In separate_wider_delim(), allow for specifying the direction to split by delim in

[olivroy]

There is no way with separate wider delim to use the spaces from the end. The docs say to use separate_wider_regex(), but I wonder if the delim function could add an option.

Of course, a workaround is to construct a more complex regex and to use separate_wider_regex(), but that could become clunky?

 dplyr::tibble(
  col = c(
    "A label with space 1.0 2.31",
    "Non-space label 1.22 3.21"
  )
) |> 
  tidyr::separate_wider_delim(
    cols = col,
    delim = " ",
    names = c("label", "n1", "n2"),
    too_many = "merge"
  )
#> # A tibble: 2 × 3
#>   label     n1    n2                 
#>   <chr>     <chr> <chr>              
#> 1 A         label with space 1.0 2.31
#> 2 Non-space label 1.22 3.21

^{Created on 2024-05-01 with reprex v2.1.0}

[DavisVaughan]

Seems like a reasonable feature request, another idea for now is to double reverse the string

dplyr::tibble(
  col = c(
    "A label with space 1.0 2.31",
    "Non-space label 1.22 3.21"
  )
) |>
  dplyr::mutate(col = stringi::stri_reverse(col)) |>
  tidyr::separate_wider_delim(
    cols = col,
    delim = " ",
    names = c("n1", "n2", "label"),
    too_many = "merge"
  ) |>
    dplyr::mutate(dplyr::across(c(n1, n2, label), stringi::stri_reverse))
#> # A tibble: 2 × 3
#>   n1    n2    label             
#>   <chr> <chr> <chr>             
#> 1 2.31  1.0   A label with space
#> 2 3.21  1.22  Non-space label

^{Created on 2024-10-24 with reprex v2.1.1}

[DavisVaughan]

I don't think it would go in too_many though, I think it would be a new argument to separate_wider_delim() like direction = c("forward", "backward") that specifies the direction to split by the delim in

---

Need to keep in mind that delim is allowed to be stringr::regex() and friends, which may complicate things.