I put together a sugar function trimws to mirror base::trimws (added in R 3.2.0). For those not familiar with this, it removes either leading (which = "left"), trailing (which = "right"), or leading and trailing (which = "both") whitespace (defined as any of ' ', '\t', '\r', or '\n') from strings, returning NA on NA input.

In my fork, usage looks like this:

#include <Rcpp.h>
using namespace Rcpp;

// [[Rcpp::export]]
CharacterVector trim(CharacterVector x, const char* which = "both")
{
    return trimws(x, which);
}

/*** R

trim("  x y z \t \n \r   ")
# [1] "x y z"

trim("  x y z \t \n \r   ", "left")
# [1] "x y z \t \n \r   "

trim("  x y z \t \n \r   ", "right")
# [1] "  x y z"

*/

Other features:

• Default which argument is "both", as in base::trimws
• Only the first letter of which is checked ('b', 'l', and 'r' are valid first letters) to mimic the use of match.arg in R
• It is also overloaded for the CharacterMatrix and String classes
• Much faster than the base version, which uses regex to accomplish this

If this seems like a worthwhile addition, I can submit a PR shortly.