Graff is a graph database for python, implemented using sqlalchemy for compatibility with a wide range of database storage engines. It is also inspired by sqlalchemy in the way it allows queries to be built generatively. But for most uses, the user does not need to have knowledge of either SQL or sqlalchemy.
Graff evolved from the tangos database project for cosmological simulations.
Usage examples
Initialise a graph database in RAM:
import graff
mydb = graff.Connection()Initialise a graph database in the specified mySQL database:
mydb = graff.Connection('mysql+pymysql://mysql_user:mysql_password@localhost/mysql_database_name')(For more information see the SQLAlchemy documentation around dialects and engines.)
Initialise a graph database in RAM and populate it with a random network of people and friend relationships:
import graff.testing
mydb = graff.testing.init_friends_network(n_people=10, n_connections=100)Get the 'nodes' corresponding to all people:
mydb.query_node("person").all()Get the names of all people:
mydb.query_node("person").return_property("name").all()Count the number of friends-of-friends connections:
mydb.query_node("person").follow("likes").follow("likes").count()Get the names of all friends-of-friends, and print the first 100 pairs:
fof = mydb.query_node("person").return_property("name").\
follow("likes").follow("likes").\
return_property("name").all()
for name_a, name_b in fof[:100]:
print(name_b, "is a friend of a friend of", name_a)Get all known properties of the first person in the database:
mydb.query_node("person").return_properties().first()Return the edge objects linking people:
mydb.query_node("person").edge("likes").all()Return a property from the edge as well as the nodes:
results = mydb.query_node("person").return_property("name").\
edge("likes").return_property("num_messages").\
node().return_property("name").all()
for name_a,num,name_b in results:
print(name_a, "has sent", num, "messages to", name_b)
