On Feb 2, 2011, at 9:17 AM, Tom Atkins wrote: > Thanks for the quick reply! Your suggestion works fine. Perfect for my low > traffic server with lots of small sites. I'll look into getting Apache to > manage redirects if (when!) it gets busier. > > The redirect is useful as if a user comes to the www. version of the site and > then to the no www version they lose their session as the cookies don't > match. With your suggestion the session is maintained regardless of the URL > entered.
I'm glad to here it. Did you use the short version or the long version? If you used the short version, would you mind trying out the long version and letting us know how it works? I'd like to include the technique in the documentation. BTW, this technique (or something very like it) should work fine with the regex-based rewrite logic; the request.env values that we're using are basically http request header items, slightly cleaned up. > > On 2 February 2011 16:43, Jonathan Lundell <[email protected]> wrote: > On Feb 2, 2011, at 7:52 AM, Tom Atkins wrote: > > Thanks for your work on the new routers Jonathan - it makes life much > > easier. > > > > Quick question, say I have: > > > > routers = dict( > > BASE = dict( > > domains = { > > 'domain1.com' : 'app1', > > 'domain2.com' : 'app2', > > } > > ), > > ) > > > > But would also like www.domain1.com to map to app1 but also change the URL > > to domain1.com (stripping the www part) what is the best way to achieve > > that? > > > > Ideally I'd like it to work the same way as this .htaccess: > > > > RewriteEngine On > > RewriteCond %{HTTP_HOST} ^www\.(.+)$ [NC] > > RewriteRule ^(.*)$ http://%1/$1 [R=301,L] > > > > This silently redirects to the non www domain and sends a 301 to any search > > engines to make it clear that the only site is at domain1.com (nothing at > > www.domain1.com). > > > > If possible, it's better to do the redirection through .htaccess. It's more > efficient, since web2py never has to run. > > This should also work, if you don't really need the redirect, so it's also > efficient: > > routers = dict( > BASE = dict( > domains = { > 'domain1.com' : 'app1', > 'www.domain1.com' : 'app1', > 'domain2.com' : 'app2', > } > ), > ) > > My impression is that search engines (Google anyway) are fairly smart about > encountering the same content at related domains, and IIRC Google has a > protocol for telling them which address is preferred. > > > If you still want to do it through web2py, I think I'd do the rewrite as > above (for both domain1 and www.domain1) and in models/0redirect.py (or at > any rate, before any other application code runs), something like: > > if request.env.http_host == 'www.domain1.com': > redirect("http://domain1.com%s"; % request.env.request_uri, 301) > > or more generally: > > redirects = { > 'www.domain1.com' : 'domain1.com', > ... > } > > if request.env.http_host in redirects: > scheme = request.env.get('wsgi_url_scheme', 'http').lower() > uri = "%s://%s%s" % (scheme, redirects[request.env.http_host], > request.env.request_uri) > redirect(uri, 301) > > None of that is tested, and you'd want to check it all, including appropriate > escaping (if it's not done already). But you get the idea. It also doesn't > take port overrides into consideration. > >
