moeFlix/inc/seriesScraper.php

<?php

ignore_user_abort(true);

class SeriesScraper {
  private $apiURL = "http://api.themoviedb.org/3/search/tv?api_key=a39779a38e0619f8ae58b09f64522597&query=";
  private $bannerURL = "https://image.tmdb.org/t/p/original";

  private $outputText = "";

  public function __construct() {

  }


  public function getOutputText() {
    return $this->outputText;
  }

  public static function getEpisodeNumberByFilename($filename) {
    preg_match("/([a-zA-Z0-9äöüÄÖÜß\-\.\,\w]+)-[S|s]([0-9]+)[E|e]([0-9]+)-([a-zA-Z0-9äöüÄÖÜß\-\.\,\w\:]+)\.([a-zA-Z0-9\(\).*\w\,]+)/", $filename, $output);

    return ltrim($output[3], 0);
  }

  public static function remove($seriesID) {
    $series = Model::getSeriesBySeriesID($seriesID);
    $seasons = Model::getSeasonsBySeriesID($seriesID);
    foreach($seasons as $season) {
      $GLOBALS['db']->deleteRows("series-episodes", "season-id", $season['id']);
    }
    $GLOBALS['db']->deleteRows("series-seasons", "series-id", $seriesID);
    $GLOBALS['db']->deleteRows("series", "id", $seriesID);

    // TODO: Remove files
  }

  public function update($seriesID) {
    $series = Model::getSeriesBySeriesID($seriesID);

    self::remove($seriesID);

    $this->downloadSeriesByID($series[0]['moviedb-id'], $series[0]['path'], $series[0]['source']);
  }

  public function scrapeFolder($sourceID) {
    if(!file_exists("img/posters")) {
        mkdir("img/posters");
    }

    $source = $GLOBALS['db']->getAllAssoc("sources", "id", $sourceID);

    $dirList = scandir($source[0]['path']);
    $dirList = array_diff($dirList, array('.', '..', 'formatting.txt', '.Trash-1000'));

    foreach($dirList as $dir) {
      $this->outputText .= "<b>" . $dir . "</b><br>" . PHP_EOL;

      if($GLOBALS['db']->countRows("series", "path", $dir) > 0) {
        $this->outputText .= "Exists, skipping..<br><br>" . PHP_EOL;
        continue;
      }

      $series = json_decode(curl_download($this->apiURL . urlencode($dir) . "&language=de&include_image_language=de"), true);

      if($series['total_results'] == 1) {
        $this->outputText .= "Found 1 series, downloading...<br>" . PHP_EOL;

        $this->downloadSeriesByID($series['results'][0]['id'], $dir, $sourceID);
      } else if($series['total_results'] < 1) {
        $this->outputText .= "<span style=\"color: red;\">Not found!!</span><br>" . PHP_EOL;
      } else { // multiple search results
        usort($series['results'], array("MovieScraper", "sortByPopularity")); // sort results by popularity, so that you don't have to scroll like 500000x
        foreach($series['results'] as $result) {
          $this->outputText .= "<pre>" . print_r($result, true) . "</pre>" . PHP_EOL;
          $this->outputText .= "<a href=\"?view=scrape&action=scrapeSingleTV&moviedbID=" . $result['id'] . "&path=" . urlencode($dir) . "&sourceID=" . $sourceID . "\" target=\"_blank\">Load</a><br>" . PHP_EOL; // TODO: this
        }
      }

      $this->outputText .= "<br>" . PHP_EOL;

      /*

      Array
        (
            [page] => 1
            [results] => Array
                (
                    [0] => Array
                        (
                            [poster_path] => /hDlezfMSw8oXIFNZ7B9fFLrV8kd.jpg
                            [popularity] => 1.006821
                            [id] => 15826
                            [backdrop_path] => /meQN6iuOulLwOV8LNO6S9z3bJBY.jpg
                            [vote_average] => 7
                            [overview] => 1000 Ways to Die is an anthology television series that [...]
                            [first_air_date] => 2009-02-04
                            [origin_country] => Array
                                (
                                    [0] => US
                                )

