2 * Copyright (c) 2003-2008 Tim Kientzle
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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.
28 #ifndef ARCHIVE_ENTRY_H_INCLUDED
29 #define ARCHIVE_ENTRY_H_INCLUDED
32 * Note: archive_entry.h is for use outside of libarchive; the
33 * configuration headers (config.h, archive_platform.h, etc.) are
34 * purely internal. Do NOT use HAVE_XXX configuration macros to
35 * control the behavior of this header! If you must conditionalize,
36 * use predefined compiler and/or platform macros.
39 #include <sys/types.h>
40 #include <stddef.h> /* for wchar_t */
43 /* Get appropriate definitions of standard POSIX-style types. */
44 /* These should match the types used in 'struct stat' */
45 #if defined(_WIN32) && !defined(__CYGWIN__)
46 #define __LA_INT64_T __int64
47 #define __LA_UID_T unsigned int
48 #define __LA_GID_T unsigned int
49 #define __LA_DEV_T unsigned int
50 #define __LA_MODE_T unsigned short
53 #define __LA_INT64_T int64_t
54 #define __LA_UID_T uid_t
55 #define __LA_GID_T gid_t
56 #define __LA_DEV_T dev_t
57 #define __LA_MODE_T mode_t
61 * XXX Is this defined for all Windows compilers? If so, in what
62 * header? It would be nice to remove the __LA_INO_T indirection and
63 * just use plain ino_t everywhere. Likewise for the other types just
66 #define __LA_INO_T ino_t
70 * On Windows, define LIBARCHIVE_STATIC if you're building or using a
71 * .lib. The default here assumes you're building a DLL. Only
72 * libarchive source should ever define __LIBARCHIVE_BUILD.
74 #if ((defined __WIN32__) || (defined _WIN32) || defined(__CYGWIN__)) && (!defined LIBARCHIVE_STATIC)
75 # ifdef __LIBARCHIVE_BUILD
77 # define __LA_DECL __attribute__((dllexport)) extern
79 # define __LA_DECL __declspec(dllexport)
83 # define __LA_DECL __attribute__((dllimport)) extern
85 # define __LA_DECL __declspec(dllimport)
89 /* Static libraries on all platforms and shared libraries on non-Windows. */
98 * Description of an archive entry.
100 * You can think of this as "struct stat" with some text fields added in.
102 * TODO: Add "comment", "charset", and possibly other entries that are
103 * supported by "pax interchange" format. However, GNU, ustar, cpio,
104 * and other variants don't support these features, so they're not an
105 * excruciatingly high priority right now.
107 * TODO: "pax interchange" format allows essentially arbitrary
108 * key/value attributes to be attached to any entry. Supporting
109 * such extensions may make this library useful for special
110 * applications (e.g., a package manager could attach special
111 * package-management attributes to each entry).
113 struct archive_entry;
116 * File-type constants. These are returned from archive_entry_filetype()
117 * and passed to archive_entry_set_filetype().
119 * These values match S_XXX defines on every platform I've checked,
120 * including Windows, AIX, Linux, Solaris, and BSD. They're
121 * (re)defined here because platforms generally don't define the ones
122 * they don't support. For example, Windows doesn't define S_IFLNK or
123 * S_IFBLK. Instead of having a mass of conditional logic and system
124 * checks to define any S_XXX values that aren't supported locally,
125 * I've just defined a new set of such constants so that
126 * libarchive-based applications can manipulate and identify archive
127 * entries properly even if the hosting platform can't store them on
130 * These values are also used directly within some portable formats,
131 * such as cpio. If you find a platform that varies from these, the
132 * correct solution is to leave these alone and translate from these
133 * portable values to platform-native values when entries are read from
134 * or written to disk.
