Search code examples
rmacrosstringify

R macro: how to f(x) --> g("x")


How can I use defmacro (from gtools) to arrange it so that when I type f(x) the result is the return value of g("x") ?

I suppose (because I am a C programmer) that defmacro is involved... but I would be happy to achieve the goal with or without defmacro.

Unfortunately f <- defmacro(x, expr={ g(quote(x)) }) is hit and miss... it works for some functions, but fails for others, like for example

g <- function(v) {
  eval(parse(text=paste0("f0 = lm(Sale_Price ~ ", v, ", data = d1)")))
  r0 <- data.frame(predict(f0, d1), d1$Sale_Price)
  colnames(r0) <- c(v, "Sale_Price")
  ggplot() +
  labs(x= v, y= "Sale_Price") +
  geom_point(data = r0, aes(get(v), Sale_Price), colour="black") +
  geom_smooth(data = r0, method = "lm", aes(x = get(v), y Sale_Price), colour="blue", se = FALSE)
}

I can modify the above definition so that f <- defmacro(x, expr={ g(quote(x)) }) will succeed... but that is not my question. I want to know how in general it can be arranged so that when I type f(x) the result is the return value of g("x") for arbitrary user-defined g (but I would appreciate an answer which works with mild restrictions on g).


Solution

  • when I type f(x) the result is the return value of g("x") for arbitrary user-defined g

    f <- function(x, g) {
      g(as.character(substitute(x)))
    } 
    
    #using paste as an example of g    
    f(a, paste)
    #[1] "a"
    

    Do not use eval(parse()). Forget it exists until you are much more advanced in your knowledge of the language. You can do something like this:

    form <- as.formula(sprintf("Sale_Price ~ %s", v))
    f0 <- lm(form, data = d1)
    

    Also, study help("aes_string"). You don't need get.