yt_dlp/extractor/chilloutzone.py

   1 import base64
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     clean_html,
   6     int_or_none,
   7     traverse_obj,
   8 )
   9
  10
  11 class ChilloutzoneIE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:www\.)?chilloutzone\.net/video/(?P<id>[\w-]+)\.html'
  13     _TESTS = [{
  14         'url': 'https://www.chilloutzone.net/video/enemene-meck-alle-katzen-weg.html',
  15         'md5': 'a76f3457e813ea0037e5244f509e66d1',
  16         'info_dict': {
  17             'id': 'enemene-meck-alle-katzen-weg',
  18             'ext': 'mp4',
  19             'title': 'Enemene Meck - Alle Katzen weg',
  20             'description': 'Ist das der Umkehrschluss des Niesenden Panda-Babys?',
  21             'duration': 24,
  22         },
  23     }, {
  24         'note': 'Video hosted at YouTube',
  25         'url': 'https://www.chilloutzone.net/video/eine-sekunde-bevor.html',
  26         'info_dict': {
  27             'id': '1YVQaAgHyRU',
  28             'ext': 'mp4',
  29             'title': '16 Photos Taken 1 Second Before Disaster',
  30             'description': 'md5:58a8fcf6a459fe0a08f54140f0ad1814',
  31             'uploader': 'BuzzFeedVideo',
  32             'uploader_id': '@BuzzFeedVideo',
  33             'upload_date': '20131105',
  34             'availability': 'public',
  35             'thumbnail': 'https://i.ytimg.com/vi/1YVQaAgHyRU/maxresdefault.jpg',
  36             'tags': 'count:41',
  37             'like_count': int,
  38             'playable_in_embed': True,
  39             'channel_url': 'https://www.youtube.com/channel/UCpko_-a4wgz2u_DgDgd9fqA',
  40             'chapters': 'count:6',
  41             'live_status': 'not_live',
  42             'view_count': int,
  43             'categories': ['Entertainment'],
  44             'age_limit': 0,
  45             'channel_id': 'UCpko_-a4wgz2u_DgDgd9fqA',
  46             'duration': 100,
  47             'uploader_url': 'http://www.youtube.com/@BuzzFeedVideo',
  48             'channel_follower_count': int,
  49             'channel': 'BuzzFeedVideo',
  50         },
  51     }, {
  52         'url': 'https://www.chilloutzone.net/video/icon-blending.html',
  53         'md5': '2f9d6850ec567b24f0f4fa143b9aa2f9',
  54         'info_dict': {
  55             'id': 'LLNkHpSjBfc',
  56             'ext': 'mp4',
  57             'title': 'The Sunday Times   Making of Icons',
  58             'description': 'md5:b9259fcf63a1669e42001e5db677f02a',
  59             'uploader': 'MadFoxUA',
  60             'uploader_id': '@MadFoxUA',
  61             'upload_date': '20140204',
  62             'channel_id': 'UCSZa9Y6-Vl7c11kWMcbAfCw',
  63             'channel_url': 'https://www.youtube.com/channel/UCSZa9Y6-Vl7c11kWMcbAfCw',
  64             'comment_count': int,
  65             'uploader_url': 'http://www.youtube.com/@MadFoxUA',
  66             'duration': 66,
  67             'live_status': 'not_live',
  68             'channel_follower_count': int,
  69             'playable_in_embed': True,
  70             'view_count': int,
  71             'like_count': int,
  72             'thumbnail': 'https://i.ytimg.com/vi/LLNkHpSjBfc/maxresdefault.jpg',
  73             'categories': ['Comedy'],
  74             'availability': 'public',
  75             'tags': [],
  76             'channel': 'MadFoxUA',
  77             'age_limit': 0,
  78         },
  79     }, {
  80         'url': 'https://www.chilloutzone.net/video/ordentlich-abgeschuettelt.html',
  81         'info_dict': {
  82             'id': 'ordentlich-abgeschuettelt',
  83             'ext': 'mp4',
  84             'title': 'Ordentlich abgeschüttelt',
  85             'description': 'md5:d41541966b75d3d1e8ea77a94ea0d329',
  86             'duration': 18,
  87         },
  88     }]
  89
  90     def _real_extract(self, url):
  91         video_id = self._match_id(url)
  92         webpage = self._download_webpage(url, video_id)
  93         b64_data = self._html_search_regex(
  94             r'var cozVidData\s*=\s*"([^"]+)"', webpage, 'video data')
  95         info = self._parse_json(base64.b64decode(b64_data).decode(), video_id)
  96
  97         video_url = info.get('mediaUrl')
  98         native_platform = info.get('nativePlatform')
  99
 100         if native_platform and info.get('sourcePriority') == 'native':
 101             native_video_id = info['nativeVideoId']
 102             if native_platform == 'youtube':
 103                 return self.url_result(native_video_id, 'Youtube')
 104             elif native_platform == 'vimeo':
 105                 return self.url_result(f'https://vimeo.com/{native_video_id}', 'Vimeo')
 106
 107         elif not video_url:
 108             # Possibly a standard youtube embed?
 109             # TODO: Investigate if site still does this (there are no tests for it)
 110             return self.url_result(url, 'Generic')
 111
 112         return {
 113             'id': video_id,
 114             'url': video_url,
 115             'ext': 'mp4',
 116             **traverse_obj(info, {
 117                 'title': 'title',
 118                 'description': ('description', {clean_html}),
 119                 'duration': ('videoLength', {int_or_none}),
 120                 'width': ('videoWidth', {int_or_none}),
 121                 'height': ('videoHeight', {int_or_none}),
 122             }),
 123         }