FIX Multibyte URL routing

Was double url encoding (once in database value, then again in request)
Fixes https://github.com/silverstripe/silverstripe-framework/issues/8723

code/Controllers/ContentController.php

@@ -25,6 +25,7 @@ use SilverStripe\Security\Security;
 use SilverStripe\SiteConfig\SiteConfig;
 use SilverStripe\Versioned\Versioned;
 use SilverStripe\View\ArrayData;
+use SilverStripe\View\Parsers\URLSegmentFilter;
 use SilverStripe\View\Requirements;
 use SilverStripe\View\SSViewer;
 use Translatable;
@@ -195,11 +196,16 @@ class ContentController extends Controller
             if (class_exists('Translatable')) {
                 Translatable::disable_locale_filter();
             }
+
+            $filter = URLSegmentFilter::create();
+
             // look for a page with this URLSegment
             $child = SiteTree::get()->filter([
                 'ParentID' => $this->ID,
-                'URLSegment' => rawurlencode($action),
+                // url encode unless it's multibyte (already pre-encoded in the database)
+                'URLSegment' => $filter->getAllowMultibyte() ? $action : rawurlencode($action),
             ])->first();
+
             if (class_exists('Translatable')) {
                 Translatable::enable_locale_filter();
             }

code/Controllers/ModelAsController.php

@@ -16,6 +16,7 @@ use SilverStripe\Core\Injector\Injector;
 use SilverStripe\Dev\Debug;
 use SilverStripe\ORM\DataObject;
 use SilverStripe\ORM\DB;
+use SilverStripe\View\Parsers\URLSegmentFilter;
 use Translatable;
 
 /**
@@ -127,8 +128,14 @@ class ModelAsController extends Controller implements NestedController
             Translatable::disable_locale_filter();
         }
 
+        // url encode unless it's multibyte (already pre-encoded in the database)
+        $filter = URLSegmentFilter::create();
+        if (!$filter->getAllowMultibyte()) {
+            $URLSegment = rawurlencode($URLSegment);
+        }
+
         // Select child page
-        $conditions = array('"SiteTree"."URLSegment"' => rawurlencode($URLSegment));
+        $conditions = array('"SiteTree"."URLSegment"' => $URLSegment);
         if (SiteTree::config()->get('nested_urls')) {
             $conditions[] = array('"SiteTree"."ParentID"' => 0);
         }

tests/php/Controllers/ModelAsControllerTest.php

@@ -11,6 +11,7 @@ use SilverStripe\Control\Director;
 use SilverStripe\Control\Controller;
 use SilverStripe\Dev\FunctionalTest;
 use Page;
+use SilverStripe\View\Parsers\URLSegmentFilter;
 
 class ModelAsControllerTest extends FunctionalTest
 {
@@ -321,4 +322,37 @@ class ModelAsControllerTest extends FunctionalTest
             'The page should not be found since its parent has not been published, in this case http://<yousitename>/root/sub-root or http://<yousitename>/sub-root'
         );
     }
+
+    public function testAllowMultibyte()
+    {
+        Config::modify()->set(URLSegmentFilter::class, 'default_allow_multibyte', true);
+
+        $parent = new Page();
+        $parent->Title = 'Multibyte test';
+        $parent->URLSegment = 'بلاگ';
+        $parent->write();
+        $parent->copyVersionToStage(Versioned::DRAFT, Versioned::LIVE);
+
+        $child = new Page();
+        $child->Title = 'Multibyte test';
+        $child->URLSegment = 'فضة';
+        $child->ParentID = $parent->ID;
+        $child->write();
+        $child->copyVersionToStage(Versioned::DRAFT, Versioned::LIVE);
+
+        // Emulate browser behaviour around multibyte URL encodings
+        $response = $this->get(rawurlencode('بلاگ'));
+        $this->assertEquals(
+            $response->getStatusCode(),
+            200,
+            'Routes toplevel paths'
+        );
+
+        $response = $this->get(join('/', [rawurlencode('بلاگ'), rawurlencode('فضة')]));
+        $this->assertEquals(
+            $response->getStatusCode(),
+            200,
+            'Routes nested paths'
+        );
+    }
 }