_ Git - youtube-dl/blob - youtube_dl/extractor/periscope.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     parse_iso8601,
   7     unescapeHTML,
   8 )
   9
  10
  11 class PeriscopeIE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:www\.)?periscope\.tv/w/(?P<id>[^/?#]+)'
  13     _TEST = {
  14         'url': 'https://www.periscope.tv/w/aJUQnjY3MjA3ODF8NTYxMDIyMDl2zCg2pECBgwTqRpQuQD352EMPTKQjT4uqlM3cgWFA-g==',
  15         'md5': '65b57957972e503fcbbaeed8f4fa04ca',
  16         'info_dict': {
  17             'id': '56102209',
  18             'ext': 'mp4',
  19             'title': 'Bec Boop - 🚠✈️🇬🇧 Fly above #London in Emirates Air Line cable car at night 🇬🇧✈️🚠 #BoopScope 🎀💗',
  20             'timestamp': 1438978559,
  21             'upload_date': '20150807',
  22             'uploader': 'Bec Boop',
  23             'uploader_id': '1465763',
  24         },
  25         'skip': 'Expires in 24 hours',
  26     }
  27
  28     def _real_extract(self, url):
  29         video_id = self._match_id(url)
  30
  31         replay = self._download_json(
  32             'https://api.periscope.tv/api/v2/getAccessPublic?token=%s' % video_id, video_id)
  33
  34         video_url = replay['replay_url']
  35
  36         webpage = self._download_webpage(url, video_id)
  37
  38         broadcast_data = self._parse_json(
  39             unescapeHTML(self._html_search_meta(
  40                 'broadcast-data', webpage, 'broadcast data', fatal=True)),
  41             video_id)
  42
  43         broadcast = broadcast_data['broadcast']
  44         status = broadcast['status']
  45
  46         uploader = broadcast.get('user_display_name') or broadcast_data.get('user', {}).get('display_name')
  47         uploader_id = broadcast.get('user_id') or broadcast_data.get('user', {}).get('id')
  48
  49         title = '%s - %s' % (uploader, status) if uploader else status
  50         timestamp = parse_iso8601(broadcast.get('created_at'))
  51
  52         thumbnails = [{
  53             'url': broadcast[image],
  54         } for image in ('image_url', 'image_url_small') if broadcast.get(image)]
  55
  56         return {
  57             'id': broadcast.get('id') or video_id,
  58             'url': video_url,
  59             'ext': 'mp4',
  60             'protocol': 'm3u8_native',
  61             'title': title,
  62             'timestamp': timestamp,
  63             'uploader': uploader,
  64             'uploader_id': uploader_id,
  65             'thumbnails': thumbnails,
  66         }