yt_dlp/extractor/gamestar.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     int_or_none,
   8     remove_end,
   9 )
  10
  11
  12 class GameStarIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?game(?P<site>pro|star)\.de/videos/.*,(?P<id>[0-9]+)\.html'
  14     _TESTS = [{
  15         'url': 'http://www.gamestar.de/videos/trailer,3/hobbit-3-die-schlacht-der-fuenf-heere,76110.html',
  16         'md5': 'ee782f1f8050448c95c5cacd63bc851c',
  17         'info_dict': {
  18             'id': '76110',
  19             'ext': 'mp4',
  20             'title': 'Hobbit 3: Die Schlacht der Fünf Heere - Teaser-Trailer zum dritten Teil',
  21             'description': 'Der Teaser-Trailer zu Hobbit 3: Die Schlacht der Fünf Heere zeigt einige Szenen aus dem dritten Teil der Saga und kündigt den...',
  22             'thumbnail': r're:^https?://.*\.jpg$',
  23             'timestamp': 1406542380,
  24             'upload_date': '20140728',
  25             'duration': 17,
  26         }
  27     }, {
  28         'url': 'http://www.gamepro.de/videos/top-10-indie-spiele-fuer-nintendo-switch-video-tolle-nindies-games-zum-download,95316.html',
  29         'only_matching': True,
  30     }, {
  31         'url': 'http://www.gamestar.de/videos/top-10-indie-spiele-fuer-nintendo-switch-video-tolle-nindies-games-zum-download,95316.html',
  32         'only_matching': True,
  33     }]
  34
  35     def _real_extract(self, url):
  36         mobj = self._match_valid_url(url)
  37         site = mobj.group('site')
  38         video_id = mobj.group('id')
  39
  40         webpage = self._download_webpage(url, video_id)
  41
  42         # TODO: there are multiple ld+json objects in the webpage,
  43         # while _search_json_ld finds only the first one
  44         json_ld = self._parse_json(self._search_regex(
  45             r'(?s)<script[^>]+type=(["\'])application/ld\+json\1[^>]*>(?P<json_ld>[^<]+VideoObject[^<]+)</script>',
  46             webpage, 'JSON-LD', group='json_ld'), video_id)
  47         info_dict = self._json_ld(json_ld, video_id)
  48         info_dict['title'] = remove_end(
  49             info_dict['title'], ' - Game%s' % site.title())
  50
  51         view_count = int_or_none(json_ld.get('interactionCount'))
  52         comment_count = int_or_none(self._html_search_regex(
  53             r'<span>Kommentare</span>\s*<span[^>]+class=["\']count[^>]+>\s*\(\s*([0-9]+)',
  54             webpage, 'comment count', fatal=False))
  55
  56         info_dict.update({
  57             'id': video_id,
  58             'url': 'http://gamestar.de/_misc/videos/portal/getVideoUrl.cfm?premium=0&videoId=' + video_id,
  59             'ext': 'mp4',
  60             'view_count': view_count,
  61             'comment_count': comment_count
  62         })
  63
  64         return info_dict