136 #define AE_IFMT 0170000
137 #define AE_IFREG 0100000
138 #define AE_IFLNK 0120000
139 #define AE_IFSOCK 0140000
140 #define AE_IFCHR 0020000
141 #define AE_IFBLK 0060000
142 #define AE_IFDIR 0040000
143 #define AE_IFIFO 0010000
146 * Basic object manipulation
149 __LA_DECL struct archive_entry *archive_entry_clear(struct archive_entry *);
150 /* The 'clone' function does a deep copy; all of the strings are copied too. */
151 __LA_DECL struct archive_entry *archive_entry_clone(struct archive_entry *);
152 __LA_DECL void archive_entry_free(struct archive_entry *);
153 __LA_DECL struct archive_entry *archive_entry_new(void);
156 * Retrieve fields from an archive_entry.
158 * There are a number of implicit conversions among these fields. For
159 * example, if a regular string field is set and you read the _w wide
160 * character field, the entry will implicitly convert narrow-to-wide
161 * using the current locale. Similarly, dev values are automatically
162 * updated when you write devmajor or devminor and vice versa.
164 * In addition, fields can be "set" or "unset." Unset string fields
165 * return NULL, non-string fields have _is_set() functions to test
166 * whether they've been set. You can "unset" a string field by
167 * assigning NULL; non-string fields have _unset() functions to
170 * Note: There is one ambiguity in the above; string fields will
171 * also return NULL when implicit character set conversions fail.
172 * This is usually what you want.
174 __LA_DECL time_t archive_entry_atime(struct archive_entry *);
175 __LA_DECL long archive_entry_atime_nsec(struct archive_entry *);
176 __LA_DECL int archive_entry_atime_is_set(struct archive_entry *);
177 __LA_DECL time_t archive_entry_birthtime(struct archive_entry *);
178 __LA_DECL long archive_entry_birthtime_nsec(struct archive_entry *);
179 __LA_DECL int archive_entry_birthtime_is_set(struct archive_entry *);
180 __LA_DECL time_t archive_entry_ctime(struct archive_entry *);
181 __LA_DECL long archive_entry_ctime_nsec(struct archive_entry *);
182 __LA_DECL int archive_entry_ctime_is_set(struct archive_entry *);
183 __LA_DECL dev_t archive_entry_dev(struct archive_entry *);
184 __LA_DECL dev_t archive_entry_devmajor(struct archive_entry *);
185 __LA_DECL dev_t archive_entry_devminor(struct archive_entry *);
186 __LA_DECL __LA_MODE_T archive_entry_filetype(struct archive_entry *);
187 __LA_DECL void archive_entry_fflags(struct archive_entry *,
188 unsigned long * /* set */,
189 unsigned long * /* clear */);
190 __LA_DECL const char *archive_entry_fflags_text(struct archive_entry *);
191 __LA_DECL __LA_GID_T archive_entry_gid(struct archive_entry *);
192 __LA_DECL const char *archive_entry_gname(struct archive_entry *);
193 __LA_DECL const wchar_t *archive_entry_gname_w(struct archive_entry *);
194 __LA_DECL const char *archive_entry_hardlink(struct archive_entry *);
195 __LA_DECL const wchar_t *archive_entry_hardlink_w(struct archive_entry *);
196 __LA_DECL __LA_INO_T archive_entry_ino(struct archive_entry *);
197 __LA_DECL __LA_MODE_T archive_entry_mode(struct archive_entry *);
198 __LA_DECL time_t archive_entry_mtime(struct archive_entry *);
199 __LA_DECL long archive_entry_mtime_nsec(struct archive_entry *);
200 __LA_DECL int archive_entry_mtime_is_set(struct archive_entry *);
201 __LA_DECL unsigned int archive_entry_nlink(struct archive_entry *);
202 __LA_DECL const char *archive_entry_pathname(struct archive_entry *);
203 __LA_DECL const wchar_t *archive_entry_pathname_w(struct archive_entry *);
204 __LA_DECL dev_t archive_entry_rdev(struct archive_entry *);
205 __LA_DECL dev_t archive_entry_rdevmajor(struct archive_entry *);
206 __LA_DECL dev_t archive_entry_rdevminor(struct archive_entry *);
207 __LA_DECL const char *archive_entry_sourcepath(struct archive_entry *);
208 __LA_DECL __LA_INT64_T archive_entry_size(struct archive_entry *);
209 __LA_DECL int archive_entry_size_is_set(struct archive_entry *);
210 __LA_DECL const char *archive_entry_strmode(struct archive_entry *);
211 __LA_DECL const char *archive_entry_symlink(struct archive_entry *);
212 __LA_DECL const wchar_t *archive_entry_symlink_w(struct archive_entry *);
213 __LA_DECL __LA_UID_T archive_entry_uid(struct archive_entry *);
214 __LA_DECL const char *archive_entry_uname(struct archive_entry *);
215 __LA_DECL const wchar_t *archive_entry_uname_w(struct archive_entry *);
218 * Set fields in an archive_entry.
