utf-8 decoding in java

Question

I'm trying to pass parameters from a PHP middle tier to a java backend that understands J2EE. I'm writing the controller code in Groovy. In there, I'm trying to decode some parameter that will likely contain international characters.

I am really puzzled by the results of my debugging this problem so far, hence I wanted to share it with you in the hope that someone will be able to give the correct interpretation of my results.

For the sake of my little test, the parameter I'm passing is "déjeuner". Just to be sure, System.out.println("déjeuner") correctly gives me:

déjeuner

in the console

Now following are the char/dec and hex values of each char of the original string:

next char: d 100 64
next char: ? -61 c3
next char: ? -87 a9
next char: j 106 6a
next char: e 101 65
next char: u 117 75
next char: n 110 6e
next char: e 101 65
next char: r 114 72

note that the c3a9 sequence in UTF-8 is the wished-for character: http://www.fileformat.info/info/unicode/char/00e9/index.htm

Now if I try to read this string as an UTF-8 string, as in stmt.getBytes("UTF-8"), I suddenly end up having a 11 bytes sequence, as follows:

64 c3 83 c2 a9 6a 65 75 6e 65 72

whereas stmt.getBytes("iso-8859-1") gives me 9 bytes:

64 c3 a9 6a 65 75 6e 65 72

note the c3a9 sequence here!

now if I try to convert the UTF-8 sequence to UTF-8, as in

new String(stmt.getBytes("UTF-8"), "UTF-8");

I get:

next char: d 100 64
next char: ? -61 c3
next char: ? -87 a9
next char: j 106 6a
next char: e 101 65
next char: u 117 75
next char: n 110 6e
next char: e 101 65
next char: r 114 72

note the c3a9 sequence

while

new String(stmt.getBytes("iso-8859-1"), "UTF-8")

results in:

next char: d 100 64
next char: ? -23 e9
next char: j 106 6a
next char: e 101 65
next char: u 117 75
next char: n 110 6e
next char: e 101 65
next char: r 114 72

note the e9 which in utf-8 (and ascii) is, again, the 'é' character that I'm longing for.

Unfortunately, in neither case am I ending up with a proper string that would display like the literal string "déjeuner". Strangely enough, the byte sequences both seem correct though.

Answer 1

When dealing with Strings, always remember: byte != char. So in your first example, you have the char c3, not the byte c3 which is a huge difference: The byte would be part of the UTF-8 sequence but the char already is Unicode. So when you convert that to UTF-8, the Unicode character c3 must become the byte sequence c3 83.

So the question is: How did you get the String? There must be a bug in that code which doesn't properly handle UTF-8 encoded byte sequences.

The reason why ISO-8859-1 usually works is that this encoding doesn't modify any char with a code point < 256 (i.e. anything between 0 and 255), so UTF-8 encoded byte sequences won't be modified.

Your last example is also wrong: The char e9 is é in ISO-8859-1 and Unicode. In UTF-8, it's not valid since it's not a byte and since it's the byte c3 prefix is missing. That said, it correctly represents the Unicode string you seek.