Is it possible to deactivate figure handling from quarto ?

[FabienSe]

Hello !

I would like to use the extension native_numbering from pandoc to have native numbering of figure and table in a docx file which is generated thanks to quarto.
Sadly it is not working well with how is handling cross-reference and figure by quarto.
I would love to let pandoc handles the management of the figure but keep the amazing features (generation from python) provided by quarto on top of pandoc.

So is it possible to configure quarto to not do extra work on figure? This way the native_numbering extension can work.

Thanks a lot !

[cscheid]

I believe that's not possible today. With that said, we are currently reworking the crossref system. We'll keep this in mind and see if there's a simple way to disable the processing that doesn't impact the rest of our pipeline.

See my response below. This should be supported on quarto. Alas, that doesn't quite appear to work. Back to this being a feature request...

[cscheid]

Do note that you'll currently need to do some extra processing to "correct" the lack of crossref processing. For example, this is the output you'll get:

---
title: test
crossref: false
---

This is a @crossref test.

```{r}
#| label: fig-test
#| caption: This is a test figure.
plot(1:20)
```

So figures with captions are not processed correctly, etc.

[FabienSe]

Ha nice Thanks. I put this setting in my document.
I tried to render my doc with : quarto render doc.qmd --to docx+native_numbering

The pandoc's extension native_numbering does not seems to work even with no crossref processing. :(

Or may be it is not the way to pass this kind of parameter to pandoc though quarto ?

[cscheid]

Your quarto render command looks correct to me. It's likely that other parts of Quarto's filter chain interfere with native_numbering, unfortunately.

[cderv]

@cscheid as I suspected we could have an issue with variants, I checked something.

I believe we could have an issue in how we parse format and triage between supported Markdown variants and what we call modifiers (e.g pdf+draft and pdf+final).

This is what we get when running the command above

> quarto render doc.qmd --to docx+native_numbering

processing file: doc.qmd

output file: doc.knit.md

pandoc
  to: docx
  output-file: doc.docx
  default-image-extension: png

metadata
  title: test

Output created: doc.docx

No native_numbering in the to field for pandoc

From debugging within VSCode it seems we are seing docx+native_numbering not as the +native_numbering variants and then build the wrong format string.

Talking about this part of the code
https://github.com/quarto-dev/quarto-cli/blob/main/src/core/pandoc/pandoc-formats.ts#L171-L216

called
https://github.com/quarto-dev/quarto-cli/blob/main/src/core/pandoc/pandoc-formats.ts#L171-L12

I feel something is not right there but not sure as it seems the list to detect variants was built on the Markdown format supported ones.
https://github.com/quarto-dev/quarto-cli/blob/main/src/core/pandoc/pandoc-formats.ts#L218

@dragonstyle knows better this part of the code. If this is confirmed, we should transfer to an issue probably

Note that doing

format: docx
variant: +native_numbering

seems to correctly pass the variant as quarto command prints

pandoc 
  to: docx+native_numbering

but we document variant only for markdown output.

Hope it helps

[FabienSe]

Thanks a lot for your answer.

Ok +native_numbering is not working with the internals of the filter from quarto. I tried it with you proposition.
My original thought was to use the native numbering to have a list of figures.
I found another way to generate it in the docs document.

I created a custom filter which insert a TOC which filter on the style of the captions.
I found this idea in another thread of discussion : iandol/scrivomatic#48 (comment)

Thanks for this one @iandol and @iusgit.