Public feed #600134

Unnamed

public php v1 · immutable
Raw Edit
·published 2007-07-02 17:25 UTC
rendered paste body
<?phpinclude ('./crawler.conf');include ('./class.database.php');$database = new database();function insert_link($url) {	global $database;	$query = $database->query("SELECT count(url) FROM `queue` WHERE `url` = '".$url."'") ;	$result = $query->fetchRow() ;	if ($result[0] < 1) {		$database->query("INSERT INTO `queue` (`url`, `time`) VALUE ('".$url."', ".time().")") ;		echo "INSERT > ".$url."\n";		return true;	}	echo "ALREADY EXIST > ".$url."\n";	return false;}function processed($url, $status) {	global $database;	$query = $database->query("UPDATE `queue` SET `status` = '".$status."' WHERE `url` = '".$url."' LIMIT 1");}function search_links($url) {	global $database;	echo "SEARCHING LINKS FROM : ".$url."...\n";	if ($str = file_get_contents($url)) {		$motif = "`href=(\"|\')([^\"\']*)`";		preg_match_all($motif, $str, $out);		$count=count($out[0]);		if ($count > 0) {			/*print_r($out);*/			for($i=0;$i<$count;$i++)			{				if (!preg_match("`\:\/\/`", $out[2][$i])) {					$out[2][$i] = $url.$out[2][$i];				}					insert_link($out[2][$i]);				processed($out[2][$i], 0);			}		}	processed($url, 1);	}	else {		insert_link($url);		processed($url, 2);	}	}$url = "http://daxey.net";if (!insert_link($url)) {	$query = $database->query("SELECT url FROM queue WHERE status = 0 ORDER BY Rand() LIMIT 1");	$result = $query->fetchRow() ;	$url = $result[0];}while(1) {	$query = $database->query("SELECT url FROM queue WHERE status = 0 ORDER BY Rand() LIMIT 1");	$result = $query->fetchRow() ;	$url = $result[0];	search_links($url);}?>