Webapp2

================================
- Added WSGIApplication.app and WSGIApplication.request, class attributes set
on each request to reference currently active app and request.
WSGIApplication.app is an alias to WSGIApplication.active_instance.

- Fixed double escaping of + in urlunsplit(). Thanks, alkis.

- IMPROVED: configuration now behaves exactly like a dictionary, still
auto-loading configuration values when needed and honoring required configs.
For example, we always used this::

bar = self.app.get_config('foo', 'bar')

Now it is also possible to use direct access and dict methods::

bar = self.app.config['foo']['bar']
or...
bar = self.app.config['foo'].get('bar')
or...
bar = self.app.config.get('foo').get('bar')

The previous get_config() method works as always.

=============================
- Fix: Anchors in generated URLs are quoted using urlib.quote, instead of
urlib.quote_plus.

- In Router.dispatch(), if an exception occurs while handling an exception,
raise it instead of trying to handle it again.

- Fixed bug when writing a unicode string to Response and charset is not set.
Thanks to martinc for the patch.

- Changed: the app won't fallback anymore to the exception handler set for
status 500 if handlers for other status are not set.

- Changed: exceptions are only logged when unhandled. It is up to exception
handlers to log them when appropriate.

===============================
- When a URL matches, some attributes are set in the request object:
- request.route: the matched route
- request.route_args: the matched positional arguments, a tuple
- request.route_kwargs: the matched keyword arguments, a dict

- WSGIApplication.handle_exception() doesn't automatically raises the exception
when in debug mode: it is up to the error handler to raise it if in dev; it
will be raised if no error handler is defined to handle it anyway.

- Added attributes, WSGIApplication.registry, Request.registry and
Request.context, dictionaries for objects in use during the app or request
lifetimes.

- Before passing the request method to the RequestHandler, '-' is replaced
by '_', so that a method like WebDav's 'VERSION-CONTROL' can be supported.

- Config.get() now only returns the passed default value when key is defined.
This is the intended, more predictable behavior: default is a default for
the key, not the module. For example::

If config['foo']['bar'] is not set, return 'baz'.
config.get('foo', 'bar', default='baz')

If config['foo'] is not set, return None. Default is ignored here.
config.get('foo', default='baz')

- Router initialization now receives the app as parameter, so that extended
routes can access app's config.

=============================
- Better compatibility with webapp:
- webapp2.WSGIapplication can be used with webapp.RequestHandler.
- webapp.WSGIapplication can be used with webapp2.RequestHandler.

Although the functionality becomes limited in both cases, this should help
migration.

- Review of Response based on observations from
http://pythonpaste.org/webob/differences.htmlwebapp-response:

- Response.out is now a reference to self, to use webob.Response's neat
.write() method which can handle both string and unicode.
- Response.clear() now sets .body = '' instead of .app_iter = [].

- Added Response.write(), for compatibility with StringIO behavior in webapp
when calling write() passing non-basestring values (issue 2).

- Removed url_escape() and url_unescape(). Unused or almost unused.

- ErrorHandlers can now be defined as strings to be lazily loaded, as they
now use the same dispatch mechanism of other handlers.

Backwards compatibility warning
-------------------------------
- The method handle_exception() is called from app-wide error handlers.
Previously, get() was called.

===============================
- Removed router parameter from get_routes(), get_match_routes(),
get_build_routes(). This simplifies multi-routes quite a bit.

=============================
- redirect() and redirect_to() now accept a keyword argument 'abort' to raise
an exception to do the redirect.

- '_netloc' can be passed to url_for() build URLs for a given domain or
subdomain.

- Added BaseRoute, an interface for custom routes. Several improvements make
the routing system more extensible, while the default Route class sticks to
the basics.

- Nested routes are now possible. As an example, `extras/routes.py` has several
classes that accept nested routes or extend routing in other ways:

- PathPrefixRoute: the idea of this route is to set a base path for other
routes::

app = WSGIApplication([
PathPrefixRoute('/users/<user:\w+>', [
Route('/', UserOverviewHandler, 'user-overview'),
Route('/profile', UserProfileHandler, 'user-profile'),
Route('/projects', UserProjectsHandler, 'user-projects'),
]),
])

The example above is the same as setting the following routes, just more
convenient as you can reuse the path prefix::

app = WSGIApplication([
Route('/users/<user:\w+>/', UserOverviewHandler, 'user-overview'),
Route('/users/<user:\w+>/profile', UserProfileHandler, 'user-profile'),
Route('/users/<user:\w+>/projects', UserProjectsHandler, 'user-projects'),
])

- NamePrefixRoute: Same as PathPrefixRoute, but prefixes the names of routes.

- HandlerPrefixRoute: Same as PathPrefixRoute, but prefixes the handlers of
routes.

- DomainRoute: a route used to restrict route matches to a given domain or
subdomain.

For example, to restrict routes to a subdomain of the appspot domain::

SUBDOMAIN_RE = '^([^.]+)\.app-id\.appspot\.com$'

app = WSGIApplication([
DomainRoute(SUBDOMAIN_RE, [
Route('/foo', 'FooHandler', 'subdomain-thing'),
]),
Route('/bar', 'BarHandler', 'normal-thing'),
])

- ImprovedRoute: a route with redirect_to and strict_slash.

- `redirect_to`: if set, the route is used to redirect to a URL. The value
can be a URL string or a callable that returns a URL. These two are
equivalent::

route = Route('/foo', RedirectHandler, defaults={'url': '/bar'})
route = Route('/foo', redirect_to='/bar')

- `strict_slash`: if True, redirects access to the same URL with different
trailing slash to the strict path defined in the rule. For example, take
these rules::

route = Route('/foo', FooHandler, strict_slash=True)
route = Route('/bar/', BarHandler, strict_slash=True)

Because **strict_slash** is True, this is what will happen:

- Access to ``/foo`` will execute ``FooHandler`` normally.
- Access to ``/bar/`` will execute ``BarHandler`` normally.
- Access to ``/foo/`` will redirect to ``/foo``.
- Access to ``/bar`` will redirect to ``/bar/``.