Quarto RevealJS presentations with figures generated from typst packages

[hros]

Description

I want to add figures of quantum circuits generated from the typst Quil package
It would be best to have the typst code for each circuit in a separate file, and have Quarto call typst to generate the figures, which could be used by the presentation(s) (perhaps the directory of the circuits and the generated figures would be a shared directory/repository for collaboration)

I went back-and-forth with chatGPT, but it is still "almost".... I'd appreciate advice on best-practices for this

[mcanouil]

What's the question?

Reveal.js is HTML.

If you want to include a figure, then generate that figure beforehand as there is no Typst engine at the moment.
You can use Typst embedded in Quarto with quarto typst.
You could also define a Lua filter to execute Typst on code cells and save the output as image then include it. Something like https://github.com/pandoc-ext/diagram

[hros]

Can you please elaborate on the first option:

You can use Typst embedded in Quarto with quarto typst

How would I combine the typst code for the figure with the presentation code?

[mcanouil]

I was referring to manual compilation: You write the image outside of Reveal.js and later on include it when compiled.

[cderv]

This reminds me of the tikz code cell engine in knitr: https://github.com/yihui/knitr-examples/blob/master/058-engine-tikz.Rmd
or the texPreview R package that bring a custom engine to compile some cells content to tex output: https://yonicd.github.io/texPreview/

So you could leverage Knitr's custom engine: https://bookdown.org/yihui/rmarkdown-cookbook/custom-engine.html
to have one that run typst on file content from a cell.

There is even in knitr an exec engine that could probably be leverage as-is: https://github.com/yihui/knitr-examples/blob/master/124-exec-engine.Rmd

Though, maybe you don't need an engine and you could use R package typr: https://christophertkenny.com/typr/
This would be a great place to have a typst knitr engine I guess.

Anyhow, this is quite specific questions and tricks and chatGPT won't know this, but if you read the docs linked you have some clues to create what you need I think.

Those are alternatives to use engine: knitr with Quarto, and leverage some knitr feature.

Using Lua filter as mentioned is another options, which is not computation engine specific.