def test(self):
with open(self.CODE, "w+", encoding="utf-8") as fobj:
fobj.write("import sys\n")
fobj.write("import os\n")
fobj.write("with open(sys.argv[1], 'a+') as fobj:\n")
fobj.write(" fobj.write('foo')\n")
ret = main(["config", "cache.type", "hardlink"])
self.assertEqual(ret, 0)
self.assertEqual(ret, 0)
ret = main([
"run",
"-d",
self.CODE,
"-o",
self.FOO,
"--single-stage",
"python",
self.CODE,
self.FOO,
])
self.assertEqual(ret, 0)
self.assertFalse(os.access(self.FOO, os.W_OK))
self.assertTrue(system.is_hardlink(self.FOO))
with open(self.FOO, encoding="utf-8") as fd:
self.assertEqual(fd.read(), "foo")
ret = main([
"run",
"--force",
"--no-run-cache",
"--single-stage",
"-d",
self.CODE,
"-o",
self.FOO,
"python",
self.CODE,
self.FOO,
])
self.assertEqual(ret, 0)
self.assertFalse(os.access(self.FOO, os.W_OK))
self.assertTrue(system.is_hardlink(self.FOO))
with open(self.FOO, encoding="utf-8") as fd:
self.assertEqual(fd.read(), "foo")
def test(self):
ret = main(["config", "cache.type", "hardlink"])
self.assertEqual(ret, 0)
ret = main(["add", self.FOO])
self.assertEqual(ret, 0)
self.assertFalse(os.access(self.FOO, os.W_OK))
self.assertTrue(system.is_hardlink(self.FOO))
ret = main(["unprotect", self.FOO])
self.assertEqual(ret, 0)
ret = main(["add", self.FOO])
self.assertEqual(ret, 0)
self.assertFalse(os.access(self.FOO, os.W_OK))
self.assertTrue(system.is_hardlink(self.FOO))
def test_add_optimization_for_hardlink_on_empty_files(tmp_dir, dvc, mocker):
dvc.odb.local.cache_types = ["hardlink"]
tmp_dir.gen({"foo": "", "bar": "", "lorem": "lorem", "ipsum": "ipsum"})
m = mocker.spy(LocalFileSystem, "is_hardlink")
stages = dvc.add(["foo", "bar", "lorem", "ipsum"])
assert m.call_count == 8
assert m.call_args != call(tmp_dir / "foo")
assert m.call_args != call(tmp_dir / "bar")
for stage in stages[:2]:
# hardlinks are not created for empty files
assert not system.is_hardlink(stage.outs[0].fs_path)
for stage in stages[2:]:
assert system.is_hardlink(stage.outs[0].fs_path)
for stage in stages:
assert os.path.exists(stage.path)
assert os.path.exists(stage.outs[0].cache_path)
tmp_dir.dvc_gen({"foo": "foo", "bar": "bar"})
shutil.copy("bar", "foo")
copy_spy = mocker.spy(dvc.odb.local.fs, "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):
dvc.config["cache"]["type"] = link
tmp_dir.dvc_gen({"foo": "foo", "bar": "bar"})
os.remove("foo")
getattr(dvc.odb.local.fs, link)((tmp_dir / "bar").fs_path,
(tmp_dir / "foo").fs_path)