Use quarto just for code execution

[dmca-glasgow]

Hi there 👋,

I'm hoping to integrate Quarto into my own project, which is aimed at producing more accessible STEM coursework.

I'm wondering if it's possible to use Quarto solely for the code-execution aspect.

EDIT: completely rewriting my question for clarity

For example, if I have the following code:

```{r}
a <- c(1, 4, 2)
a
```

| Col1 | Col2 | Col3 |
|------|------|------|
| A    | B    | C    |
| E    | F    | G    |
| A    | G    | G    |

: First Table {#tbl-first}

See @tbl-first for details.

I am hoping for the following result:

``` r
a <- c(1, 4, 2)
a
```

[1] 1 4 2

| Col1 | Col2 | Col3 |
|------|------|------|
| A    | B    | C    |
| E    | F    | G    |
| A    | G    | G    |

: First Table {#tbl-first}

See @tbl-first for details.

Where the code block is executed, but the Markdown is left as is.

My reason for this is my project has it's own LaTeX-to-Markdown and Markdown-to-accessible-HTML processors, and I'm hoping to use Quarto along with those, for those who need it.

Sidenote: for bonus points, it would also love the ability to wrap the code output in some way, for example:

``` r
a <- c(1, 4, 2)
a
```

:::code-output
[1] 1 4 2
:::

Thanks.

[mcanouil]

This is (mostly) Pandoc behaviour for GFM format. You can use gfm-raw_html variant to strip raw HTML.
Or you can use the other Pandoc Markdown variants/formats.

[dmca-glasgow]

Thanks for the response @mcanouil, so there's no option which will bypass Pandoc, and just give me back the same Markdown but with the code blocks executed and inlined?

[mcanouil]

Why would a change to the format lead Quarto to lose its main feature? Did you try?

[dmca-glasgow]

Hi @mcanouil, my question was badly written in a rush yesterday. I have rewritten it. I hope it's clearer now. Thanks.

[mcanouil]

You could target the intermediate markdown keep-md: true but Quarto does not control what is emitted by code.
So if a code emits HTML, you'll end up with HTML.

Beyond that, you have access to all Markdown variants and extensions from Pandoc.

Note that all code is transformed to Pandoc AST (the schematic representation of a document), so it has to be written back and this uses opinionated writers.

About your new "sidenote", write your own Lua filter to do that or use engine specific features:

---
title: "Untitled"
format:
  commonmark: default
---

```{r}
#| class-output: "my-output"
head(mtcars)
```

Some **markdown** text and an inline equation $y = mx + c$.

But again, there is no garanties that the markdown will be left untouched if it does not follow the writer's standard.

[dmca-glasgow]

Thanks for your help @mcanouil, unfortunately it seems like Quarto isn't the right fit for my project. I previously used knitr directly for this purpose but it also had it's own problems. Maybe I will try using Jupyter directly. Thanks again.

[dmca-glasgow]

In case it may help someone else, @mcanouil suggestion of the keep-md: true option is basically exactly what I was hoping for!

I thought all Markdown was parsed into Pandoc's AST before the code execution and inlining phase, but it turns out not to be the case, here is the intermediate file's contents:

---
title: "My Document"
execute:
  keep-md: true
---


::: {.cell}

```{.r .cell-code}
a <- c(1, 4, 2)
a
```

::: {.cell-output .cell-output-stdout}

```
[1] 1 4 2
```
:::
:::

| Col1 | Col2 | Col3 |
|------|------|------|
| A    | B    | C    |
| E    | F    | G    |
| A    | G    | G    |

: First Table {#tbl-first}

See @tbl-first for details.

Also note that the output format can be gfm or commonmark or indeed html or probably anything, as this file is generated before being passed to Pandoc 🚀

[cderv]

A note that the keep-md output is mainly provided for debugging purposes. The intermediate markdown in it will contain some markup that are meant to be handled by Quarto Lua processing layer.

If the .qmd input is simple, this will probably get you an ok output. But if the .qmd contains some Quarto features, the keep-md will have some mark up that are not in the source.

If one desires Markdown output, then using format: markdown or other markdown formats + variants should be the way to go. You could even write a Quarto extension for your custom format that produce the markdown you expect before your processing.

For example, this is how Hugo support has been provided in Quarto: https://quarto.org/docs/output-formats/hugo.html

Anyhow, if this work for you - great! Just be aware of what you are using while quarto keeps evolving.

I previously used knitr directly for this purpose but it also had it's own problems

Quarto uses knitr internally which some tweaked configurations, and hooks to allow quarto feature. I don't know the problem you faced, but tweaking knitr is also a possibility if you only target R cells.

[dmca-glasgow]

Thanks for the input @cderv. What I am trying to achieve, through Test Driven Development, while I build out my accessibility tool, is to write unit tests for LaTeX syntax that is tested as valid against pdflatex, then the expected converted Markdown, tested against Quarto as valid syntax, then the accessible HTML, tested against AxE/pa11y/others. There will be users writing LaTeX as input and others writing (Q)Markdown input.

I am following the Quarto docs under "authoring": https://quarto.org/docs/authoring/markdown-basics.html

Converting LaTeX to Markdown, I will follow this as the reference Markdown syntax to generate for example, referencing a figure, and all the rest. But I need this syntax to be the output of Quarto markdown not the input, in this case, because I will be advising Stats users to first use Quarto for code execution, then run through my tool for Markdown to accessible HTML.

If there are issues with the intermediate file, they should come up in the next few days as I build out my unit tests. If I find anything specific I'll ask here! Thanks.

[mcanouil]

For reference:

• Is it possible to avoid HTML blocks around references in `format: gfm`? #408