How to include R expressions in (quarto) YAML

using YAML with embedded R expressions for cleaner, reusable Quarto configs.

When writing R Markdown or Quarto files, I’ve always disliked cluttering them with long file paths, fixed parameters, or small snippets of code. Many of these values are constants or things I don’t plan to change often, sometimes not even across projects. So I started looking for a way to move such information into a separate YAML file and refer to it from within Quarto. You could of course do this with a regular R script and source(), but I found that YAML makes these kinds of project-wide parameters easier to organize and read.

Embedding R in YAML

While digging around, I came across this helpful StackOverflow post, which pointed out that YAML can evaluate R expressions directly if marked with !expr. This means your config file can contain values that are not just static strings, but dynamically computed.

For example, my _paper.yaml file looks like this:

experiments:
  first: 
    path: !expr 'paste0("C:/Users/", Sys.info()[6], "/OneDrive/Science/experiments/first")'
    nsubject: !expr 'nrow(read.csv(file.path(paste0("C:/Users/", Sys.info()[6], "/OneDrive/Science/experiments/first"), "results.csv")))'
  second: !expr 'paste0("C:/Users/", Sys.info()[6], "/OneDrive/Science/experiments/second")'
  third: ...

Reading it in R

To load the config file, you just need to enable expression evaluation when calling yaml::read_yaml():

config <- yaml::read_yaml('_paper.yaml', eval.expr = TRUE)

Now, config behaves like a nested list, and you can access parameters directly. For example:

config$experiments$first
config$experiments$first$nsubject

This makes it easy to centralize all project-level constants and keep your Quarto documents cleaner.

Why This Helps

• Cleaner documents: no repeated paths or magic numbers in your .qmd files.
• Reusability: the same YAML file can be shared across multiple projects.
• Flexibility: values can be static (like participant counts) or computed on the fly (like paths that depend on the username).

It might feel a little fragile, since I am evaluating R code hidden in a config file, but it’s been reliable and I am finding it quite useful. For instance, I sometimes even compute the number of participants automatically from an Ibex CSV and store it in the YAML—then refer to it in my Quarto text.