issues with how unstructured ref data is put in XML file from bib file

[danielskatz]

It appears now (see https://github.com/openjournals/joss-reviews/issues/1371#issuecomment-586962197, though the old XML file from that time is no longer apparent to me) that the process of generating XML data for unstructured references (ones that do not have DOIs) takes all the elements of the item from the bib file and strings them together. Ideally, this would be more selective, and would just pick certain fields (e.g., author, title, url, journal, ...) and would ignore other fields (e.g. Date-Added, Date-Modified, Bdsk-Url-1, Urldate, ...)

[arfon]

@tarleb - any thoughts on how this could be improved? We currently do exactly as @danielskatz describes here. For the PDF we use Pandoc + the APA CSL which presumably handles some of this complexity for formatting the citations?

[tarleb]

I believe this could be done by changing lines 1347 to 1367 of the apa.csl definition, but I haven't tested that yet.