MaGog to RSS

magogrss.php

<?php

/**
 * MaGog to RSS script for game update notifications.
 *
 * A script to scrape results from the
 * {@link [http://www.an-ovel.com/pages/magog.php] [<MaGog GOG search engine>]} and convert them
 * into an RSS feed. 
 * 
 * @author Kiran Welle
 */

/**
 * Specify the MaGog snapshot GUID to use.
 */
if( $_REQUEST['snap'] !== null ) {
	$MAGOG_SNAPSHOT = $_REQUEST['snap'];
} else {
	die("Please specify a MaGog snapshot to check.");
}

/**
 * Show games updated within this time period.
 */
$DAYS_SINCE_UPDATE = 1;

/**
 * Replace special characters to clean up the game titles.
 */
function clean_game_title( $title ) {
	$title = str_replace("®", "&#169;", htmlspecialchars($title));
	return mb_convert_encoding( $title, "UTF-8" );
}

/*
function guid_from_url( $url ) {
	$guid = split("/",$url);
	return $guid[count($guid)-1];
}
*/

/**
 * Encase the game description in a CDATA block to keep feed readers from complaining.
 */
function clean_description( $description ) {
	$innerHTML= '';
	foreach ($description->childNodes as $child) {
		$innerHTML .= $child->ownerDocument->saveXML( $child ); 
	}
	return "<![CDATA[".$innerHTML."]]>";
}

/**
 * Clean things up for the XML import. This should probably be replaced by JSON at some point.
 */
function clean_xml_document( $xml ) {
	$p = $xml->getElementsByTagName('p');
	if( $p->length > 0 )
		foreach( $p as $node ) {
			if( $node->getAttribute('class') == "nogames" )
				return false;
		}
	
	$table = $xml->getElementsByTagName('span');
	foreach($table as $node) {
		if( $node->getAttribute("class") == "dl_match" ) {
			$node->setAttribute("style","font-weight: bold; color: red;");
			$node->removeAttribute("class");
			
			$parent = $node->parentNode;
			if( $parent->tagName == "small" ) {
				if( $node->nextSibling->nodeName == "#text" && $node->nextSibling->nodeValue == "; " )
					$parent->removeChild($node->nextSibling);
				$element = $xml->createElement('b');
				$para = $xml->createElement('p');
				$para->appendChild($element);
				$parent->replaceChild($para, $node);
				$element->appendChild($node);
			}
		}
	}
	$table = $xml->getElementsByTagName('tr');
	$table->item(0)->parentNode->removeChild($table->item(0));
	$table->item($table->length - 1)->parentNode->removeChild($table->item($table->length - 1));
	return $table;
}

/**
 * Format the MaGog URL and retrieve the results from MaGog.
 */
function setup_xml_document( &$url ) {
	global $DAYS_SINCE_UPDATE;
	if( $DAYS_SINCE_UPDATE > 0 )
		$date = date('dmY', strtotime('-'.$DAYS_SINCE_UPDATE.' days'));
	else
		$date = date('dmY');
	$url = preg_replace( '/flt=Dsa~([0-9]{8})/i', "flt=Dsa~$date", $url );

	// Create a new DOM Document to hold our webpage structure
	$xml = new DOMDocument();

	// Load the url's contents into the DOM
	if( $xml->loadHTMLFile($url) === false ) {
		http_response_code(404);
		die( "Retrieval of <b>$url<b> failed." );
	}

	return clean_xml_document($xml);
}

/**
 * Set up the headers for the file.
 */
function setup_headers() {
	header("Content-Type: application/rss+xml");
	echo '<?xml version="1.0" encoding="UTF-8" ?>'."\n";
	echo '<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">'."\n";
}

/**
 * Set up the meta data for the RSS feed.
 */
function get_channel( $url ) {
?>
	<channel>
		<title>GOG Game Updates via MaGog</title>
		<description>Recent game updates on GOG.com made available by MaGog.</description>
		<link><?php echo str_replace("&", "&amp;", $url) ?></link>
		<atom:link href="http://<?php echo $_SERVER[HTTP_HOST].$_SERVER[REQUEST_URI]; ?>" rel="self" type="application/rss+xml" />
<?php
}

/**
 * Retrieve and format a game.
 */
function get_item( $node ) {
?>
		<item>
			<title><?php echo clean_game_title($node->childNodes->item(1)->nodeValue) ?></title>
			<description><?php echo clean_description($node->nextSibling->firstChild); ?></description>
			<link><?php echo $node->childNodes->item(1)->firstChild->getAttribute("href"); ?></link>
			<guid><?php echo $node->childNodes->item(1)->firstChild->getAttribute("href"); ?></guid>
		</item>
<?php
}

/**
 * Run the thing. Maybe this will all become a class at some point?
 */
	$url = 'http://www.an-ovel.com/cgi-bin/magog.cgi?ver=422&scp=gdsp&dsp=fD&ord=&flt=Dsa~01012000~3an~owned%2C~&opt=d&myf='.$MAGOG_SNAPSHOT;
	$table = setup_xml_document($url);
	setup_headers();
	get_channel($url);
	
	if( $table !== false ) {
		$i = 0;
		foreach($table as $node) {
			$i++;
			if( $i % 2 == 0 ) continue;
			get_item( $node );
		}
	}
?>
	</channel>
</rss>