Specifying An Encoding

Question

Right, while this is in my head, and I don't forget it myself... ;-)
String myString = "blah";
byte[] myWeirdEncodingBytes = myString.getBytes("myWeirdEncoding");myString is in UTF-16, no matter what. The encoding specified in the getBytes() method is the target encoding. So myWeirdEncodingBytes is an array of bytes stored in a weird encoding, rather than UTF-16.
String myNewString = new String(myWeirdEncodingBytes, "myWeirdEncoding");myNewString is UTF-16 again, no matter what. The encoding specified in the constructor is the source encoding. So myWeirdEncodingBytes are still in a weird encoding. But building a new String out of them requires an encoding conversion to occur, as the String must be in UTF-16. So we need to let the constructor know which encoding to convert from.
A Java byte can be in any encoding it likes - it's simply eight raw bits of data stored in an arbitrary order. Anything extending a Java object is stored in UTF-8, except for String, which keeps its internal characters in UTF-16 encoding.
In short, if you pull data into Java from an external source, the encoding it sits in within Java sticks to the rules above. If it first hits Java as a byte stream (say from an InputStream), it's going to be in whatever encoding it was already in. If it first hits Java as a String (say from request.getParameter() in a servlet or JSP), it will be in UTF-16, as Java will convert its encoding immediately.
The same holds when writing back out. If you write out a series of bytes (such as with OutputStream), the resulting output is in the encoding the bytes were in to begin with. If you're outputting a String (although this is a lot rarer), it will be in UTF-16.
Hope that clears a few people's heads (mine included!)....
Martin Hughes

Answer 1

I think that you should also point out that random data cannot be stored in a string (easily). In general:
  byte[] inputData = <some random data>;
  String inputString = new String(data, "an encoding");
  byte[] outputData = inputString.getBytes("an encoding");
  assert(inputData.equals(outputData)); // ! FALSEeven if "an encoding" is the same in both places. Why? Because some byte (or bytes) in the input stream may not be a valid character for "an encoding". In that case, the resulting character is undefined.
However:
  String inputString = <some random string>;
  byte[] inputData = inputString.getBytes("an encoding");
  String outputString = new String(inputData, "an encoding");
  assert(inputString.equals(outputString)); // ! TRUEis guaranteed, as long as "an encoding" supports all of the characters in inputString.
(Someone please tell me if I'm wrong!)