I'm writting a servlet filter that manipulates the http response body
(injecting HTML).  It works fine with pages using the English charset,
but when processing a page with double-byte chars, some of the
characters are junk.

When processing the OutputStream, I create a ByteArrayOutputStream

baStream = new ByteArrayOutputStream();

then I create a string (forcing it to UTF-8) with that stream:

String str = new String(baStream.toByteArray(), "UTF-8");

I then manipulate that string using standard regex, then output it back
to the browser:

outStream.write(str.getBytes());

The problem is I don't know a lot about how charsets work in Java.  I
do know that Java's native string charset is UTF-16, but beyond that,
I'm not sure how to make sure that what comes into my servlet filter is
what goes out.

Thanks in advance!

Rich