Split string based on alternating character in R

Question

I\'m trying to figure out an efficient way to go about splitting a string like

\"111110000011110000111000\"

into a vector

[1] \         


        

Answer 1

Original Approach: Here is a stringi approach that incorporates rle().

x <- "111110000011110000111000"
library(stringi)

cs <- cumsum(
    rle(stri_split_boundaries(x, type = "character")[[1L]])$lengths
)
stri_sub(x, c(1L, head(cs + 1L, -1L)), cs)
# [1] "11111" "00000" "1111"  "0000"  "111"   "000"  

Or, you can use the length argument in stri_sub()

rl <- rle(stri_split_boundaries(x, type = "character")[[1L]])
with(rl, {
    stri_sub(x, c(1L, head(cumsum(lengths) + 1L, -1L)), length = lengths)
})
# [1] "11111" "00000" "1111"  "0000"  "111"   "000"  

Updated for Efficiency: After realizing that base::strsplit() is faster than stringi::stri_split_boundaries(), here is a more efficient version of my previous answer using only base functions.

set.seed(24)
x3 <- stri_rand_strings(1L, 1e6L)

system.time({
    cs <- cumsum(rle(strsplit(x3, NULL)[[1L]])[[1L]])
    substring(x3, c(1L, head(cs + 1L, -1L)), cs)
})
#   user  system elapsed 
#  0.686   0.012   0.697