This introduces the DOM, the DiffX Object Model, which is a hierarchical

representation of DiffX files. It can be used to assemble a DiffX file

section-by-section, or to load in a DiffX file and process or manipulate

it.
There's an object for each type of section, organized into a tree. These

are divided into container and content sections, just as in the

specification.
To ease use, consumers really only have to deal with the container

sections (DiffX, DiffXChangeSection, and DiffXFileSection). The

content and options for each child content section is accessible via

forwarding attributes. For example, a DiffXChangeSection's preamble

text can be accessed directly through a preamble attribute, and the

preamble encoding can be accessed through preamble_encoding.
All attributes are type-safe, ensuring there are no problems when

writing a DiffX file. The classes are also limited in what attributes

can be set by defining __slots__, which helps to avoid accidentally

setting the wrong attribute name.
These objects manage data storage. All reading/writing is still done via

the underlying stream reader/writer, with specializations to convert

to-and-from the DOM.
It's expected that most users of the pydiffx module will use the DOM

implementaton, as it provides the easiest way of working with DiffX.

Future updates are planned to automatically calculate insert/line/file

statistics for metadata and to extract information from vendor-specific

Unified Diff variants.

Unit tests pass on Python 2 and 3.
Used this with an experimental diff parser for Review Board and diff

generator for RBTools.