moeFlix/inc/movieScraper.php

<?php

ignore_user_abort(true);

class MovieScraper {
  private $apiURL = "http://api.themoviedb.org/3/search/movie?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 getMovieNameFromFilename($filename) {
    preg_match("/([a-zA-Z0-9äöüÄÖÜ.]+).([0-9][0-9][0-9][0-9]).([a-z0-9]+)/", $filename, $output);
    return str_replace(".", " ", $output[1]);
  }

  public static function sortByPopularity($movieA, $movieB) {    
    if($movieA['popularity'] == $movieB['popularity']) {
      return 0;
    }

    return ($movieA['popularity'] < $movieB['popularity']) ? 1 : -1;
  }

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

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

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

    foreach($fileList as $file) {
      $this->outputText .= "<b>" . self::getMovieNameFromFilename($file) . "</b><br>" . PHP_EOL;

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

      $movie = json_decode(curl_download($this->apiURL . urlencode(self::getMovieNameFromFilename($file)) . "&language=de&include_image_language=de"), true);

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

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

      $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 downloadMovieByID($movieID, $path, $sourceID) {

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

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

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

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

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

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