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