Access full data in grouped dplyr summarize without relying on dplyr internals

user_9349302 0
Viewed 103

I have a function that is supposed to be used in a grouped dplyr::summarise(). I want to access both the data of each group as well as the full data. The former can easily be achieved with dplyr::across(). For the latter, I only see the option to rely on an internal function mask <- dplyr:::peek_mask() and then do mask$full_data().

My question: Is there any way to access the full data without relying on an internal function?

Below is a simple example, the actual function is rather complex, but in its essence it compares the grouped data to the full data.

Added for clarification: Ideally, the user can call this function foo and it produces the result inside dplyr::summarise without the need of specifying further arguments or adding calls.

Any help appreciated.

library(dplyr)

# function to be used inside summarize
foo <- function() {
  mask <- dplyr:::peek_mask() # replace this function without relying on dplyr internals
  full_data <- mask$full_data() 
  tibble(group_rows = nrow(across()),
         full_data = nrow(full_data))
}

mtcars %>% 
 group_by(gear) %>% 
 summarise(foo()) 
#> `summarise()` ungrouping output (override with `.groups` argument)
#> # A tibble: 3 x 3
#>    gear group_rows full_data
#>   <dbl>      <int>     <int>
#> 1     3         15        32
#> 2     4         12        32
#> 3     5          5        32

Created on 2020-12-09 by the reprex package (v0.3.0)

dplyr r
1 Answers

An option is cur_data

foo <- function() {
 
  tibble(group_rows = nrow(across()),
         full_data = nrow(cur_data())) 
        
}

mtcars %>% 
 group_by(gear) %>% 
 summarise(foo(), .groups = 'drop') %>%
 mutate(full_data  = sum(full_data))

-output

# A tibble: 3 x 3
#   gear group_rows full_data
#  <dbl>      <int>     <int>
#1     3         15        32
#2     4         12        32
#3     5          5        32
user_37322710
Related