def test_open(tmp_dir, dvc):
tmp_dir.gen("foo", "foo")
dvc.add("foo")
(tmp_dir / "foo").unlink()
fs = DataFileSystem(repo=dvc)
with fs.open("foo", "r") as fobj:
assert fobj.read() == "foo"
def test_open_dirty_hash(tmp_dir, dvc):
tmp_dir.dvc_gen("file", "file")
(tmp_dir / "file").write_text("something")
fs = DataFileSystem(repo=dvc)
with fs.open("file", "r") as fobj:
# NOTE: Unlike DvcFileSystem, DataFileSystem should not
# be affected by a dirty workspace.
assert fobj.read() == "file"
def test_open_dirty_no_hash(tmp_dir, dvc):
tmp_dir.gen("file", "file")
(tmp_dir / "file.dvc").write_text("outs:\n- path: file\n")
fs = DataFileSystem(repo=dvc)
# NOTE: Unlike DvcFileSystem, DataFileSystem should not
# be affected by a dirty workspace.
with pytest.raises(FileNotFoundError):
with fs.open("file", "r"):
pass
def test_open_no_remote(tmp_dir, dvc):
tmp_dir.dvc_gen("file", "file")
(tmp_dir / "file").unlink()
remove(dvc.odb.local.cache_dir)
fs = DataFileSystem(repo=dvc)
with pytest.raises(FileNotFoundError) as exc_info:
with fs.open("file", "r"):
pass
assert isinstance(exc_info.value.__cause__, NoRemoteError)
def test_open_in_history(tmp_dir, scm, dvc):
tmp_dir.gen("foo", "foo")
dvc.add("foo")
dvc.scm.add(["foo.dvc", ".gitignore"])
dvc.scm.commit("foo")
tmp_dir.gen("foo", "foofoo")
dvc.add("foo")
dvc.scm.add(["foo.dvc", ".gitignore"])
dvc.scm.commit("foofoo")
for rev in dvc.brancher(revs=["HEAD~1"]):
if rev == "workspace":
continue
fs = DataFileSystem(repo=dvc)
with fs.open("foo", "r") as fobj:
assert fobj.read() == "foo"
def open_by_relpath(self, path, remote=None, mode="r", encoding=None):
"""Opens a specified resource as a file descriptor"""
from dvc.fs.data import DataFileSystem
from dvc.fs.dvc import DvcFileSystem
if os.path.isabs(path):
fs = DataFileSystem(repo=self, workspace="local")
fs_path = path
else:
fs = DvcFileSystem(repo=self, subrepos=True)
fs_path = fs.from_os_path(path)
try:
with fs.open(
fs_path,
mode=mode,
encoding=encoding,
remote=remote,
) as fobj:
yield fobj
except FileNotFoundError as exc:
raise FileMissingError(path) from exc
except IsADirectoryError as exc:
raise DvcIsADirectoryError(f"'{path}' is a directory") from exc