php 抓取

<?php
set_time_limit(0);

$data = array();
$memKey = 'Outnews_nagovnews';

$url = 'http://www.cqna.gov.cn/html/zjna/jrna/';
$contents = file_get_contents($url);
$contents = iconv('gbk', 'utf-8', $contents);
$res = UtilsExt::run()->cutStr($contents, '<div class="sortlist">', '</table></td>');

$tmp = array();
$exp = explode('<li>', $res);
foreach($exp as $k=>$v){
	if($k>0){
		$tmp['time'] = UtilsExt::run()->cutStr($v, '<span>', '</span>');
		$tmp['title'] = UtilsExt::run()->cutStr($v, 'title="', '"');
		$tmp['url'] = 'http://www.cqna.gov.cn'.UtilsExt::run()->cutStr($v, '<a href="', '"');
		
		$con = file_get_contents($tmp['url']);
		$con = iconv('gbk', 'utf-8', $con);
		$tmp['contents'] = UtilsExt::run()->cutStr($con, '<div id="showcontent">', '</div></td>');
		$tmp['contents'] = str_replace('</div><div>', "\r\n", $tmp['contents']);
		$tmp['contents'] = str_replace('<div>', '', $tmp['contents']);
		$tmp['contents'] = strip_tags(str_replace('</div>', '', $tmp['contents']));
		
		$data[] = $tmp;
	}
}

if(class_exists('Memcached')){
	$memcache = new Memcached;
}else{
	$memcache = new Memcache;
}

$memcache->addServer('127.0.0.1', 11211);
$memcache->set($memKey, $data);

if(!class_exists('Memcached')) $memcache->close();
die;

class UtilsExt{
	public static function run($class=__CLASS__){
		return new $class();
	}
	
	public function cutStr($str, $begin, $end){
		$exp1 = explode($begin, $str);
		$exp2 = explode($end, $exp1[1]);
		
		return $exp2[0];
	}
}
?>