XML documents are just text files, so you could read and write them using ordinary file I/O. But you'd miss the benefits of XML if you did that. Valid XML documents have a lot of structure to them, and we want to read them in a way that lets us check their validity, and also preserve the information about what fields they have and how they are laid out.

What we need is a program that reads a flat XML file and generates a tree data structure in memory containing all the information from the file. Ideally, this program should be general enough to build that structure for all possible valid XML files. Processing an XML file is called “parsing” it. Parsing is the computer science term (borrowed from compiler terminology) for reading something that has a fixed grammar, and checking that it corresponds to its grammar. The program is known as an “XML parser.” The parser provides a service to application programs. Application programs hand the parser a stream of XML from a document file or URL, the parser does its work and then hands back a tree of Java objects that represents or “models” the document.

An XML parser that works this way is said to be a “Document Object Model” or “DOM” parser. The key aspect is that once the DOM parser starts, it carries on until the end and then hands back a complete tree representing the XML file. The DOM parser is very general and doesn't know anything about your customized XML tags. So how does it give you a tree that represents your XML? Well, the DOM API has some interfaces that allow any kind of data to be held in a tree. The parser has some classes that implement those interfaces, and it instantiates objects of those classes.

It's all kept pretty flexible, and allows different parsers to be plugged in and out without affecting your application code. Similarly, you get information out of the tree by calling routines specified in the DOM API. The Node interface is the primary datatype for the Document Object Model. It represents a single node in the document tree, and provides methods for navigating to child Node. Most of the other interfaces, like Document, Element, Entity, and Attr, extend Node. In the next section we will review the code for a simple program that uses a DOM parser. DOM parsers can be and are written in any language, but we are only concerned with Java implementations here.