def _collect_paths(
repo: "Repo",
targets: Iterable[str],
recursive: bool = False,
rev: str = None,
):
from dvc.tree.repo import RepoTree
path_infos = [PathInfo(os.path.abspath(target)) for target in targets]
tree = RepoTree(repo)
target_infos = []
for path_info in path_infos:
if recursive and tree.isdir(path_info):
target_infos.extend(tree.walk_files(path_info))
if not tree.exists(path_info):
if not recursive:
if rev == "workspace" or rev == "":
logger.warning(
"'%s' was not found in current workspace.", path_info,
)
else:
logger.warning(
"'%s' was not found at: '%s'.", path_info, rev,
)
continue
target_infos.append(path_info)
return target_infos
def _collect_paths(
repo: Repo,
targets: Iterable[str],
recursive: bool = False,
rev: str = None,
):
path_infos = {PathInfo(os.path.abspath(target)) for target in targets}
tree = RepoTree(repo)
target_infos = set()
for path_info in path_infos:
if recursive and tree.isdir(path_info):
target_infos.update(set(tree.walk_files(path_info)))
if not tree.exists(path_info):
if not recursive:
logger.warning(
"'%s' was not found at: '%s'.",
path_info,
rev,
)
continue
target_infos.add(path_info)
return target_infos
def _collect_metrics(repo, targets, recursive):
if targets:
target_infos = [
PathInfo(os.path.abspath(target)) for target in targets
]
tree = RepoTree(repo)
rec_files = []
if recursive:
for target_info in target_infos:
if tree.isdir(target_info):
rec_files.extend(list(tree.walk_files(target_info)))
result = [t for t in target_infos if tree.isfile(t)]
result.extend(rec_files)
return result
metrics = set()
for stage in repo.stages:
for out in stage.outs:
if not out.metric:
continue
metrics.add(out.path_info)
return list(metrics)
def test_isdir_mixed(tmp_dir, dvc):
tmp_dir.gen({"dir": {"foo": "foo", "bar": "bar"}})
dvc.add(str(tmp_dir / "dir" / "foo"))
tree = RepoTree(dvc)
assert tree.isdir("dir")
assert not tree.isfile("dir")
def test_repo_tree_no_subrepos(tmp_dir, dvc, scm):
tmp_dir.scm_gen(
{"dir": {
"repo.txt": "file to confuse RepoTree"
}},
commit="dir/repo.txt",
)
tmp_dir.dvc_gen({"lorem": "lorem"}, commit="add foo")
subrepo = tmp_dir / "dir" / "repo"
make_subrepo(subrepo, scm)
subrepo.dvc_gen({"foo": "foo", "dir1": {"bar": "bar"}}, commit="FOO")
subrepo.scm_gen({"ipsum": "ipsum"}, commit="BAR")
# using tree that does not have dvcignore
dvc.tree._reset()
tree = RepoTree(dvc, subrepos=False, fetch=True)
expected = [
tmp_dir / ".dvcignore",
tmp_dir / ".gitignore",
tmp_dir / "lorem",
tmp_dir / "lorem.dvc",
tmp_dir / "dir",
tmp_dir / "dir" / "repo.txt",
]
actual = []
for root, dirs, files in tree.walk(tmp_dir, dvcfiles=True):
for entry in dirs + files:
actual.append(os.path.normpath(os.path.join(root, entry)))
expected = [str(path) for path in expected]
assert set(actual) == set(expected)
assert len(actual) == len(expected)
assert tree.isfile(tmp_dir / "lorem") is True
assert tree.isfile(tmp_dir / "dir" / "repo" / "foo") is False
assert tree.isdir(tmp_dir / "dir" / "repo") is False
assert tree.isdir(tmp_dir / "dir") is True
assert tree.isdvc(tmp_dir / "lorem") is True
assert tree.isdvc(tmp_dir / "dir" / "repo" / "dir1") is False
assert tree.exists(tmp_dir / "dir" / "repo.txt") is True
assert tree.exists(tmp_dir / "repo" / "ipsum") is False
def test_exists_isdir_isfile_dirty(tmp_dir, dvc):
tmp_dir.dvc_gen(
{"datafile": "data", "datadir": {"foo": "foo", "bar": "bar"}}
)
tree = RepoTree(dvc)
shutil.rmtree(tmp_dir / "datadir")
(tmp_dir / "datafile").unlink()
root = PathInfo(tmp_dir)
assert tree.exists(root / "datafile")
assert tree.exists(root / "datadir")
assert tree.exists(root / "datadir" / "foo")
assert tree.isfile(root / "datafile")
assert not tree.isfile(root / "datadir")
assert tree.isfile(root / "datadir" / "foo")
assert not tree.isdir(root / "datafile")
assert tree.isdir(root / "datadir")
assert not tree.isdir(root / "datadir" / "foo")
# NOTE: creating file instead of dir and dir instead of file
tmp_dir.gen({"datadir": "data", "datafile": {"foo": "foo", "bar": "bar"}})
assert tree.exists(root / "datafile")
assert tree.exists(root / "datadir")
assert not tree.exists(root / "datadir" / "foo")
assert tree.exists(root / "datafile" / "foo")
assert not tree.isfile(root / "datafile")
assert tree.isfile(root / "datadir")
assert not tree.isfile(root / "datadir" / "foo")
assert tree.isfile(root / "datafile" / "foo")
assert tree.isdir(root / "datafile")
assert not tree.isdir(root / "datadir")
assert not tree.isdir(root / "datadir" / "foo")
assert not tree.isdir(root / "datafile" / "foo")
def collect(
self,
targets: List[str] = None,
revs: List[str] = None,
recursive: bool = False,
) -> Dict[str, Dict]:
"""Collects all props and data for plots.
