Build your own website thumbnail generator with Django

On many website you can see a awesome link preview effect using a thumbnail of the page. There is some web services such as Websnapr or Thumbalizr providing an API to make your own webpages screenshots but it can be annoying to rely on third party projects: free accounts can only display a small among of screenshots and, cannot customize the size of the image, you cannot render Flash animations or Javascript…

Webkit Logo

We will build a website thumbnail generator using CutyCapt, a Webkit based free software designed to make screenshots of web pages, and Django, the powerful Python web framework.

The first step is to download and install CutyCapt. On a Debian-based system (such as Ubuntu) just type the following commands:

sudo apt-get install subversion libqt4-webkit libqt4-dev g++
svn co https://cutycapt.svn.sourceforge.net/svnroot/cutycapt
cd cutycapt/CutyCapt
qmake
make
./CutyCapt --url=//dunglas.fr --out=example.png

Try to open the example.png (i.e: eog example.png), if the install is OK you must see a screenshot of this blog.

I assume your Django installation is working fine. Start a new project if needed (django-admin.py startproject mysite) and create an application called webthumb

with python manage.py startapp in your project.

Django Reinhardt

Edit your urls.py file to add a new pattern to the tuple urlpatterns. It will match screenshots requests and call the thumb view:

(r'^thumb/(?P.*)', 'mysite.webthumb.views.thumb'),

Now create a view called thumb in webthumb/view.py with this code:

import os
import subprocess
import md5
from django.http import HttpResponse

CUTYCAPT = '/home/dunglas/cutycapt/CutyCapt/CutyCapt'
THUMBS_DIR  = '/home/dunglas/Documents/Projets/kakofony/thumbs/'

def thumb(request, url):
    hash = md5.new(url).hexdigest()
    path = THUMBS_DIR + hash + '.png'
    print path

    if not os.path.isfile(path):
        try:
            subprocess.check_call([CUTYCAPT,\
                '--url=' + url,\
                '--min-width=200',\
                '--out=' + path])
        except subprocess.CalledProcessError:
            return django.http.HttpResponseServerError

    img = open(path, 'rb').read()
    return django.http.HttpResponse(img, mimetype='image/png')

Adapt the variables CUTYCAPT and THUMBS_DIR with your settings. Of course the directory pointed by THUMBS_DIR must be writeable (chmod 777).

You can now embed website thumbnails in your websites just with this HTML markup <img src="http://localhost/thumb///dunglas.fr" alt="Lapin Blanc" />.

Enjoy !