QuerySet relational filtering workaround (as of 5/7/2012. neo4django 03b247d7911753)

gistfile1.py

# a hack for `NodeA.objects.filter(nodeb__in=nodebnodes)`

from neo4django.db import connections
from neo4django.db.models.script_utils import LazyNode

# ...
b_nodes = set([...]) #your NodeB instances
b_node_ids = [b.id for b in b_nodes]

type_node_a = NodeA._type_node()

cypher_query = 'START b=node({b_ids}), t=node({type_node_id}) MATCH b -[r]-> a, t -[`<<INSTANCE>>`]->a RETURN a'
table = connections['default'].cypher(cypher_query, b_ids = b_node_ids, type_node_id = type_node_a.id)

models = [NodeA._neo4j_instance(LazyNode.from_dict(r[0])) for r in table['data']]

Big thanks for this example and the issue comment. Although the ORM layer is great for supporting other Django extensions that assume RDBMS, I would have to agree that Cypher query is a huge plus.

Little question: I've reviewed a few other neo4j extensions, both in the Rails and Python, and for most I've seen them simply put a property in the node that represents the class (ie element_type = 'Person') as opposed to the direction that neo4django takes, which is a class type reference node (ie model_name='Person' - [:<>] -> nodes). I have never been able to find a document that explains the pros and cons of both methods or Lucene comparison against the other. I also assumed that the way neo4django does it was the old way, before built integrated search engines were available such as Neo4j and Lucene. Do you mind sharing your views?