Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/hitbox.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     clean_html,
   9     parse_iso8601,
  10     float_or_none,
  11     int_or_none,
  12     compat_str,
  13 )
  14
  15
  16 class HitboxIE(InfoExtractor):
  17     IE_NAME = 'hitbox'
  18     _VALID_URL = r'https?://(?:www\.)?hitbox\.tv/video/(?P<id>[0-9]+)'
  19     _TEST = {
  20         'url': 'http://www.hitbox.tv/video/203213',
  21         'info_dict': {
  22             'id': '203213',
  23             'title': 'hitbox @ gamescom, Sub Button Hype extended, Giveaway - hitbox News Update with Oxy',
  24             'alt_title': 'hitboxlive - Aug 9th #6',
  25             'description': '',
  26             'ext': 'mp4',
  27             'thumbnail': 're:^https?://.*\.jpg$',
  28             'duration': 215.1666,
  29             'resolution': 'HD 720p',
  30             'uploader': 'hitboxlive',
  31             'view_count': int,
  32             'timestamp': 1407576133,
  33             'upload_date': '20140809',
  34             'categories': ['Live Show'],
  35         },
  36         'params': {
  37             # m3u8 download
  38             'skip_download': True,
  39         },
  40     }
  41
  42     def _extract_metadata(self, url, video_id):
  43         thumb_base = 'https://edge.sf.hitbox.tv'
  44         metadata = self._download_json(
  45             '%s/%s' % (url, video_id), video_id)
  46
  47         date = 'media_live_since'
  48         media_type = 'livestream'
  49         if metadata.get('media_type') == 'video':
  50             media_type = 'video'
  51             date = 'media_date_added'
  52
  53         video_meta = metadata.get(media_type, [])[0]
  54         title = video_meta.get('media_status')
  55         alt_title = video_meta.get('media_title')
  56         description = clean_html(
  57             video_meta.get('media_description') or
  58             video_meta.get('media_description_md'))
  59         duration = float_or_none(video_meta.get('media_duration'))
  60         uploader = video_meta.get('media_user_name')
  61         views = int_or_none(video_meta.get('media_views'))
  62         timestamp = parse_iso8601(video_meta.get(date), ' ')
  63         categories = [video_meta.get('category_name')]
  64         thumbs = [
  65             {'url': thumb_base + video_meta.get('media_thumbnail'),
  66              'width': 320,
  67              'height': 180},
  68             {'url': thumb_base + video_meta.get('media_thumbnail_large'),
  69              'width': 768,
  70              'height': 432},
  71         ]
  72
  73         return {
  74             'id': video_id,
  75             'title': title,
  76             'alt_title': alt_title,
  77             'description': description,
  78             'ext': 'mp4',
  79             'thumbnails': thumbs,
  80             'duration': duration,
  81             'uploader': uploader,
  82             'view_count': views,
  83             'timestamp': timestamp,
  84             'categories': categories,
  85         }
  86
  87     def _real_extract(self, url):
  88         video_id = self._match_id(url)
  89
  90         metadata = self._extract_metadata(
  91             'https://www.hitbox.tv/api/media/video',
  92             video_id)
  93
  94         player_config = self._download_json(
  95             'https://www.hitbox.tv/api/player/config/video/%s' % video_id,
  96             video_id)
  97
  98         clip = player_config.get('clip')
  99         video_url = clip.get('url')
 100         res = clip.get('bitrates', [])[0].get('label')
 101
 102         metadata['resolution'] = res
 103         metadata['url'] = video_url
 104         metadata['protocol'] = 'm3u8'
 105
 106         return metadata
 107
 108
 109 class HitboxLiveIE(HitboxIE):
 110     IE_NAME = 'hitbox:live'
 111     _VALID_URL = r'https?://(?:www\.)?hitbox\.tv/(?!video)(?P<id>.+)'
 112     _TEST = {
 113         'url': 'http://www.hitbox.tv/dimak',
 114         'info_dict': {
 115             'id': 'dimak',
 116             'ext': 'mp4',
 117             'description': 'md5:c9f80fa4410bc588d7faa40003fc7d0e',
 118             'timestamp': int,
 119             'upload_date': compat_str,
 120             'title': compat_str,
 121             'uploader': 'Dimak',
 122         },
 123         'params': {
 124             # live
 125             'skip_download': True,
 126         },
 127     }
 128
 129     def _real_extract(self, url):
 130         video_id = self._match_id(url)
 131
 132         metadata = self._extract_metadata(
 133             'https://www.hitbox.tv/api/media/live',
 134             video_id)
 135
 136         player_config = self._download_json(
 137             'https://www.hitbox.tv/api/player/config/live/%s' % video_id,
 138             video_id)
 139
 140         formats = []
 141         cdns = player_config.get('cdns')
 142         servers = []
 143         for cdn in cdns:
 144             base_url = cdn.get('netConnectionUrl')
 145             host = re.search('.+\.([^\.]+\.[^\./]+)/.+', base_url).group(1)
 146             if base_url not in servers:
 147                 servers.append(base_url)
 148                 for stream in cdn.get('bitrates'):
 149                     label = stream.get('label')
 150                     if label != 'Auto':
 151                         formats.append({
 152                             'url': '%s/%s' % (base_url, stream.get('url')),
 153                             'ext': 'mp4',
 154                             'vbr': stream.get('bitrate'),
 155                             'resolution': label,
 156                             'rtmp_live': True,
 157                             'format_note': host,
 158                             'page_url': url,
 159                             'player_url': 'http://www.hitbox.tv/static/player/flowplayer/flowplayer.commercial-3.2.16.swf',
 160                         })
 161
 162         self._sort_formats(formats)
 163         metadata['formats'] = formats
 164         metadata['is_live'] = True
 165         metadata['title'] = self._live_title(metadata.get('title'))
 166         return metadata