Please add a "type.convert" argument to tstrsplit()

[rmscriven]

In relation to the post

http://stackoverflow.com/questions/29268941/how-to-split-a-string-variable-in-variables-in-r-non-rectangular/29268972#29268972

it would be convenient and useful if there was a type.convert argument to the new tstrsplit() function, for obvious reasons like easy conversion of numerics as characters to actual numerics after splitting.

I was thinking something as simple as

tstrsplit <- function(..., fill = NA, type.convert = TRUE) {
    x <- transpose(strsplit(...), fill = fill, ignore.empty = FALSE)
    if(type.convert) lapply(x, type.convert, as.is = TRUE) else x
}

x <- c("", "1-7-9", "3", "2-4-6-8")
tstrsplit(x, "-", fixed = TRUE)
# [[1]]
# [1] NA  1  3  2
# 
# [[2]]
# [1] NA  7 NA  4
# 
# [[3]]
# [1] NA  9 NA  6
# 
# [[4]]
# [1] NA NA NA  8

But I'm not sure how your internals work and might be better-off done internally. Of course, other things to think about are the other arguments to type.convert(), but they aren't available in read.table() so that might not matter.

[arunsrinivasan]

Also convert from factor to character by default as spotted by @rmscriven here.

[DavidArenburg]

Also maybe add simplify argument too in case you want a matrix back instead of a list (though it's not a problem when within the data.table parent frame)