An Object Graph Mapper (OGM) for the neo4j graph database, built on the awesome py2neo.
- Familiar Django model style definitions.
- Powerful query API.
- Enforce your schema through cardinality restrictions.
- Full transaction support.
- Hooks including (optional) Django signals support.
Available on readthedocs.
- Python 2.7, 3.4
- neo4j 2.0, 2.1, 2.2
Install from pypi (recommended):
$ pip install neomodel
To install from github:
$ pip install git+git://github.com/robinedwards/neomodel.git@HEAD#egg=neomodel-dev
Please note that if you are utilizing Neo4j version 2.2 or newer there are some additional setup steps necessary relating to authentication. As of version 2.2 Neo4j authentication is activated by default on new instances. Please follow the outstanding documentation provided by py2neo's Authentication section to setup new credentials. If you are utilizing a hosted service this is most likely already taken care for you.
There is one modification that is necessary when transitioning from version 1.x to 2.x relating to direct cypher queries. In version 1.x performing a cypher query always returned at least one array within an array. This enabled some assumptions to be made when querying for single object directly. Since you could assume the result would always be a list within a list the result of a response could be safely accessed like the following:
query = 'MATCH (a:Profile {username: "%s"}) RETURN a' % (username)
res, col = db.cypher_query(query)
end_result = res[0][0]
If there wasn't a Profile with the given username nothing would fail out and
end_result would be set to an empty list.
In version 2.x we lean a bit more on py2neo to handle the response. This results
in what py2neo refers to as a RecordList.
A RecordList is a list of Record objects which provide
a few more capabilities on accessing the results. The draw back is
that now accessing the above query at res[0][0] will result in an IndexError
due to the new response being an empty RecordList. Py2neo has however provided
a solution to this which is the one method. This method can be called on the
cypher query response to achieve the old res[0][0] functionality. So in
version 2 the above query would look like:
query = 'MATCH (a:Profile {username: "%s"}) RETURN a' % (username)
res, col = db.cypher_query(query)
end_result = res.one()
This will result in end_result being set to None.
Ideas, bugs, tests and pull requests always welcome.
Make sure you have a fresh virtualenv and nose installed:
$ pip install nose
A Neo4j database to run the tests on, (it will wipe this database):
$ export NEO4J_REST_URL=http://localhost:7474/db/data # (the default)
Install neomodel for development and run the suite:
$ python setup.py develop $ nosetests -s
