python - Draw more information on graph\nodes using PyGraphviz

Question

I want to create a graph and draw it, so far so good, but the problem is that i want to draw more information on each node. I saw i can save attributes to nodes\edges, but how do i draw the attributes? i'm using PyGraphviz witch uses Graphviz.

Answer 1

An example would be

import pygraphviz as pgv
from pygraphviz import *
G=pgv.AGraph()
ndlist = [1,2,3]
for node in ndlist:
    label = "Label #" + str(node)
    G.add_node(node, label=label)
G.layout()
G.draw('example.png', format='png')

but make sure you explicitly add the attribute label for extra information to show as Martin mentioned https://stackoverflow.com/a/15456323/1601580.

Answer 2

You can only add supported attributes to nodes and edges. These attributes have specific meaning to GrpahViz.

To show extra information on edges or nodes, use the label attribute.

Answer 3

If you already have a graph with some attribute you want to label you can use this:

def draw_nx_with_pygraphviz_attribtes_as_labels(g, attribute_name, path2file=None):
    import matplotlib.pyplot as plt
    import matplotlib.image as mpimg

    # https://stackoverflow.com/questions/15345192/draw-more-information-on-graph-nodes-using-pygraphviz

    if path2file is None:
        path2file = './example.png'
        path2file = Path(path2file).expanduser()

    g = nx.nx_agraph.to_agraph(g)
    # to label in pygrapviz make sure to have the AGraph obj have the label attribute set on the nodes
    g = str(g)
    g = g.replace(attribute_name, 'label')  # it only
    print(g)
    g = pgv.AGraph(g)
    g.layout()
    g.draw(path2file)

    # https://stackoverflow.com/questions/20597088/display-a-png-image-from-python-on-mint-15-linux
    img = mpimg.imread(path2file)
    plt.imshow(img)
    plt.show()

    # remove file https://stackoverflow.com/questions/6996603/how-to-delete-a-file-or-folder
    path2file.unlink()

# -- tests

def test_draw():
    # import pylab
    import networkx as nx
    g = nx.Graph()
    g.add_node('Golf', size='small')
    g.add_node('Hummer', size='huge')
    g.add_node('Soccer', size='huge')
    g.add_edge('Golf', 'Hummer')
    draw_nx_with_pygraphviz_attribtes_as_labels(g, attribute_name='size')

if __name__ == '__main__':
    test_draw()

result:

in particular note that the two huge's did not become a self loop and they are TWO different nodes (e.g. two sports can be huge but they are not the same sport/entity).

related but plotting with nx: Plotting networkx graph with node labels defaulting to node name