_ Git - youtube-dl/blob - youtube_dl/extractor/baidu.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 compat_urlparse
   8
   9
  10 class BaiduVideoIE(InfoExtractor):
  11     IE_DESC = '百度视频'
  12     _VALID_URL = r'http://v\.baidu\.com/(?P<type>[a-z]+)/(?P<id>\d+)\.htm'
  13     _TESTS = [{
  14         'url': 'http://v.baidu.com/comic/1069.htm?frp=bdbrand&q=%E4%B8%AD%E5%8D%8E%E5%B0%8F%E5%BD%93%E5%AE%B6',
  15         'info_dict': {
  16             'id': '1069',
  17             'title': '中华小当家 TV版国语',
  18             'description': 'md5:40a9c1b1c7f4e05d642e7bb1c84eeda0',
  19         },
  20         'playlist_count': 52,
  21     }, {
  22         'url': 'http://v.baidu.com/show/11595.htm?frp=bdbrand',
  23         'info_dict': {
  24             'id': '11595',
  25             'title': 're:^奔跑吧兄弟',
  26             'description': 'md5:1bf88bad6d850930f542d51547c089b8',
  27         },
  28         'playlist_mincount': 12,
  29     }]
  30
  31     def _call_api(self, path, category, playlist_id):
  32         return self._download_json('http://app.video.baidu.com/%s/?worktype=adnative%s&id=%s' % (path, category, playlist_id), playlist_id)
  33
  34     def _real_extract(self, url):
  35         category, playlist_id = re.match(self._VALID_URL, url).groups()
  36         if category == 'show':
  37             category = 'tvshow'
  38         if category == 'tv':
  39             category = 'tvplay'
  40
  41         playlist_detail = self._call_api('xqinfo', category, playlist_id)
  42
  43         playlist_title = playlist_detail['title']
  44         playlist_description = playlist_detail.get('intro')
  45
  46         episodes_detail = self._call_api('xqsingle', category, playlist_id)
  47
  48         entries = []
  49         for episode in episodes_detail['videos']:
  50             episode_id = '%s_%s' % (playlist_id, episode['episode'])
  51
  52             entries.append(self.url_result(
  53                 episode['url'], video_title=episode['title']))
  54
  55         return self.playlist_result(
  56             entries, playlist_id, playlist_title, playlist_description)