]> jfr.im git - yt-dlp.git/blame - youtube_dlc/extractor/baidu.py
Merge branch 'myvideo_ge' of https://github.com/fonkap/youtube-dl into fonkap-myvideo_ge
[yt-dlp.git] / youtube_dlc / extractor / baidu.py
CommitLineData
21f63302
YCH
1# coding: utf-8
2from __future__ import unicode_literals
3
4import re
5
6from .common import InfoExtractor
a1d9f6c5 7from ..utils import unescapeHTML
21f63302
YCH
8
9
10class BaiduVideoIE(InfoExtractor):
446e7645 11 IE_DESC = '百度视频'
5886b38d 12 _VALID_URL = r'https?://v\.baidu\.com/(?P<type>[a-z]+)/(?P<id>\d+)\.htm'
21f63302
YCH
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',
957e0db1 17 'title': '中华小当家 TV版国语',
a1d9f6c5 18 'description': 'md5:51be07afe461cf99fa61231421b5397c',
21f63302
YCH
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 },
957e0db1 28 'playlist_mincount': 12,
21f63302
YCH
29 }]
30
03116772
YCH
31 def _call_api(self, path, category, playlist_id, note):
32 return self._download_json('http://app.video.baidu.com/%s/?worktype=adnative%s&id=%s' % (
33 path, category, playlist_id), playlist_id, note)
957e0db1 34
21f63302 35 def _real_extract(self, url):
957e0db1 36 category, playlist_id = re.match(self._VALID_URL, url).groups()
21f63302 37 if category == 'show':
957e0db1 38 category = 'tvshow'
39 if category == 'tv':
40 category = 'tvplay'
21f63302 41
03116772
YCH
42 playlist_detail = self._call_api(
43 'xqinfo', category, playlist_id, 'Download playlist JSON metadata')
21f63302 44
957e0db1 45 playlist_title = playlist_detail['title']
a1d9f6c5 46 playlist_description = unescapeHTML(playlist_detail.get('intro'))
21f63302 47
03116772
YCH
48 episodes_detail = self._call_api(
49 'xqsingle', category, playlist_id, 'Download episodes JSON metadata')
21f63302 50
b7546397
YCH
51 entries = [self.url_result(
52 episode['url'], video_title=episode['title']
53 ) for episode in episodes_detail['videos']]
21f63302
YCH
54
55 return self.playlist_result(
56 entries, playlist_id, playlist_title, playlist_description)