gistfile1.aw

<?php
//парсинг баша.орг.ру

#error_reporting(0);
set_time_limit(0);
ini_set('memory_limit', '512M');

$db = new SQLite3('bash');

#$db->query('DELETE FROM quotes;');

for ($i = 1; $i <= 100; $i++){
	$url = 'http://bash.im/byrating/' . $i;
	if ( url_exists($url) ){
		$content = file_get_contents($url);
		
		preg_match_all('/<span id="v([0-9]+)" class="rating">([0-9]+)<\/span>/i', $content, $matches);
		$ids = $matches[1];
		$ratings = $matches[2];
		
		preg_match_all('/<div class="text">(.*)<\/div>/im', $content, $matches);
		$texts = $matches[1];
		
		preg_match_all('/<span class="date">(.*)<\/span>/i', $content, $matches);
		$text_dates = $matches[1];
		
		foreach($ids as $key => $id){
			//заменяем html сущности
			$texts[$key] = str_replace(array('<br />', '<br/>', '<br>'), "\n", $texts[$key]);
			$texts[$key] = str_replace(array('&quot;', '&lt;', '&gt;', '&amp;', '&#8206;', '&#039;'), array('"', '<', '>', '&', '', '\''), $texts[$key]);
		
			$sql = 'INSERT INTO quotes(id, text, date, rating) VALUES (\'' . $id . '\', \'' . sqlite_escape_string( iconv('windows-1251', 'utf-8', $texts[$key]) ) . '\', \'' . strtotime($text_dates[$key]) . '\', \'' .$ratings[$key] . '\' );';
			
			$db->query( $sql );
		}

	}
}
$db->close();
echo 'finished!';

function url_exists($url) {
    $file_headers = @get_headers($url);
	return $file_headers[0] == 'HTTP/1.1 200 OK';		
}