R中的函数，用于验证data.frame上特定列的存在

Question

I'd like to validate that a data.frame contains columns with specific names. Ideally this would be a utility function that I can just pass the data.frame and expected column names and the function will raise an error if the data.frame does not contain the expected columns. I have written my own function below, however, this seems like something that would already exist in the R ecosystem.

My questions are:

1. Does such a function (or one-liner) already exist either in base R or in a common package?
2. If not, any suggestions for my function (below)?

Example of the function I have written to do this:

validate_df_columns <- function(df, columns) {
    chr_df <- deparse(substitute(df))
    chr_columns <- paste(columns, collapse = ", ")
    if (!('data.frame' %in% class(df))) {
        stop(paste("Argument", df, "must be a data.frame."))
    }
    if (sum(colnames(df) %in% columns) != length(columns)) {
        stop(paste(chr_df, "must contain the columns", chr_columns))
    }
}

validate_df_columns(data.frame(a=1:3, b=4:6), c("a", "b", "c'"))
## Error in validate_df_columns(data.frame(a = 1:3, b = 4:6), c("a", "b",  : 
##   data.frame(a = 1:3, b = 4:6) must contain the columns a, b, c'

Answer 1

The packages tibble and rlang , part of tidyverse have a function to check this :

library(tibble) # or library(rlang) or library(tidyverse)
has_name(iris, c("Species","potatoe"))
# [1]  TRUE FALSE

Technically it lives in rlang and its code is just :

function (x, name) 
{
    name %in% names2(x)
}

where rlang::names2 is an enhanced version of base::names which returns a vector of empty strings rather than NULL when the object doesn't have names.

Here's a way to rewrite your function :

validate_df_columns <- function(df, columns){
if (!is.data.frame(df)) {
    stop(paste("Argument", deparse(substitute(df)), "must be a data.frame."))
}
  if(!all(i <- rlang::has_name(df,columns)))
    stop(sprintf(
      "%s doesn't contain: %s",
      deparse(substitute(df)),
      paste(columns[!i], collapse=", ")))
}

validate_df_columns(iris, c("Species","potatoe","banana"))
# Error in validate_df_columns(iris, c("Species", "potatoe", "banana")) : 
# iris doesn't contain: potatoe, banana

Using deparse(substitute(...)) here makes little sense to me though, as it's not used interactively, clearer in my opinion to just say "df" .

Answer 2

The %in% operator works with pairs of vectors, so there is already a one-liner we can use here. Consider:

df <- data.frame(a=c(1:3), b=c(4:6), c=c(7:9))
names <- c("a", "c", "blah", "doh")
names[names %in% names(df)]

[1] "a" "c"

If you want to assert that the data frame contains all the input names, then just use:

length(names %in% names(df)) == length(names)     # to check all inputs are present
length(names %in% names(df)) == length(names(df)) # to check that input matches df