[5.3] Routing: Allow to mark parsed URLs as tainted (#44455)

Author: Hackwar

libraries/src/Router/Router.php

@@ -80,6 +80,16 @@ class Router
      */
     protected $cache = [];
 
+    /**
+     * Flag to mark the last parsed URL as tainted
+     * If a URL could be read, but has errors, this
+     * flag can be set to true to mark the URL as erroneous.
+     *
+     * @var    bool
+     * @since  __DEPLOY_VERSION__
+     */
+    protected $tainted = false;
+
     /**
      * Router instances container.
      *
@@ -140,6 +150,9 @@ public static function getInstance($client, $options = [])
      */
     public function parse(&$uri, $setVars = false)
     {
+        // Reset the tainted flag
+        $this->tainted = false;
+
         // Do the preprocess stage of the URL parse process
         $this->processParseRules($uri, self::PROCESS_BEFORE);
 
@@ -362,6 +375,34 @@ public function getRules()
         return $this->rules;
     }
 
+    /**
+     * Set the currently parsed URL as tainted
+     * If a URL can be parsed, but not all parts were correct,
+     * (for example an ID was found, but the alias was wrong) the parsing
+     * can be marked as tainted. When the URL is marked as tainted, the router
+     * has to have returned correct data to create the right URL afterwards and
+     * can later do additional processing, like redirecting to the right URL.
+     * If the URL is demonstrably wrong, it should still throw a 404 exception.
+     *
+     * @since  __DEPLOY_VERSION__
+     */
+    public function setTainted()
+    {
+        $this->tainted = true;
+    }
+
+    /**
+     * Return if the last parsed URL was tainted.
+     *
+     * @return  bool
+     *
+     * @since  __DEPLOY_VERSION__
+     */
+    public function isTainted()
+    {
+        return $this->tainted;
+    }
+
     /**
      * Process the parsed router variables based on custom defined rules
      *

plugins/system/sef/src/Extension/Sef.php

@@ -100,6 +100,30 @@ public function onAfterRoute()
             return;
         }
 
+        $router = $this->getSiteRouter();
+
+        /**
+         * The URL was successfully parsed, but is "tainted", e.g. parts of
+         * it were recoverably wrong. So we take the parsed variables, build
+         * a new URL and redirect to that.
+         */
+        if ($router->isTainted()) {
+            $parsedVars = $router->getVars();
+
+            if ($app->getLanguageFilter()) {
+                $parsedVars['lang'] = $parsedVars['language'];
+                unset($parsedVars['language']);
+            }
+
+            $newRoute = Route::_($parsedVars, false);
+            $origUri  = clone Uri::getInstance();
+            $route    = $origUri->toString(['path', 'query']);
+
+            if ($route !== $newRoute) {
+                $app->redirect($newRoute, 301);
+            }
+        }
+
         // Enforce removing index.php with a redirect
         if ($app->get('sef_rewrite') && $this->params->get('indexphp')) {
             $this->removeIndexphp();