def _get_by_pid(self, pid, sysmeta=None, mn_client=None, cn_client=None):
""" Return DataObject
"""
if pid is None:
raise Exception("Missing pid")
fname = utils.get_object_by_pid(pid, resolve=True, mn_client=mn_client, cn_client=cn_client)
if fname:
meta = sysmeta
if not meta:
meta = utils.get_sysmeta_by_pid(pid, True)
url = utils.create_get_url_for_pid(mn_client.base_url, pid)
return DataObject(pid, False, fname, url, meta, meta.formatId, None)
else:
return None
def load(self):
""" Get the object referred to by pid and make sure it is a
package.
"""
if self.pid is None:
raise Exception("Missing pid")
sysmeta = utils.get_sysmeta_by_pid(self.pid)
if not sysmeta:
raise Exception('Couldn\'t find "%s" in DataONE.' % self.pid)
if sysmeta.formatId != RDFXML_FORMATID:
raise Exception('Package must be in RDF/XML format (not "%s").' % sysmeta.formatId)
rdf_xml_file = utils.get_object_by_pid(self.pid)
if not self._parse_rdf_xml(rdf_xml_file):
raise Exception('Unable to load package "%s".' % self.pid)
self.original_pid = self.pid
self.sysmeta = sysmeta
self.scimeta = self._download_object(self.scimeta)
loaded_scidata = {}
for pid, scidata in self.scidata_dict.iteritems():
loaded_scidata[pid] = self._download_object(scidata)
self.scidata_dict = loaded_scidata
return self
