Also trim non-breakable spaces from auto-generated titles

Author: mudrd8mz

classes/autotitle.php

@@ -33,15 +33,13 @@ class autotitle {
      */
     public static function extract_title(string $summary): string {
 
-        $summary = trim($summary);
-
         if ($summary === '') {
             return '';
         }
 
         $dom = new \DOMDocument();
         libxml_use_internal_errors(true);
-        $dom->loadHTML($summary);
+        $dom->loadHTML('<?xml encoding="utf-8" ?>' . $summary);
         libxml_clear_errors();
 
         return static::find_first_nonempty_text_node_value($dom);
@@ -59,7 +57,7 @@ public static function extract_title(string $summary): string {
     public static function find_first_nonempty_text_node_value(\DOMNode $node): string {
 
         if ($node->nodeType == XML_TEXT_NODE) {
-            $text = trim($node->textContent);
+            $text = (string) preg_replace('/^\s+|\s+$/u', '', $node->textContent);
 
             if ($text !== '') {
                 return $text;

tests/block_course_contents_test.php

@@ -79,6 +79,15 @@ public function extract_title_data(): array {
                 'summary' => 'Hello  <br> students',
                 'title' => 'Hello',
             ],
+            'Multi-byte' => [
+                'summary' => "\u{a0}µ déjà\u{a0}ěščřžýáíé &nbsp; \u{a0}",
+                'title' => "µ déjà\u{a0}ěščřžýáíé",
+            ],
+            'Non-breakable spaces' => [
+                'summary' => '<h2>&nbsp;<span style="font-family:tahoma">Lesson 4: Your first game &nbsp;' . "\xC2\xA0" .
+                    "\xc2\xa0" . '&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;</span> </h2>',
+                'title' => 'Lesson 4: Your first game',
+            ],
         ];
     }
 }