@netmikey
This snippet works in jaxb-api

@Test
public void testHexBinary() {
    final String originStr = "6f1087c8105312e302e30820408a00005001588306312e3021000000a5049f1c611a4d02156501ff";
        //\u006f\u0010\ufffd\ufffd\u0010\u0053\u0012\ufffd\u0002\ufffd\u0008\u0020\u0040\ufffd\u0000\u0000\u0050\u0001\u0058\ufffd\u0006\u0031\u002e\u0030\u0021\u0000\u0000\u0000\ufffd\u0004\ufffd\u001c\u0061\u001a\u004d\u0002\u0015\u0065\u0001\ufffd
    final byte[] decoded = DatatypeConverter.parseHexBinary(originStr);
    final String encoded = DatatypeConverter.printHexBinary(decoded);
    Assert.assertEquals(originStr, encoded.toLowerCase());
}

Maybe is a real bug in eclipse-ee4j/jaxb-ri ??

Regards,
Antonio.

@antoniosanct

Not sure I was able to make my point clear...

To have your test reproduce what actually happens at runtime, it should test HexBinaryAdapter (this in turn calls DataTypeConcerter) and most importantly include the newline and whitespace content of the element like this (binary content shortened for readability):

@Test
public void testHexBinary() {
    final String originStr = "\n    6f1087c8105312e302e30820408a00005001588306312e";
    // the following fails:
    final byte[] decoded = new HexBinaryAdapter().unmarshal(originStr);
}

@netmikey
My point is there is no problem about hex binary conversion. Although you maybe think that hex binary adapter should trim the input string before unmarshalling?

All characters included between node tags, including non-visible characters, are valid characters for hex binary conversion.

@lukasj has a better top-down vision about JAXB api. He maybe can help you about the problem.

Regards,
Antonio.

All characters included between node tags, including non-visible characters, are valid characters for hex binary conversion.

I'm afraid they're not:

@Test
public void testHexBinaryConversion() {
    // OK
    new jakarta.xml.bind.DatatypeConverterImpl().parseHexBinary("1234567890abcdef");
    // java.lang.IllegalArgumentException: hexBinary needs to be even-length
    new jakarta.xml.bind.DatatypeConverterImpl().parseHexBinary("\n1234567890abcdef");
    // java.lang.IllegalArgumentException: contains illegal character for hexBinary
    new jakarta.xml.bind.DatatypeConverterImpl().parseHexBinary("\n   1234567890abcdef");
}

@netmikey
You're right about non-visible characters, my mistake! However, I've my doubts about if jaxb-api have to clean up such characters. The solution is easy: set a trimmed string in HexBinaryAdapter.

Regards,
Antonio.

What do you mean by "set a trimmed string in HexBinaryAdapter"? I'm not calling HexBinaryAdapter manually. I'm just using the library and expect it to unmarshal an XML document without errors.

Something like:

JAXBContext context = JAXBContext.newInstance(Book.class);
context.createUnmarshaller().unmarshal(new FileReader("./book.xml"));

@netmikey
You don't! But you could maybe create a PR fixing this issue at this project!

Regards,
Antonio.

Oh, yes, of course. Where do you think trimming the hex binary string would make the most sense then? Since this particular behavior seems to be hexBinary-specific, my first impulse would be adding it in HexBinaryAdapter before passing it to DatatypeConverter.

Would you be open for such a PR?