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) {
      $episodes = $GLOBALS['db']->getAllAssoc("series-episodes", "season-id", $season['id']);
      foreach($episodes as $episode) {
        unlink("img/posters/" . $episode['thumb']);
      }
      $GLOBALS['db']->deleteRows("series-episodes", "season-id", $season['id']);
    }
    $seasons = $GLOBALS['db']->getAllAssoc("series-seasons", "series-id", $seriesID);
    foreach($seasons as $season) {
      unlink("img/posters/" . $season['poster']);
    }
    $GLOBALS['db']->deleteRows("series-seasons", "series-id", $seriesID);
    $series = $GLOBALS['db']->getAllAssoc("series", "id", $seriesID);
    unlink("img/posters/" . $series[0]['poster']);
    unlink("img/posters/" . $series[0]['backdrop']);
    $GLOBALS['db']->deleteRows("series", "id", $seriesID);

      header("Location: " . $GLOBALS['conf']['baseURL'] . "?view=admin");
  }

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

    self::remove($seriesID);

    $this->downloadSeriesByID($series[0]['moviedb-id'], $series[0]['path'], $series[0]['source']);
    header("Location: " . $GLOBALS['conf']['baseURL'] . "?view=admin");
  }

  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', 'queue.hbq'));

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

      if($GLOBALS['db']->countRows("series", "path", $dir) > 0) {
        $this->outputText .= "Exists, updating..<br>" . PHP_EOL;
        self::updateByID($GLOBALS['db']->getString("id", "series", "path", $dir));
        $this->outputText .= "<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 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 /><br />" . PHP_EOL; // TODO: this
        }
      }

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

  }

  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'],
      $path,
      ltrim($series['poster_path'], "/"),
      ltrim($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'],
        ltrim($series['seasons'][$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'],
          ltrim($episodes[$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.";
  }

  public function updateByID($seriesID) {
    $seriesPath = $GLOBALS['db']->customQuery("SELECT `series`.`path` AS 'seriespath', `sources`.`path` AS 'sourcepath' FROM `series`, `sources` WHERE `series`.`id` = " . $seriesID . " AND `series`.`source` = `sources`.`id`");
    $seriesPath = implode("/", array_reverse($seriesPath[0]));

    $series   = $GLOBALS['db']->getAllAssoc("series", "id", $seriesID);
    $seriesDB = json_decode(curl_download("http://api.themoviedb.org/3/tv/" . $series[0]['moviedb-id'] . "?api_key=a39779a38e0619f8ae58b09f64522597&language=de&include_image_language=de"), true);

    if($series[0]['poster'] != ltrim($seriesDB['poster_path'], "/")) { // poster
      //error_log("New Poster, old: " . $series[0]['poster'] . " new: " . ltrim($seriesDB['poster_path'], "/"));
      if(file_exists("img/posters/" . $series[0]['poster']) && !is_dir("img/posters/" . $series[0]['poster'])) {
        unlink("img/posters/" . $series[0]['poster']);
      }
      file_put_contents("img/posters" . $seriesDB['poster_path'], fopen($this->bannerURL . $seriesDB['poster_path'], 'r'));
      $GLOBALS['db']->updateRow("series", "poster", '"' . ltrim($seriesDB['poster_path'], "/") . '"', "id", $seriesID);
    }

    if($series[0]['backdrop'] != ltrim($seriesDB['backdrop_path'], "/")) { // backdrop
      //error_log("New Backdrop, old: " . $series[0]['backdrop'] . " new: " . ltrim($seriesDB['backdrop_path'], "/"));
      if(file_exists("img/posters/" . $series[0]['backdrop']) && !is_dir("img/posters/" . $series[0]['backdrop'])) {
        unlink("img/posters/" . $series[0]['backdrop']);
      }
      file_put_contents("img/posters" . $seriesDB['backdrop_path'], fopen($this->bannerURL . $seriesDB['backdrop_path'], 'r'));
      $GLOBALS['db']->updateRow("series", "backdrop", '"' . ltrim($seriesDB['backdrop_path'], "/") . '"', "id", $seriesID);
    }

