Split vector in R every time vector element changes

Question

I need to split a vector of repeated groups of elements every time the element value changes. For example:

test_vector <- c("string1", "string1", "string1", "string2", 
  "string2", "string1", "string1", "string3")

must become:

$`1`
[1] "string1" "string1" "string1"

$`2`
[1] "string2" "string2"

$`3`
[1] "string1" "string1"

$`4`
[1] "string3"

If I try split(test_vector, test_vector) I get the wrong output:

$string1
[1] "string1" "string1" "string1" "string1" "string1"

$string2
[1] "string2" "string2"

$string3
[1] "string3"

I wrote some code which achieves this but it seems unnecessarily long and I feel like I'm missing something out there that's much simpler:

# find indices where splitting will occur:
split_points <- rep(F, length(test_vector))
for (i in 1:length(test_vector)) {
  if (i != 1) {
    if (test_vector[i] != test_vector[i-1]) {
      split_points[i] <- T
    }
  }
}
split_points <- c(1, which(split_points))

# create split vector:
split_code <- rep(1, length(test_vector))
for ( j in 1:length(split_points) ) {

  if (j!=length(split_points)) {
    split_code[
      split_points[j]:(split_points[j+1]-1)
    ] <- j
  } else {
    split_code[
      split_points[j]:length(test_vector)
    ] <- j
  }

}

split_result <- split(test_vector, split_code)
$`1`
[1] "string1" "string1" "string1"

$`2`
[1] "string2" "string2"

$`3`
[1] "string1" "string1"

$`4`
[1] "string3"

If anyone could help me find a simpler solution this would be much appreciated!

Answer 1

In base R , we can use rle to get the run-length-encoding of the vector

grp <- with(rle(test_vector), rep(seq_along(values), lengths))

Use that to split the vector

split(test_vector, grp)

---

With data.table , rleid gives the id based on the difference between adjacent elements

library(data.table)
split(test_vector, rleid(test_vector))

Answer 2

f = cumsum(c(TRUE, test_vector[-length(test_vector)] != test_vector[-1]))
split(test_vector, f)

OR

with(rle(test_vector), Map(rep, values, lengths))

Answer 3

A base R option is to use findInterval + cumsum + rle , ie,

res <- split(test_vector,
             findInterval(seq_along(test_vector),
                          cumsum(rle(test_vector)$lengths),
                          left.open = TRUE))

such that

> res
$`1`
[1] "string1" "string1" "string1"

$`2`
[1] "string2" "string2"

$`3`
[1] "string1" "string1"

$`4`
[1] "string3"