Strings returned by the library should be unicode
When processing the strings from the description the library
currently returns str instances. It would be way better if
the strings returned were unicode objects.
In order to do so we should thus specify the encoding of the
file when parsing it.