<?php
require_once('library/HTML5/Parser.php');
+require_once('library/HTMLPurifier.auto.php');
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' => '');
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');
}
}
-
$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;
}
}
$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;
}
}
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();
}