XmlUnit order based on value

I am comparing a saved example xml with a live marshalled xml in my JUnit testing. Validating the presence of a key value pair in the xml.

I am making use of XmlUnit 2.1.0 specifically

My xml is as follows:

<entries>
        <entry>
            <key>delete</key>
            <value>ENABLED</value>
        </entry>
        <entry>
            <key>view</key>
            <value>DISABLED</value>
        </entry>
        <entry>
            <key>create</key>
            <value>DISABLED</value>
        </entry>
    </entries>

The order of the entries can vary. I'm unsure how to get it to validate correctly since it sees a different key value as a difference in the xml even though it's just an order change.

I am asserting similarity with the follow assertion in JUnit:

 assertThat(marshalledXml, isSimilarTo(Input.fromFile("path/to/example.xml").ignoreWhitespace().ignoreComments());

I suspect I may need to make use of XPath matchers or the DefaultNodeMatchers with an ElementSelector.

Solution

Yes, you need to provide an ElementSelector that "knows" which nodes to pick for comparison in your specific case.

For most of the document the name of the element seems to be what you should use. At least that's true for entries, key and value. For entry elements you want to compare those elements, that have matching nested text in the key element that is their immediate child, right?

I think this translates to

ElementSelectors.conditionalBuilder()
    .whenElementIsNamed("entry")
    .thenUse(ElementSelectors.byXPath("./key", ElementSelectors.byNameAndText))
    .elseUse(ElementSelectors.byName)
    .build();

See https://github.com/xmlunit/user-guide/wiki/SelectingNodes for a more detailed discussion of the ElementSelector options. Your XML is pretty close to the table example used in the introduction and discussed in the next sections.