From d00745c41113894e54f3ff71d73006410e3e9e20 Mon Sep 17 00:00:00 2001
From: Mynacol <git@mynacol.xyz>
Date: Thu, 28 Aug 2025 22:10:00 +0000
Subject: [PATCH] [GolemBridge] Add and remove more stuff

Fix the inclusion of tables and add ul/ol elements.
This includes new unwanted stuff that is explicitly removed.
---
 bridges/GolemBridge.php | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/bridges/GolemBridge.php b/bridges/GolemBridge.php
index d8b18070..e401508a 100644
--- a/bridges/GolemBridge.php
+++ b/bridges/GolemBridge.php
@@ -139,8 +139,9 @@ class GolemBridge extends FeedExpander
 
         // delete known bad elements and unwanted gallery images
         foreach (
-            $article->find('div[id*="adtile"], #job-market, #seminars, iframe, .go-article-header__title, .go-article-header__kicker,
-                        .gbox_affiliate, div.toc, .go-button-bar, .go-alink-list, .go-teaser-block, .go-vh, .go-gallery .[data-active="false"]') as $bad
+            $article->find('div[id*="adtile"], #job-market, #seminars, iframe, .go-article-header__title, .go-article-header__kicker, .go-label--sponsored,
+                        .gbox_affiliate, div.toc, .go-button-bar, .go-alink-list, .go-teaser-block, .go-vh, .go-paywall, .go-index-link, .go-pagination__list,
+                        .go-gallery .[data-active="false"]') as $bad
         ) {
             $bad->remove();
         }
@@ -165,8 +166,8 @@ class GolemBridge extends FeedExpander
         }
 
         foreach (
-            $article->find('div.go-article-header__intro, p, h1, h2, h3, pre, .go-media img[src*="."], .go-media figcaption,
-                    div[class*="golem_tablediv"], iframe, video') as $element
+            $article->find('div.go-article-header__intro, p, h1, h2, h3, pre, ul, ol, .go-media img[src*="."], .go-media figcaption,
+                    table, iframe, video') as $element
         ) {
             if (!str_contains($prevcontent, $element)) {
                 $item .= $element;