FeedsYoutubeParser.inc in Feeds: YouTube Parser 7.2

Feeds parser class for YouTube.

File

FeedsYoutubeParser.inc

<?php

/**
 * @file
 * Feeds parser class for YouTube.
 */

/**
 * Class definition for YouTube Parser.
 *
 * Parses RSS or Atom feeds returned from YouTube API like.
 */
class FeedsYoutubeParser extends FeedsParser {

  /**
   * Parse the extra mapping sources provided by this parser.
   *
   * @param $fetcher_result FeedsFetcherResult
   * @param $source FeedsSource
   *
   * @see FeedsParser::parse()
   */
  public function parse(FeedsSource $source, FeedsFetcherResult $fetcher_result) {
    $youtube_feed = $fetcher_result
      ->getRaw();
    $result = new FeedsParserResult();

    /**
     * @see common_syndication_parser_parse()
     */
    if (!defined('LIBXML_VERSION') || version_compare(phpversion(), '5.1.0', '<')) {
      @($sxml = simplexml_load_string($youtube_feed, NULL));
    }
    else {
      @($sxml = simplexml_load_string($youtube_feed, NULL, LIBXML_NOERROR | LIBXML_NOWARNING | LIBXML_NOCDATA));
    }

    // Got a malformed XML.
    if ($sxml === FALSE || is_null($sxml)) {
      throw new Exception(t('FeedsYoutubeParser: Malformed XML source.'));
    }

    // Run parsing if the feed is Atom or RSS
    if ($this
      ->isAtomFeed($sxml)) {
      $result = $this
        ->parseAtom($sxml, $source, $fetcher_result);
    }
    elseif ($this
      ->isRssFeed($sxml)) {
      $result = $this
        ->parseRss20($sxml, $source, $fetcher_result);
    }
    else {
      throw new Exception(t('FeedsYoutubeParser: Unknown type of feed.'));
    }
    return $result;
  }

  /**
   * Check if given feed object is an Atom feed.
   *
   * @param SimpleXMLElement $sxml
   *
   * @return boolean
   *   TRUE if given SimpleXML object is Atom feed or FALSE
   */
  protected function isAtomFeed(SimpleXMLElement $sxml) {
    return $sxml
      ->getName() == 'feed';
  }

  /**
   * Check if given feed object is a RSS feed.
   *
   * @param SimpleXMLElement $sxml
   *
   * @return boolean
   *   TRUE if given SimpleXML object is RSS feed or FALSE
   */
  protected function isRssFeed(SimpleXMLElement $sxml) {
    return $sxml
      ->getName() == 'rss';
  }

  /**
   * Add the extra mapping sources provided by this parser.
   */
  public function getMappingSources() {
    return parent::getMappingSources() + array(
      'feed_title' => array(
        'name' => t('Feed title'),
        'description' => t('The title of the pulled feed.'),
      ),
      'guid' => array(
        'name' => t('GUID'),
      ),
      'video_id' => array(
        'name' => t('Video ID'),
        'description' => t('YouTube video unique ID.'),
      ),
      'title' => array(
        'name' => t('Video title'),
        'description' => t('Video title.'),
      ),
      'author' => array(
        'name' => t('Author'),
        'description' => t('Author or uploader of the video.'),
      ),
      'updated_datetime' => array(
        'name' => t('Updated on (Datetime)'),
      ),
      'updated_timestamp' => array(
        'name' => t('Updated on (Timestamp)'),
      ),
      'published_datetime' => array(
        'name' => t('Published on (Datetime)'),
      ),
      'published_timestamp' => array(
        'name' => t('Published on (Timestamp)'),
      ),
      'uploaded_datetime' => array(
        'name' => t('Uploaded on (Datetime)'),
      ),
      'uploaded_timestamp' => array(
        'name' => t('Uploaded on (Timestamp)'),
      ),
      'description' => array(
        'name' => t('Description'),
      ),
      'thumbnail' => array(
        'name' => t('Thumbnail'),
        'description' => t('URL of the thumbnail of the video.'),
      ),
      'category' => array(
        'name' => t('Category'),
      ),
      'tags' => array(
        'name' => t('Tags'),
        'description' => t('This can be imported directly with Taxonomy "tags" vocabularies.'),
      ),
      'watch_page' => array(
        'name' => t('Watch page'),
        'description' => t('The URL of the video.'),
      ),
      'url' => array(
        'name' => t('Video URL'),
        'description' => t('The URL of the video.'),
      ),
      'duration' => array(
        'name' => t('Duration (Formatted)'),
        'description' => t('Duration of the video in HH:MM:SS format.'),
      ),
      'duration_raw' => array(
        'name' => t('Duration (Seconds)'),
        'description' => t('Duration of the video in number of seconds.'),
      ),
      'fav_count' => array(
        'name' => t('Favorite count'),
      ),
      'view_count' => array(
        'name' => t('View count'),
      ),
      'rating' => array(
        'name' => t('Rating'),
      ),
      'raters_number' => array(
        'name' => t('Number of raters'),
      ),
      'likes' => array(
        'name' => t('Number of likes'),
      ),
      'dislikes' => array(
        'name' => t('Number of dislikes'),
      ),
    );
  }

