How to edit an R markdown YAML header programmatically?

Question

Take for example this Rmd file - https://github.com/rstudio/learnr/blob/master/inst/tutorials/ex-setup-r/ex-setup-r.Rmd

The YAML header has this -

output:
  learnr::tutorial:
    progressive: true
    allow_skip: true

I would like to change this to -

output: 
  ioslides_presentation:
    widescreen: true

Is there a way to make this edit programmatically i.e. Can I write some function that takes an Rmd file as input, edits the YAML header, and produces a new Rmd file?

Thanks!

r2evans · Accepted Answer

I think a quick function could do this.

change_yaml_matter <- function(input_file, ..., output_file) {
  input_lines <- readLines(input_file)
  delimiters <- grep("^---\s*$", input_lines)
  if (!length(delimiters)) {
    stop("unable to find yaml delimiters")
  } else if (length(delimiters) == 1L) {
    if (delimiters[1] == 1L) {
      stop("cannot find second delimiter, first is on line 1")
    } else {
      # found just one set, assume it is *closing* the yaml matter;
      # fake a preceding line of delimiter
      delimiters <- c(0L, delimiters[1])
    }
  }
  delimiters <- delimiters[1:2]
  yaml_list <- yaml::yaml.load(input_lines[ (delimiters[1]+1):(delimiters[2]-1) ])

  dots <- list(...)
  yaml_list <- c(yaml_list[ setdiff(names(yaml_list), names(dots)) ], dots)

  output_lines <- c(
    if (delimiters[1] > 0) input_lines[1:(delimiters[1])],
    strsplit(yaml::as.yaml(yaml_list), "
")[[1]],
    input_lines[ -(1:(delimiters[2]-1)) ]
  )

  if (missing(output_file)) {
    return(output_lines)
  } else {
    writeLines(output_lines, con = output_file)
    return(invisible(output_lines))
  }
}

Where ... is whatever you want it to be. Meaning: if you want to replace the output: component of the yaml matter, then you give a named list as output=list(...).

If I use the rmarkdown document I used in a previous answer, then unchanged, it looks like this:

readLines("~/StackOverflow/1883604/62095186.Rmd")
#  [1] "---"                                              
#  [2] "title: Hello"                                     
#  [3] "output: html_document"                            
#  [4] "params:"                                          
#  [5] "  intab: TRUE"                                    
#  [6] "---"                                              
#  [7] ""                                                 
#  [8] "# Headline 1"                                     
#  [9] ""                                                 
# [10] "## Headline 2 `r if (params$intab) "{.tabset}"`"
# [11] ""                                                 
# [12] "### Headline 3 in a tab"                          
# [13] ""                                                 
# [14] "### Headline 4 in a tab"                          
# [15] ""                                                 
# [16] "### Headline 5 in a tab"                          
# [17] ""                                                 
# [18] ""

And to change the output portion, I add a nested named list as:

change_yaml_matter("~/StackOverflow/1883604/62095186.Rmd", 
                   output=list(ioslides_presentation=list(widescreen=TRUE)))
#  [1] "---"                                              
#  [2] "title: Hello"                                     
#  [3] "params:"                                          
#  [4] "  intab: yes"                                     
#  [5] "output:"                                          
#  [6] "  ioslides_presentation:"                         
#  [7] "    widescreen: yes"                              
#  [8] "---"                                              
#  [9] ""                                                 
# [10] "# Headline 1"                                     
# [11] ""                                                 
# [12] "## Headline 2 `r if (params$intab) "{.tabset}"`"
# [13] ""                                                 
# [14] "### Headline 3 in a tab"                          
# [15] ""                                                 
# [16] "### Headline 4 in a tab"                          
# [17] ""                                                 
# [18] "### Headline 5 in a tab"                          
# [19] ""                                                 
# [20] ""

You can change just about any portion of the yaml matter. (The only things you cannot change, I suspect, are if you happen to have yaml parameters named input_file or output_file. If you actually have Rmd files with those yaml top-level parameters, then you can easily rename the named arguments here to be something else, such as Mxyzptlk and something else ... you're unlikely to see those in production.)

Notes:

This did not save anything to a file, you have to do that yourself. Add output_file="path/to/new.RMd" to your call, and it will write a new file.
When you do include output_file= in the arguments, if you choose to not catch the return value, it will appear to return nothing. This is due to invisible in my return; if you really want to see and save, either capture to a variable and look at that, or wrap the function call in parens, as in (change_yaml_matter(...)).

The trick for YAML is to know that yaml:: will treat every top-level as the named element of a list, and its contents are recursively lists in the same manner. For instance,

str(yaml::yaml.load("
---
top1:
  level2a:
    level3a: 123
    level3b: 456
  level2b: 789
top2: quux
---"))
# List of 2
#  $ top1:List of 2
#   ..$ level2a:List of 2
#   .. ..$ level3a: int 123
#   .. ..$ level3b: int 456
#   ..$ level2b: int 789
#  $ top2: chr "quux"

To assign new values, just provide nested named lists.

How to edit an R markdown YAML header programmatically?

Answers (2)

Related Questions