2 from __future__ import unicode_literals
7 import xml.etree.ElementTree as ET
9 from .common import InfoExtractor
17 class BiliBiliIE(InfoExtractor):
18 _VALID_URL = r'http://www\.bilibili\.(?:tv|com)/video/av(?P<id>[0-9]+)/'
21 'url': 'http://www.bilibili.tv/video/av1074402/',
22 'md5': '2c301e4dab317596e837c3e7633e7d86',
24 'id': '1074402_part1',
28 'upload_date': '20140420',
29 'thumbnail': 're:^https?://.+\.jpg',
32 'url': 'http://www.bilibili.com/video/av1041170/',
35 'title': '【BD1080P】刀语【诸神&异域】',
40 def _real_extract(self, url):
41 video_id = self._match_id(url)
42 webpage = self._download_webpage(url, video_id)
44 if self._search_regex(r'(此视频不存在或被删除)', webpage, 'error message', default=None):
45 raise ExtractorError('The video does not exist or was deleted', expected=True)
46 video_code = self._search_regex(
47 r'(?s)<div itemprop="video".*?>(.*?)</div>', webpage, 'video code')
49 title = self._html_search_meta(
50 'media:title', video_code, 'title', fatal=True)
51 duration_str = self._html_search_meta(
52 'duration', video_code, 'duration')
53 if duration_str is None:
56 duration_mobj = re.match(
57 r'^T(?:(?P<hours>[0-9]+)H)?(?P<minutes>[0-9]+)M(?P<seconds>[0-9]+)S$',
60 int_or_none(duration_mobj.group('hours'), default=0) * 3600 +
61 int(duration_mobj.group('minutes')) * 60 +
62 int(duration_mobj.group('seconds')))
63 upload_date = unified_strdate(self._html_search_meta(
64 'uploadDate', video_code, fatal=False))
65 thumbnail = self._html_search_meta(
66 'thumbnailUrl', video_code, 'thumbnail', fatal=False)
68 cid = self._search_regex(r'cid=(\d+)', webpage, 'cid')
72 lq_page = self._download_webpage(
73 'http://interface.bilibili.com/v_cdn_play?appkey=1&cid=%s' % cid,
75 note='Downloading LQ video info'
78 err_info = json.loads(lq_page)
80 'BiliBili said: ' + err_info['error_text'], expected=True)
84 lq_doc = ET.fromstring(lq_page)
85 lq_durls = lq_doc.findall('./durl')
87 hq_doc = self._download_xml(
88 'http://interface.bilibili.com/playurl?appkey=1&cid=%s' % cid,
90 note='Downloading HQ video info',
93 if hq_doc is not False:
94 hq_durls = hq_doc.findall('./durl')
95 assert len(lq_durls) == len(hq_durls)
97 hq_durls = itertools.repeat(None)
100 for lq_durl, hq_durl in zip(lq_durls, hq_durls):
104 'url': lq_durl.find('./url').text,
105 'filesize': int_or_none(
106 lq_durl.find('./size'), get_attr='text'),
113 'url': hq_durl.find('./url').text,
114 'filesize': int_or_none(
115 hq_durl.find('./size'), get_attr='text'),
117 self._sort_formats(formats)
120 'id': '%s_part%d' % (video_id, i),
123 'duration': duration,
124 'upload_date': upload_date,
125 'thumbnail': thumbnail,
131 '_type': 'multi_video',