A modular document pipeline processing library.
If it makes sense to model your application as a set of documents (maybe only one) being filtered through a pipeline of modules each modifying those documents (or a shared context) then Tuberia might be for you.
Part flow control, part process manager, all Tuberia. Built around a minimal core, you can add functional modules to perform I/O, parse contents, restructure, or otherwise transform a set of input documents in a configured way. Aside from a library of common modules available from Tuberia, it is easy to create your own either from scratch or based around existing functionality.