2024-05-27 13:28:36 +02:00
|
|
|
<html>
|
|
|
|
|
<body>
|
|
|
|
|
|
|
|
|
|
<?php
|
2024-05-28 16:13:52 +02:00
|
|
|
|
2024-06-03 17:03:19 +02:00
|
|
|
require 'settings.php';
|
2024-06-03 17:29:30 +02:00
|
|
|
require 'TPEpubCreator.php';
|
2024-06-03 17:03:19 +02:00
|
|
|
|
2024-06-04 09:21:33 +02:00
|
|
|
date_default_timezone_set('Europe/Paris');
|
2024-06-03 17:03:19 +02:00
|
|
|
$date = (new DateTime('today'))->format('Ymd');
|
2024-05-28 16:13:52 +02:00
|
|
|
|
2024-06-03 17:29:30 +02:00
|
|
|
// Le Monde
|
|
|
|
|
if (isset($_POST['lemonde']) && $_POST['lemonde'])
|
|
|
|
|
{
|
|
|
|
|
$url = $_POST['lmurl'];
|
2024-05-28 16:13:52 +02:00
|
|
|
|
|
|
|
|
// extract url from curl command
|
|
|
|
|
if (str_starts_with($url, 'curl '))
|
|
|
|
|
{
|
|
|
|
|
$url = explode("'", $url)[1];
|
|
|
|
|
echo '<p>extracted url from curl command:</p><p>' . $url . '</p>';
|
|
|
|
|
}
|
|
|
|
|
|
2024-05-27 13:28:36 +02:00
|
|
|
$epub = new TPEpubCreator();
|
|
|
|
|
$epub->temp_folder = 'temp/';
|
2024-06-03 21:28:52 +02:00
|
|
|
$epub->epub_file = 'epub/lemonde.epub';
|
2024-06-03 17:03:19 +02:00
|
|
|
$epub->title = 'Le Monde ' . $date ;
|
2024-05-28 16:13:52 +02:00
|
|
|
|
2024-06-03 17:03:19 +02:00
|
|
|
if ($lm_includecover)
|
|
|
|
|
{
|
|
|
|
|
// todo get correct cover according to date and time
|
|
|
|
|
$coverurl = 'https://www.lemonde.fr/thumbnail/journal/'. $date .'/1000/1490';
|
|
|
|
|
$epub->AddImage( $coverurl, 'image/jpeg', true );
|
|
|
|
|
}
|
2024-05-28 16:13:52 +02:00
|
|
|
|
2024-05-27 13:28:36 +02:00
|
|
|
// cache json in case url expires
|
|
|
|
|
$tempjsonpath = 'temp/' . hash('md5', $url) . '.json';
|
|
|
|
|
$json = '';
|
|
|
|
|
if (file_exists($tempjsonpath))
|
|
|
|
|
{
|
|
|
|
|
$json = file_get_contents($tempjsonpath);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
$json = file_get_contents($url);
|
|
|
|
|
file_put_contents($tempjsonpath, $json);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
$publication = json_decode($json);
|
|
|
|
|
$content = array_filter($publication->Content, function($item) { return $item->Category == 'Le Monde'; });
|
|
|
|
|
usort($content, function ($a, $b) { return $a->PageNumber - $b->PageNumber; });
|
2024-05-28 16:13:52 +02:00
|
|
|
|
2024-05-27 13:28:36 +02:00
|
|
|
foreach ($content as $article)
|
|
|
|
|
{
|
2024-05-29 21:35:13 +02:00
|
|
|
$articlebody = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'text/xml'; });
|
|
|
|
|
$articlebody = array_values($articlebody)[0];
|
|
|
|
|
|
2024-05-27 13:28:36 +02:00
|
|
|
if ($articlebody->Title && $articlebody->HtmlText)
|
|
|
|
|
{
|
|
|
|
|
$pagecontent = '<h1>'. strip_tags($articlebody->Title) . '</h1>';
|
2024-06-03 17:03:19 +02:00
|
|
|
$pagecontent .= '<p>Page ' . $article->PageNumber . '</p>';
|
2024-05-29 21:35:13 +02:00
|
|
|
|
|
|
|
|
$author = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'author/xml'; });
|
|
|
|
|
$author = array_values($author)[0];
|
|
|
|
|
if ($author->Author)
|
|
|
|
|
{
|
|
|
|
|
$pagecontent .= $author->Author;
|
|
|
|
|
}
|
|
|
|
|
|
2024-05-27 13:28:36 +02:00
|
|
|
if ($articlebody->Introduction)
|
|
|
|
|
{
|
|
|
|
|
$pagecontent .= '<b>' . $articlebody->Introduction . '</b>';
|
|
|
|
|
}
|
2024-05-29 21:35:13 +02:00
|
|
|
|
2024-06-03 17:03:19 +02:00
|
|
|
if ($lm_includeimages)
|
2024-05-29 21:35:13 +02:00
|
|
