3

I created a directed graph using networkx.DiGraph then plotted it using networkx.draw_spring(graph), so all the nodes of the graph have some details stored in a list of dictionaries.

How to add something like a "tooltip" to view these details on mouse hover on each node? If this is possible, how to make this "tooltip" always visible for all nodes, not just by hovering?

busybear
  • 10,194
  • 1
  • 25
  • 42
sudofix
  • 63
  • 2
  • 5

2 Answers2

5

Always visible

To label all the nodes, you just need to use annotate. Something like this

import matplotlib.pyplot as plt
import networkx as nx

G = nx.path_graph(5)
attrs = {0: {'attr1': 20, 'attr2': 'nothing'}, 1: {'attr2': 3}, 2: {'attr1': 42}, 3: {'attr3': 'hello'}, 4: {'attr1': 54, 'attr3': '33'}}
nx.set_node_attributes(G, attrs)
nx.draw(G)

for node in G.nodes:
    xy = pos[node]
    annot.xy = xy
    node_attr = G.nodes[node]
    text = '\n'.join(f'{k}: {v}' for k, v in G.nodes[node].items())
    text = f'node {node}\n' + text
    ax.annotate(text, xy=xy)

On hover

Here's a working example of getting tooltip on hover. This is based off tooltips using standard matplotlib plots here. I used draw_networkx_nodes to get the objects used for hovering and displaying tooltips instead of using draw_spring. But you can manually define the position with spring_layout.

import matplotlib.pyplot as plt
import networkx as nx

G = nx.path_graph(5)
attrs = {0: {'attr1': 20, 'attr2': 'nothing'}, 1: {'attr2': 3}, 2: {'attr1': 42}, 3: {'attr3': 'hello'}, 4: {'attr1': 54, 'attr3': '33'}}
nx.set_node_attributes(G, attrs)

fig, ax = plt.subplots()
pos = nx.spring_layout(G)
nodes = nx.draw_networkx_nodes(G, pos=pos, ax=ax)
nx.draw_networkx_edges(G, pos=pos, ax=ax)

annot = ax.annotate("", xy=(0,0), xytext=(20,20),textcoords="offset points",
                    bbox=dict(boxstyle="round", fc="w"),
                    arrowprops=dict(arrowstyle="->"))
annot.set_visible(False)

def update_annot(ind):
    node = ind["ind"][0]
    xy = pos[node]
    annot.xy = xy
    node_attr = {'node': node}
    node_attr.update(G.nodes[node])
    text = '\n'.join(f'{k}: {v}' for k, v in node_attr.items())
    annot.set_text(text)

def hover(event):
    vis = annot.get_visible()
    if event.inaxes == ax:
        cont, ind = nodes.contains(event)
        if cont:
            update_annot(ind)
            annot.set_visible(True)
            fig.canvas.draw_idle()
        else:
            if vis:
                annot.set_visible(False)
                fig.canvas.draw_idle()

fig.canvas.mpl_connect("motion_notify_event", hover)

plt.show()

enter image description here

busybear
  • 10,194
  • 1
  • 25
  • 42
3

FYI @busybear @ sudofix

This only works if you have your nodes starting from 0.

if you do:

import matplotlib.pyplot as plt
import networkx as nx

nodes = list(range(5))

edges = []
for e1,e2 in zip(nodes[:-1],nodes[1:]):
    edges.append((e1,e2))

G = nx.Graph()
G.add_nodes_from(nodes)
G.add_edges_from(edges)

attrs = {}
for node in G.nodes:
    attrs[node] = {'attr1': node, 'attr2': 'hello', 'attr3': 33}

and leave the rest as

nx.set_node_attributes(G, attrs)

fig, ax = plt.subplots()
pos = nx.spring_layout(G)
nodes = nx.draw_networkx_nodes(G, pos=pos, ax=ax)
nx.draw_networkx_edges(G, pos=pos, ax=ax)

annot = ax.annotate("", xy=(0,0), xytext=(20,20),textcoords="offset points",
                    bbox=dict(boxstyle="round", fc="w"),
                    arrowprops=dict(arrowstyle="->"))
annot.set_visible(False)

def update_annot(ind):
    node = ind["ind"][0]
    xy = pos[node]
    annot.xy = xy
    node_attr = {'node': node}
    node_attr.update(G.nodes[node])
    text = '\n'.join(f'{k}: {v}' for k, v in node_attr.items())
    annot.set_text(text)

def hover(event):
    vis = annot.get_visible()
    if event.inaxes == ax:
        cont, ind = nodes.contains(event)
        if cont:
            update_annot(ind)
            annot.set_visible(True)
            fig.canvas.draw_idle()
        else:
            if vis:
                annot.set_visible(False)
                fig.canvas.draw_idle()

fig.canvas.mpl_connect("motion_notify_event", hover)

plt.show()

everything works great.

But if you change nodes = list(range(5)) to be nodes = list(range(1,6)) it does not work, because node = ind["ind"][0] returns the position of the nodes in G.nodes and not the name of the node, so accessing pos[node] and G.nodes[node] gets the wrong position (it is shifted by 1).

The solution is to create a mapping like

idx_to_node_dict = {}
for idx, node in enumerate(G.nodes):
    idx_to_node_dict[idx] = node

and fix the function to use it as:

def update_annot(ind):
    node_idx = ind["ind"][0]
    node = idx_to_node_dict[node_idx]
Daniel Gilon
  • 56
  • 3