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 ..compat import (
   8     compat_urllib_request,
   9 )
  10 from ..utils import (
  11     parse_iso8601,
  12 )
  13
  14
  15 class SportDeutschlandIE(InfoExtractor):
  16     _VALID_URL = r'https?://sportdeutschland\.tv/(?P<sport>[^/?#]+)/(?P<id>[^?#/]+)(?:$|[?#])'
  17     _TESTS = [{
  18         'url': 'http://sportdeutschland.tv/badminton/live-li-ning-badminton-weltmeisterschaft-2014-kopenhagen',
  19         'info_dict': {
  20             'id': 'live-li-ning-badminton-weltmeisterschaft-2014-kopenhagen',
  21             'ext': 'mp4',
  22             'title': 're:Li-Ning Badminton Weltmeisterschaft 2014 Kopenhagen',
  23             'categories': ['Badminton'],
  24             'view_count': int,
  25             'thumbnail': 're:^https?://.*\.jpg$',
  26             'description': 're:Die Badminton-WM 2014 aus Kopenhagen bei Sportdeutschland\.TV',
  27             'timestamp': int,
  28             'upload_date': 're:^201408[23][0-9]$',
  29         },
  30         'params': {
  31             'skip_download': 'Live stream',
  32         },
  33     }, {
  34         'url': 'http://sportdeutschland.tv/li-ning-badminton-wm-2014/lee-li-ning-badminton-weltmeisterschaft-2014-kopenhagen-herren-einzel-wei-vs',
  35         'info_dict': {
  36             'id': 'lee-li-ning-badminton-weltmeisterschaft-2014-kopenhagen-herren-einzel-wei-vs',
  37             'ext': 'mp4',
  38             'upload_date': '20140825',
  39             'description': 'md5:60a20536b57cee7d9a4ec005e8687504',
  40             'timestamp': 1408976060,
  41             'title': 'Li-Ning Badminton Weltmeisterschaft 2014 Kopenhagen: Herren Einzel, Wei Lee vs. Keun Lee',
  42             'thumbnail': 're:^https?://.*\.jpg$',
  43             'view_count': int,
  44             'categories': ['Li-Ning Badminton WM 2014'],
  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://splink.tv/api/permalinks/%s/%s' % (
  54             sport_id, video_id)
  55         req = compat_urllib_request.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         categories = list(data.get('section', {}).get('tags', {}).values())
  62         asset = data['asset']
  63         assets_info = self._download_json(asset['url'], video_id)
  64
  65         formats = []
  66         smil_url = assets_info['video']
  67         if '.smil' in smil_url:
  68             m3u8_url = smil_url.replace('.smil', '.m3u8')
  69             formats.extend(
  70                 self._extract_m3u8_formats(m3u8_url, video_id, ext='mp4'))
  71
  72             smil_doc = self._download_xml(
  73                 smil_url, video_id, note='Downloading SMIL metadata')
  74             base_url = smil_doc.find('./head/meta').attrib['base']
  75             formats.extend([{
  76                 'format_id': 'rmtp',
  77                 'url': base_url,
  78                 'play_path': n.attrib['src'],
  79                 'ext': 'flv',
  80                 'preference': -100,
  81                 'format_note': 'Seems to fail at example stream',
  82             } for n in smil_doc.findall('./body/video')])
  83         else:
  84             formats.append({'url': smil_url})
  85
  86         self._sort_formats(formats)
  87
  88         return {
  89             'id': video_id,
  90             'formats': formats,
  91             'title': asset['title'],
  92             'thumbnail': asset.get('image'),
  93             'description': asset.get('teaser'),
  94             'categories': categories,
  95             'view_count': asset.get('views'),
  96             'rtmp_live': asset.get('live'),
  97             'timestamp': parse_iso8601(asset.get('date')),
  98         }