Website should work with wsgiref

[chadwhitacre]

I started building a little app using Aspen as a library instead of a server (see #283). I am unable to use wsgiref.simple_server, however. When I run the following and hit it ttw I get spinnage:

#!/usr/bin/env python                                                                                       
from aspen.website import Website                                                                           
from wsgiref.simple_server import make_server                                                               

make_server('0.0.0.0', 8080, Website()).serve_forever()

I believe the issue is from the fp.read call under _read_raw in request.py.

[chadwhitacre]

I ended up getting the little app I was writing to work like so:

#!/usr/bin/env python
import os

from aspen import Response, resources
from aspen.website import Website
from wsgiref.simple_server import make_server


website = Website()
algorithm = website.algorithm


def total_hack(environ):
    # Workaround for https://github.com/gittip/aspen-python/issues/289

    class Dummy:
        def read(*a,**kw):
            return ''

    environ['wsgi.input'] = Dummy()


def favicon(request):
    if request.line.uri.path.raw == '/favicon.ico':
        raise Response(404)

IA_SPT = os.path.join(os.path.dirname(__file__), 'ia.spt')

def serve_ia_spt(request):
    request.fs = IA_SPT
    resource = resources.get(request)
    response = resource.respond(request)
    return {'response': response}


algorithm.functions = [ total_hack
                      , algorithm['parse_environ_into_request']
                      , algorithm['tack_website_onto_request']
                      , favicon
                      , serve_ia_spt
                      , algorithm['log_traceback_for_exception']
                       ]


server = make_server('0.0.0.0', os.environ.get('PORT', 8536), website)
server.serve_forever()

[pjz]

Yeah, it's because aspen is calling read() when there's 0 bytes waiting, so it blocks. Need to check content-length first to see how many bytes are expected, and read that many bytes.