  /**
   *  Display seconds as HH:MM:SS, with leading 0's.
   *
   *  @param $seconds
   *    The number of seconds to display.
   */
  public function secsToTime($seconds) {

    // Number of seconds in an hour.
    $unith = 3600;

    // Number of seconds in a minute.
    $unitm = 60;

    // '/' given value by num sec in hour... output = HOURS
    $hh = intval($seconds / $unith);

    // Multiply number of hours by seconds, then subtract from given value.
    // Output = REMAINING seconds.
    $ss_remaining = $seconds - $hh * 3600;

    // Take remaining seconds and divide by seconds in a min... output = MINS.
    $mm = intval($ss_remaining / $unitm);

    // Multiply number of mins by seconds, then subtract from remaining seconds.
    // Output = REMAINING seconds.
    $ss = $ss_remaining - $mm * 60;
    $output = '';

    // If we have any hours, then prepend that to our output.
    if ($hh) {
      $output .= "{$hh}:";
    }

    // Create a safe-for-output MM:SS.
    $output .= sprintf($hh ? "%02d:%02d" : "%d:%02d", $mm, $ss);
    return $output;
  }

  /**
   * Parse Atom feed.
   *
   * @param SimpleXMLElement $sxml
   * @param FeedsFetcherResult $fetcher_result
   * @param FeedsSource $source
   */
  private function parseAtom(SimpleXMLElement $sxml, FeedsSource $source, FeedsFetcherResult $fetcher_result) {
    $fetcher_result->title = $feed_title = (string) $sxml->title;
    $result = new FeedsParserResult();

    // Iterate over entries in feed
    // TODO: This is not DRY - extract things which is same in Atom and RSS20 to common method
    foreach ($sxml->entry as $entry) {

      // Get nodes in media: namespace for media information.
      $media = $entry
        ->children('http://search.yahoo.com/mrss/');

      // Get video player URL.
      $attrs = $media->group->player
        ->attributes();
      $watch = str_replace('&feature=youtube_gdata_player', '', $attrs['url']);

      // Get video thumbnail.
      $attrs = $media->group->thumbnail[0]
        ->attributes();
      $thumbnail = (string) $attrs['url'];

      // Get <yt:duration> node for video length.
      $yt = $media
        ->children('http://gdata.youtube.com/schemas/2007');
      $attrs = $yt->duration
        ->attributes();
      $length = $attrs['seconds'];

      // Get video ID.
      $id = (string) $yt->videoid;

      // Get video uploaded date.
      $uploaded = (string) $yt->uploaded;

      // Get <yt:statistics> node for viewer statistics.
      $yt = $entry
        ->children('http://gdata.youtube.com/schemas/2007');
      $attrs = $yt->statistics
        ->attributes();
      $viewCount = $attrs['viewCount'];
      $favCount = $attrs['favoriteCount'];

      // Get <yt:rating> node for likes and dislikes statistics.
      $likes = 0;
      $dislikes = 0;
      if (isset($yt->rating) && ($attrs = $yt->rating
        ->attributes())) {
        $likes = (int) $attrs['numLikes'];
        $dislikes = (int) $attrs['numDislikes'];
      }

      // Get <gd:rating> node for video ratings.
      $gd = $entry
        ->children('http://schemas.google.com/g/2005');
      $rating = 0;
      $raters_number = 0;
      if ($gd->rating) {
        $attrs = $gd->rating
          ->attributes();
        $rating = (double) $attrs['average'];
        $raters_number = (int) $attrs['numRaters'];
      }
      $updated = (string) $entry->updated;
      $published = (string) $entry->published;
      $item = array(
        'feed_title' => $feed_title,
        'guid' => (string) $entry->id,
        'video_id' => $id,
        'url' => 'http://www.youtube.com/watch?v=' . $id,
        'watch_page' => $watch,
        'title' => (string) $media->group->title,
        'author' => (string) $entry->author->name,
        'description' => (string) $media->group->description,
        'thumbnail' => $thumbnail,
        'category' => (string) $media->group->category,
        'tags' => explode(',', $media->group->keywords),
        'embedded_player' => '',
        'duration' => $this
          ->secsToTime($length),
        'duration_raw' => (int) $length,
        'view_count' => (string) $viewCount,
        'fav_count' => (string) $favCount,
        'rating' => (string) $rating,
        'raters_number' => $raters_number,
        'updated_datetime' => date('Y-m-d H:i:s', strtotime($updated)),
        'updated_timestamp' => strtotime($updated),
        'published_datetime' => date('Y-m-d H:i:s', strtotime($published)),
        'published_timestamp' => strtotime($published),
        'uploaded_datetime' => date('Y-m-d H:i:s', strtotime($uploaded)),
        'uploaded_timestamp' => strtotime($uploaded),
        'likes' => $likes,
        'dislikes' => $dislikes,
      );

