From a0bbbd69789c140debb087b8f5a6a40d373d2dc5 Mon Sep 17 00:00:00 2001
From: sysadminstory <sysadminstory@users.noreply.github.com>
Date: Sat, 18 Dec 2021 11:19:58 +0100
Subject: [PATCH] [RadioMelodieBridge] Fix to use the new website layout
 (#2330)

---
 bridges/RadioMelodieBridge.php | 117 +++++++++++++++++++++++----------
 1 file changed, 83 insertions(+), 34 deletions(-)

diff --git a/bridges/RadioMelodieBridge.php b/bridges/RadioMelodieBridge.php
index 6807f589..703eb6e0 100644
--- a/bridges/RadioMelodieBridge.php
+++ b/bridges/RadioMelodieBridge.php
@@ -12,12 +12,7 @@ class RadioMelodieBridge extends BridgeAbstract {
 	public function collectData(){
 		$html = getSimpleHTMLDOM(self::URI . '/actu/')
 			or returnServerError('Could not request Radio Melodie.');
-		$list = $html->find('div[class=displayList]', 0)->children();
-
-		$dateFormat = '%A  %e %B %Y à %H:%M';
-		// Set locale and Timezone to parse the date
-		setlocale (LC_TIME, 'fr_FR.utf8');
-		date_default_timezone_set('Europe/Paris');
+		$list = $html->find('div[class=listArticles]', 0)->children();
 
 		foreach($list as $element) {
 			if($element->tag == 'a') {
@@ -34,7 +29,7 @@ class RadioMelodieBridge extends BridgeAbstract {
 				$picture = array();
 
 				// Get the Main picture URL
-				$picture[] = self::URI . $article->find('div[id=pictureTitleSupport]', 0)->find('img', 0)->src;
+				$picture[] = self::URI . $article->find('figure[class=photoviewer]', 0)->find('img', 0)->src;
 				$audioHTML = $article->find('audio');
 
 				// Add the audio element to the enclosure
@@ -57,26 +52,14 @@ class RadioMelodieBridge extends BridgeAbstract {
 					$article->save();
 				}
 
-				// Remove Radio Melodie Logo
-				$logoHTML = $article->find('div[id=logoArticleRM]', 0);
-				$logoHTML->outertext = '';
-				$article->save();
-
-				$author = $article->find('p[class=AuthorName]', 0)->plaintext;
+				// Extract the author
+				$author = $article->find('div[class=author]', 0)->children(1)->children(0)->plaintext;
 
 				// Handle date to timestamp
-				$dateHTML = $article->find('p[class=date]', 0)->plaintext;
-				preg_match('/\| ([^-]*)( - .*|)$/', $dateHTML, $matches);
+				$dateHTML = $article->find('div[class=author]', 0)->children(1)->plaintext;
+				preg_match('/([a-z]{4,10}[ ]{1,2}[0-9]{1,2} [\p{L}]{4,10} [0-9]{4} à [0-9]{2}:[0-9]{2})/mus', $dateHTML, $matches);
 				$dateText = $matches[1];
-				$dateArray = strptime($dateText, $dateFormat);
-				$timestamp = mktime(
-					$dateArray['tm_hour'],
-					$dateArray['tm_min'],
-					$dateArray['tm_sec'],
-					$dateArray['tm_mon'] + 1,
-					$dateArray['tm_mday'],
-					$dateArray['tm_year'] + 1900
-				);
+				$timestamp = $this->parseDate($dateText);
 
 				$item['enclosures'] = array_merge($picture, $audio);
 				$item['author'] = $author;
@@ -86,22 +69,15 @@ class RadioMelodieBridge extends BridgeAbstract {
 					$item['timestamp'] = $timestamp;
 				}
 
-				// Header Image
-				$header = '<img src="' . $picture[0] . '"/>';
-
-				// Remove the Date and Author part
-				$textDOM->find('div[class=AuthorDate]', 0)->outertext = '';
-
-				// Remove Facebook javascript
-				$textDOM->find('script[src^=https://connect.facebook.net]', 0)->outertext = '';
+				// Remove the share article part
+				$textDOM->find('div[class=share]', 0)->outertext = '';
 
 				// Rewrite relative Links
 				$textDOM = defaultLinkTo($textDOM, self::URI . '/');
 
 				$article->save();
-				//$this->rewriteAudioPlayers($textDOM);
 				$text = $textDOM->innertext;
-				$item['content'] = '<h1>' . $item['title'] . '</h1>' . $dateHTML . '<br/>' . $header . $text;
+				$item['content'] = '<h1>' . $item['title'] . '</h1>' . $dateText . '<br/>' . $text;
 				$this->items[] = $item;
 			}
 		}
@@ -143,4 +119,77 @@ class RadioMelodieBridge extends BridgeAbstract {
 		}
 
 	}
+
+	/*
+	 * Function to parse the article date
+	 */
+	private function parseDate($date_fr)
+	{
+		// French date texts
+		$search_fr = array(
+			'janvier',
+			'février',
+			'mars',
+			'avril',
+			'mai',
+			'juin',
+			'juillet',
+			'août',
+			'septembre',
+			'octobre',
+			'novembre',
+			'décembre',
+			'lundi',
+			'mardi',
+			'mercredi',
+			'jeudi',
+			'vendredi',
+			'samedi',
+			'dimanche'
+		);
+
+		// English replacement date text
+		$replace_en = array(
+			'january',
+			'february',
+			'march',
+			'april',
+			'may',
+			'june',
+			'july',
+			'august',
+			'september',
+			'october',
+			'november',
+			'december',
+			'monday',
+			'tuesday',
+			'wednesday',
+			'thursday',
+			'friday',
+			'saturday',
+			'sunday'
+		);
+
+		$dateFormat = 'l j F Y \à H:i';
+
+		// Convert the date from French to English
+		$date_en = str_replace($search_fr, $replace_en, $date_fr);
+
+		// Parse the date and convert it to an array
+		$date_array = date_parse_from_format($dateFormat, $date_en);
+
+		// Convert the array to a unix timestamp
+		$timestamp = mktime(
+			$date_array['hour'],
+			$date_array['minute'],
+			$date_array['second'],
+			$date_array['month'],
+			$date_array['day'],
+			$date_array['year']
+		);
+
+		return $timestamp;
+
+	}
 }