LaTeX input proposal & discussion.

[argent0]

General considerations

Current nested figures handling.

This is for pandoc 2.14.0.2

The current pandoc version can already handle nested figures in some way. Considering the input:

fig.tex

\begin{figure}
  \begin{subfigure}[b]{0.5\textwidth}
    \begin{subfigure}[b]{0.5\textwidth}
      \centering
      \includegraphics{test/media/rId25.jpg}
      \caption{Inner 1.1}
    \end{subfigure}
    \begin{subfigure}[b]{0.5\textwidth}
      \centering
      \includegraphics{test/media/rId25.jpg}
      \caption{Inner 1.2}
    \end{subfigure}
    \caption{Inner 1}
    \label{fig:inner1}
  \end{subfigure}
  \begin{subfigure}[b]{0.5\textwidth}
    \includegraphics{test/media/rId25.jpg}
    \caption{$Inner 2$}
    \label{fig:inner2}
  \end{subfigure}
  \caption{Outer caption}
  \label{fig:outer}
\end{figure}

It produces:

$ pandoc -f latex -t native fig.tex
[Para [Image ("",[],[]) [Str "Inner",Space,Str "1.1"] ("test/media/rId25.jpg","fig:")]
,Para [Image ("",[],[]) [Str "Inner",Space,Str "1.2"] ("test/media/rId25.jpg","fig:")]
,Para [Image ("fig:inner2",[],[]) [Math InlineMath "Inner 2"] ("test/media/rId25.jpg","fig:")]]

This is, all figures get flattened into a sequence.

This means that:

1. The subfigure environment is already handled by pandoc.
2. Reading the subfigure environment doesn't mean requiring the subcaption LaTeX package.

One issue with figures in general

Currently pandoc has issues with the parsing of figures in general. That is not necessarily nested. This is illustrated in the following example:

fig-simple.tex

\begin{document}

  \begin{figure}
    \includegraphics{test/media/rId25.jpg}
    \caption{CAP}
    \label{LAB}
  \end{figure}


  \begin{figure}
    \label{LAB}
    \includegraphics{test/media/rId25.jpg}
    \caption{CAP}
  \end{figure}


  \begin{figure}
    \includegraphics{test/media/rId25.jpg}
    \label{LAB}
    \caption{CAP}
  \end{figure}

\end{document}

All those figures render the same with pdflatex. So, one would expect all figures to produce the same AST but the actual output is:

$ pandoc -f latex -t native fig-simple.tex

[Para [Image ("LAB",[],[]) [Str "CAP"] ("test/media/rId25.jpg","fig:")]
,Para [Span ("LAB",[],[("label","LAB")]) [Str "[LAB]"],SoftBreak,Image ("LAB",[],[]) [Str "CAP"] ("test/media/rId25.jpg","fig:")]
,Para [Image ("LAB",[],[]) [Str "CAP"] ("test/media/rId25.jpg","fig:"),SoftBreak,Span ("LAB",[],[("label","LAB")]) [Str "[LAB]"]]]

Notice the SoftBreak elements generated for the second and third figures. As far as I can tell, this depends on the order of the label control sequence inside the figure environment.

$ pandoc -f latex -t html fig-simple.tex
<figure>
<img src="test/media/rId25.jpg" id="LAB" alt="CAP" /><figcaption aria-hidden="true">CAP</figcaption>
</figure>
<p><span id="LAB" label="LAB">[LAB]</span> <img src="test/media/rId25.jpg" title="fig:" id="LAB" alt="CAP" /></p>
<p><img src="test/media/rId25.jpg" title="fig:" id="LAB" alt="CAP" /> <span id="LAB" label="LAB">[LAB]</span></p>

New native figures & LaTeX

I've repurposed the native_figure extensions to work with LaTex.

Some issues

I have similar issues as before though. For example, fig-simple.tex generates:

$ pandoc -f latex+native_figures -t native fig-simple.tex
[Figure ("LAB",[],[]) (Caption Nothing [Plain [Str "CAP"]]) [Plain [Image ("",[],[]) [] ("test/media/rId25.jpg","")]]
,Figure ("LAB",[],[]) (Caption Nothing [Plain [Str "CAP"]]) [Plain [Image ("",[],[]) [] ("test/media/rId25.jpg","")]]
,Figure ("LAB",[],[]) (Caption Nothing [Plain [Str "CAP"]]) [Para [Image ("",[],[]) [Str "image"] ("test/media/rId25.jpg",""),SoftBreak,Span ("LAB",[],[("label","LAB")]) [Str "[LAB]"]]]]

Which in html translates to:

$ pandoc -f latex+native_figures -t html fig-simple.tex
<figure id="LAB">
<img src="test/media/rId25.jpg" />
<figcaption>CAP</figcaption>
</figure>
<figure id="LAB">
<img src="test/media/rId25.jpg" />
<figcaption>CAP</figcaption>
</figure>
<figure id="LAB">
<p><img src="test/media/rId25.jpg" alt="image" /> <span id="LAB" label="LAB">[LAB]</span></p>
<figcaption>CAP</figcaption>
</figure>

Notice the <span> in for the third figure.

Nested figures

For the nested figure case (fig.tex), I get:

$ pandoc -f latex+native_figures -t native fig.tex
[Figure ("fig:outer",[],[]) (Caption Nothing [Plain [Str "Outer",Space,Str "caption"]])
	[ Figure ("fig:inner1",[],[]) (Caption Nothing [Plain [Str "Inner",Space,Str "1"]])
		[ Figure ("",[],[]) (Caption Nothing [Plain [Str "Inner",Space,Str "1.1"]])
			[Plain [Image ("",[],[]) [] ("test/media/rId25.jpg","")]]
		, Figure ("",[],[]) (Caption Nothing [Plain [Str "Inner",Space,Str "1.2"]])
			[Plain [Image ("",[],[]) [] ("test/media/rId25.jpg","")]]]
	, Figure ("fig:inner2",[],[]) (Caption Nothing [Plain [Math InlineMath "Inner 2"]])
		[ Plain [Image ("",[],[]) [] ("test/media/rId25.jpg","")]]]]

For HTML output this becomes:

$ pandoc -f latex+native_figures -t html fig.tex
<figure id="fig:outer">
	<figure id="fig:inner1">
		<figure>
			<img src="test/media/rId25.jpg" />
			<figcaption>Inner 1.1</figcaption>
		</figure>
		<figure>
			<img src="test/media/rId25.jpg" />
			<figcaption>Inner 1.2</figcaption>
		</figure>
		<figcaption>Inner 1</figcaption>
	</figure>
	<figure id="fig:inner2">
		<img src="test/media/rId25.jpg" />
		<figcaption><span class="math inline"><em>I</em><em>n</em><em>n</em><em>e</em><em>r</em>2</span></figcaption>
	</figure>
	<figcaption>Outer caption</figcaption>
</figure>

Closing

Any idea about what's happening with the reader is wellcome. As usual, the code can be found here