def test_get_hash_granular(tmp_dir, dvc):
tmp_dir.dvc_gen(
{"dir": {"foo": "foo", "bar": "bar", "subdir": {"data": "data"}}}
)
fs = DataFileSystem(repo=dvc)
subdir = "dir/subdir"
assert fs.info(subdir).get("md5") is None
_, _, obj = stage(dvc.odb.local, subdir, fs, "md5", dry_run=True)
assert obj.hash_info == HashInfo(
"md5", "af314506f1622d107e0ed3f14ec1a3b5.dir"
)
data = posixpath.join(subdir, "data")
assert fs.info(data)["md5"] == "8d777f385d3dfec8815d20f7496026dc"
_, _, obj = stage(dvc.odb.local, data, fs, "md5", dry_run=True)
assert obj.hash_info == HashInfo("md5", "8d777f385d3dfec8815d20f7496026dc")
def test_get_hash_dir(tmp_dir, dvc, mocker):
tmp_dir.dvc_gen(
{"dir": {"foo": "foo", "bar": "bar", "subdir": {"data": "data"}}}
)
fs = DataFileSystem(repo=dvc)
hash_file_spy = mocker.spy(dvc_data.hashfile.hash, "hash_file")
assert fs.info("dir")["md5"] == "8761c4e9acad696bee718615e23e22db.dir"
assert not hash_file_spy.called
def test_get_hash_dirty_dir(tmp_dir, dvc):
tmp_dir.dvc_gen({"dir": {"foo": "foo", "bar": "bar"}})
(tmp_dir / "dir" / "baz").write_text("baz")
fs = DataFileSystem(repo=dvc)
expected = "5ea40360f5b4ec688df672a4db9c17d1.dir"
assert fs.info("dir").get("md5") == expected
_, _, obj = stage(dvc.odb.local, "dir", fs, "md5", dry_run=True)
assert obj.hash_info == HashInfo("md5", expected)
def test_get_hash_dirty_file(tmp_dir, dvc):
tmp_dir.dvc_gen("file", "file")
(tmp_dir / "file").write_text("something")
fs = DataFileSystem(repo=dvc)
expected = "8c7dd922ad47494fc02c388e12c00eac"
assert fs.info("file").get("md5") == expected
_, _, obj = stage(dvc.odb.local, "file", fs, "md5", dry_run=True)
assert obj.hash_info == HashInfo("md5", expected)
def test_get_hash_file(tmp_dir, dvc):
tmp_dir.dvc_gen({"foo": "foo"})
fs = DataFileSystem(repo=dvc)
assert fs.info("foo")["md5"] == "acbd18db4cc2f85cedef654fccc4a4d8"