Le mardi 30 avril 2019 à 04:43 -0200, Ken Mankoff a écrit : > Hi Emmanuel, > > I have looked into PDF and DOCX export repeatedly over the years with > different versions of Org and Pandoc and Google Docs and keep finding > that the best method is Org -> LaTeX, and then LaTeX -> DOCX with > Pandoc.
That's also basically the advice given by John Kitchin. > I use this babel block to achieve this: > > #+BEGIN_SRC sh :results verbatim :var fn=(file-name-sans-extension > (buffer-name)) > bibexport -o ${fn}.bib ${fn}.aux > pandoc -f latex -i ${fn}.tex -t DOCX -o ${fn}.docx --bibliography > ./${fn}.bib > #+END_SRC > > The first line may not be needed. That extracts the references used > in this document from my single massive (3500 element) bib file. Whew ! How do you manage this ? I gave up using JabRef fo a large collection, and use Zotero (which allows me to share my bibliographic base with Wordheads co-workers.1 have about 14 000 references dispatched in one (small) personal library and three (large: academic physicians are (have to be) chatty...) shared libraries. I create a problem specific collection to collect the relevant references, and export that. I found GetterBibtex's auto-export feature to be very useful. > The advantage of this Org -> LaTeX -> Pandoc -> DOCX method is that > if you need to patch something for Pandoc (e.g. to fix your code > captions or something) you could insert a NoWeb <<fix-tex-for- > pandoc>> block before the pandoc line that generates a second, > corrected, LaTeX file. Hmmm... this way lies madness : it multiplies the possible sources for a bug. I'd rather avoid that, and fight for a *correct* LaTeX source, which is not the case at the moment. > > Note that I don't use org-ref, just the standard Org referencing > code, but I don't see why this would change anything, if you're able > to generate PDFs that meet your requirements. Org-ref has its points : it offers the ability to create/update and maintain a document-specific .bib file, ythanks to a set of interfaces to PubMed/DOI/etc.... I don't give up on the idea of suggesting to John Kitchin an interface to Zotero : a very crude interface r(Totxt) exists, that can be used to communicate with a local Zotero process. But this is now too rough (remaining bugs, etc..) to prepare a patch- request (I'm not much of a programmer...). Back to the problem at hand : I'm still bugged by the fact that ox- pandoc does the right thing in the present case (as well as the built- in HTML exporter, by the way), and that the built-in LaTeX exporter is buggy. I just posted a formal bug report limited to this. Cordially yours, -- Emmanuel Charpentier