FeedsExQueryPathHtml.inc in Feeds extensible parsers 7

Contains FeedsExQueryPathHtml.

File

src/FeedsExQueryPathHtml.inc

<?php

/**
 * @file
 * Contains FeedsExQueryPathHtml.
 */

/**
 * Parses HTML documents with QueryPath.
 */
class FeedsExQueryPathHtml extends FeedsExQueryPathXml {

  /**
   * {@inheritdoc}
   */
  protected $encoderClass = 'FeedsExHtmlEncoder';

  /**
   * {@inheritdoc}
   */
  protected function setUp(FeedsSource $source, FeedsFetcherResult $fetcher_result) {

    // Change some parser settings.
    $this->queryPathOptions['use_parser'] = 'html';
  }

  /**
   * {@inheritdoc}
   */
  protected function getRawValue(QueryPath $node) {
    return $node
      ->html();
  }

  /**
   * {@inheritdoc}
   */
  protected function prepareDocument(FeedsSource $source, FeedsFetcherResult $fetcher_result) {
    $raw = $this
      ->prepareRaw($fetcher_result);
    if ($this->config['use_tidy'] && extension_loaded('tidy')) {
      $raw = tidy_repair_string($raw, $this
        ->getTidyConfig(), 'utf8');
    }
    return FeedsExXmlUtility::createHtmlDocument($raw);
  }

  /**
   * {@inheritdoc}
   */
  protected function getTidyConfig() {
    return array(
      'merge-divs' => FALSE,
      'merge-spans' => FALSE,
      'join-styles' => FALSE,
      'drop-empty-paras' => FALSE,
      'wrap' => 0,
      'tidy-mark' => FALSE,
      'escape-cdata' => TRUE,
      'word-2000' => TRUE,
    );
  }

}

Classes

Name	Description
FeedsExQueryPathHtml	Parses HTML documents with QueryPath.