def walk_files(self, fs: BaseFileSystem, path_info: AnyPath, **kwargs):
if fs.scheme == Schemes.LOCAL:
for root, _, files in self.walk(fs, path_info, **kwargs):
for file in files:
# NOTE: os.path.join is ~5.5 times slower
yield PathInfo(f"{root}{os.sep}{file}")
else:
yield from fs.walk_files(path_info)
def test_list_hashes(dvc):
odb = ObjectDB(BaseFileSystem(), None)
odb.path_info = PathInfo("foo")
with mock.patch.object(odb, "_list_paths", return_value=["12/3456",
"bar"]):
hashes = list(odb.list_hashes())
assert hashes == ["123456"]
def test_list_hashes(dvc):
cache = CloudCache(BaseFileSystem(dvc, {}))
cache.fs.path_info = PathInfo("foo")
with mock.patch.object(cache,
"_list_paths",
return_value=["12/3456", "bar"]):
hashes = list(cache.list_hashes())
assert hashes == ["123456"]
def test_list_paths(dvc):
path_info = PathInfo("foo")
odb = ObjectDB(BaseFileSystem(), path_info)
with mock.patch.object(odb.fs, "walk_files", return_value=[]) as walk_mock:
for _ in odb._list_paths():
pass
walk_mock.assert_called_with(path_info, prefix=False)
for _ in odb._list_paths(prefix="000"):
pass
walk_mock.assert_called_with(path_info / "00" / "0", prefix=True)
def test_cmd_error(dvc):
config = {}
cmd = "sed 'hello'"
ret = "1"
err = "sed: expression #1, char 2: extra characters after command"
with mock.patch.object(
BaseFileSystem,
"remove",
side_effect=RemoteCmdError("base", cmd, ret, err),
):
with pytest.raises(RemoteCmdError):
BaseFileSystem(**config).remove("file")
def test_list_paths(dvc):
cache = CloudCache(BaseFileSystem(dvc, {}))
cache.fs.path_info = PathInfo("foo")
with mock.patch.object(cache.fs, "walk_files",
return_value=[]) as walk_mock:
for _ in cache._list_paths():
pass
walk_mock.assert_called_with(cache.fs.path_info, prefix=False)
for _ in cache._list_paths(prefix="000"):
pass
walk_mock.assert_called_with(cache.fs.path_info / "00" / "0",
prefix=True)
def test_hashes_exist(object_exists, traverse, dvc):
cache = CloudCache(BaseFileSystem(dvc, {}))
# remote does not support traverse
cache.fs.CAN_TRAVERSE = False
with mock.patch.object(cache, "list_hashes",
return_value=list(range(256))):
hashes = set(range(1000))
cache.hashes_exist(hashes)
object_exists.assert_called_with(hashes, None, None)
traverse.assert_not_called()
cache.fs.CAN_TRAVERSE = True
# large remote, small local
object_exists.reset_mock()
traverse.reset_mock()
with mock.patch.object(cache, "list_hashes",
return_value=list(range(256))):
hashes = list(range(1000))
cache.hashes_exist(hashes)
# verify that _cache_paths_with_max() short circuits
# before returning all 256 remote hashes
max_hashes = math.ceil(
cache._max_estimation_size(hashes) /
pow(16, cache.fs.TRAVERSE_PREFIX_LEN))
assert max_hashes < 256
object_exists.assert_called_with(frozenset(range(max_hashes, 1000)),
None, None)
traverse.assert_not_called()
# large remote, large local
object_exists.reset_mock()
traverse.reset_mock()
cache.fs.JOBS = 16
with mock.patch.object(cache, "list_hashes",
return_value=list(range(256))):
hashes = list(range(1000000))
cache.hashes_exist(hashes)
object_exists.assert_not_called()
traverse.assert_called_with(
256 * pow(16, cache.fs.TRAVERSE_PREFIX_LEN),
set(range(256)),
None,
None,
)
def test_list_hashes_traverse(_path_to_hash, list_hashes, dvc):
odb = ObjectDB(BaseFileSystem())
odb.fs.path_info = PathInfo("foo")
# parallel traverse
size = 256 / odb.fs._JOBS * odb.fs.LIST_OBJECT_PAGE_SIZE
list(odb.list_hashes_traverse(size, {0}))
for i in range(1, 16):
list_hashes.assert_any_call(prefix=f"{i:03x}",
progress_callback=CallableOrNone)
for i in range(1, 256):
list_hashes.assert_any_call(prefix=f"{i:02x}",
progress_callback=CallableOrNone)
# default traverse (small remote)
size -= 1
list_hashes.reset_mock()
list(odb.list_hashes_traverse(size - 1, {0}))
list_hashes.assert_called_with(prefix=None,
progress_callback=CallableOrNone)
def test_is_dir_hash(hash_, result):
assert BaseFileSystem.is_dir_hash(hash_) == result
def test_missing_deps(pkg, msg, mocker):
requires = {"missing": "missing"}
mocker.patch.object(BaseFileSystem, "REQUIRES", requires)
mocker.patch("dvc.utils.pkg.PKG", pkg)
with pytest.raises(RemoteMissingDepsError, match=msg):
BaseFileSystem()
