From 566287d767700a5753e5827c963600ac74535b26 Mon Sep 17 00:00:00 2001
From: quenousimporte <quenousimporte@riseup.net>
Date: Thu, 22 Aug 2024 12:08:39 +0200
Subject: [PATCH] add mediapart from home page

---
 epub.php | 127 +++++++++++++++++++++++++++++++++++++++++++++++--------
 1 file changed, 109 insertions(+), 18 deletions(-)

diff --git a/epub.php b/epub.php
index 77bb448..d004a04 100644
--- a/epub.php
+++ b/epub.php
@@ -157,7 +157,7 @@
 			$content = '<h1>' . $article->title . '</h1>';
 			//$content .= '<b>' . $article->abstract . '</b>';
 			$content .= '<div>Page ' . $article->page . '</div>';
-	
+
 			if (count($article->rubrics))
 			{
 				$content .= '<div>Rubriques: ' . implode(',', $article->rubrics) . '</div>';
@@ -175,19 +175,19 @@
 					{
 						if ($item->class == 'quote')
 						{
-							$content .= '<blockquote>' .$item->content . '</blockquote>'; 
+							$content .= '<blockquote>' .$item->content . '</blockquote>';
 						}
 						else if ($item->class == 'paragraphTitle')
 						{
-							$content .= '<h2>' .$item->content . '</h2>'; 
+							$content .= '<h2>' .$item->content . '</h2>';
 						}
 						else if ($item->class == 'introduction')
 						{
-							$content .= '<p><b>' .$item->content . '</b></p>'; 
+							$content .= '<p><b>' .$item->content . '</b></p>';
 						}
 						else
 						{
-							$content .= '<p>' .$item->content . '</p>'; 
+							$content .= '<p>' .$item->content . '</p>';
 						}
 					}
 				}
@@ -206,11 +206,11 @@
 				else
 				{
 					$epub->AddPage('<img style="width: 100%" src="' . $path . '">', false, 'Page ' . $page, true);
-				}				
+				}
 				$page++;
 			}
 			$epub->AddPage($content, false, $article->title);
-			
+
 		}
 		write_epub($epub);
 	}
@@ -291,19 +291,19 @@
 		write_epub($epub);
 	}
 
-	// Mediapart
+	if (isset($_POST['forcempcookies']) && $_POST['forcempcookies'])
+	{
+		$old = $mp_sessionid;
+		$mp_sessionid = get_mp_cookie();
+		$settings = file_get_contents('settings.php');
+		$settings = str_replace($old, $mp_sessionid, $settings);
+		file_put_contents('settings.php', $settings);
+		echo '<div>Updated mpsessid in settings.</div>';
+	}
+
+	// Mediapart RSS
 	if (isset($_POST['mediapart']) && $_POST['mediapart'])
 	{
-		if (isset($_POST['forcempcookies']) && $_POST['forcempcookies'])
-		{
-			$old = $mp_sessionid;
-			$mp_sessionid = get_mp_cookie();
-			$settings = file_get_contents('settings.php');
-			$settings = str_replace($old, $mp_sessionid, $settings);
-			file_put_contents('settings.php', $settings);
-			echo '<div>Updated mpsessid in settings.</div>';
-		}
-
 		$feedurl = 'https://www.mediapart.fr/articles/feed';
 		$opts = [
 		'http' => [
@@ -386,6 +386,94 @@
 		write_epub($epub);
 	}
 
+	// Mediapart home page
+	if (isset($_POST['mphome']) && $_POST['mphome'])
+	{
+		$homeurl = 'https://www.mediapart.fr';
+		$opts = [
+		'http' => [
+			'method' => "GET",
+			'header' => "Accept-language: en\nCookie: MPSESSID=" . $mp_sessionid,
+		]
+		];
+		$context = stream_context_create($opts);
+
+		$epub = new TPEpubCreator();
+		$epub->temp_folder = 'temp/';
+		$epub->epub_file = 'epub/mediaparthome.epub';
+		$epub->title = 'Mediapart home ' . $date;
+
+		$home = file_get_contents($homeurl, false, $context);
+		$doc = new DOMDocument();
+		$doc->loadHTML($home);
+		$finder = new DomXPath($doc);
+
+		$links = $finder->query('//section[contains(@class, "_hot")]//a[not(contains(@href, "/dossier/")) and not(contains(@href, "/studio/")) and not(contains(@href, "fil-")) and (@href != "/journal/series")]/@href');
+
+		foreach ($links as $link)
+		{
+			$url = 'https://www.mediapart.fr' . $link->value;
+
+			$article = file_get_contents($url, false, $context);
+			$doc = new DOMDocument();
+			$doc->loadHTML($article);
+			$finder = new DomXPath($doc);
+
+			$title = $finder->query('//h1')->item(0)->textContent;
+			$author = $finder->query('//a[contains(@href, "biographie")]')->item(0)->textContent;
+			$date = $finder->query('//time')->item(0)->textContent;
+			$summary = $finder->query('//p[contains(@class, "news__heading__top__intro")]')->item(0)->textContent;
+
+			// strip images
+			$toremove = $finder->query('//svg');
+			foreach ($toremove as $elt)
+			{
+				$elt->parentNode->removeChild($elt);
+			}
+			$toremove = $finder->query('//figure');
+			foreach ($toremove as $elt)
+			{
+				$elt->parentNode->removeChild($elt);
+			}
+			$toremove = $finder->query('//span[@class="screen-reader-only"]');
+			foreach ($toremove as $elt)
+			{
+				$elt->parentNode->removeChild($elt);
+			}
+
+			$result = '<h1>' . $title . '</h1>';
+			$result .= '<p>' . $author . '</p>';
+			$result .= '<p>' . $date . '</p>';
+			$result .= '<p><b>' . $summary . '</b></p>';
+
+			$nodes = $finder->query('//div[contains(@class, "paywall-restricted-content")]');
+			if (!$nodes->length)
+			{
+				// articles accès libre
+				$nodes = $finder->query('//div[contains(@class, "news__body__center__article")]');
+			}
+
+			if (!$nodes->length)
+			{
+				echo 'warning: could not get content of "' . $title . '"<br>';
+			}
+			else
+			{
+				$node = $nodes->item(0);
+
+				$innerHTML = '';
+				foreach ($node->childNodes as $childNode){
+					$innerHTML .= $childNode->ownerDocument->saveHTML($childNode);
+				}
+				$result .= '<div>' . strip_tags($innerHTML, '<p><b><h2><i>') . '</div>';
+
+				$epub->AddPage($result, false, $title);
+			}
+		}
+
+		write_epub($epub);
+	}
+
 	// New York Times
 	if (isset($_POST['nyt']) && $_POST['nyt'])
 	{
@@ -565,6 +653,9 @@
 	<input id="mediapart" name="mediapart" type="checkbox">
 	<label for="mediapart">From RSS</label>
 	<br>
+	<input id="mphome" name="mphome" type="checkbox">
+	<label for="mphome">From home page</label>
+	<br>
 	<input id="forcempcookies" name="forcempcookies" type="checkbox">
 	<label for="forcempcookies">Force cookies</label>