Use the final URL from the finished request instead of the provided URL for…

Use the final URL from the finished request instead of the provided URL for filename generation, to properly deal with redirects.

library/network/get_url

@@ -183,11 +183,11 @@ def url_do_get(module, url, dest, use_proxy, last_mod_time):
     try:
         r = urllib2.urlopen(request)
         info.update(r.info())
+        info['url'] = r.geturl()  # The URL goes in too, because of redirects.
         info.update(dict(msg="OK (%s bytes)" % r.headers.get('Content-Length', 'unknown'), status=200))
     except urllib2.HTTPError, e:
         # Must not fail_json() here so caller can handle HTTP 304 unmodified
         info.update(dict(msg=str(e), status=e.code))
-        return r, info
     except urllib2.URLError, e:
         code = getattr(e, 'code', -1)
         module.fail_json(msg="Request failed: %s" % str(e), status_code=code)
@@ -287,11 +287,14 @@ def main():
 
     # Now the request has completed, we can finally generate the final
     # destination file name from the info dict.
+
     if dest_is_dir:
         filename = extract_filename_from_headers(info)
         if not filename:
             # Fall back to extracting the filename from the URL.
-            filename = url_filename(url)
+            # Pluck the URL from the info, since a redirect could have changed
+            # it.
+            filename = url_filename(info['url'])
         dest = os.path.join(dest, filename)
 
     md5sum_src   = None