yt_dlp/extractor/moviepilot.py

   1 from .dailymotion import DailymotionIE
   2 from .common import InfoExtractor
   3 from ..utils import (
   4     parse_iso8601,
   5     try_get,
   6 )
   7
   8 import re
   9
  10
  11 class MoviepilotIE(InfoExtractor):
  12     _IE_NAME = 'moviepilot'
  13     _IE_DESC = 'Moviepilot trailer'
  14     _VALID_URL = r'https?://(?:www\.)?moviepilot\.de/movies/(?P<id>[^/]+)'
  15
  16     _TESTS = [{
  17         'url': 'https://www.moviepilot.de/movies/interstellar-2/',
  18         'info_dict': {
  19             'id': 'x7xdut5',
  20             'display_id': 'interstellar-2',
  21             'ext': 'mp4',
  22             'title': 'Interstellar',
  23             'thumbnail': r're:https://\w+\.dmcdn\.net/v/SaXev1VvzitVZMFsR/x720',
  24             'timestamp': 1400491705,
  25             'description': 'md5:7dfc5c1758e7322a7346934f1f0c489c',
  26             'uploader': 'Moviepilot',
  27             'like_count': int,
  28             'view_count': int,
  29             'uploader_id': 'x6nd9k',
  30             'upload_date': '20140519',
  31             'duration': 140,
  32             'age_limit': 0,
  33             'tags': ['Alle Trailer', 'Movie', 'Third Party'],
  34         },
  35     }, {
  36         'url': 'https://www.moviepilot.de/movies/interstellar-2/trailer',
  37         'only_matching': True,
  38     }, {
  39         'url': 'https://www.moviepilot.de/movies/interstellar-2/kinoprogramm/berlin',
  40         'only_matching': True,
  41     }, {
  42         'url': 'https://www.moviepilot.de/movies/queen-slim/trailer',
  43         'info_dict': {
  44             'id': 'x7xj6o7',
  45             'display_id': 'queen-slim',
  46             'title': 'Queen & Slim',
  47             'ext': 'mp4',
  48             'thumbnail': r're:https://\w+\.dmcdn\.net/v/SbUM71WtomSjVmI_q/x720',
  49             'timestamp': 1571838685,
  50             'description': 'md5:73058bcd030aa12d991e4280d65fbebe',
  51             'uploader': 'Moviepilot',
  52             'like_count': int,
  53             'view_count': int,
  54             'uploader_id': 'x6nd9k',
  55             'upload_date': '20191023',
  56             'duration': 138,
  57             'age_limit': 0,
  58             'tags': ['Movie', 'Verleih', 'Neue Trailer'],
  59         },
  60     }, {
  61         'url': 'https://www.moviepilot.de/movies/der-geiger-von-florenz/trailer',
  62         'info_dict': {
  63             'id': 'der-geiger-von-florenz',
  64             'title': 'Der Geiger von Florenz',
  65             'ext': 'mp4',
  66         },
  67         'skip': 'No trailer for this movie.',
  68     }, {
  69         'url': 'https://www.moviepilot.de/movies/muellers-buero/',
  70         'info_dict': {
  71             'id': 'x7xcw1i',
  72             'display_id': 'muellers-buero',
  73             'title': 'Müllers Büro',
  74             'ext': 'mp4',
  75             'description': 'md5:57501251c05cdc61ca314b7633e0312e',
  76             'timestamp': 1287584475,
  77             'age_limit': 0,
  78             'duration': 82,
  79             'upload_date': '20101020',
  80             'thumbnail': r're:https://\w+\.dmcdn\.net/v/SaMes1WfAm1d6maq_/x720',
  81             'uploader': 'Moviepilot',
  82             'like_count': int,
  83             'view_count': int,
  84             'tags': ['Alle Trailer', 'Movie', 'Verleih'],
  85             'uploader_id': 'x6nd9k',
  86         },
  87     }]
  88
  89     def _real_extract(self, url):
  90         video_id = self._match_id(url)
  91
  92         webpage = self._download_webpage(f'https://www.moviepilot.de/movies/{video_id}/trailer', video_id)
  93
  94         duration = try_get(
  95             re.match(r'P(?P<hours>\d+)H(?P<mins>\d+)M(?P<secs>\d+)S',
  96                      self._html_search_meta('duration', webpage, fatal=False) or ''),
  97             lambda mobj: sum(float(x) * y for x, y in zip(mobj.groups(), (3600, 60, 1))))
  98         # _html_search_meta is not used since we don't want name=description to match
  99         description = self._html_search_regex(
 100             '<meta[^>]+itemprop="description"[^>]+content="([^>"]+)"', webpage, 'description', fatal=False)
 101
 102         return {
 103             '_type': 'url_transparent',
 104             'ie_key': DailymotionIE.ie_key(),
 105             'display_id': video_id,
 106             'title': self._og_search_title(webpage),
 107             'url': self._html_search_meta('embedURL', webpage),
 108             'thumbnail': self._html_search_meta('thumbnailURL', webpage),
 109             'description': description,
 110             'duration': duration,
 111             'timestamp': parse_iso8601(self._html_search_meta('uploadDate', webpage), delimiter=' ')
 112         }