From 799c93a3c60075c4514ec180053b89b62793939a Mon Sep 17 00:00:00 2001
From: jNullj <iontankatchker@gmail.com>
Date: Mon, 10 Jan 2022 12:44:18 +0200
Subject: [PATCH] [ExplosmBridge] Rewrite to work without feedburner (#2417)

* [ExplosmBridge] Rewrite to work without feedburner

re-wrote the bridge to scrap from the new explosm site as the old method of using feedburner is not working anymore, feedburner is stuck on dec/22 when the explosm site changed.
---
 bridges/ExplosmBridge.php | 57 ++++++++++++++++++++++++++++++++-------
 1 file changed, 47 insertions(+), 10 deletions(-)

diff --git a/bridges/ExplosmBridge.php b/bridges/ExplosmBridge.php
index 876f10e5..cfe42195 100644
--- a/bridges/ExplosmBridge.php
+++ b/bridges/ExplosmBridge.php
@@ -1,22 +1,59 @@
 <?php
-class ExplosmBridge extends FeedExpander {
+class ExplosmBridge extends BridgeAbstract {
 
 	const MAINTAINER = 'bockiii';
 	const NAME = 'Explosm Bridge';
 	const URI = 'https://www.explosm.net/';
 	const CACHE_TIMEOUT = 4800; //2hours
 	const DESCRIPTION = 'Returns the last 5 comics';
+	const PARAMETERS = array(
+		'Get latest posts' => array(
+			'limit' => array(
+				'name' => 'Posts limit',
+				'type' => 'number',
+				'title' => 'Maximum number of items to return',
+				'defaultValue' => 5
+				)
+			)
+		);
 
 	public function collectData(){
-		$this->collectExpandableDatas('https://feeds.feedburner.com/Explosm');
-	}
+		$limit = $this->getInput('limit');
+		$latest = getSimpleHTMLDOM('https://explosm.net/comics/latest');
+		$image = $latest->find('div[id=comic]', 0)->find('img', 0)->getAttribute('src');
+		$date_string = $latest->find('p[class*=Author__P]', 0)->innertext;
+		$next_data_string = $latest->find('script[id=__NEXT_DATA__]', 0)->innertext;
+		$exp = '/{\\\"latest\\\":\[{\\\"slug\\\":\\\"(.*?)\\ /';
+		$reg_array = array();
+		preg_match($exp, $next_data_string, $reg_array);
+		$comic_id = $reg_array[1];
+		$comic_id = substr($comic_id, 0, strpos($comic_id, '\\'));
+		$item = array();
+		$item['uri'] = $this::URI . 'comics/' . $comic_id;
+		$item['uid'] = $this::URI . 'comics/' . $comic_id;
+		$item['title'] = 'Comic for ' . $date_string;
+		$item['timestamp'] = strtotime($date_string);
+		$item['author'] = $latest->find('p[class*=Author__P]', 2)->innertext;
+		$item['content'] = '<img src="' . $image . '" />';
+		$this->items[] = $item;
 
-	protected function parseItem($feedItem){
-		$item = parent::parseItem($feedItem);
-		$comicpage = getSimpleHTMLDOM($item['uri']);
-		$image = $comicpage->find('div[id=comic-wrap]', 0)->find('img', 0)->getAttribute('src');
-		$item['content'] = '<img src="https:' . $image . '" />';
-
-		return $item;
+		$next_comic = substr($this::URI, 0, -1)
+			. $latest->find('div[class*=MainComic__Selector]', 0)->find('a', 0)->getAttribute('href');
+		// use index 1 as the latest comic was already found
+		for ($i = 1; $i <= $limit; $i++) {
+			$this_comic = getSimpleHTMLDOM($next_comic);
+			$image = $this_comic->find('div[id=comic]', 0)->find('img', 0)->getAttribute('src');
+			$date_string = $this_comic->find('p[class*=Author__P]', 0)->innertext;
+			$item = array();
+			$item['uri'] = $next_comic;
+			$item['uid'] = $next_comic;
+			$item['title'] = 'Comic for ' . $date_string;
+			$item['timestamp'] = strtotime($date_string);
+			$item['author'] = $this_comic->find('p[class*=Author__P]', 2)->innertext;
+			$item['content'] = '<img src="' . $image . '" />';
+			$this->items[] = $item;
+			$next_comic = substr($this::URI, 0, -1)
+				. $this_comic->find('div[class*=MainComic__Selector]', 0)->find('a', 0)->getAttribute('href'); // get next comic link
+		}
 	}
 }