I'm trying to get the number of nodes of a Neo4j graph database using Python, but I don't find any method or property to do that.
Does anybody how can I get this information?
Other Python packages like NetworkX has a method to get this information.
>>> G = nx.Graph() # or DiGraph, MultiGraph, MultiDiGraph, etc
>>> G.add_path([0,1,2])
>>> len(G)
3
Update:
Since I first wrote this, the answer has changed. The database now keeps exact counts of total nodes, as well as counts by label. Unlike most databases, this is not a heuristic, these counters are transactionally kept in sync with the rest of the datastore.
This means you can get exact node counts in O(1) time from Neo4j. You get access to them by asking Cypher:
MATCH (n) RETURN count(*)
Original reply:
There are two ways to get the number of nodes in a neo4j database. The first one is to actually iterate through all the nodes, and counting them.
Alternative two is to use the "number of node ids in use" statistic provided by the db kernel, which does not guarantee to be exact, but will be at least the number of nodes in use. In a high-load db it will be higher, since it also contains ids of deleted nodes that have not been reclaimed yet.
Alt one is reasonably exact (depending on how many are created/deleted while you iterate), but can be super slow. Alt two is potentially way off, but is a O(1) operation.
You currently don't have much choice, because alt one is the only one that works. It isn't officially supported, so doing it today looks a bit dirty:
from neo4j import GraphDatabase
db = GraphDatabase('..')
node_count = sum(1 for _ in db.getAllNodes().iterator())
I've added two issues for this, one to add support for accessing management info (eg. support the alt two method), and one to add support for these use cases:
node_count = sum(1 for _ in db.nodes)
node_count = len(db.nodes)
Follow these issues here:
https://github.com/neo4j/python-embedded/issues/7
https://github.com/neo4j/python-embedded/issues/6
Please let us now if you run into any other trouble with neo4j-embedded, add a ticket to the github issues if you discover any bugs or think of any other enhancements!