name:

A nice readable name for the format.

content_types:

A list of mime types; the first one will be preferred.

type:

A string giving some indication what kind of Python object this produces. 'python' is just a generic python object; other kinds of objects should use the module or class name.

load, dump_iter:

load and dump_iter take two arguments -- either a file object or the Python object to be dumped, and then a content type.

want_unicode:

if true, then load and dump produce/consume unicode objects

secure:

if true, then this format can decode arbitrary strings without having to trust the source

choose_mimetype:

if given, this takes a Python object, preferred content type, and header dict and returns the content type

doc:

Any extra documentation you want associated with the object

Takes the WSGI environment and parses the request for this format, possibly using a shortcut to avoid decoding.

Parses the app_iter.

Note: does not call app_iter.close()

Parses the HTTPResponse-like object

Returns a WSGI application that serves the given data, with an optional explicit content_type and optional additional headers.