]>
Commit | Line | Data |
---|---|---|
9ac0a675 PH |
1 | from __future__ import unicode_literals |
2 | ||
7b2212e9 | 3 | import re |
4 | ||
5 | from .common import InfoExtractor | |
1cc79574 | 6 | from ..compat import ( |
47af21e8 | 7 | compat_urllib_parse_unquote, |
7b2212e9 | 8 | compat_urllib_parse_urlparse, |
1cc79574 PH |
9 | ) |
10 | from ..utils import ( | |
5c2266df | 11 | sanitized_Request, |
9767726b | 12 | str_to_int, |
1cc79574 | 13 | unified_strdate, |
7b2212e9 | 14 | ) |
9767726b | 15 | from ..aes import aes_decrypt_text |
7b2212e9 | 16 | |
9ac0a675 | 17 | |
7b2212e9 | 18 | class SpankwireIE(InfoExtractor): |
28b83495 | 19 | _VALID_URL = r'https?://(?:www\.)?(?P<url>spankwire\.com/[^/]*/video(?P<id>[0-9]+)/?)' |
59e6acc7 | 20 | _TESTS = [{ |
551c7837 S |
21 | # download URL pattern: */<height>P_<tbr>K_<video_id>.mp4 |
22 | 'url': 'http://www.spankwire.com/Buckcherry-s-X-Rated-Music-Video-Crazy-Bitch/video103545/', | |
23 | 'md5': '8bbfde12b101204b39e4b9fe7eb67095', | |
24 | 'info_dict': { | |
25 | 'id': '103545', | |
26 | 'ext': 'mp4', | |
27 | 'title': 'Buckcherry`s X Rated Music Video Crazy Bitch', | |
28 | 'description': 'Crazy Bitch X rated music video.', | |
29 | 'uploader': 'oreusz', | |
30 | 'uploader_id': '124697', | |
31 | 'upload_date': '20070507', | |
32 | 'age_limit': 18, | |
33 | } | |
34 | }, { | |
35 | # download URL pattern: */mp4_<format_id>_<video_id>.mp4 | |
36 | 'url': 'http://www.spankwire.com/Titcums-Compiloation-I/video1921551/', | |
37 | 'md5': '09b3c20833308b736ae8902db2f8d7e6', | |
38 | 'info_dict': { | |
39 | 'id': '1921551', | |
40 | 'ext': 'mp4', | |
41 | 'title': 'Titcums Compiloation I', | |
42 | 'description': 'cum on tits', | |
43 | 'uploader': 'dannyh78999', | |
44 | 'uploader_id': '3056053', | |
45 | 'upload_date': '20150822', | |
46 | 'age_limit': 18, | |
47 | }, | |
48 | }] | |
7b2212e9 | 49 | |
50 | def _real_extract(self, url): | |
51 | mobj = re.match(self._VALID_URL, url) | |
28b83495 | 52 | video_id = mobj.group('id') |
7b2212e9 | 53 | |
5c2266df | 54 | req = sanitized_Request('http://www.' + mobj.group('url')) |
7b2212e9 | 55 | req.add_header('Cookie', 'age_verified=1') |
56 | webpage = self._download_webpage(req, video_id) | |
57 | ||
fdb4d278 S |
58 | title = self._html_search_regex( |
59 | r'<h1>([^<]+)', webpage, 'title') | |
0ad97bbc | 60 | description = self._html_search_regex( |
44705366 | 61 | r'(?s)<div\s+id="descriptionContent">(.+?)</div>', |
fdb4d278 | 62 | webpage, 'description', fatal=False) |
9767726b | 63 | thumbnail = self._html_search_regex( |
d3c9af84 | 64 | r'playerData\.screenShot\s*=\s*["\']([^"\']+)["\']', |
fdb4d278 | 65 | webpage, 'thumbnail', fatal=False) |
9767726b S |
66 | |
67 | uploader = self._html_search_regex( | |
fdb4d278 S |
68 | r'by:\s*<a [^>]*>(.+?)</a>', |
69 | webpage, 'uploader', fatal=False) | |
9767726b | 70 | uploader_id = self._html_search_regex( |
90076b61 | 71 | r'by:\s*<a href="/(?:user/viewProfile|Profile\.aspx)\?.*?UserId=(\d+).*?"', |
fdb4d278 S |
72 | webpage, 'uploader id', fatal=False) |
73 | upload_date = unified_strdate(self._html_search_regex( | |
74 | r'</a> on (.+?) at \d+:\d+', | |
75 | webpage, 'upload date', fatal=False)) | |
7b2212e9 | 76 | |
fdb4d278 S |
77 | view_count = str_to_int(self._html_search_regex( |
78 | r'<div id="viewsCounter"><span>([\d,\.]+)</span> views</div>', | |
79 | webpage, 'view count', fatal=False)) | |
80 | comment_count = str_to_int(self._html_search_regex( | |
44705366 | 81 | r'<span\s+id="spCommentCount"[^>]*>([\d,\.]+)</span>', |
fdb4d278 S |
82 | webpage, 'comment count', fatal=False)) |
83 | ||
551c7837 S |
84 | videos = re.findall( |
85 | r'playerData\.cdnPath([0-9]{3,})\s*=\s*(?:encodeURIComponent\()?["\']([^"\']+)["\']', webpage) | |
86 | heights = [int(video[0]) for video in videos] | |
87 | video_urls = list(map(compat_urllib_parse_unquote, [video[1] for video in videos])) | |
7b2212e9 | 88 | if webpage.find('flashvars\.encrypted = "true"') != -1: |
1434184c | 89 | password = self._search_regex( |
fdb4d278 S |
90 | r'flashvars\.video_title = "([^"]+)', |
91 | webpage, 'password').replace('+', ' ') | |
92 | video_urls = list(map( | |
93 | lambda s: aes_decrypt_text(s, password, 32).decode('utf-8'), | |
94 | video_urls)) | |
7b2212e9 | 95 | |
96 | formats = [] | |
551c7837 | 97 | for height, video_url in zip(heights, video_urls): |
a56f9de1 | 98 | path = compat_urllib_parse_urlparse(video_url).path |
551c7837 S |
99 | _, quality = path.split('/')[4].split('_')[:2] |
100 | f = { | |
101 | 'url': video_url, | |
102 | 'height': height, | |
103 | } | |
104 | tbr = self._search_regex(r'^(\d+)[Kk]$', quality, 'tbr', default=None) | |
105 | if tbr: | |
106 | f.update({ | |
107 | 'tbr': int(tbr), | |
108 | 'format_id': '%dp' % height, | |
59e6acc7 | 109 | }) |
110 | else: | |
551c7837 S |
111 | f['format_id'] = quality |
112 | formats.append(f) | |
9ac0a675 | 113 | self._sort_formats(formats) |
7b2212e9 | 114 | |
750e9833 FV |
115 | age_limit = self._rta_search(webpage) |
116 | ||
7b2212e9 | 117 | return { |
118 | 'id': video_id, | |
9767726b | 119 | 'title': title, |
7b2212e9 | 120 | 'description': description, |
9767726b S |
121 | 'thumbnail': thumbnail, |
122 | 'uploader': uploader, | |
123 | 'uploader_id': uploader_id, | |
124 | 'upload_date': upload_date, | |
125 | 'view_count': view_count, | |
126 | 'comment_count': comment_count, | |
7b2212e9 | 127 | 'formats': formats, |
750e9833 | 128 | 'age_limit': age_limit, |
7b2212e9 | 129 | } |