Re: Emphasis etc. in #+CAPTION:

[Carsten Dominik <carsten.dominik@gmail.com>]

On Mar 29, 2010, at 3:14 AM, Nick Dokos wrote:

> Thomas S. Dye <tsd@tsdye.com> wrote:
>
>> I see in the Org Manual that:
>>
>>    You can make words *bold*, /italic/, _underlined_, =code= and
>>    ~verbatim~, and, if you must, ‘+strike-through+’. Text in the  
>> code
>>    and verbatim string is not processed for Org-mode specific syntax,
>>    it is exported verbatim.
>>
>> So a caption like this doesn't export italicized a, b, c, and d
>> (presumably because this is a code string):
>
> Hi Tom,
>
> IIUC, it's not because it's a code string but simply because the  
> caption
> is not processed for emphasis (or anything else for that matter - see
> below). I think the above text in the manual covers the generic case,
> but captions were probably added afterwards and they didn't get the
> emphasis treatment.
>
>>
>>    #+CAPTION: Pumice collected from pāhoehoe flows at Manini`ōwali:
>>    /a/, pieces that sink in fresh water; /b/, pieces without obvious
>>    crystallization or coating (Category A); /c/, pieces with obvious
>>    crystallization but no coating (Category B); /d/, pieces with
>>    obvious coating (Category C).  The scale is 1 cm.
>>
>> Is there a common work-around for this that will produce correct  
>> LaTeX
>> and HTML using the org-mode exporters?
>>
>
> Can't think of any workaround within org-mode. Post-processing the
> output to replace /a/ with \emph{a} for LaTeX and <i>a</i> for HTML,
> perhaps restricted to captions, seems possible (but ugly).
>
>> And, this is just a question and not a feature request or a request
>> for changes to org-mode, would it be hard to change org-mode's
>> behavior so that markup like /a/ in the #+CAPTION: string exports
>> correctly marked-up text to LaTeX and HTML?
>>
>
> I think this is the way it works currently:
>
> in org-exp.el:org-export-attach-captions-and-attributes() sets  text  
> properties
> on the following line (the first line of a table or the line  
> containing a link)
> as follows:
>
>        ...
> 	(add-text-properties (point-at-bol) end
> 			     (list 'org-caption cap
> 				   'org-attributes attr
> 				   'org-label label))
>
> The various exporters then retrieve the caption with something like  
> the
> following code (this is one of the three calls in org-latex.el; the
> other calls in this exporter, and the other exporters as well, are
> similar):
>
>    (let*
>        ...
>        (caption (org-find-text-property-in-string 'org-caption raw- 
> path))
>        ...
>
> This just gets the value of the given text property without further
> elaboration. So I'd guess it's possible to change these calls so that
> these values are passed through various filters (emphasis is one of
> them, but maybe there are others?) before the exporters get them,  
> but it
> is not done today.
>
> But maybe I'm full of it, and somebody will come up with a better  
> idea.

Full of good analysis you are.  May the force be with you.

Well, it is just like Nick says.  I am grabbing the caption
early, before all that processing happens unfortunately.

And, unfortunately, the Org exporters are not modular enough
to easily fix this entirely.

What I *have* done now is implement basic processing, like emphasis.
But I am afraid that footnotes and link formatting, for example, do
not work.

Hope that make it good enough.

- Carsten