presse/lemonde.php

91 lines
2.6 KiB
PHP
Raw Normal View History

2024-05-27 13:28:36 +02:00
<html>
<body>
<?php
2024-05-27 13:28:36 +02:00
$url = '';
$coverurl = '';
if (isset($_POST['url']))
{
require('TPEpubCreator.php');
2024-05-27 13:28:36 +02:00
$url = $_POST['url'];
// extract url from curl command
if (str_starts_with($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/';
$epub->epub_file = $_POST['epub'];
$epub->title = $_POST['title'];
2024-05-27 13:28:36 +02:00
if ($_POST['cover'])
{
$coverurl = $_POST['cover'];
$coverdata = file_get_contents($coverurl);
file_put_contents('temp/cover.jpg', $coverdata);
$epub->AddImage( 'temp/cover.jpg', 'image/jpeg', true );
}
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 - $b->PageNumber; });
2024-05-27 13:28:36 +02:00
foreach ($content as $article)
{
$articlebody = array_filter($article->ContentItem, function($item) { return $item->ContentType == 'text/xml'; })[0];
if ($articlebody->Title && $articlebody->HtmlText)
{
$pagecontent = '<h1>'. strip_tags($articlebody->Title) . '</h1>';
if ($articlebody->Introduction)
{
$pagecontent .= '<b>' . $articlebody->Introduction . '</b>';
}
$pagecontent .= $articlebody->HtmlText;
$epub->AddPage($pagecontent, false, strip_tags($articlebody->Title));
// echo('<h1>'. strip_tags($articlebody->Title) . '</h1>');
// echo($articlebody->HtmlText);
}
}
2024-05-27 13:28:36 +02:00
if ( ! $epub->error ) {
$epub->CreateEPUB();
2024-05-27 13:28:36 +02:00
if ( ! $epub->error ) {
echo 'Success: Download your book <a href="' . $epub->epub_file . '">here</a>.';
}
2024-05-27 13:28:36 +02:00
} else {
echo $epub->error;
}
}
2024-05-27 13:28:36 +02:00
?>
<form method="post">
<p>Go to <a href="https://journal.lemonde.fr" target="_blank">https://journal.lemonde.fr</a> and hit f12 to get GetPublicationContentItems json and cover jpg urls.</p>
Json url: <input name="url" value="<?php echo $url; ?>"><br>
Cover url: <input name="cover" value="<?php echo $coverurl; ?>"><br>
Title: <input name="title" value="Le Monde"><br>
Epub file name: <input name="epub" value="lemonde.epub"><br>
<input type="submit">
</form>
</body>
</html>