`separate_longer_delim` errors with "can't recycle input of size X to size Y,"

[llrs]

In some cases separate_longer_delim doesn't work.

I initially thought that this was due to a different number of values inside each column but in the examples it combines it correctly.
It doesn't seem to come from the delimiter, as it happens both with "," and " ", the order of the rows doesn't affect it too (having first a single value and then one to split).

library("tidyr")
packageVersion("tidyr")
#> [1] '1.3.0'
dfr <- data.frame(a = "a,b,c", b = "A,B")
separate_longer_delim(dfr, c(a, b), delim = ",")
#> Error in `separate_longer_delim()`:
#> ! In row 1, can't recycle input of size 3 to size 2.
#> Backtrace:
#>     ▆
#>  1. └─tidyr::separate_longer_delim(dfr, c(a, b), delim = ",")
#>  2.   └─tidyr:::map_unchop(...)
#>  3.     └─tidyr::unchop(...)
#>  4.       └─tidyr:::df_unchop(...)
#>  5.         └─purrr::reduce(x_sizes, unchop_sizes2, error_call = error_call)
#>  6.           └─purrr:::reduce_impl(.x, .f, ..., .init = .init, .dir = .dir)
#>  7.             └─tidyr (local) fn(out, elt, ...)
#>  8.               └─cli::cli_abort(...)
#>  9.                 └─rlang::abort(...)
dfr2 <- data.frame(a = "a b c", b = "A B")
separate_longer_delim(dfr2, c(a, b), delim = " ")
#> Error in `separate_longer_delim()`:
#> ! In row 1, can't recycle input of size 3 to size 2.
#> Backtrace:
#>     ▆
#>  1. └─tidyr::separate_longer_delim(dfr2, c(a, b), delim = " ")
#>  2.   └─tidyr:::map_unchop(...)
#>  3.     └─tidyr::unchop(...)
#>  4.       └─tidyr:::df_unchop(...)
#>  5.         └─purrr::reduce(x_sizes, unchop_sizes2, error_call = error_call)
#>  6.           └─purrr:::reduce_impl(.x, .f, ..., .init = .init, .dir = .dir)
#>  7.             └─tidyr (local) fn(out, elt, ...)
#>  8.               └─cli::cli_abort(...)
#>  9.                 └─rlang::abort(...)

# From the examples works
df <- tibble(id = 1:3, x = c("x", "x y", "x y z"), y = c("a", "a b", "a b c"))
separate_longer_delim(df, c(x, y), delim = " ")
#> # A tibble: 6 × 3
#>      id x     y    
#>   <int> <chr> <chr>
#> 1     1 x     a    
#> 2     2 x     a    
#> 3     2 y     b    
#> 4     3 x     a    
#> 5     3 y     b    
#> 6     3 z     c
# But not when there is only one row
df2 <- tibble(id = 1, x = "x y z", y = "a b")
separate_longer_delim(df2, c(x, y), delim = " ")
#> Error in `separate_longer_delim()`:
#> ! In row 1, can't recycle input of size 3 to size 2.
#> Backtrace:
#>     ▆
#>  1. └─tidyr::separate_longer_delim(df2, c(x, y), delim = " ")
#>  2.   └─tidyr:::map_unchop(...)
#>  3.     └─tidyr::unchop(...)
#>  4.       └─tidyr:::df_unchop(...)
#>  5.         └─purrr::reduce(x_sizes, unchop_sizes2, error_call = error_call)
#>  6.           └─purrr:::reduce_impl(.x, .f, ..., .init = .init, .dir = .dir)
#>  7.             └─tidyr (local) fn(out, elt, ...)
#>  8.               └─cli::cli_abort(...)
#>  9.                 └─rlang::abort(...)

^{Created on 2023-06-28 with reprex v2.0.2}

[llrs]

I think the problem is with map_unchop as stringr::str_split(dfr2$a, stringr::fixed(" ")) works correctly but somehow separate_longer_delim(dfr2, c(a, b), delim = "\\ ") doesn't produce an error.

[hadley]

In the examples, the number of elements in each row is the same in both columns. In your example, the number of elements is different across the columns. I'm not sure what you'd expect the output to look like in this case.

[victorrssx]

@hadley I think this could be an enhancement. I understand how the function works up to this point (i.e., in the case of two columns the number of elements must be equal across them). But in some cases you may be looking for all the combinations between the elements in each row where this condition does not hold. Currently, the expected output would be achieved by running separate_longer_delim twice. With an improvement the current behavior would become a particular case of a general one.

tibble(id = 1, x = "x y z", y = "a b") %>% 
separate_longer_delim(x, delim = " ") %>% 
separate_longer_delim(y, delim = " ")

# A tibble: 6 × 3
     id x     y    
  <dbl> <chr> <chr>
1     1 x     a    
2     1 x     b    
3     1 y     a    
4     1 y     b    
5     1 z     a    
6     1 z     b