yt_dlp/extractor/periscope.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     int_or_none,
   4     parse_iso8601,
   5     unescapeHTML,
   6 )
   7
   8
   9 class PeriscopeBaseIE(InfoExtractor):
  10     _M3U8_HEADERS = {
  11         'Referer': 'https://www.periscope.tv/'
  12     }
  13
  14     def _call_api(self, method, query, item_id):
  15         return self._download_json(
  16             'https://api.periscope.tv/api/v2/%s' % method,
  17             item_id, query=query)
  18
  19     def _parse_broadcast_data(self, broadcast, video_id):
  20         title = broadcast.get('status') or 'Periscope Broadcast'
  21         uploader = broadcast.get('user_display_name') or broadcast.get('username')
  22         title = '%s - %s' % (uploader, title) if uploader else title
  23         is_live = broadcast.get('state').lower() == 'running'
  24
  25         thumbnails = [{
  26             'url': broadcast[image],
  27         } for image in ('image_url', 'image_url_small') if broadcast.get(image)]
  28
  29         return {
  30             'id': broadcast.get('id') or video_id,
  31             'title': title,
  32             'timestamp': parse_iso8601(broadcast.get('created_at')),
  33             'uploader': uploader,
  34             'uploader_id': broadcast.get('user_id') or broadcast.get('username'),
  35             'thumbnails': thumbnails,
  36             'view_count': int_or_none(broadcast.get('total_watched')),
  37             'tags': broadcast.get('tags'),
  38             'is_live': is_live,
  39         }
  40
  41     @staticmethod
  42     def _extract_common_format_info(broadcast):
  43         return broadcast.get('state').lower(), int_or_none(broadcast.get('width')), int_or_none(broadcast.get('height'))
  44
  45     @staticmethod
  46     def _add_width_and_height(f, width, height):
  47         for key, val in (('width', width), ('height', height)):
  48             if not f.get(key):
  49                 f[key] = val
  50
  51     def _extract_pscp_m3u8_formats(self, m3u8_url, video_id, format_id, state, width, height, fatal=True):
  52         m3u8_formats = self._extract_m3u8_formats(
  53             m3u8_url, video_id, 'mp4',
  54             entry_protocol='m3u8_native'
  55             if state in ('ended', 'timed_out') else 'm3u8',
  56             m3u8_id=format_id, fatal=fatal, headers=self._M3U8_HEADERS)
  57         if len(m3u8_formats) == 1:
  58             self._add_width_and_height(m3u8_formats[0], width, height)
  59         for f in m3u8_formats:
  60             f.setdefault('http_headers', {}).update(self._M3U8_HEADERS)
  61         return m3u8_formats
  62
  63
  64 class PeriscopeIE(PeriscopeBaseIE):
  65     IE_DESC = 'Periscope'
  66     IE_NAME = 'periscope'
  67     _VALID_URL = r'https?://(?:www\.)?(?:periscope|pscp)\.tv/[^/]+/(?P<id>[^/?#]+)'
  68     _EMBED_REGEX = [r'<iframe[^>]+src=([\'"])(?P<url>(?:https?:)?//(?:www\.)?(?:periscope|pscp)\.tv/(?:(?!\1).)+)\1']
  69     # Alive example URLs can be found here https://www.periscope.tv/
  70     _TESTS = [{
  71         'url': 'https://www.periscope.tv/w/aJUQnjY3MjA3ODF8NTYxMDIyMDl2zCg2pECBgwTqRpQuQD352EMPTKQjT4uqlM3cgWFA-g==',
  72         'md5': '65b57957972e503fcbbaeed8f4fa04ca',
  73         'info_dict': {
  74             'id': '56102209',
  75             'ext': 'mp4',
  76             'title': 'Bec Boop - 🚠✈️🇬🇧 Fly above #London in Emirates Air Line cable car at night 🇬🇧✈️🚠 #BoopScope 🎀💗',
  77             'timestamp': 1438978559,
  78             'upload_date': '20150807',
  79             'uploader': 'Bec Boop',
  80             'uploader_id': '1465763',
  81         },
  82         'skip': 'Expires in 24 hours',
  83     }, {
  84         'url': 'https://www.periscope.tv/w/1ZkKzPbMVggJv',
  85         'only_matching': True,
  86     }, {
  87         'url': 'https://www.periscope.tv/bastaakanoggano/1OdKrlkZZjOJX',
