archive_read_disk(3) BSD Library Functions Manual archive_read_disk(3)
archive_read_disk_entry_from_file, archive_read_disk_gname, archive_read_disk_uname,
archive_read_disk_set_standard_lookup, archive_read_close, archive_read_finish -- functions
for reading objects from disk
struct archive *
archive_read_disk_set_symlink_logical(struct archive *);
archive_read_disk_set_symlink_physical(struct archive *);
archive_read_disk_set_symlink_hybrid(struct archive *);
archive_read_disk_gname(struct archive *, gid_t);
archive_read_disk_uname(struct archive *, uid_t);
archive_read_disk_set_gname_lookup(struct archive *, void *,
const char *(*lookup)(void *, gid_t), void (*cleanup)(void *));
archive_read_disk_set_uname_lookup(struct archive *, void *,
const char *(*lookup)(void *, uid_t), void (*cleanup)(void *));
archive_read_disk_set_standard_lookup(struct archive *);
archive_read_disk_entry_from_file(struct archive *, struct archive_entry *, int fd,
const struct stat *);
archive_read_close(struct archive *);
archive_read_finish(struct archive *);
These functions provide an API for reading information about objects on disk. In particu-
lar, they provide an interface for populating struct archive_entry objects.
Allocates and initializes a struct archive object suitable for reading object infor-
mation from disk.
This sets the mode used for handling symbolic links. The ``logical'' mode follows
all symbolic links. The ``physical'' mode does not follow any symbolic links. The
``hybrid'' mode currently behaves identically to the ``logical'' mode.
Returns a user or group name given a gid or uid value. By default, these always
return a NULL string.
These allow you to override the functions used for user and group name lookups. You
may also provide a void * pointer to a private data structure and a cleanup function
for that data. The cleanup function will be invoked when the struct archive object
is destroyed or when new lookup functions are registered.
This convenience function installs a standard set of user and group name lookup
functions. These functions use getpwid(3) and getgrid(3) to convert ids to names,
defaulting to NULL if the names cannot be looked up. These functions also implement
a simple memory cache to reduce the number of calls to getpwid(3) and getgrid(3).
Populates a struct archive_entry object with information about a particular file.
The archive_entry object must have already been created with archive_entry_new(3)
and at least one of the source path or path fields must already be set. (If both
are set, the source path will be used.)
Information is read from disk using the path name from the struct archive_entry
object. If a file descriptor is provided, some information will be obtained using
that file descriptor, on platforms that support the appropriate system calls.
If a pointer to a struct stat is provided, information from that structure will be
used instead of reading from the disk where appropriate. This can provide perfor-
mance benefits in scenarios where struct stat information has already been read from
the disk as a side effect of some other operation. (For example, directory traver-
sal libraries often provide this information.)
Where necessary, user and group ids are converted to user and group names using the
currently registered lookup functions above. This affects the file ownership fields
and ACL values in the struct archive_entry object.
This currently does nothing.
Invokes archive_write_close() if it was not invoked manually, then releases all
More information about the struct archive object and the overall design of the library can
be found in the libarchive(3) overview.
The following illustrates basic usage of the library by showing how to use it to copy an
item on disk into an archive.
file_to_archive(struct archive *a, const char *name)
struct archive *ard;
struct archive_entry *entry;
ard = archive_read_disk_new();
entry = archive_entry_new();
fd = open(name, O_RDONLY);
if (fd < 0)
archive_read_disk_entry_from_file(ard, entry, fd, NULL);
while ((bytes_read = read(fd, buff, sizeof(buff))) > 0)
archive_write_data(a, buff, bytes_read);
Most functions return ARCHIVE_OK (zero) on success, or one of several negative error codes
for errors. Specific error codes include: ARCHIVE_RETRY for operations that might succeed
if retried, ARCHIVE_WARN for unusual conditions that do not prevent further operations, and
ARCHIVE_FATAL for serious errors that make remaining operations impossible. The
archive_errno(3) and archive_error_string(3) functions can be used to retrieve an appropri-
ate error code and a textual error message. (See archive_util(3) for details.)
archive_read_disk_new() returns a pointer to a newly-allocated struct archive object or NULL
if the allocation failed for any reason.
archive_read_disk_gname() and archive_read_disk_uname() return const char * pointers to the
textual name or NULL if the lookup failed for any reason. The returned pointer points to
internal storage that may be reused on the next call to either of these functions; callers
should copy the string if they need to continue accessing it.
archive_read(3), archive_write(3), archive_write_disk(3), tar(1), libarchive(3)
The libarchive library first appeared in FreeBSD 5.3. The archive_read_disk interface was
added to libarchive 2.6 and first appeared in FreeBSD 8.0.
The libarchive library was written by Tim Kientzle <firstname.lastname@example.org>.
The ``standard'' user name and group name lookup functions are not the defaults because
getgrid(3) and getpwid(3) are sometimes too large for particular applications. The current
design allows the application author to use a more compact implementation when appropriate.
The full list of metadata read from disk by archive_read_disk_entry_from_file() is necessar-
The archive_read_disk_entry_from_file() function reads as much information as it can from
disk. Some method should be provided to limit this so that clients who do not need ACLs,
for instance, can avoid the extra work needed to look up such information.
This API should provide a set of methods for walking a directory tree. That would make it a
direct parallel of the archive_read(3) API. When such methods are implemented, the
``hybrid'' symbolic link mode will make sense.
BSD March 10, 2009 BSD