220 * Note that string 'set' functions do not copy the string, only the pointer.
221 * In contrast, 'copy' functions do copy the object pointed to.
223 * Note: As of libarchive 2.4, 'set' functions do copy the string and
224 * are therefore exact synonyms for the 'copy' versions. The 'copy'
225 * names will be retired in libarchive 3.0.
228 __LA_DECL void archive_entry_set_atime(struct archive_entry *, time_t, long);
229 __LA_DECL void archive_entry_unset_atime(struct archive_entry *);
230 __LA_DECL void archive_entry_set_birthtime(struct archive_entry *, time_t, long);
231 __LA_DECL void archive_entry_unset_birthtime(struct archive_entry *);
232 __LA_DECL void archive_entry_set_ctime(struct archive_entry *, time_t, long);
233 __LA_DECL void archive_entry_unset_ctime(struct archive_entry *);
234 __LA_DECL void archive_entry_set_dev(struct archive_entry *, dev_t);
235 __LA_DECL void archive_entry_set_devmajor(struct archive_entry *, dev_t);
236 __LA_DECL void archive_entry_set_devminor(struct archive_entry *, dev_t);
237 __LA_DECL void archive_entry_set_filetype(struct archive_entry *, unsigned int);
238 __LA_DECL void archive_entry_set_fflags(struct archive_entry *,
239 unsigned long /* set */, unsigned long /* clear */);
240 /* Returns pointer to start of first invalid token, or NULL if none. */
241 /* Note that all recognized tokens are processed, regardless. */
242 __LA_DECL const char *archive_entry_copy_fflags_text(struct archive_entry *,
244 __LA_DECL const wchar_t *archive_entry_copy_fflags_text_w(struct archive_entry *,
246 __LA_DECL void archive_entry_set_gid(struct archive_entry *, __LA_GID_T);
247 __LA_DECL void archive_entry_set_gname(struct archive_entry *, const char *);
248 __LA_DECL void archive_entry_copy_gname(struct archive_entry *, const char *);
249 __LA_DECL void archive_entry_copy_gname_w(struct archive_entry *, const wchar_t *);
250 __LA_DECL int archive_entry_update_gname_utf8(struct archive_entry *, const char *);
251 __LA_DECL void archive_entry_set_hardlink(struct archive_entry *, const char *);
252 __LA_DECL void archive_entry_copy_hardlink(struct archive_entry *, const char *);
253 __LA_DECL void archive_entry_copy_hardlink_w(struct archive_entry *, const wchar_t *);
254 __LA_DECL void archive_entry_set_ino(struct archive_entry *, unsigned long);
255 __LA_DECL void archive_entry_set_link(struct archive_entry *, const char *);
256 __LA_DECL void archive_entry_copy_link(struct archive_entry *, const char *);
257 __LA_DECL void archive_entry_copy_link_w(struct archive_entry *, const wchar_t *);
258 __LA_DECL int archive_entry_update_link_utf8(struct archive_entry *, const char *);
259 __LA_DECL void archive_entry_set_mode(struct archive_entry *, __LA_MODE_T);
260 __LA_DECL void archive_entry_set_mtime(struct archive_entry *, time_t, long);
261 __LA_DECL void archive_entry_unset_mtime(struct archive_entry *);
262 __LA_DECL void archive_entry_set_nlink(struct archive_entry *, unsigned int);
263 __LA_DECL void archive_entry_set_pathname(struct archive_entry *, const char *);
264 __LA_DECL void archive_entry_copy_pathname(struct archive_entry *, const char *);
265 __LA_DECL void archive_entry_copy_pathname_w(struct archive_entry *, const wchar_t *);
266 __LA_DECL int archive_entry_update_pathname_utf8(struct archive_entry *, const char *);
267 __LA_DECL void archive_entry_set_perm(struct archive_entry *, __LA_MODE_T);
