It crawls the web site and scrapes web pages using WebDriver. Then it compares the HTML source & screenshot, giving the percentages changed that can be measured against a baseline and threshold. A really smart coworker developed the diff tool & algorithms.
It’s great for comparing different versions of the site (say development against production) or doing a pixel comparison in different browsers.
The MongoDB document store is a great fit because we can serialize our hierarchical data and persist it without creating a DB schema, and the send JSON from Mongo to the dashboard where it is rendered with only a only a thin web service wrapper layer.
It seems like after our presentation to several testers, a couple managers and development architects that there’s a lot of excitement for using the tool. I can’t make any promises, but we may even be able to open source it.