Rss парсер - PHP-Fusion SF Russian Support Site



if(substr_count($_SERVER['REQUEST_URI'],'weblinks.php') > 0 ) {

openside("Лента новостей BBC");



$cashe_filename = $_SERVER['DOCUMENT_ROOT']."/bbc_news.txt";



// wrap string by len, save existing words

function wrap_by_len($str, $max_len, $endimiter = '…')

{ 

if( strlen($str) <= ($max_len - strlen($endimiter)) ) return $str ;



$result_str = '';

$words = explode(" ", $str);

$max_len -= strlen($endimiter);



for( $i = 0; strlen($result_str) < $max_len; $i++ )

$result_str .= $words[$i].' ';



$result_str = rtrim($result_str," ,");



return $result_str.$endimiter ;

}



/**********************************************************

* Parse XML data into an array structure *

* Usage: array parse_rss ( string data ) *

**********************************************************/

function parse_rss($reg_exp, $xml_data) {

preg_match_all($reg_exp, $xml_data, $temp);

return array(

'count'=>count($temp[0]),

'title'=>$temp[1],

'link'=>$temp[2],

'desc'=>$temp[3]

);

}



/**********************************************************

* Parse Array data into an HTML structure *

* Usage: string parse_rss ( array data ) *

**********************************************************/

function output_rss($pattern, $rss_data, $counter, $maxlen ) 

{

$result = '';



for($i=0; $i<$rss_data['count'] && $i<$counter; $i++) 

{

$rss_data['desc'][$i] = wrap_by_len($rss_data['desc'][$i], $maxlen);



$result .= sprintf($pattern,

$rss_data['link'][$i],

html_entity_decode($rss_data['title'][$i]),

html_entity_decode($rss_data['desc'][$i])

);

}

return $result ;

}



/**********************************************************

* Settings *

**********************************************************/

$cashe_good_time = 1800; // in seconds

$counter = 6;

$maxlen = 200;

$url = 'http://www.bbc.co.uk/russian/rolling_news/index.xml';



$reg_exp = '#<item>.*?<title>(.*?)<\/title>.*?';

$reg_exp .='<link>(.*?)<\/link>.*?<description>';

$reg_exp .='(.*?)<\/description>.*?<\/item>#si';



$pattern = '<p><noindex><a rel="nofollow" target="_blank" href="%s">%s</a></noindex><br>%s</p>';



/**********************************************************

* Main script *

**********************************************************/



$refresh_cashefile = true ;

$data_is_output = false ;



// check cashe file modifications if file exist nad readalbe

if( file_exists($cashe_filename) )

{

$fileage = time() - filemtime( $cashe_filename ); 



// good cashe state, out cashe content and out here 

if( $fileage < $cashe_good_time ) 

{

$refresh_cashefile = false ;

echo file_get_contents($cashe_filename);

$data_is_output = true ;

}

}



if( !$data_is_output )

{

if( $xml_data = file_get_contents($url) ) 

{

$rss_data = parse_rss($reg_exp, $xml_data);

$content = output_rss($pattern, $rss_data, $counter, $maxlen);



if( extension_loaded('mb_string') ) 

$content = mb_convert_encoding($content, "utf-8", "windows-1251");

elseif( extension_loaded('iconv') ) 

$content = iconv('utf-8', 'windows-1251', $content);



// need refresh cashe file' content

if( $refresh_cashefile )

{

$fp = fopen($cashe_filename, "w"); 



if( $fp )

{

fwrite($fp, $content); 

fclose($fp); 

}

}



// out content

echo $content ;

}

}

closeside();

}