presse/mediapart.php

<?php

$sessionid = 'value of cookie MPSESSID';
require('TPEpubCreator.php');
$feedurl = 'https://www.mediapart.fr/articles/feed';
$opts = [
  'http' => [
    'method' => "GET",
    'header' => "Accept-language: en\nCookie: MPSESSID=" . $sessionid,
  ]
];
$context = stream_context_create($opts);

$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'mediapart.epub';
$epub->title = 'Mediapart';

// load feeds
$feed = file_get_contents($feedurl);
$xml = new SimpleXMLElement($feed);
$items = $xml->xpath("/rss/channel/item");

setlocale(LC_ALL, 'fr_FR.UTF8', 'fr_FR','fr','fr','fra','fr_FR@euro');
$localedate = strftime("%A %d %B %Y");

foreach ($items as $item)
{
	$title = $item->title;

	$category = $item->xpath('dc:subject')[0];
	$author = $item->xpath('dc:creator')[0];
	$summary = $item->description;

	$article = file_get_contents($item->link, false, $context);
	$doc = new DOMDocument();
	$doc->loadHTML($article);
	$finder = new DomXPath($doc);

	// clean images
	$figures = $finder->query('//figure');
	foreach ($figures as $figure)
	{
		$figure->parentNode->removeChild($figure);
	}

	// clean "à lire aussi" and inline quotes
	$asides = $finder->query('//aside');
	foreach ($asides as $aside)
	{
		$aside->parentNode->removeChild($aside);
	}

	// improve section titles
	$asides = $finder->query('//h2');
	foreach ($asides as $aside)
	{
		$aside->textContent = "\n*" . $aside->textContent . "*\n";
		// todo keep h2
	}

	$maintext = $finder->query('//div[contains(@class, "paywall-restricted-content")]');
	$result = '<h1>' . $title . '</h1>';
	$result .= '<p>' . $author . '</p>';
	$result .= '<p>' . $item->pubDate . '</p>';
	$result .= '<p><b>' . $summary . '</b></p>';

	// todo use html instead
	$result .= '<div>' . str_replace("\n", '<br>', $maintext->item(0)->textContent) . '</div>';

	//echo $result;

	$epub->AddPage($result, false, $title);

	//break;
}

if ( ! $epub->error )
{
	$epub->CreateEPUB();

	if ( ! $epub->error ) {
		echo 'Success: Download your book <a href="' . $epub->epub_file . '">here</a>.';
	}

	} else {
	echo $epub->error;
}
?>