/**
* Converts links to trackable links and tokens
*
* @param EmailSendEvent $event
*
* @return array
*/
protected function convertTrackableLinks(EmailSendEvent $event)
{
// Get a list of tokens for the tokenized link conversion
$currentTokens = $event->getTokens();
/** @var \Mautic\PageBundle\Model\RedirectModel $redirectModel */
$redirectModel = $this->factory->getModel('page.redirect');
// Parse the content for links
$body = $event->getContent();
// Find links using DOM to only find <a> tags
$libxmlPreviousState = libxml_use_internal_errors(true);
libxml_use_internal_errors(true);
$dom = new \DOMDocument();
$dom->loadHTML('<?xml encoding="UTF-8">' . $body);
libxml_clear_errors();
libxml_use_internal_errors($libxmlPreviousState);
$links = $dom->getElementsByTagName('a');
$foundLinks = $tokenizedLinks = array();
foreach ($links as $link) {
$url = $link->getAttribute('href');
// The editor will have converted & to & but DOMDocument will have converted them back so this must be accounted for
$url = str_replace('&', '&', $url);
$this->validateLink($url, $currentTokens, $foundLinks);
}
// Process plain text as well
$plainText = $event->getPlainText();
if (!empty($plainText)) {
// Plaintext links
preg_match_all('@(?<![.*">])\\b(?:(?:https?|ftp|file)://|[a-z]\\.)[-A-Z0-9+&#/%=~_|$?!:,.]*[A-Z0-9+&#/%=~_|$]@i', $plainText, $matches);
if (!empty($matches[0])) {
foreach ($matches[0] as $url) {
// Remove anything left on at the end; just in case
$url = preg_replace('/^\\PL+|\\PL\\z/', '', trim($url));
$this->validateLink($url, $currentTokens, $foundLinks);
}
}
}
$trackedLinks = array();
if (!empty($foundLinks)) {
$links = $redirectModel->getRedirectListByUrls($foundLinks, $this->emailEntity);
foreach ($links as $url => $link) {
if (!$link->getId() && !isset($persistEntities[$url])) {
$persistEntities[$url] = $link;
}
$trackedLinks[$url] = $link;
}
}
if (!empty($persistEntities)) {
// Save redirect entities
$redirectModel->getRepository()->saveEntities($persistEntities);
}
unset($foundLinks, $links, $persistEntities);
return $trackedLinks;
}
/**
* Parses content for URLs and tokens.
*
* @param EmailSendEvent $event
* @param $emailId
*
* @return mixed
*/
protected function parseContentForUrls(EmailSendEvent $event, $emailId)
{
static $convertedContent = [];
// Prevent parsing the exact same content over and over
if (!isset($convertedContent[$event->getContentHash()])) {
$html = $event->getContent();
$text = $event->getPlainText();
$contentTokens = $event->getTokens();
list($content, $trackables) = $this->pageTrackableModel->parseContentForTrackables([$html, $text], $contentTokens, $emailId ? 'email' : null, $emailId);
list($html, $text) = $content;
unset($content);
if ($html) {
$event->setContent($html);
}
if ($text) {
$event->setPlainText($text);
}
$convertedContent[$event->getContentHash()] = $trackables;
// Don't need to preserve Trackable or Redirect entities in memory
$this->em->clear('Mautic\\PageBundle\\Entity\\Redirect');
$this->em->clear('Mautic\\PageBundle\\Entity\\Trackable');
unset($html, $text, $trackables);
}
return $convertedContent[$event->getContentHash()];
}
/**
* Converts links to trackable links and tokens
*
* @param EmailSendEvent $event
*
* @return array
*/
protected function convertTrackableLinks(EmailSendEvent $event)
{
$taggedDoNotTrack = array();
// Get a list of tokens for the tokenized link conversion
$currentTokens = $event->getTokens();
// Parse the content for links
$body = $event->getContent();
// Find links using DOM to only find <a> tags
$libxmlPreviousState = libxml_use_internal_errors(true);
libxml_use_internal_errors(true);
$dom = new \DOMDocument();
$dom->loadHTML('<?xml encoding="UTF-8">' . $body);
libxml_clear_errors();
libxml_use_internal_errors($libxmlPreviousState);
$links = $dom->getElementsByTagName('a');
$foundLinks = $tokenizedLinks = array();
/** @var \DOMElement $link */
foreach ($links as $link) {
$url = $link->getAttribute('href');
// The editor will have converted & to & but DOMDocument will have converted them back so this must be accounted for
$url = str_replace('&', '&', $url);
// Check for a do not track
if ($link->hasAttribute('mautic:disable-tracking')) {
$taggedDoNotTrack[$url] = true;
continue;
}
$this->validateLink($url, $currentTokens, $foundLinks);
}
// Process plain text as well
$plainText = $event->getPlainText();
if (!empty($plainText)) {
// Plaintext links
preg_match_all('/((https?|ftps?):\\/\\/)([a-zA-Z0-9-\\.{}]*[a-zA-Z0-9=}]*)(\\??)([^\\s\\]]+)?/i', $plainText, $matches);
if (!empty($matches[0])) {
foreach ($matches[0] as $url) {
$url = trim($url);
// Validate the link if it has not already been marked as do not track by attribute in the HTML version
if (!isset($taggedDoNotTrack[$url])) {
$this->validateLink($url, $currentTokens, $foundLinks);
}
}
}
}
/** @var \Mautic\PageBundle\Model\RedirectModel $redirectModel */
$redirectModel = $this->factory->getModel('page.redirect');
if (!empty($foundLinks)) {
$links = $redirectModel->getRedirectListByUrls($foundLinks, $this->emailEntity);
foreach ($links as $url => $link) {
if (!$link->getId() && !isset($persistEntities[$url])) {
$persistEntities[$url] = $link;
}
$this->emailTrackedLinkSettings['trackedLinks'][$url] = $link;
}
}
if (!empty($persistEntities)) {
// Save redirect entities
$redirectModel->getRepository()->saveEntities($persistEntities);
}
unset($foundLinks, $links, $persistEntities);
}
