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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import time
   5 import hashlib
   6 import json
   7
   8 from .adobepass import AdobePass
   9 from ..compat import compat_HTTPError
  10 from ..utils import (
  11     int_or_none,
  12     parse_age_limit,
  13     str_or_none,
  14     parse_duration,
  15     ExtractorError,
  16     extract_attributes,
  17 )
  18
  19
  20 class VicelandIE(AdobePass):
  21     _VALID_URL = r'https?://(?:www\.)?viceland\.com/[^/]+/video/[^/]+/(?P<id>[a-f0-9]+)'
  22     _TEST = {
  23         # FIXME: fill the test after fixing delegation problem
  24         'url': 'https://www.viceland.com/en_us/video/cyberwar-trailer/57608447973ee7705f6fbd4e',
  25         'info_dict': {
  26             'id': '57608447973ee7705f6fbd4e',
  27             'ext': 'mp4',
  28         },
  29         'params': {
  30             # m3u8 download
  31             'skip_download': True,
  32         },
  33         'add_ie': ['UplynkPreplay', 'Uplynk'],
  34     }
  35
  36     def _real_extract(self, url):
  37         video_id = self._match_id(url)
  38
  39         webpage = self._download_webpage(url, video_id)
  40         watch_hub_data = extract_attributes(self._search_regex(
  41             r'(?s)(<watch-hub\s*.+?</watch-hub>)', webpage, 'watch hub'))
  42         video_id = watch_hub_data['vms-id']
  43         title = watch_hub_data['video-title']
  44
  45         query = {}
  46         if watch_hub_data.get('video-locked') == '1':
  47             resource = self._get_mvpd_resource(
  48                 'VICELAND', title, video_id,
  49                 watch_hub_data.get('video-rating'))
  50             query['tvetoken'] = self._extract_mvpd_auth(url, video_id, 'VICELAND', resource)
  51
  52         # signature generation algorithm is reverse engineered from signatureGenerator in
  53         # webpack:///../shared/~/vice-player/dist/js/vice-player.js in
  54         # https://www.viceland.com/assets/common/js/web.vendor.bundle.js
  55         exp = int(time.time()) + 14400
  56         query.update({
  57             'exp': exp,
  58             'sign': hashlib.sha512(('%s:GET:%d' % (video_id, exp)).encode()).hexdigest(),
  59         })
  60
  61         try:
  62             preplay = self._download_json('https://www.viceland.com/en_us/preplay/%s' % video_id, video_id, query=query)
  63         except ExtractorError as e:
  64             if isinstance(e.cause, compat_HTTPError) and e.cause.code == 400:
  65                 error = json.loads(e.cause.read().decode())
  66                 raise ExtractorError('%s said: %s' % (self.IE_NAME, error['details']), expected=True)
  67
  68         video_data = preplay['video']
  69         base = video_data['base']
  70         uplynk_preplay_url = preplay['preplayURL']
  71         episode = video_data.get('episode', {})
  72         channel = video_data.get('channel', {})
  73
  74         subtitles = {}
  75         cc_url = preplay.get('ccURL')
  76         if cc_url:
  77             subtitles['en'] = [{
  78                 'url': cc_url,
  79             }]
  80
  81         return {
  82             '_type': 'url_transparent',
  83             'url': uplynk_preplay_url,
  84             'id': video_id,
  85             'title': title,
  86             'description': base.get('body'),
  87             'thumbnail': watch_hub_data.get('cover-image') or watch_hub_data.get('thumbnail'),
  88             'duration': parse_duration(video_data.get('video_duration') or watch_hub_data.get('video-duration')),
  89             'timestamp': int_or_none(video_data.get('created_at')),
  90             'age_limit': parse_age_limit(video_data.get('video_rating')),
  91             'series': video_data.get('show_title') or watch_hub_data.get('show-title'),
  92             'episode_number': int_or_none(episode.get('episode_number') or watch_hub_data.get('episode')),
  93             'episode_id': str_or_none(episode.get('id') or video_data.get('episode_id')),
  94             'season_number': int_or_none(watch_hub_data.get('season')),
  95             'season_id': str_or_none(episode.get('season_id')),
  96             'uploader': channel.get('base', {}).get('title') or watch_hub_data.get('channel-title'),
  97             'uploader_id': str_or_none(channel.get('id')),
  98             'subtitles': subtitles,
  99             'ie_key': 'UplynkPreplay',
 100         }