From db42f2786c6f362b7ce9f152578592ee4c92ae9f Mon Sep 17 00:00:00 2001
From: ORelio <ORelio@users.noreply.github.com>
Date: Tue, 1 Apr 2025 00:42:08 +0200
Subject: [PATCH] [FeedExpander] Add prepareXml() overridable function (#4485)

* FeedExpander: Remove tailing content in XML

- Move preprocessing code into overridable preprocessXml()
- Auto-remove trailing data after root xml node

* FeedExpander: Add PR reference with use case

* FeedExpander: Code linting

* [FeedExpander] Keep content at end of document for now

Will add back later if more sites have the same issue

* [FeedExpander] prepareXml: Add type hints
---
 lib/FeedExpander.php | 26 ++++++++++++++++++--------
 1 file changed, 18 insertions(+), 8 deletions(-)

diff --git a/lib/FeedExpander.php b/lib/FeedExpander.php
index ef001af1..1a47851b 100644
--- a/lib/FeedExpander.php
+++ b/lib/FeedExpander.php
@@ -22,14 +22,7 @@ abstract class FeedExpander extends BridgeAbstract
         if ($xmlString === '') {
             throw new \Exception(sprintf('Unable to parse xml from `%s` because we got the empty string', $url), 10);
         }
-        // prepare/massage the xml to make it more acceptable
-        $problematicStrings = [
-            '&nbsp;',
-            '&raquo;',
-            '&rsquo;',
-        ];
-        $xmlString = str_replace($problematicStrings, '', $xmlString);
-
+        $xmlString = $this->prepareXml($xmlString);
         $feedParser = new FeedParser();
         try {
             $this->feed = $feedParser->parseFeed($xmlString);
@@ -59,6 +52,23 @@ abstract class FeedExpander extends BridgeAbstract
         return $item;
     }
 
+    /**
+    * Prepare XML document to make it more acceptable by the parser
+    * This method can be overriden by bridges to change this behavior
+    *
+    * @return string
+    */
+    protected function prepareXml(string $xmlString): string
+    {
+        // Remove problematic escape sequences
+        $problematicStrings = [
+            '&nbsp;',
+            '&raquo;',
+            '&rsquo;',
+        ];
+        return str_replace($problematicStrings, '', $xmlString);
+    }
+
     public function getURI()
     {
         return $this->feed['uri'] ?? parent::getURI();