def test_path_object_and_str_are_valid_types_get_mtime_and_size(
path, repo_dir):
dvcignore = DvcIgnoreFilter(repo_dir.root_dir)
time, size = get_mtime_and_size(path, dvcignore)
object_time, object_size = get_mtime_and_size(PathInfo(path), dvcignore)
assert time == object_time
assert size == object_size
def test_nobranch(self):
tree = WorkingTree(self._root_dir)
dvcignore = DvcIgnoreFilter(self._root_dir)
self.assertWalkEqual(
tree.walk(".", dvcignore=dvcignore),
[
(
self._root_dir,
["data_dir"],
["bar", "тест", "code.py", "foo"],
),
(join(self._root_dir, "data_dir"), ["data_sub_dir"], ["data"]),
(
join(self._root_dir, "data_dir", "data_sub_dir"),
[],
["data_sub"],
),
],
)
self.assertWalkEqual(
tree.walk(join("data_dir", "data_sub_dir"), dvcignore=dvcignore),
[(
join(self._root_dir, "data_dir", "data_sub_dir"),
[],
["data_sub"],
)],
)
def dvc_walk(
top,
topdown=True,
onerror=None,
followlinks=False,
ignore_file_handler=None,
):
"""
Proxy for `os.walk` directory tree generator.
Utilizes DvcIgnoreFilter functionality.
"""
ignore_filter = None
if topdown:
from dvc.ignore import DvcIgnoreFilter
ignore_filter = DvcIgnoreFilter(
top, ignore_file_handler=ignore_file_handler)
for root, dirs, files in os.walk(top,
topdown=topdown,
onerror=onerror,
followlinks=followlinks):
if ignore_filter:
dirs[:], files[:] = ignore_filter(root, dirs, files)
yield root, dirs, files
def _walk(
self,
tree,
topdown=True,
ignore_file_handler=None,
dvc_ignore_filter=None,
):
dirs, nondirs = [], []
for i in tree:
if i.mode == GIT_MODE_DIR:
dirs.append(i.name)
else:
nondirs.append(i.name)
if topdown:
if not dvc_ignore_filter:
dvc_ignore_filter = DvcIgnoreFilter(
tree.abspath, ignore_file_handler=ignore_file_handler)
dirs, nondirs = dvc_ignore_filter(tree.path, dirs, nondirs)
yield os.path.normpath(tree.abspath), dirs, nondirs
for i in dirs:
for x in self._walk(
tree[i],
topdown=True,
ignore_file_handler=ignore_file_handler,
dvc_ignore_filter=dvc_ignore_filter,
):
yield x
if not topdown:
yield os.path.normpath(tree.abspath), dirs, nondirs
def dvcignore(self):
from dvc.ignore import DvcIgnoreFilter, DvcIgnoreFilterNoop
root = self.dvcignore_root or self.tree_root
if not self.use_dvcignore:
return DvcIgnoreFilterNoop(self, root)
self.use_dvcignore = False
ret = DvcIgnoreFilter(self, root)
self.use_dvcignore = True
return ret
def test_subdir(self):
dvcignore = DvcIgnoreFilter(self.root_dir)
tree = WorkingTree(self._root_dir)
self.assertWalkEqual(
tree.walk(join("data_dir", "data_sub_dir"), dvcignore=dvcignore),
[(
join(self._root_dir, "data_dir", "data_sub_dir"),
[],
["data_sub"],
)],
)
def test(self):
dvcignore = DvcIgnoreFilter(self.root_dir)
file_time, file_size = get_mtime_and_size(self.DATA, dvcignore)
dir_time, dir_size = get_mtime_and_size(self.DATA_DIR, dvcignore)
actual_file_size = os.path.getsize(self.DATA)
actual_dir_size = os.path.getsize(self.DATA) + os.path.getsize(
self.DATA_SUB)
self.assertIs(type(file_time), str)
self.assertIs(type(file_size), str)
self.assertEqual(file_size, str(actual_file_size))
self.assertIs(type(dir_time), str)
self.assertIs(type(dir_size), str)
self.assertEqual(dir_size, str(actual_dir_size))
def test_ignore_collecting_dvcignores(repo_dir, dname):
top_ignore_file = os.path.join(
repo_dir.root_dir, os.path.dirname(dname), DvcIgnore.DVCIGNORE_FILE
)
repo_dir.create(top_ignore_file, os.path.basename(dname))
ignore_file = os.path.join(
repo_dir.root_dir, dname, DvcIgnore.DVCIGNORE_FILE
)
repo_dir.create(ignore_file, repo_dir.FOO)
assert DvcIgnoreFilter(repo_dir.root_dir).ignores == {
DvcIgnoreDirs([".git", ".hg", ".dvc"]),
DvcIgnorePatterns(top_ignore_file),
}
def test(self):
dvcignore = DvcIgnoreFilter(self.root_dir)
tree = WorkingTree(self._root_dir)
self.assertWalkEqual(
tree.walk(self._root_dir, dvcignore=dvcignore),
[
(
self._root_dir,
["data_dir"],
["code.py", "bar", "тест", "foo"],
),
(join(self._root_dir, "data_dir"), ["data_sub_dir"], ["data"]),
(
join(self._root_dir, "data_dir", "data_sub_dir"),
[],
["data_sub"],
),
],
)
def test_ignore_on_branch(git, dvc_repo, repo_dir):
dvc_repo.add(repo_dir.DATA_DIR)
dvc_repo.scm.commit("add data dir")
branch_name = "branch_one"
dvc_repo.scm.checkout(branch_name, create_new=True)
repo_dir.create(DvcIgnore.DVCIGNORE_FILE, to_posixpath(repo_dir.DATA_SUB))
dvc_repo.scm.add([DvcIgnore.DVCIGNORE_FILE])
git.index.commit("add ignore")
dvc_repo.scm.checkout("master")
git_tree = dvc_repo.scm.get_tree(branch_name)
branch_data_files = set(
itertools.chain.from_iterable([
files for _, _, files in dvc_repo.tree.walk(
repo_dir.DATA_DIR,
dvcignore=DvcIgnoreFilter(repo_dir.root_dir, git_tree),
)
]))
assert branch_data_files == {"data"}
def dvcignore(self):
return DvcIgnoreFilter(self.root_dir)
def dvcignore(self) -> DvcIgnoreFilter:
return DvcIgnoreFilter(self.fs, self.root_dir)