Returns a structure like:
{rev: {plots.csv: {
props: {x: ..., "header": ..., ...},
data: "...data as a string...",
}}}
Data parsing is postponed, since it's affected by props.
"""
from dvc.tree.repo import RepoTree
targets = [targets] if isinstance(targets, str) else targets or []
data = {}
for rev in self.repo.brancher(revs=revs):
# .brancher() adds unwanted workspace
if revs is not None and rev not in revs:
continue
rev = rev or "workspace"
tree = RepoTree(self.repo)
plots = _collect_plots(self.repo, targets, rev, recursive)
for path_info, props in plots.items():
if rev not in data:
data[rev] = {}
if tree.isdir(path_info):
plot_files = []
for pi in tree.walk_files(path_info):
plot_files.append(
(pi, relpath(pi, self.repo.root_dir))
)
else:
plot_files = [
(path_info, relpath(path_info, self.repo.root_dir))
]
for path, repo_path in plot_files:
data[rev].update({repo_path: {"props": props}})
# Load data from git or dvc cache
try:
with tree.open(path) as fd:
data[rev][repo_path]["data"] = fd.read()
except FileNotFoundError:
# This might happen simply because cache is absent
pass
return data
def test_isdir_isfile(tmp_dir, dvc):
tmp_dir.gen({"datafile": "data", "datadir": {"foo": "foo", "bar": "bar"}})
tree = RepoTree(dvc)
assert tree.isdir("datadir")
assert not tree.isfile("datadir")
assert not tree.isdvc("datadir")
assert not tree.isdir("datafile")
assert tree.isfile("datafile")
assert not tree.isdvc("datafile")
dvc.add(["datadir", "datafile"])
shutil.rmtree(tmp_dir / "datadir")
(tmp_dir / "datafile").unlink()
assert tree.isdir("datadir")
assert not tree.isfile("datadir")
assert tree.isdvc("datadir")
assert not tree.isdir("datafile")
assert tree.isfile("datafile")
assert tree.isdvc("datafile")
def _get_checksum(self, locked=True):
from dvc.tree.repo import RepoTree
with self._make_repo(locked=locked) as repo:
try:
return repo.find_out_by_relpath(self.def_path).info["md5"]
except OutputNotFoundError:
path = PathInfo(os.path.join(repo.root_dir, self.def_path))
# we want stream but not fetch, so DVC out directories are
# walked, but dir contents is not fetched
tree = RepoTree(repo, stream=True)
# We are polluting our repo cache with some dir listing here
if tree.isdir(path):
return self.repo.cache.local.tree.get_hash(path,
tree=tree)[1]
return tree.get_file_hash(path)
def test_subrepos(tmp_dir, scm, dvc):
tmp_dir.scm_gen(
{"dir": {
"repo.txt": "file to confuse RepoTree"
}},
commit="dir/repo.txt",
)
subrepo1 = tmp_dir / "dir" / "repo"
subrepo2 = tmp_dir / "dir" / "repo2"
for repo in [subrepo1, subrepo2]:
make_subrepo(repo, scm)
subrepo1.dvc_gen({"foo": "foo", "dir1": {"bar": "bar"}}, commit="FOO")
subrepo2.dvc_gen({
"lorem": "lorem",
"dir2": {
"ipsum": "ipsum"
}
},
commit="BAR")
dvc.tree._reset()
tree = RepoTree(dvc, subrepos=True, fetch=True)
def assert_tree_belongs_to_repo(ret_val):
method = tree._get_repo
def f(*args, **kwargs):
r = method(*args, **kwargs)
assert r.root_dir == ret_val.root_dir
return r
return f
with mock.patch.object(
tree,
"_get_repo",
side_effect=assert_tree_belongs_to_repo(subrepo1.dvc),
):
assert tree.exists(subrepo1 / "foo") is True
assert tree.exists(subrepo1 / "bar") is False
assert tree.isfile(subrepo1 / "foo") is True
assert tree.isfile(subrepo1 / "dir1" / "bar") is True
assert tree.isfile(subrepo1 / "dir1") is False
assert tree.isdir(subrepo1 / "dir1") is True
assert tree.isdir(subrepo1 / "dir1" / "bar") is False
assert tree.isdvc(subrepo1 / "foo") is True
with mock.patch.object(
tree,
"_get_repo",
side_effect=assert_tree_belongs_to_repo(subrepo2.dvc),
):
assert tree.exists(subrepo2 / "lorem") is True
assert tree.exists(subrepo2 / "ipsum") is False
assert tree.isfile(subrepo2 / "lorem") is True
assert tree.isfile(subrepo2 / "dir2" / "ipsum") is True
assert tree.isfile(subrepo2 / "dir2") is False
assert tree.isdir(subrepo2 / "dir2") is True
assert tree.isdir(subrepo2 / "dir2" / "ipsum") is False
assert tree.isdvc(subrepo2 / "lorem") is True
