For context, I'm trying to handle mouse events on instances of Collection from matplotlib. matplotlib handles the mouse events too, the problem is that I'm doing this from networkx, which uses matplotlib.pyplot.scatter to plot nodes on a graph. I'm not modifying matplotlib itself, because I don't think these features would be useful to matplotlib.
I also want to be able to pick and choose which events are handled and allow users to choose what happens when a mouse event is detected. So, for example, if I want something to be draggable, I would allow it to handle mouse click, motion and unclick events. However, if I want it to just give me information about a node, it just needs to handle the mouse click event.
matplotlib does have an example here, where they wrap the instance in a Draggable class, connect to the mouse events and have functions to handle behaviour. In my case, I wouldn't know which events I would want to handle or I might not implement the callback functions for each event.
As I see it at the moment, one option is to create methods that could be bound to the instance:
import types
import matplotlib.pyplot as plt
def on_click(self, mouse_click_event):
'''callback on click event, mouse_click_event is passed
automatically by mpl_connect'''
pass
node_collection = plt.scatter(...) # Has a lot of arguments, returns a PathCollection
node_collection.on_click = types.MethodType(on_click, node_collection)
node_collection.figure.canvas.mpl_connect('button_press_event',
node_collection.on_click)
I think it would be relatively easy to make a series of functions that could connect the mouse events to the collection and add bound methods for the callbacks without everything clashing.
Is there a better way to be doing this? I've read up about class decorators, metaclasses and abstract base classes, but I don't think they are appropriate for my particular problem. Is there something else I haven't heard of (or more likely something that I have heard of and didn't figure to use...)?
