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