def _get_all_paths(self):
paths = []
ignore_file_handler = DvcIgnoreFileHandler(self.dvc.tree)
for root, dirs, files in self.dvc.tree.walk(
self.dvc.root_dir, ignore_file_handler=ignore_file_handler):
for dname in dirs:
paths.append(os.path.join(root, dname))
for fname in files:
paths.append(os.path.join(root, fname))
return paths
def stages(self, from_directory=None, check_dag=True):
"""
Walks down the root directory looking for Dvcfiles,
skipping the directories that are related with
any SCM (e.g. `.git`), DVC itself (`.dvc`), or directories
tracked by DVC (e.g. `dvc add data` would skip `data/`)
NOTE: For large repos, this could be an expensive
operation. Consider using some memoization.
"""
from dvc.stage import Stage
if not from_directory:
from_directory = self.root_dir
elif not os.path.isdir(from_directory):
raise TargetNotDirectoryError(from_directory)
stages = []
outs = []
ignore_file_handler = DvcIgnoreFileHandler(self.tree)
for root, dirs, files in self.tree.walk(
from_directory, ignore_file_handler=ignore_file_handler):
for fname in files:
path = os.path.join(root, fname)
if not Stage.is_valid_filename(path):
continue
stage = Stage.load(self, path)
for out in stage.outs:
outs.append(out.path + out.sep)
stages.append(stage)
def filter_dirs(dname, root=root):
if dname in ('image', 'image-aug'):
return False
path = os.path.join(root, dname)
if path in (self.dvc_dir, self.scm.dir):
return False
for out in outs:
if path == os.path.normpath(out) or path.startswith(out):
return False
return True
dirs[:] = list(filter(filter_dirs, dirs))
if check_dag:
self.check_dag(stages)
return stages
def setUp(self):
super(TestDvcIgnore, self).setUp()
self.ignore_file_handler = DvcIgnoreFileHandler(self.dvc.tree)