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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     unified_strdate,
   8 )
   9
  10
  11 class HitboxIE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:www\.)?hitbox\.tv/video/(?P<id>[0-9]+)'
  13     _TEST = {
  14         'url': 'http://www.hitbox.tv/video/203213',
  15         'info_dict': {
  16             'id': '203213',
  17             'title': 'hitbox @ gamescom, Sub Button Hype extended, Giveaway - hitbox News Update with Oxy',
  18             'alt_title': 'hitboxlive - Aug 9th #6',
  19             'description': '\n',
  20             'ext': 'mp4',
  21             'thumbnail': 're:^https?://.*\.jpg$',
  22             'duration': 215,
  23             'resolution': 'HD 720p',
  24             'uploader_id': 'hitboxlive',
  25             'view_count': int,
  26             'upload_date': '20140809',
  27             'categories': ['Live Show'],
  28         },
  29         'params': {
  30             # m3u8 download
  31             'skip_download': True,
  32         },
  33     }
  34
  35     def _extract_metadata(self, url, video_id):
  36         thumb_base = 'https://edge.sf.hitbox.tv'
  37         metadata = self._download_json(
  38             '%s/%s' % (url, video_id), video_id
  39         )
  40
  41         date = 'media_live_since'
  42         media_type = 'livestream'
  43         if metadata.get('media_type') == 'video':
  44             media_type = 'video'
  45             date = 'media_date_added'
  46
  47         video_meta = metadata.get(media_type, [])[0]
  48         title = video_meta.get('media_status')
  49         alt_title = video_meta.get('media_title')
  50         description = video_meta.get('media_description_md')
  51         duration = int(float(video_meta.get('media_duration')))
  52         uploader = video_meta.get('media_user_name')
  53         views = int(video_meta.get('media_views'))
  54         upload_date = unified_strdate(video_meta.get(date))
  55         categories = [video_meta.get('category_name')]
  56         thumbs = [
  57             {'url': thumb_base + video_meta.get('media_thumbnail'),
  58              'width': 320,
  59              'height': 180},
  60             {'url': thumb_base + video_meta.get('media_thumbnail_large'),
  61              'width': 768,
  62              'height': 432},
  63         ]
  64
  65         return {
  66             'id': video_id,
  67             'title': title,
  68             'alt_title': alt_title,
  69             'description': description,
  70             'ext': 'mp4',
  71             'thumbnails': thumbs,
  72             'duration': duration,
  73             'uploader_id': uploader,
  74             'view_count': views,
  75             'upload_date': upload_date,
  76             'categories': categories,
  77         }
  78
  79     def _real_extract(self, url):
  80         video_id = self._match_id(url)
  81
  82         metadata = self._extract_metadata(
  83             'https://www.hitbox.tv/api/media/video',
  84             video_id
  85         )
  86
  87         player_config = self._download_json(
  88             'https://www.hitbox.tv/api/player/config/video/%s' % (video_id),
  89             video_id
  90         )
  91
  92         clip = player_config.get('clip')
  93         video_url = clip.get('url')
  94         res = clip.get('bitrates', [])[0].get('label')
  95
  96         metadata['resolution'] = res
  97         metadata['url'] = video_url
  98         metadata['protocol'] = 'm3u8'
  99
 100         return metadata
 101
 102
 103 class HitboxLiveIE(HitboxIE):
 104     _VALID_URL = r'https?://(?:www\.)?hitbox\.tv/(?!video)(?P<id>.+)'
 105     _TEST = {
 106         'url': 'http://www.hitbox.tv/dimak',
 107         'info_dict': {
 108             'id': 'dimak',
 109             'ext': 'mp4',
 110             'description': str,
 111             'upload_date': str,
 112             'title': str,
 113             'uploader_id': 'Dimak',
 114         },
 115         'params': {
 116             # live
 117             'skip_download': True,
 118         },
 119     }
 120
 121     def _real_extract(self, url):
 122         video_id = self._match_id(url)
 123
 124         metadata = self._extract_metadata(
 125             'https://www.hitbox.tv/api/media/live',
 126             video_id
 127         )
 128
 129         player_config = self._download_json(
 130             'https://www.hitbox.tv/api/player/config/live/%s' % (video_id),
 131             video_id
 132         )
 133
 134         formats = []
 135         cdns = player_config.get('cdns')
 136         servers = []
 137         for cdn in cdns:
 138             base_url = cdn.get('netConnectionUrl')
 139             host = re.search('.+\.([^\.]+\.[^\./]+)/.+', base_url).group(1)
 140             if base_url not in servers:
 141                 servers.append(base_url)
 142                 for stream in cdn.get('bitrates'):
 143                     label = stream.get('label')
 144                     if label != 'Auto':
 145                         formats.append({
 146                             'url': '%s/%s' % (base_url, stream.get('url')),
 147                             'ext': 'mp4',
 148                             'vbr': stream.get('bitrate'),
 149                             'resolution': label,
 150                             'rtmp_live': True,
 151                             'format_note': host,
 152                             'page_url': url,
 153                             'player_url': 'http://www.hitbox.tv/static/player/flowplayer/flowplayer.commercial-3.2.16.swf',
 154                         })
 155
 156         self._sort_formats(formats)
 157         metadata['formats'] = formats
 158         metadata['is_live'] = True
 159         metadata['title'] = self._live_title(metadata.get('title'))
 160         return metadata