Vectorised solution

reverse <- function(text) {
  strsplit(text, "") |>
    lapply(rev) |>
    sapply(paste, collapse = "")
}

The Exercism test suite only provides a single string in each test, expecting a single string to be returned. This is compatible with other language tracks, but conflicts with R's emphasis on vector processing.

Suppose you want to do this:

test_that("a vector", {
  text <- c("", "robot", "Ramen", "I'm hungry!", "racecar", "drawer")
  expected <- c("", "tobor", "nemaR", "!yrgnuh m'I", "racecar", "reward")
  expect_equal(reverse(text), expected)
})

In fact, this is perfectly normal in R. However, of the other approaches discussed, only stri_reverse() passes the vector test.

To generalize the other functions to handle vectors correctly, we need the lapply ("list apply") and sapply ("simplified apply") functions, as in the code at the top of the page.

Programmers unused to these functions may find the code difficult to understand. For clarity, we will break it down in stages, using intermediate variables that can be examined.

> text <- c("abc", "def")
> chars <- strsplit(text, "")
> chars
[[1]]
[1] "a" "b" "c"

[[2]]
[1] "d" "e" "f"

> typeof(chars)
[1] "list"

The strsplit() function always returns a list of vectors.

For the split-reverse approach using single string inputs, we extracted the vector at index 1 by appending [[1]]. THere is no need for that here: the list can be used directly by lapply().

> reversed <- lapply(chars, rev)
> reversed
[[1]]
[1] "c" "b" "a"

[[2]]
[1] "f" "e" "d"

The lapply() function always returns a list, which is next passed to sapply():

> result <- sapply(reversed, paste, collapse = "")
> result
[1] "cba" "fed"
> typeof(result)
[1] "character"

The sapply() function tries to simplify its output, rather than automatically returning a list like lapply(). In this case, it produces a vector of character strings, which is what we need.