John Vandenberg

Hmm, I see that `errors` is already being used in the fallback: ```py else: data = dict(errors=response.data, title=problem_title, status=problem_status, type=problem_type) ```

These macros are the most common appearing in the titles of my dataset that are missing from citeproc's parser. See also #55 for a larger list of missing macros.

See also #54 for Bibtex date parsing issues (i.e. date issues earlier in the pipeline than what this patch fixes).

This patch could be extended further to support literal name, as that fixes another two tests. https://github.com/jayvdb/citeproc-py/commit/d820ad4e48431b01ed50610157695a8e792ee73b

Just noting that lxml apparently crashed on pypy according to this commit https://github.com/brechtm/citeproc-py/commit/370e1187c79f8b6229195f37e91a1e853cd892b7 Would be good to make that reproducible and raise a bug upstream. We could look for a...

Thanks for the update @njsmith . After working around #63, the previous lxml problems have disappeared (or moved from sight). However something else is making the build fail. https://travis-ci.org/jayvdb/citeproc-py/builds/116795654 PyPy...

https://bugs.launchpad.net/lxml/+bug/1442427 "lxml fails to install on pypy3" has been marked wont fix with comment ``` Marking as "won't fix" because it's up to PyPy to come up with a working...

The `global` declaration is not needed anyway.

No particular rush for me on this one, as I have a workaround (https://github.com/jayvdb/pubpdf/commit/6e0d9e5131d5b449f98928245ef3f1a332313040), but other uses might be surprised by the wrong number appearing for the issue.

See #57 for one complexity of BibTex mapping to CSL.