]>
Commit | Line | Data |
---|---|---|
28746fbd PH |
1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
04b32c8f | 4 | import hashlib |
520e7533 | 5 | import re |
28746fbd PH |
6 | |
7 | from .common import InfoExtractor | |
04b32c8f | 8 | from ..compat import compat_parse_qs |
28746fbd | 9 | from ..utils import ( |
6461f2b7 YCH |
10 | int_or_none, |
11 | float_or_none, | |
04b32c8f | 12 | unified_timestamp, |
1f85029d | 13 | urlencode_postdata, |
28746fbd PH |
14 | ) |
15 | ||
16 | ||
17 | class BiliBiliIE(InfoExtractor): | |
1f85029d | 18 | _VALID_URL = r'https?://(?:www\.|bangumi\.|)bilibili\.(?:tv|com)/(?:video/av|anime/v/)(?P<id>\d+)' |
28746fbd | 19 | |
4875ff68 | 20 | _TEST = { |
28746fbd | 21 | 'url': 'http://www.bilibili.tv/video/av1074402/', |
412abb87 | 22 | 'md5': '9fa226fe2b8a9a4d5a69b4c6a183417e', |
28746fbd | 23 | 'info_dict': { |
04b32c8f | 24 | 'id': '1074402', |
412abb87 | 25 | 'ext': 'mp4', |
28746fbd | 26 | 'title': '【金坷垃】金泡沫', |
6461f2b7 | 27 | 'description': 'md5:ce18c2a2d2193f0df2917d270f2e5923', |
412abb87 | 28 | 'duration': 308.315, |
6461f2b7 | 29 | 'timestamp': 1398012660, |
28746fbd | 30 | 'upload_date': '20140420', |
ec85ded8 | 31 | 'thumbnail': r're:^https?://.+\.jpg', |
d90e4030 | 32 | 'uploader': '菊子桑', |
6461f2b7 | 33 | 'uploader_id': '156160', |
28746fbd | 34 | }, |
4875ff68 | 35 | } |
28746fbd | 36 | |
baa3e184 RA |
37 | _APP_KEY = '84956560bc028eb7' |
38 | _BILIBILI_KEY = '94aba54af9065f71de72f5508f1cd42e' | |
6461f2b7 | 39 | |
520e7533 | 40 | def _real_extract(self, url): |
04b32c8f | 41 | video_id = self._match_id(url) |
6461f2b7 YCH |
42 | webpage = self._download_webpage(url, video_id) |
43 | ||
1f85029d | 44 | if 'anime/v' not in url: |
7be15d40 P |
45 | cid = compat_parse_qs(self._search_regex( |
46 | [r'EmbedPlayer\([^)]+,\s*"([^"]+)"\)', | |
1f85029d | 47 | r'<iframe[^>]+src="https://secure\.bilibili\.com/secure,([^"]+)"'], |
7be15d40 P |
48 | webpage, 'player parameters'))['cid'][0] |
49 | else: | |
1f85029d YCH |
50 | js = self._download_json( |
51 | 'http://bangumi.bilibili.com/web_api/get_source', video_id, | |
52 | data=urlencode_postdata({'episode_id': video_id}), | |
53 | headers={'Content-Type': 'application/x-www-form-urlencoded; charset=UTF-8'}) | |
7be15d40 | 54 | cid = js['result']['cid'] |
04b32c8f YCH |
55 | |
56 | payload = 'appkey=%s&cid=%s&otype=json&quality=2&type=mp4' % (self._APP_KEY, cid) | |
57 | sign = hashlib.md5((payload + self._BILIBILI_KEY).encode('utf-8')).hexdigest() | |
6d00a2dc | 58 | |
04b32c8f YCH |
59 | video_info = self._download_json( |
60 | 'http://interface.bilibili.com/playurl?%s&sign=%s' % (payload, sign), | |
61 | video_id, note='Downloading video info page') | |
28746fbd | 62 | |
d90e4030 | 63 | entries = [] |
c4a21bc9 | 64 | |
04b32c8f | 65 | for idx, durl in enumerate(video_info['durl']): |
497f5fd9 | 66 | formats = [{ |
04b32c8f YCH |
67 | 'url': durl['url'], |
68 | 'filesize': int_or_none(durl['size']), | |
497f5fd9 | 69 | }] |
86d68f90 | 70 | for backup_url in durl.get('backup_url', []): |
6461f2b7 | 71 | formats.append({ |
04b32c8f | 72 | 'url': backup_url, |
6461f2b7 | 73 | # backup URLs have lower priorities |
04b32c8f | 74 | 'preference': -2 if 'hd.mp4' in backup_url else -3, |
6461f2b7 YCH |
75 | }) |
76 | ||
77 | self._sort_formats(formats) | |
497f5fd9 | 78 | |
c4a21bc9 | 79 | entries.append({ |
04b32c8f YCH |
80 | 'id': '%s_part%s' % (video_id, idx), |
81 | 'duration': float_or_none(durl.get('length'), 1000), | |
55af2b26 | 82 | 'formats': formats, |
58a84b8c | 83 | }) |
28746fbd | 84 | |
6461f2b7 YCH |
85 | title = self._html_search_regex('<h1[^>]+title="([^"]+)">', webpage, 'title') |
86 | description = self._html_search_meta('description', webpage) | |
04b32c8f YCH |
87 | timestamp = unified_timestamp(self._html_search_regex( |
88 | r'<time[^>]+datetime="([^"]+)"', webpage, 'upload time', fatal=False)) | |
1f85029d | 89 | thumbnail = self._html_search_meta(['og:image', 'thumbnailUrl'], webpage) |
6461f2b7 YCH |
90 | |
91 | # TODO 'view_count' requires deobfuscating Javascript | |
d90e4030 | 92 | info = { |
04b32c8f | 93 | 'id': video_id, |
d90e4030 | 94 | 'title': title, |
6461f2b7 YCH |
95 | 'description': description, |
96 | 'timestamp': timestamp, | |
7be15d40 | 97 | 'thumbnail': thumbnail, |
04b32c8f | 98 | 'duration': float_or_none(video_info.get('timelength'), scale=1000), |
28746fbd | 99 | } |
d90e4030 | 100 | |
6461f2b7 YCH |
101 | uploader_mobj = re.search( |
102 | r'<a[^>]+href="https?://space\.bilibili\.com/(?P<id>\d+)"[^>]+title="(?P<name>[^"]+)"', | |
103 | webpage) | |
104 | if uploader_mobj: | |
105 | info.update({ | |
106 | 'uploader': uploader_mobj.group('name'), | |
107 | 'uploader_id': uploader_mobj.group('id'), | |
108 | }) | |
109 | ||
110 | for entry in entries: | |
111 | entry.update(info) | |
112 | ||
d90e4030 | 113 | if len(entries) == 1: |
d90e4030 | 114 | return entries[0] |
115 | else: | |
ad73083f YCH |
116 | for idx, entry in enumerate(entries): |
117 | entry['id'] = '%s_part%d' % (video_id, (idx + 1)) | |
118 | ||
6461f2b7 | 119 | return { |
d90e4030 | 120 | '_type': 'multi_video', |
520e7533 | 121 | 'id': video_id, |
6461f2b7 YCH |
122 | 'title': title, |
123 | 'description': description, | |
d90e4030 | 124 | 'entries': entries, |
6461f2b7 | 125 | } |