def _collect_used_dir_cache(self,
remote=None,
force=False,
jobs=None,
filter_info=None) -> Optional["Tree"]:
"""Fetch dir cache and return used object IDs for this out."""
try:
self.get_dir_cache(jobs=jobs, remote=remote)
except DvcException:
logger.debug(f"failed to pull cache for '{self}'")
try:
objects.check(self.odb, self.odb.get(self.hash_info))
except FileNotFoundError:
msg = ("Missing cache for directory '{}'. "
"Cache for files inside will be lost. "
"Would you like to continue? Use '-f' to force.")
if not force and not prompt.confirm(msg.format(self.path_info)):
raise CollectCacheError(
"unable to fully collect used cache"
" without cache for directory '{}'".format(self))
return None
obj = self.get_obj()
if filter_info and filter_info != self.path_info:
prefix = filter_info.relative_to(self.path_info).parts
obj = obj.filter(prefix)
return obj
def collect_used_dir_cache(
self, remote=None, force=False, jobs=None, filter_info=None
):
"""Get a list of `info`s related to the given directory.
- Pull the directory entry from the remote cache if it was changed.
Example:
Given the following commands:
$ echo "foo" > directory/foo
$ echo "bar" > directory/bar
$ dvc add directory
It will return a NamedCache like:
nc = NamedCache()
nc.add(self.scheme, 'c157a79031e1', 'directory/foo')
nc.add(self.scheme, 'd3b07384d113', 'directory/bar')
"""
cache = NamedCache()
try:
self.get_dir_cache(jobs=jobs, remote=remote)
except DvcException:
logger.debug(f"failed to pull cache for '{self}'")
if self.cache.changed_cache_file(self.hash_info):
msg = (
"Missing cache for directory '{}'. "
"Cache for files inside will be lost. "
"Would you like to continue? Use '-f' to force."
)
if not force and not prompt.confirm(msg.format(self.path_info)):
raise CollectCacheError(
"unable to fully collect used cache"
" without cache for directory '{}'".format(self)
)
else:
return cache
path = str(self.path_info)
filter_path = str(filter_info) if filter_info else None
is_win = os.name == "nt"
for entry in self.dir_cache:
checksum = entry[self.tree.PARAM_CHECKSUM]
entry_relpath = entry[self.tree.PARAM_RELPATH]
if is_win:
entry_relpath = entry_relpath.replace("/", os.sep)
entry_path = os.path.join(path, entry_relpath)
if (
not filter_path
or entry_path == filter_path
or entry_path.startswith(filter_path + os.sep)
):
cache.add(self.scheme, checksum, entry_path)
return cache
def collect_used_dir_cache(
self,
remote=None,
force=False,
jobs=None,
filter_info=None) -> Dict[Optional["ObjectDB"], Set["HashFile"]]:
"""Fetch dir cache and return used objects for this out."""
try:
self.get_dir_cache(jobs=jobs, remote=remote)
except DvcException:
logger.debug(f"failed to pull cache for '{self}'")
try:
objects.check(self.odb, self.odb.get(self.hash_info))
except (FileNotFoundError, ObjectFormatError):
msg = ("Missing cache for directory '{}'. "
"Cache for files inside will be lost. "
"Would you like to continue? Use '-f' to force.")
if not force and not prompt.confirm(msg.format(self.path_info)):
raise CollectCacheError(
"unable to fully collect used cache"
" without cache for directory '{}'".format(self))
return {}
obj = self.get_obj(filter_info=filter_info, copy=True)
self._set_obj_names(obj)
return {None: {obj}}
def _collect_used_dir_cache(
self, remote=None, force=False, jobs=None, filter_info=None
):
"""Get a list of `info`s related to the given directory.
- Pull the directory entry from the remote cache if it was changed.
Example:
Given the following commands:
$ echo "foo" > directory/foo
$ echo "bar" > directory/bar
$ dvc add directory
It will return a NamedCache like:
nc = NamedCache()
nc.add(self.scheme, 'c157a79031e1', 'directory/foo')
nc.add(self.scheme, 'd3b07384d113', 'directory/bar')
"""
cache = NamedCache()
if self.cache.changed_cache_file(self.checksum):
try:
self.repo.cloud.pull(
NamedCache.make("local", self.checksum, str(self)),
jobs=jobs,
remote=remote,
show_checksums=False,
)
except DvcException:
logger.debug("failed to pull cache for '{}'".format(self))
if self.cache.changed_cache_file(self.checksum):
msg = (
"Missing cache for directory '{}'. "
"Cache for files inside will be lost. "
"Would you like to continue? Use '-f' to force."
)
if not force and not prompt.confirm(msg.format(self.path_info)):
raise CollectCacheError(
"unable to fully collect used cache"
" without cache for directory '{}'".format(self)
)
else:
return cache
for entry in self.dir_cache:
checksum = entry[self.remote.PARAM_CHECKSUM]
info = self.path_info / entry[self.remote.PARAM_RELPATH]
if not filter_info or info.isin_or_eq(filter_info):
cache.add(self.scheme, checksum, str(info))
return cache
def collect_used_dir_cache(
self, remote=None, force=False, jobs=None, filter_info=None
):
"""Get a list of `info`s related to the given directory.
- Pull the directory entry from the remote cache if it was changed.
Example:
Given the following commands:
$ echo "foo" > directory/foo
$ echo "bar" > directory/bar
$ dvc add directory
It will return a NamedCache like:
nc = NamedCache()
nc.add(self.scheme, 'c157a79031e1', 'directory/foo')
nc.add(self.scheme, 'd3b07384d113', 'directory/bar')
"""
cache = NamedCache()
try:
self.get_dir_cache(jobs=jobs, remote=remote)
except DvcException:
logger.debug(f"failed to pull cache for '{self}'")
try:
objects.check(self.odb, self.odb.get(self.hash_info))
except (FileNotFoundError, ObjectFormatError):
msg = (
"Missing cache for directory '{}'. "
"Cache for files inside will be lost. "
"Would you like to continue? Use '-f' to force."
)
if not force and not prompt.confirm(msg.format(self.path_info)):
raise CollectCacheError(
"unable to fully collect used cache"
" without cache for directory '{}'".format(self)
)
return cache
path = str(self.path_info)
filter_path = str(filter_info) if filter_info else None
for entry_key, entry_obj in self.obj:
entry_path = os.path.join(path, *entry_key)
if (
not filter_path
or entry_path == filter_path
or entry_path.startswith(filter_path + os.sep)
):
cache.add(self.scheme, entry_obj.hash_info.value, entry_path)
return cache