]>
Commit | Line | Data |
---|---|---|
6599c725 S |
1 | import re |
2 | ||
1cc79574 | 3 | from ..utils import ( |
2fb3deec S |
4 | int_or_none, |
5 | str_to_int, | |
1d45a23b | 6 | ) |
8804f10e | 7 | from .keezmovies import KeezMoviesIE |
2fb3deec | 8 | |
1d45a23b | 9 | |
8804f10e | 10 | class Tube8IE(KeezMoviesIE): |
58e7071a | 11 | _VALID_URL = r'https?://(?:www\.)?tube8\.com/(?:[^/]+/)+(?P<display_id>[^/]+)/(?P<id>\d+)' |
7dde358a S |
12 | _TESTS = [{ |
13 | 'url': 'http://www.tube8.com/teen/kasia-music-video/229795/', | |
14 | 'md5': '65e20c48e6abff62ed0c3965fff13a39', | |
15 | 'info_dict': { | |
16 | 'id': '229795', | |
17 | 'display_id': 'kasia-music-video', | |
18 | 'ext': 'mp4', | |
19 | 'description': 'hot teen Kasia grinding', | |
20 | 'uploader': 'unknown', | |
21 | 'title': 'Kasia music video', | |
22 | 'age_limit': 18, | |
23 | 'duration': 230, | |
6599c725 S |
24 | 'categories': ['Teen'], |
25 | 'tags': ['dancing'], | |
26 | }, | |
a616f654 | 27 | }, { |
7dde358a S |
28 | 'url': 'http://www.tube8.com/shemale/teen/blonde-cd-gets-kidnapped-by-two-blacks-and-punished-for-being-a-slutty-girl/19569151/', |
29 | 'only_matching': True, | |
30 | }] | |
1d45a23b | 31 | |
06993715 S |
32 | @staticmethod |
33 | def _extract_urls(webpage): | |
34 | return re.findall( | |
35 | r'<iframe[^>]+\bsrc=["\']((?:https?:)?//(?:www\.)?tube8\.com/embed/(?:[^/]+/)+\d+)', | |
36 | webpage) | |
37 | ||
1d45a23b | 38 | def _real_extract(self, url): |
8804f10e | 39 | webpage, info = self._extract_info(url) |
2fb3deec | 40 | |
8804f10e S |
41 | if not info['title']: |
42 | info['title'] = self._html_search_regex( | |
43 | r'videoTitle\s*=\s*"([^"]+)', webpage, 'title') | |
2fb3deec | 44 | |
2fb3deec | 45 | description = self._html_search_regex( |
79facb27 | 46 | r'(?s)Description:</dt>\s*<dd>(.+?)</dd>', webpage, 'description', fatal=False) |
2fb3deec | 47 | uploader = self._html_search_regex( |
6ebdfe43 | 48 | r'<span class="username">\s*(.+?)\s*<', |
2fb3deec S |
49 | webpage, 'uploader', fatal=False) |
50 | ||
7dde358a | 51 | like_count = int_or_none(self._search_regex( |
6ebdfe43 | 52 | r'rupVar\s*=\s*"(\d+)"', webpage, 'like count', fatal=False)) |
7dde358a | 53 | dislike_count = int_or_none(self._search_regex( |
6ebdfe43 | 54 | r'rdownVar\s*=\s*"(\d+)"', webpage, 'dislike count', fatal=False)) |
7dde358a | 55 | view_count = str_to_int(self._search_regex( |
79facb27 | 56 | r'Views:\s*</dt>\s*<dd>([\d,\.]+)', |
7dde358a S |
57 | webpage, 'view count', fatal=False)) |
58 | comment_count = str_to_int(self._search_regex( | |
59 | r'<span id="allCommentsCount">(\d+)</span>', | |
60 | webpage, 'comment count', fatal=False)) | |
1d45a23b | 61 | |
6599c725 | 62 | category = self._search_regex( |
79facb27 | 63 | r'Category:\s*</dt>\s*<dd>\s*<a[^>]+href=[^>]+>([^<]+)', |
6599c725 S |
64 | webpage, 'category', fatal=False) |
65 | categories = [category] if category else None | |
66 | ||
67 | tags_str = self._search_regex( | |
79facb27 | 68 | r'(?s)Tags:\s*</dt>\s*<dd>(.+?)</(?!a)', |
6599c725 S |
69 | webpage, 'tags', fatal=False) |
70 | tags = [t for t in re.findall( | |
71 | r'<a[^>]+href=[^>]+>([^<]+)', tags_str)] if tags_str else None | |
72 | ||
8804f10e | 73 | info.update({ |
2fb3deec | 74 | 'description': description, |
2fb3deec | 75 | 'uploader': uploader, |
2fb3deec S |
76 | 'view_count': view_count, |
77 | 'like_count': like_count, | |
78 | 'dislike_count': dislike_count, | |
79 | 'comment_count': comment_count, | |
6599c725 S |
80 | 'categories': categories, |
81 | 'tags': tags, | |
8804f10e S |
82 | }) |
83 | ||
84 | return info |