]>
Commit | Line | Data |
---|---|---|
9b27e6c3 | 1 | # -*- coding: utf-8 -*- |
c060b774 PH |
2 | from __future__ import unicode_literals |
3 | ||
ae287755 PH |
4 | import re |
5 | ||
6 | from .common import InfoExtractor | |
ae287755 PH |
7 | |
8 | ||
9 | class TumblrIE(InfoExtractor): | |
3da0e1f8 | 10 | _VALID_URL = r'http://(?P<blog_name>.*?)\.tumblr\.com/(?:post|video)/(?P<id>[0-9]+)(?:$|[/?#])' |
62f1f950 | 11 | _TESTS = [{ |
c060b774 | 12 | 'url': 'http://tatianamaslanydaily.tumblr.com/post/54196191430/orphan-black-dvd-extra-behind-the-scenes', |
c060b774 PH |
13 | 'md5': '479bb068e5b16462f5176a6828829767', |
14 | 'info_dict': { | |
62f1f950 PP |
15 | 'id': '54196191430', |
16 | 'ext': 'mp4', | |
17 | 'title': 'tatiana maslany news, Orphan Black || DVD extra - behind the scenes ↳...', | |
681b9caa | 18 | 'description': 'md5:37db8211e40b50c7c44e95da14f630b7', |
62f1f950 | 19 | 'thumbnail': 're:http://.*\.jpg', |
6f5ac90c | 20 | } |
62f1f950 PP |
21 | }, { |
22 | 'url': 'http://5sostrum.tumblr.com/post/90208453769/yall-forgetting-the-greatest-keek-of-them-all', | |
23 | 'md5': 'bf348ef8c0ef84fbf1cbd6fa6e000359', | |
24 | 'info_dict': { | |
25 | 'id': '90208453769', | |
26 | 'ext': 'mp4', | |
681b9caa | 27 | 'title': '5SOS STRUM ;]', |
62f1f950 PP |
28 | 'description': 'md5:dba62ac8639482759c8eb10ce474586a', |
29 | 'thumbnail': 're:http://.*\.jpg', | |
30 | } | |
8f947841 YCH |
31 | }, { |
32 | 'url': 'http://naked-yogi.tumblr.com/post/118312946248/naked-smoking-stretching', | |
33 | 'md5': 'de07e5211d60d4f3a2c3df757ea9f6ab', | |
34 | 'info_dict': { | |
35 | 'id': 'Wmur', | |
36 | 'ext': 'mp4', | |
37 | 'title': 'naked smoking & stretching', | |
38 | 'upload_date': '20150506', | |
39 | 'timestamp': 1430931613, | |
40 | }, | |
41 | 'add_ie': ['Vidme'], | |
c5895d5d YCH |
42 | }, { |
43 | 'url': 'http://camdamage.tumblr.com/post/98846056295/', | |
44 | 'md5': 'a9e0c8371ea1ca306d6554e3fecf50b6', | |
45 | 'info_dict': { | |
46 | 'id': '105463834', | |
47 | 'ext': 'mp4', | |
48 | 'title': 'Cam Damage-HD 720p', | |
49 | 'uploader': 'John Moyer', | |
50 | 'uploader_id': 'user32021558', | |
51 | }, | |
52 | 'add_ie': ['Vimeo'], | |
62f1f950 | 53 | }] |
ae287755 PH |
54 | |
55 | def _real_extract(self, url): | |
56 | m_url = re.match(self._VALID_URL, url) | |
57 | video_id = m_url.group('id') | |
58 | blog = m_url.group('blog_name') | |
59 | ||
60 | url = 'http://%s.tumblr.com/post/%s/' % (blog, video_id) | |
c909e582 | 61 | webpage, urlh = self._download_webpage_handle(url, video_id) |
c5895d5d | 62 | |
681b9caa JMF |
63 | iframe_url = self._search_regex( |
64 | r'src=\'(https?://www\.tumblr\.com/video/[^\']+)\'', | |
c909e582 YCH |
65 | webpage, 'iframe url', default=None) |
66 | if iframe_url is None: | |
67 | return self.url_result(urlh.geturl(), 'Generic') | |
68 | ||
3f302bca YCH |
69 | iframe = self._download_webpage(iframe_url, video_id, |
70 | 'Downloading iframe page') | |
681b9caa | 71 | video_url = self._search_regex(r'<source src="([^"]+)"', |
9e1a5b84 | 72 | iframe, 'video url') |
ae287755 PH |
73 | |
74 | # The only place where you can get a title, it's not complete, | |
75 | # but searching in other places doesn't work for all videos | |
3da0e1f8 PH |
76 | video_title = self._html_search_regex( |
77 | r'(?s)<title>(?P<title>.*?)(?: \| Tumblr)?</title>', | |
78 | webpage, 'title') | |
ae287755 | 79 | |
3da0e1f8 PH |
80 | return { |
81 | 'id': video_id, | |
681b9caa JMF |
82 | 'url': video_url, |
83 | 'ext': 'mp4', | |
84 | 'title': video_title, | |
ae849ca1 YCH |
85 | 'description': self._og_search_description(webpage, default=None), |
86 | 'thumbnail': self._og_search_thumbnail(webpage, default=None), | |
3da0e1f8 | 87 | } |