Converting image to string for streaming to web browser

Hi matplotlib users,

I'm using matplotlib for an application for producing tsunami travel time
contour maps. Recently I discovered the CherryPy (cherrypy.org) web server
framework which allows you to very easily embed a web server in your
application. This allows the application to be accessed remotely through a
web browser.

At present the application produces the contour plots when it receives input
from a web browser that it should do so. The contour plot png image is then
saved to disk and served to the user.

CherryPy is able to serve/stream the image directly to the browser if it is
converted to a string using an appropriate encoding (png, rgb maybe). PIL is
able to do this conversion:

http://www.theorganization.net/mt/archives/2005/05/30/ive_finally_got_it.html

I would like to do something similar with my matplotlib image (i.e. convert it
to a string with an appropriate encoding) instead of writing it to the
harddisk. Does anyone know if that is possible?

Kind regards,
Jesper

I've just been working on something similar (although not for web use)
myself and came up with the following solution although it's possible
there's a better way.

Using the agg backend you can obtain an RGBA buffer or RGB string which
can then be loaded as a PIL Image for processing. I've adapted a the
examples/agg_oo.py to demonstrate.

···

On Wed, 2005-08-24 at 12:58 +0200, Jesper Larsen wrote:

CherryPy is able to serve/stream the image directly to the browser if it is
converted to a string using an appropriate encoding (png, rgb maybe). PIL is
able to do this conversion:

http://www.theorganization.net/mt/archives/2005/05/30/ive_finally_got_it.html

I would like to do something similar with my matplotlib image (i.e. convert it
to a string with an appropriate encoding) instead of writing it to the
harddisk. Does anyone know if that is possible?

----
from matplotlib.backends.backend_agg \
    import FigureCanvasAgg as FigureCanvas
from matplotlib.figure import Figure
import Image

fig = Figure()
canvas = FigureCanvas(fig)
ax = fig.add_subplot(111)
ax.plot([1,2,3])
ax.set_title('hi mom')
ax.grid(True)
ax.set_xlabel('time')
ax.set_ylabel('volts')
canvas.draw()
size = canvas.get_width_height()
usebuffer = True
if usebuffer:
    # Load the agg buffer directly as the source of the PIL image
    # - could be less stable as agg and PIL share memory.
    buf = canvas.buffer_rgba()
    im = Image.frombuffer('RGBA', size, buf, 'raw', 'RGBA', 0, 1)
else:
    # Save the agg buffer to a string and load this into the PIL image.
    buf = canvas.tostring_rgb()
    im = Image.fromstring('RGB', size, buf, 'raw', 'RGB', 0, 1)

im.show()
----

Hope this helps,
Nick