def test_cache_checksums(dvc):
remote = BaseRemote(dvc, {})
remote.path_info = PathInfo("foo")
with mock.patch.object(
remote, "list_cache_paths", return_value=["12/3456", "bar"]
):
checksums = list(remote.cache_checksums())
assert checksums == ["123456"]
def test_cache_checksums_traverse(path_to_checksum, cache_checksums, dvc):
remote = BaseRemote(dvc, {})
remote.path_info = PathInfo("foo")
# parallel traverse
size = 256 / remote.JOBS * remote.LIST_OBJECT_PAGE_SIZE
list(remote._cache_checksums_traverse(size, {0}))
for i in range(1, 16):
cache_checksums.assert_any_call(prefix="{:03x}".format(i),
progress_callback=CallableOrNone)
for i in range(1, 256):
cache_checksums.assert_any_call(prefix="{:02x}".format(i),
progress_callback=CallableOrNone)
# default traverse (small remote)
size -= 1
cache_checksums.reset_mock()
list(remote._cache_checksums_traverse(size - 1, {0}))
cache_checksums.assert_called_with(prefix=None,
progress_callback=CallableOrNone)
def cache_exists(self, *args, **kwargs):
return BaseRemote.cache_exists(self, *args, **kwargs)
def test_cache_exists(object_exists, traverse, dvc):
remote = BaseRemote(dvc, {})
# remote does not support traverse
remote.CAN_TRAVERSE = False
with mock.patch.object(
remote, "cache_checksums", return_value=list(range(256))
):
checksums = set(range(1000))
remote.cache_exists(checksums)
object_exists.assert_called_with(checksums, None, None)
traverse.assert_not_called()
remote.CAN_TRAVERSE = True
# large remote, small local
object_exists.reset_mock()
traverse.reset_mock()
with mock.patch.object(
remote, "cache_checksums", return_value=list(range(256))
):
checksums = list(range(1000))
remote.cache_exists(checksums)
# verify that _cache_paths_with_max() short circuits
# before returning all 256 remote checksums
max_checksums = math.ceil(
remote._max_estimation_size(checksums)
/ pow(16, remote.TRAVERSE_PREFIX_LEN)
)
assert max_checksums < 256
object_exists.assert_called_with(
frozenset(range(max_checksums, 1000)), None, None
)
traverse.assert_not_called()
# large remote, large local
object_exists.reset_mock()
traverse.reset_mock()
remote.JOBS = 16
with mock.patch.object(
remote, "cache_checksums", return_value=list(range(256))
):
checksums = list(range(1000000))
remote.cache_exists(checksums)
object_exists.assert_not_called()
traverse.assert_called_with(
256 * pow(16, remote.TRAVERSE_PREFIX_LEN),
set(range(256)),
None,
None,
)
def test_is_dir_checksum(checksum, result):
assert BaseRemote.is_dir_checksum(checksum) == result