Ask HN: How can I take screenshots of websites programmatically?

run a browser in a vm and invoke a screenshot utility.

run a browser in a vm and script a 'print to pdf'

run a browser in an X session and do a screen grab of X.

There are services that do this, such as http://www.shrinktheweb.com/

Sorry to answer with a trivial Google search, but: http://www.websnapr.com/ might do something for you.

I think this can be done by using the Silverlight WebBrowserBrush control and the WriteableBitmap class. http://msdn.microsoft.com/en-us/library/ff457752(v=VS.95).as... http://msdn.microsoft.com/en-us/library/system.windows.media...

if you need screenshots from multiple browsers selenium is the way to go. saucelabs are great if you don't want to host it yourself.

> Couldn't really find much after googling

http://www.google.com/search?q=take+screenshots+of+websites+...

http://www.google.com/search?q=webpage+screenshot+script

There are a couple of web services that offer the exact thing you're looking for. Websnapr is one, Browsershots is (to my knowledge) the oldest competitor in the space.

I tried using Browsershots years ago, and found it really quite slow, so I instead implemented a local version instead. If you're using Linux, you can programmatically launch a browser, take an X-Screenshot, and run that through ImageMagick to crop out browser chrome and OS things.

Offhandedly, I don't know how you'd go about splicing together multiple screenshots to get the long tail of the page, but I'm sure it's doable too, with thought.

That said, websnapr looks faster than I remember browsershots being, and not knowing your requirements, it may be good enough anyway.

I use self-hosted Browsershots with some modifications. With 4 ShotFactories running, a slow disk, 5 thumbnail sizes, and non-optimal reuse settings (reuse browser only once, reuse VNC server only once) it creates about 10 thumbnails per minute.

It runs headless and I only had to interact with the VNC server years ago when I originally set it up - so that I could configure Firefox with some addons.

(edit: maybe I should have clarified: I only use the Browsershots code, not the service)

What I've done using Xvnc + Perl:

1) Run an Xvnc session.

2) Run a browser inside the xvnc session DISPLAY=:2 /usr/bin/firefox http://url.com/ or whatever.

3) Connect to the vnc session with Net::VNC (vnccapture has example code) and take a screenshot.

4) Crop out the browser UI with Image::Imlib2

I needed to run a full browser because I wanted Flash elements to load. I had something like 20 xvnc/browser instances running simultaneously, which was more than enough for me. It could scale indefinitely though.

You can roll your own with Qt's Webkit module; load the page programmatically and then render the frame into a QImage and save that as a png/jpg etc.

You can use the Python bindings and hack something together in under 50 lines of code. Hook it up to a message queue or some kind of cron job and you're good to go.

Here's some code for doing this with Qt Webkit for a local app:

http://labs.qt.nokia.com/2008/11/03/thumbnail-preview-of-web...

What you'd want to do is make a little server that does the same thing but can serve the resulting image, or just a simple utility that makes the thumbnails on demand and just saves them as images and call that from your web app.