presse/lemonde.php

116 lines
3.4 KiB
PHP

<html>
<body>
<?php
$url = '';
$coverdate = '';
if (isset($_POST['url']))
{
require('TPEpubCreator.php');
$url = $_POST['url'];
$coverdate = $_POST['date'];
// extract url from curl command
if (str_starts_with($url, 'curl '))
{
$url = explode("'", $url)[1];
echo '<p>extracted url from curl command:</p><p>' . $url . '</p>';
}
$epub = new TPEpubCreator();
$epub->temp_folder = 'temp/';
$epub->epub_file = 'lemonde' . $coverdate . '.epub';
$epub->title = 'Le Monde ' . $coverdate ;
$coverurl = 'https://www.lemonde.fr/thumbnail/journal/'. $coverdate .'/1000/1490';
$epub->AddImage( $coverurl, 'image/jpeg', true );
// 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 - $b->PageNumber; });
foreach ($content as $article)
{
$articlebody = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'text/xml'; });
$articlebody = array_values($articlebody)[0];
if ($articlebody->Title && $articlebody->HtmlText)
{
$pagecontent = '<h1>'. strip_tags($articlebody->Title) . '</h1>';
$author = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'author/xml'; });
$author = array_values($author)[0];
if ($author->Author)
{
$pagecontent .= $author->Author;
}
if ($articlebody->Introduction)
{
$pagecontent .= '<b>' . $articlebody->Introduction . '</b>';
}
/*$images = array_values(array_filter($article->ContentItem, function($item) { return $item->ContentType == 'graphic/jpeg' || $item->ContentType == 'image/jpeg'; }));
foreach ($images as $image)
{
$imageid = $image->ContentItemId;
$imageurl = preg_replace('/GetPublicationContentItems-.*\.json/', 'Image-MEDIUM-' . $imageid . '.jpg', $url);
$tempcontent = file_get_contents($imageurl);
file_put_contents('temp/' . $imageid, $tempcontent);
//$epub->AddImage('temp/' . $imageid, false, false );
$pagecontent .= '<div><img src="' . $imageurl . '"></div>';
if ($image->HtmlText)
{
$pagecontent .= $image->HtmlText;
}
}*/
$pagecontent .= $articlebody->HtmlText;
$epub->AddPage($pagecontent, false, strip_tags($articlebody->Title));
}
}
if ( ! $epub->error ) {
$epub->CreateEPUB();
if ( ! $epub->error ) {
echo 'Success: Download your book <a href="' . $epub->epub_file . '">here</a>.';
}
else
{
echo $epub->error;
}
} else {
echo $epub->error;
}
}
?>
<form method="post">
<p>Go to <a href="https://journal.lemonde.fr" target="_blank">https://journal.lemonde.fr</a>, hit f12, open edition, and get GetPublicationContentItems json url.</p>
Json url: <input name="url" value="<?php echo $url; ?>"><br>
Date: <input name="date" value="<?php echo $coverdate ? $coverdate : (new DateTime('tomorrow'))->format('Ymd'); /* or today if before 1pm */?>"><br>
<input type="submit">
</form>
</body>
</html>