WIP: Validate documents against the openmensa schema

[klemens]

This is useful for development, as you may not notice when the generated xml is not valid (as I did in #37).

The first commit adds validating support to parse.py, which makes lxml a mandatory dependency for dev.

I also think it would be useful to add validation to the travis builds, maybe by replacing the wget commands with a python script that validates the returned documents. What do you think?

[mswart]

Looks good as fallback layer. I am unsure whether it is good to have the schema copied into this project. Independent from that I have updated PyOpenMensa (c3fd2ec) to raise ValueErrors on empty category names and empty notes: mswart/pyopenmensa@05d9fd6.

For now I would keep wget on travis. My goal would be to adjust PyOpenMensa to prevent all invalid XML generation.

[klemens]

You are right, adding the schema to the repository isn't so great. I added it, because downloading it on every call to parse.py seemed even worse. However making sure PyOpenMensa produces valid documents is the far better option.

I think validating on travis would still be useful, especially as you are using it to directly deploy on a successful build. Instead of using a python script for validation we could also just pipe the files through xmllint and download the schema once for every build, so we don't have to add it to the repo.