public function crawlMetaData($websiteUrl, $keyInput = '', $pageContent = '', $returVal = false)
{
if (empty($pageContent)) {
if (!preg_match('/\\w+/', $websiteUrl)) {
return;
}
if (!stristr($websiteUrl, 'http://')) {
$websiteUrl = "http://" . $websiteUrl;
}
$spider = new Spider();
$ret = $spider->getContent($websiteUrl);
} else {
$ret['page'] = $pageContent;
$metaInfo = array();
}
if (!empty($ret['page'])) {
if (empty($keyInput)) {
# meta title
preg_match('/<TITLE>(.*?)<\\/TITLE>/si', $ret['page'], $matches);
if (!empty($matches[1])) {
if ($returVal) {
$metaInfo['page_title'] = $matches[1];
} else {
WebsiteController::addInputValue($matches[1], 'webtitle');
}
}
# meta description
preg_match('/<META.*?name="description".*?content="(.*?)"/si', $ret['page'], $matches);
if (empty($matches[1])) {
preg_match("/<META.*?name='description'.*?content='(.*?)'/si", $ret['page'], $matches);
}
if (empty($matches[1])) {
preg_match('/<META content="(.*?)" name="description"/si', $ret['page'], $matches);
}
if (!empty($matches[1])) {
if ($returVal) {
$metaInfo['page_description'] = $matches[1];
} else {
WebsiteController::addInputValue($matches[1], 'webdescription');
}
}
}
# meta keywords
preg_match('/<META.*?name="keywords".*?content="(.*?)"/si', $ret['page'], $matches);
if (empty($matches[1])) {
preg_match("/<META.*?name='keywords'.*?content='(.*?)'/si", $ret['page'], $matches);
}
if (empty($matches[1])) {
preg_match('/<META content="(.*?)" name="keywords"/si', $ret['page'], $matches);
}
if (!empty($matches[1])) {
if ($returVal) {
$metaInfo['page_keywords'] = $matches[1];
} else {
WebsiteController::addInputValue($matches[1], 'webkeywords');
}
}
}
return $metaInfo;
}
