yt_dlp/extractor/teamcoco.py

   1 import json
   2
   3 from .turner import TurnerBaseIE
   4 from ..utils import (
   5     determine_ext,
   6     ExtractorError,
   7     int_or_none,
   8     mimetype2ext,
   9     parse_duration,
  10     parse_iso8601,
  11     qualities,
  12 )
  13
  14
  15 class TeamcocoIE(TurnerBaseIE):
  16     _VALID_URL = r'https?://(?:\w+\.)?teamcoco\.com/(?P<id>([^/]+/)*[^/?#]+)'
  17     _TESTS = [
  18         {
  19             'url': 'http://teamcoco.com/video/mary-kay-remote',
  20             'md5': '55d532f81992f5c92046ad02fec34d7d',
  21             'info_dict': {
  22                 'id': '80187',
  23                 'ext': 'mp4',
  24                 'title': 'Conan Becomes A Mary Kay Beauty Consultant',
  25                 'description': 'Mary Kay is perhaps the most trusted name in female beauty, so of course Conan is a natural choice to sell their products.',
  26                 'duration': 495.0,
  27                 'upload_date': '20140402',
  28                 'timestamp': 1396407600,
  29             }
  30         }, {
  31             'url': 'http://teamcoco.com/video/louis-ck-interview-george-w-bush',
  32             'md5': 'cde9ba0fa3506f5f017ce11ead928f9a',
  33             'info_dict': {
  34                 'id': '19705',
  35                 'ext': 'mp4',
  36                 'description': 'Louis C.K. got starstruck by George W. Bush, so what? Part one.',
  37                 'title': 'Louis C.K. Interview Pt. 1 11/3/11',
  38                 'duration': 288,
  39                 'upload_date': '20111104',
  40                 'timestamp': 1320405840,
  41             }
  42         }, {
  43             'url': 'http://teamcoco.com/video/timothy-olyphant-drinking-whiskey',
  44             'info_dict': {
  45                 'id': '88748',
  46                 'ext': 'mp4',
  47                 'title': 'Timothy Olyphant Raises A Toast To “Justified”',
  48                 'description': 'md5:15501f23f020e793aeca761205e42c24',
  49                 'upload_date': '20150415',
  50                 'timestamp': 1429088400,
  51             },
  52             'params': {
  53                 'skip_download': True,  # m3u8 downloads
  54             }
  55         }, {
  56             'url': 'http://teamcoco.com/video/full-episode-mon-6-1-joel-mchale-jake-tapper-and-musical-guest-courtney-barnett?playlist=x;eyJ0eXBlIjoidGFnIiwiaWQiOjl9',
  57             'info_dict': {
  58                 'id': '89341',
  59                 'ext': 'mp4',
  60                 'title': 'Full Episode - Mon. 6/1 - Joel McHale, Jake Tapper, And Musical Guest Courtney Barnett',
  61                 'description': 'Guests: Joel McHale, Jake Tapper, And Musical Guest Courtney Barnett',
  62             },
  63             'params': {
  64                 'skip_download': True,  # m3u8 downloads
  65             },
  66             'skip': 'This video is no longer available.',
  67         }, {
  68             'url': 'http://teamcoco.com/video/the-conan-audiencey-awards-for-04/25/18',
  69             'only_matching': True,
  70         }, {
  71             'url': 'http://teamcoco.com/italy/conan-jordan-schlansky-hit-the-streets-of-florence',
  72             'only_matching': True,
  73         }, {
  74             'url': 'http://teamcoco.com/haiti/conan-s-haitian-history-lesson',
  75             'only_matching': True,
  76         }, {
  77             'url': 'http://teamcoco.com/israel/conan-hits-the-streets-beaches-of-tel-aviv',
  78             'only_matching': True,
  79         }, {
  80             'url': 'https://conan25.teamcoco.com/video/ice-cube-kevin-hart-conan-share-lyft',
  81             'only_matching': True,
  82         }
  83     ]
  84     _RECORD_TEMPL = '''id
  85         title
  86         teaser
  87         publishOn
  88         thumb {
  89           preview
  90         }
  91         tags {
  92           name
  93         }
  94         duration
  95         turnerMediaId
  96         turnerMediaAuthToken'''
  97
  98     def _graphql_call(self, query_template, object_type, object_id):