268 __LA_DECL void archive_entry_set_rdev(struct archive_entry *, dev_t);
269 __LA_DECL void archive_entry_set_rdevmajor(struct archive_entry *, dev_t);
270 __LA_DECL void archive_entry_set_rdevminor(struct archive_entry *, dev_t);
271 __LA_DECL void archive_entry_set_size(struct archive_entry *, __LA_INT64_T);
272 __LA_DECL void archive_entry_unset_size(struct archive_entry *);
273 __LA_DECL void archive_entry_copy_sourcepath(struct archive_entry *, const char *);
274 __LA_DECL void archive_entry_set_symlink(struct archive_entry *, const char *);
275 __LA_DECL void archive_entry_copy_symlink(struct archive_entry *, const char *);
276 __LA_DECL void archive_entry_copy_symlink_w(struct archive_entry *, const wchar_t *);
277 __LA_DECL void archive_entry_set_uid(struct archive_entry *, __LA_UID_T);
278 __LA_DECL void archive_entry_set_uname(struct archive_entry *, const char *);
279 __LA_DECL void archive_entry_copy_uname(struct archive_entry *, const char *);
280 __LA_DECL void archive_entry_copy_uname_w(struct archive_entry *, const wchar_t *);
281 __LA_DECL int archive_entry_update_uname_utf8(struct archive_entry *, const char *);
283 * Routines to bulk copy fields to/from a platform-native "struct
284 * stat." Libarchive used to just store a struct stat inside of each
285 * archive_entry object, but this created issues when trying to
286 * manipulate archives on systems different than the ones they were
289 * TODO: On Linux, provide both stat32 and stat64 versions of these functions.
291 __LA_DECL const struct stat *archive_entry_stat(struct archive_entry *);
292 __LA_DECL void archive_entry_copy_stat(struct archive_entry *, const struct stat *);
296 * ACL routines. This used to simply store and return text-format ACL
297 * strings, but that proved insufficient for a number of reasons:
298 * = clients need control over uname/uid and gname/gid mappings
299 * = there are many different ACL text formats
300 * = would like to be able to read/convert archives containing ACLs
301 * on platforms that lack ACL libraries
303 * This last point, in particular, forces me to implement a reasonably
304 * complete set of ACL support routines.
306 * TODO: Extend this to support NFSv4/NTFS permissions. That should
307 * allow full ACL support on Mac OS, in particular, which uses
308 * POSIX.1e-style interfaces to manipulate NFSv4/NTFS permissions.
312 * Permission bits mimic POSIX.1e. Note that I've not followed POSIX.1e's
313 * "permset"/"perm" abstract type nonsense. A permset is just a simple
314 * bitmap, following long-standing Unix tradition.
316 #define ARCHIVE_ENTRY_ACL_EXECUTE 1
317 #define ARCHIVE_ENTRY_ACL_WRITE 2
318 #define ARCHIVE_ENTRY_ACL_READ 4
320 /* We need to be able to specify either or both of these. */
321 #define ARCHIVE_ENTRY_ACL_TYPE_ACCESS 256
322 #define ARCHIVE_ENTRY_ACL_TYPE_DEFAULT 512
324 /* Tag values mimic POSIX.1e */
325 #define ARCHIVE_ENTRY_ACL_USER 10001 /* Specified user. */
326 #define ARCHIVE_ENTRY_ACL_USER_OBJ 10002 /* User who owns the file. */
327 #define ARCHIVE_ENTRY_ACL_GROUP 10003 /* Specified group. */
328 #define ARCHIVE_ENTRY_ACL_GROUP_OBJ 10004 /* Group who owns the file. */
329 #define ARCHIVE_ENTRY_ACL_MASK 10005 /* Modify group access. */
330 #define ARCHIVE_ENTRY_ACL_OTHER 10006 /* Public. */
333 * Set the ACL by clearing it and adding entries one at a time.
