Groovy MarkupBuilder name conflict

Question

I have this code:

String buildCatalog(Catalog catalog) {
    def writer = new StringWriter()
    def xml = new MarkupBuilder(writer)
    xml.catalog(xmlns:\'         


        

Answer 1

There might be a better way, but one trick is to call invokeMethod directly:

String buildCatalog(Catalog catalog) {
    def writer = new StringWriter()
    def xml = new MarkupBuilder(writer)
    xml.catalog(xmlns:'http://www.sybrium.com/XMLSchema/NodeCatalog') {
        delegate.invokeMethod('identity', [{
            groupId(catalog.groupId)
            artifactId(catalog.artifactId)
            version(catalog.version)
        }])
    }

    return writer.toString();
}

This is effectively what Groovy is doing behind the scenes. I couldn't get delegate.identity or owner.identity to work, which are the usual tricks.

---

Edit: I figured out what's going on.

Groovy adds a method with a signature of identity(Closure c) to every object.

This means that when you tried to dynamically invoke the identity element on the XML builder, while passing in a single closure argument, it was calling the identity() method, which is like calling delegate({...}) on the outer closure.

Using the invokeMethod trick forces Groovy to bypass the Meta Object Protocol and treat the method as a dynamic method, even though the identity method already exists on the MetaObject.

Knowing this, we can put together a better, more legible solution. All we have to do is change the signature of the method, like so:

String buildCatalog(Catalog catalog) {
    def writer = new StringWriter()
    def xml = new MarkupBuilder(writer)
    xml.catalog(xmlns:'http://www.sybrium.com/XMLSchema/NodeCatalog') {
        // NOTE: LEAVE the empty map here to prevent calling the identity method!
        identity([:]) {
            groupId(catalog.groupId)
            artifactId(catalog.artifactId)
            version(catalog.version)
        }
    }

    return writer.toString();
}

This is much more readable, it's clearer the intent, and the comment should (hopefully) prevent anyone from removing the "unnecessary" empty map.