]>
Commit | Line | Data |
---|---|---|
2aebbcce | 1 | from __future__ import unicode_literals |
2 | ||
2aebbcce | 3 | from .common import InfoExtractor |
e0977d76 S |
4 | from ..compat import ( |
5 | compat_chr, | |
6 | compat_ord, | |
7 | compat_urllib_parse_unquote, | |
8 | ) | |
5946cda7 S |
9 | from ..utils import ( |
10 | int_or_none, | |
11 | parse_iso8601, | |
f6ff52b4 | 12 | urljoin, |
5946cda7 | 13 | ) |
2aebbcce | 14 | |
15 | ||
16 | class BeegIE(InfoExtractor): | |
17 | _VALID_URL = r'https?://(?:www\.)?beeg\.com/(?P<id>\d+)' | |
18 | _TEST = { | |
19 | 'url': 'http://beeg.com/5416503', | |
8814ae42 | 20 | 'md5': 'a1a1b1a8bc70a89e49ccfd113aed0820', |
2aebbcce | 21 | 'info_dict': { |
22 | 'id': '5416503', | |
23 | 'ext': 'mp4', | |
24 | 'title': 'Sultry Striptease', | |
5946cda7 S |
25 | 'description': 'md5:d22219c09da287c14bed3d6c37ce4bc2', |
26 | 'timestamp': 1391813355, | |
27 | 'upload_date': '20140207', | |
28 | 'duration': 383, | |
29 | 'tags': list, | |
7ca2e11f | 30 | 'age_limit': 18, |
2aebbcce | 31 | } |
32 | } | |
33 | ||
34 | def _real_extract(self, url): | |
5946cda7 | 35 | video_id = self._match_id(url) |
3baa62e8 | 36 | |
3afef2e3 S |
37 | webpage = self._download_webpage(url, video_id) |
38 | ||
39 | cpl_url = self._search_regex( | |
f6ff52b4 | 40 | r'<script[^>]+src=(["\'])(?P<url>(?:/static|(?:https?:)?//static\.beeg\.com)/cpl/\d+\.js.*?)\1', |
3afef2e3 S |
41 | webpage, 'cpl', default=None, group='url') |
42 | ||
f6ff52b4 GF |
43 | cpl_url = urljoin(url, cpl_url) |
44 | ||
3afef2e3 S |
45 | beeg_version, beeg_salt = [None] * 2 |
46 | ||
47 | if cpl_url: | |
48 | cpl = self._download_webpage( | |
49 | self._proto_relative_url(cpl_url), video_id, | |
50 | 'Downloading cpl JS', fatal=False) | |
51 | if cpl: | |
2a048f98 S |
52 | beeg_version = int_or_none(self._search_regex( |
53 | r'beeg_version\s*=\s*([^\b]+)', cpl, | |
54 | 'beeg version', default=None)) or self._search_regex( | |
3afef2e3 S |
55 | r'/(\d+)\.js', cpl_url, 'beeg version', default=None) |
56 | beeg_salt = self._search_regex( | |
2a048f98 | 57 | r'beeg_salt\s*=\s*(["\'])(?P<beeg_salt>.+?)\1', cpl, 'beeg salt', |
3afef2e3 S |
58 | default=None, group='beeg_salt') |
59 | ||
f6ff52b4 | 60 | beeg_version = beeg_version or '2185' |
2a048f98 | 61 | beeg_salt = beeg_salt or 'pmweAkq8lAYKdfWcFCUj0yoVgoPlinamH5UE1CB3H' |
3afef2e3 | 62 | |
6e736d86 S |
63 | for api_path in ('', 'api.'): |
64 | video = self._download_json( | |
65 | 'https://%sbeeg.com/api/v6/%s/video/%s' | |
66 | % (api_path, beeg_version, video_id), video_id, | |
67 | fatal=api_path == 'api.') | |
68 | if video: | |
69 | break | |
d63cfc3f S |
70 | |
71 | def split(o, e): | |
72 | def cut(s, x): | |
73 | n.append(s[:x]) | |
74 | return s[x:] | |
75 | n = [] | |
76 | r = len(o) % e | |
77 | if r > 0: | |
78 | o = cut(o, r) | |
79 | while len(o) > e: | |
80 | o = cut(o, e) | |
81 | n.append(o) | |
82 | return n | |
3baa62e8 | 83 | |
e0977d76 | 84 | def decrypt_key(key): |
5299bc3f | 85 | # Reverse engineered from http://static.beeg.com/cpl/1738.js |
3afef2e3 | 86 | a = beeg_salt |
e0977d76 | 87 | e = compat_urllib_parse_unquote(key) |
d63cfc3f S |
88 | o = ''.join([ |
89 | compat_chr(compat_ord(e[n]) - compat_ord(a[n % len(a)]) % 21) | |
e0977d76 | 90 | for n in range(len(e))]) |
d63cfc3f | 91 | return ''.join(split(o, 3)[::-1]) |
e0977d76 S |
92 | |
93 | def decrypt_url(encrypted_url): | |
94 | encrypted_url = self._proto_relative_url( | |
4511c197 | 95 | encrypted_url.replace('{DATA_MARKERS}', ''), 'https:') |
e0977d76 S |
96 | key = self._search_regex( |
97 | r'/key=(.*?)%2Cend=', encrypted_url, 'key', default=None) | |
98 | if not key: | |
99 | return encrypted_url | |
100 | return encrypted_url.replace(key, decrypt_key(key)) | |
101 | ||
5946cda7 S |
102 | formats = [] |
103 | for format_id, video_url in video.items(): | |
5d0f84d3 S |
104 | if not video_url: |
105 | continue | |
5946cda7 S |
106 | height = self._search_regex( |
107 | r'^(\d+)[pP]$', format_id, 'height', default=None) | |
108 | if not height: | |
109 | continue | |
110 | formats.append({ | |
e0977d76 | 111 | 'url': decrypt_url(video_url), |
5946cda7 S |
112 | 'format_id': format_id, |
113 | 'height': int(height), | |
114 | }) | |
3baa62e8 | 115 | self._sort_formats(formats) |
2aebbcce | 116 | |
5946cda7 S |
117 | title = video['title'] |
118 | video_id = video.get('id') or video_id | |
119 | display_id = video.get('code') | |
120 | description = video.get('desc') | |
5f6a1245 | 121 | |
5946cda7 S |
122 | timestamp = parse_iso8601(video.get('date'), ' ') |
123 | duration = int_or_none(video.get('duration')) | |
2aebbcce | 124 | |
5946cda7 | 125 | tags = [tag.strip() for tag in video['tags'].split(',')] if video.get('tags') else None |
2aebbcce | 126 | |
127 | return { | |
128 | 'id': video_id, | |
5946cda7 | 129 | 'display_id': display_id, |
2aebbcce | 130 | 'title': title, |
131 | 'description': description, | |
5946cda7 S |
132 | 'timestamp': timestamp, |
133 | 'duration': duration, | |
134 | 'tags': tags, | |
3baa62e8 | 135 | 'formats': formats, |
3afef2e3 | 136 | 'age_limit': self._rta_search(webpage), |
2aebbcce | 137 | } |