  99         find_object = 'find' + object_type
 100         return self._download_json(
 101             'https://teamcoco.com/graphql', object_id, data=json.dumps({
 102                 'query': query_template % (find_object, object_id)
 103             }).encode(), headers={
 104                 'Content-Type': 'application/json',
 105             })['data'][find_object]
 106
 107     def _real_extract(self, url):
 108         display_id = self._match_id(url)
 109
 110         response = self._graphql_call('''{
 111   %%s(slug: "%%s") {
 112     ... on RecordSlug {
 113       record {
 114         %s
 115       }
 116     }
 117     ... on PageSlug {
 118       child {
 119         id
 120       }
 121     }
 122     ... on NotFoundSlug {
 123       status
 124     }
 125   }
 126 }''' % self._RECORD_TEMPL, 'Slug', display_id)
 127         if response.get('status'):
 128             raise ExtractorError('This video is no longer available.', expected=True)
 129
 130         child = response.get('child')
 131         if child:
 132             record = self._graphql_call('''{
 133   %%s(id: "%%s") {
 134     ... on Video {
 135       %s
 136     }
 137   }
 138 }''' % self._RECORD_TEMPL, 'Record', child['id'])
 139         else:
 140             record = response['record']
 141         video_id = record['id']
 142
 143         info = {
 144             'id': video_id,
 145             'display_id': display_id,
 146             'title': record['title'],
 147             'thumbnail': record.get('thumb', {}).get('preview'),
 148             'description': record.get('teaser'),
 149             'duration': parse_duration(record.get('duration')),
 150             'timestamp': parse_iso8601(record.get('publishOn')),
 151         }
 152
 153         media_id = record.get('turnerMediaId')
 154         if media_id:
 155             self._initialize_geo_bypass({
 156                 'countries': ['US'],
 157             })
 158             info.update(self._extract_ngtv_info(media_id, {
 159                 'accessToken': record['turnerMediaAuthToken'],
 160                 'accessTokenType': 'jws',
 161             }))
 162         else:
 163             video_sources = self._download_json(
 164                 'https://teamcoco.com/_truman/d/' + video_id,
 165                 video_id)['meta']['src']
 166             if isinstance(video_sources, dict):
 167                 video_sources = video_sources.values()
 168
 169             formats = []
 170             get_quality = qualities(['low', 'sd', 'hd', 'uhd'])
 171             for src in video_sources:
 172                 if not isinstance(src, dict):
 173                     continue
 174                 src_url = src.get('src')
 175                 if not src_url:
 176                     continue
 177                 format_id = src.get('label')
 178                 ext = determine_ext(src_url, mimetype2ext(src.get('type')))
 179                 if format_id == 'hls' or ext == 'm3u8':
 180                     # compat_urllib_parse.urljoin does not work here
 181                     if src_url.startswith('/'):
 182                         src_url = 'http://ht.cdn.turner.com/tbs/big/teamcoco' + src_url
 183                     formats.extend(self._extract_m3u8_formats(
 184                         src_url, video_id, 'mp4', m3u8_id=format_id, fatal=False))
 185                 else:
 186                     if src_url.startswith('/mp4:protected/'):
 187                         # TODO Correct extraction for these files
 188                         continue
 189                     tbr = int_or_none(self._search_regex(
 190                         r'(\d+)k\.mp4', src_url, 'tbr', default=None))
 191
 192                     formats.append({
 193                         'url': src_url,
 194                         'ext': ext,
 195                         'tbr': tbr,
 196                         'format_id': format_id,
 197                         'quality': get_quality(format_id),
 198                     })
 199             info['formats'] = formats
 200
 201         return info