def run(self, package_path=None, xml_path=None):
package_path, xml_path = self.parse_params(package_path, xml_path)
ip = self.get_information_package()
package_path = package_path if package_path is not None else ip.object_path
xml_path = xml_path if xml_path is not None else os.path.splitext(
package_path)[0] + '.xml'
generate_package_mets(ip, package_path, xml_path)
return xml_path
def GeneratePackageMets(self, package_path=None, xml_path=None):
package_path, xml_path = self.parse_params(package_path, xml_path)
ip = self.get_information_package()
package_path = package_path if package_path is not None else ip.object_path
xml_path = xml_path if xml_path is not None else os.path.splitext(
package_path)[0] + '.xml'
generate_package_mets(ip, package_path, xml_path)
msg = 'Generated {xml}'.format(xml=xml_path)
self.create_success_event(msg)
return xml_path
def StorageMigration(self, storage_method, temp_path):
ip = self.get_information_package()
container_format = ip.get_container_format()
storage_method = StorageMethod.objects.get(pk=storage_method)
try:
storage_target = storage_method.enabled_target
except StorageTarget.DoesNotExist:
raise ValueError('No writeable target available for {}'.format(storage_method))
dir_path = os.path.join(temp_path, ip.object_identifier_value)
container_path = os.path.join(temp_path, ip.object_identifier_value + '.{}'.format(container_format))
aip_xml_path = os.path.join(temp_path, ip.object_identifier_value + '.xml')
aic_xml_path = os.path.join(temp_path, ip.aic.object_identifier_value + '.xml')
if storage_target.master_server and not storage_target.remote_server:
# we are on remote host
src_container = True
else:
# we are not on master, access from existing storage object
storage_object = ip.get_fastest_readable_storage_object()
if storage_object.container:
storage_object.read(container_path, self.get_processtask())
else:
storage_object.read(dir_path, self.get_processtask())
src_container = storage_object.container
dst_container = storage_method.containers
# If storage_object is "long term" and storage_method is not (or vice versa),
# then we have to do some "conversion" before we go any further
if src_container and not dst_container:
# extract container
if container_format == 'tar':
with tarfile.open(container_path) as tar:
tar.extractall(temp_path)
elif container_format == 'zip':
with zipfile.ZipFile(container_path) as zipf:
zipf.extractall(temp_path)
else:
raise ValueError('Invalid container format: {}'.format(container_format))
elif not src_container and dst_container:
# create container, aip xml and aic xml
if container_format == 'tar':
with tarfile.open(container_path, 'w') as new_tar:
new_tar.format = settings.TARFILE_FORMAT
new_tar.add(dir_path)
elif container_format == 'zip':
zip_directory(dirname=dir_path, zipname=container_path, compress=False)
else:
raise ValueError('Invalid container format: {}'.format(container_format))
generate_package_mets(ip, container_path, aip_xml_path)
generate_aic_mets(ip, aic_xml_path)
if dst_container or storage_target.remote_server:
src = [
container_path,
aip_xml_path,
aic_xml_path,
]
else:
src = [dir_path]
if storage_target.remote_server:
# we are on master, copy files to remote
host, user, passw = storage_target.remote_server.split(',')
dst = urljoin(host, reverse('informationpackage-add-file-from-master'))
requests_session = requests.Session()
requests_session.verify = settings.REQUESTS_VERIFY
requests_session.auth = (user, passw)
for s in src:
copy_file(s, dst, requests_session=requests_session)
obj_id = ip.preserve(src, storage_target, dst_container, self.get_processtask())
Notification.objects.create(
message="Migrated {} to {}".format(ip.object_identifier_value, storage_method.name),
level=logging.INFO,
user_id=self.responsible,
refresh=True,
)
return obj_id
def run(self):
generate_package_mets(self.get_information_package())