How to detect if data.frame is grouped by dplyr from subfunction? - r

Question

I have an R package where some functions are designed to be typically called within dplyr functions mutate or summarize.
newdata <- dplyr::mutate(group_by(olddata, col1), newcol = myfunc(col1))
However, sometimes users might forget to group their data before putting it into the mutate or summarize call.
newdata <- dplyr::mutate(olddata, newcol = myfunc(col1))
When the data frame is not grouped first, the package functions will produce largely nonsensical results. However, there won't be any errors or warnings per se, which could leave users uncertain about the cause of the issue.
I'd like to add a Warning() within the myfunc code when myfunc detects that the input data isn't coming from a grouped data.frame. However, I can't figure out how myfunc could detect if the data is coming from a grouped data.frame. It appears that mutate only passes a vector to myfunc, so both dplyr::is.grouped_df and inherits(x, "grouped_df") return false.
What I would like:
myfunc <- function(x) {if(comes.from.grouped.df) {print("grouped")} else {print("ungrouped")}}
mutate(olddata, newcol = myfunc(col1))
'ungrouped'
mutate(group_by(olddata, col1), newcol = myfunc(col1))
'grouped'
'grouped'
'grouped'

Answer 1

If you want your function used within a specific context, and emit a warning if the data frame is not grouped, then you can do:
library(tidyverse)
myfunc <- function(x) {
if(all(ls(envir = parent.frame()) == "~")) {
ss <- sys.status()
funcs <- sapply(ss$sys.calls, function(x) deparse(as.list(x)[[1]]))
wf <- which(funcs == "mutate")
if(length(wf) == 0) stop("`myfunc` must be called from inside `mutate`")
wf <- max(wf)
data <- eval(substitute(.data), ss$sys.frames[[wf]])
if(!inherits(data, "grouped_df")) {
warning("`myfunc` called on an ungrouped data frame / tibble.")
}
return(x^2)
}
stop("`myfunc` must be called from inside `mutate`")
}
Used outside mutate, we get an error:
myfunc(1:10)
#> Error in myfunc(1:10): `myfunc` must be called from inside `mutate`
With an ungrouped data frame or tibble we get a warning:
tibble(iris) %>%
mutate(x = myfunc(Sepal.Length))
#> Warning in myfunc(Sepal.Length): `myfunc` called on an ungrouped data frame /
#> tibble.
#> # A tibble: 150 x 6
#> Sepal.Length Sepal.Width Petal.Length Petal.Width Species x
#> <dbl> <dbl> <dbl> <dbl> <fct> <dbl>
#> 1 5.1 3.5 1.4 0.2 setosa 26.0
#> 2 4.9 3 1.4 0.2 setosa 24.0
#> 3 4.7 3.2 1.3 0.2 setosa 22.1
#> 4 4.6 3.1 1.5 0.2 setosa 21.2
#> 5 5 3.6 1.4 0.2 setosa 25
#> 6 5.4 3.9 1.7 0.4 setosa 29.2
#> 7 4.6 3.4 1.4 0.3 setosa 21.2
#> 8 5 3.4 1.5 0.2 setosa 25
#> 9 4.4 2.9 1.4 0.2 setosa 19.4
#> 10 4.9 3.1 1.5 0.1 setosa 24.0
#> # ... with 140 more rows
And it runs without complaint if the tibble is grouped:
tibble(iris) %>%
group_by(Species) %>%
mutate(x = myfunc(Sepal.Length))
#> # A tibble: 150 x 6
#> # Groups: Species [3]
#> Sepal.Length Sepal.Width Petal.Length Petal.Width Species x
#> <dbl> <dbl> <dbl> <dbl> <fct> <dbl>
#> 1 5.1 3.5 1.4 0.2 setosa 26.0
#> 2 4.9 3 1.4 0.2 setosa 24.0
#> 3 4.7 3.2 1.3 0.2 setosa 22.1
#> 4 4.6 3.1 1.5 0.2 setosa 21.2
#> 5 5 3.6 1.4 0.2 setosa 25
#> 6 5.4 3.9 1.7 0.4 setosa 29.2
#> 7 4.6 3.4 1.4 0.3 setosa 21.2
#> 8 5 3.4 1.5 0.2 setosa 25
#> 9 4.4 2.9 1.4 0.2 setosa 19.4
#> 10 4.9 3.1 1.5 0.1 setosa 24.0
#> # ... with 140 more rows
Created on 2023-02-15 with reprex v2.0.2