]>
Commit | Line | Data |
---|---|---|
d740f7e1 | 1 | from __future__ import unicode_literals |
2 | ||
3 | import re | |
4 | ||
5 | from .common import InfoExtractor | |
0aef0771 | 6 | from ..utils import ( |
b9bc1cff | 7 | int_or_none, |
0aef0771 | 8 | NO_DEFAULT, |
b9bc1cff | 9 | parse_duration, |
0aef0771 S |
10 | str_to_int, |
11 | ) | |
d740f7e1 | 12 | |
13 | ||
14 | class DrTuberIE(InfoExtractor): | |
38db52ad | 15 | _VALID_URL = r'https?://(?:(?:www|m)\.)?drtuber\.com/(?:video|embed)/(?P<id>\d+)(?:/(?P<display_id>[\w-]+))?' |
519d8970 | 16 | _TESTS = [{ |
d740f7e1 | 17 | 'url': 'http://www.drtuber.com/video/1740434/hot-perky-blonde-naked-golf', |
18 | 'md5': '93e680cf2536ad0dfb7e74d94a89facd', | |
19 | 'info_dict': { | |
20 | 'id': '1740434', | |
33422c05 | 21 | 'display_id': 'hot-perky-blonde-naked-golf', |
d740f7e1 | 22 | 'ext': 'mp4', |
76d1466b | 23 | 'title': 'hot perky blonde naked golf', |
eade1d7e | 24 | 'like_count': int, |
eade1d7e | 25 | 'comment_count': int, |
2914e5f0 | 26 | 'categories': ['Babe', 'Blonde', 'Erotic', 'Outdoor', 'Softcore', 'Solo'], |
ec85ded8 | 27 | 'thumbnail': r're:https?://.*\.jpg$', |
11fc065c | 28 | 'age_limit': 18, |
d740f7e1 | 29 | } |
519d8970 S |
30 | }, { |
31 | 'url': 'http://www.drtuber.com/embed/489939', | |
32 | 'only_matching': True, | |
38db52ad W |
33 | }, { |
34 | 'url': 'http://m.drtuber.com/video/3893529/lingerie-blowjob-from-beautiful-teen', | |
35 | 'only_matching': True, | |
519d8970 | 36 | }] |
d740f7e1 | 37 | |
37e7a71c S |
38 | @staticmethod |
39 | def _extract_urls(webpage): | |
40 | return re.findall( | |
41 | r'<iframe[^>]+?src=["\'](?P<url>(?:https?:)?//(?:www\.)?drtuber\.com/embed/\d+)', | |
42 | webpage) | |
43 | ||
d740f7e1 | 44 | def _real_extract(self, url): |
5ad28e7f | 45 | mobj = self._match_valid_url(url) |
d740f7e1 | 46 | video_id = mobj.group('id') |
519d8970 | 47 | display_id = mobj.group('display_id') or video_id |
d740f7e1 | 48 | |
519d8970 S |
49 | webpage = self._download_webpage( |
50 | 'http://www.drtuber.com/video/%s' % video_id, display_id) | |
d740f7e1 | 51 | |
0f4a5a73 PV |
52 | video_data = self._download_json( |
53 | 'http://www.drtuber.com/player_config_json/', video_id, query={ | |
54 | 'vid': video_id, | |
55 | 'embed': 0, | |
56 | 'aid': 0, | |
57 | 'domain_id': 0, | |
58 | }) | |
59 | ||
60 | formats = [] | |
61 | for format_id, video_url in video_data['files'].items(): | |
62 | if video_url: | |
63 | formats.append({ | |
64 | 'format_id': format_id, | |
65 | 'quality': 2 if format_id == 'hq' else 1, | |
66 | 'url': video_url | |
67 | }) | |
68 | self._sort_formats(formats) | |
d740f7e1 | 69 | |
b9bc1cff J |
70 | duration = int_or_none(video_data.get('duration')) or parse_duration( |
71 | video_data.get('duration_format')) | |
72 | ||
d740f7e1 | 73 | title = self._html_search_regex( |
9d15be3a PV |
74 | (r'<h1[^>]+class=["\']title[^>]+>([^<]+)', |
75 | r'<title>([^<]+)\s*@\s+DrTuber', | |
76 | r'class="title_watch"[^>]*><(?:p|h\d+)[^>]*>([^<]+)<', | |
367976d4 S |
77 | r'<p[^>]+class="title_substrate">([^<]+)</p>', |
78 | r'<title>([^<]+) - \d+'), | |
76d1466b | 79 | webpage, 'title') |
d740f7e1 | 80 | |
81 | thumbnail = self._html_search_regex( | |
82 | r'poster="([^"]+)"', | |
83 | webpage, 'thumbnail', fatal=False) | |
84 | ||
0aef0771 | 85 | def extract_count(id_, name, default=NO_DEFAULT): |
86511ea4 S |
86 | return str_to_int(self._html_search_regex( |
87 | r'<span[^>]+(?:class|id)="%s"[^>]*>([\d,\.]+)</span>' % id_, | |
0aef0771 | 88 | webpage, '%s count' % name, default=default, fatal=False)) |
86511ea4 S |
89 | |
90 | like_count = extract_count('rate_likes', 'like') | |
0aef0771 | 91 | dislike_count = extract_count('rate_dislikes', 'dislike', default=None) |
86511ea4 | 92 | comment_count = extract_count('comments_count', 'comment') |
eade1d7e | 93 | |
2914e5f0 | 94 | cats_str = self._search_regex( |
0aef0771 S |
95 | r'<div[^>]+class="categories_list">(.+?)</div>', |
96 | webpage, 'categories', fatal=False) | |
97 | categories = [] if not cats_str else re.findall( | |
98 | r'<a title="([^"]+)"', cats_str) | |
d740f7e1 | 99 | |
100 | return { | |
101 | 'id': video_id, | |
1bf8cf5c | 102 | 'display_id': display_id, |
0f4a5a73 | 103 | 'formats': formats, |
d740f7e1 | 104 | 'title': title, |
105 | 'thumbnail': thumbnail, | |
eade1d7e S |
106 | 'like_count': like_count, |
107 | 'dislike_count': dislike_count, | |
108 | 'comment_count': comment_count, | |
d740f7e1 | 109 | 'categories': categories, |
11fc065c | 110 | 'age_limit': self._rta_search(webpage), |
b9bc1cff | 111 | 'duration': duration, |
d740f7e1 | 112 | } |