X-Git-Url: https://git.mxchange.org/?a=blobdiff_plain;f=mod%2Fparse_url.php;h=9bb0bc46408e76f3a090d512c3d01fc63bc2bafa;hb=2abcf76ec17a9a7754c399cdde9a4449308a4b02;hp=aa71893ab47ebade0e5043976157f57d174daf59;hpb=0afc71a95a54f6b648a551584c2f7fc7a6c16880;p=friendica.git

diff --git a/mod/parse_url.php b/mod/parse_url.php
index aa71893ab4..9bb0bc4640 100644
--- a/mod/parse_url.php
+++ b/mod/parse_url.php
@@ -1,29 +1,64 @@
 <?php
 
 require_once('library/HTML5/Parser.php');
+require_once('library/HTMLPurifier.auto.php');
 
 function parse_url_content(&$a) {
 
-	$url = trim($_GET['url']);
+	logger('parse_url: ' . $_GET['url']);
 
-	$template = "<a href=\"%s\" >%s</a>%s";
+	$url = trim(hex2bin($_GET['url']));
 
-	if($url) 
+	logger('parse_url: ' . $url);
+
+	$text = null;
+
+	$template = "<br /><a href=\"%s\" >%s</a>%s<br />";
+
+
+	$arr = array('url' => $url, 'text' => '');
+
+	call_hooks('parse_link', $arr);
+
+	if(strlen($arr['text'])) {
+		echo $arr['text'];
+		killme();
+	}
+
+	if($url) {
 		$s = fetch_url($url);
-	else {
+	} else {
 		echo '';
 		killme();
 	}
-	
+
+	logger('parse_url: data: ' . $s, LOGGER_DATA);
+
 	if(! $s) {
 		echo sprintf($template,$url,$url,'');
 		killme();
 	}
 
-	$dom = HTML5_Parser::parse($s);
+	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);
 
-	if(! $dom)
-		return $ret;
+	$purifier = new HTMLPurifier($config);
+	$s = $purifier->purify($s);
+
+//	logger('parse_url: purified: ' . $s, LOGGER_DATA);
+
+	$dom = @HTML5_Parser::parse($s);
+
+	if(! $dom) {
+		echo sprintf($template,$url,$url,'');
+		killme();
+	}
 
 	$items = $dom->getElementsByTagName('title');
 
@@ -34,22 +69,56 @@ function parse_url_content(&$a) {
 		}
 	}
 
-	$items = $dom->getElementsByTagName('p');
-	if($items) {
-		foreach($items as $item) {
-			$text = $item->textContent;
-			$text = strip_tags($text);
-			if(strlen($text) < 100)
-				continue;
-			$text = substr($text,0,250) . '...' ;
-			break;
+	$divs = $dom->getElementsByTagName('div');
+	if($divs) {
+		foreach($divs as $div) {
+			$class = $div->getAttribute('class');
+			if($class && (stristr($class,'article') || stristr($class,'content'))) {
+				$items = $div->getElementsByTagName('p');
+				if($items) {
+					foreach($items as $item) {
+						$text = $item->textContent;
+						if(stristr($text,'<script')) {
+							$text = '';
+							continue;
+						}
+						$text = strip_tags($text);
+						if(strlen($text) < 100) {
+							$text = '';
+							continue;
+						}
+						$text = substr($text,0,250) . '...' ;
+						break;
+					}
+				}
+			}
+			if($text)
+				break;
+		}
+	}
+
+	if(! $text) {
+		$items = $dom->getElementsByTagName('p');
+		if($items) {
+			foreach($items as $item) {
+				$text = $item->textContent;
+				if(stristr($text,'<script'))
+					continue;
+				$text = strip_tags($text);
+				if(strlen($text) < 100) {
+					$text = '';
+					continue;
+				}
+				$text = substr($text,0,250) . '...' ;
+				break;
+			}
 		}
 	}
 
 	if(strlen($text)) {
-		$text = '<br />' . $text;
+		$text = '<br /><br /><blockquote>' . $text . '</blockquote><br />';
 	}
 
-	echo sprintf($template,$url,$title,$text);
+	echo sprintf($template,$url,($title) ? $title : $url,$text);
 	killme();
-}
\ No newline at end of file
+}