Over the past years, we (@PierreSelim, @Lena and me) have worked on tools for analysing collections of media on Wikimedia Commons. Our most advanced one so far is [[ https://github.com/Commonists/wm_metrics | wm-metrics ]], which also has a [[ http://tools.wmflabs.org/wm-metrics/ | Web interface on ToolLabs ]], and we started some deep rewrites of the backend with [[ https://github.com/Commonists/MediaCollectionDB | MediaCollectionDB]].
For this hackathon, I worked on setting up a proper testing and quality infrastructure for MediaCollectionDB, with continuous integrations, linting, unit test runner, test coverage, and autodocumentation on ReadTheDocs.
I also worked on refactoring the backend code of wm-metrics, taking steps to decouple the database access layer from the computing code ; and separate the generic code from the concept of a FDC round.