Fix bug where HTML entities in URLs were not decoded on HTML to plain…

… text conversion (#9312)
roundcube · Jan 27, 2024 · 549f99c · 549f99c
1 parent cadc82e
commit 549f99c
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 0 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -3,6 +3,7 @@
 ## Unreleased
 
 - Makefile: Use phpDocumentor v3.4 for the Framework docs (#9313)
+- Fix bug where HTML entities in URLs were not decoded on HTML to plain text conversion (#9312)
 
 ## Release 1.6.6
 

diff --git a/program/lib/Roundcube/rcube_html2text.php b/program/lib/Roundcube/rcube_html2text.php
@@ -758,6 +758,7 @@ public function tags_preg_callback($matches)
         case 'a':
             // Remove spaces in URL (#1487805)
             $url = str_replace(' ', '', $matches[3]);
+            $url = html_entity_decode($url, \ENT_HTML5, RCUBE_CHARSET);
             return $this->_handle_link($url, $matches[4]);
         }
     }

diff --git a/tests/Framework/Html2text.php b/tests/Framework/Html2text.php
@@ -179,6 +179,20 @@ function test_links()
         $res = $ht->get_text();
 
         $this->assertSame($expected, $res, 'Skip link with href == content');
+
+        // HTML entities in links
+        $html     = '<a href="http://test.com?test1&amp;test2">test3&amp;test4</a>';
+        $expected = 'test3&test4 [1]
+
+Links:
+------
+[1] http://test.com?test1&test2
+';
+
+        $ht = new rcube_html2text($html, false, rcube_html2text::LINKS_END);
+        $res = $ht->get_text();
+
+        $this->assertSame($expected, $res, 'Links with HTML entities');
     }
 
     /**