[mod_python] How to get path from URL

On 28/02/2005, at 3:52 AM, Shawn Harrison wrote:

> Gerald Lee wrote [02/27/05 1:54 AM]:
>> The url is: http://localhost/ebiz/admin/index.py/index
>> I can get string "http://localhost/ebiz/admin/", who can help me?
>
> I got curious, so I came up with the following:
>
> i = len(req.document_root())
> path = req.filename[i+1:]	# +1 is for the fwd slash
> p = 'http://' + '/'.join([req.hostname, path])
>
> # same thing, functional style
> p = ('http://'
>     + '/'.join([req.hostname, 
> req.filename.strip('/')[len(req.document_root().strip('/'))+1:]]))
>
>
> This makes no provision for the case where uri = req.hostname.

Don't understand your last comment, but I don't believe what you have 
will
always work. The problem is that you use the physical filesystem path 
info.
This will fail for ~user access and possibly other cases where a 
separate
physical directory is interposed into an arbitrary location in the URL
namespace.

Take the following for example:

   req.filename = /Users/grahamd/Sites/hclass/page
   req.document_root() = /usr/local/apache-2.0/htdocs
   req.uri = /~grahamd/hclass/page/method2
   req.unparsed_uri = /~grahamd/hclass/page/method2
   req.path_info = /method2

Your code yields:

   http://localhost/age

That it got so close was purely because virtual URL was similar length
to physical path name.

What you want is:

   p = 'http://' + req.hostname + req.uri[:-len(req.path_info)]

Which yields:

   http://localhost/~grahamd/hclass/page

This will not work though if not the standard server port or if "https" 
is
used.

One is therefore better off trying to work out the relative location to 
the
directory and avoid absolute URLs if possible.

Graham