<?php
/*
PROTOTYPE DEFINITION:
function proxy_test( $proxies, $test_url, $maxtime = 5, $test_expression = '' )

	This tests an array of proxies 
	The array needs to list the proxies  in XXX.XXX.XXX.XXX:PORT format

	$proxies		- array of proxies in XXX.XXX.XXX.XXX:PORT format
	$test_url 		- url to test the proxy with ie 'http://www.google.com'
	$maxtime		- maximum time in seconds that a proxy is allowed to respond, if it returns later than this don't list it
	$test_expression 	- regular expression to test the returned text from the webpage to ensure its valid

	returns array of proxies that meet the test guidelines
	in the following format XXX.XXX.XXX.XXX:PORT


function samair_proxies()
	This function grabs all the proxies from SAMAIR
	and returns then in an array indexed by PORT the proxies listen too

	example returned array

	Array(
		[80] => Array( 
			'10.10.10.111:80',
			'190.10.10.111:80',
			'211.10.10.111:80',
			'20.10.10.111:80'
		),
		[8080] => Array( 
			'10.10.10.111:8080',
			'190.10.10.111:8080',
			'211.10.10.111:8080',
			'20.10.10.111:8080'
		)
	)	

*/


$untested_proxies = samair_proxies();
$proxies = array();

foreach( $untested_proxies as $testlist )
{

	$proxies = array_merge( $proxies,  proxy_test( $testlist, 'http://google.com', 5 ));
}

print_r( $proxies  );


/* 
Function Definitions 
------------------------------------------
*/


// This function grabs all the proxies from SAMAIR
// and returns then in an array indexed by PORT the proxies listen too

	
	function samair_proxies()
	{
// grab and output the SAMIAR proxy lists
		$all_proxies = array();

		$pages = 20;

		for( $i = 0; $i < $pages; $i++ )
		{
			$url = 'http://www.samair.ru/proxy/proxy-' . $i . '.htm';
	
			$ch = curl_init();
			curl_setopt( $ch, CURLOPT_URL, $url  );
			curl_setopt( $ch, CURLOPT_USERAGENT, '' );
			curl_setopt( $ch, CURLOPT_RETURNTRANSFER, 1 );
//			curl_setopt( $ch, CURLOPT_PROXY, '' );

			$data = curl_exec( $ch );
			$data = strip_tags( $data );

// chop the top

			preg_match_all( '/\d+\.\d+\.\d+\.\d+:\d+/', $data, $these_proxies );
			$all_proxies = array_merge( $all_proxies, $these_proxies[0] );
		}

		$proxies_count = count( $all_proxies );
		$proxies_by_port = array();


		for( $i = 0; $i < $proxies_count; $i++ )
		{
			$proxy = explode( ':', $all_proxies[$i] );

			if( $proxies_by_port[ $proxy[1] ] )
			{
				array_push( $proxies_by_port[$proxy[1]] , $all_proxies[$i] );
			} else {
				$proxies_by_port[ $proxy[1]  ] = array();
				array_push( $proxies_by_port[ $proxy[1] ] , $all_proxies[$i] );
			}
		}
	
		return $proxies_by_port;
	} 
 

/*

This tests an array of proxies 
The array needs to list the proxies  in XXX.XXX.XXX.XXX:PORT format

$proxies		- array of proxies in XXX.XXX.XXX.XXX:PORT format
$test_url 		- url to test the proxy with ie 'http://www.google.com'
$maxtime		- maximum time in seconds that a proxy is allowed to respond, if it returns later than this don't list it
$test_expression 	- regular expression to test the returned text from the webpage to ensure its valid

*/

	function proxy_test( $proxies, $test_url, $maxtime = 5, $test_expression = '' )
	{
		if( is_array( $proxies ) )
		{
// array of proxie sips and ports as expected
		} else {
// scalar 
// check for value and push onto array
			if( $proxies )
			{
				$temp = $proxies;
				$proxies = array();
				array_push( $proxies, $temp );
			}	
		}

// store proxies that meet out criteria
		$sat_proxies = array();

		foreach( $proxies as $ip )
		{
			$ch = curl_init();
			curl_setopt( $ch, CURLOPT_URL, $test_url );
			curl_setopt( $ch, CURLOPT_USERAGENT, 'USER AGENT HERE' );
			curl_setopt( $ch, CURLOPT_RETURNTRANSFER, 1 );
			curl_setopt( $ch, CURLOPT_CONNECTTIMEOUT, $maxtime ); // time to wait for response set to 5 seconds initially, a bit high IMO

// proxy to test
			curl_setopt( $ch, CURLOPT_PROXY, $ip );

			$time_start = time();
			$data = curl_exec( $ch );
			$time_finish = time();
			$response_time = $time_finish - $time_start;


			$good_proxy = 0;
			if( $response_time < ( $maxtime + 1 ) )
			{

// if a regular expression was passed 
// we need to test for it in the data returned
// this ensures that the proxy inst' just sending
// a default message page back
				if( $test_expression )
				{
					if( preg_match( $test_expresion, $data ) )
					{
//  response time was good and data was found on page 
						$good_proxy = 1;
					}
				} else {
//  response time was good
					$good_proxy = 1;
				}
			}

			if( $good_proxy )
			{
				array_push( $sat_proxies, $ip );
			}
		}


		return $sat_proxies;
	}


?>