_ Git - youtube-dl/blob - youtube_dl/extractor/gameone.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 xpath_with_ns
   8
   9 NAMESPACE_MAP = {
  10     'media': 'http://search.yahoo.com/mrss/',
  11 }
  12
  13 RAW_MP4_URL = 'http://cdn.riptide-mtvn.com/'
  14
  15 class GameOneIE(InfoExtractor):
  16     _VALID_URL = r'https?://(?:www\.)?gameone\.de/tv/(?P<id>\d+)'
  17     _TESTS = {
  18         'url': 'http://www.gameone.de/tv/288',
  19         'md5': '136656b7fb4c9cb4a8e2d500651c499b',
  20         'info_dict': {
  21             'id': '288',
  22             'ext': 'mp4',
  23             'title': 'Game One - Folge 288',
  24             'duration': 1238,
  25             'thumbnail': 'http://s3.gameone.de/gameone/assets/video_metas/teaser_images/000/643/636/big/640x360.jpg',
  26         }
  27     }
  28
  29     def _real_extract(self, url):
  30         mobj = re.match(self._VALID_URL, url)
  31         video_id = mobj.group('id')
  32
  33         webpage = self._download_webpage(url, video_id)
  34         og_video = self._og_search_video_url(webpage, secure=False)
  35         mrss_url = self._search_regex(r'mrss=([^&]+)', og_video, 'mrss')
  36
  37         mrss = self._download_xml(mrss_url, video_id, 'Downloading mrss')
  38         title = mrss.find('.//item/title').text
  39         thumbnail = mrss.find('.//item/image').get('url')
  40         content = mrss.find(xpath_with_ns('.//media:content', NAMESPACE_MAP))
  41         content_url = content.get('url')
  42
  43         content = self._download_xml(content_url, video_id, 'Downloading media:content')
  44         rendition_items = content.findall('.//rendition')
  45         duration = int(rendition_items[0].get('duration'))
  46         formats = [
  47                 {
  48                     'url': re.sub(r'.*/(r2)', RAW_MP4_URL + r'\1', r.find('./src').text),
  49                     'width': int(r.get('width')),
  50                     'height': int(r.get('height')),
  51                     'tbr': int(r.get('bitrate')),
  52                 }
  53             for r in rendition_items
  54         ]
  55
  56         return {
  57             'id': video_id,
  58             'title': title,
  59             'thumbnail': thumbnail,
  60             'duration': duration,
  61             'formats': formats,
  62         }