The Parser Model

The parser model here follows the model in section 8.2.1 of the HTML5 specification, though we do not assume a networking layer.

 [ InputStream ]    // Generic support for reading input.
       ||
  [ Scanner ]       // Breaks down the stream into characters.
       ||
 [ Tokenizer ]      // Groups characters into syntactic
       ||
[ Tree Builder ]    // Organizes units into a tree of objects
       ||
 [ DOM Document ]     // The final state of the parsed document.

InputStream

This is an interface with at least two concrete implementations:

• StringInputStream: Reads an HTML5 string.
• FileInputStream: Reads an HTML5 file.

Scanner

This is a mechanical piece of the parser.

Tokenizer

This follows section 8.4 of the HTML5 spec. It is (roughly) a recursive descent parser. (Though there are plenty of optimizations that are less than purely functional.

EventHandler and DOMTree

EventHandler is the interface for tree builders. Since not all implementations will necessarily build trees, we've chosen a more generic name.

The event handler emits tokens during tokenization.

The DOMTree is an event handler that builds a DOM tree. The output of the DOMTree builder is a DOMDocument.

DOMDocument

PHP has a DOMDocument class built-in (technically, it's part of libxml.) We use that, thus rendering the output of this process compatible with SimpleXML, QueryPath, and many other XML/HTML processing tools.

For cases where the HTML5 is a fragment of a HTML5 document a DOMDocumentFragment is returned instead. This is another built-in class.