Capturing functions in R

Here’s a little code snippet used to capture a function and its arguments. A use-case is to define the function when building metacode.

The function get_function() takes the name of an R function that exists in memory and returns the function definition as it was be defined in code. Two trailing lines are removed: <environment> is the environment in which the function was defined, typically namespace:base for base R functions or namespace:pkg for a function from package pkg; <bytecode> indicates that the function has been compiled into bytecode.

get_function <- function(fn) {
  
  ## capture the function
  fn_txt <- capture.output(print(get(fn)))
  
  if (length(fn_txt) > 0) {
    
    fn_txt[1] <- paste(fn, "<-", fn_txt[1])
    
    if (grepl("<environment", fn_txt[length(fn_txt)])) {
      fn_txt <- fn_txt[-length(fn_txt)]
    }
    if (grepl("<bytecode", fn_txt[length(fn_txt)])) {
      fn_txt <- fn_txt[-length(fn_txt)]
    }
    fn_txt <- paste(fn_txt, collapse = "\n")
  }
  return(fn_txt)
  
}

Example

my_function <- function(x) {
  # return x + 1
  x + 1
}

cat(get_function("my_function"))

my_function <- function (x)
{
    # return x + 1
    x + 1
}

Why not use deparse?

deparse() and deparse() are base R functions that return unevaluated R expressions into strings. Passing a function name to deparse() leads to a similar output, however any comments are removed from the output.

fn_deparse <- deparse(my_function)
fn_deparse <- paste(fn_deparse, collapse = "\n")
cat(fn_deparse)

function (x) 
{
    x + 1
}