php curl模块模拟登录后采集页面实例

$curl = curl_init();
$cookie_jar = tempnam('./tmp','cookie');
curl_setopt($curl, CURLOPT_URL,'http://www.111cn.net/  checkUser.jsp');//这里写上处理登录的界面
curl_setopt($curl, CURLOPT_POST, 1);
$request = 'user=xxx&password=xxx';
curl_setopt($curl, CURLOPT_POSTFIELDS, $request);//传 递数据
curl_setopt($curl, CURLOPT_COOKIEJAR, $cookie_jar);// 把返回来的cookie信息保存在$cookie_jar文件中
curl_setopt($curl, CURLOPT_RETURNTRANSFER, 1);//设定返回 的数据是否自动显示
curl_setopt($curl, CURLOPT_HEADER, false);//设定是否显示头信 息
curl_setopt($curl, CURLOPT_NOBODY, false);//设定是否输出页面 内容
curl_exec($curl);//返回结果
curl_close($curl); //关闭

$curl2 = curl_init();
curl_setopt($curl2, CURLOPT_URL, 'http://www.111cn.net/ aaa.php');//登陆后要从哪个页面获取信息
curl_setopt($curl2, CURLOPT_HEADER, false);
curl_setopt($curl2, CURLOPT_RETURNTRANSFER, 1);
curl_setopt($curl2, CURLOPT_COOKIEFILE, $cookie_jar);
$content = curl_exec($curl2);

$content = strstr($orders,'<div class="products">');//查找首次出现

<div class="products">

<?php

function get_qq_status($uin)
{
 error_reporting(0);
 $f=file_get_contents('http://wpa.qq.com/pa?p=1:'.$uin.':4');
 if(!$f) return(true);
 foreach($http_response_header as $val)
 {
  if(strpos($val,'Content-Length')!==false)
  {
   return(intval(substr($val,16,50))>1000);
  }
 }
}

?>

<?
function tphp_qq_online( $uin )
{
    $reques  = "GET /pa?p=1:".$uin.":1 HTTP/1.1rn";
    $reques .= "Host: wpa.qq.comrn";
    $reques .= "User-Agent: PHP_QQ_SPYrnrn";
 
    if ( !( $socket = socket_create( AF_INET, SOCK_STREAM, SOL_TCP ) ) ) return(-1);
    if ( !( socket_connect( $socket, "wpa.qq.com", 80 ) ) ) return(-1);
    if ( !( socket_write( $socket, $reques ) ) ) return(-1);
    if ( !( $respon = socket_read( $socket, 1024, PHP_BINARY_READ ) ) ) return(-1);;
    socket_close( $socket );
    
    $field = explode( "rn", $respon ); 
    for ( $i=0; $i<count($field); $i++ ) {
        if ( strncasecmp($field[$i], "Location:", 9) == 0 ) {
            if ( strpos( $field[$i], "online") ) {
                $ret = 1;
            } else if ( strpos( $field[$i], "offline") ) {
                $ret = 0;
            } else { 
                $ret = -1;
            } // if
            break;
        } // if
    } // for
    
    return( $ret );
}
/* }}} */
 
 

echo tphp_qq_online( 561272831 );

?>

//获取QQ状态
function getQQState($qq){
    $url ='http://wpa.qq.com/pa?p=2:'.$qq.':41&r=' . time ();
 $headInfo = get_headers($url,1);
 $length = $headInfo['Content-Length'];
 if ($length==1243) {
  return true;
 }else {
  return false;
 }
}
//获取QQ昵称
function getQQNick($qq){
 $str = file_get_contents('http://r.qzone.qq.com/cgi-bin/user/cgi_personal_card?uin='.$qq);
 $pattern = '/'.preg_quote('"nickname":"','/').'(.*?)'.preg_quote('",','/').'/i';
 preg_match ( $pattern,$str, $result );
 return $result[1];
}
//获取QQ姓名
function getQQName($qq){
 //$qqArr = include 'friendArr.php';//预先设置的
 //$username = $qqArr[$qq];
 if (!$username) {
  $username = getQQNick($qq);
 }
 return $username;
}

<?php
$fp = fopen ("templets.html","a");
if ($fp){
$fup = fread ($fp,filesize("templets.html"));
$fp2 = fopen ("html.shtml","w");
if ($fwrite ($fp2,$fup)){
$fclose ($fp);
$fcolse ($fp2);
die ("写入模板成功");
} else {
fclose ($fp);
die ("写入模板失败!");
}
}
?>

<?php
$s_fname = "93e.php";
$o_fname = "93e.htm";
ob_end_clean();
ob_start();
include($s_fname);
$length = ob_get_length();
$buffer = ob_get_contents();
$buffer = eregi_replace("r","",$buffer);
ob_end_clean();
 
$fp = fopen($o_fname,"w+");
fwrite($fp,$buffer);
fclose($fp);
?>

RewriteRule ^post/([a-z0-9-]+.html)$  wp-post.php?$1$2

ob_start();
$qstring = isset($_SERVER["QUERY_STRING"]) ? $_SERVER["QUERY_STRING"] : "";
define("HTML_FILE",  $_SERVER['DOCUMENT_ROOT']."/post/".$qstring);

if (file_exists(HTML_FILE))
{
 $lcft = filemtime(HTML_FILE);
 if (($lcft + 3600) > time())  //判断上次生成HTML文件是否超过1小时，若没有才直接输出文件内容
 {
  echo(file_get_contents(HTML_FILE));
  exit(0);
 }
}

define("HTMLMETA","<!-- this is a real static html file created at ".date("Y-m-d H:i:s")." -->");
$buffer = ob_get_flush();
$fp = fopen(HTML_FILE, "w");
if ($fp)
{
 fwrite($fp, $buffer.HTMLMETA);
 fclose($fp);
}

/**
     * 获取CSS中图片地址,并且保存到本地
     */
    class getInCssImage
    {  
        /**
         *  图片保存下来
         * @param $cssUrl css的url地址
         * @param $dir 保存图片的目录
         * @return void
         */
        static public function saveImage($cssUrl, $dir)
        {  
            $content = file_get_contents($cssUrl); 
            $patterns = '/images(.*).(jpg|gif|png)/'; //正则根据不同地址需要变换
            preg_match_all($patterns, $content, $matches);
            $imagesUrls = $matches[0];
            if (!is_dir($dir))
                mkdir(dirname(__FILE__). '/'. $dir, 0777);
            foreach($imagesUrls as $image)
            {  
                ob_start();
                $imageUrl = "http://www.xx.com/".$image; //这个地址本来用程序给获取的。偷懒了下
                readfile($imageUrl);
                $img  = ob_get_contents();
                ob_end_clean();
                $size = strlen($img);
                $localImage = $dir. strchr($image, '/'); //存到本地的图片地址
                $fp = fopen($localImage, 'a');
                fwrite($fp, $img);
                fclose($fp);
            }  
    }  
        }
    }

$content = getInCssImage::saveImage('/css/css.css', 'image');