How to paste parameters in R functions using `substitute` and `eval` to make data.table work like dplyr?

Question

I have a working R function written as:

library(data.table)
mutate_dt = function (.data, ..., by) 
{
    dt = as.data.table(.data)
    substitute(dt[, `:=`(...), by][]) %>% eval()
}

However, I could not use this function in other functions. For example:

t_f = function(dt,x){
  mutate_dt(dt,a = x)
}

t_f(iris,1)
#> Error in eval(jsub, SDenv, parent.frame()): object 'x' not found

Is there any way that I can make it work?

Answer 1

You need to ensure that the expression is evaluated in the correct environment. R makes it easy by providing the convenience function eval.parent:

library(data.table)
mutate_dt = function (.data, ..., by) 
{
  dt = as.data.table(.data)
  eval.parent(substitute(dt[, `:=`(...), by][]))
}

mutate_dt(iris,a = 1)
#works

t_f = function(dt,x){
  mutate_dt(dt,a = x)
}

t_f(iris,2)
#works