334 * Unlike the POSIX.1e ACL routines, you must specify the type
335 * (access/default) for each entry. Internally, the ACL data is just
336 * a soup of entries. API calls here allow you to retrieve just the
337 * entries of interest. This design (which goes against the spirit of
338 * POSIX.1e) is useful for handling archive formats that combine
339 * default and access information in a single ACL list.
341 __LA_DECL void archive_entry_acl_clear(struct archive_entry *);
342 __LA_DECL void archive_entry_acl_add_entry(struct archive_entry *,
343 int /* type */, int /* permset */, int /* tag */,
344 int /* qual */, const char * /* name */);
345 __LA_DECL void archive_entry_acl_add_entry_w(struct archive_entry *,
346 int /* type */, int /* permset */, int /* tag */,
347 int /* qual */, const wchar_t * /* name */);
350 * To retrieve the ACL, first "reset", then repeatedly ask for the
351 * "next" entry. The want_type parameter allows you to request only
352 * access entries or only default entries.
354 __LA_DECL int archive_entry_acl_reset(struct archive_entry *, int /* want_type */);
355 __LA_DECL int archive_entry_acl_next(struct archive_entry *, int /* want_type */,
356 int * /* type */, int * /* permset */, int * /* tag */,
357 int * /* qual */, const char ** /* name */);
358 __LA_DECL int archive_entry_acl_next_w(struct archive_entry *, int /* want_type */,
359 int * /* type */, int * /* permset */, int * /* tag */,
360 int * /* qual */, const wchar_t ** /* name */);
363 * Construct a text-format ACL. The flags argument is a bitmask that
364 * can include any of the following:
366 * ARCHIVE_ENTRY_ACL_TYPE_ACCESS - Include access entries.
367 * ARCHIVE_ENTRY_ACL_TYPE_DEFAULT - Include default entries.
368 * ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID - Include extra numeric ID field in
369 * each ACL entry. (As used by 'star'.)
370 * ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT - Include "default:" before each
373 #define ARCHIVE_ENTRY_ACL_STYLE_EXTRA_ID 1024
374 #define ARCHIVE_ENTRY_ACL_STYLE_MARK_DEFAULT 2048
375 __LA_DECL const wchar_t *archive_entry_acl_text_w(struct archive_entry *,
378 /* Return a count of entries matching 'want_type' */
379 __LA_DECL int archive_entry_acl_count(struct archive_entry *, int /* want_type */);
382 * Private ACL parser. This is private because it handles some
383 * very weird formats that clients should not be messing with.
384 * Clients should only deal with their platform-native formats.
385 * Because of the need to support many formats cleanly, new arguments
386 * are likely to get added on a regular basis. Clients who try to use
387 * this interface are likely to be surprised when it changes.
391 * TODO: Move this declaration out of the public header and into
392 * a private header. Warnings above are silly.
394 __LA_DECL int __archive_entry_acl_parse_w(struct archive_entry *,
395 const wchar_t *, int /* type */);
398 * extended attributes
401 __LA_DECL void archive_entry_xattr_clear(struct archive_entry *);
402 __LA_DECL void archive_entry_xattr_add_entry(struct archive_entry *,
403 const char * /* name */, const void * /* value */,
407 * To retrieve the xattr list, first "reset", then repeatedly ask for the
411 __LA_DECL int archive_entry_xattr_count(struct archive_entry *);
412 __LA_DECL int archive_entry_xattr_reset(struct archive_entry *);
413 __LA_DECL int archive_entry_xattr_next(struct archive_entry *,
414 const char ** /* name */, const void ** /* value */, size_t *);
417 * Utility to match up hardlinks.
