From a7a1095487700270b56b33c974f2638c30d53d53 Mon Sep 17 00:00:00 2001
From: quenousimporte <quenousimporte@riseup.net>
Date: Thu, 29 Aug 2024 18:20:47 +0200
Subject: [PATCH] add politis (rss)

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

diff --git a/epub.php b/epub.php
index a7153eb..312580b 100644
--- a/epub.php
+++ b/epub.php
@@ -545,6 +545,110 @@
 		}
 		write_epub($epub);
 	}
+
+	// Politis
+	if (isset($_POST['politis']) && $_POST['politis'])
+	{
+		// cookies?
+		/*$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/politis.epub';
+		$epub->title = 'Politis - RSS ' . $today;
+
+		// load feeds
+		$feedurl = 'https://www.politis.fr/flux-rss-politis-fr/';
+		$feed = file_get_contents($feedurl);
+		$xml = new SimpleXMLElement($feed);
+		$items = $xml->xpath("/rss/channel/item");
+
+		// toc-based to test
+		/*
+		action=get_sommaire
+		id_produit=310761
+		nonce=89e48e1696
+		https://www.politis.fr/wp-admin/admin-ajax.php
+		*/
+
+
+		foreach ($items as $item)
+		{
+			if (new DateTime($item->pubDate) < new DateTime('-7 days'))
+			{
+				continue;
+			}
+			// restrict to last week
+			$url = $item->link;
+
+			$article = file_get_contents($url); //, false, $context);
+			$doc = new DOMDocument();
+			$doc->loadHTML($article);
+			$finder = new DomXPath($doc);
+
+			$title = $finder->query('//h1')->item(0)->textContent;
+			$author = $finder->query('//div[@class="auteur_date"]/span[@class="nom"]')->item(0)->textContent;
+			$date = $finder->query('//div[@class="auteur_date"]/span[@class="date"]')->item(0)->textContent;
+			$summary = $finder->query('//div[@class="extrait"]')->item(0)->textContent;
+
+			// strip images
+			/*$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);
+			}*/
+
+			$toremove = $finder->query('//script');
+			foreach ($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 = $finder->query('//main/section[@class="section_top" or @class="section_content"]');
+			if (!$nodes->length)
+			{
+				echo 'warning: could not get content of "' . $title . '"<br>';
+			}
+			else
+			{
+				// top
+				$node = $nodes->item(0);
+				$nodehtml = get_clean_html($node);
+				$result .= $nodehtml;
+
+				// content
+				$node = $nodes->item(1);
+				$nodehtml = get_clean_html($node);
+				$result .= $nodehtml;
+
+				$epub->AddPage($result, false, $title);
+			}
+		}
+
+		write_epub($epub);
+	}
+
 ?>
 
 <h1>Generate epub</h1>
@@ -581,6 +685,11 @@
 	<input id="nyttoday" name="nyttoday" type="checkbox">
 	<label for="nyttoday">From today's paper embedded script</label>
 
+	<h2>Politis</h2>
+	<input id="politis" name="politis" type="checkbox">
+	<label for="politis">From RSS (last 7 days)</label>
+	<br>
+
 	<h2>Admin</h2>
 	<input id="emptycache" name="emptycache" type="checkbox" checked>
 	<label for="emptycache">Empty cache</label>