Code Block Formatting/Fixers?

[acederberg]

Description

Hi everybody! Thank you again for this wonderful project.

Use Case

I was wondering if there is any known way to clean up poorly formatted documents. For instance, using injected formatters with conform.nvim I can get python, yaml, and other languages fixed in my notebook through nvim. I would like to run this on some files in bulk in github actions - however this might be quite tricky or impossible with nvim.

I would like to note that I do not think the quarto command itself should do this, but am asking if there is a solution that already exists and if not, how to proceed possibly implementing a solution.

As for my thoughts, I think it might be possible using pandoc filters and updating the files in place.

Apparently, prettier is capable of fixing code blocks in markdown however only js based plugins can be used for this.

What Would it Do?

This functionality could turn the following poorly formatted quarto document

---
title: Poorly formatted front matter
categories:
- poorly
- formated
---

blah blah blah

```{.python}
print( "poorly" , "formated"
)
```

into

---
title: Poorly formatted front matter
categories: [poorly, formated]
---

blah blah blah

```{python}
print("poorly", "formated")
```

when the user specifies to use black and yamlfix.

[mcanouil]

I would like to note that I do not think the quarto command itself should do this, but am asking if there is a solution that already exists and if not, how to proceed possibly implementing a solution.

That's very unlikely as Quarto supports any possible programming languages either via knitr or Jupyter kernels.
Quarto cannot possibly embed languages and formatter.
As a reminder, Quarto is language agnostic. Using R, Python, Julia, etc. is a user action so is the formatting.

This being said, you can build your own system as some users already did, see https://github.com/shafayetShafee/black-formatter for Black (Python).

Any formatter and their settings is opinionated so it's best you do what you think is best in that regard.

[acederberg]

Thanks!

This being said, you can build your own system as some users already did, see https://github.com/shafayetShafee/black-formatter for Black (Python).

Something like that would be cool. I think I might implement a pandoc filter that:

1. Can be configured to use certain tools on certain languages and do nothing on others, e.g. only use yamlfix on yaml codeblocks.
2. Configuration of the particular formatter should be left to the user and not configured in the filter configuration itself. E.g. in pyproject.toml for yamlfix and black.
3. Outputs the reformatted code into the respective qmd file instead of outputting to html so that source files are fixed and not just the output.

One final question. To make this work, is it possible to load a qmd file and output it to a qmd file with pandoc? I tried

quarto pandoc -o ./tables.qmd ./tables.qmd

and got

[WARNING] Could not deduce format from file extension .qmd
  Defaulting to markdown
[WARNING] Could not deduce format from file extension .qmd
  Defaulting to html

[mcanouil]

qmd is "markdown with code cells (computation)".
If Pandoc was able to understand that there would not be a Quarto😅
Anyhow Quarto is "markdown" as far as Pandoc is concerned.

In the case of formatters, you usually want a replacement in place.

[acederberg]

If Pandoc was able to understand that there would not be a Quarto😅
Anyhow Quarto is "markdown" as far as Pandoc is concerned.

True that. I will just let pandoc treat qmd files the same as markdown. I will see what I can cook up and share if it is worth while.

[cscheid]

@acederberg These are deep waters, just FYI. You'll need to use a custom Pandoc reader like we do, otherwise Pandoc will not recognize code blocks that use the {...} language syntax.