HTML Input proposal & discussion

[argent0]

Hi, I've been working on reading figures from HTML input. The code can be found here. Details follow:

General

The HTML reader can now produce the Figure Block element. This can be used further down the line in filters and writers.

In order to keep the old behavior intact, I decide to implement the functionality through an extension native_figures. Like native_divs.

Pandoc 2.14.0.2 behavior

Current pandoc's figure tag handling can be divided in two cases:

Only img and figcaption tags are kept in the AST

$ pandoc -f html -t native
<figure class="important">
  <img src="../media/rId25.jpg" />
  <ul> <li> ITEM </li> </ul>
  <figcaption> CAP1 </figcaption>
</figure>
^D
[Para [Image ("",[],[]) [Str "CAP1"] ("../media/rId25.jpg","fig:")]]

$ pandoc -f html -t html
...
^D
<figure>
<img src="../media/rId25.jpg" alt="CAP1" /><figcaption aria-hidden="true">CAP1</figcaption>
</figure>

We see that img and figcaption are preserved but the ul is absent from the internal representation. This is the case for any other tag.

Only the first img and figcaption tags are kept in the AST

Of course then semantics of multiple figcaption tags could be considered questionable, but still:

$ pandoc -f html -t native
<figure class="important">
  <img src="../media/rId25.jpg" />
  <img src="../media/rId25.jpg" />
  <figcaption> CAP1 </figcaption>
  <figcaption> CAP2 </figcaption>
</figure>
^D
[Para [Image ("",[],[]) [Str "CAP1"] ("../media/rId25.jpg","fig:")]]

Other remarks

It's also worth mentioning that the class is missing from the output.

My Fork

Example Run

$ pandoc -f html+native_figures -t native
<figure class="important">

  <img src="../media/rId25.jpg" />

  <img src="../media/rId25.jpg" />

  <figcaption> CAP1 </figcaption>
  <figcaption> CAP2 </figcaption>
</figure>

[Figure ("",["important"],[]) (Caption Nothing [Plain [Str "CAP1"],Plain [Str "CAP2"]]) [Plain [Image ("",[],[]) [] ("../media/rId25.jpg",""),SoftBreak,Image ("",[],[]) [] ("../media/rId25.jpg","")]]]

$ pandoc -f html+native_figures -t html
...
<figure class="important">
<img src="../media/rId25.jpg" /> <img src="../media/rId25.jpg" />
<figcaption>CAP1
CAP2</figcaption>
</figure>

Other remarks

In this version all img and figcaption tags are included in the AST. And the round trip conversion is very similar.

<p><img .../></p> Loses its special meaning though.