`dplyr::mutate()` but only if the provided columns are new in the data frame

tidyverse
1

I want to use dplyr::mutate() to only update provided columns that are new in the data frame:

mutate_if_missing(
  cars,
  speed = "don't overwrite", # Existing column
  length = "new value"       # New column
)
#>   speed dist    length
#> 1     4    2 new value
#> 2     4   10 new value
#> 3     7    4 new value
#> 4     7   22 new value
#> 5     8   16 new value
#> 6     9   10 new value

I didn't find a way to do that with dplyr verbs, so I created a helper function mutate_if_missing(), which works fine:

library(dplyr)
library(rlang)

# Helper function
mutate_if_missing <- function(.data, ...) {
  # Convert ... to list
  args <- rlang::list2(...)
  # Select arguments that are new columns
  new_columns <- args[!names(args) %in% colnames(.data)]
  # Splice the filtered arguments into dplyr::mutate()
  dplyr::mutate(.data, !!!new_columns)
}

mutate_if_missing(
  head(cars),
  speed = "don't overwrite",
  length = "new value"
)
#>   speed dist    length
#> 1     4    2 new value
#> 2     4   10 new value
#> 3     7    4 new value
#> 4     7   22 new value
#> 5     8   16 new value
#> 6     9   10 new value

Created on 2025-05-24 with reprex v2.1.1

The function fails however, when the provided arguments use .data:

library(dplyr)
library(rlang)

# Helper function
mutate_if_missing <- function(.data, ...) {
  # Convert ... to list
  args <- rlang::list2(...)
  # Select arguments that are new columns
  new_columns <- args[!names(args) %in% colnames(.data)]
  # Splice the filtered arguments into dplyr::mutate()
  dplyr::mutate(.data, !!!new_columns)
}

mutate_if_missing(
  head(cars),
  length = dplyr::if_else(.data$speed > 4, "> 4", "<= 4")
)
#> Error:
#> ! Can't subset `.data` outside of a data mask context.

# Works fine with dplyr::mutate()
mutate(
  head(cars),
  length = dplyr::if_else(.data$speed > 4, "> 4", "<= 4")
)
#>   speed dist length
#> 1     4    2   <= 4
#> 2     4   10   <= 4
#> 3     7    4  > 4
#> 4     7   22   > 4
#> 5     8   16    > 4
#> 6     9   10    > 4

Created on 2025-05-24 with reprex v2.1.1

Any suggestions on how to resolve this? Approaches using just dplyr verbs that that don't require the helper function would be even better.

2

Found it, this can be solved by using rlang::quos() rather than rlang::list2():

library(dplyr)
library(rlang)

# Helper function
mutate_if_missing <- function(.data, ...) {
  # Convert ... to list
  args <- rlang::quos(...)
  # Select arguments that are new columns
  new_columns <- args[!names(args) %in% colnames(.data)]
  # Splice the filtered arguments into dplyr::mutate()
  dplyr::mutate(.data, !!!new_columns)
}

mutate_if_missing(
  head(cars),
  length = dplyr::if_else(.data$speed > 4, "> 4", "<= 4")
)
#>   speed dist length
#> 1     4    2   <= 4
#> 2     4   10   <= 4
#> 3     7    4    > 4
#> 4     7   22    > 4
#> 5     8   16    > 4
#> 6     9   10    > 4

Created on 2025-05-24 with reprex v2.1.1

1 Like