/**
* @see wcf\system\ICronjob::execute()
*/
public function execute(Cronjob $cronjob) {
$filename = FileUtil::downloadFileFromHttp('http://www.woltlab.com/spiderlist/spiderlist.xml', 'spiders');
$xml = new XML();
$xml->load($filename);
$xpath = $xml->xpath();
// fetch spiders
$spiders = $xpath->query('/spiderlist/spider');
if (!empty($spiders)) {
// delete old entries
$sql = "DELETE FROM wcf".WCF_N."_spider";
$statement = WCF::getDB()->prepareStatement($sql);
$statement->execute();
$statementParameters = array();
foreach ($spiders as $spider) {
$identifier = StringUtil::toLowerCase($spider->getAttribute('ident'));
$name = $xpath->query('name', $spider)->item(0);
$info = $xpath->query('info', $spider)->item(0);
$statementParameters[$identifier] = array(
'spiderIdentifier' => $identifier,
'spiderName' => $name->nodeValue,
'spiderURL' => $info ? $info->nodeValue : ''
);
}
if (!empty($statementParameters)) {
$sql = "INSERT INTO wcf".WCF_N."_spider
(spiderIdentifier, spiderName, spiderURL)
VALUES (?, ?, ?)";
$statement = WCF::getDB()->prepareStatement($sql);
foreach ($statementParameters as $parameters) {
$statement->execute(array(
$parameters['spiderIdentifier'],
$parameters['spiderName'],
$parameters['spiderURL']
));
}
}
// clear spider cache
SpiderCacheBuilder::getInstance()->reset();
}
// delete tmp file
@unlink($filename);
}
/**
* @see \wcf\system\ICronjob::execute()
*/
public function execute(Cronjob $cronjob)
{
$filename = FileUtil::downloadFileFromHttp('http://assets.woltlab.com/spiderlist/typhoon/list.xml', 'spiders');
$xml = new XML();
$xml->load($filename);
$xpath = $xml->xpath();
// fetch spiders
$spiders = $xpath->query('/ns:data/ns:spider');
if (!empty($spiders)) {
$existingSpiders = SpiderCacheBuilder::getInstance()->getData();
$statementParameters = array();
foreach ($spiders as $spider) {
$identifier = mb_strtolower($spider->getAttribute('ident'));
$name = $xpath->query('ns:name', $spider)->item(0);
$info = $xpath->query('ns:url', $spider)->item(0);
$statementParameters[$identifier] = array('spiderIdentifier' => $identifier, 'spiderName' => $name->nodeValue, 'spiderURL' => $info ? $info->nodeValue : '');
}
if (!empty($statementParameters)) {
$sql = "INSERT INTO\t\t\twcf" . WCF_N . "_spider\n\t\t\t\t\t\t\t\t\t(spiderIdentifier, spiderName, spiderURL)\n\t\t\t\t\tVALUES\t\t\t\t(?, ?, ?)\n\t\t\t\t\tON DUPLICATE KEY UPDATE\t\tspiderName = VALUES(spiderName),\n\t\t\t\t\t\t\t\t\tspiderURL = VALUES(spiderURL)";
$statement = WCF::getDB()->prepareStatement($sql);
WCF::getDB()->beginTransaction();
foreach ($statementParameters as $parameters) {
$statement->execute(array($parameters['spiderIdentifier'], $parameters['spiderName'], $parameters['spiderURL']));
}
WCF::getDB()->commitTransaction();
}
// delete obsolete entries
$sql = "DELETE FROM wcf" . WCF_N . "_spider WHERE spiderIdentifier = ?";
$statement = WCF::getDB()->prepareStatement($sql);
foreach ($existingSpiders as $spider) {
if (!isset($statementParameters[$spider->spiderIdentifier])) {
$statement->execute(array($spider->spiderIdentifier));
}
}
// clear spider cache
SpiderCacheBuilder::getInstance()->reset();
}
// delete tmp file
@unlink($filename);
}
/**
* Returns the spider object
*
* @return \wcf\data\spider\Spider
*/
public function getSpider()
{
if (!$this->spiderID) {
return null;
}
if ($this->spider === null) {
$spiderList = SpiderCacheBuilder::getInstance()->getData();
$this->spider = $spiderList[$this->spiderID];
}
return $this->spider;
}
/**
* Returns the spider id for given user agent.
*
* @param string $userAgent
* @return mixed
*/
protected function getSpiderID($userAgent)
{
$spiderList = SpiderCacheBuilder::getInstance()->getData();
$userAgent = strtolower($userAgent);
foreach ($spiderList as $spider) {
if (strpos($userAgent, $spider->spiderIdentifier) !== false) {
return $spider->spiderID;
}
}
return null;
}
