def copy(self, src, dst, overwrite=False, chunk_size=1024*64):
"""An optimized copy() that will skip mounting an archive if one is involved
as either the src or dst. This allows the file containing the archive to be
copied."""
src_is_archive = libarchive.is_archive_name(src)
# If src path is a mounted archive, unmount it.
if src_is_archive and self.ismount(src):
self.unmount(src)
# Now delegate the path, if the path is an archive, don't remount it.
srcfs, _ignored, src = self._delegate(src, auto_mount=(not src_is_archive))
# Follow the same steps for dst.
dst_is_archive = libarchive.is_archive_name(dst)
if dst_is_archive and self.ismount(dst):
self.unmount(dst)
dstfs, _ignored, dst = self._delegate(dst, auto_mount=(not dst_is_archive))
# srcfs, src and dstfs, dst are now the file system and path for our src and dst.
if srcfs is dstfs and srcfs is not self:
# Both src and dst are on the same fs, let it do the copy.
srcfs.copy(src, dst, overwrite=overwrite, chunk_size=chunk_size)
else:
# Src and dst are on different file systems. Just do the copy...
srcfd = None
try:
srcfd = srcfs.open(src, 'rb')
dstfs.setcontents(dst, srcfd, chunk_size=chunk_size)
except ResourceNotFoundError:
if srcfs.exists(src) and not dstfs.exists(dirname(dst)):
raise ParentDirectoryMissingError(dst)
finally:
if srcfd:
srcfd.close()
def test_formats(self):
self.assertEqual(is_archive_name('foo'), None)
self.assertEqual(is_archive_name('foo.txt'), None)
self.assertEqual(is_archive_name('foo.txt.gz'), None)
self.assertEqual(is_archive_name('foo.tar.gz'), 'tar')
self.assertEqual(is_archive_name('foo.tar.bz2'), 'tar')
self.assertEqual(is_archive_name('foo.zip'), 'zip')
self.assertEqual(is_archive_name('foo.rar'), 'rar')
self.assertEqual(is_archive_name('foo.iso'), 'iso')
self.assertEqual(is_archive_name('foo.rpm'), 'cpio')
def test_formats(self):
self.assertEqual(is_archive_name('foo'), None)
self.assertEqual(is_archive_name('foo.txt'), None)
self.assertEqual(is_archive_name('foo.txt.gz'), None)
self.assertEqual(is_archive_name('foo.tar.gz'), 'tar')
self.assertEqual(is_archive_name('foo.tar.bz2'), 'tar')
self.assertEqual(is_archive_name('foo.zip'), 'zip')
self.assertEqual(is_archive_name('foo.rar'), 'rar')
self.assertEqual(is_archive_name('foo.iso'), 'iso')
self.assertEqual(is_archive_name('foo.rpm'), 'cpio')
def remove(self, path, **kwargs):
"""A remove() override that deletes an archive directly. It is not fooled
by a mounted archive. If the path is not an archive, the call is delegated."""
if libarchive.is_archive_name(path) and self.ismount(path):
self.unmount(path)
fs, _mount_path, delegate_path = self._delegate(path, auto_mount=False)
return fs.remove(delegate_path)
def open(self, path, *args, **kwargs):
"""An open() override that opens an archive. It is not fooled by mounted
archives. If the path is a mounted archive, it is unmounted and the archive
file is opened and returned."""
if libarchive.is_archive_name(path) and self.ismount(path):
self.unmount(path)
fs, _mount_path, delegate_path = self._delegate(path, auto_mount=False)
return fs.open(delegate_path, *args, **kwargs)
def rename(self, src, dst):
"""An rename() implementation that ensures the rename does not span
file systems. It also ensures that an archive can be renamed (without
trying to mount either the src or destination paths)."""
src_is_archive = libarchive.is_archive_name(src)
# If src path is a mounted archive, unmount it.
if src_is_archive and self.ismount(src):
self.unmount(src)
# Now delegate the path, if the path is an archive, don't remount it.
srcfs, _ignored, src = self._delegate(src, auto_mount=(not src_is_archive))
# Follow the same steps for dst.
dst_is_archive = libarchive.is_archive_name(dst)
if dst_is_archive and self.ismount(dst):
self.unmount(dst)
dstfs, _ignored, dst = self._delegate(dst, auto_mount=(not dst_is_archive))
# srcfs, src and dstfs, dst are now the file system and path for our src and dst.
if srcfs is dstfs and srcfs is not self:
# Both src and dst are on the same fs, let it do the copy.
return srcfs.rename(src, dst)
raise OperationFailedError("rename resource", path=src)
def remove(self, path):
# In case one of our mounted file systems backing archive is being
# deleted, unmout it before continuing. Once unmounted, the archive
# can be deleted by root fs, otherwise, the ArchiveFS will be asked
# to remove itself, which it cannot do.
if self.ismount(path) and libarchive.is_archive_name(path):
self.unmount(path)
# Send the delete directoy to the root filesystem. This avoids
# being delegated, and the fs we just unmounted being remounted.
return self.rootfs.remove(path)
# Otherwise, just delegate to the responsible fs.
return super(ArchiveMountFS, self).remove(path)
def _delegate(self, path):
for ppath in recursepath(path)[1:]:
# Don't mount again...
if self.ismount(ppath):
break
if libarchive.is_archive_name(ppath):
# It looks like an archive, try mounting it.
full_path = self.mount_tree['/'].fs.getsyspath(ppath)
try:
self.mountdir(ppath, ArchiveFS(full_path, 'r'))
except:
pass # Must NOT have been an archive after all
# Stop recursing path, we support just one archive per path!
# No nested archives yet!
break
return super(ArchiveMountFS, self)._delegate(path)
def _delegate(self, path, auto_mount=True):
"""A _delegate() override that will automatically mount archives that are
encountered in the path. For example, the path /foo/bar.zip/baz.txt contains
the archive path /foo/bar.zip. If this archive can be mounted by ArchiveFS,
it will be. Then the file system call will be delegated to that mounted file
system, which will act upon /baz.txt within the archive. This is lazy
initialization which means users of this class need not crawl the file system
for archives and mount them all up-front.
This behavior can be overridden by self.auto_mount=False or by passing the
auto_mount=False keyword argument.
"""
if self.auto_mount and auto_mount:
for ppath in recursepath(path)[1:]:
if self.ismount(ppath):
# If something is already mounted here, no need to continue.
break
if libarchive.is_archive_name(ppath):
# It looks like an archive, we might mount it.
# First check that the size is acceptable.
if self.max_size:
if self.rootfs.exists(ppath) and \
self.rootfs.getsize(ppath) > self.max_size:
break
# Looks good, the proof is in the pudding, so let's try to
# mount this *supposed* archive...
full_path = self.rootfs.getsyspath(ppath)
try:
# TODO: it would be really nice if we could open the path using
# self.rootfs.open(), that way we could support archives on a file
# system other than osfs (even nested archives). However, the libarchive
# wrapper is not sophisticated enough to handle a Python file-like object,
# it uses an actual fd.
self.mountdir(ppath, ArchiveFS(full_path, 'r'))
# That worked!! Stop recursing path, we support just one archive per path!
break
except:
# Must NOT have been an archive after all, but maybe
# there is one deeper in the directory...
continue
return super(ArchiveMountFS, self)._delegate(path)
def isarchive(path):
try:
return libarchive.is_archive(path)
except:
return libarchive.is_archive_name(path)
def isarchive(path):
try:
return libarchive.is_archive(path)
except:
return libarchive.is_archive_name(path)
