Should num_range select helper work together with e.g. ends_with or get an additional argument "suffix = "?

[deschen1]

(note - copy from: tidyverse/dplyr#5764)

I was just puzzled when I tried to select certain columns of my data frame that a) start with a letter, followed by a number (for which the num_range helper would be ideal) and b) also end with a certain string, e.g. "_rec" (for which ends_with would be perfect).

However, using this approach, a data frame with 0 columns is returned. I guess that num_range is only suited for full matches, however the help function doesn't say a word about it and I also think num_range would benefit from an option to not only add a prefix, but also a suffix.

library(tidyverse)
# Data
df <- data.frame(x1 = c(1, 2),
                 x2 = c(3, 4),
                 x3 = c(5, 6),
                 x4y = c(7, 8),
                 x1_rec = c(9, 10),
                 x2_rec = c(11, 12),
                 x4y_rec = c(13, 14))

#   x1 x2 x3 x4y x1_rec x2_rec x4y_rec
# 1  1  3  5   7      9     11      13
# 2  2  4  6   8     10     12      14

# Code that should work, but doesn't
df %>% select(num_range("x", 1:2) & ends_with("_rec"))

# data frame with 0 columns and 2 rows

# Expected output
#   x1_rec x2_rec
# 1      9     11
# 2     10     12

What I could imagine is really just an extension of the num_range function, e.g.:

num_range(prefix = "x",
          range = 1:2,
          suffix = "_rec")

[lionel-]

Thanks for the suggestion. In this case I'd rather not add features to num_range() given that matches() provides enough flexibility.

[deschen1]

Understood, but in this case there's still the remaining bug (IMO) that the combination of num_range with other select helpers does not work as e.g. shown above with:

df %>% select(num_range("x", 1:2) & ends_with("_rec"))

So I'd hope for a respective re-opening of the issue.

Not sure though if that is a tidyselect problem or a dplyr/select problem.

[lionel-]

ok reopening so I can take a look at your reprex when I get back to work on tidyselect.

[hadley]

library(dplyr, warn.conflicts = FALSE)

df <- tibble(
  x1 = c(1, 2),
  x2 = c(3, 4),
  x3 = c(5, 6),
  x4y = c(7, 8),
  x1_rec = c(9, 10),
  x2_rec = c(11, 12),
  x4y_rec = c(13, 14)
)

df %>% select(num_range("x", 1:2)) |> names()
#> [1] "x1" "x2"
df %>% select(ends_with("_rec")) |> names()
#> [1] "x1_rec"  "x2_rec"  "x4y_rec"

^{Created on 2022-08-12 by the reprex package (v2.0.1)}

It's because num_range() only returns x1 and x2. I think it's probably fine to add a suffix argument to num_range().