Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/sportdeutschland.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     parse_iso8601,
   9     sanitized_Request,
  10 )
  11
  12
  13 class SportDeutschlandIE(InfoExtractor):
  14     _VALID_URL = r'https?://sportdeutschland\.tv/(?P<sport>[^/?#]+)/(?P<id>[^?#/]+)(?:$|[?#])'
  15     _TESTS = [{
  16         'url': 'http://sportdeutschland.tv/badminton/live-li-ning-badminton-weltmeisterschaft-2014-kopenhagen',
  17         'info_dict': {
  18             'id': 'live-li-ning-badminton-weltmeisterschaft-2014-kopenhagen',
  19             'ext': 'mp4',
  20             'title': 're:Li-Ning Badminton Weltmeisterschaft 2014 Kopenhagen',
  21             'categories': ['Badminton'],
  22             'view_count': int,
  23             'thumbnail': r're:^https?://.*\.jpg$',
  24             'description': r're:Die Badminton-WM 2014 aus Kopenhagen bei Sportdeutschland\.TV',
  25             'timestamp': int,
  26             'upload_date': 're:^201408[23][0-9]$',
  27         },
  28         'params': {
  29             'skip_download': 'Live stream',
  30         },
  31     }, {
  32         'url': 'http://sportdeutschland.tv/li-ning-badminton-wm-2014/lee-li-ning-badminton-weltmeisterschaft-2014-kopenhagen-herren-einzel-wei-vs',
  33         'info_dict': {
  34             'id': 'lee-li-ning-badminton-weltmeisterschaft-2014-kopenhagen-herren-einzel-wei-vs',
  35             'ext': 'mp4',
  36             'upload_date': '20140825',
  37             'description': 'md5:60a20536b57cee7d9a4ec005e8687504',
  38             'timestamp': 1408976060,
  39             'duration': 2732,
  40             'title': 'Li-Ning Badminton Weltmeisterschaft 2014 Kopenhagen: Herren Einzel, Wei Lee vs. Keun Lee',
  41             'thumbnail': r're:^https?://.*\.jpg$',
  42             'view_count': int,
  43             'categories': ['Li-Ning Badminton WM 2014'],
  44
  45         }
  46     }]
  47
  48     def _real_extract(self, url):
  49         mobj = re.match(self._VALID_URL, url)
  50         video_id = mobj.group('id')
  51         sport_id = mobj.group('sport')
  52
  53         api_url = 'http://proxy.vidibusdynamic.net/sportdeutschland.tv/api/permalinks/%s/%s?access_token=true' % (
  54             sport_id, video_id)
  55         req = sanitized_Request(api_url, headers={
  56             'Accept': 'application/vnd.vidibus.v2.html+json',
  57             'Referer': url,
  58         })
  59         data = self._download_json(req, video_id)
  60
  61         asset = data['asset']
  62         categories = [data['section']['title']]
  63
  64         formats = []
  65         smil_url = asset['video']
  66         if '.smil' in smil_url:
  67             m3u8_url = smil_url.replace('.smil', '.m3u8')
  68             formats.extend(
  69                 self._extract_m3u8_formats(m3u8_url, video_id, ext='mp4'))
  70
  71             smil_doc = self._download_xml(
  72                 smil_url, video_id, note='Downloading SMIL metadata')
  73             base_url_el = smil_doc.find('./head/meta')
  74             if base_url_el:
  75                 base_url = base_url_el.attrib['base']
  76             formats.extend([{
  77                 'format_id': 'rmtp',
  78                 'url': base_url if base_url_el else n.attrib['src'],
  79                 'play_path': n.attrib['src'],
  80                 'ext': 'flv',
  81                 'preference': -100,
  82                 'format_note': 'Seems to fail at example stream',
  83             } for n in smil_doc.findall('./body/video')])
  84         else:
  85             formats.append({'url': smil_url})
  86
  87         self._sort_formats(formats)
  88
  89         return {
  90             'id': video_id,
  91             'formats': formats,
  92             'title': asset['title'],
  93             'thumbnail': asset.get('image'),
  94             'description': asset.get('teaser'),
  95             'duration': asset.get('duration'),
  96             'categories': categories,
  97             'view_count': asset.get('views'),
  98             'rtmp_live': asset.get('live'),
  99             'timestamp': parse_iso8601(asset.get('date')),
 100         }