|
{
|
2024-06-03 17:03:19 +02:00
|
|
|
$images = array_values(array_filter($article->ContentItem, function($item) { return $item->ContentType == 'graphic/jpeg' || $item->ContentType == 'image/jpeg'; }));
|
|
|
|
|
foreach ($images as $image)
|
2024-05-29 21:35:13 +02:00
|
|
|
{
|
2024-06-03 17:03:19 +02:00
|
|
|
$imageid = $image->ContentItemId;
|
|
|
|
|
$imageurl = preg_replace('/GetPublicationContentItems-.*\.json/', 'Image-MEDIUM-' . $imageid . '.jpg', $url);
|
2024-06-03 21:28:52 +02:00
|
|
|
|
2024-06-03 17:03:19 +02:00
|
|
|
$tempcontent = file_get_contents($imageurl);
|
|
|
|
|
file_put_contents('temp/' . $imageid, $tempcontent);
|
2024-06-03 21:28:52 +02:00
|
|
|
|
2024-06-03 17:03:19 +02:00
|
|
|
//$epub->AddImage('temp/' . $imageid, false, false );
|
2024-06-03 21:28:52 +02:00
|
|
|
|
2024-06-03 17:03:19 +02:00
|
|
|
$pagecontent .= '<div><img src="' . $imageurl . '"></div>';
|
|
|
|
|
if ($image->HtmlText)
|
|
|
|
|
{
|
|
|
|
|
$pagecontent .= $image->HtmlText;
|
|
|
|
|
}
|
2024-05-29 21:35:13 +02:00
|
|
|
}
|
2024-06-03 17:03:19 +02:00
|
|
|
}
|
2024-05-29 21:35:13 +02:00
|
|
|
|
2024-05-27 13:28:36 +02:00
|
|
|
$pagecontent .= $articlebody->HtmlText;
|
|
|
|
|
$epub->AddPage($pagecontent, false, strip_tags($articlebody->Title));
|
|
|
|
|
}
|
|
|
|
|
}
|
2024-05-28 16:13:52 +02:00
|
|
|
|
2024-05-27 13:28:36 +02:00
|
|
|
if ( ! $epub->error ) {
|
|
|
|
|
$epub->CreateEPUB();
|
2024-05-28 16:13:52 +02:00
|
|
|
|
2024-05-27 13:28:36 +02:00
|
|
|
if ( ! $epub->error ) {
|
2024-06-03 17:23:39 +02:00
|
|
|
echo 'Success: ' . $epub->epub_file . ' created.<br>';
|
2024-05-27 13:28:36 +02:00
|
|
|
}
|
2024-05-29 21:35:13 +02:00
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
echo $epub->error;
|
|
|
|
|
}
|
2024-05-28 16:13:52 +02:00
|
|
|
|
2024-05-27 13:28:36 +02:00
|
|
|
} else {
|
|
|
|
|
echo $epub->error;
|
|
|
|
|
}
|
|
|
|
|
}
|
2024-05-28 16:13:52 +02:00
|
|
|
|
2024-06-03 17:38:46 +02:00
|
|
|
// Mediapart
|
|
|
|
|
if (isset($_POST['mediapart']) && $_POST['mediapart'])
|
|
|
|
|
{
|
|
|
|
|
$feedurl = 'https://www.mediapart.fr/articles/feed';
|
|
|
|
|
$opts = [
|
|
|
|
|
'http' => [
|
|
|
|
|
'method' => "GET",
|
2024-06-03 21:28:52 +02:00
|
|
|
'header' => "Accept-language: en\nCookie: MPSESSID=" . $mp_sessionid,
|
2024-06-03 17:38:46 +02:00
|
|
|
]
|
|
|
|
|
];
|
|
|
|
|
$context = stream_context_create($opts);
|
|
|
|
|
|
|
|
|
|
$epub = new TPEpubCreator();
|
|
|
|
|
$epub->temp_folder = 'temp/';
|
2024-06-03 21:28:52 +02:00
|
|
|
$epub->epub_file = 'epub/mediapart.epub';
|
|
|
|
|
$epub->title = 'Mediapart ' . $date;
|
2024-06-03 17:38:46 +02:00
|
|
|
|
|
|
|
|
// load feeds
|
|
|
|
|
$feed = file_get_contents($feedurl);
|
|
|
|
|
$xml = new SimpleXMLElement($feed);
|
|
|
|
|
$items = $xml->xpath("/rss/channel/item");
|
|
|
|
|
|
|
|
|
|
foreach ($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);
|
|
|
|
|
|
|
|
|
|
// 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>' . $item->pubDate . '</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);
|
|
|
|
|
}
|
|
|
|
|
$outerHTML = $node->ownerDocument->saveHTML($node);
|
|
|
|
|
$textcontent = $node->textContent;
|
|
|
|
|
|
|
|
|
|
$result .= '<div>' . strip_tags($innerHTML, '<p><b><h2><i>') . '</div>';
|
|
|
|
|
|
|
|
|
|
$epub->AddPage($result, false, $title);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ( ! $epub->error ) {
