skweez.net

<?php
// Heise Feed Version 0.4
header("Content-type: text/xml; charset=UTF-8");
define('CACHEFOLDER','feed');
define('FEEDINTERVAL',600);
define('MAXARTIKELS',100);
 
$feedUrl = urlencode('http://www.heise.de/newsticker/heise-atom.xml');
 
if( $_GET["url"] ) {
        $feedUrl = urlencode($_GET["url"]);
}
 
if( $_GET["do"]=="reload" ) $do_reload = true;
if( $_GET["do"]=="sync" ) $do_sync = true;
 
# Do not diplay xml errors
libxml_use_internal_errors(true);
 
function getID($url) {
        preg_match('/-([0-9]+)\.html/', $url, $id);
        return $id[1];
}
 
function getArtikel($url, $id, $date, $forceReload) {
        if(!$forceReload and file_exists(CACHEFOLDER."/".$date."-".$id.".txt")) {
                $artikelFertig = file_get_contents(CACHEFOLDER."/".$date."-".$id.".txt");
        } else {
                $doc = new DOMDocument();
                $artikel = mb_convert_encoding(file_get_contents($url), 'HTML-ENTITIES', "UTF-8");
                $doc->loadHTML($artikel);
                $divs = $doc->getElementsByTagName('div');
 
                $artikelFertig = 'Konnte Artikelinhalt nicht finden.';
 
                $newDoc = new DOMDocument();
 
                foreach ( $divs as $div ) {
                        if( $div->hasAttribute('class') && $div->getAttribute('class') == 'meldung_wrapper' ) {
                                $newDoc->appendChild($newDoc->importNode($div,true));
                        }
                }
                $artikelFertig = $newDoc->saveHTML();
                $artikelFertig = strip_tags($artikelFertig, '<span><a><pre><b><br><em><ul><li><hr><p><img><strong><table><tbod<><td><tr>');
                $artikelFertig = str_replace("\"/","\"http://www.heise.de/",$artikelFertig);
                $artikelFertig = htmlspecialchars($artikelFertig);
 
                if(!file_put_contents(CACHEFOLDER."/".$date."-".$id.".txt",$artikelFertig)) die();
        }
        $files = glob(CACHEFOLDER."/*.txt");
        while(count($files) > MAXARTIKELS) {
                unlink($files[2]);
                $files = glob(CACHEFOLDER."/*.txt");
        }
        return $artikelFertig;
}
 
$cacheFile = CACHEFOLDER."/".$feedUrl.".txt";
 
if($do_reload or $do_sync or !file_exists($cacheFile) or time() - filemtime($cacheFile) > FEEDINTERVAL) {
        $xml = DOMDocument::load(urldecode($feedUrl));
        $entrys = $xml->getElementsByTagName('entry');
 
        $xml->getElementsByTagName('link')->item(1)->attributes->item(1)->nodeValue="http://example.com/feed.php";
 
        for ($i=0; $i < $entrys->length; $i++) {
                $entry = $entrys->item($i);
                $url = str_replace('/from/atom10', '', $entry->getElementsByTagName('id')->item(0)->nodeValue);
                $id = getID($url);
                $date = $entry->getElementsByTagName('updated')->item(0)->nodeValue;
 
                # Set unique id.
                $entry->getElementsByTagName('id')->item(0)->nodeValue = 'http://www.heise.de/'.$id;
 
                # Create content element and fill it with content
                $content = $xml->createElement('content');
                $entry->appendChild($content);
 
                $entry->getElementsByTagName('content')->item(0)->setAttribute('type','html');
 
                $forceReload = false;
                if($do_reload or preg_match("/update/i",$entry->getElementsByTagName('title')->item(0)->nodeValue)) {
                        $forceReload = true;
                }
 
                $artikelInhalt = getArtikel($url, $id, $date, $forceReload);
                $entrys->item($i)->getElementsByTagName('content')->item(0)->nodeValue = $artikelInhalt;
        }
        $feed = $xml->saveXML();
        file_put_contents($cacheFile, $feed);
        echo $feed;
} else {
        echo file_get_contents($cacheFile);
}
 
# Clear xml errors
libxml_clear_errors();
?>