def test(self):
self.swap_foo_with_bar()
stages = self.dvc.reproduce(self.foo_stage.path)
self.assertTrue(filecmp.cmp(self.FOO, self.BAR, shallow=False))
self.assertEqual(stages[0].outs[0].hash_info.value,
file_md5(self.BAR, self.dvc.fs))
def test_state(tmp_dir, dvc):
tmp_dir.gen("foo", "foo content")
path = tmp_dir / "foo"
hash_info = HashInfo("md5", file_md5(path, dvc.fs))
state = State(dvc.root_dir, dvc.tmp_dir, dvc.dvcignore)
state.save(path, dvc.fs, hash_info)
assert state.get(path, dvc.fs)[1] == hash_info
path.unlink()
path.write_text("1")
assert state.get(path, dvc.fs) == (None, None)
hash_info = HashInfo("md5", file_md5(path, dvc.fs))
state.save(path, dvc.fs, hash_info)
assert state.get(path, dvc.fs)[1] == hash_info
def test_add(tmp_dir, dvc):
(stage, ) = tmp_dir.dvc_gen({"foo": "foo"})
md5 = file_md5("foo", dvc.fs)
assert stage is not None
assert isinstance(stage, Stage)
assert os.path.isfile(stage.path)
assert len(stage.outs) == 1
assert len(stage.deps) == 0
assert stage.cmd is None
assert stage.outs[0].hash_info == HashInfo("md5", md5)
assert stage.md5 is None
assert (tmp_dir / "foo.dvc").parse() == {
"outs": [{
"md5": "acbd18db4cc2f85cedef654fccc4a4d8",
"path": "foo",
"size": 3,
}]
}
def test(self):
cmd = "python {} {} {}".format(self.CODE, self.FOO, "out")
deps = [self.FOO, self.CODE]
outs = [os.path.join(self.dvc.root_dir, "out")]
outs_no_cache = []
fname = "out.dvc"
self.dvc.add(self.FOO)
stage = self.dvc.run(
cmd=cmd,
deps=deps,
outs=outs,
outs_no_cache=outs_no_cache,
fname=fname,
single_stage=True,
)
self.assertTrue(filecmp.cmp(self.FOO, "out", shallow=False))
self.assertTrue(os.path.isfile(stage.path))
self.assertEqual(stage.cmd, cmd)
self.assertEqual(len(stage.deps), len(deps))
self.assertEqual(len(stage.outs), len(outs + outs_no_cache))
self.assertEqual(stage.outs[0].fspath, outs[0])
self.assertEqual(stage.outs[0].hash_info.value,
file_md5(self.FOO, self.dvc.fs))
self.assertTrue(stage.path, fname)
with self.assertRaises(OutputDuplicationError):
self.dvc.run(
cmd=cmd,
deps=deps,
outs=outs,
outs_no_cache=outs_no_cache,
fname="duplicate" + fname,
single_stage=True,
)
def test_pre_push_hook(self, tmp_dir, scm, dvc, tmp_path_factory):
temp = tmp_path_factory.mktemp("external")
git_remote = temp / "project.git"
storage_path = temp / "dvc_storage"
with dvc.config.edit() as conf:
conf["remote"]["store"] = {"url": os.fspath(storage_path)}
conf["core"]["remote"] = "store"
tmp_dir.dvc_gen("file", "file_content", "commit message")
file_checksum = file_md5("file", dvc.fs)
expected_storage_path = (
storage_path / file_checksum[:2] / file_checksum[2:]
)
scm.gitpython.repo.clone(os.fspath(git_remote))
scm.gitpython.repo.create_remote("origin", os.fspath(git_remote))
dvc.install()
assert not expected_storage_path.is_file()
scm.gitpython.repo.git.push("origin", "master")
assert expected_storage_path.is_file()
assert expected_storage_path.read_text() == "file_content"