Open danmackinlay opened 7 years ago
I've just implemented citations using the data-cite
attribute in nbsphinx
: https://github.com/spatialaudio/nbsphinx/pull/255.
This works in HTML and LaTeX output, but the citation is displayed quite differently in the Notebook app and on nbviewer
.
Ideally, there should be some kind of JavaScript-based BibTeX processor in the Notebook app (or rather in JupyterLab) which would be able to process a local BibTeX file. And of course a similar thing should be used on nbviewer
.
I would also be interested in any new developments!
Currently if you wish to write citations in a jupyter notebook you must choose between the HTML Way
which is not supported by any major citation management software I'm aware of,
and the LaTeX Way
which looks rubbish in the HTML version, but will export ok, and which is available as a drag-and-drop output format from, for example, Zotero.
@chrisjsewell implemented a regex-based converter for his
nbconvert
addon, ipypublish.Additionally @takluyver implementd cite2c provide a JS-based extension which constructs this markup automatically, but it requires public references and only works for Zotero, which much as i like it, is not the only citation manager.
I wonder if this is the kind of thing that nbconvert should be offering natively? Or some other way of making the process of citing, which is a major use case for these processes, a little bit easier?
The
ipypublish
workflow has its own difficulties of course, but it seems like a commonly reported pain paint - every time I present jupyter to my academic colleagues, for example, I have a conversation where they say "Wow! I could almost write my papers in jupyter! but how do i cite my references?" and I say "er, you laboriously type in cite keys from some bibtex file wrapped in non-obvious html markup", and the conversation ends rapidly.Understanding this is a non-trivial design problem, I'm nonetheless curious is there are any developments or plans here, or if cite2c and ipypublish are the best way forward?