def __init__(self,
archive_format=ZIP,
prefix=None,
description=None,
predicate=None,
client_id=None):
"""CollectionArchiveGenerator constructor.
Args:
archive_format: May be ArchiveCollectionGenerator.ZIP or
ArchiveCollectionGenerator.TAR_GZ. Defaults to ZIP.
prefix: Name of the folder inside the archive that will contain all the
generated data.
description: String describing archive's contents. It will be included
into the auto-generated MANIFEST file. Defaults to 'Files archive
collection'.
predicate: If not None, only the files matching the predicate will be
archived, all others will be skipped. The predicate receives a
db.ClientPath as input.
client_id: The client_id to use when exporting a flow results collection.
Raises:
ValueError: if prefix is None.
"""
super(Aff4CollectionArchiveGenerator, self).__init__()
if archive_format == self.ZIP:
self.archive_generator = utils.StreamingZipGenerator(
compression=zipfile.ZIP_DEFLATED)
elif archive_format == self.TAR_GZ:
self.archive_generator = utils.StreamingTarGenerator()
else:
raise ValueError("Unknown archive format: %s" % archive_format)
if not prefix:
raise ValueError("Prefix can't be None.")
self.prefix = prefix
self.description = description or "Files archive collection"
self.total_files = 0
self.archived_files = 0
self.ignored_files = []
self.failed_files = []
self.predicate = predicate or (lambda _: True)
self.client_id = client_id
def __init__(self, flow: rdf_flow_objects.Flow,
archive_format: ArchiveFormat):
self.flow = flow
self.archive_format = archive_format
if archive_format == ArchiveFormat.ZIP:
self.archive_generator = utils.StreamingZipGenerator(
compression=zipfile.ZIP_DEFLATED)
extension = "zip"
elif archive_format == ArchiveFormat.TAR_GZ:
self.archive_generator = utils.StreamingTarGenerator()
extension = "tar.gz"
else:
raise ValueError(f"Unknown archive format: {archive_format}")
self.prefix = "{}_{}_{}".format(
flow.client_id.replace(".", "_"), flow.flow_id, flow.flow_class_name)
self.filename = f"{self.prefix}.{extension}"
self.num_archived_files = 0
