sanitize: properly handle cached content in archived articles

This commit is contained in:
Andrew Dolgov 2017-02-10 12:11:09 +03:00
parent c4ebf01e69
commit 7818bfde0b
1 changed files with 40 additions and 42 deletions

View File

@ -893,23 +893,22 @@
$doc->loadHTML($charset_hack . $res); $doc->loadHTML($charset_hack . $res);
$xpath = new DOMXPath($doc); $xpath = new DOMXPath($doc);
$entries = $xpath->query('(//a[@href]|//img[@src]|//video/source[@src])');
$ttrss_uses_https = parse_url(get_self_url_prefix(), PHP_URL_SCHEME) === 'https'; $ttrss_uses_https = parse_url(get_self_url_prefix(), PHP_URL_SCHEME) === 'https';
$rewrite_base_url = $site_url ? $site_url : SELF_URL_PATH;
$entries = $xpath->query('(//a[@href]|//img[@src]|//video/source[@src])');
foreach ($entries as $entry) { foreach ($entries as $entry) {
if ($site_url) {
if ($entry->hasAttribute('href')) { if ($entry->hasAttribute('href')) {
$entry->setAttribute('href', $entry->setAttribute('href',
rewrite_relative_url($site_url, $entry->getAttribute('href'))); rewrite_relative_url($rewrite_base_url, $entry->getAttribute('href')));
$entry->setAttribute('rel', 'noopener noreferrer'); $entry->setAttribute('rel', 'noopener noreferrer');
} }
if ($entry->hasAttribute('src')) { if ($entry->hasAttribute('src')) {
$src = rewrite_relative_url($site_url, $entry->getAttribute('src')); $src = rewrite_relative_url($rewrite_base_url, $entry->getAttribute('src'));
$extension = $entry->tagName == 'source' ? '.mp4' : '.png'; $extension = $entry->tagName == 'source' ? '.mp4' : '.png';
$cached_filename = CACHE_DIR . '/images/' . sha1($src) . $extension; $cached_filename = CACHE_DIR . '/images/' . sha1($src) . $extension;
@ -962,7 +961,6 @@
$entry->parentNode->replaceChild($p, $entry); $entry->parentNode->replaceChild($p, $entry);
} }
} }
}
if (strtolower($entry->nodeName) == "a") { if (strtolower($entry->nodeName) == "a") {
$entry->setAttribute("target", "_blank"); $entry->setAttribute("target", "_blank");