From 1b584b455106d8346f23ad7b4c7ed38fb0b646dd Mon Sep 17 00:00:00 2001
From: tillcash <tillcash@users.noreply.github.com>
Date: Thu, 14 Aug 2025 18:01:47 +0530
Subject: [PATCH] [CybernewsBridge] add bridge (#4665)

* [CybernewsBridge] add bridge

* [CybernewsBridge] fix lint

* [CybernewsBridge] add header

* [CybernewsBridge] fix url

* [CybernewsBridge] fix url 2

* [CybernewsBridge] revert header

* [CybernewsBridge] refactor

* [CybernewsBridge] final

* [CybernewsBridge] lint
---
 bridges/CybernewsBridge.php | 114 ++++++++++++++++++++++++++++++++++++
 1 file changed, 114 insertions(+)
 create mode 100644 bridges/CybernewsBridge.php

diff --git a/bridges/CybernewsBridge.php b/bridges/CybernewsBridge.php
new file mode 100644
index 00000000..64146f6f
--- /dev/null
+++ b/bridges/CybernewsBridge.php
@@ -0,0 +1,114 @@
+<?php
+
+declare(strict_types=1);
+
+class CybernewsBridge extends BridgeAbstract
+{
+    const NAME          = 'Cybernews';
+    const URI           = 'https://cybernews.com';
+    const DESCRIPTION   = 'Fetches the latest news from Cybernews';
+    const MAINTAINER    = 'tillcash';
+    const CACHE_TIMEOUT = 3600; // 1 hour
+    const MAX_ARTICLES  = 5;
+
+    public function collectData()
+    {
+        $sitemapXml = getContents(self::URI . '/news-sitemap.xml');
+
+        if (!$sitemapXml) {
+            throwServerException('Unable to retrieve Cybernews sitemap');
+        }
+
+        $sitemap = simplexml_load_string($sitemapXml, null, LIBXML_NOCDATA);
+
+        if (!$sitemap) {
+            throwServerException('Unable to parse Cybernews sitemap');
+        }
+
+        foreach ($sitemap->url as $entry) {
+            $url        = trim((string) $entry->loc);
+            $lastmod    = trim((string) $entry->lastmod);
+
+            if (!$url) {
+                continue;
+            }
+
+            $pathParts  = explode('/', trim(parse_url($url, PHP_URL_PATH), '/'));
+            $category   = isset($pathParts[0]) && $pathParts[0] !== '' ? $pathParts[0] : '';
+
+            // Skip non-English versions
+            if (in_array($category, ['nl', 'de'], true)) {
+                continue;
+            }
+
+            $namespaces = $entry->getNamespaces(true);
+            $title      = '';
+
+            if (isset($namespaces['news'])) {
+                $news = $entry->children($namespaces['news'])->news;
+
+                if ($news) {
+                    $title = trim((string) $news->title);
+                }
+            }
+
+            if (!$title) {
+                continue;
+            }
+
+            $this->items[] = [
+                'title'      => $title,
+                'uri'        => $url,
+                'uid'        => $url,
+                'timestamp'  => strtotime($lastmod),
+                'categories' => $category ? [$category] : [],
+                'content'    => $this->fetchFullArticle($url),
+            ];
+
+            if (count($this->items) >= self::MAX_ARTICLES) {
+                break;
+            }
+        }
+    }
+
+    private function fetchFullArticle(string $url): string
+    {
+        $html = getSimpleHTMLDOMCached($url);
+
+        if (!$html) {
+            return 'Unable to fetch article content';
+        }
+
+        $article = $html->find('article', 0);
+
+        if (!$article) {
+            return 'Unable to parse article content';
+        }
+
+        // Remove unnecessary elements
+        $removeSelectors = [
+            'script',
+            'style',
+            'div.links-bar',
+            'div.google-news-cta',
+            'div.a-wrapper',
+            'div.embed_youtube',
+        ];
+
+        foreach ($removeSelectors as $selector) {
+            foreach ($article->find($selector) as $element) {
+                $element->outertext = '';
+            }
+        }
+
+        // Handle lazy-loaded images
+        foreach ($article->find('img') as $img) {
+            if (!empty($img->{'data-src'})) {
+                $img->src = $img->{'data-src'};
+                unset($img->{'data-src'});
+            }
+        }
+
+        return $article->innertext;
+    }
+}