From 586d707ae48b74b150eb101f2e9986d0bd18d89f Mon Sep 17 00:00:00 2001
From: July <phantop@tuta.io>
Date: Sat, 9 Sep 2023 03:19:09 -0400
Subject: [PATCH] [ArsTechnicaBridge] Add new bridge (#3657)

---
 bridges/ArsTechnicaBridge.php | 71 +++++++++++++++++++++++++++++++++++
 1 file changed, 71 insertions(+)
 create mode 100644 bridges/ArsTechnicaBridge.php

diff --git a/bridges/ArsTechnicaBridge.php b/bridges/ArsTechnicaBridge.php
new file mode 100644
index 00000000..1e3e6379
--- /dev/null
+++ b/bridges/ArsTechnicaBridge.php
@@ -0,0 +1,71 @@
+<?php
+
+class ArsTechnicaBridge extends FeedExpander
+{
+    const MAINTAINER = 'phantop';
+    const NAME = 'Ars Technica';
+    const URI = 'https://arstechnica.com/';
+    const DESCRIPTION = 'Returns the latest articles from Ars Technica';
+    const PARAMETERS = [[
+            'section' => [
+                'name' => 'Site section',
+                'type' => 'list',
+                'defaultValue' => 'index',
+                'values' => [
+                    'All' => 'index',
+                    'Apple' => 'apple',
+                    'Board Games' => 'cardboard',
+                    'Cars' => 'cars',
+                    'Features' => 'features',
+                    'Gaming' => 'gaming',
+                    'Information Technology' => 'technology-lab',
+                    'Science' => 'science',
+                    'Staff Blogs' => 'staff-blogs',
+                    'Tech Policy' => 'tech-policy',
+                    'Tech' => 'gadgets',
+                    ]
+            ]
+    ]];
+
+    public function collectData()
+    {
+        $url = 'https://feeds.arstechnica.com/arstechnica/' . $this->getInput('section');
+        $this->collectExpandableDatas($url);
+    }
+
+    protected function parseItem($newItem)
+    {
+        $item = parent::parseItem($newItem);
+
+        $item_html = getSimpleHTMLDOMCached($item['uri'] . '&amp');
+        $item_html = defaultLinkTo($item_html, self::URI);
+        $item['content'] = $item_html->find('.amp-wp-article-content', 0);
+
+        // remove various ars advertising
+        $item['content']->find('#social-left', 0)->remove();
+        foreach ($item['content']->find('.ars-component-buy-box') as $ad) {
+            $ad->remove();
+        }
+        foreach ($item['content']->find('i-amphtml-sizer') as $ad) {
+            $ad->remove();
+        }
+        foreach ($item['content']->find('.sidebar') as $ad) {
+            $ad->remove();
+        }
+
+        foreach ($item['content']->find('a') as $link) { //remove amp redirect links
+            $url = $link->getAttribute('href');
+            if (str_contains($url, 'go.redirectingat.com')) {
+                $url = extractFromDelimiters($url, 'url=', '&amp');
+                $url = urldecode($url);
+                $link->setAttribute('href', $url);
+            }
+        }
+
+        $item['content'] = backgroundToImg(str_replace('data-amp-original-style="background-image', 'style="background-image', $item['content']));
+
+        $item['uid'] = explode('=', $item['uri'])[1];
+
+        return $item;
+    }
+}