Method to save networkx graph to json graph?

Each Answer to this Q is separated by one/two green lines.

Seems like there should be a method in networkx to export the json graph format, but I don’t see it. I imagine this should be easy to do with nx.to_dict_of_dicts(), but would require a bit of manipulation. Anyone know of a simple and elegant solution?

This documentation contains a full description

A simple example is this:

import networkx as nx
from networkx.readwrite import json_graph

DG = nx.DiGraph()
DG.add_edge('a', 'b')
print json_graph.dumps(DG)

You can also take a look at the Javascript/SVG/D3 nice example on adding physics to the graph visualization.

Here is a JSON approach that I just did, together with code to read the results back in. It saves the node and edge attributes, in case you need that.

import simplejson as json
import networkx as nx
G = nx.DiGraph()
# add nodes, edges, etc to G ...

def save(G, fname):
    json.dump(dict(nodes=[[n, G.node[n]] for n in G.nodes()],
                   edges=[[u, v, G.edge[u][v]] for u,v in G.edges()]),
              open(fname, 'w'), indent=2)

def load(fname):
    G = nx.DiGraph()
    d = json.load(open(fname))
    G.add_nodes_from(d['nodes'])
    G.add_edges_from(d['edges'])
    return G

Generally I use the following code :

import networkx as nx; 
from networkx.readwrite import json_graph;
G = nx.Graph();
G.add_node(...)
G.add_edge(...)
....
json_graph.node_link_data(G)

it will create json formatted graph in which the nodes are in nodes and edges in links
in addition to other information about the graph (directionality, … etc)

Try this:

# Save graph
nx.write_gml(G, "path_where_graph_should_be_saved.gml")

# Read graph
G = nx.read_gml('path_to_graph_graph.gml')

The rest of the solutions didn’t work for me. From the networkx 2.2 documentation:

nx.write_gpickle(G, "test.gpickle")
G = nx.read_gpickle("test.gpickle")

Are the nodes and edges enough information? If so, you could write your own function:

json.dumps(dict(nodes=graph.nodes(), edges=graph.edges()))


The answers/resolutions are collected from stackoverflow, are licensed under cc by-sa 2.5 , cc by-sa 3.0 and cc by-sa 4.0 .