]>
Commit | Line | Data |
---|---|---|
bb198c95 PH |
1 | from __future__ import unicode_literals |
2 | ||
314368c8 | 3 | import base64 |
cd8b8302 PH |
4 | import re |
5 | ||
6 | from .common import InfoExtractor | |
314368c8 | 7 | from ..utils import qualities |
cd8b8302 PH |
8 | |
9 | ||
10 | class TeamcocoIE(InfoExtractor): | |
dfb2cb5c | 11 | _VALID_URL = r'http://teamcoco\.com/video/(?P<video_id>[0-9]+)?/?(?P<display_id>.*)' |
bb799e81 | 12 | _TESTS = [ |
9e1a5b84 JW |
13 | { |
14 | 'url': 'http://teamcoco.com/video/80187/conan-becomes-a-mary-kay-beauty-consultant', | |
9e1a5b84 JW |
15 | 'md5': '3f7746aa0dc86de18df7539903d399ea', |
16 | 'info_dict': { | |
17d2712d PH |
17 | 'id': '80187', |
18 | 'ext': 'mp4', | |
9e1a5b84 | 19 | 'title': 'Conan Becomes A Mary Kay Beauty Consultant', |
641eb10d NJ |
20 | 'description': 'Mary Kay is perhaps the most trusted name in female beauty, so of course Conan is a natural choice to sell their products.', |
21 | 'age_limit': 0, | |
9e1a5b84 JW |
22 | } |
23 | }, { | |
24 | 'url': 'http://teamcoco.com/video/louis-ck-interview-george-w-bush', | |
9e1a5b84 JW |
25 | 'md5': 'cde9ba0fa3506f5f017ce11ead928f9a', |
26 | 'info_dict': { | |
17d2712d PH |
27 | 'id': '19705', |
28 | 'ext': 'mp4', | |
314368c8 NJ |
29 | 'description': 'Louis C.K. got starstruck by George W. Bush, so what? Part one.', |
30 | 'title': 'Louis C.K. Interview Pt. 1 11/3/11', | |
641eb10d | 31 | 'age_limit': 0, |
9e1a5b84 | 32 | } |
bb799e81 | 33 | } |
bb799e81 | 34 | ] |
3811c567 NJ |
35 | _VIDEO_ID_REGEXES = ( |
36 | r'"eVar42"\s*:\s*(\d+)', | |
37 | r'Ginger\.TeamCoco\.openInApp\("video",\s*"([^"]+)"', | |
38 | r'"id_not"\s*:\s*(\d+)' | |
39 | ) | |
cd8b8302 PH |
40 | |
41 | def _real_extract(self, url): | |
42 | mobj = re.match(self._VALID_URL, url) | |
dfb2cb5c PH |
43 | |
44 | display_id = mobj.group('display_id') | |
45 | webpage = self._download_webpage(url, display_id) | |
5f6a1245 | 46 | |
314368c8 | 47 | video_id = mobj.group('video_id') |
dfb2cb5c | 48 | if not video_id: |
04ee53ec | 49 | video_id = self._html_search_regex( |
3811c567 | 50 | self._VIDEO_ID_REGEXES, webpage, 'video id') |
cd8b8302 | 51 | |
314368c8 NJ |
52 | embed_url = 'http://teamcoco.com/embed/v/%s' % video_id |
53 | embed = self._download_webpage( | |
54 | embed_url, video_id, 'Downloading embed page') | |
55 | ||
56 | encoded_data = self._search_regex( | |
57 | r'"preload"\s*:\s*"([^"]+)"', embed, 'encoded data') | |
58 | data = self._parse_json( | |
59 | base64.b64decode(encoded_data.encode('ascii')).decode('utf-8'), video_id) | |
cd8b8302 | 60 | |
e7e6b54d | 61 | formats = [] |
314368c8 NJ |
62 | get_quality = qualities(['500k', '480p', '1000k', '720p', '1080p']) |
63 | for filed in data['files']: | |
64 | m_format = re.search(r'(\d+(k|p))\.mp4', filed['url']) | |
e7e6b54d JMF |
65 | if m_format is not None: |
66 | format_id = m_format.group(1) | |
67 | else: | |
314368c8 | 68 | format_id = filed['bitrate'] |
befdc8f3 | 69 | tbr = ( |
314368c8 NJ |
70 | int(filed['bitrate']) |
71 | if filed['bitrate'].isdigit() | |
befdc8f3 PH |
72 | else None) |
73 | ||
e7e6b54d | 74 | formats.append({ |
314368c8 | 75 | 'url': filed['url'], |
e7e6b54d | 76 | 'ext': 'mp4', |
befdc8f3 | 77 | 'tbr': tbr, |
e7e6b54d | 78 | 'format_id': format_id, |
314368c8 | 79 | 'quality': get_quality(format_id), |
e7e6b54d | 80 | }) |
befdc8f3 PH |
81 | |
82 | self._sort_formats(formats) | |
cd8b8302 | 83 | |
e7e6b54d | 84 | return { |
bb198c95 | 85 | 'id': video_id, |
dfb2cb5c | 86 | 'display_id': display_id, |
e7e6b54d | 87 | 'formats': formats, |
314368c8 NJ |
88 | 'title': data['title'], |
89 | 'thumbnail': data.get('thumb', {}).get('href'), | |
90 | 'description': data.get('teaser'), | |
641eb10d | 91 | 'age_limit': self._family_friendly_search(webpage), |
e7e6b54d | 92 | } |