Draw Networkx Graph with Pyvis

pyvis_for_networkx.py

def draw_graph3(networkx_graph,notebook=True,output_filename='graph.html',show_buttons=True,only_physics_buttons=False):
    """
    This function accepts a networkx graph object,
    converts it to a pyvis network object preserving its node and edge attributes,
    and both returns and saves a dynamic network visualization.
    
    Valid node attributes include:
        "size", "value", "title", "x", "y", "label", "color".
        
        (For more info: https://pyvis.readthedocs.io/en/latest/documentation.html#pyvis.network.Network.add_node)
        
    Valid edge attributes include:
        "arrowStrikethrough", "hidden", "physics", "title", "value", "width"
        
        (For more info: https://pyvis.readthedocs.io/en/latest/documentation.html#pyvis.network.Network.add_edge)
        
    
    Args:
        networkx_graph: The graph to convert and display
        notebook: Display in Jupyter?
        output_filename: Where to save the converted network
        show_buttons: Show buttons in saved version of network?
        only_physics_buttons: Show only buttons controlling physics of network?
    """
    
    # import
    from pyvis import network as net
    
    # make a pyvis network
    pyvis_graph = net.Network(notebook=notebook)
    
    # for each node and its attributes in the networkx graph
    for node,node_attrs in networkx_graph.nodes(data=True):
        pyvis_graph.add_node(str(node),**node_attrs)
        
    # for each edge and its attributes in the networkx graph
    for source,target,edge_attrs in networkx_graph.edges(data=True):
        # if value/width not specified directly, and weight is specified, set 'value' to 'weight'
        if not 'value' in edge_attrs and not 'width' in edge_attrs and 'weight' in edge_attrs:
            # place at key 'value' the weight of the edge
            edge_attrs['value']=edge_attrs['weight']
        # add the edge
        pyvis_graph.add_edge(str(source),str(target),**edge_attrs)
        
    # turn buttons on
    if show_buttons:
        if only_physics_buttons:
            pyvis_graph.show_buttons(filter_=['physics'])
        else:
            pyvis_graph.show_buttons()
    
    # return and also save
    return pyvis_graph.show(output_filename)


##
# For example:
##

# make a new neworkx network
import networkx as nx
G=nx.Graph()

# add nodes and edges (color can be html color name or hex code)
G.add_node('a',color='red',size=4)
G.add_node('b',color='#30a1a5',size=3)
G.add_node('c',color='green',size=1)
G.add_edge('a','b',weight=1023)
G.add_edge('a','c',weight=435)
G.add_edge('b','c',weight=100)

# draw
draw_graph3(G)

Hi @quadrismegistus,

I added a few more parameters to this function. One can now pass set_options() and .Network() parameters to pyvis.
I don't know how pull requests work with gists so I am putting my code here:
https://gist.github.com/maciejkos/e3bc958aac9e7a245dddff8d86058e17

Best,
-M.

Thanks for this! One thing to note - if your nodes are not strings or integers this will fail. Convert your nodes to a unique string in line 34 and line 43 to support plotting these.

Updated! Thanks @DomHudson!

Hi @quadrismegistus

May I know if there is a way to get this working when "x", "y", and "z" are present?

Seems the new version of pyVis doesn't even need this. You just call their function after you've built the network G:

from pyvis.network import Network
vis = Network('500px', '500px')
vis.from_nx(G)
vis.show('pyVis.html')

It's all automated :)

Seems the new version of pyVis doesn't even need this. You just call their function after you've built the network G:

from pyvis.network import Network
vis = Network('500px', '500px')
vis.from_nx(G)
vis.show('pyVis.html')

It's all automated :)

Yes, thanks!

Here is the documentation: https://pyvis.readthedocs.io/en/latest/tutorial.html#networkx-integration