Hubbub parser architecture
==========================

Introduction
------------

  Hubbub is a flexible HTML parser. It offers two interfaces:
  
    * a SAX-style event interface
    * a DOM-style tree-based interface

Overview
--------

  Hubbub is comprised of two parts:
  
    * a tokeniser
    * a tree builder

  Tokeniser
  ---------
  
    The tokeniser divides the data held in the document buffer into chunks. 
    It sends SAX-style events for each chunk. 
  
  Tree builder
  ------------
  
    The tree builder constructs a DOM-like tree from the SAX events emitted by 
    the tokeniser. The exact representation of the tree is up to the client,
    which must provide a number of tree building handler functions.

Memory usage and ownership
--------------------------

  Memory usage within the library is well defined, as is ownership of allocated
  memory.
  
  Raw input data provided by the library client is owned by the client. 

  SAX events which refer to document segments contain direct references to
  internal data. Token objects are transient and data within them are no 
  longer valid once the event handler has returned control to the tokeniser.
  All data returned by a SAX event is owned by the library.

  The tree builder will use client callbacks to create the objects used
  within the tree. Tree objects may be reference counted (the client may
  do nothing in the ref/unref callbacks and use garbage collection instead).
  The resultant tree is owned by the client.

Parse errors
------------

  Notification of parse errors is made through a dedicated event. This event
  contains the line/column offset of the error location, along with a message
  detailing the error.

Exceptional circumstances
-------------------------

  Exceptional circumstances (such as memory exhaustion) are reported 
  immediately.

  The parser's state in such situations is undefined. There is no recovery
  mechanism. 

  Therefore, if the client is able to recover from the exceptional
  circumstance (e.g. by making more free memory available) the only valid
  way to proceed is to create a new parser instance and start parsing from
  scratch. The client should ensure that they destroy the old parser instance
  and any DOM tree produced by it, to avoid resource leaks.