We recently redid our website. The new site included a new design and much more content explaining what we do. We wanted a quick way to check over everything and make sure we didn't miss any spelling errors or typos. First I started looking for a web service that could scan the site for spelling errors. I found spellr.us, which is nice but would only catch errors once they were live. It also can't scan all of the pages which require being logged in.
I was pairing with Avdi who thought we should just run Aspell, which worked out great. We were originally trying to just create a simple Emacs macro to go through all our HTML files and check them but in the end created simple Rake tasks, which makes it really easy to integrate spellcheck into CI. After Avdi figured out the commands we needed to use on each file to get the information we needed from Aspell, it was easy to just wrap the command using Rake's FileList. To keep everyone on the same setup, we created a local dictionary of words to ignore or accept and keep that checked into source control as well.
The final solution grabs all the files you want to spell check, then runs them through Aspell with HTML filtering. We have two tasks: one that runs in interactive mode the the user can fix mistakes and one mode for CI that just fails if it finds any errors.
