TruncateHtmlExtension.php

<?php

/*
 * This file is part of the Twig Extension TruncateHtmlExtension.
 *
 * Author Leon Radley <[email protected]>
 * Rewrite by Instabledesign
 *
 *
 * Truncate Html string without stripping tags
 * register in Resources/config/services.yml with:
 *
 * services:
 *   truncatehtml.twig.extension:
 *     class: Radley\TwigExtensionBundle\Extension\TruncateHtmlExtension
 *     tags:
 *       - { name: twig.extension }
 *
 * Usage:
 * {{ htmlstring|truncatehtml(500)|raw }}
 */

namespace Radley\TwigExtensionBundle\Extension;

class TruncateHtmlExtension extends \Twig_Extension
{
  public function getName() {
    return 'truncatehtml';
  }

  public function getFilters()
  {
    return array('truncatehtml' => new \Twig_Filter_Method($this, 'truncatehtml'));
  }

  public function truncatehtml($html, $limit, $elipsis = '')
  {
    return TruncateHtmlString::cut($html, $limit, $elipsis);
  }
}

TruncateHtmlString.php

<?php

/*
 * This file is part of the Twig Extension TruncateHtmlExtension.
 *
 * Author Leon Radley <[email protected]>
 * Rewrite by Instabledesign
 *
 */
namespace Radley\TwigExtensionBundle\Extension;

class TruncateHtmlString {

  /**
   * Return semantic truncate HTML
   *
   * @param string $html HTML code
   * @param int $length The length of HTML truncate
   * @param string $elipsis The elipsis string put at the end of last word
   * @param string $encoding The encoding HTML
   *
   * @return string Return semantic truncate HTML string
   */
  public static function cut($html, $length = 500, $elipsis = '', $encoding = 'UTF-8') {
    $fromXML = new \DomDocument();
    $fromXML->loadXML('<div>'.$html.'</div>');

    $toXML = new \DomDocument();
    self::searchEnd(
      $fromXML->documentElement,
      $toXML,
      $toXML,
      $length,
      $elipsis,
      $encoding
    );
    return $toXML->saveHTML();
  }

  /**
   * Delete node child recursively
   *
   * @param mixed
   */
  private static function deleteNodeChildren($node) {
    while (isset($node->firstChild)) {
      self::deleteNodeChildren($node->firstChild);
      $node->removeChild($node->firstChild);
    }
  }

  /**
   * Search the HTML truncate point recursively
   *
   * @param DomElement $domElement
   * @param mixed $domElementParent
   * @param DomDocument $toXML
   * @param integer $length
   * @param string $elipsis
   * @param integer &$charCount
   * @param string $encoding
   *
   * @return boolean
   */
  private static function searchEnd(\DomElement $domElement, $domElementParent, \DomDocument $toXML, $length, $elipsis, &$charCount = 0, $encoding = 'UTF-8') {
    foreach ($domElement->childNodes as $child) {
      if($child->nodeType != 3) {
        $newElement = $toXML->importNode($child, true);
        if (0 === count($child->childNodes)) {
          $domElementParent->appendChild($newElement);
          continue;
        }

        self::deleteNodeChildren($newElement);

        $domElementParent->appendChild($newElement);

        if (false === self::searchEnd($child, $newElement, $toXML, $length, $elipsis, $charCount)) {
          continue;
        }

        return true;
      }

      if (mb_strlen($child->nodeValue, $encoding) + $charCount >= $length) {
        $newElement = $toXML->importNode($child);
        $newElement->nodeValue = substr($newElement->nodeValue, 0, $length - $charCount).$elipsis;

        $domElementParent->appendChild($newElement);

        return true;
      }

      $newElement = $toXML->importNode($child);

      $domElementParent->appendChild($newElement);

      $charCount += mb_strlen($newElement->nodeValue, $encoding);
    }

    return false;
  }
}