]>
Commit | Line | Data |
---|---|---|
008af866 | 1 | from .common import InfoExtractor |
432cd484 S |
2 | from ..utils import ( |
3 | int_or_none, | |
4 | url_or_none, | |
5 | ) | |
f54aee02 PH |
6 | |
7 | ||
008af866 | 8 | class CliphunterIE(InfoExtractor): |
efc86777 | 9 | IE_NAME = 'cliphunter' |
008af866 | 10 | |
5886b38d | 11 | _VALID_URL = r'''(?x)https?://(?:www\.)?cliphunter\.com/w/ |
b6d3a996 PH |
12 | (?P<id>[0-9]+)/ |
13 | (?P<seo>.+?)(?:$|[#\?]) | |
14 | ''' | |
aadd3ce2 | 15 | _TESTS = [{ |
efc86777 | 16 | 'url': 'http://www.cliphunter.com/w/1012420/Fun_Jynx_Maze_solo', |
3e055aa5 | 17 | 'md5': 'b7c9bbd4eb3a226ab91093714dcaa480', |
efc86777 | 18 | 'info_dict': { |
f54aee02 | 19 | 'id': '1012420', |
3e055aa5 | 20 | 'ext': 'flv', |
efc86777 | 21 | 'title': 'Fun Jynx Maze solo', |
ec85ded8 | 22 | 'thumbnail': r're:^https?://.*\.jpg$', |
f54aee02 | 23 | 'age_limit': 18, |
aadd3ce2 YCH |
24 | }, |
25 | 'skip': 'Video gone', | |
26 | }, { | |
27 | 'url': 'http://www.cliphunter.com/w/2019449/ShesNew__My_booty_girlfriend_Victoria_Paradices_pussy_filled_with_jizz', | |
28 | 'md5': '55a723c67bfc6da6b0cfa00d55da8a27', | |
29 | 'info_dict': { | |
30 | 'id': '2019449', | |
31 | 'ext': 'mp4', | |
32 | 'title': 'ShesNew - My booty girlfriend, Victoria Paradice\'s pussy filled with jizz', | |
ec85ded8 | 33 | 'thumbnail': r're:^https?://.*\.jpg$', |
aadd3ce2 YCH |
34 | 'age_limit': 18, |
35 | }, | |
36 | }] | |
008af866 P |
37 | |
38 | def _real_extract(self, url): | |
c9f08154 | 39 | video_id = self._match_id(url) |
008af866 P |
40 | webpage = self._download_webpage(url, video_id) |
41 | ||
f54aee02 PH |
42 | video_title = self._search_regex( |
43 | r'mediaTitle = "([^"]+)"', webpage, 'title') | |
44 | ||
39fa4cc1 S |
45 | gexo_files = self._parse_json( |
46 | self._search_regex( | |
47 | r'var\s+gexoFiles\s*=\s*({.+?});', webpage, 'gexo files'), | |
48 | video_id) | |
3e055aa5 NJ |
49 | |
50 | formats = [] | |
39fa4cc1 | 51 | for format_id, f in gexo_files.items(): |
432cd484 | 52 | video_url = url_or_none(f.get('url')) |
39fa4cc1 S |
53 | if not video_url: |
54 | continue | |
55 | fmt = f.get('fmt') | |
56 | height = f.get('h') | |
57 | format_id = '%s_%sp' % (fmt, height) if fmt and height else format_id | |
58 | formats.append({ | |
432cd484 | 59 | 'url': video_url, |
39fa4cc1 S |
60 | 'format_id': format_id, |
61 | 'width': int_or_none(f.get('w')), | |
62 | 'height': int_or_none(height), | |
63 | 'tbr': int_or_none(f.get('br')), | |
64 | }) | |
f54aee02 PH |
65 | self._sort_formats(formats) |
66 | ||
67 | thumbnail = self._search_regex( | |
68 | r"var\s+mov_thumb\s*=\s*'([^']+)';", | |
69 | webpage, 'thumbnail', fatal=False) | |
f54aee02 | 70 | |
008af866 P |
71 | return { |
72 | 'id': video_id, | |
73 | 'title': video_title, | |
74 | 'formats': formats, | |
f54aee02 PH |
75 | 'age_limit': self._rta_search(webpage), |
76 | 'thumbnail': thumbnail, | |
008af866 | 77 | } |