You could call it Webob augmented itty . It borrows it's form but allow for the use of a passed in Request or Response object. By default, it uses the Request and Response from WebOb, as well as WebOb's.
We were working with itty and hit our head a few times. We really liked the way it worked though and thought we might fit it with our favorite request and response objects. 106 lines later, wee was born as the dispatcher for a wsgi app.
Here is a super simple pony app to show the basics of wee. Pretty much the same as itty except we use full regex strings:
import wee from wsgiref.simple_server import make_server @wee.get(r'^/$') def ponytime(request): return "It's Pony time!" @wee.post(r'^/$'): def make_pony(request): ... make a pony ... @wee.put('^/(?P<pony_name>\w+)$'): def add_unicorn_horn(request, pony_name=None): ... change a pony ... @wee.delete('^/(?P<pony_name>\w+)$'): def kill_a_pony(request, pony_name): ... delete a pony ... srv = make_server(host, port, wee.make_app()) srv.serve_forever()
Using a prefix
Sometimes you want to hang a wee app on an existing app tree:
import wee @wee.get('^/$') def logical_root(request): return "I'm an index" registry = wee.PrefixRegistry(prefix='/my/crazy/existing/dispatch/tree') wee.make_app(registry=prefix)
Experimental REST Container support
There is some rough support for creating simple CRUD containers like so:
import wee @wee.rest("^/candymountain") class UnicornStable(wee.Resource): subtype = 'unicorn_id' def get(self): ... your list of unicorns ... def post(self): name = self.request.POST['unicorn-name'] ... make a unicorn ... def getitem(self, unicorn_id): ... serve a unicorn ... def put(self, unicorn_id): ... change a unicorn ... def delete(self, unicorn_id): ... kill a unicorn ...
The rest verb generates a series of regexes to dispatch upon for the appropriate verbs with a special care to separate 'get' (/) and 'getitem' (/some_id).
Other Differences from itty
The only other main difference is that dispatch is scoped by the module that defines the handler. This means multiple wee apps can run in the same process without clobbering each other.
We don't give you are run command either, nor any adapters for popular frameworks. Maybe later.
coverage.py says 100% currently, but we could have more unittests vs. stack tests.
Wait I'm Confused ...
"When do I use itty and when do I use wee?"
You probably shouldn't use either. This is an experiment, so seems to be itty.
If you are brave though, use wee when you want to do some itty or sinatra like, but are using Webob. Use itty when you want some totally contained in itself w/ no dependencies that requires little configuration that you want to run as quickly as possible.
Matt George and Whit Morriss -- Ax work on Wee