def __exit__(self, *kwargs):
super().__exit__(*kwargs)
repo = Repo(get_path("root"))
repo.add(files)
for file in files:
repo.commit(file)
repo.push(files)
class TestCheckoutCorruptedCacheDir(TestDvc):
def test(self):
# NOTE: using 'copy' so that cache and link don't have same inode
ret = main(["config", "cache.type", "copy"])
self.assertEqual(ret, 0)
self.dvc = DvcRepo(".")
stages = self.dvc.add(self.DATA_DIR)
self.assertEqual(len(stages), 1)
self.assertEqual(len(stages[0].outs), 1)
out = stages[0].outs[0]
# NOTE: modifying cache file for one of the files inside the directory
# to check if dvc will detect that the cache is corrupted.
_, entry_hash = next(
self.dvc.cache.local.load_dir_cache(out.hash_info).items())
cache = os.fspath(
self.dvc.cache.local.tree.hash_to_path_info(entry_hash.value))
os.chmod(cache, 0o644)
with open(cache, "w+") as fobj:
fobj.write("1")
with pytest.raises(CheckoutError):
self.dvc.checkout(force=True)
self.assertFalse(os.path.exists(cache))
class TestCheckoutCorruptedCacheDir(TestDvc):
def test(self):
# NOTE: using 'copy' so that cache and link don't have same inode
ret = main(["config", "cache.type", "copy"])
self.assertEqual(ret, 0)
self.dvc = DvcRepo(".")
stages = self.dvc.add(self.DATA_DIR)
self.assertEqual(len(stages), 1)
self.assertEqual(len(stages[0].outs), 1)
out = stages[0].outs[0]
# NOTE: modifying cache file for one of the files inside the directory
# to check if dvc will detect that the cache is corrupted.
entry = self.dvc.cache.local.load_dir_cache(out.checksum)[0]
checksum = entry[self.dvc.cache.local.PARAM_CHECKSUM]
cache = self.dvc.cache.local.get(checksum)
os.chmod(cache, 0o644)
with open(cache, "w+") as fobj:
fobj.write("1")
with pytest.raises(CheckoutError):
self.dvc.checkout(force=True)
self.assertFalse(os.path.exists(cache))
def test_update_import(dvc_repo, erepo):
src = "version"
dst = src
stage = dvc_repo.imp(erepo.root_dir, src, dst, rev="branch")
assert os.path.exists(dst)
assert os.path.isfile(dst)
with open(dst, "r+") as fobj:
assert fobj.read() == "branch"
# update data
repo = Repo(erepo.root_dir)
saved_dir = os.getcwd()
os.chdir(erepo.root_dir)
repo.scm.checkout("branch")
os.unlink("version")
erepo.create("version", "updated")
repo.add("version")
repo.scm.add([".gitignore", "version.dvc"])
repo.scm.commit("updated")
repo.scm.checkout("master")
repo.scm.close()
os.chdir(saved_dir)
# Caching in external repos doesn't see upstream updates within single
# cli call, so we need to clean the caches to see the changes.
clean_repos()
assert dvc_repo.status([stage.path]) == {}
dvc_repo.update(stage.path)
assert dvc_repo.status([stage.path]) == {}
assert os.path.exists(dst)
assert os.path.isfile(dst)
with open(dst, "r+") as fobj:
assert fobj.read() == "updated"
def test_update_import(dvc_repo, erepo):
src = "version"
dst = src
stage = dvc_repo.imp(erepo.root_dir, src, dst, rev="branch")
assert os.path.exists(dst)
assert os.path.isfile(dst)
with open(dst, "r+") as fobj:
assert fobj.read() == "branch"
# update data
repo = Repo(erepo.root_dir)
saved_dir = os.getcwd()
os.chdir(erepo.root_dir)
repo.scm.checkout("branch")
os.unlink("version")
erepo.create("version", "updated")
repo.add("version")
repo.scm.add([".gitignore", "version.dvc"])
repo.scm.commit("updated")
repo.scm.checkout("master")
repo.scm.git.close()
os.chdir(saved_dir)
assert dvc_repo.status(stage.path) == {}
dvc_repo.update(stage.path)
assert dvc_repo.status(stage.path) == {}
assert os.path.exists(dst)
assert os.path.isfile(dst)
with open(dst, "r+") as fobj:
assert fobj.read() == "updated"
class TestGCMultipleDvcRepos(TestDvcGit):
def _check_cache(self, num):
total = 0
for root, dirs, files in os.walk(os.path.join(".dvc", "cache")):
total += len(files)
self.assertEqual(total, num)
def setUp(self):
super().setUp()
self.additional_path = TestDir.mkdtemp()
self.additional_git = Repo.init(self.additional_path)
self.additional_dvc = DvcRepo.init(self.additional_path)
cache_path = os.path.join(self._root_dir, ".dvc", "cache")
config_path = os.path.join(self.additional_path, ".dvc",
"config.local")
cfg = configobj.ConfigObj()
cfg.filename = config_path
cfg["cache"] = {"dir": cache_path}
cfg.write()
self.additional_dvc = DvcRepo(self.additional_path)
def test(self):
# ADD FILE ONLY IN MAIN PROJECT
fname = "only_in_first"
with open(fname, "w+") as fobj:
fobj.write("only in main repo")
stages = self.dvc.add(fname)
self.assertEqual(len(stages), 1)
# ADD FILE IN MAIN PROJECT THAT IS ALSO IN SECOND PROJECT
fname = "in_both"
with open(fname, "w+") as fobj:
fobj.write("in both repos")
stages = self.dvc.add(fname)
self.assertEqual(len(stages), 1)
cwd = os.getcwd()
os.chdir(self.additional_path)
# ADD FILE ONLY IN SECOND PROJECT
fname = os.path.join(self.additional_path, "only_in_second")
with open(fname, "w+") as fobj:
fobj.write("only in additional repo")
stages = self.additional_dvc.add(fname)
self.assertEqual(len(stages), 1)
# ADD FILE IN SECOND PROJECT THAT IS ALSO IN MAIN PROJECT
fname = os.path.join(self.additional_path, "in_both")
with open(fname, "w+") as fobj:
fobj.write("in both repos")
stages = self.additional_dvc.add(fname)
self.assertEqual(len(stages), 1)
os.chdir(cwd)
self._check_cache(3)
self.dvc.gc(repos=[self.additional_path], workspace=True)
self._check_cache(3)
self.dvc.gc(workspace=True)
self._check_cache(2)