419 * The 'struct archive_entry_linkresolver' is a cache of archive entries
420 * for files with multiple links. Here's how to use it:
421 * 1. Create a lookup object with archive_entry_linkresolver_new()
422 * 2. Tell it the archive format you're using.
423 * 3. Hand each archive_entry to archive_entry_linkify().
424 * That function will return 0, 1, or 2 entries that should
426 * 4. Call archive_entry_linkify(resolver, NULL) until
427 * no more entries are returned.
428 * 5. Call archive_entry_link_resolver_free(resolver) to free resources.
430 * The entries returned have their hardlink and size fields updated
431 * appropriately. If an entry is passed in that does not refer to
432 * a file with multiple links, it is returned unchanged. The intention
433 * is that you should be able to simply filter all entries through
436 * To make things more efficient, be sure that each entry has a valid
437 * nlinks value. The hardlink cache uses this to track when all links
438 * have been found. If the nlinks value is zero, it will keep every
439 * name in the cache indefinitely, which can use a lot of memory.
441 * Note that archive_entry_size() is reset to zero if the file
442 * body should not be written to the archive. Pay attention!
444 struct archive_entry_linkresolver;
447 * There are three different strategies for marking hardlinks.
448 * The descriptions below name them after the best-known
449 * formats that rely on each strategy:
451 * "Old cpio" is the simplest, it always returns any entry unmodified.
452 * As far as I know, only cpio formats use this. Old cpio archives
453 * store every link with the full body; the onus is on the dearchiver
454 * to detect and properly link the files as they are restored.
455 * "tar" is also pretty simple; it caches a copy the first time it sees
456 * any link. Subsequent appearances are modified to be hardlink
457 * references to the first one without any body. Used by all tar
458 * formats, although the newest tar formats permit the "old cpio" strategy
459 * as well. This strategy is very simple for the dearchiver,
460 * and reasonably straightforward for the archiver.
461 * "new cpio" is trickier. It stores the body only with the last
462 * occurrence. The complication is that we might not
463 * see every link to a particular file in a single session, so
464 * there's no easy way to know when we've seen the last occurrence.
465 * The solution here is to queue one link until we see the next.
466 * At the end of the session, you can enumerate any remaining
467 * entries by calling archive_entry_linkify(NULL) and store those
468 * bodies. If you have a file with three links l1, l2, and l3,
469 * you'll get the following behavior if you see all three links:
470 * linkify(l1) => NULL (the resolver stores l1 internally)
471 * linkify(l2) => l1 (resolver stores l2, you write l1)
472 * linkify(l3) => l2, l3 (all links seen, you can write both).
473 * If you only see l1 and l2, you'll get this behavior:
474 * linkify(l1) => NULL
476 * linkify(NULL) => l2 (at end, you retrieve remaining links)
477 * As the name suggests, this strategy is used by newer cpio variants.
478 * It's noticably more complex for the archiver, slightly more complex
479 * for the dearchiver than the tar strategy, but makes it straightforward
480 * to restore a file using any link by simply continuing to scan until
481 * you see a link that is stored with a body. In contrast, the tar
482 * strategy requires you to rescan the archive from the beginning to
483 * correctly extract an arbitrary link.
486 __LA_DECL struct archive_entry_linkresolver *archive_entry_linkresolver_new(void);
487 __LA_DECL void archive_entry_linkresolver_set_strategy(
488 struct archive_entry_linkresolver *, int /* format_code */);
489 __LA_DECL void archive_entry_linkresolver_free(struct archive_entry_linkresolver *);
490 __LA_DECL void archive_entry_linkify(struct archive_entry_linkresolver *,
491 struct archive_entry **, struct archive_entry **);
497 /* This is meaningless outside of this header. */
500 #endif /* !ARCHIVE_ENTRY_H_INCLUDED */