def test_staging_file(tmp_dir, dvc):
from dvc.data import check
from dvc.data.stage import stage
from dvc.data.transfer import transfer
tmp_dir.gen("foo", "foo")
fs = LocalFileSystem()
local_odb = dvc.odb.local
staging_odb, _, obj = stage(local_odb, (tmp_dir / "foo").fs_path, fs,
"md5")
assert not local_odb.exists(obj.hash_info)
assert staging_odb.exists(obj.hash_info)
with pytest.raises(FileNotFoundError):
check(local_odb, obj)
check(staging_odb, obj)
transfer(staging_odb, local_odb, {obj.hash_info}, hardlink=True)
check(local_odb, obj)
check(staging_odb, obj)
path = local_odb.hash_to_path(obj.hash_info.value)
assert fs.exists(path)
def test_get_hash_dirty_file(tmp_dir, dvc):
from dvc.data import check
from dvc.data.stage import get_file_hash
from dvc.objects.errors import ObjectFormatError
tmp_dir.dvc_gen("file", "file")
file_hash_info = HashInfo("md5", "8c7dd922ad47494fc02c388e12c00eac")
(tmp_dir / "file").write_text("something")
something_hash_info = HashInfo("md5", "437b930db84b8079c2dd804a71936b5f")
clean_staging()
# file is modified in workspace
# get_file_hash(file) should return workspace hash, not DVC cached hash
fs = RepoFileSystem(repo=dvc)
assert fs.info((tmp_dir / "file").fs_path).get("md5") is None
staging, _, obj = stage(dvc.odb.local, (tmp_dir / "file").fs_path, fs,
"md5")
assert obj.hash_info == something_hash_info
check(staging, obj)
# file is removed in workspace
# any staged object referring to modified workspace obj is now invalid
(tmp_dir / "file").unlink()
with pytest.raises(ObjectFormatError):
check(staging, obj)
# get_file_hash(file) should return DVC cached hash
assert fs.info((tmp_dir / "file").fs_path)["md5"] == file_hash_info.value
_, hash_info = get_file_hash((tmp_dir / "file").fs_path,
fs,
"md5",
state=dvc.state)
assert hash_info == file_hash_info
# tmp_dir/file can be staged even though it is missing in workspace since
# repofs will use the DVC cached hash (and refer to the local cache object)
_, _, obj = stage(dvc.odb.local, (tmp_dir / "file").fs_path, fs, "md5")
assert obj.hash_info == file_hash_info