1ARCHIVE_WRITE_DISK(3) BSD Library Functions Manual ARCHIVE_WRITE_DISK(3)
2
4 archive_write_disk_new, archive_write_disk_set_options,
5 archive_write_disk_set_skip_file, archive_write_disk_set_group_lookup,
6 archive_write_disk_set_standard_lookup,
7 archive_write_disk_set_user_lookup — functions for creating objects on
8 disk
9
11 Streaming Archive Library (libarchive, -larchive)
12
14 #include <archive.h>
15 struct archive *
17 archive_write_disk_new(void);
18 int
20 archive_write_disk_set_options(struct archive *, int flags);
21 int
23 archive_write_disk_set_skip_file(struct archive *, dev_t, ino_t);
24 int
26 archive_write_disk_set_group_lookup(struct archive *, void *,
27 gid_t (*)(void *, const char *gname, gid_t gid),
28 void (*cleanup)(void *));
29 int
31 archive_write_disk_set_standard_lookup(struct archive *);
32 int
34 archive_write_disk_set_user_lookup(struct archive *, void *,
35 uid_t (*)(void *, const char *uname, uid_t uid),
36 void (*cleanup)(void *));
37
39 These functions provide a complete API for creating objects on disk from
40 struct archive_entry descriptions. They are most naturally used when ex‐
41 tracting objects from an archive using the archive_read() interface. The
42 general process is to read struct archive_entry objects from an archive,
43 then write those objects to a struct archive object created using the
44 archive_write_disk() family functions. This interface is deliberately
45 very similar to the archive_write() interface used to write objects to a
46 streaming archive.
47 archive_write_disk_new()
49 Allocates and initializes a struct archive object suitable for
50 writing objects to disk.
51 archive_write_disk_set_skip_file()
53 Records the device and inode numbers of a file that should not be
54 overwritten. This is typically used to ensure that an extraction
55 process does not overwrite the archive from which objects are be‐
56 ing read. This capability is technically unnecessary but can be
57 a significant performance optimization in practice.
58 archive_write_disk_set_options()
60 The options field consists of a bitwise OR of one or more of the
61 following values:
62 ARCHIVE_EXTRACT_ACL
63 Attempt to restore Access Control Lists. By default, ex‐
64 tended ACLs are ignored.
65 ARCHIVE_EXTRACT_CLEAR_NOCHANGE_FFLAGS
66 Before removing a file system object prior to replacing
67 it, clear platform-specific file flags which might pre‐
68 vent its removal.
69 ARCHIVE_EXTRACT_FFLAGS
70 Attempt to restore file attributes (file flags). By de‐
71 fault, file attributes are ignored. See chattr(1)
72 (Linux) or chflags(1) (FreeBSD, Mac OS X) for more infor‐
73 mation on file attributes.
74 ARCHIVE_EXTRACT_MAC_METADATA
75 Mac OS X specific. Restore metadata using copyfile(3).
76 By default, copyfile(3) metadata is ignored.
77 ARCHIVE_EXTRACT_NO_OVERWRITE
78 Existing files on disk will not be overwritten. By de‐
79 fault, existing regular files are truncated and overwrit‐
80 ten; existing directories will have their permissions up‐
81 dated; other pre-existing objects are unlinked and recre‐
82 ated from scratch.
83 ARCHIVE_EXTRACT_OWNER
84 The user and group IDs should be set on the restored
85 file. By default, the user and group IDs are not re‐
86 stored.
87 ARCHIVE_EXTRACT_PERM
88 Full permissions (including SGID, SUID, and sticky bits)
89 should be restored exactly as specified, without obeying
90 the current umask. Note that SUID and SGID bits can only
91 be restored if the user and group ID of the object on
92 disk are correct. If ARCHIVE_EXTRACT_OWNER is not speci‐
93 fied, then SUID and SGID bits will only be restored if
94 the default user and group IDs of newly-created objects
95 on disk happen to match those specified in the archive
96 entry. By default, only basic permissions are restored,
97 and umask is obeyed.
98 ARCHIVE_EXTRACT_SAFE_WRITES
99 Extract files atomically, by first creating a unique tem‐
100 porary file and then renaming it to its required destina‐
101 tion name. This avoids a race where an application might
102 see a partial file (or no file) during extraction.
103 ARCHIVE_EXTRACT_SECURE_NOABSOLUTEPATHS
104 Refuse to extract an absolute path. The default is to
105 not refuse such paths.
106 ARCHIVE_EXTRACT_SECURE_NODOTDOT
107 Refuse to extract a path that contains a .. element any‐
108 where within it. The default is to not refuse such
109 paths. Note that paths ending in .. always cause an er‐
110 ror, regardless of this flag.
111 ARCHIVE_EXTRACT_SECURE_SYMLINKS
112 Refuse to extract any object whose final location would
113 be altered by a symlink on disk. This is intended to
114 help guard against a variety of mischief caused by ar‐
115 chives that (deliberately or otherwise) extract files
116 outside of the current directory. The default is not to
117 perform this check. If ARCHIVE_EXTRACT_UNLINK is speci‐
118 fied together with this option, the library will remove
119 any intermediate symlinks it finds and return an error
120 only if such symlink could not be removed.
