mirror of
https://tt-rss.org/git/tt-rss.git
synced 2024-07-01 12:40:50 +02:00
f3774b9d65
This primarily occurs when modifying article content. If 'saveXML' is used following 'loadHTML' there is the possibility of strangeness, such as a self-closing anchor tag. Note that the DOMDocument used in 'classes/feeditem/atom.php' came from 'loadXML', but we use 'saveHTML' since we're returning HTML content.
52 lines
1.0 KiB
PHP
52 lines
1.0 KiB
PHP
<?php
|
|
class Af_Fsckportal extends Plugin {
|
|
|
|
private $host;
|
|
|
|
function about() {
|
|
return array(1.0,
|
|
"Remove feedsportal spamlinks from article content",
|
|
"fox");
|
|
}
|
|
|
|
function init($host) {
|
|
$this->host = $host;
|
|
|
|
$host->add_hook($host::HOOK_ARTICLE_FILTER, $this);
|
|
}
|
|
|
|
function hook_article_filter($article) {
|
|
|
|
$doc = new DOMDocument();
|
|
|
|
$charset_hack = '<head>
|
|
<meta http-equiv="Content-Type" content="text/html; charset=utf-8"/>
|
|
</head>';
|
|
|
|
@$doc->loadHTML($charset_hack . $article["content"]);
|
|
|
|
if ($doc) {
|
|
$xpath = new DOMXPath($doc);
|
|
$entries = $xpath->query('(//img[@src]|//a[@href])');
|
|
|
|
foreach ($entries as $entry) {
|
|
if (preg_match("/feedsportal.com/", $entry->getAttribute("src"))) {
|
|
$entry->parentNode->removeChild($entry);
|
|
} else if (preg_match("/feedsportal.com/", $entry->getAttribute("href"))) {
|
|
$entry->parentNode->removeChild($entry);
|
|
}
|
|
}
|
|
|
|
$article["content"] = $doc->saveHTML();
|
|
|
|
}
|
|
|
|
return $article;
|
|
}
|
|
|
|
function api_version() {
|
|
return 2;
|
|
}
|
|
|
|
}
|