No figure caption with hugo-md

[dmenne]

Bug description

Format hugo-md does not create captions, only alt-text

Steps to reproduce

---
title: Hugo Captions
format: 
  html:
    fig-height: 3
  hugo-md:
    fig-height: 2.5
---


```{r}
#| label: fig-hugo
#| echo: FALSE
#| fig-cap: "This should be a numbered caption"
plot(rnorm(100))
```

With format: html, there is a numbered caption.

```
<figure class="quarto-float quarto-float-fig figure">
<div aria-describedby="fig-hugo-caption-0ceaefa1-69ba-4598-a22c-09a6ac19f8ca">
<img src="hugo_caption_files/figure-html/fig-hugo-1.png" class="img-fluid figure-img" width="672">
</div>
<figcaption class="quarto-float-caption-bottom quarto-float-caption quarto-float-fig" id="fig-hugo-caption-0ceaefa1-69ba-4598-a22c-09a6ac19f8ca">
Figure&nbsp;1: This should be a numbered caption
</figcaption>
</figure>
```

With format: hugo-md, there is no caption, only an alt-text.

```
<img
src="hugo_caption.markdown_strict_files/figure-markdown_strict/fig-hugo-1.png"
id="fig-hugo" alt="Figure 1: This should be a numbered caption" />`
```

Expected behavior

Hugo needs figure captions

Actual behavior

No figure captions

Your environment

No response

Quarto check output

Quarto 1.6.36

[dmenne]

It would be possible to delegate this to a Hugo-hook (https://sebastiandedeyne.com/captioned-images-with-markdown-render-hooks-in-hugo/), but why bother because html does a nice job.

[cderv]

I think this is something we may not be handling well for cross reference figure.

This is the markdown produced by knitr as intermediate for Quarto + Pandoc

![This should be a numbered caption](test.markdown_strict_files/figure-markdown_strict/hugo-1.png){#fig-hugo width=768}

If fig- label is not used then a caption is produced.

![This should be a numbered caption](test.markdown_strict_files/figure-markdown_strict/hugo-1.png){width=768}

leads to

<figure>
<img src="test.markdown_strict_files/figure-markdown_strict/hugo-1.png" width="768" alt="This should be a numbered caption" />
<figcaption aria-hidden="true">This should be a numbered caption</figcaption>
</figure>

So our processing is making the caption undealt with. Here is another reprex of what I think the problem is with cross ref figure for hugo-md

---
title: Hugo Captions
format: hugo-md
keep-md: true
---

::: {#fig-hugo}

![]({{< placeholder >}})

Caption
:::

This leads to only fig-alt.

<img
src="data:image/png;base64,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"
id="fig-hugo" alt="Figure 1: Caption" />

It is using alt and not caption. Hugo would support another syntax with HTML 5 and figcaption.

I believe this ends up like this because we do process Hugo markdown floatRefTarget specifically as Img

quarto-cli/src/resources/filters/layout/hugo.lua

Lines 80 to 96 in 178c75e

	local el = quarto.utils.match("Plain/[1]/Image")(float.content)
	if el then
	local text = pandoc.write(pandoc.Pandoc({
	pandoc.Image(
	pandoc.utils.stringify(quarto.utils.as_inlines(float.caption_long)),
	el.src,
	"",
	pandoc.Attr(float.identifier, {}, {}))}), "html")
	-- we use "html" here because otherwise
	-- Pandoc appears to not want to emit the right thing
	-- when the output is inside a table
	-- Pandoc is also emitting bizarre (but ultimately harmless) newline entities inside tables
	-- if we add the newline here, but if we don't add newlines,
	-- then images by themselves don't have line breaks after them
	local result = pandoc.RawInline("html", text .. "\n")
	return result
	end

and not Figure with pandoc.Figure() so no Caption added by Pandoc.

Here is waht Pandoc does

1. First is a figure, and HTML for figure is produced
2. Second is a image, and only HTML for image is produced

❯ quarto pandoc --to gfm
![Some Text](fig-1.png){#fig-hugo}

![Some Text 2](fig-2.png){#fig-hugo2}\
^Z
<figure id="fig-hugo">
<img src="fig-1.png" alt="Some Text" />
<figcaption aria-hidden="true">Some Text</figcaption>
</figure>

<img src="fig-2.png" id="fig-hugo2" alt="Some Text 2" />

This is recent parsing change in Pandoc 3 I believe

❯ quarto pandoc --to native
![Some Text](fig-1.png){#fig-hugo}

![Some Text 2](fig-2.png){#fig-hugo2}\
^Z
[ Figure
    ( "fig-hugo" , [] , [] )
    (Caption
       Nothing [ Plain [ Str "Some" , Space , Str "Text" ] ])
    [ Plain
        [ Image
            ( "" , [] , [] )
            [ Str "Some" , Space , Str "Text" ]
            ( "fig-1.png" , "" )
        ]
    ]
, Para
    [ Image
        ( "fig-hugo2" , [] , [] )
        [ Str "Some" , Space , Str "Text" , Space , Str "2" ]
        ( "fig-2.png" , "" )
    , LineBreak
    ]
]

So we could probably adapt Quarto output for Hugo, as floatreftarget should probably output as a figure with caption.