topntailvettorso.R

torso <- function(d, n=5L, ...) {
	stopifnot(length(n) == 1L)
	center <- as.integer(nrow(d)/2)
	left <- as.integer(center-n/3)
	right <- as.integer(center + n/3)
	d[seq.int(from= left, to = right, by = 1L),]
}

topntail <- function(d, n=4L) {
	rbind(head(d,n), tail(d,n))
}

vet <- function(d, n=4L) {
	rbind(head(d,n), torso(d,n), tail(d,n))
}

torso(iris)
topntail(iris)
vet(iris)

Nice idea. This might work a little better for the torso function:

torso <- function(d, n=5L) {
    stopifnot(length(n) == 1L)
    center <- as.integer(nrow(d)/2)
    left <- as.integer(center - n/2)
    d[ seq_len(n) + left - 1L, ]
}

Using seq_len negates the need for calculating right, and ensures you'll always get the right number of rows (notice how torso(iris,5) currently give 4 rows instead of the 5 requested). Also n/2 seems more appropriate for calculating left than n/3. Was there a reason for the 3 there?

In fact, depending on your taste, it might be worth simplifying a little more by skipping the center assignment too:

torso <- function(d, n=5L) {
    stopifnot(length(n) == 1L)
    left <- as.integer(nrow(d)/2 - n/2)
    d[ seq_len(n) + left - 1L, ]
}