
243 		/* No more arguments, so no more options. */
244 		if (bsdtar->argv[0] == NULL)
245 			return (-1);
246 		/* Doesn't start with '-', so no more options. */
247 		if (bsdtar->argv[0][0] != '-')
248 			return (-1);
249 		/* "--" marks end of options; consume it and return. */
250 		if (strcmp(bsdtar->argv[0], "--") == 0) {
251 			++bsdtar->argv;
252 			--bsdtar->argc;
253 			return (-1);
254 		}
255 		/* Get next word for parsing. */
256 		bsdtar->getopt_word = *bsdtar->argv++;
257 		--bsdtar->argc;
258 		if (bsdtar->getopt_word[1] == '-') {
259 			/* Set up long option parser. */
260 			bsdtar->getopt_state = state_long;
261 			bsdtar->getopt_word += 2; /* Skip leading '--' */
262 		} else {
263 			/* Set up short option parser. */
264 			bsdtar->getopt_state = state_short;
265 			++bsdtar->getopt_word;  /* Skip leading '-' */
266 		}
267 	}
268 
269 	/*
270 	 * We're parsing a group of POSIX-style single-character options.
271 	 */
272 	if (bsdtar->getopt_state == state_short) {
273 		/* Peel next option off of a group of short options. */
274 		opt = *bsdtar->getopt_word++;
275 		if (opt == '\0') {
276 			/* End of this group; recurse to get next option. */
277 			bsdtar->getopt_state = state_next_word;
278 			return bsdtar_getopt(bsdtar);
279 		}
280 
281 		/* Does this option take an argument? */
282 		p = strchr(short_options, opt);
283 		if (p == NULL)
284 			return ('?');
285 		if (p[1] == ':')
286 			required = 1;
287 
288 		/* If it takes an argument, parse that. */
289 		if (required) {
290 			/* If arg is run-in, bsdtar->getopt_word already points to it. */
291 			if (bsdtar->getopt_word[0] == '\0') {
292 				/* Otherwise, pick up the next word. */
293 				bsdtar->getopt_word = *bsdtar->argv;
294 				if (bsdtar->getopt_word == NULL) {
295 					lafe_warnc(0,
296 					    "Option -%c requires an argument",
297 					    opt);
298 					return ('?');
299 				}
300 				++bsdtar->argv;
301 				--bsdtar->argc;
302 			}
303 			if (opt == 'W') {
304 				bsdtar->getopt_state = state_long;
305 				long_prefix = "-W "; /* For clearer errors. */
306 			} else {
307 				bsdtar->getopt_state = state_next_word;
308 				bsdtar->argument = bsdtar->getopt_word;
309 			}
310 		}
311 	}
312 
313 	/* We're reading a long option, including -W long=arg convention. */
314 	if (bsdtar->getopt_state == state_long) {
315 		/* After this long option, we'll be starting a new word. */
316 		bsdtar->getopt_state = state_next_word;
317 
318 		/* Option name ends at '=' if there is one. */
319 		p = strchr(bsdtar->getopt_word, '=');
320 		if (p != NULL) {
321 			optlength = (size_t)(p - bsdtar->getopt_word);
322 			bsdtar->argument = (char *)(uintptr_t)(p + 1);
323 		} else {
324 			optlength = strlen(bsdtar->getopt_word);
325 		}
326 
327 		/* Search the table for an unambiguous match. */
328 		for (popt = tar_longopts; popt->name != NULL; popt++) {
329 			/* Short-circuit if first chars don't match. */
330 			if (popt->name[0] != bsdtar->getopt_word[0])
331 				continue;
332 			/* If option is a prefix of name in table, record it.*/
333 			if (strncmp(bsdtar->getopt_word, popt->name, optlength) == 0) {
334 				match2 = match; /* Record up to two matches. */
335 				match = popt;
336 				/* If it's an exact match, we're done. */
337 				if (strlen(popt->name) == optlength) {
338 					match2 = NULL; /* Forget the others. */
339 					break;
340 				}
341 			}
342 		}
343 
344 		/* Fail if there wasn't a unique match. */
345 		if (match == NULL) {
346 			lafe_warnc(0,
347 			    "Option %s%s is not supported",
348 			    long_prefix, bsdtar->getopt_word);
349 			return ('?');
350 		}
351 		if (match2 != NULL) {
352 			lafe_warnc(0,
353 			    "Ambiguous option %s%s (matches --%s and --%s)",
354 			    long_prefix, bsdtar->getopt_word, match->name, match2->name);
355 			return ('?');
356 		}
357 
358 		/* We've found a unique match; does it need an argument? */
359 		if (match->required) {
360 			/* Argument required: get next word if necessary. */
361 			if (bsdtar->argument == NULL) {
362 				bsdtar->argument = *bsdtar->argv;
363 				if (bsdtar->argument == NULL) {
364 					lafe_warnc(0,
365 					    "Option %s%s requires an argument",
366 					    long_prefix, match->name);
367 					return ('?');
368 				}
369 				++bsdtar->argv;
370 				--bsdtar->argc;
371 			}
372 		} else {
373 			/* Argument forbidden: fail if there is one. */
374 			if (bsdtar->argument != NULL) {
375 				lafe_warnc(0,
376 				    "Option %s%s does not allow an argument",
377 				    long_prefix, match->name);
378 				return ('?');
379 			}
380 		}
381 		return (match->equivalent);
382 	}
383 
384 	return (opt);
385 }
386