On 2021-07-03, Matt Price wrote:
> I've added some comments in the issue you linked to, but in the meantime
> I've also come up with what seems to be at least a semi-viable hack for
> adding native CSL citation support to org-re-reveal. It involves creating
> two new variables and then let-setting `citeproc-fmt--formatters-alist` in
> `org-re-reveal-export-to-html`. Something similar could presumably be done
> in other derived backends.
>
> I find it quite hackish and I don't know whether perhaps some more general
> solution could be found, but in any case here is the code, which I have
> inserted into org-re-reveal.el locally:
Thank you for sharing your code, Matt!
What is the general state of this new citation handling with respect
to export backends? Did you create the settings for HTML from
scratch or did you take inspiration from HTML export functionality?
I guess that basic support should go into ox-html, while small
modifications could than take care of specifics for reveal.js
presentations...
Best wishes
Jens
Thanks Jens!
So, there is a rather complicated chain of dependencies and I'm not sure how best to procee in the specific case of org-re-reveal.
Here is how I understand the current state of affairs (and I think Bruce, Nicolas, and Andras are all likely to have corrections to my explanation):
- the new syntax and processors are available in the wip-cite-new branch of hte org git repo, and will likely be merged into the master branch pretty soon.
- at present, and likely for some time, the best citation processor for html export is defined in `oc-csl.el` and relies on the citeproc-el library. My code will only have an effect if the org file contains lines like
--------------
#+bibliography: food-studies.bib
#+cite_export: csl "/home/matt/src/styles/chicago-author-date-mod.csl"
----------------
- oc-csl.el relies on citeproc-el to actually process the citations.
- citeproc-el is a fairly complex package with many moving parts. It supports processing of citations in html and latex backends, but not in any others (I think this is right!)
The latter is what we need to change for our export -- we want to change just one of the lambda functions used by the formatter for the duration of export, without updating the global value of citeproc-fmt--formatters-alist.
Because all these variables are lexically bound -- making the resultant closure values pretty confusing for me to inspect -- and because the process of creating a formatter is a bit complex, I was a little confused about how best to modify the code, and I'm still not quite sure how to go about it. It's pretty easy to do this using the dash library:
-----
(let ((org-re-reveal-citeproc-fmt-alist
(--map-when (eq (car it) 'cited-item-no)
`(cited-item-no . ,(lambda (x y) (concat "<a href="" x "</a>")))
citeproc-fmt--html-alist)))
[do some exporting stuff])
-------
or just
-----
(defcustom org-re-reveal-citeproc-fmt-alist
(--map-when (eq (car it) 'cited-item-no)
`(cited-item-no . ,(lambda (x y) (concat "<a href="" x "</a>")))
citeproc-fmt--html-alist)
:group org-re-reveal
:type alist)
------
But I'm not sure how best to do it without that.
In any case, the code I have provided here and in gitlab (
https://gitlab.com/oer/org-re-reveal/-/merge_requests/33) is quite hackish and doubtless likely to error out in many circumstances. Probably I should, in org-re-reveal-export-to-html, at least test if the function citeproc-formatter-fun-create is bound, so:
----------
(let ((citeproc-fmt--formatters-alist
(and (functionp 'citeproc-formatter-create)
`((html . ,(citeproc-formatter-create
:rt (citeproc-formatter-fun-create org-re-reveal-citeproc-fmt-alist)
:bib #'citeproc-fmt--html-bib-formatter))))))
....do stuff)
(actually, I simplified my merge request to use all this code).
Does that help the explanation? And what do you think?