2 from __future__ import unicode_literals
6 from .common import InfoExtractor
7 from ..compat import compat_parse_qs
15 class BiliBiliIE(InfoExtractor):
16 _VALID_URL = r'http://www\.bilibili\.(?:tv|com)/video/av(?P<id>[0-9]+)/'
19 'url': 'http://www.bilibili.tv/video/av1074402/',
20 'md5': '2c301e4dab317596e837c3e7633e7d86',
26 'upload_date': '20140420',
27 'thumbnail': 're:^https?://.+\.jpg',
31 def _real_extract(self, url):
32 video_id = self._match_id(url)
33 webpage = self._download_webpage(url, video_id)
35 video_code = self._search_regex(
36 r'(?s)<div itemprop="video".*?>(.*?)</div>', webpage, 'video code')
38 title = self._html_search_meta(
39 'media:title', video_code, 'title', fatal=True)
40 duration_str = self._html_search_meta(
41 'duration', video_code, 'duration')
42 if duration_str is None:
45 duration_mobj = re.match(
46 r'^T(?:(?P<hours>[0-9]+)H)?(?P<minutes>[0-9]+)M(?P<seconds>[0-9]+)S$',
49 int_or_none(duration_mobj.group('hours'), default=0) * 3600 +
50 int(duration_mobj.group('minutes')) * 60 +
51 int(duration_mobj.group('seconds')))
52 upload_date = unified_strdate(self._html_search_meta(
53 'uploadDate', video_code, fatal=False))
54 thumbnail = self._html_search_meta(
55 'thumbnailUrl', video_code, 'thumbnail', fatal=False)
57 player_params = compat_parse_qs(self._html_search_regex(
58 r'<iframe .*?class="player" src="https://secure\.bilibili\.(?:tv|com)/secure,([^"]+)"',
59 webpage, 'player params'))
61 if 'cid' in player_params:
62 cid = player_params['cid'][0]
64 lq_doc = self._download_xml(
65 'http://interface.bilibili.cn/v_cdn_play?cid=%s' % cid,
67 note='Downloading LQ video info'
69 lq_durl = lq_doc.find('.//durl')
73 'url': lq_durl.find('./url').text,
74 'filesize': int_or_none(
75 lq_durl.find('./size'), get_attr='text'),
78 hq_doc = self._download_xml(
79 'http://interface.bilibili.cn/playurl?cid=%s' % cid,
81 note='Downloading HQ video info',
84 if hq_doc is not False:
85 hq_durl = hq_doc.find('.//durl')
90 'url': hq_durl.find('./url').text,
91 'filesize': int_or_none(
92 hq_durl.find('./size'), get_attr='text'),
95 raise ExtractorError('Unsupported player parameters: %r' % (player_params,))
97 self._sort_formats(formats)
102 'duration': duration,
103 'upload_date': upload_date,
104 'thumbnail': thumbnail,