You can't do anything with the tags in a DOM document, you can only work with the nodes. So that means you have to think of the document as a tree consisting of nodes, rather than as a text document consisting of tags and text.
In this case the document consists of an <a> element with three children. You want to convert that to a document whose root element is one of those three children. You will have to decide whether you want to create a new document, or whether you want to modify the existing document.
As for your second question, I don't think there's a DOM method to return all the descendant nodes of a node. You can express that easily in XPath, if you want to take that route, or you could write the recursive method if you like writing
Java code better.