To illustrate the state pattern, let's build an XML parsing tool. The context class will be the parser itself. It will take a string as input and place the tool in an initial parsing state. The various parsing states will eat characters, looking for a specific value, and when that value is found, change to a different state. The goal is to create a tree of node objects for each tag and its contents. To keep things manageable, we'll parse only a subset of XML – tags and tag names. We won't be able to handle attributes on tags. It will parse text content of tags, but won't attempt to parse mixed content, which has tags inside of text. Here is an example simplified XML file that we'll be able to parse:
<book> <author>Dusty Phillips</author> ...