121 ARCHIVE_EXTRACT_SPARSE
122 Scan data for blocks of NUL bytes and try to recreate
123 them with holes. This results in sparse files, indepen‐
124 dent of whether the archive format supports or uses them.
125 ARCHIVE_EXTRACT_TIME
126 The timestamps (mtime, ctime, and atime) should be re‐
127 stored. By default, they are ignored. Note that restor‐
128 ing of atime is not currently supported.
129 ARCHIVE_EXTRACT_UNLINK
130 Existing files on disk will be unlinked before any at‐
131 tempt to create them. In some cases, this can prove to
132 be a significant performance improvement. By default,
133 existing files are truncated and rewritten, but the file
134 is not recreated. In particular, the default behavior
135 does not break existing hard links.
136 ARCHIVE_EXTRACT_XATTR
137 Attempt to restore extended file attributes. By default,
138 they are ignored. See xattr(7) (Linux), xattr(2) (Mac OS
139 X), or getextattr(8) (FreeBSD) for more information on
140 extended file attributes.
141 archive_write_disk_set_group_lookup(),
143 archive_write_disk_set_user_lookup()
144 The struct archive_entry objects contain both names and ids that
145 can be used to identify users and groups. These names and ids
146 describe the ownership of the file itself and also appear in ACL
147 lists. By default, the library uses the ids and ignores the
148 names, but this can be overridden by registering user and group
149 lookup functions. To register, you must provide a lookup func‐
150 tion which accepts both a name and id and returns a suitable id.
151 You may also provide a void * pointer to a private data structure
152 and a cleanup function for that data. The cleanup function will
153 be invoked when the struct archive object is destroyed.
154 archive_write_disk_set_standard_lookup()
156 This convenience function installs a standard set of user and
157 group lookup functions. These functions use getpwnam(3) and
158 getgrnam(3) to convert names to ids, defaulting to the ids if the
159 names cannot be looked up. These functions also implement a sim‐
160 ple memory cache to reduce the number of calls to getpwnam(3) and
161 getgrnam(3).
162 More information about the struct archive object and the overall design
163 of the library can be found in the libarchive(3) overview. Many of these
164 functions are also documented under archive_write(3).
165
167 Most functions return ARCHIVE_OK (zero) on success, or one of several
168 non-zero error codes for errors. Specific error codes include:
169 ARCHIVE_RETRY for operations that might succeed if retried, ARCHIVE_WARN
170 for unusual conditions that do not prevent further operations, and
171 ARCHIVE_FATAL for serious errors that make remaining operations impossi‐
172 ble.
173 archive_write_disk_new() returns a pointer to a newly-allocated struct
175 archive object.
176 archive_write_data() returns a count of the number of bytes actually
178 written, or -1 on error.
179
181 Detailed error codes and textual descriptions are available from the
182 archive_errno() and archive_error_string() functions.
183
185 tar(1), archive_read(3), archive_write(3), libarchive(3)
186
188 The libarchive library first appeared in FreeBSD 5.3. The
189 archive_write_disk interface was added to libarchive 2.0 and first ap‐
190 peared in FreeBSD 6.3.
191
193 The libarchive library was written by Tim Kientzle <kientzle@acm.org>.
194
196 Directories are actually extracted in two distinct phases. Directories
197 are created during archive_write_header(), but final permissions are not
198 set until archive_write_close(). This separation is necessary to cor‐
199 rectly handle borderline cases such as a non-writable directory contain‐
200 ing files, but can cause unexpected results. In particular, directory
201 permissions are not fully restored until the archive is closed. If you
202 use chdir(2) to change the current directory between calls to
203 archive_read_extract() or before calling archive_read_close(), you may
204 confuse the permission-setting logic with the result that directory per‐
205 missions are restored incorrectly.
206 The library attempts to create objects with filenames longer than
208 PATH_MAX by creating prefixes of the full path and changing the current
209 directory. Currently, this logic is limited in scope; the fixup pass
210 does not work correctly for such objects and the symlink security check
211 option disables the support for very long pathnames.
212 Restoring the path aa/../bb does create each intermediate directory. In
214 particular, the directory aa is created as well as the final object bb.
215 In theory, this can be exploited to create an entire directory hierarchy
216 with a single request. Of course, this does not work if the
217 ARCHIVE_EXTRACT_NODOTDOT option is specified.
218 Implicit directories are always created obeying the current umask. Ex‐
220 plicit objects are created obeying the current umask unless
221 ARCHIVE_EXTRACT_PERM is specified, in which case they current umask is
222 ignored.
223 SGID and SUID bits are restored only if the correct user and group could
225 be set. If ARCHIVE_EXTRACT_OWNER is not specified, then no attempt is
226 made to set the ownership. In this case, SGID and SUID bits are restored
227 only if the user and group of the final object happen to match those
228 specified in the entry.
229 The “standard” user-id and group-id lookup functions are not the defaults
231 because getgrnam(3) and getpwnam(3) are sometimes too large for particu‐
232 lar applications. The current design allows the application author to
233 use a more compact implementation when appropriate.
234 There should be a corresponding archive_read_disk interface that walks a
236 directory hierarchy and returns archive entry objects.
237BSD January 19, 2020 BSD