X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=mod%2Fparse_url.php;h=46c6b46e97998e3db59b542d11b3b15b937c8fe4;hb=cfc7f8308f5e1756d1ea7b2fadd317821442d16c;hp=15a6aced0e0fda8fcff6d765c17aea3cf94baa00;hpb=7c5a7a94d4d224d23f7c922b86a60a6ffe53fc9f;p=friendica.git

diff --git a/mod/parse_url.php b/mod/parse_url.php
index 15a6aced0e..46c6b46e97 100644
--- a/mod/parse_url.php
+++ b/mod/parse_url.php
@@ -1,6 +1,7 @@
 <?php
 
 require_once('library/HTML5/Parser.php');
+require_once('library/HTMLPurifier.auto.php');
 
 function parse_url_content(&$a) {
 
@@ -12,7 +13,7 @@ function parse_url_content(&$a) {
 
 	$text = null;
 
-	$template = "<a href=\"%s\" >%s</a>\n%s";
+	$template = "<br /><a href=\"%s\" >%s</a>%s<br />";
 
 
 	$arr = array('url' => $url, 'text' => '');
@@ -31,16 +32,33 @@ function parse_url_content(&$a) {
 		killme();
 	}
 
+	logger('parse_url: data: ' . $s, LOGGER_DATA);
 
 	if(! $s) {
 		echo sprintf($template,$url,$url,'');
 		killme();
 	}
 
+	if(strpos($s,'<title>')) {
+		$title = substr($s,strpos($s,'<title>')+7,64);
+		if(strpos($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');
 
@@ -51,26 +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 $item) {
-						if($item->getElementsByTagName('script'))
-							continue;
 						$text = $item->textContent;
+						if(stristr($text,'<script')) {
+							$text = '';
+							continue;
+						}
 						$text = strip_tags($text);
-						if(strlen($text) < 100)
+						if(strlen($text) < 100) {
+							$text = '';
 							continue;
+						}
 						$text = substr($text,0,250) . '...' ;
 						break;
 					}
 				}
 			}
+			if($text)
+				break;
 		}
 	}
 
@@ -78,12 +101,14 @@ function parse_url_content(&$a) {
 		$items = $dom->getElementsByTagName('p');
 		if($items) {
 			foreach($items as $item) {
-				if($item->getElementsByTagName('script'))
-					continue;
 				$text = $item->textContent;
+				if(stristr($text,'<script'))
+					continue;
 				$text = strip_tags($text);
-				if(strlen($text) < 100)
+				if(strlen($text) < 100) {
+					$text = '';
 					continue;
+				}
 				$text = substr($text,0,250) . '...' ;
 				break;
 			}
@@ -91,9 +116,9 @@ function parse_url_content(&$a) {
 	}
 
 	if(strlen($text)) {
-		$text = '<br />' . $text;
+		$text = '<br /><br />' . $text;
 	}
 
-	echo sprintf($template,$url,$title,$text);
+	echo sprintf($template,$url,($title) ? $title : $url,$text);
 	killme();
 }