GitLab

Categories with spaces in should use underscores in the category
name in the URL. These have now been fixed.

After changing the names to be Unicode internally, I forgot that the
URL encoding was still needed for these names when we request from
the server. This should now be fixed.

There was some confusion in the naming of the files and the internal
`name` properties. Now it should be the case that internally we deal
with unicode names, and on the command line we supply UTF-8 names.

The command line is obviously UTF-8 in order to ensure that we can supply
all the names we need on the command line, but internally the choice
to use unicode was driven by what I feel is the more sane of the
choices - that the native python representation is used internally,
and only at presentation time does the user need to worry about how
that is encoding.

That also affects fsname, which will now be forced to UTF-8, rather
than just taking the wiki name, which might be percent encoded.

The code has been restructured so that the library is able to be
used on its own, with the JSON functions and the CLI as separate
bits.

Simple documentation added, and a BSD license added to the repo.

We now have a load of command line options to read and extract information
from the Rosetta Code data. It isn't especially nice, and if anyone actually
wanted more control they'd directly use the structures.

So that I can test the compiler, the reader now will write out a bunch
of files in a directory. This should let me test Norcroft against the
examples. It can also optionally write out the intro and task sections
from the page, which might be useful in understanding what's intended.

Categorys group a number of related Tasks.
Tasks set out a goal to implement.
Languages hold a single language within the Task.
CodeBlocks may be given which contain the code to implement the
Task in a given Language.

Being able to request a single Task is the first step to extracting the
data from Rosetta Code.

The Task object will hold that information, and at the moment it just
contains the bare markdown used to exit the page.