How to make tidyr::complete work when nesting the same column twice?

For some reason, my code repeats a column's name inside complete (it's inside a function and the argumet may be a column already present).

library(tidyverse)
# Works
mtcars %>% complete(nesting(cyl, gear), vs)

# Doesn't work
mtcars %>% complete(nesting(cyl, cyl), vs)

# Error in `dplyr::full_join()`:
# ! Join columns in `y` must be present in the data.
# ✖ Problem with `cyl...1` and `cyl...2`.

I don't want the column cyl to be repeated in the result.

How to make this work elegantly?

Edit: If the idea is to remove duplicated column names in nesting, you could use your own function:

my_nesting <- function(...){
  nesting(..., .name_repair = "minimal") %>%
    subset(select = !duplicated(colnames(.)))
}

mtcars %>% complete(my_nesting(cyl = cyl, cyl = cyl), vs) 
#Works

Why is it not working?

_{Not really an answer, but this is too long to be a comment.}

In your example case, I'm not sure what's the point of repeating the column twice, since the combination of two identical columns is the same as the combination of one of the two (check below).

with the default behavior

Under the hood, complete uses full_join and expand. expand works with duplicate names:

data = mtcars
out <- expand(data, nesting(cyl, cyl), vs)

But full_join does not because names have changed:

dplyr::full_join(out, data, by = names(out))

Indeed, the default behavior of nesting (and expand) when there are identical names is .name_repair = "check_unique":

with(mtcars, nesting(cyl, cyl)) 

#   cyl...1 cyl...2
# 1       4       4
# 2       6       6
# 3       8       8

with .name_repair = "minimal"

If you specify .name_repair = "minimal" on top of overwriting the columns name with identical names, you'll have the same name:

with(mtcars, nesting(cyl = cyl, cyl = cyl, .name_repair = "minimal"))

#     cyl   cyl
# 1     4     4
# 2     6     6
# 3     8     8

But, here full_join is blocking the process again, because names must be unique:

out <- expand(data, nesting(cyl = cyl, cyl = cyl, .name_repair = "minimal"), vs, .name_repair = "minimal")
dplyr::full_join(out, data, by = names(out))

#Error in `dplyr::full_join()`:
#! Input columns in `x` must be unique.
#✖ Problem with `cyl`.

So, what can be done?

• Not so much: there does not seem to be an apparent reason to nest identical columns since the unique combinations will be the same.
• You could create other variables in your original data.frame to match the column names created through the process.

You may replace nesting with select. select does not allow duplicate columns so this would solve your issue.

library(dplyr)
library(tidyr)

mtcars %>% complete(select(., cyl, gear), vs)

# A tibble: 38 × 11
#     cyl  gear    vs   mpg  disp    hp  drat    wt  qsec    am  carb
#   <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
# 1     4     3     0  NA    NA      NA NA    NA     NA      NA    NA
# 2     4     3     1  21.5 120.     97  3.7   2.46  20.0     0     1
# 3     4     4     0  NA    NA      NA NA    NA     NA      NA    NA
# 4     4     4     1  22.8 108      93  3.85  2.32  18.6     1     1
# 5     4     4     1  24.4 147.     62  3.69  3.19  20       0     2
# 6     4     4     1  22.8 141.     95  3.92  3.15  22.9     0     2
# 7     4     4     1  32.4  78.7    66  4.08  2.2   19.5     1     1
# 8     4     4     1  30.4  75.7    52  4.93  1.62  18.5     1     2
# 9     4     4     1  33.9  71.1    65  4.22  1.84  19.9     1     1
#10     4     4     1  27.3  79      66  4.08  1.94  18.9     1     1
# ℹ 28 more rows
# ℹ Use `print(n = ...)` to see more rows

Comparing the result of select with nesting to make sure they give identical results.

identical(mtcars %>% complete(nesting(cyl, gear), vs), 
          mtcars %>% complete(select(., cyl, gear), vs))
#[1] TRUE

identical(mtcars %>% complete(nesting(cyl), vs),
          mtcars %>% complete(select(., cyl, cyl), vs))
#[1] TRUE