108 lines
No EOL
5.4 KiB
PHP
108 lines
No EOL
5.4 KiB
PHP
<?php
|
|
// URL à scraper (remplacez par l'URL souhaitée)
|
|
$url = 'https://www.lemonde.fr/societe/article/2025/07/16/au-c-ur-des-emeutes-de-nouvelle-caledonie-quand-la-mort-de-banane-abattu-par-le-gign-lance-le-siege-de-saint-louis_6621496_3224.html';
|
|
|
|
// Fonction pour récupérer les données Open Graph
|
|
function getOpenGraphData($url) {
|
|
$ogData = [
|
|
'title' => '',
|
|
'description' => '',
|
|
'image' => '',
|
|
'site_name' => '',
|
|
'url' => $url
|
|
];
|
|
|
|
// Configuration du contexte pour éviter les erreurs SSL
|
|
$context = stream_context_create([
|
|
'http' => [
|
|
'method' => 'GET',
|
|
'header' => 'User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36',
|
|
'timeout' => 10
|
|
],
|
|
'ssl' => [
|
|
'verify_peer' => false,
|
|
'verify_peer_name' => false
|
|
]
|
|
]);
|
|
|
|
// Récupérer le HTML
|
|
$html = @file_get_contents($url, false, $context);
|
|
|
|
if ($html === false) {
|
|
return $ogData;
|
|
}
|
|
|
|
// Parser les meta tags Open Graph
|
|
preg_match_all('/<meta\s+property=["\']og:([^"\']+)["\']\s+content=["\']([^"\']+)["\']/i', $html, $matches);
|
|
|
|
if (!empty($matches[1])) {
|
|
foreach ($matches[1] as $index => $property) {
|
|
$content = $matches[2][$index];
|
|
switch ($property) {
|
|
case 'title':
|
|
$ogData['title'] = htmlspecialchars($content);
|
|
break;
|
|
case 'description':
|
|
$ogData['description'] = htmlspecialchars($content);
|
|
break;
|
|
case 'image':
|
|
$ogData['image'] = $content;
|
|
break;
|
|
case 'site_name':
|
|
$ogData['site_name'] = htmlspecialchars($content);
|
|
break;
|
|
}
|
|
}
|
|
}
|
|
|
|
// Fallback: si pas de og:title, utiliser <title>
|
|
if (empty($ogData['title'])) {
|
|
preg_match('/<title>([^<]+)<\/title>/i', $html, $titleMatch);
|
|
if (!empty($titleMatch[1])) {
|
|
$ogData['title'] = htmlspecialchars($titleMatch[1]);
|
|
}
|
|
}
|
|
|
|
// Fallback: si pas de site_name, utiliser le domaine
|
|
if (empty($ogData['site_name'])) {
|
|
$parsed = parse_url($url);
|
|
$ogData['site_name'] = $parsed['host'] ?? '';
|
|
}
|
|
|
|
return $ogData;
|
|
}
|
|
|
|
$ogData = getOpenGraphData($url);
|
|
|
|
// Toujours utiliser le domaine de l'URL pour site_name
|
|
$parsed = parse_url($url);
|
|
$ogData['site_name'] = $parsed['host'] ?? '';
|
|
?>
|
|
|
|
<div class="card--open-graph">
|
|
<?php if (!empty($ogData['image'])): ?>
|
|
<figure>
|
|
<img src="<?= $ogData['image'] ?>" alt="<?= $ogData['title'] ?>">
|
|
</figure>
|
|
<?php endif; ?>
|
|
|
|
<div class="content">
|
|
<?php if (!empty($ogData['site_name'])): ?>
|
|
<span class="site-name"><?= $ogData['site_name'] ?></span>
|
|
<?php endif; ?>
|
|
|
|
<?php if (!empty($ogData['title'])): ?>
|
|
|
|
<h3 class="title">
|
|
<a href="<?= $ogData['url'] ?>" target="_blank">
|
|
<?= $ogData['title'] ?>
|
|
</a>
|
|
</h3>
|
|
<?php endif; ?>
|
|
|
|
<?php if (!empty($ogData['description'])): ?>
|
|
<p class="description"><?= $ogData['description'] ?></p>
|
|
<?php endif; ?>
|
|
</div>
|
|
<a class="link-block" href="<?= $ogData['url'] ?>" target="_blank"></a>
|
|
</div>
|