gistfile1.txt

<?php
set_time_limit(111111);

$db_host = "localhost";
$db_user = "root";
$db_password = "";
$db_dbname = "spr";
$db_tablename = "data";

$log_file = "log.txt";
$f = fopen($log_file, "w+");


mysql_connect($db_host, $db_user, $db_password) or die("Can't connect to mysql");
mysql_select_db($db_dbname) or die("Can't connect to database $db_dbname");

@mysql_query("DROP TABLE $db_tablename");
mysql_query("CREATE TABLE $db_tablename(
			id INT AUTO_INCREMENT PRIMARY KEY,
			section VARCHAR(100),
			subsection VARCHAR(100),
			name VARCHAR(500),
			address VARCHAR(500),
			timetable VARCHAR(500),
			site VARCHAR(100))
			DEFAULT CHARSET=utf8 COLLATE=utf8_unicode_ci");

function getPage($url)
{
	$ch = curl_init (); 
	curl_setopt ($ch , CURLOPT_URL , $url);
	curl_setopt ($ch , CURLOPT_USERAGENT , "Mozilla/5.0 (Windows; U; Windows NT 5.1; ru-RU; rv:1.7.12) Gecko/20050919 Firefox/1.0.7"); 
	curl_setopt ($ch , CURLOPT_RETURNTRANSFER , 1 );
	curl_setopt ($ch,CURLOPT_HTTPHEADER,array ("Content-Type: text/xml; charset=windows-1251","Expect: 100-continue"));
	$content = curl_exec($ch);
	$content = iconv("cp1251", "UTF-8", $content);
	curl_close($ch); 
	return $content;
}

function addItemToBase($section_name, $subsection_name, $name, $address, $timetable, $site)
{
	global $db_tablename;
	$query = "INSERT INTO ".$db_tablename. " SET section='".mysql_escape_string($section_name)."', 
				subsection='".mysql_escape_string($subsection_name)."',  
				name='".mysql_escape_string($name)."',  
				address='".mysql_escape_string($address)."', 
				timetable='".mysql_escape_string($timetable)."',  
				site='".mysql_escape_string($site)."'";
	mysql_query($query) or die(mysql_error());
}

function parseItem($url, $section_name, $subsection_name)
{
	$url = "http://www.spr.ru/".$url.".html";
	$name = $address = $timetable = $site = "";
	$text = getPage($url);
	preg_match("#<H1 class='main' style='font-size:24px;'>(.+?)</H1>#sui", $text, $name);
	$name = str_replace("&nbsp", "", htmlspecialchars_decode(strip_tags($name[1])));
	preg_match("#<font class='view-orange'>Адрес</font>&nbsp;<font class='view-font'>(.+?)&nbsp;&nbsp;<noindex>#sui", $text, $address);
	$address = str_replace("&nbsp", "", htmlspecialchars_decode(strip_tags($address[1])));
	preg_match("#<font class='view-orange'>Часы работы</font><font class='font'>&nbsp;</font><font class='view-font'>(.+?)</font>#sui", $text, $timetable);
	$timetable = $timetable[1];
	preg_match("#<font class='view-orange'>Сайт</font><font class='font'>&nbsp;</font><font class='view-font'>(.+?)</font>#sui", $text, $site);
	$site = $site[1];

	addItemToBase($section_name, $subsection_name, $name, $address, $timetable, $site);
}

function parseCity($url)
{
	global $f;
	
	$text = getPage($url);
	preg_match_all("#<noindex><a rel='nofollow' href='(.+?)' title='(.+?)' style='font-size:14px;'>(.+?)</a></noindex><br>#sui", $text, $sections, PREG_SET_ORDER);
	foreach($sections as $section)
	{
		$section_url = $section[1];
		$section_name = $section[3];
	
		$section_text = getPage($section_url);
		
		preg_match_all("#<font style='font-size:16px; color:222222;'>(.+?)<br><br>#sui", $section_text, $subsections_content, PREG_SET_ORDER);
		foreach($subsections_content as $subsection_item)
		{
			preg_match_all("#<a href='(.+?)' title='(.+?)' style='line-height:17px; font-size:14px;'>(.+?)</a>#sui", $subsection_item[1], $subsections, PREG_SET_ORDER);
			foreach($subsections as $subsection)
			{
				$subsection_url = $subsection[1];
				$subsection_name = $subsection[3];
				
				$subsection_text = getPage($subsection_url);
				preg_match_all('#firms.push\(\["(.+?)","(.+?)","(.+?)",(.+?)\]\);#sui', $subsection_text, $items, PREG_SET_ORDER);
				$count += count($items);
				foreach($items as $item)
					parseItem($item[3], $section_name, $subsection_name);
			}
		}
	}
	fwrite($f, $url."---->".$count);
}

$text = getPage("http://www.spr.ru/");
preg_match('#<ul class="Control">(.+?)</ul>#sui', $text, $regions_content);
preg_match_all('#<li><a href="(.+?)" rel=\'nofollow\'>(.+?)</a></li>#sui', $regions_content[0], $regions, PREG_SET_ORDER);
foreach($regions as $region)
{
	$url = $region[1];
	$url = "http://www.spr-cfo.ru/";
	$text = getPage($url);
	preg_match_all("#<td valign='top' width='95%'>(.+?)</td>#sui", $text, $temp, PREG_SET_ORDER);
	foreach($temp as $letter)
	{
		preg_match_all("#<a href='(.+?)' title='(.+?)'>(.+?)</a><font style='color:676d74;'>(.+?)</font>#sui", $letter[1], $elems, PREG_SET_ORDER);
		foreach($elems as $elem)
			$sections[] = $elem[1];
	}
	preg_match_all("#<font class='list'>(.+?)</font>#sui", $text, $sections2, PREG_SET_ORDER);
	foreach($sections2 as $section)
	{
		preg_match_all("#<a href='(.+?)' title='(.+?)'>(.+?)</a><br>#sui", $section[1], $items, PREG_SET_ORDER);
		foreach($items as $item)
			$sections[] = $item[1];
	}
	
	foreach($sections as $section)
		parseCity($section);
	exit();
}

fclose($f);

?>