WebFinger aliases with 'index.php/'

lib/util.php

@@ -1428,6 +1428,37 @@ function common_fake_local_fancy_url($url)
     return $fancy_url;
 }
 
+// FIXME: Maybe this should also be able to handle non-fancy URLs with index.php?p=...
+function common_fake_local_nonfancy_url($url)
+{
+    /**
+     * This is a hacky fix to make URIs NOT generated with "index.php/" match against
+     * locally stored URIs WITH that. The reverse from the above.
+     *
+     * It will also "repair" index.php URLs with multiple / prepended. Like https://some.example///index.php/user/1
+     */
+    if (!preg_match(
+                // [1] protocol part, we can only rewrite http/https anyway.
+                '/^(https?:\/\/)' .
+                // [2] site name.
+                // FIXME: Dunno how this acts if we're aliasing ourselves with a .onion domain etc.
+                '('.preg_quote(common_config('site', 'server'), '/').')' .
+                // [3] site path, or if that is empty just '/' (to retain the /)
+                '('.preg_quote(common_config('site', 'path') ?: '/', '/').')' .
+                // [4] should be empty (might contain one or more / and then maybe also index.php). Will be overwritten.
+                // [5] will have the extracted actual URL part (besides site path)
+                '((?!index.php\/)\/*(?:index.php\/)?)(.*)$/', $url, $matches)) {
+        // if preg_match failed to match
+        throw new Exception('No known change could be made to the URL.');
+    }
+
+    $matches[4] = 'index.php/'; // inject the index.php/ rewritethingy
+
+    // remove the first element, which is the full matching string
+    array_shift($matches);
+    return implode($matches);
+}
+
 function common_inject_session($url, $serverpart = null)
 {
     if (!common_have_session()) {

plugins/WebFinger/WebFingerPlugin.php

@@ -105,11 +105,16 @@ class WebFingerPlugin extends Plugin
                 $profile = $user->getProfile();
             } catch (NoResultException $e) {
                 try {
-                    // common_fake_local_fancy_url can throw an exception
-                    $fancy_url = common_fake_local_fancy_url($resource);
+                    try {   // if it's a /index.php/ url
+                        // common_fake_local_fancy_url can throw an exception
+                        $alt_url = common_fake_local_fancy_url($resource);
+                    } catch (Exception $e) {    // let's try to create a fake local /index.php/ url
+                        // this too if it can't do anything about the URL
+                        $alt_url = common_fake_local_nonfancy_url($resource);
+                    }
 
                     // and this will throw a NoResultException if not found
-                    $user = User::getByUri($fancy_url);
+                    $user = User::getByUri($alt_url);
                     $profile = $user->getProfile();
                 } catch (Exception $e) {
                     // if our rewrite hack didn't work, try to get something by profile URL

plugins/WebFinger/lib/webfingerresource.php

@@ -47,17 +47,33 @@ abstract class WebFingerResource
             // getUrl failed because no valid URL could be returned, just ignore it
         }
 
-        // We claim that we are for example https://site.example/user/1 even if the client
-        // requests https://site.example/index.php/user/1 due to behaviour seen in the wild.
+        /**
+         * Here we add some hacky hotfixes for remote lookups that have been taught the
+         * (at least now) wrong URI but it's still obviously the same user. Such as:
+         * - https://site.example/user/1 even if the client requests https://site.example/index.php/user/1
+         * - https://site.example/user/1 even if the client requests https://site.example//index.php/user/1
+         * - https://site.example/index.php/user/1 even if the client requests https://site.example/user/1
+         * - https://site.example/index.php/user/1 even if the client requests https://site.example///index.php/user/1
+         */
+
+
         foreach(array_keys($aliases) as $alias) {
             try {
                 // get a "fancy url" version of the alias, even without index.php/
-                $fancy_url = common_fake_local_fancy_url($alias);
+                $alt_url = common_fake_local_fancy_url($alias);
                 // store this as well so remote sites can be sure we really are the same profile
-                $aliases[$fancy_url] = true;
+                $aliases[$alt_url] = true;
             } catch (Exception $e) {
-                // in case we couldn't make a "fake local fancy URL", just continue the foreach-loop
-                continue;
+                // Apparently we couldn't rewrite that, the $alias was as the function wanted it to be
+            }
+
+            try {
+                // get a non-"fancy url" version of the alias, i.e. add index.php/
+                $alt_url = common_fake_local_nonfancy_url($alias);
+                // store this as well so remote sites can be sure we really are the same profile
+                $aliases[$alt_url] = true;
+            } catch (Exception $e) {
+                // Apparently we couldn't rewrite that, the $alias was as the function wanted it to be
             }
         }