views:

4173

answers:

4

I've setup a static website on GAE using hints found elsewhere, but can't figure out how to return a 404 error. My app.yaml file looks like

- url: (.*)/
  static_files: static\1/index.html
  upload: static/index.html

- url: /
  static_dir: static

with all the static html/jpg files stored under the static directory. The above works for files that exist, but returns a null length file if they don't. The answer is probably to write a python script to return a 404 error, but how do you set things up to serve the static files that exist but run the script for files that don't?

Here is the log from fetching a non-existent file (nosuch.html) on the development application server:

ERROR    2008-11-25 20:08:34,084 dev_appserver.py] Error encountered reading file "/usr/home/ctuffli/www/tufflinet/static/nosuch.html":
[Errno 2] No such file or directory: '/usr/home/ctuffli/www/tufflinet/static/nosuch.html'
INFO     2008-11-25 20:08:34,088 dev_appserver.py] "GET /nosuch.html HTTP/1.1" 404 -
+15  A: 

You need to register a catch-all script handler. Append this at the end of your app.yaml:

- url: /.*
  script: main.py

In main.py you will need to put this code:

from google.appengine.ext import webapp
from google.appengine.ext.webapp.util import run_wsgi_app

class NotFoundPageHandler(webapp.RequestHandler):
    def get(self):
        self.error(404)
        self.response.out.write('<Your 404 error html page>')

application = webapp.WSGIApplication([('/.*', NotFoundPageHandler)],
                                     debug=True)

def main():
    run_wsgi_app(application)

if __name__ == "__main__":
    main()

Replace <Your 404 error html page> with something meaningful. Or better use a template, you can read how to do that here.

Please let me know if you have problems setting this up.

Alexander Kojevnikov
It looks like the last line in my original app.yaml matches before the one you suggest adding
ctuffli
This works, but I prefer Zachary's suggestion to use a static handler for the 404 page (see his answer below).
David Underhill
I'm trying to do it this way but if I set the 404 error, the html is not displayed whether I use a template or not.
Richard Nienaber
does not work for cases when u've different urls like domain.com/admin/.* , domain.com/.* etc
rafek
To handle that (custom 404 for each section) case, you add a `('/admin/.*', NotFoundPageHandler')` in you app setup for each handler in app.yaml.
Robert Kluin
+8  A: 

A significantly simpler way to do this without requiring any CPU cycles is to place this handler at the bottom of your app.yaml

- url: /.*
    static_files: views/404.html
    upload: views/404.html

This then allows you to place a static 404.html file in your views directory. No need for a python handler. Anything that isn't handled in your app.yaml already will hit that.

Zachary Spencer
Brilliant, thanks.
fiXedd
There isn't any way to return a 404 HTTP response code this way, is there?
fiXedd
Yah, all you have to do is add the header to your HTML file.
Zachary Spencer
Maybe it is brilliant.. but it does not work in case of when u have nadlers/controllers for i.e. domain.com/admin/ and different for domain.com/ and u want to handle domain.com/admin/notexistent and also domain.com/nonexistent 404 error
rafek
This is wrong. You can't make static content return a status code other than 200.
Nick Johnson
+6  A: 

google app engine now has Custom Error Responses

so you can now add an error_handlers section to your app.yaml, as in this example:

error_handlers:

- file: default_error.html

- error_code: over_quota
    file: over_quota.html
jonmiddleton
+1  A: 

The dev_appserver is already returning 404 responses for anything that doesn't match the mapping, or does match the mapping but doesn't exist. The 404 response itself has no body, but it's still a 404:

$ wget -O - http://127.0.0.1:8080/foo
--2010-10-28 10:54:51--  http://127.0.0.1:8080/foo
Connecting to 127.0.0.1:8080... connected.
HTTP request sent, awaiting response... 404 
2010-10-28 10:54:51 ERROR 404: (no description).

$ wget -O - http://127.0.0.1:8080/foo/
--2010-10-28 10:54:54--  http://127.0.0.1:8080/foo/
Connecting to 127.0.0.1:8080... connected.
HTTP request sent, awaiting response... 404 
2010-10-28 10:54:54 ERROR 404: (no description).

If you want to return a more user-friendly error page, follow jonmiddleton's advice and specify a custom 404 page.

Nick Johnson
It does not work that way I think error_code can be ^(default|over_quota|dos_api_denial|timeout)$ and default one does NOT handle 404 error
rafek
@rafek Then specify a catchall handler, as Alexander suggests. Either way, the correct response code is already being returned by the default handlers, just without an explanatory message for users.
Nick Johnson
@Nick: if I have /admin/.* handler and (all) /.* handler and do what Alexander suggests.. then the case with domain.com/admin/stuff-that-doesnt-exist is not handled.. coz /admin/.* handler in app.yaml handles it.. and then I should have catch all handler for every script - that's awfull code duplication.
rafek
Then limit your handlers to one, or have a module that contains a 404 handler that you import in each handler, and set as the catchall.
Nick Johnson
Limit my handlers to one?! Not an option. Importing - may be interesting..
rafek
One handler _script_ - as many handlers as you want. Most people use one per 'app'.
Nick Johnson