]>
Commit | Line | Data |
---|---|---|
ebfe352b JMF |
1 | from __future__ import unicode_literals |
2 | ||
825e0984 PH |
3 | import re |
4 | ||
5 | from .common import InfoExtractor | |
2441c1aa | 6 | from .youtube import YoutubeIE |
3052a30d S |
7 | from ..utils import ( |
8 | int_or_none, | |
9 | url_or_none, | |
10 | ) | |
825e0984 PH |
11 | |
12 | ||
13 | class BreakIE(InfoExtractor): | |
2441c1aa | 14 | _VALID_URL = r'https?://(?:www\.)?break\.com/video/(?P<display_id>[^/]+?)(?:-(?P<id>\d+))?(?:[/?#&]|$)' |
f78c01f6 | 15 | _TESTS = [{ |
ebfe352b | 16 | 'url': 'http://www.break.com/video/when-girls-act-like-guys-2468056', |
ebfe352b JMF |
17 | 'info_dict': { |
18 | 'id': '2468056', | |
19 | 'ext': 'mp4', | |
20 | 'title': 'When Girls Act Like D-Bags', | |
d7bb8884 | 21 | 'age_limit': 13, |
221ce325 RA |
22 | }, |
23 | }, { | |
2441c1aa S |
24 | # youtube embed |
25 | 'url': 'http://www.break.com/video/someone-forgot-boat-brakes-work', | |
221ce325 | 26 | 'info_dict': { |
2441c1aa | 27 | 'id': 'RrrDLdeL2HQ', |
221ce325 | 28 | 'ext': 'mp4', |
2441c1aa S |
29 | 'title': 'Whale Watching Boat Crashing Into San Diego Dock', |
30 | 'description': 'md5:afc1b2772f0a8468be51dd80eb021069', | |
31 | 'upload_date': '20160331', | |
32 | 'uploader': 'Steve Holden', | |
33 | 'uploader_id': 'sdholden07', | |
221ce325 | 34 | }, |
2441c1aa S |
35 | 'params': { |
36 | 'skip_download': True, | |
37 | } | |
f78c01f6 S |
38 | }, { |
39 | 'url': 'http://www.break.com/video/ugc/baby-flex-2773063', | |
40 | 'only_matching': True, | |
41 | }] | |
825e0984 PH |
42 | |
43 | def _real_extract(self, url): | |
2441c1aa | 44 | display_id, video_id = re.match(self._VALID_URL, url).groups() |
221ce325 | 45 | |
2441c1aa | 46 | webpage = self._download_webpage(url, display_id) |
221ce325 | 47 | |
2441c1aa S |
48 | youtube_url = YoutubeIE._extract_url(webpage) |
49 | if youtube_url: | |
50 | return self.url_result(youtube_url, ie=YoutubeIE.ie_key()) | |
51 | ||
52 | content = self._parse_json( | |
221ce325 | 53 | self._search_regex( |
2441c1aa S |
54 | r'(?s)content["\']\s*:\s*(\[.+?\])\s*[,\n]', webpage, |
55 | 'content'), | |
221ce325 | 56 | display_id) |
d397c0b3 | 57 | |
221ce325 | 58 | formats = [] |
2441c1aa | 59 | for video in content: |
3052a30d S |
60 | video_url = url_or_none(video.get('url')) |
61 | if not video_url: | |
221ce325 | 62 | continue |
2441c1aa S |
63 | bitrate = int_or_none(self._search_regex( |
64 | r'(\d+)_kbps', video_url, 'tbr', default=None)) | |
d397c0b3 | 65 | formats.append({ |
2441c1aa | 66 | 'url': video_url, |
221ce325 | 67 | 'format_id': 'http-%d' % bitrate if bitrate else 'http', |
221ce325 | 68 | 'tbr': bitrate, |
d397c0b3 | 69 | }) |
2441c1aa | 70 | self._sort_formats(formats) |
d397c0b3 | 71 | |
2441c1aa S |
72 | title = self._search_regex( |
73 | (r'title["\']\s*:\s*(["\'])(?P<value>(?:(?!\1).)+)\1', | |
74 | r'<h1[^>]*>(?P<value>[^<]+)'), webpage, 'title', group='value') | |
d397c0b3 | 75 | |
2441c1aa S |
76 | def get(key, name): |
77 | return int_or_none(self._search_regex( | |
78 | r'%s["\']\s*:\s*["\'](\d+)' % key, webpage, name, | |
79 | default=None)) | |
221ce325 | 80 | |
2441c1aa S |
81 | age_limit = get('ratings', 'age limit') |
82 | video_id = video_id or get('pid', 'video id') or display_id | |
d397c0b3 | 83 | |
ebfe352b JMF |
84 | return { |
85 | 'id': video_id, | |
221ce325 RA |
86 | 'display_id': display_id, |
87 | 'title': title, | |
2441c1aa S |
88 | 'thumbnail': self._og_search_thumbnail(webpage), |
89 | 'age_limit': age_limit, | |
d397c0b3 | 90 | 'formats': formats, |
ebfe352b | 91 | } |