mod_spam_filter: resolve http redirects so we're not fooled by url sh…

…orteners

mod_spam_filter/src/mod_spam_filter.erl

@@ -412,11 +412,28 @@ extract_urls(Body) ->
     Options = [global, {capture, all, binary}],
     case re:run(Body, RE, Options) of
 	{match, Captured} when is_list(Captured) ->
-	    {urls, lists:flatten(Captured)};
+      Urls = resolve_redirects(lists:flatten(Captured), []),
+	    {urls, Urls};
 	nomatch ->
 	    none
     end.
 
+-spec resolve_redirects([url()], [url()]) -> [url()].
+resolve_redirects([], Result) -> Result;
+resolve_redirects([Url | Rest], Acc) ->
+    case httpc:request(get, {Url, [{"user-agent", "curl/8.7.1"}]}, [{autoredirect, false}], []) of
+        {ok, {{_, Moved, _}, Headers, _Body}} when Moved >= 300, Moved < 400 ->
+            Location = proplists:get_value("location", Headers),
+            case lists:member(Location, Acc) of
+                false ->
+                    resolve_redirects([Location | Rest], [Url | Acc]);
+                true ->
+                    resolve_redirects(Rest, [Url | Acc])
+            end;
+        _Res ->
+            resolve_redirects(Rest, [Url | Acc])
+    end.
+
 -spec extract_jids(binary()) -> {jids, [ljid()]} | none.
 extract_jids(Body) ->
     RE = <<"\\S+@\\S+">>,