ext/flexmailer/src/ClickTracker/HtmlClickTracker.php

   1 <?php
   2 /*
   3  +--------------------------------------------------------------------+
   4  | Copyright CiviCRM LLC. All rights reserved.                        |
   5  |                                                                    |
   6  | This work is published under the GNU AGPLv3 license with some      |
   7  | permitted exceptions and without any warranty. For full license    |
   8  | and copyright information, see https://civicrm.org/licensing       |
   9  +--------------------------------------------------------------------+
  10  */
  11 namespace Civi\FlexMailer\ClickTracker;
  12
  13 class HtmlClickTracker implements ClickTrackerInterface {
  14
  15   public function filterContent($msg, $mailing_id, $queue_id) {
  16     return self::replaceHrefUrls($msg,
  17       function ($url) use ($mailing_id, $queue_id) {
  18         if (strpos($url, '{') !== FALSE) {
  19           return $url;
  20         }
  21         $data = \CRM_Mailing_BAO_TrackableURL::getTrackerURL(
  22           $url, $mailing_id, $queue_id);
  23         $data = htmlentities($data, ENT_NOQUOTES);
  24         return $data;
  25       }
  26     );
  27   }
  28
  29   /**
  30    * Find any HREF-style URLs and replace them.
  31    *
  32    * @param string $html
  33    * @param callable $replace
  34    *   Function(string $oldHtmlUrl) => string $newHtmlUrl.
  35    * @return mixed
  36    *   String, HTML.
  37    */
  38   public static function replaceHrefUrls($html, $replace) {
  39     $useNoFollow = TRUE;
  40     $callback = function ($matches) use ($replace, $useNoFollow) {
  41       $replacement = $replace($matches[2]);
  42
  43       // See: https://github.com/civicrm/civicrm-core/pull/12561
  44       // If we track click-throughs on a link, then don't encourage search-engines to traverse them.
  45       // At a policy level, I'm not sure I completely agree, but this keeps things consistent.
  46       // You can tell if we're tracking a link because $replace() yields a diff URL.
  47       $noFollow = '';
  48       if ($useNoFollow && $replacement !== $matches[2]) {
  49         $noFollow = " rel='nofollow'";
  50       }
  51
  52       return $matches[1] . $replacement . $matches[3] . $noFollow;
  53     };
  54
  55     // Find anything like href="..." or href='...' inside a tag.
  56     $tmp = preg_replace_callback(
  57       ';(\<[^>]*href *= *")([^">]+)(");', $callback, $html);
  58     return preg_replace_callback(
  59       ';(\<[^>]*href *= *\')([^\'>]+)(\');', $callback, $tmp);
  60   }
  61
  62   //  /**
  63   //   * Find URL expressions; replace them with tracked URLs.
  64   //   *
  65   //   * @param string $msg
  66   //   * @param int $mailing_id
  67   //   * @param int|string $queue_id
  68   //   * @param bool $html
  69   //   * @return string
  70   //   *   Updated $msg
  71   //   */
  72   //  public static function scanAndReplace_old($msg, $mailing_id, $queue_id, $html = FALSE) {
  73   //
  74   //    $protos = '(https?|ftp)';
  75   //    $letters = '\w';
  76   //    $gunk = '/#~:.?+=&%@!\-';
  77   //    $punc = '.:?\-';
  78   //    $any = "{$letters}{$gunk}{$punc}";
  79   //    if ($html) {
  80   //      $pattern = "{\\b(href=([\"'])?($protos:[$any]+?(?=[$punc]*[^$any]|$))([\"'])?)}im";
  81   //    }
  82   //    else {
  83   //      $pattern = "{\\b($protos:[$any]+?(?=[$punc]*[^$any]|$))}eim";
  84   //    }
  85   //
  86   //    $trackURL = \CRM_Mailing_BAO_TrackableURL::getTrackerURL('\\1', $mailing_id, $queue_id);
  87   //    $replacement = $html ? ("href=\"{$trackURL}\"") : ("\"{$trackURL}\"");
  88   //
  89   //    $msg = preg_replace($pattern, $replacement, $msg);
  90   //    if ($html) {
  91   //      $msg = htmlentities($msg, ENT_NOQUOTES);
  92   //    }
  93   //    return $msg;
  94   //  }
  95
  96 }