From 7dad6af3b190ea42b9f1c5de7728f8c3f36ba5f6 Mon Sep 17 00:00:00 2001
From: Kevin Jilissen <info@kevinjilissen.nl>
Date: Tue, 25 Nov 2025 08:04:44 +0100
Subject: [PATCH 1/2] Re-encode submission source in UTF-8

Try to re-encode the submission source if it currently is not valid in
UTF-8. Add some extra safeguard by forcing another sanity check re-encode
from UTF-8 to the original encoding, which filters out binary blobs and
verifies that the re-encoding is non-destructive.
---
 .../Controller/Jury/SubmissionController.php    |  2 +-
 webapp/src/Twig/TwigExtension.php               |  4 ++--
 webapp/src/Utils/Utils.php                      | 17 +++++++++++++++++
 3 files changed, 20 insertions(+), 3 deletions(-)

diff --git a/webapp/src/Controller/Jury/SubmissionController.php b/webapp/src/Controller/Jury/SubmissionController.php
index 0bd7bac6dc..3097765723 100644
--- a/webapp/src/Controller/Jury/SubmissionController.php
+++ b/webapp/src/Controller/Jury/SubmissionController.php
@@ -920,7 +920,7 @@ public function sourceAction(
             $files[$f->getFilename()][$submitId] = [
                 'rank' => $f->getRank(),
                 'filename' => $f->getFilename(),
-                'source'   => mb_check_encoding($f->getSourcecode(), 'UTF-8') ? $f->getSourcecode() : "Could not display file as UTF-8, is it binary?",
+                'source'   => Utils::reencodeUtf8($f->getSourcecode()),
             ];
 
             // Keep track of the single filename within a submission for handling renaming.
diff --git a/webapp/src/Twig/TwigExtension.php b/webapp/src/Twig/TwigExtension.php
index fcfae5a4b4..20faf786d6 100644
--- a/webapp/src/Twig/TwigExtension.php
+++ b/webapp/src/Twig/TwigExtension.php
@@ -892,7 +892,7 @@ public function codeEditor(
 HTML;
         $rank   = $index;
         $id     = sprintf('editor%s', $rank);
-        $source = mb_check_encoding($code, 'UTF-8') ? $code : "Could not display file as UTF-8, is it binary?";
+        $source = Utils::reencodeUtf8($code);
         if ($elementToUpdate) {
             $extraForEdit = <<<JS
 editor.getModel().onDidChangeContent(() => {
@@ -942,7 +942,7 @@ public function getMonacoModel(SubmissionFile $file): string
         }
         $this->renderedSources[$file->getSubmitfileid()] = true;
 
-        $source = mb_check_encoding($file->getSourcecode(), 'UTF-8') ? $file->getSourcecode() : "Could not display file as UTF-8, is it binary?";
+        $source = Utils::reencodeUtf8($file->getSourcecode());
         return sprintf(
             <<<JS
 monaco.editor.createModel(
diff --git a/webapp/src/Utils/Utils.php b/webapp/src/Utils/Utils.php
index f24d65e246..8e86b7855f 100644
--- a/webapp/src/Utils/Utils.php
+++ b/webapp/src/Utils/Utils.php
@@ -1053,6 +1053,23 @@ public static function jsonEncode(mixed $data): string
         return json_encode($data, JSON_PRESERVE_ZERO_FRACTION | JSON_UNESCAPED_SLASHES | JSON_THROW_ON_ERROR);
     }
 
+    public static function reencodeUtf8(string $source): string {
+        $detectOrder = array_unique(array_merge(mb_detect_order(), mb_list_encodings()));
+        $encoding = mb_detect_encoding($source, $detectOrder, strict: true);
+        if ($encoding !== false) {
+            $encoded = $encoding !== 'UTF-8' ? mb_convert_encoding($source, 'UTF-8', $encoding) : $source;
+            // Some binary files are strictly valid in an encoding but fail to re-encoding correctly.
+            // A successive strict call to mb_detect_encoding still says it is validly encoded, but rendering the string fails.
+            // This will filter these files.
+            $sanity = mb_convert_encoding($encoded, $encoding, 'UTF-8');
+            if ($source === $sanity) {
+                return $encoded;
+            }
+        }
+        return "Could not display file as UTF-8.\n"
+             . "Check the supported PHP mbstring encodings on the `Config checker` page if you did not expect this file to be binary.";
+    }
+
     /**
      * @return array<string, string>
      */

From 224dc7e28832b291313e629eae31d8c10868c2e6 Mon Sep 17 00:00:00 2001
From: Kevin Jilissen <info@kevinjilissen.nl>
Date: Thu, 27 Nov 2025 10:56:01 +0100
Subject: [PATCH 2/2] Add a configuration check for encoding detection

---
 webapp/src/Service/CheckConfigService.php | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

diff --git a/webapp/src/Service/CheckConfigService.php b/webapp/src/Service/CheckConfigService.php
index 476038a9b3..5f1c76f4b3 100644
--- a/webapp/src/Service/CheckConfigService.php
+++ b/webapp/src/Service/CheckConfigService.php
@@ -52,6 +52,7 @@ public function runAll(): array
             'php_version' => $this->checkPhpVersion(),
             'php_extensions' => $this->checkPhpExtensions(),
             'php_settings' => $this->checkPhpSettings(),
+            'mbstring_settings' => $this->checkMbstringSettings(),
             'mysql_settings' => $this->checkMysqlSettings(),
         ];
 
@@ -197,6 +198,23 @@ public function checkPhpSettings(): ConfigCheckItem
         );
     }
 
+    public function checkMbstringSettings(): ConfigCheckItem
+    {
+        $this->stopwatch->start(__FUNCTION__);
+
+        $desc = "\nFor submission source code displaying, the detect order should at least contain all expected submission encodings:\n";
+        $desc .= sprintf("  - `mbstring.detect_order` should at least contain `UTF-8` (now set to `%s`).\n", implode(', ', mb_detect_order()));
+        $desc .= sprintf("  - Supported encodings: `%s`.\n", implode(', ', mb_list_encodings()));
+        $result = in_array('UTF-8', mb_detect_order()) ? 'O' : 'W';
+
+        $this->stopwatch->stop(__FUNCTION__);
+        return new ConfigCheckItem(
+            caption: 'PHP multibyte string settings',
+            result: $result,
+            desc: $desc
+        );
+    }
+
     public function checkMysqlSettings(): ConfigCheckItem
     {
         $this->stopwatch->start(__FUNCTION__);