From 8c86c7869110444ae1832617c4bbdb8a163b845f Mon Sep 17 00:00:00 2001
From: Mynacol <git@mynacol.xyz>
Date: Fri, 5 Sep 2025 15:12:00 +0000
Subject: [PATCH] [GolemBridge] Actually fix pagination

Commit 3f33d0e31206c4313278438dbe5454bbf0db43fc modified the logic to
accommodate for the website change.
The commit d00745c41113894e54f3ff71d73006410e3e9e20 broke
pagination again by removing the pagination elements from the article
content.
This commit finally fixes pagination again by doing all pre-processing
before calling `extractContent()`.

Tested with the following articles:
https://www.golem.de/news/supercomputer-jupiter-eingeweiht-europaeisch-ist-hochleistungsrechnen-richtig-gedacht-2509-199789.html
https://www.golem.de/news/ankers-soundcore-sleep-a30-im-praxistest-diese-anc-stoepsel-verbessern-meinen-schlaf-2509-199735.html
---
 bridges/GolemBridge.php | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/bridges/GolemBridge.php b/bridges/GolemBridge.php
index b6d583bb..f122aae0 100644
--- a/bridges/GolemBridge.php
+++ b/bridges/GolemBridge.php
@@ -94,15 +94,16 @@ class GolemBridge extends FeedExpander
                 }
             }
 
-            $item['content'] .= $this->extractContent($articlePage, $item['content']);
-
             // next page
-            $nextUri = $articlePage->find('li.go-pagination__item--next>a', 0);
+            $nextUri = $articlePage->find('li.go-pagination__item--next a', 0);
             if ($nextUri) {
                 $uri = $nextUri->href;
             } else {
                 $uri = null;
             }
+
+            // Only extract the content (and remove content) after all pre-processing is done
+            $item['content'] .= $this->extractContent($articlePage, $item['content']);
         }
 
         return $item;