Fix imports and general cleanup
[youtube-dl] / youtube_dl / extractor / bilibili.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_parse_qs
8 from ..utils import (
9     ExtractorError,
10     int_or_none,
11     unified_strdate,
12 )
13
14
15 class BiliBiliIE(InfoExtractor):
16     _VALID_URL = r'http://www\.bilibili\.(?:tv|com)/video/av(?P<id>[0-9]+)/'
17
18     _TEST = {
19         'url': 'http://www.bilibili.tv/video/av1074402/',
20         'md5': '2c301e4dab317596e837c3e7633e7d86',
21         'info_dict': {
22             'id': '1074402',
23             'ext': 'flv',
24             'title': '【金坷垃】金泡沫',
25             'duration': 308,
26             'upload_date': '20140420',
27             'thumbnail': 're:^https?://.+\.jpg',
28         },
29     }
30
31     def _real_extract(self, url):
32         video_id = self._match_id(url)
33         webpage = self._download_webpage(url, video_id)
34
35         video_code = self._search_regex(
36             r'(?s)<div itemprop="video".*?>(.*?)</div>', webpage, 'video code')
37
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:
43             duration = None
44         else:
45             duration_mobj = re.match(
46                 r'^T(?:(?P<hours>[0-9]+)H)?(?P<minutes>[0-9]+)M(?P<seconds>[0-9]+)S$',
47                 duration_str)
48             duration = (
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)
56
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'))
60
61         if 'cid' in player_params:
62             cid = player_params['cid'][0]
63
64             lq_doc = self._download_xml(
65                 'http://interface.bilibili.cn/v_cdn_play?cid=%s' % cid,
66                 video_id,
67                 note='Downloading LQ video info'
68             )
69             lq_durl = lq_doc.find('.//durl')
70             formats = [{
71                 'format_id': 'lq',
72                 'quality': 1,
73                 'url': lq_durl.find('./url').text,
74                 'filesize': int_or_none(
75                     lq_durl.find('./size'), get_attr='text'),
76             }]
77
78             hq_doc = self._download_xml(
79                 'http://interface.bilibili.cn/playurl?cid=%s' % cid,
80                 video_id,
81                 note='Downloading HQ video info',
82                 fatal=False,
83             )
84             if hq_doc is not False:
85                 hq_durl = hq_doc.find('.//durl')
86                 formats.append({
87                     'format_id': 'hq',
88                     'quality': 2,
89                     'ext': 'flv',
90                     'url': hq_durl.find('./url').text,
91                     'filesize': int_or_none(
92                         hq_durl.find('./size'), get_attr='text'),
93                 })
94         else:
95             raise ExtractorError('Unsupported player parameters: %r' % (player_params,))
96
97         self._sort_formats(formats)
98         return {
99             'id': video_id,
100             'title': title,
101             'formats': formats,
102             'duration': duration,
103             'upload_date': upload_date,
104             'thumbnail': thumbnail,
105         }