|
|
|
|
|
$epub->CreateEPUB();
|
|
|
|
|
|
|
|
|
|
if ( ! $epub->error ) {
|
|
|
|
|
echo 'Success: ' . $epub->epub_file . ' created.<br>';
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
echo $epub->error;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
} else {
|
|
|
|
|
echo $epub->error;
|
2024-06-03 21:28:52 +02:00
|
|
|
}
|
2024-06-03 17:38:46 +02:00
|
|
|
}
|
|
|
|
|
|
2024-06-03 21:28:52 +02:00
|
|
|
// New York Times
|
|
|
|
|
if (isset($_POST['nyt']) && $_POST['nyt'])
|
|
|
|
|
{
|
|
|
|
|
|
|
|
|
|
$feedurl = 'https://rss.nytimes.com/services/xml/rss/nyt/HomePage.xml';
|
|
|
|
|
$opts = [
|
|
|
|
|
'http' => [
|
|
|
|
|
'method' => "GET",
|
|
|
|
|
'header' => "Accept-language: en\nCookie: NYT-S=" . $nyt_sessionid,
|
|
|
|
|
]
|
|
|
|
|
];
|
|
|
|
|
$context = stream_context_create($opts);
|
|
|
|
|
|
|
|
|
|
$epub = new TPEpubCreator();
|
|
|
|
|
$epub->temp_folder = 'temp/';
|
|
|
|
|
$epub->epub_file = 'epub/newyorktimes.epub';
|
|
|
|
|
$epub->title = 'The New York Times ' . $date;
|
|
|
|
|
|
|
|
|
|
// load feeds
|
|
|
|
|
$feed = file_get_contents($feedurl);
|
|
|
|
|
$xml = new SimpleXMLElement($feed);
|
|
|
|
|
$items = $xml->xpath("/rss/channel/item");
|
|
|
|
|
|
|
|
|
|
foreach ($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);
|
|
|
|
|
|
|
|
|
|
// 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>' . $item->pubDate . '</p>';
|
|
|
|
|
$result .= '<p><b>' . $summary . '</b></p>';
|
|
|
|
|
|
|
|
|
|
$nodes = $finder->query('//section[@name="articleBody"]');
|
|
|
|
|
/*if (!$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);
|
|
|
|
|
}
|
|
|
|
|
$outerHTML = $node->ownerDocument->saveHTML($node);
|
|
|
|
|
$textcontent = $node->textContent;
|
|
|
|
|
|
|
|
|
|
$result .= '<div>' . strip_tags($innerHTML, '<p><b><h2><i>') . '</div>';
|
|
|
|
|
|
|
|
|
|
$epub->AddPage($result, false, $title);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if ( ! $epub->error ) {
|
|
|
|
|
$epub->CreateEPUB();
|
|
|
|
|
|
|
|
|
|
if ( ! $epub->error ) {
|
|
|
|
|
echo 'Success: ' . $epub->epub_file . ' created.<br>';
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
echo $epub->error;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
} else {
|
|
|
|
|
echo $epub->error;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
2024-06-03 17:03:19 +02:00
|
|
|
// list existing files
|
|
|
|
|
$files = glob('epub/*');
|
|
|
|
|
foreach ($files as $file)
|
|
|
|
|
{
|
2024-06-03 21:28:52 +02:00
|
|
|
echo '<a href="' . $file . '">' . $file . '</a> ' . date('F d Y H:i:s', filemtime($file)) . '<br>';
|
2024-06-03 17:03:19 +02:00
|
|
|
}
|
|
|
|
|
|
2024-05-27 13:28:36 +02:00
|
|
|
?>
|
|
|
|
|
|
2024-06-03 21:28:52 +02:00
|
|
|
<div>
|
|
|
|
|
Generate epub:<br>
|
2024-05-27 13:28:36 +02:00
|
|
|
<form method="post">
|
2024-06-03 17:38:46 +02:00
|
|
|
<input input name="lemonde" type="checkbox">Le Monde. GetPublicationContentItems url: <input name="lmurl"><br>
|
|
|
|
|
<input input name="mediapart" type="checkbox">Mediapart.<br>
|
2024-06-03 21:28:52 +02:00
|
|
|
<input input name="nyt" type="checkbox">The New York Times.<br>
|
2024-05-27 13:28:36 +02:00
|
|
|
<input type="submit">
|
|
|
|
|
</form>
|
2024-06-03 21:28:52 +02:00
|
|
|
</div>
|
2024-05-27 13:28:36 +02:00
|
|
|
</body>
|
|
|
|
|
</html>
|
