I’m doing an OPW internship at Wikimedia, and last week my first patches were merged in.

My project is part of Wikimedia’s Parsoid, which translates back and forth between wikitext (the markup language Wikipedia is written in) and HTML. The testing framework translates around 160,000 Wikipedia pages from wikitext to HTML and back to wikitext, and then uses diffs to compare the results with the original articles.

You can see test outcomes here. You may notice that some of the views aren’t very informative, for example the performance stats view. I’m going to make some UI improvements, including data visualizations. I’m starting by using templates to separate out the parts of the code responsible only for presentation.