Quarto Website Filters?

[jlgraves-ubc]

Hi all,

I'm trying to make a filter (i.e. an extension) that adds some custom elements to the navbar in Quarto.

However, I'm struggling since I don't seem to be able to get Pandoc/Lua to identify the div elements or ul elements that contain the navbar as Div elements in the AST - they are being identified as RawBlock elements. This means I can't do anything to them.

Is there something special about the website project time that might be doing this? That I could adjust or change?

Thanks!

[dragonstyle]

Hi - sorry for the trouble! We add the navbar in a post processor, so unfortunately it isn't visible in LUA AST. There isn't a great way to get at it aside from javascript right now, though in the long run we would like to give you access to post processors as well (not sure when we'll get there).

In the meantime is there something specific we should consider adding to navbars to support your scenario more directly?

[jlgraves-ubc]

It's not "better" in an absolute sense, but it does do basically everything JupyterBook does, at the cost of a little more overhead.

• Quarto is more flexible than JupyterBook, allowing more types of sites with different features and themes, but requires more configuration which takes time to set up and tweak.
• JupyterBook is more tightly integrated with JupyterNotebook formats, which allows it to do some things that Quarto currently doesn't do (see the above discussion), but on the other hand it means that it's less capable with other formats and relies heavily on kernels for rendering and output.
• Quarto supports better cross-referencing, LaTeX output, and rendering targets, but is slower in general to render large projects.
• JupyterBook is better integrated with conda making it easier to deploy on a custom server as a website, but Quarto offers more robust CLI options for non-interactive deployment.

For me, I think Quarto is better for my purposes because:

1. Working with a large team with .ipynb format objects is not feasible on a Git system, due to the way the format stores metadata per cell. JupyterNotebooks are an output format, not an input format for large projects.
2. Flexibility and user interface is key, and Quarto's ability to adapt is much better; it's basically half-way between a standard Notebook and a full-blown static site generator, which is kind of perfect for me.
3. LaTeX and BibTeX support with multiple render formats is essential for teaching and research; I can write nice-looking slides in RevealJS, and render handouts in LaTeX, along with textbook chapters or articles all from the same content sources without multiple programs or even significantly different code.

[jav-ed]

Thank you very much for your quick and detailed reply. I am also missing the feature of downloading .ipynb files directly from the webpage view and having access to the navigation bar.

[jlgraves-ubc]

Yeah - you can see the workaround we use to deploy on a JupyterHub on the website here:

https://comet.arts.ubc.ca/

It's not well-documented in the examples on the Quarto website, but you can also render to .ipynb if you have a .qmd format of a file, which can give you some options by using custom _includes in headers and stuff to pull in those files with a contextual link, but it will require some lua coding most likely or Javascript.

[jav-ed]

Thank you, Jonathan :)

[andrewpbray]

@jlgraves-ubc Thanks for the thorough breakdown of the distinctions as you see them - very helpful!