[vgtv] Add new extractor
[youtube-dl] / youtube_dl / extractor / vube.py
1 from __future__ import unicode_literals
2
3 import re
4
5 from .common import InfoExtractor
6 from ..utils import int_or_none
7
8
9 class VubeIE(InfoExtractor):
10     IE_NAME = 'vube'
11     IE_DESC = 'Vube.com'
12     _VALID_URL = r'http://vube\.com/(?:[^/]+/)+(?P<id>[\da-zA-Z]{10})\b'
13
14     _TESTS = [
15         {
16             'url': 'http://vube.com/Chiara+Grispo+Video+Channel/YL2qNPkqon',
17             'md5': 'db7aba89d4603dadd627e9d1973946fe',
18             'info_dict': {
19                 'id': 'YL2qNPkqon',
20                 'ext': 'mp4',
21                 'title': 'Chiara Grispo - Price Tag by Jessie J',
22                 'description': 'md5:8ea652a1f36818352428cb5134933313',
23                 'thumbnail': 'http://frame.thestaticvube.com/snap/228x128/102e7e63057-5ebc-4f5c-4065-6ce4ebde131f.jpg',
24                 'uploader': 'Chiara.Grispo',
25                 'uploader_id': '1u3hX0znhP',
26                 'timestamp': 1388743358,
27                 'upload_date': '20140103',
28                 'duration': 170.56
29             }
30         },
31         {
32             'url': 'http://vube.com/SerainaMusic/my-7-year-old-sister-and-i-singing-alive-by-krewella/UeBhTudbfS?t=s&n=1',
33             'md5': '5d4a52492d76f72712117ce6b0d98d08',
34             'info_dict': {
35                 'id': 'UeBhTudbfS',
36                 'ext': 'mp4',
37                 'title': 'My 7 year old Sister and I singing "Alive" by Krewella',
38                 'description': 'md5:40bcacb97796339f1690642c21d56f4a',
39                 'thumbnail': 'http://frame.thestaticvube.com/snap/228x128/102265d5a9f-0f17-4f6b-5753-adf08484ee1e.jpg',
40                 'uploader': 'Seraina',
41                 'uploader_id': 'XU9VE2BQ2q',
42                 'timestamp': 1396492438,
43                 'upload_date': '20140403',
44                 'duration': 240.107
45             }
46         }
47     ]
48
49     def _real_extract(self, url):
50         mobj = re.match(self._VALID_URL, url)
51         video_id = mobj.group('id')
52
53         video = self._download_json(
54             'http://vube.com/api/v2/video/%s' % video_id, video_id, 'Downloading video JSON')
55
56         public_id = video['public_id']
57
58         formats = [
59             {
60                 'url': 'http://video.thestaticvube.com/video/%s/%s.mp4' % (fmt['media_resolution_id'], public_id),
61                 'height': int(fmt['height']),
62                 'abr': int(fmt['audio_bitrate']),
63                 'vbr': int(fmt['video_bitrate']),
64                 'format_id': fmt['media_resolution_id']
65             } for fmt in video['mtm'] if fmt['transcoding_status'] == 'processed'
66         ]
67
68         self._sort_formats(formats)
69
70         title = video['title']
71         description = video.get('description')
72         thumbnail = video['thumbnail_src']
73         if thumbnail.startswith('//'):
74             thumbnail = 'http:' + thumbnail
75         uploader = video['user_alias']
76         uploader_id = video['user_url_id']
77         timestamp = int(video['upload_time'])
78         duration = video['duration']
79         view_count = video.get('raw_view_count')
80         like_count = video.get('total_likes')
81         dislike_count= video.get('total_hates')
82
83         comment = self._download_json(
84             'http://vube.com/api/video/%s/comment' % video_id, video_id, 'Downloading video comment JSON')
85
86         comment_count = int_or_none(comment.get('total'))
87
88         return {
89             'id': video_id,
90             'formats': formats,
91             'title': title,
92             'description': description,
93             'thumbnail': thumbnail,
94             'uploader': uploader,
95             'uploader_id': uploader_id,
96             'timestamp': timestamp,
97             'duration': duration,
98             'view_count': view_count,
99             'like_count': like_count,
100             'dislike_count': dislike_count,
101             'comment_count': comment_count,
102         }