Returning MatPotLib image as string

Question

I am using matplotlib in a django app and would like to directly return the rendered image. So far I can go plt.savefig(...), then return the location of the image

What I want to do is:

return HttpResponse(plt.renderfig(...), mimetype="image/png")

Any ideas?

Answer 1

Employ ducktyping and pass a object of your own, in disguise of file object

class MyFile(object):
    def __init__(self):
        self._data = ""
    def write(self, data):
        self._data += data

myfile = MyFile()
fig.savefig(myfile)
print myfile._data

you can use myfile = StringIO.StringIO() instead in real code and return data in reponse e.g.

output = StringIO.StringIO()
fig.savefig(output)
contents = output.getvalue()
return HttpResponse(contents , mimetype="image/png")

Answer 2

what about cStringIO?

import pylab
import cStringIO
pylab.plot([3,7,2,1])
output = cStringIO.StringIO()
pylab.savefig('test.png', dpi=75)
pylab.savefig(output, dpi=75)
print output.getvalue() == open('test.png', 'rb').read() # True

Answer 3

Django's HttpResponse object supports file-like API and you can pass a file-object to savefig.

response = HttpResponse(mimetype="image/png")
# create your image as usual, e.g. pylab.plot(...)
pylab.savefig(response, format="png")
return response

Hence, you can return the image directly in the HttpResponse.

Answer 4

There is a recipe in the Matplotlib Cookbook that does exactly this. At its core, it looks like:

def simple(request):
    from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
    from matplotlib.figure import Figure

    fig=Figure()
    ax=fig.add_subplot(111)
    ax.plot(range(10), range(10), '-')
    canvas=FigureCanvas(fig)
    response=django.http.HttpResponse(content_type='image/png')
    canvas.print_png(response)
    return response

Put that in your views file, point your URL to it, and you're off and running.

Edit: As noted, this is a simplified version of a recipe in the cookbook. However, it looks like there is a difference between calling print_png and savefig, at least in the initial test that I did. Calling fig.savefig(response, format='png') gave an image with that was larger and had a white background, while the original canvas.print_png(response) gave a slightly smaller image with a grey background. So, I would replace the last few lines above with:

    canvas=FigureCanvas(fig)
    response=django.http.HttpResponse(content_type='image/png')
    fig.savefig(response, format='png')
    return response

You still need to have the canvas instantiated, though.