Double event registered on mouse-click if legend is outside axes

Question

I have to plot several data sets in one plot. It is useful to be able to highlight one or more of the plots in order to compare them. For this, I toggle the line style of th

Answer 1

My diving into the legend artist has found that a legend line is in the children tree of a legend twice when the legend has the bbox_to_anchor set.

I asked about this here with my solution where I watched for a NEW mouseevent and kept track of the artists that had already been handled by my callback.

I've asked for comments if anyone thinks there's a more elegant way to handle this "feature"

I'm not sure this is a bug. But, it seems unique to legends where the children lines are held in the .lines attribute and deep in the packing boxes data structure - the get_children method finds both of these. Luckily they are the same object rather than a copy so I could check for a line already been handled.

Answer 2

Another approach similar to tcaswell's solution counting events. Only 3 lines added to the original legend picking example. It uses python's function attributes.

def onpick(event):
    if onpick.count % 2 == 0:    #### LINE ADDED ####  
        # on the pick event, find the orig line corresponding to the
        # legend proxy line, and toggle the visibility
        legline = event.artist
        origline = lined[legline]
        vis = not origline.get_visible()
        origline.set_visible(vis)
        # Change the alpha on the line in the legend so we can see what lines
        # have been toggled
        if vis:
            legline.set_alpha(1.0)
        else:
            legline.set_alpha(0.2)
        fig.canvas.draw()
    onpick.count += 1    #### LINE ADDED ####

onpick.count = 0    #### LINE ADDED ####

Answer 3

If you monkey patch Artist.pick with the following:

matplotlib.artist.Artist.orig_pick = matplotlib.artist.Artist.pick
def nu_pick(self, me):
    print self
    matplotlib.artist.Artist.orig_pick(self, me)

matplotlib.artist.Artist.pick = nu_pick

You can look at how the artists recurse on a pick event. (Each Artist object calls pick on it's self and then on all of it's children). For reasons I don't understand, there are two copies of each line in the drawing area of the legend (and it behaves differently when it is inside and outside).

A way-hacky solution is to just count how many times the leglines have been hit, and only toggle on the odd ones:

import pylab
import numpy

# Create data for plotting
t = numpy.linspace(0, 1.0, 100) 
a = numpy.sin(2*numpy.pi*t)

# Set up figure
fig = pylab.figure()
ax = pylab.subplot(111)

# Plot figures    
lines = []    
for i in range(5):
    line = ax.plot(t, (i+1)*a, linestyle=':', picker=5, label='line%d'%(i+1)) 
    lines.append(line[0]) # Save plot lines

# Create legend
leg = ax.legend(bbox_to_anchor=(1.01, 1), loc=2) # Does not work as expected
#leg = ax.legend() # Works!!

# Get legend lines
leglines = leg.get_lines() 
# Set event for legend lines
for line in leglines:
    line.set_picker(5)

# Create a 2 way mapping between legend lines <-> plot lines    
line2leg = dict(zip(lines+leglines, leglines+lines))
count_dict = dict((l, 0) for l in lines )
# Define event function
def onpick(event):
    thisline = event.artist
    print event
    print thisline
    if thisline in lines:
        print 'lines'
        count_dict[thisline] = 0
    elif thisline in leglines:
        print 'leglines'
        thisline = line2leg[thisline]
        count_dict[thisline] += 1
    print 'added'
    if (count_dict[thisline] % 2) == 1:
        print count_dict[thisline]
        return
    print 'tested'
    if thisline.get_linestyle()==':':
        print ": -> -" # For debugging
        thisline.set_linestyle('-')
        line2leg[thisline].set_linestyle('-')
    else:
        print "- -> :" # For debugging
        thisline.set_linestyle(':')
        line2leg[thisline].set_linestyle(':')
    fig.canvas.draw()

# connect event function    
fig.canvas.mpl_connect('pick_event', onpick)
pylab.show()

(I left all my de-bugging statements in).

Pretty sure this is a bug, if you don't want to create an issue on github I will.