Re: [PATCH][oc-csl] Improve reference parsing

[Ihor Radchenko]

[ Adding Org ML back to CC ]

András Simonyi <andras.simonyi@gmail.com> writes:

> On Thu, 19 Jan 2023 at 10:56, Ihor Radchenko <yantar92@posteo.net> wrote:
>> So, do I understand correctly that italics, bold, subscript,
>> superscript, small-caps, and nocase must be passed to the CSL processor
>> in a format understood by CSL? Everything else could just be left in Org
>> and later exported according to actual export settings?
>
> Unfortunately, the situation is a bit different --  the solution I see
> the most viable is to export the affixes and the locator to a form in
> which
> the markup elements you listed (plus links, which I haven't mentioned
> but are also supported) are in the required CSL
> input form, but the rest is in plain text.

What do you mean by plain text? Plain text as in Org or plain text as in
ASCII export?

> Anything else would be way
> more complicated to handle in Citeproc and I don't
> really see the benefits/use-case either (we are talking about elements
> within a citation). This would "only" require a custom backend
> exporting CSL-supported elements in the html-like CSL format and
> everything else which is allowed by the Org syntax
> in plain text.

I am not much concerned about CSL format itself.
I am concerned about the formatted citation returned back to Org by CSL.

Consider the following:

[cite:Prefix *bold* +strike-through+ @key]

It will be interpreted by Org export as

(citation
    (:style nil ...)
    (citation-reference
     (:key "key"... :prefix
           ("Prefix "
            (bold
             (... :post-blank 1 ...)
             "bold")
            (strike-through
             (... :post-blank 1 ...)
             "strike-through")))))

Now, consider that the user has a custom export filter that decorates
"+strike-through+" like "!!strike-through!!" upon export.

If we pass the original citation to the CSL, will the export filter be
applied?

Also, what if user decorates a CSL locator with Org markup like
strike-through?
             
>> May we:
>> 1. Convert the Org markup supported by CSL into CSL-understood HTML
>> format
>> 2. Convert all other Org markup into verbatim
>
> I'm not sure what you mean by verbatim -- leaving it as Org markup?

Whatever prevents CSL from altering the text. (Like escaping "_" you
mentioned earlier)

>> 3. Convert back non-verbatim markup altered by CSL into Org
>> 4. Perform exporting Org->current export backend as usual.
>
> If verbatim is Org then step 3 could be rather complicated, we'd need
> to identify
> the Org fragments in citeproc's HTML and LaTeX output when those
> backends are used.

But can't CSL output in Org format? Isn't the whole CSL thing supposed
to work for arbitrary export backend, not just HTML and LaTeX?

> Also I'd worry that the result would not pass through Citeproc's
> post-processing steps --
> there is now a user-customizable hook variable for citation post-processing
> which acts on the internal representations.

I envision the conversion back to Org to happen after _all_ the
Citeproc's processing, be it user-customized or not.

> All in all I'd first concentrate on the use-case: is there anything
> important left out
> if we go with simply using a custom backend to export the CSL-supported markup
> in CSL input format and everything else as plain text, then do what we
> do know, namely
> either simply insert the Citeproc-formatted output into the exported document
> without any post-processing (currently this is for LaTeX and HTML), or
> parse and export
> with Org when the Org Citeproc formatter is used (currently for all
> other formats),

What I imagine is doing "parse and export with Org" all the time,
including HTML and LaTeX export.

-- 
Ihor Radchenko // yantar92,
Org mode contributor,
Learn more about Org mode at <https://orgmode.org/>.
Support Org development at <https://liberapay.com/org-mode>,
or support my work at <https://liberapay.com/yantar92>