                            [genre_ids] => Array
                                (
                                    [0] => 99
                                )

                            [original_language] => en
                            [vote_count] => 1
                            [name] => 1000 Ways to Die
                            [original_name] => 1000 Ways to Die
                        )

                )

            [total_results] => 1
            [total_pages] => 1
        )

        */
    }

  }

  public function downloadSeriesByID($seriesID, $path, $sourceID) {

    $source = $GLOBALS['db']->getAllAssoc("sources", "id", $sourceID);

    $series = json_decode(curl_download("http://api.themoviedb.org/3/tv/" . $seriesID . "?api_key=a39779a38e0619f8ae58b09f64522597&language=de&include_image_language=de"), true);

    //$this->outputText .= "<pre>" . print_r($series, true) . "</pre>";

    // Download poster, backdrop
    if(!file_exists("img/posters" . $series['poster_path'])) {
      file_put_contents("img/posters" . $series['poster_path'], fopen($this->bannerURL . $series['poster_path'], 'r'));
    }

    if(!file_exists("img/posters" . $series['backdrop_path'])) {
      file_put_contents("img/posters" . $series['backdrop_path'], fopen($this->bannerURL . $series['backdrop_path'], 'r'));
    }

    $cols = array(
      "moviedb-id",
      "name",
      "path",
      "poster",
      "backdrop",
      "overview",
      "source"
    );

    $vals = array(
      $series['id'],
      $series['name'],
      $path,
      $series['poster_path'],
      $series['backdrop_path'],
      $series['overview'],
      $sourceID
    );

    $dbSeriesID = $GLOBALS['db']->insertRow("series", $cols, $vals);

    $localSeasons = scandir($source[0]['path'] . "/" . $path);
    $localSeasons = array_diff($localSeasons, array('.', '..', 'formatting.txt', '.Trash-1000'));

    foreach($localSeasons as $localSeason) {
      $season = array_search(ltrim(ltrim($localSeason, "S"), 0), array_column($series['seasons'], 'season_number'));

      $cols = array(
        "series-id",
        "number",
        "poster"
      );

      $vals = array(
        $dbSeriesID,
        $series['seasons'][$season]['season_number'],
        $series['seasons'][$season]['poster_path']
      );

      $dbSeasonID = $GLOBALS['db'] -> insertRow("series-seasons", $cols, $vals);

      if(!file_exists("img/posters" . $series['seasons'][$season]['poster_path'])) {
        file_put_contents("img/posters" . $series['seasons'][$season]['poster_path'], fopen($this->bannerURL . $series['seasons'][$season]['poster_path'], 'r'));
      }

      $localEpisodes = scandir($source[0]['path'] . "/" . $path . "/" . $localSeason);
      $localEpisodes = array_diff($localEpisodes, array('.', '..', 'formatting.txt', '.Trash-1000'));

      $dbEpisodes = get_object_vars(json_decode(curl_download("http://api.themoviedb.org/3/tv/" . $seriesID . "/season/" . ltrim(ltrim($localSeason, "S"), 0) . "?api_key=a39779a38e0619f8ae58b09f64522597&language=de&include_image_language=de")));

      $episodes = array();

      foreach($dbEpisodes['episodes'] as $episode) {
        $episodes[] = get_object_vars($episode);
      }

      foreach($localEpisodes as $localEpisode) {
        $episode = array_search($this::getEpisodeNumberByFilename($localEpisode), array_column($episodes, 'episode_number'));

        $cols = array(
          "season-id",
          "number",
          "name",
          "thumb",
          "path"
        );

        $vals = array(
          $dbSeasonID,
          $episodes[$episode]['episode_number'],
          $episodes[$episode]['name'],
          $episodes[$episode]['still_path'],
          $localEpisode
        );

        $GLOBALS['db'] -> insertRow("series-episodes", $cols, $vals);

        if(!file_exists("img/posters" . $episodes[$episode]['still_path'])) {
          file_put_contents("img/posters" . $episodes[$episode]['still_path'], fopen($this->bannerURL . $episodes[$episode]['still_path'], 'r'));
        }
      }
    }

    $this->outputText .= "Done.";
  }
}