采集中国代理服务器网

php代码

<?php
/**
 * 采集中国代理服务器网 最新列表
 */
class proxy
{
    /* 需采集列表   */
    public $list;

    /* 代理列表 保存路径 */
    public $save_path = &#39;proxy.txt&#39;;

    /* 获取采集列表 */
    function get_list($page)
    {
        $url = &#39;http://www.cnproxy.com/proxy(*).html&#39;;

        // 处理列表
        $this->list = preg_replace(&#39;/\(\*\)/&#39;, $page, $url);
        return $this->list;
    }

    /* 采集代理内容 */
    function get($page)
    {
        $this->get_list($page);
        $file = stripslashes(file_get_contents($this->list));
        $zz = &#39;/<tr><td>([0-9\.]+)<SCRIPT type=text\/javascript>document\.write\(":"([\+a-z]+)\)<\/SCRIPT><\/td><td>[\w]+<\/td><td>[\w\,]+<\/td>/is&#39;;
        preg_match_all($zz, $file, $temp);
        unset($temp[0]);
        $th = array(&#39;z&#39;, &#39;m&#39;, &#39;k&#39;, &#39;l&#39;, &#39;d&#39;, &#39;x&#39;, &#39;i&#39;, &#39;w&#39;, &#39;q&#39;, &#39;b&#39;);
        $th2 = array(3, 4, 2, 9, 0, 5, 7, 6, 8, 1);
        foreach ($temp[2] as $k=>$v) {
            $v = preg_replace("/[\+]+/", &#39;&#39;, $v);
            $s = str_replace($th, $th2, $v);
            $re .= $temp[1][$k] . &#39;:&#39; . $s . "\r\n";
        }
        $this->save($re);
        return true;
    }

    /* 保存 */
    function save($re)
    {
        return file_put_contents($this->save_path, $re, FILE_APPEND);
    }

    /* 读取 */
    function read()
    {
        return file_get_contents($this->save_path);
    }

}

// 初始化采集类
$p = new proxy;

$start = 1;
$end = 10;


// 控制
if($_GET[&#39;a&#39;] == &#39;start&#39;) {
    echo &#39;正在发送采集请求&#39;;
    echo &#39;<meta http-equiv="Refresh" content="3;URL=?p=1">&#39;;
} elseif(isset($_GET[&#39;p&#39;])) {
    $i = $_GET[&#39;p&#39;]++;
    if($i >= $end+1) {
        exit(&#39;<meta http-equiv="Refresh" content="0;URL=?a=end">&#39;);
    } else {
        echo &#39;正在请求列表 &#39;. $i .&#39; > &#39;. $end;
        if($p->get($i)) {
            echo &#39;<meta http-equiv="Refresh" content="3;URL=?p=&#39;.$_GET[&#39;p&#39;]++.&#39;">&#39;;
        }
    }
}  elseif($_GET[&#39;a&#39;] == &#39;end&#39;) {
    echo &#39;采集完毕&#39;;
} else {
    echo &#39;<form>
            <input type="hidden" name="a" value="start" />
            <input type="submit" value="开始采集" />
          </form>&#39;;
}
?>