Eu w/Japanese Win2K can't enter password

Question

I haven't been able to debug this issue yet. I was just wondering if someone already knew what this might be. Maybe some kind of character set issue?
A couple Japanese End Users are failing my one-way hash password protected application even though they are entering the same English password as everyone else.
The code looks like this:
      char[] passwordAttempt = edtPassword.getPassword();
      byte[] digest = new byte[0];
      try {
         MessageDigest algorithm = MessageDigest.getInstance("SHA-1");
         algorithm.reset();
         byte[] bytes = new byte[passwordAttempt.length];
         for (int j = 0; j < passwordAttempt.length; j++) {
            bytes[j] = (byte)passwordAttempt[j];
         algorithm.update(bytes);
         digest = algorithm.digest();
      } catch (NoSuchAlgorithmException nsae) {
         Trace.ex(nsae);
      passwordValid = true;
      if(digest.length != password.length) {
         passwordValid = false;
      else {
         for(int i = 0; i < digest.length; i++) {
            if (digest[i] != password) {
passwordValid = false;
break;
This code works fine on all the English-based pc's. I do know that no exceptions are being thrown.
Any suggestions?

Answer 1

      char[] passwordAttempt =
edtPassword.getPassword();
byte[] digest = new byte[0];
try {
MessageDigest algorithm =
gorithm = MessageDigest.getInstance("SHA-1");
algorithm.reset();
byte[] bytes = new
tes = new byte[passwordAttempt.length];
for (int j = 0; j < passwordAttempt.length;
t.length; j++) {
bytes[j] = (byte)passwordAttempt[j];
}You are converting UNICODE characters to bytes by ingoring them most significant byte of a character. This cannot be the right approach but if all the users are typing in ASCII characters for the password then I can't see that it should make any difference.
To make it more general, the best that I can come up with is
byte[] passwordBytes = new String(edtPassword.getPassword()).getBytes("ASCII");
though this looks awkward!
Using this you should end up with exactly the same MD5 when using only ASCII characters in a passwords. If you wanted to allow anything a password then you could use "UTF-8" instead of "ASCII"; this would still give the samve result for ASCII characters but will allow most other languages to be used.