PHP小偷关于抓取页面不同编码的解决方法！-PHP教程-爱易网页

日期：2012-05-15　浏览次数：20480 次

<?
function get_sub_content($str, $start, $end){
if ( $start == '' $end == '' ){
return "页面元素已经改变！";
}
$str = explode($start, $str);
$str = explode($end, $str[1]);
return $str[0];
}
function my_encoding($data,$to){
$encode_arr = array('UTF-8','ASCII','GBK','GB2312','BIG5','JIS','eucjp-win','sjis-win','EUC-JP');
$encoded = mb_detect_encoding($data, $encode_arr);
$data = mb_convert_encoding($data,$to,$encoded);
return $data;
}
$doc = file_get_contents("http://video.baidu.com/v?ct=0&word=周杰伦%20site%3Awww%2Etudou%2Ecom&db=0&ty=0&rn=20&pn=0&fbl=1024");
$doc = my_encoding($doc,"utf-8");
$doc =get_sub_content($doc,"<div id=\"result\">","<br clear=");
$str_replace = explode("<div class=x>",$doc);
echo "<?xml version=\"1.0\" encoding=\"UTF-8\"?>";
echo "<data>";
for ($i=1; $i<=count($str_replace)-1; $i++){
echo "<video>";
echo "<name>";
echo "<![CDATA[".get_sub_content($str_replace[$i],"title=\"","\"")."]]>";
echo "</name>";
echo "<pageurl>";
echo "<![CDATA[".get_sub_content($str_replace[$i],"<a href=\"","\" onmousedown=")."]]>";
echo "</pageurl>";
echo "</video>";
}
echo "</data>";
?>