HTTP Basic auth encoding (again)

[mkofahl]

This issue is based on #19, which targets the character encoding used during http authentication.

I expect the following change to be useless, because enc is the codepage independent base64 string. No need to require any special character set.

milton2/milton-api/src/main/java/io/milton/http/Auth.java

Line 225 in 2feff46

byte[] bytes = Base64.decodeBase64(enc.getBytes(basicParserCharset));

The actual question is the character encoding of the base64 decoded byte sequence stored in bytes. When storing this byte sequence as String (UTF-16), the platform's default charset matters:

milton2/milton-api/src/main/java/io/milton/http/Auth.java

Line 226 in 2feff46

String s = new String(bytes);

I suggest implementing some simple encoding detection at this point and constructing the string platform independent similar to:

if(isUTF8(bytes)) {
    s = new String(bytes, Charset.forName("UTF-8"));
}
else {
    final String latin = new String(bytes, Charset.forName("ISO-8859-1"));
    final byte[] lbytes = latin.getBytes(Charset.forName("UTF-8"));
    s = new String(lbytes);
}

Actually, the list of expected charsets should be configurable.

[bradmac]

👍