c81175f532
... and any other language where the characters take multiple bytes in UTF-8. This fix was worked out by Dušan Ristić. I am just committing it.