开源中国个人帐号信息抓取实例

<?php
//需要调用到php包，从oschina中检索下载则可
include "Snoopy.class.php";  // 抓取网页信息， 支持通过http代码方式，
include "simple_html_dom.php";  //html分析包

$snoopy = new Snoopy;
// $snoopy->proxy_host="###";
// $snoopy->proxy_port="8888";
// $snoopy->user="***";
// $snoopy->pass="****";
$snoopy->agent = "(Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:24.0) Gecko/20131117 Firefox/24.0 PaleMoon/24.1.2)";
$snoopy->referer = "http://www.oschina.net/";

$snoopy->cookies["oscid"] = '******';


$snoopy->fetch("https://www.oschina.net/home/login?goto_page=http%3A%2F%2Fwww.oschina.net%2F");


//认证  需要认证才能有权限获取到数据
$submit_url = "http://www.oschina.net/action/user/hash_login";
$submit_vars["email"] = "*****";
$submit_vars["pwd"] = SHA1("*******");
$submit_vars["save_login"] = 1;
$submit_vars["submit"] = "现在登录";
$snoopy->submit($submit_url,$submit_vars);
echo "认证情况\n";

// print $snoopy->results;


//认证通过后，开始获取数据
$snoopy->fetch("http://my.oschina.net/******/admin/inbox");
echo "开源中国个人帐号信息测试程序\n";
echo "时间:".date("Y-m-d h:i:s")."\n";
// echo iconv("UTF-8","GBK//IGNORE",$snoopy->results);
$result=$snoopy->results;

// print $result;

//simple_html_dom解释包使用实例
$SHTML=new simple_html_dom();
$html = str_get_html($result);
$messS=$html->find('ul.Msgs li[id]');
foreach($messS as $mess){
	$aT=$SHTML->load($mess);
	echo "===================\n";
	echo $aT."\n";
	echo "-----------\n下面是上面html信息的分析提取处理:\n";
	echo '信息id:  '.$aT->find('li[id]',0)->id.''."\n";
	echo '信息:  '.trim($aT->find('div.msg',0)->plaintext)."\n";
	echo '日期:  '.trim($aT->find('div.bottom span.date',0)->plaintext).' '."\n";
	print("===================\n");
}