1

I'm trying to evaluate a string as a formula:

In dplyr it would look like this: 

dt = data.table(a = 1:10)
expr = 'sum(a)'

dt %>% 
  mutate(b := !!parse_expr(expr))

However when I try with data.table I'm getting an error: 

dt[, b := parse_expr(expr)]

Error in [.data.table(dt, , :=(b, parse_expr(expr))) : RHS of assignment is not NULL, not an an atomic vector (see ?is.atomic) and not a list column.

Rafael
  • 2,762
  • 1
  • 18
  • 55

1 Answers1

1

Instead of parse_expr, eval(parse can be used

dt[, b := eval(parse(text = expr))]

Or wrap with eval on parse_expr as the !! is doing the evaluation in tidyverse

dt[, b := eval(rlang::parse_expr(expr)) ]
akrun
  • 789,025
  • 32
  • 460
  • 575
  • @R.M. With `rlang` and tidyverse, they use a consistent pattern with `!!`, but that wouldn't work outside the tidyverse env or else you may need `as_mapper` – akrun Nov 22 '19 at 19:26