A Drupal migration_tools obtainer for finding long plain text.

ObtainLongPlainText.php

<?php

namespace Drupal\migration_tools\Obtainer;

use Drupal\migration_tools\Obtainer\ObtainHtml;
use Drupal\migration_tools\StringTools;

/**
 * Obtainers for Long Plain Text.
 *
 * @package Drupal\migration_tools\Obtainer
 */
class ObtainLongPlainText extends ObtainHtml {

  /**
   * Plucker to turn html into long plain text for nth selector on the page.
   *
   * @param string $selector
   *   The selector to find.
   * @param int $n
   *   (optional) The depth to find.  Default: first item n=1.
   *
   * @return string
   *   The text found.
   */
  protected function pluckPlain($selector, $n = 1, $pluck = TRUE) {
    $text = '';
    $n = ($n > 0) ? $n - 1 : 0;
    if (!empty($selector)) {
      $elements = $this->queryPath->find($selector);
      /** @var \QueryPath\DOMQuery $element */
      foreach ((is_object($elements)) ? $elements : [] as $i => $element) {
        if ($i == $n) {
          if ($pluck) {
            $this->setElementToRemove($element);
          }
          $plain_text = '';
          /** @var \QueryPath\DOMQuery $item */
          foreach ($element->contents() as $item) {
            if ($item->tag() == 'p') {
              if (strlen($plain_text) > 1 && substr($plain_text, -2) != PHP_EOL . PHP_EOL) {
                $plain_text .= PHP_EOL . PHP_EOL;
              }
              $text = $item->innerHTML() . PHP_EOL . PHP_EOL;
            }
            else {
              $text = $item->html();
            }
            $sections = self::splitOnBr($text);
            $plain_text .= strip_tags(implode(PHP_EOL, $sections));
          }
          $this->setCurrentFindMethod("pluckPlain($selector, " . ++$n . ')');
          break;
        }
      }
    }

    return $plain_text;
  }

  /**
   * Finder to turn html into long plain text for nth selector on the page.
   *
   * @param string $selector
   *   The selector to find.
   * @param int $n
   *   (optional) The depth to find.  Default: first item n=1.
   *
   * @return string
   *   The text found.
   */
  protected function findPlain($selector, $n = 1) {
    return $this->pluckPlain($selector, $n, FALSE);
  }

  /**
   * {@inheritdoc}
   */
  public static function cleanString($string) {
    // There are also numeric html special chars, let's change those.
    $string = StringTools::decodeHtmlEntityNumeric($string);
    // Checking again in case another process rendered it non UTF-8.
    $is_utf8 = mb_check_encoding($string, 'UTF-8');

    if (!$is_utf8) {
      $string = StringTools::fixEncoding($string);
    }

    // Remove white space-like things from the ends and decodes html entities.
    $string = StringTools::superTrim($string);

    return $string;
  }

}