def get_file_node(rootdir, path, hash_cache, uid_map=None, gid_map=None):
metadata = get_default_metadata(rootdir, path, uid_map=uid_map, gid_map=gid_map)
native_path = utils.build_native_path(rootdir, path)
stat = os.lstat(native_path)
metadata['hash'] = hash_cache[path]
metadata['size'] = stat.st_size
return FileNode(**metadata)
def compute_digests(rootdir, paths, num_threads=None):
"""Return a dict with (path, digest)."""
# First, we write the temporary file with filenames in the format
# expected by hashdeep
rootdir = os.path.normpath(rootdir)
fd, tempfilename = tempfile.mkstemp()
with os.fdopen(fd, 'w') as tmpfile:
for p in paths:
tmpfile.write(utils.build_native_path(rootdir, p))
tmpfile.write('\n')
# Run hashdeep -c sha1,sha256 -f tempfilename -l -d (-j num_threads)
cmd = ['hashdeep', '-c', 'sha1,sha256', '-f', tempfilename, '-l', '-d']
if num_threads is not None:
cmd.extend(['-j', str(num_threads)])
output = subprocess.check_output(cmd)
# Delete temporary file
os.remove(tempfilename)
# Parse output
root = ET.fromstring(output)
res = {}
for fileobj in root:
if fileobj.tag != 'fileobject':
continue
name = None
sha1 = None
sha256 = None
for child in fileobj:
if child.tag == 'hashdigest':
if child.attrib['type'] == 'SHA1':
sha1 = child.text
elif child.attrib['type'] == 'SHA256':
sha256 = child.text
else:
raise ValueError('Unexpected hash type "{}".'.format(
child.attrib['type']))
if child.tag == 'filename':
name = utils.get_path_from_native_path(rootdir, child.text)
if not name or not sha1 or not sha256:
raise ValueError('Could not extract all required information from '
'digest.')
res[name] = Digest(sha1, sha256)
keys = res.keys()
if len(keys) != len(paths) or set(keys) != set(paths):
raise ValueError('List of filenames returned by hashdeep does not '
'match the input list.')
return res
def get_default_metadata(rootdir, path, uid_map=None, gid_map=None):
"""Returns a dictonary, whose keys are default_metadata (as
defined above) and entries are the results of the corresponding
os.stat call on native_path."""
if uid_map is None:
uid_map = utils.get_uid_name_map()
if gid_map is None:
gid_map = utils.get_gid_name_map()
native_path = utils.build_native_path(rootdir, path)
stat = os.lstat(native_path)
metadata = {}
metadata['name'] = os.path.basename(path)
# TODO: use strings instead of UNIX epoch seconds
# TODO: fields we are currently not using: st_mode, st_ino, st_dev, st_nlink
# - is this intentional?
metadata['mtime'] = stat.st_mtime
metadata['user'] = uid_map.get(stat.st_uid, str(stat.st_uid))
metadata['group'] = gid_map.get(stat.st_gid, str(stat.st_gid))
metadata['permissions'] = '%04o' % stat.st_mode
return metadata
def get_symlink_node(rootdir, path, uid_map=None, gid_map=None):
metadata = get_default_metadata(rootdir, path, uid_map=uid_map, gid_map=gid_map)
native_path = utils.build_native_path(rootdir, path)
metadata['link_target'] = os.readlink(native_path)
return SymlinkNode(**metadata)
def test_build_native_path_simple_1():
rootdir = '/home/x'
path = '/foo'
res = utils.build_native_path(rootdir, path)
assert res == '/home/x/foo'
def test_build_native_path_non_normalized_path():
rootdir = '/home/x'
path = '/foo/../bar'
with pytest.raises(ValueError):
res = utils.build_native_path(rootdir, path)
def test_build_native_path_non_absolute_rootdir():
rootdir = 'root'
path = '/'
with pytest.raises(ValueError):
res = utils.build_native_path(rootdir, path)
def test_build_native_path_simple_base_rootdir_base_path():
rootdir = '/'
path = '/'
res = utils.build_native_path(rootdir, path)
assert res == '/'