      // Populate the FeedsFetcherResult object with the parsed results.
      $result->items[] = $item;
    }
    return $result;
  }

  /**
   * Parse RSS 2.0 feed.
   *
   * @param SimpleXMLElement $sxml
   * @param FeedsFetcherResult $fetcher_result
   * @param FeedsSource $source
   */
  private function parseRss20(SimpleXMLElement $sxml, FeedsSource $source, FeedsFetcherResult $fetcher_result) {

    // XML was parsed successfully, so we can begin to process items.
    $result = new FeedsParserResult();
    $fetcher_result->title = (string) $sxml->channel->title;
    $fetcher_result->description = (string) $sxml->channel->description;
    $fetcher_result->link = (string) $sxml->channel->link;
    $feed_title = (string) $sxml->channel->title;

    // Iterate over entries in feed.
    // TODO: This is not DRY - extract things which is same in Atom and RSS20 to common method
    foreach ($sxml
      ->xpath('//item') as $entry) {

      // Get Atom nodes.
      $atom = $entry
        ->children('http://www.w3.org/2005/Atom');
      $updated = $atom->updated;

      // Get video ID.
      $id = end(explode('/', $entry->guid));

      // Get nodes in media: namespace for media information.
      $media = $entry
        ->children('http://search.yahoo.com/mrss/');

      // Get video player URL
      $attrs = $media->group->player
        ->attributes();
      $player = (string) $attrs['url'];

      // Get video thumbnail.
      $attrs = $media->group->thumbnail[0]
        ->attributes();
      $thumbnail = (string) $attrs['url'];

      // Get <yt:duration> node for video length.
      $yt = $media
        ->children('http://gdata.youtube.com/schemas/2007');
      $attrs = $yt->duration
        ->attributes();
      $length = (int) $attrs['seconds'];

      // Get <yt:stats> node for viewer statistics.
      $yt = $entry
        ->children('http://gdata.youtube.com/schemas/2007');
      $attrs = $yt->statistics
        ->attributes();
      $viewCount = (int) $attrs['viewCount'];
      $favCount = (int) $attrs['favoriteCount'];

      // Get <gd:rating> node for video ratings.
      $gd = $entry
        ->children('http://schemas.google.com/g/2005');
      $rating = 0;
      if ($gd->rating) {
        $attrs = $gd->rating
          ->attributes();
        $rating = (double) $attrs['average'];
        $raters_number = (int) $attrs['numRaters'];
      }
      $item = array(
        'feed_title' => $feed_title,
        'guid' => (string) $entry->guid,
        'video_id' => $id,
        'url' => 'http://www.youtube.com/watch?v=' . $id,
        'watch_page' => 'http://www.youtube.com/watch?v=' . $id,
        'title' => html_entity_decode((string) $media->group->title),
        'author' => (string) $entry->author,
        'description' => html_entity_decode((string) $media->group->description),
        'thumbnail' => $thumbnail,
        'category' => (string) $media->group->category,
        'tags' => explode(',', (string) $media->group->keywords),
        'embedded_player' => $player,
        'duration' => $this
          ->secsToTime($length),
        'duration_raw' => $length,
        'view_count' => $viewCount,
        'fav_count' => $favCount,
        'rating' => $rating,
        'raters_number' => $raters_number,
        'updated_datetime' => date('Y-m-d H:i:s', strtotime($updated)),
        'updated_timestamp' => strtotime($updated),
        'published_datetime' => date('Y-m-d H:i:s', strtotime($entry->pubDate)),
        'published_timestamp' => strtotime($entry->pubDate),
      );

      // Populate the FeedsFetcherResult object with the parsed results.
      $result->items[] = $item;
    }
    return $result;
  }

}

Classes

Name	Description
FeedsYoutubeParser	Class definition for YouTube Parser.