The best way to validate XML in a unit test?

If your program validates the XML against the DTD during normal execution, then you should just get the DTD from wherever your program will get it.

If not and the DTD is extremely short (only a few lines), then storing it as a string in your code is probably okay.

Otherwise, I'd put it in an external file and have your unit test read it from that file.

I've used XmlUnit in the past and found it to be useful.

It can be used to validate XML against a schema or compare your XML against a string. It is clever enough to understand XML's parsing rules.For example it knows that "<e1/>" is equivalent to "<e1></e1>" and can be configured to ignore or include whitespace.

Using a DTD in the unit test to test its validity is one thing, testing for the correct content is another.

You can use the DTD to check for the validity of the generated xml which I would simply read the way you do in your program. I personally would not include it inline (as a String); there is always a dependency between you application code and the unit test. When the generated xml changes, the DTD will also change.

To test for the correct content I would go for XMLUnit.

Asserting xml using XMLUnit:

XMLUnit.setIgnoreWhitespace(true);XMLUnit.setIgnoreDiffBetweenTextAndCDATA(true);Diff diff = new Diff(expectedDocument, obtainedDocument);XMLAssert.assertXMLIdentical("xml invalid", diff, true);

One thing you might come across is the fact that the generated xml might contain changing identifiers (id/uid attributes or alike). This can be solved by using a DifferenceListener when asserting the generated xml.

Example implementation of such DifferenceListener:

public class IgnoreVariableAttributesDifferenceListener implements DifferenceListener {    private final List<String> IGNORE_ATTRS;    private final boolean ignoreAttributeOrder;    public IgnoreVariableAttributesDifferenceListener(List<String> attributesToIgnore, boolean ignoreAttributeOrder) {        this.IGNORE_ATTRS = attributesToIgnore;        this.ignoreAttributeOrder = ignoreAttributeOrder;    }    @Override    public int differenceFound(Difference difference) {        // for attribute value differences, check for ignored attributes        if (difference.getId() == DifferenceConstants.ATTR_VALUE_ID) {            if (IGNORE_ATTRS.contains(difference.getControlNodeDetail().getNode().getNodeName())) {                return RETURN_IGNORE_DIFFERENCE_NODES_IDENTICAL;            }        }        // attribute order mismatch (optionally ignored)        else if (difference.getId() == DifferenceConstants.ATTR_SEQUENCE_ID && ignoreAttributeOrder) {            return RETURN_IGNORE_DIFFERENCE_NODES_IDENTICAL;        }        // attribute missing / not expected        else if (difference.getId() == DifferenceConstants.ATTR_NAME_NOT_FOUND_ID) {            if (IGNORE_ATTRS.contains(difference.getTestNodeDetail().getValue())) {                return RETURN_IGNORE_DIFFERENCE_NODES_IDENTICAL;            }        }        return RETURN_ACCEPT_DIFFERENCE;    }    @Override    public void skippedComparison(Node control, Node test) {        // nothing to do    }}

using DifferenceListener:

    XMLUnit.setIgnoreWhitespace(true);    XMLUnit.setIgnoreDiffBetweenTextAndCDATA(true);    Diff diff = new Diff(expectedDocument, obtainedDocument);    diff.overrideDifferenceListener(new IgnoreVariableAttributesDifferenceListener(Arrays.asList("id", "uid"), true));    XMLAssert.assertXMLIdentical("xml invalid", diff, true);