def get_all_staging_paths(
self,
invalid_chars: Sequence[str] = ("/", ),
include_collection_name: bool = False,
) -> List[str]:
safe_element_identifiers = []
for element_identifiers, extension, *_ in self.element_identifiers_extensions_paths_and_metadata_files:
datatype = self.datatypes_registry.get_datatype_by_extension(
extension)
if datatype:
extension = getattr(datatype, 'file_ext_export_alias',
extension)
current_element_identifiers = []
for element_identifier in element_identifiers:
max_len = 254 - len(extension)
if include_collection_name:
max_len = max_len - (len(self.name or "") + 1)
assert max_len >= 1, 'Could not stage element, element identifier is too long'
current_element_identifier = filesystem_safe_string(
element_identifier,
max_len=max_len,
invalid_chars=invalid_chars)
if include_collection_name and self.name:
current_element_identifier = f"{filesystem_safe_string(self.name, invalid_chars=invalid_chars)}{os.path.sep}{current_element_identifier}"
current_element_identifiers.append(current_element_identifier)
safe_element_identifiers.append(
f'{os.path.sep.join(current_element_identifiers)}.{extension}')
return safe_element_identifiers
def staging_path(self):
"""
Strip leading dots, unicode null chars, replace `/` with `_`, truncate at 255 characters.
Not safe for commandline use, would need additional sanitization.
"""
max_len = 254 - len(self.file_ext)
safe_element_identifier = filesystem_safe_string(self.element_identifier, max_len=max_len)
return f"{safe_element_identifier}.{self.file_ext}"
def all_staging_paths(self):
safe_element_identifiers = []
for element_identifiers, extension, *_ in self.element_identifiers_extensions_paths_and_metadata_files:
datatype = self.datatypes_registry.get_datatype_by_extension(extension)
if datatype:
extension = getattr(datatype, 'file_ext_export_alias', extension)
current_element_identifiers = []
for element_identifier in element_identifiers:
current_element_identifiers.append(filesystem_safe_string(element_identifier, max_len=254 - len(extension)))
safe_element_identifiers.append(f'{os.path.sep.join(current_element_identifiers)}.{extension}')
return safe_element_identifiers