def to_dict(self):
return {
'start': self.data['start'],
'end': self.data['end'],
'length': self.length,
'types': self.types,
'labels': self.labels,
'start_entity_id': self.start_entity_id,
'end_collection_id': self.end_collection_id,
'nodes': unwind(self.data['paths']),
}
def from_data(cls, start_entity, end_collection_id, paths, types,
labels, start, end):
obj = cls()
obj.start_entity_id = start_entity.id
obj.end_collection_id = end_collection_id
obj.labels = unwind(labels)
obj.types = unwind(types)
lengths = map(len, paths)
obj.length = min(lengths) - 1
average = (float(sum(lengths)) - len(lengths)) / float(len(lengths))
obj.weight = len(paths) * (1.0 / max(1.0, average))
obj.data = {
'start': start,
'end': end,
'paths': paths
}
db.session.add(obj)
return obj
def generate_paths(graph, entity, ignore_types=SKIP_TYPES):
"""Generate all possible paths which end in a different collection."""
Path.delete_by_entity(entity.id)
if graph is None or entity.state != entity.STATE_ACTIVE:
return
log.info("Generating graph path cache: %r", entity)
# TODO: should max path length be configurable?
q = "MATCH pth = (start:Aleph:Entity)-[*1..3]-(end:Aleph:Entity) " \
"MATCH (start)-[startpart:PART_OF]->(startcoll:Collection) " \
"MATCH (end)-[endpart:PART_OF]->(endcoll:Collection) " \
"WHERE start.fingerprint = {entity_fp} AND " \
"startpart.alephCanonical = {entity_id} AND " \
"startcoll.alephCollection <> endcoll.alephCollection AND " \
"all(r IN relationships(pth) WHERE NOT type(r) IN {ignore_types}) " \
"WITH DISTINCT start, end, " \
" COLLECT(DISTINCT extract(x IN nodes(pth) | x.id)) AS paths, " \
" COLLECT(DISTINCT extract(x IN nodes(pth) | labels(x))) AS labels, " \
" COLLECT(DISTINCT extract(r IN relationships(pth) | type(r))) AS types, " \
" COLLECT(DISTINCT endcoll.alephCollection) AS end_collection_id " \
"RETURN start, end, paths, types, labels, end_collection_id "
count = 0
for row in graph.run(q,
entity_id=entity.id,
entity_fp=entity.fingerprint,
ignore_types=ignore_types):
labels = unwind(row.get('labels'))
labels = [l for l in labels if l != BASE_NODE]
types = unwind(row.get('types'))
if len(types) == 1 and 'AKA' in types:
continue
Path.from_data(entity, row.get('end_collection_id'), row.get('paths'),
types, labels, NodeType.dict(row.get('start')),
NodeType.dict(row.get('end')))
count += 1
db.session.commit()
# TODO: send email to collection owners?
log.info("Generated %s paths for %r", count, entity)