I want a function taking an unevaluated expression containing variable names into a new expression containing string for the variable names.

c(a,b)

should be converted to

c("a","b")

How can this be done?

Let me explain in more detail what I want to achieve.

my_tibble <-
  tibble(a=rep(c("a","b"),each=4),
              b=rep(rep(c("x","y"),each=2),2),
              x=seq_along(b)
  )

summary_groupwise <- function(data,group_var,var_to_summarise,result_var){
  data |>
    group_by({{group_var}}) |>
    summarise({{result_var}} := sum({{var_to_summarise}}))
}

This function works

summary_groupwise(my_tibble,a,x,group_sum)

Now I want to group by multiple variables

summary_groupwise_multiple <- function(data,group_vars,var_to_summarise,result_var){
  data |>
    group_by(across(all_of({{group_vars}}))) |>
    summarise({{result_var}} := sum({{var_to_summarise}}))
}

This works:

summary_groupwise_multiple(my_tibble,c("a","b"),x,group_sums)

To make summary_groupwise_multiple work, the names of the grouping variables have to be given as strings.

For consistency of the syntax, I would like the following call to work

summary_groupwise_multiple(my_tibble,c(a,b),x,group_sums)

How can is fiddle with the parameter group_vars so the call works without giving the names of the variables as strings.

CodePudding user response：

If the expressions are passed as a vector i.e. with c(), an option is

f1 <- function(expr) {dput(sapply(as.list(substitute(expr))[-1], deparse))}
> f1(c(a, b))
c("a", "b")

CodePudding user response：

fun <- function(x)dput(as.character(substitute(x)[-1]))

fun(c(a, b, c))
c("a", "b", "c")