def __init__(
self,
stage,
path,
info=None,
remote=None,
cache=True,
metric=False,
persist=False,
):
self._validate_output_path(path)
# This output (and dependency) objects have too many paths/urls
# here is a list and comments:
#
# .def_path - path from definition in stage file
# .path_info - PathInfo/URLInfo structured resolved path
# .fspath - local only, resolved
# .__str__ - for presentation purposes, def_path/relpath
#
# By resolved path, which contains actual location,
# should be absolute and don't contain remote:// refs.
self.stage = stage
self.repo = stage.repo if stage else None
self.def_path = path
self.info = info
self.remote = remote or self.REMOTE(self.repo, {})
self.use_cache = False if self.IS_DEPENDENCY else cache
self.metric = False if self.IS_DEPENDENCY else metric
self.persist = persist
self.path_info = self._parse_path(remote, path)
if self.use_cache and self.cache is None:
raise RemoteCacheRequiredError(self.path_info)
def __init__(
self,
stage,
path,
info=None,
cache=True,
metric=False,
plot=False,
persist=False,
checkpoint=False,
live=False,
desc=None,
isexec=False,
remote=None,
):
self.repo = stage.repo if stage else None
fs_cls, fs_config, path_info = get_cloud_fs(self.repo, url=path)
self.fs = fs_cls(**fs_config)
if (
self.fs.scheme == "local"
and stage
and path_isin(path, stage.repo.root_dir)
):
self.def_path = relpath(path, stage.wdir)
else:
self.def_path = path
self._validate_output_path(path, stage)
# This output (and dependency) objects have too many paths/urls
# here is a list and comments:
#
# .def_path - path from definition in DVC file
# .path_info - PathInfo/URLInfo structured resolved path
# .fspath - local only, resolved
# .__str__ - for presentation purposes, def_path/relpath
#
# By resolved path, which contains actual location,
# should be absolute and don't contain remote:// refs.
self.stage = stage
self.hash_info = HashInfo.from_dict(info)
self.use_cache = False if self.IS_DEPENDENCY else cache
self.metric = False if self.IS_DEPENDENCY else metric
self.plot = False if self.IS_DEPENDENCY else plot
self.persist = persist
self.checkpoint = checkpoint
self.live = live
self.desc = desc
self.path_info = self._parse_path(self.fs, path_info)
if self.use_cache and self.odb is None:
raise RemoteCacheRequiredError(self.path_info)
self.obj = None
self.isexec = False if self.IS_DEPENDENCY else isexec
self.remote = remote
def get_dir_checksum(self, path_info, tree=None):
if not self.cache:
raise RemoteCacheRequiredError(path_info)
dir_info = self._collect_dir(path_info, tree=None)
if tree:
# don't save state entry for path_info if it is a tree path
path_info = None
return self._save_dir_info(dir_info, path_info)
def __init__(
self,
stage,
path,
info=None,
fs=None,
cache=True,
metric=False,
plot=False,
persist=False,
checkpoint=False,
live=False,
desc=None,
isexec=False,
):
self._validate_output_path(path, stage)
# This output (and dependency) objects have too many paths/urls
# here is a list and comments:
#
# .def_path - path from definition in DVC file
# .path_info - PathInfo/URLInfo structured resolved path
# .fspath - local only, resolved
# .__str__ - for presentation purposes, def_path/relpath
#
# By resolved path, which contains actual location,
# should be absolute and don't contain remote:// refs.
self.stage = stage
self.repo = stage.repo if stage else None
self.def_path = path
self.hash_info = HashInfo.from_dict(info)
if fs:
self.fs = fs
else:
self.fs = self.FS_CLS(self.repo, {})
self.use_cache = False if self.IS_DEPENDENCY else cache
self.metric = False if self.IS_DEPENDENCY else metric
self.plot = False if self.IS_DEPENDENCY else plot
self.persist = persist
self.checkpoint = checkpoint
self.live = live
self.desc = desc
self.path_info = self._parse_path(fs, path)
if self.use_cache and self.odb is None:
raise RemoteCacheRequiredError(self.path_info)
self.obj = None
self.isexec = False if self.IS_DEPENDENCY else isexec
try:
self.git_tracked_files = list(
self.repo.scm.backends.backends["dulwich"]().repo.open_index())
self.git_tracked_files = [
os.path.abspath(each.decode("utf-8"))
for each in self.git_tracked_files
]
except (SCMError, AttributeError, NoIndexPresent):
self.git_tracked_files = []
def get_dir_checksum(self, path_info):
if not self.cache:
raise RemoteCacheRequiredError(path_info)
dir_info = self._collect_dir(path_info)
checksum, tmp_info = self._get_dir_info_checksum(dir_info)
new_info = self.cache.checksum_to_path_info(checksum)
if self.cache.changed_cache_file(checksum):
self.cache.makedirs(new_info.parent)
self.cache.move(tmp_info, new_info, mode=self.CACHE_MODE)
self.state.save(path_info, checksum)
self.state.save(new_info, checksum)
return checksum
def __init__(
self,
stage,
path,
info=None,
tree=None,
cache=True,
metric=False,
plot=False,
persist=False,
checkpoint=False,
live=False,
desc=None,
isexec=False,
):
self._validate_output_path(path, stage)
# This output (and dependency) objects have too many paths/urls
# here is a list and comments:
#
# .def_path - path from definition in stage file
# .path_info - PathInfo/URLInfo structured resolved path
# .fspath - local only, resolved
# .__str__ - for presentation purposes, def_path/relpath
#
# By resolved path, which contains actual location,
# should be absolute and don't contain remote:// refs.
self.stage = stage
self.repo = stage.repo if stage else None
self.def_path = path
self.hash_info = HashInfo.from_dict(info)
if tree:
self.tree = tree
else:
self.tree = self.TREE_CLS(self.repo, {})
self.use_cache = False if self.IS_DEPENDENCY else cache
self.metric = False if self.IS_DEPENDENCY else metric
self.plot = False if self.IS_DEPENDENCY else plot
self.persist = persist
self.checkpoint = checkpoint
self.live = live
self.desc = desc
self.path_info = self._parse_path(tree, path)
if self.use_cache and self.cache is None:
raise RemoteCacheRequiredError(self.path_info)
self.isexec = False if self.IS_DEPENDENCY else isexec
def get_dir_hash(self, path_info, tree, **kwargs):
if not self.cache:
raise RemoteCacheRequiredError(path_info)
dir_info = self._collect_dir(path_info, tree, **kwargs)
return self._save_dir_info(dir_info, path_info)
def get_dir_checksum(self, path_info):
if not self.cache:
raise RemoteCacheRequiredError(path_info)
dir_info = self._collect_dir(path_info)
return self._save_dir_info(dir_info, path_info)
def odb(self):
odb = getattr(self.repo.odb, self.scheme)
if self.use_cache and odb is None:
raise RemoteCacheRequiredError(self.fs.scheme, self.fs_path)
return odb