从网页提取代理服务器并验证是否可用代码,PHP代码库,德仔网

代码语言

代码分类

【PHP】从网页提取代理服务器并验证是否可用

作者:灏明 / 发布于2014/12/10/ 644


	<?php
	include 'simple_html_dom.php';
	
	//ini_set('default_socket_timeout', 2);
	
	$url = 'http://cn-proxy.com/';
	
	$html = file_get_html($url);
	
	$ret = $html->find('table.sortable tr');
	
	//echo 'count(ip):', count($ret), PHP_EOL;
	
	$ips = array();
	
	foreach($ret as $key => $value)
	{
	 $tds = $value->find('td');
	 if(count($tds) < 2)
	 {
	 continue;
	 }
	 $ip = $tds[0]->innertext;
	 $port = $tds[1]->innertext;
	
	 if(filter_var($ip, FILTER_VALIDATE_IP))
	 {
	 $ips[] = array('ip' => $ip, 'port' => $port);
	 }
	}
	
	//var_dump($ips);
	
	echo '找到代理', count($ips), '个.', PHP_EOL;
	
	$validate = array();
	
	foreach($ips as $value)
	{
	 if(checkProxy($value['ip'], $value['port']))
	 {
	 $validate[] = $value;
	 }
	}
	
	echo '可用代理', count($validate), '个', PHP_EOL;
	
	foreach($validate as $value)
	{
	 visit($value['ip'], $value['port'], 'http://www.actself.me/');
	}
	
	echo 'done.', date('Y-m-d H:i:s') , PHP_EOL;
	
	function checkProxy($ip, $port)
	{
	 echo '验证 ', $ip, ":", $port, ' ', date('Y-m-d H:i:s');
	 $fd = @stream_socket_client('tcp://' . $ip . ':' . $port, $errno, $errstr, 2);
	
	 echo $fd !== false ? ' 可用' : ' 不可用';
	 echo date('Y-m-d H:i:s'), PHP_EOL;
	
	 if($fd !== false)
	 {
	 fclose($fd);
	 return true;
	 }
	 /*
	 $fd = socket_create(AF_INET, SOCK_STREAM, SOL_TCP);
	 if($fd === false)
	 {
	 echo 'socket_create failed.', PHP_EOL;
	 return false;
	 }
	 $ret = socket_connect($fd, $ip, $port);
	 socket_close($fd);
	 */
	 return false;
	}
	
	function visit($ip, $port, $url)
	{
	 $h = curl_init($url);
	 //curl_setopt($h, CURLOPT_PROXYTYPE, CURLPROXY_SOCKS5);
	 curl_setopt($h, CURLOPT_PROXYTYPE, CURLPROXY_HTTP);
	 //curl_setopt($h, CURLOPT_PROXY, '127.0.0.1:1080');
	 curl_setopt($h, CURLOPT_PROXY, $ip . ':' . $port);
	 curl_setopt($h, CURLOPT_CONNECTTIMEOUT, 2);
	 curl_setopt($h, CURLOPT_TIMEOUT, 6);
	 curl_setopt($h, CURLOPT_RETURNTRANSFER, true);
	
	 $ret = curl_exec($h);
	
	 echo $ret === false ? 'curl exec failed.' : 'curl exec successed.';
	 echo PHP_EOL;
	 if($ret === false)
	 {
	 echo curl_error($h), PHP_EOL;
	 }
	
	 curl_close($h);
	}

试试其它关键字

同语言下

可能有用的

灏明贡献的其它代码(23)