def _unprotect_file(path):
if System.is_symlink(path) or System.is_hardlink(path):
logger.debug("Unprotecting '{}'".format(path))
tmp = os.path.join(os.path.dirname(path), "." + uuid())
# The operations order is important here - if some application
# would access the file during the process of copyfile then it
# would get only the part of file. So, at first, the file should be
# copied with the temporary name, and then original file should be
# replaced by new.
copyfile(path, tmp, name="Unprotecting '{}'".format(relpath(path)))
remove(path)
os.rename(tmp, path)
else:
logger.debug("Skipping copying for '{}', since it is not "
"a symlink or a hardlink.".format(path))
os.chmod(path, os.stat(path).st_mode | stat.S_IWRITE)
def _unprotect_file(self, path):
import stat
import uuid
from dvc.system import System
from dvc.utils import copyfile, move, remove
if System.is_symlink(path) or System.is_hardlink(path):
logger.debug("Unprotecting '{}'".format(path))
tmp = os.path.join(os.path.dirname(path), "." + str(uuid.uuid4()))
move(path, tmp)
copyfile(tmp, path)
remove(tmp)
else:
logger.debug("Skipping copying for '{}', since it is not "
"a symlink or a hardlink.".format(path))
os.chmod(path, os.stat(path).st_mode | stat.S_IWRITE)
def test_hardlink_optimization(tmp_dir, dvc, ssh):
from dvc.fs import get_cloud_fs
cls, config, path_info = get_cloud_fs(dvc, **ssh.config)
fs = cls(**config)
assert isinstance(fs, SSHFileSystem)
from_info = path_info / "empty"
to_info = path_info / "link"
with fs.open(from_info, "wb"):
pass
if os.name == "nt":
link_path = "c:" + to_info.path.replace("/", "\\")
else:
link_path = to_info.path
fs.hardlink(from_info, to_info)
assert not System.is_hardlink(link_path)
def iscopy(self, path_info):
return not (System.is_symlink(path_info)
or System.is_hardlink(path_info))
def is_hardlink(path_info):
return System.is_hardlink(path_info)
tmp_dir.dvc_gen({"foo": "foo", "bar": "bar"})
shutil.copy("bar", "foo")
copy_spy = mocker.spy(dvc.cache.local.tree, "copy")
dvc.add("foo")
assert copy_spy.mock.call_count == 0
@pytest.mark.parametrize(
"link,new_link,link_test_func",
[
("hardlink", "copy", lambda path: not System.is_hardlink(path)),
("symlink", "copy", lambda path: not System.is_symlink(path)),
("copy", "hardlink", System.is_hardlink),
("copy", "symlink", System.is_symlink),
],
)
def test_should_relink_on_repeated_add(link, new_link, link_test_func, tmp_dir,
dvc):
from dvc.path_info import PathInfo
dvc.config["cache"]["type"] = link
tmp_dir.dvc_gen({"foo": "foo", "bar": "bar"})
os.remove("foo")
getattr(dvc.cache.local.tree, link)(PathInfo("bar"), PathInfo("foo"))
def is_hardlink(path):
return System.is_hardlink(path)
def test_reflink(repo_dir, ssh):
ssh.reflink("foo", "link")
assert filecmp.cmp("foo", "link")
assert not System.is_symlink("link")
assert not System.is_hardlink("link")
def test_hardlink(repo_dir, ssh):
ssh.hardlink("foo", "link")
assert System.is_hardlink("link")
def test_reflink(tmp_dir, ssh):
tmp_dir.gen("foo", "foo content")
ssh.reflink("foo", "link")
assert filecmp.cmp("foo", "link")
assert not System.is_symlink("link")
assert not System.is_hardlink("link")
def test_hardlink(tmp_dir, ssh):
tmp_dir.gen("foo", "foo content")
ssh.hardlink("foo", "link")
assert System.is_hardlink("link")
