Pandoc: LaTex/Beamer add hook before \begin{frame}... and after \end{frame}

[NMarkgraf]

I think it would be a great idea to add the possibility to call a hook macro in latex/beamer right before and after the creation of a frame.

This would make it possible to translate something like

## Exercise: a new quiz {hook=exercisemode}

Just an  example ...

## No exercise 

Nothing else

to an latex output with might change the behavior frame title styling etc:

\exercisemodetrue
\begin{frame}
\frametitle{%
\protect\hypertarget{exercise-a-new-quiz}{%
 Exercise: a new quiz }}

Just an example ...
\end{frame}
\exercisemodefalse

\begin{frame}
\frametitle{%
\protect\hypertarget{o-exercise}{%
 No exercise  }}

Nothing else ...
\end{frame}

I need something like this desperately to change the frame title style of a slide when it shows exercises..

Thanks
Norman

[cagix]

Do you really need to enclose the complete frame in the exercise mode, i.e. do you need \exercisemodetrue to appear before \begin{frame} and \exercisemodefalse to appear after \end{frame}?

If it would suffice to wrap just the content of your frame into the exercise mode, you could use a simple lua-filter like

local exercise = false

function Header(h)
    local exercisemode = pandoc.Null()

    -- handle end of previous excercise mode
    if exercise then
        exercise = false
        exercisemode = pandoc.RawBlock("latex", "\\exercisemodefalse")
    end

    -- handle current header: start exercise if attribut `hook` is set to `exercisemode`
    if h.attributes["hook"] == "exercisemode" then
        exercise = true
        return { exercisemode, h, pandoc.RawBlock("latex", "\\exercisemodetrue") }
    end

    -- else return header
    return { exercisemode, h }
end

This filter would produce the following TeX from your example (using --slide-level=2):

\begin{frame}{%
\protect\hypertarget{exercise-a-new-quiz}{%
Exercise: a new quiz}}

\exercisemodetrue

Just an example \ldots{}

\exercisemodefalse

\end{frame}

\begin{frame}{%
\protect\hypertarget{no-exercise}{%
No exercise}}

Nothing else

\end{frame}

[mb21]

If it would suffice to wrap just the content of your frame into the exercise mode

then you could also you write the raw TeX (i.e. \exercisemodetrue) as part of the markdown

[NMarkgraf]

If you write raw TeX like @mb21 suggested, this raw TeX would be part of the frame before. Because every Header of a certain level is acting greedy until the next Header of the same or higher level. So the raw TeX-Block in the AST would be part of the last Header, which means that it will be written inside of the frame before the current frame. And that is not a solution, because you need it between \end{frame} and \begin{frame}.

So that is not the right way to fix the problem.

(Yes, I do think this is a design problem of the current AST-design! ;-) )

That is also true for (any) filter suggestion, because setting anything inside of a frame won't change the current frame (and also not the next frame!) or anything inside of the frame title environment. If you write \begin{frame} everything inside will be like in a sandbox and don't harm the frame code.

So, change the behavior inside of the pandoc latex-writer is, for me, the only possible solution. -- Sorry.

Thanks to @mb21 and @cagix, but it your ideas won't work .. :-(

[jgm]

Currently we reorganize the AST for slides in the writer,
so the filter can't get to the slide structure.

Now that we have native Divs, we could consider reorganizing
things:

• the beamer, reveal.js, etc. writers would expect things to
  be already organized into a serious of Divs with class
  "slide", corresponding to the slides.

• we'd have a built-in filter that creates this Div
  structure from the regular AST in the same way as
  currently

• the writer would check to see if the AST includes the
  Div structure for slides, and call this filter if not

• but the filter could also be called manually from e.g.
  a lua filter; this would allow people to do fine-grained
  adjustments on slide structure from within a lua filter.

• writers could also just include the Div structure for
  slides manually.

It's a bit complex but worth considering.

[NMarkgraf]

I'm not sure about if this would make it easier. Complex, may be to complicated, if you would ask me.
If I understand it correctly

something like

# section
## slide
Lore ipsum ...

## slide
Lore ipsum ...

will first be converted to something like

[Div class="section" content="section",
  [Div class="slide" content="Lore Ipsum ..."],
  [Div class="slide" content="Lore ipsum ..."]
]

so that it will be possible to add something in between like

[Div class="section" content="section",
  [RawBlock ...],
  [Div class="slide" content="Lore Ipsum ..."],
  [Div class="slide" content="Lore ipsum ..."],
  [RawBlock ...]
]

or do I misunderstood something?

[jgm]

Yes, you understand the idea.

[NMarkgraf]

Okay, that would be perfect. - When it will arrive in the releases? ;-) --- Yes, i'm pushing, 'cause I need it badly! :-)

[NMarkgraf]

Sorry,

is there anything I could test or use in this direction? -- Maybe in a branch? ;-)

Yours
Norman

[jgm]

Sorry...I have limited time to work on pandoc; you shouldn't expect any
changes soon.