get_ip(); $this->check_ip($ips); echo "休眠30s" . PHP_EOL; sleep(30); } // $ips=Proxy::select(); } private function check_ip($ip_arr) { foreach ($ip_arr as $key => $value) { $response = $this->check_proxy_ip_info($value); $ips = explode(":", $value); //将结果存入数据库 if ($response) { $insert = [ 'ip' => $ips[0], 'port' => $ips[1], 'status' => 1 ]; Proxy::replace($insert, ['ip' => $ips[0], 'port' => $ips[1]]); echo $value . "\tsuccess " . PHP_EOL; } else { Proxy::update(['status' => 0], ['ip' => $ips[0], 'port' => $ips[1]]); echo $value . "\terror " . PHP_EOL; } } } //获取各大网站代理IP private function get_ip() { $ip_arr = []; echo "正在获取第一个IP列表" . PHP_EOL; $ip_arr = $this->get_xici_ip($ip_arr); echo "正在获取第二个IP列表" . PHP_EOL; $ip_arr = $this->get_kuaidaili_ip($ip_arr); return $ip_arr; } private function get_xici_ip($ip_arr) { for ($i = 1; $i <= 10; $i++) { echo sprintf("正在获取xici第%s页IP列表", $i) . PHP_EOL; $url = sprintf("http://www.xicidaili.com/nn/%s", intval($i)); $infoContent = Curl::get($url); $this->convert_encoding($infoContent); preg_match_all('/[\s\S]*?[\s\S]*?<\/td>[\s\S]*?(.*?)<\/td>[\s\S]*?(.*?)<\/td>/', $infoContent, $match); $host_arr = $match[1]; $port_arr = $match[2]; foreach ($host_arr as $key => $value) { $ip_arr[] = $host_arr[$key] . ":" . $port_arr[$key]; } sleep(2); } print_r($ip_arr); return $ip_arr; } private function get_kuaidaili_ip($ip_arr) { for ($i = 1; $i <= 10; $i++) { echo sprintf("正在获取kuaidaili第%s页IP列表", $i) . PHP_EOL; $url = 'https://www.kuaidaili.com/free/inha/' . $i; $infoContent = Curl::get($url); $this->convert_encoding($infoContent); preg_match_all('/(.*?)<\/td>/', $infoContent, $host); preg_match_all('/(.*?)<\/td>/', $infoContent, $port); $host_arr = $host[1]; $port_arr = $port[1]; foreach ($host_arr as $key => $value) { $ip_arr[] = $host_arr[$key] . ":" . $port_arr[$key]; } sleep(2); } print_r($ip_arr); return $ip_arr; } /** * +----------------------------------------------------------------------------- * 检查代理ip信息有效性 * +----------------------------------------------------------------------------- * @param string $proxy_ip [117.95.100.126:8998] * @param int $times 执行检查次数 * @return bool * @author elinx <654753115@qq.com> 2016-07-29 * +----------------------------------------------------------------------------- */ function check_proxy_ip_info($proxy_ip = '', $times = 1) { $status = false; $header = array( // "GET / HTTP/1.1", // "HOST: www.baidu.com", "accept: application/json", "accept-encoding: gzip, deflate", "accept-language: en-US,en;q=0.8", "content-type: application/json", "user-agent: Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/51.0.2704.106 Safari/537.36", ); $url = 'http://www.baidu.com/'; $result['succeed_times'] = 0; //成功次数 $result['defeat_times'] = 0; //失败次数 $result['total_spen'] = 0; //总用时 $ip = explode(":", $proxy_ip); for ($i = 0; $i < $times; $i++) { $curl = curl_init(); curl_setopt($curl, CURLOPT_URL, $url); //设置传输的url curl_setopt($curl, CURLOPT_HTTPHEADER, $header); //发送http报头 curl_setopt($curl, CURLOPT_RETURNTRANSFER, 1); curl_setopt($curl, CURLOPT_ENCODING, 'gzip,deflate'); // 解码压缩文件 curl_setopt($curl, CURLOPT_SSL_VERIFYPEER, false); //不验证证SSL书 curl_setopt($curl, CURLOPT_SSL_VERIFYHOST, false); //不验证SSL证书 curl_setopt($curl, CURLOPT_HTTPHEADER, array( 'Client_Ip: ' . $ip, )); curl_setopt($curl, CURLOPT_HTTPHEADER, array( 'X-Forwarded-For: ' . $ip, )); curl_setopt($curl, CURLOPT_PROXYTYPE, CURLPROXY_HTTP); curl_setopt($curl, CURLOPT_PROXY, $proxy_ip); curl_setopt($curl, CURLOPT_COOKIEFILE, dirname(__FILE__) . '/cookie.txt'); curl_setopt($curl, CURLOPT_COOKIEJAR, dirname(__FILE__) . '/cookie.txt'); curl_setopt($curl, CURLOPT_TIMEOUT, 10); // 设置超时限制防止死循环 // $response_header = curl_getinfo($curl); // 获取返回response报头 $content = curl_exec($curl); if (strstr($content, '百度一下，你就知道')) { $status = true; break; } else { $status = false; } // $result['list'][$i]['response_header'] = $response_header; } return $status; } private function convert_encoding(&$str) { $encode = mb_detect_encoding($str, ['GB2312', 'UTF-8', 'GBK', 'ASCII']); $str = iconv($encode, 'utf-8//TRANSLIT//IGNORE', $str); } }