add politis

epub.php

@@ -21,7 +21,7 @@
 			$nodeHTML = str_replace('</aside', '</i></b', $nodeHTML);
 			$innerHTML .= $nodeHTML;
 		}
-		return '<div>' . strip_tags($innerHTML, '<p><b><h2><i>') . '</div>';
+		return '<div>' . strip_tags($innerHTML, '<p><b><h2><i><strong>') . '</div>';
 	}
 
 	function add_mp_article($url, $context, $epub)
@@ -397,6 +397,71 @@
 		write_epub($epub);
 	}
 
+	// Politis RSS
+	if (isset($_POST['politis']) && $_POST['politis'])
+	{
+		$issuenb = '';
+		if (isset($_POST['politisnb']))
+		{
+			$issuenb = $_POST['politisnb'];
+		}
+		$feedurl = 'https://www.politis.fr/flux-rss-politis-fr/';
+		$opts = [
+		'http' => [
+			'method' => "GET",
+			//'header' => "Accept-language: en\nCookie: MPSESSID=" . $mp_sessionid, // todo
+		]
+		];
+		$context = stream_context_create($opts);
+
+		// todo add cover!
+
+		$epub = new TPEpubCreator();
+		$epub->temp_folder = 'temp/';
+		$epub->epub_file = 'epub/politis' . $issuenb . '.epub';
+		$epub->title = 'Politis ' . $issuenb . ' - RSS ' . $today;
+
+		// load feeds
+		$feed = file_get_contents($feedurl);
+		$xml = new SimpleXMLElement($feed);
+		$items = $xml->xpath("/rss/channel/item");
+
+		// reverse to respect timing
+		foreach (array_reverse($items) as $item)
+		{
+			$url = $item->link;
+			$article = file_get_contents($url, false, $context);
+
+			if ($issuenb && !str_contains($article, ' dans l’hebdo N° ' . $issuenb))
+			{
+				continue;
+			}
+
+			$doc = new DOMDocument();
+			$doc->loadHTML($article);
+			$finder = new DomXPath($doc);
+
+			$title = $finder->query('//h1')->item(0)->textContent;
+			$author = $finder->query('//span[@class="nom"]')->item(0)->textContent;
+			$date = $finder->query('//span[@class="date"]')->item(0)->textContent;
+			$summary = $finder->query('//div[@class="extrait"]')->item(0)->textContent;
+
+			$result = '<h1>' . $title . '</h1>';
+			$result .= '<p>' . $author . '</p>';
+			$result .= '<p>' . $date . '</p>';
+			$result .= '<p><b>' . $summary . '</b></p>';
+
+			$nodes = $finder->query('//div[@id="content"]');
+			$node = $nodes->item(0);
+			$nodehtml = get_clean_html($node);
+			$result .= $nodehtml;
+			$epub->AddPage($result, false, $title);
+
+		}
+
+		write_epub($epub);
+	}
+
 	// Mediapart home page
 	if (isset($_POST['mphome']) && $_POST['mphome'])
 	{
@@ -591,6 +656,11 @@
 	<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.</label>
+	<label for="politisnb">Optional issue number: </label><input id="politisnb" name="politisnb">
+
 	<h2>Admin</h2>
 	<input id="emptycache" name="emptycache" type="checkbox" checked>
 	<label for="emptycache">Empty cache</label>

index.html

@@ -68,6 +68,7 @@
 			<li><a href="https://www.cairn.info/revue-ecorev.htm">Eco Rev'</a></li>
 		</ul>
 		<li><a href="./epub">Fichiers epub</a></li>
+		<li><a href="./epub.php">Génération epub</a></li>
 	</ul>
 </body>
 </html>