    $seasons = $GLOBALS['db']->getAllAssoc("series-seasons", "series-id", $seriesID);
    $localSeasons = scandir($seriesPath);
    $localSeasons = array_diff($localSeasons, array('.', '..', 'formatting.txt', '.Trash-1000', 'queue.hbq'));

    foreach($seasons as $season) { // TODO: check for new seasons
      if(!in_array("S" . sprintf("%02d", $season['number']), $localSeasons)) {
        //error_log("Season " . "S" . sprintf("%02d", $season['number']) . " not existing anymore");
        $GLOBALS['db']->deleteRows("series-seasons", "id", $seasons['id']);
        continue;
      }

      unset($localSeasons[array_search("S" . sprintf("%02d", $season['number']), $localSeasons)]);

      $seasonDB = array_search(ltrim($season['number'], 0), array_column($seriesDB['seasons'], 'season_number'));
      $seasonDB = $seriesDB['seasons'][$seasonDB];

      if($season['poster'] != ltrim($seasonDB['poster_path'], "/")) {
        //error_log("New Season-Poster, old: " . $season['poster'] . " new: " . ltrim($seasonDB['poster_path'], "/"));
        if(file_exists("img/posters/" . $season['poster']) && !is_dir("img/posters/" . $season['poster'])) {
          unlink("img/posters/" . $season['poster']);
        }
        file_put_contents("img/posters" . $seasonDB['poster_path'], fopen($this->bannerURL . $seasonDB['poster_path'], 'r'));
        $GLOBALS['db']->updateRow("series-seasons", "poster", '"' . ltrim($seasonDB['poster_path'], "/") . '"', "id", $season['id']);
      }

      $episodesDB = json_decode(curl_download("http://api.themoviedb.org/3/tv/" . $series[0]['moviedb-id'] . "/season/" . $season['number'] . "?api_key=a39779a38e0619f8ae58b09f64522597&language=de&include_image_language=de"), true);

      $episodes = $GLOBALS['db']->getAllAssoc("series-episodes", "season-id", $season['id']);
      $localEpisodes = scandir($seriesPath . "/" . "S" . sprintf("%02d", $season['number']));
      $localEpisodes = array_diff($localEpisodes, array('.', '..', 'formatting.txt', '.Trash-1000', 'queue.hbq'));

      foreach($episodes as $episode) {
        if(!in_array($episode['path'], $localEpisodes)) {
          //error_log("Episode " . $episode['number'] . " not existing anymore");
          $GLOBALS['db']->deleteRows("series-episodes", "id", $episode['id']);
          continue;
        }

        unset($localEpisodes[array_search($episode['path'], $localEpisodes)]);

        $episodeDB = array_search($episode['number'], array_column($episodesDB['episodes'], 'episode_number'));
        $episodeDB = $episodesDB['episodes'][$episodeDB];

        if($episode['thumb'] != ltrim($episodeDB['still_path'], "/")) {
          //error_log("New Episode-Poster, old: " . $episode['thumb'] . " new: " . ltrim($episodeDB['still_path'], "/"));
          if(file_exists("img/posters/" . $episode['thumb']) && !is_dir("img/posters/" . $episode['thumb'])) {
            unlink("img/posters/" . $episode['thumb']);
          }
          file_put_contents("img/posters" . $episodeDB['still_path'], fopen($this->bannerURL . $episodeDB['still_path'], 'r'));
          $GLOBALS['db']->updateRow("series-episodes", "thumb", '"' . ltrim($episodeDB['still_path'], "/") . '"', "id", $episode['id']);
        }
      }

      if(sizeof($localEpisodes) > 0) {
        //error_log("New Episode(s); " . print_r($localEpisodes, true));
        $this->outputText .= "New Episode(s), Please update again.<br>";

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

        foreach($localEpisodes as $localEpisode) {
          $vals = array(
            $season['id'],
            ltrim(ltrim(self::getEpisodeNumberByFilename($localEpisode), "S"), 0),
            "",
            "",
            $localEpisode
          );

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

    if(sizeof($localSeasons) > 0) {
      //error_log("New Season(s): " . print_r($localSeasons, true));
      $this->outputText .= "New Season(s), Please update again.<br>";

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

      foreach($localSeasons as $localSeason) {
        $vals = array(
          $seriesID,
          ltrim(ltrim($localSeason, "S"), 0),
          "",
          1
        );
        $GLOBALS['db']->insertRow("series-seasons", $cols, $vals);
      }
    }

  }
}