def _clear_tables(self):
conn = g.engine.connect()
conn.execute('commit')
for table in Node().get_subclass_table_names():
if table != Node.__tablename__:
conn.execute('delete from {}'.format(table))
for table in Edge.get_subclass_table_names():
if table != Edge.__tablename__:
conn.execute('delete from {}'.format(table))
conn.execute('delete from _voided_nodes')
conn.execute('delete from _voided_edges')
conn.close()
def import_keywords(self):
nodes = []
for keyword in self.metadata['keywords'].split(","):
keyword=keyword.strip()
node = self.driver.nodes().labels('keyword').props({'value':keyword}).first()
if not node:
doc = self.signpost.create()
node = Node(label='keyword',node_id = doc.did,properties = {'value':keyword})
self.driver.node_merge(node=node)
print 'create new keyword %s' % keyword
nodes.append(node)
return nodes
def fake_get_nodes(dids):
nodes = []
for did in dids:
try:
file_name = files.get(did, {})["data"]["file_name"]
except ValueError:
file_name = did
nodes.append(
Node(
node_id=did,
label="file",
acl=["open"],
properties={
"file_name": file_name,
"file_size": len("fake data for {}".format(did)),
"md5sum": "fake_md5sum",
"state": "live",
},
))
return nodes
def import_metadata(self):
with self.driver.session_scope():
if not self.validate_metadata():
return
doc = self.signpost.create()
doc.urls=[self.metadata['url']]
doc.identifiers = {
'ark':self.search_identifier('ark:/31807/osdc-'+doc.did.split('-')[0])
}
doc.patch()
properties = self.metadata.copy()
del properties['url']
del properties['keywords']
node = Node(node_id=doc.did,label='dataset',properties=properties)
self.driver.node_merge(node=node)
keyword_nodes = self.import_keywords()
for keyword in keyword_nodes:
self.driver.edge_insert(Edge(node.node_id,keyword.node_id,'member_of'))
print 'metadata %s created' % doc.did