  88         'only_matching': True,
  89     }, {
  90         'url': 'https://www.periscope.tv/w/1ZkKzPbMVggJv',
  91         'only_matching': True,
  92     }]
  93
  94     def _real_extract(self, url):
  95         token = self._match_id(url)
  96
  97         stream = self._call_api(
  98             'accessVideoPublic', {'broadcast_id': token}, token)
  99
 100         broadcast = stream['broadcast']
 101         info = self._parse_broadcast_data(broadcast, token)
 102
 103         state = broadcast.get('state').lower()
 104         width = int_or_none(broadcast.get('width'))
 105         height = int_or_none(broadcast.get('height'))
 106
 107         def add_width_and_height(f):
 108             for key, val in (('width', width), ('height', height)):
 109                 if not f.get(key):
 110                     f[key] = val
 111
 112         video_urls = set()
 113         formats = []
 114         for format_id in ('replay', 'rtmp', 'hls', 'https_hls', 'lhls', 'lhlsweb'):
 115             video_url = stream.get(format_id + '_url')
 116             if not video_url or video_url in video_urls:
 117                 continue
 118             video_urls.add(video_url)
 119             if format_id != 'rtmp':
 120                 m3u8_formats = self._extract_pscp_m3u8_formats(
 121                     video_url, token, format_id, state, width, height, False)
 122                 formats.extend(m3u8_formats)
 123                 continue
 124             rtmp_format = {
 125                 'url': video_url,
 126                 'ext': 'flv' if format_id == 'rtmp' else 'mp4',
 127             }
 128             self._add_width_and_height(rtmp_format)
 129             formats.append(rtmp_format)
 130         self._sort_formats(formats)
 131
 132         info['formats'] = formats
 133         return info
 134
 135
 136 class PeriscopeUserIE(PeriscopeBaseIE):
 137     _VALID_URL = r'https?://(?:www\.)?(?:periscope|pscp)\.tv/(?P<id>[^/]+)/?$'
 138     IE_DESC = 'Periscope user videos'
 139     IE_NAME = 'periscope:user'
 140
 141     _TEST = {
 142         'url': 'https://www.periscope.tv/LularoeHusbandMike/',
 143         'info_dict': {
 144             'id': 'LularoeHusbandMike',
 145             'title': 'LULAROE HUSBAND MIKE',
 146             'description': 'md5:6cf4ec8047768098da58e446e82c82f0',
 147         },
 148         # Periscope only shows videos in the last 24 hours, so it's possible to
 149         # get 0 videos
 150         'playlist_mincount': 0,
 151     }
 152
 153     def _real_extract(self, url):
 154         user_name = self._match_id(url)
 155
 156         webpage = self._download_webpage(url, user_name)
 157
 158         data_store = self._parse_json(
 159             unescapeHTML(self._search_regex(
 160                 r'data-store=(["\'])(?P<data>.+?)\1',
 161                 webpage, 'data store', default='{}', group='data')),
 162             user_name)
 163
 164         user = list(data_store['UserCache']['users'].values())[0]['user']
 165         user_id = user['id']
 166         session_id = data_store['SessionToken']['public']['broadcastHistory']['token']['session_id']
 167
 168         broadcasts = self._call_api(
 169             'getUserBroadcastsPublic',
 170             {'user_id': user_id, 'session_id': session_id},
 171             user_name)['broadcasts']
 172
 173         broadcast_ids = [
 174             broadcast['id'] for broadcast in broadcasts if broadcast.get('id')]
 175
 176         title = user.get('display_name') or user.get('username') or user_name
 177         description = user.get('description')
 178
 179         entries = [
 180             self.url_result(
 181                 'https://www.periscope.tv/%s/%s' % (user_name, broadcast_id))
 182             for broadcast_id in broadcast_ids]
 183
 184         return self.playlist_result(entries, user_id, title, description)