The previous wc -m didn't handle multibyte characters that crossed a buffer boundary, so take a guess at making that work. (I haven't got a test case for this. I also don't know how to handle invalid sequences so just don't count them.)
1 file changed