Originally, worked only with ASCII and Unicode files. Now tries to
determine the file encoding itself. If it can't, as is the case with my
ISO-encoded test files, it guesses that the encoding is ISO Latin 1,
which is a likely guess. It now will work with these files.