Splice data.frame inputs to mutate, summarise, and transmute

[eutwt]

library(dplyr, warn.conflicts = FALSE)
library(dtplyr)

df <- tibble(x = 1)

df %>%
  mutate(data.frame(y = 2, z = 3))
#> # A tibble: 1 × 3
#>       x     y     z
#>   <dbl> <dbl> <dbl>
#> 1     1     2     3

lazy_dt(df) %>%
  mutate(data.frame(y = 2, z = 3))
#> Source: local data table [1 x 2]
#> Call:   copy(`_DT1`)[, `:=`(`data.frame(y = 2, z = 3)` = data.frame(y = 2, 
#>     z = 3))]
#> 
#>       x `data.frame(y = 2, z = 3)`
#>   <dbl> <list>                    
#> 1     1 <df [1 × 2]>              
#> 
#> # Use as.data.table()/as.data.frame()/as_tibble() to access results

^{Created on 2022-02-11 by the reprex package (v2.0.1)}

[markfairbanks]

Is this going to be possible in a lazy workflow if passed a data frame object from the environment?

Edit: As I think about this, all we would have to do is use exists()/env_has() and then check if an input is a data.frame.

library(dplyr, warn.conflicts = FALSE)

df <- tibble(x = 1)

new_cols_df <- data.frame(y = 2, z = 3)

df %>%
  mutate(new_cols_df)
#> # A tibble: 1 × 3
#>       x     y     z
#>   <dbl> <dbl> <dbl>
#> 1     1     2     3

[eutwt]

When I wrote the issue I was thinking about just checking for a data.frame/tibble/as.data.frame/as_tibble call, to allow for complex calculations inside brackets in mutate. I can't remember the example I ran into at the time where it's actually necessary/useful right now, but something like below. This is a pretty rare thing to do though, and if it is possible to reliably splice data frames not coming from an explicit in-verb call that'd be even better.

mtcars %>% 
  #lazy_dt() %>% 
  mutate({
    a <- mpg[1]
    b <- sqrt(a) + hp[4]
    data.frame(a, b)
  })