Matplotlib not working with Python 2.7 and Django on OSX

Question

I am trying to use matplotlib and mpld3 to produce some html plots on my Django report app.

Basically I have a controller for the plot that is the following:

from django.shortcuts import render
import mpld3
from matplotlib.pyplot import figure, title, bar

def cpfLogin(request):    
     mpl_figure = figure(1)
     xvalues = (1,2,3,4,5)   
     yvalues = (1,2,3,4,5)

     width = 0.5  # the width of the bars    
     title(u'Custom Bar Chart')
     bar(xvalues, yvalues, width)
     fig_html = mpld3.fig_to_html(mpl_figure)

     context = {
         'figure': fig_html,
     }

     return render(request, 'reports/CPFReport.html', context)

The code for reports/CPFReport.html is:

 {% load i18n %}

 {% block extrahead %}
     <style type="text/css">
         .chart_title {
             font-weight: bold;
             font-size: 14px;
         }
     </style>
  {% endblock %}

 {% block content %}
     <div id="content-main">
         <div class="chart_title">
             {% trans "Custom Bar Chart" %}
         </div>
         {{ figure|safe }}
     </div>
 {% endblock %}

The code is executed right and the plot is displayed correctly but after a couple of seconds the app terminates with the following error:

Assertion failed: (NSViewIsCurrentlyBuildingLayerTreeForDisplay() != currentlyBuildingLayerTree), function NSViewSetCurrentlyBuildingLayerTreeForDisplay, file /BuildRoot/Library/Caches/com.apple.xbs/Sources/AppKit/AppKit-1561.20.106/AppKit.subproj/NSView.m, line 14480.

I found out that if I comment all the code this exception is thrown when any of the matplotlib libraries are called.

Does anyone has a workaround or solution for this problem?

Answer 1

In my case I had to avoid importing :

import matplotlib.pyplot as plt

fig,ax = plt.subplots(figsize=(8,9))
l      = plt.plot(x,s, 'y-', label="line")

and substitute it with:

from matplotlib.figure import Figure

fig = Figure()
ax  = fig.add_subplot(111))
l   = ax.plot(x,s, 'y-', label="line")

Answer 2

Maybe I find the solution, just add the follow code in top.

import matplotlib
matplotlib.use('Agg')

For my case, I use python3, flask and matplotlib.

reference: https://gist.github.com/tebeka/5426211

Answer 3

Adding plt.close() after saving the figure using fig.savefig('../static/images/something.png') helped me.

Answer 4

To complete mapp mapp's answer, in this case it's linked to using matplotlib with a webserver. The solution recommended by matplotlib documentation is to use the Agg backend :

import matplotlib
matplotlib.use('Agg')
# then import pyplot and mpld3
import mpld3
from matplotlib.pyplot import figure, title, bar