X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;ds=sidebyside;f=mod%2Fparse_url.php;h=79c336ddccad66f84b5869778e5d33e099aed4ed;hb=9c3503a706d445bcbf7a14c2adcd8d344df3e46b;hp=b3b42b6cb6f2ced24b4da9be57d3d16222116c9e;hpb=3208d6afcbcf6e74ca5d0ac30be1559327e00bf9;p=friendica.git
diff --git a/mod/parse_url.php b/mod/parse_url.php
index b3b42b6cb6..79c336ddcc 100644
--- a/mod/parse_url.php
+++ b/mod/parse_url.php
@@ -1,15 +1,19 @@
%s%s";
+ $template = "
%s%s
";
$arr = array('url' => $url, 'text' => '');
@@ -21,23 +25,40 @@ function parse_url_content(&$a) {
killme();
}
- if($url)
+ if($url) {
$s = fetch_url($url);
- else {
+ } else {
echo '';
killme();
}
+ logger('parse_url: data: ' . $s, LOGGER_DATA);
if(! $s) {
echo sprintf($template,$url,$url,'');
killme();
}
+ if(strpos($s,'
')) {
+ $title = substr($s,strpos($s,'')+7,64);
+ if(strpos($title,'<') !== false)
+ $title = strip_tags(substr($title,0,strpos($title,'<')));
+ }
+
+ $config = HTMLPurifier_Config::createDefault();
+ $config->set('Cache.DefinitionImpl', null);
+
+ $purifier = new HTMLPurifier($config);
+ $s = $purifier->purify($s);
+
+// logger('parse_url: purified: ' . $s, LOGGER_DATA);
+
$dom = @HTML5_Parser::parse($s);
- if(! $dom)
- return $ret;
+ if(! $dom) {
+ echo sprintf($template,$url,$url,'');
+ killme();
+ }
$items = $dom->getElementsByTagName('title');
@@ -48,24 +69,31 @@ function parse_url_content(&$a) {
}
}
-
$divs = $dom->getElementsByTagName('div');
if($divs) {
foreach($divs as $div) {
$class = $div->getAttribute('class');
- if($class && stristr($class,'article')) {
+ if($class && (stristr($class,'article') || stristr($class,'content'))) {
$items = $div->getElementsByTagName('p');
if($items) {
foreach($items as $item) {
$text = $item->textContent;
+ if(stristr($text,'