From 9ffd89f4e3cb213921e0656e1819b2941157a07a Mon Sep 17 00:00:00 2001
From: Phantop <phantop@tuta.io>
Date: Fri, 4 Oct 2024 22:02:40 -0400
Subject: [PATCH] [ArsTechnicaBridge] Refactor to website restructuring

---
 bridges/ArsTechnicaBridge.php | 72 ++++++++++++++---------------------
 1 file changed, 28 insertions(+), 44 deletions(-)

diff --git a/bridges/ArsTechnicaBridge.php b/bridges/ArsTechnicaBridge.php
index fcb1bd4f..8bf7e41a 100644
--- a/bridges/ArsTechnicaBridge.php
+++ b/bridges/ArsTechnicaBridge.php
@@ -7,24 +7,24 @@ class ArsTechnicaBridge extends FeedExpander
     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',
-                    ]
+        '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()
@@ -36,39 +36,23 @@ class ArsTechnicaBridge extends FeedExpander
     protected function parseItem(array $item)
     {
         $item_html = getSimpleHTMLDOMCached($item['uri']);
-        $item_html = defaultLinkTo($item_html, self::URI);
-        $item['content'] = $item_html->find('.article-content', 0);
+        $parsely = $item_html->find('[name="parsely-page"]', 0)->content;
+        $parsely_json = Json::decode(html_entity_decode($parsely));
 
-        $parsely = $item_html->find('[name="parsely-page"]', 0);
-        $parsely_json = json_decode(html_entity_decode($parsely->content), true);
         $item['categories'] = $parsely_json['tags'];
-
-        $pages = $item_html->find('nav.page-numbers > .numbers > a', -2);
-        if (null !== $pages) {
-            for ($i = 2; $i <= $pages->innertext; $i++) {
-                $page_url = $item['uri'] . '&page=' . $i;
-                $page_html = getSimpleHTMLDOMCached($page_url);
-                $page_html = defaultLinkTo($page_html, self::URI);
-                $item['content'] .= $page_html->find('.article-content', 0);
-            }
-            $item['content'] = str_get_html($item['content']);
-        }
-
-        // 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('.ad_wrapper') as $ad) {
-            $ad->remove();
-        }
-        foreach ($item['content']->find('.sidebar') as $ad) {
-            $ad->remove();
+        $item['comments'] = $item_html->find('#comments a', 0)->href;
+        $item['content'] = '';
+        foreach ($item_html->find('.post-content') as $content) {
+            $item['content'] .= $content;
         }
 
         $item['content'] = backgroundToImg($item['content']);
 
-        $item['uid'] = explode('=', $item['uri'])[1];
+        // remove various ars advertising
+        $sel = '#social-left, .ars-component-buy-box, .ad_wrapper, .sidebar, .toc-container, .ars-gallery-caption-arrow';
+        foreach ($item['content']->find($sel) as $ad) {
+            $ad->remove();
+        }
 
         return $item;
     }