MyFancyRequestHandler extends the werkzeug BaseRequestHandler to provide HTTP request execution time to the server log. NB: The associated tutorial is: http://jaytaylor.com/blog/?p=305

serving.py

import time
from werkzeug.serving import BaseRequestHandler

class MyFancyRequestHandler(BaseRequestHandler):
    """Extend werkzeug request handler to suit our needs."""
    def handle(self):
        self.fancyStarted = time.time()
        rv = super(MyFancyRequestHandler, self).handle()
        return rv

    def send_response(self, *args, **kw):
        self.fancyProcessed = time.time()
        super(MyFancyRequestHandler, self).send_response(*args, **kw)

    def log_request(self, code='-', size='-'):
        duration = int((self.fancyProcessed - self.fancyStarted) * 1000)
        self.log('info', '"{0}" {1} {2} [{3}ms]'.format(self.requestline, code, size, duration))

Thanks, this was very timely for me as I just did the same thing with a secondary log using @before_request and @after_request, but this is more what I was looking for - directly appending to the access.log entries.

However, I'm trying to figure out how to override the request handler in a hosted wsgi situation where app.run() is not called directly by my code. Flask does not appear to have a way to pass in werkzeug options to the Flask() constructor. Any ideas on how to tackle that?

Like @noise, I'm trying to accomplish this from gunicorn, which doesn't app.run. The Heroku Procfile doesn't help either. There doesn't seem to be a way to set this handler after/before the call to run.