From 25e9f692614dfbfabbabec0dfe7015ca6c27ac4b Mon Sep 17 00:00:00 2001
From: dag <me@dvikan.no>
Date: Thu, 31 Mar 2022 09:49:30 +0200
Subject: [PATCH] [ElsevierBridge] fix: broken bridge (#2575)

---
 bridges/ElsevierBridge.php | 73 ++++++++++----------------------------
 1 file changed, 18 insertions(+), 55 deletions(-)

diff --git a/bridges/ElsevierBridge.php b/bridges/ElsevierBridge.php
index 08ed4c51..5ab19a1a 100644
--- a/bridges/ElsevierBridge.php
+++ b/bridges/ElsevierBridge.php
@@ -1,7 +1,7 @@
 <?php
 class ElsevierBridge extends BridgeAbstract {
 
-	const MAINTAINER = 'Pierre Mazière';
+	const MAINTAINER = 'dvikan';
 	const NAME = 'Elsevier journals recent articles';
 	const URI = 'https://www.journals.elsevier.com/';
 	const CACHE_TIMEOUT = 43200; //12h
@@ -16,63 +16,26 @@ class ElsevierBridge extends BridgeAbstract {
 		)
 	));
 
-	// Extracts the list of names from an article as string
-	private function extractArticleName($article){
-		$names = $article->find('small', 0);
-		if($names)
-			return trim($names->plaintext);
-		return '';
-	}
-
-	// Extracts the timestamp from an article
-	private function extractArticleTimestamp($article){
-		$time = $article->find('.article-info', 0);
-		if($time) {
-			$timestring = trim($time->plaintext);
-			/*
-				The format depends on the age of an article:
-				- Available online 29 July 2016
-				- July 2016
-				- May–June 2016
-			*/
-			if(preg_match('/\S*(\d+\s\S+\s\d{4})/ims', $timestring, $matches)) {
-				return strtotime($matches[0]);
-			} elseif (preg_match('/[A-Za-z]+\-([A-Za-z]+\s\d{4})/ims', $timestring, $matches)) {
-				return strtotime($matches[0]);
-			} elseif (preg_match('/([A-Za-z]+\s\d{4})/ims', $timestring, $matches)) {
-				return strtotime($matches[0]);
-			} else {
-				return 0;
-			}
-		}
-		return 0;
-	}
-
-	// Extracts the content from an article
-	private function extractArticleContent($article){
-		$content = $article->find('.article-content', 0);
-		if($content) {
-			return trim($content->plaintext);
-		}
-		return '';
-	}
-
-	public function getIcon() {
-		return 'https://cdn.elsevier.io/verona/includes/favicons/favicon-32x32.png';
-	}
-
 	public function collectData(){
-		$uri = self::URI . $this->getInput('j') . '/recent-articles/';
-		$html = getSimpleHTMLDOM($uri);
+		// Not all journals have the /recent-articles page
+		$url = sprintf('https://www.journals.elsevier.com/%s/recent-articles/', $this->getInput('j'));
+		$html = getSimpleHTMLDOM($url);
 
-		foreach($html->find('.pod-listing') as $article) {
-			$item = array();
-			$item['uri'] = $article->find('.pod-listing-header>a', 0)->getAttribute('href') . '?np=y';
-			$item['title'] = $article->find('.pod-listing-header>a', 0)->plaintext;
-			$item['author'] = $this->extractArticleName($article);
-			$item['timestamp'] = $this->extractArticleTimestamp($article);
-			$item['content'] = $this->extractArticleContent($article);
+		foreach($html->find('article') as $recentArticle) {
+			$item = [];
+			$item['uri'] = $recentArticle->find('a', 0)->getAttribute('href');
+			$item['title'] = $recentArticle->find('h2', 0)->plaintext;
+			$item['author'] = $recentArticle->find('p > span', 0)->plaintext;
+			$publicationDateString = trim($recentArticle->find('p > span', 1)->plaintext);
+			$publicationDate = DateTimeImmutable::createFromFormat('F d, Y', $publicationDateString);
+			if ($publicationDate) {
+				$item['timestamp'] = $publicationDate->getTimestamp();
+			}
 			$this->items[] = $item;
 		}
 	}
+
+	public function getIcon(): string {
+		return 'https://cdn.elsevier.io/verona/includes/favicons/favicon-32x32.png';
+	}
 }