af_comics - fix penny arcade to new markup

This commit is contained in:
Andrew Dolgov 2025-05-17 07:57:39 +03:00
parent 8aac6f2d3d
commit a931b91099
No known key found for this signature in database
GPG Key ID: 1A56B4FA25D4AF2A

View File

@ -6,55 +6,15 @@ class Af_Comics_Pa extends Af_ComicFilter {
} }
function process(&$article) { function process(&$article) {
if (str_contains($article["link"], "penny-arcade.com") && str_contains($article["title"], "Comic:")) { if (str_contains($article["link"], "penny-arcade.com/comic")) {
$doc = new DOMDocument(); $doc = new DOMDocument();
if ($doc->loadHTML(UrlHelper::fetch(['url' => $article['link']]))) { $body = UrlHelper::fetch(['url' => $article['link']]);
if ($body && $doc->loadHTML($body)) {
$xpath = new DOMXPath($doc); $xpath = new DOMXPath($doc);
$basenode = $xpath->query('(//div[@id="comicFrame"])')->item(0); $basenode = $xpath->query('(//div[@class="comic-area"])')->item(0);
if ($basenode) {
$article["content"] = $doc->saveHTML($basenode);
}
}
return true;
}
if (str_contains($article["link"], "penny-arcade.com") && str_contains($article["title"], "News Post:")) {
$doc = new DOMDocument();
$res = UrlHelper::fetch(['url' => $article['link']]);
if ($res && $doc->loadHTML($res)) {
$xpath = new DOMXPath($doc);
$entries = $xpath->query('(//div[@class="post"])');
$basenode = false;
foreach ($entries as $entry) {
$basenode = $entry;
}
$meta = $xpath->query('(//div[@class="meta"])')->item(0);
if ($meta->parentNode) { $meta->parentNode->removeChild($meta); }
$header = $xpath->query('(//div[@class="postBody"]/h2)')->item(0);
if ($header->parentNode) { $header->parentNode->removeChild($header); }
$header = $xpath->query('(//div[@class="postBody"]/div[@class="comicPost"])')->item(0);
if ($header->parentNode) { $header->parentNode->removeChild($header); }
$avatar = $xpath->query('(//div[@class="avatar"]//img)')->item(0);
if ($basenode)
$basenode->insertBefore($avatar, $basenode->firstChild);
$uninteresting = $xpath->query('(//div[@class="avatar"])');
foreach ($uninteresting as $i) {
$i->parentNode->removeChild($i);
}
if ($basenode) { if ($basenode) {
$article["content"] = $doc->saveHTML($basenode); $article["content"] = $doc->saveHTML($basenode);