_ Git - youtube-dl/blob - youtube_dl/extractor/vube.py

   1 from __future__ import unicode_literals
   2
   3 import json
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import int_or_none
   8
   9
  10 class VubeIE(InfoExtractor):
  11     IE_NAME = 'vube'
  12     IE_DESC = 'Vube.com'
  13     _VALID_URL = r'http://vube\.com/(?:[^/]+/)+(?P<id>[\da-zA-Z]{10})\b'
  14
  15     _TESTS = [
  16         {
  17             'url': 'http://vube.com/Chiara+Grispo+Video+Channel/YL2qNPkqon',
  18             'md5': 'db7aba89d4603dadd627e9d1973946fe',
  19             'info_dict': {
  20                 'id': 'YL2qNPkqon',
  21                 'ext': 'mp4',
  22                 'title': 'Chiara Grispo - Price Tag by Jessie J',
  23                 'description': 'md5:8ea652a1f36818352428cb5134933313',
  24                 'thumbnail': 're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/102e7e63057-5ebc-4f5c-4065-6ce4ebde131f\.jpg$',
  25                 'uploader': 'Chiara.Grispo',
  26                 'timestamp': 1388743358,
  27                 'upload_date': '20140103',
  28                 'duration': 170.56
  29             }
  30         },
  31         {
  32             'url': 'http://vube.com/SerainaMusic/my-7-year-old-sister-and-i-singing-alive-by-krewella/UeBhTudbfS?t=s&n=1',
  33             'md5': '5d4a52492d76f72712117ce6b0d98d08',
  34             'info_dict': {
  35                 'id': 'UeBhTudbfS',
  36                 'ext': 'mp4',
  37                 'title': 'My 7 year old Sister and I singing "Alive" by Krewella',
  38                 'description': 'md5:40bcacb97796339f1690642c21d56f4a',
  39                 'thumbnail': 're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/102265d5a9f-0f17-4f6b-5753-adf08484ee1e\.jpg$',
  40                 'uploader': 'Seraina',
  41                 'timestamp': 1396492438,
  42                 'upload_date': '20140403',
  43                 'duration': 240.107
  44             }
  45         }, {
  46             'url': 'http://vube.com/vote/Siren+Gene/0nmsMY5vEq?n=2&t=s',
  47             'md5': '0584fc13b50f887127d9d1007589d27f',
  48             'info_dict': {
  49                 'id': '0nmsMY5vEq',
  50                 'ext': 'mp4',
  51                 'title': 'Frozen - Let It Go Cover by Siren Gene',
  52                 'description': 'My rendition of "Let It Go" originally sung by Idina Menzel.',
  53                 'uploader': 'Siren Gene',
  54                 'uploader_id': 'Siren',
  55                 'thumbnail': 're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/10283ab622a-86c9-4681-51f2-30d1f65774af\.jpg$',
  56                 'duration': 221.788,
  57                 'like_count': int,
  58                 'dislike_count': int,
  59             }
  60         }
  61     ]
  62
  63     def _real_extract(self, url):
  64         mobj = re.match(self._VALID_URL, url)
  65         video_id = mobj.group('id')
  66
  67         webpage = self._download_webpage(url, video_id)
  68         data_json = self._search_regex(
  69             r'(?s)window\["(?:tapiVideoData|vubeOriginalVideoData)"\]\s*=\s*(\{.*?\n});\n',
  70             webpage, 'video data'
  71         )
  72         data = json.loads(data_json)
  73         open('/dev/shm/f', 'w').write(json.dumps(data, indent=2))
  74         video = (
  75             data.get('video') or
  76             data)
  77         assert isinstance(video, dict)
  78
  79         public_id = video['public_id']
  80
  81         formats = [
  82             {
  83                 'url': 'http://video.thestaticvube.com/video/%s/%s.mp4' % (fmt['media_resolution_id'], public_id),
  84                 'height': int(fmt['height']),
  85                 'abr': int(fmt['audio_bitrate']),
  86                 'vbr': int(fmt['video_bitrate']),
  87                 'format_id': fmt['media_resolution_id']
  88             } for fmt in video['mtm'] if fmt['transcoding_status'] == 'processed'
  89         ]
  90
  91         self._sort_formats(formats)
  92
  93         title = video['title']
  94         description = video.get('description')
  95         thumbnail = self._proto_relative_url(
  96             video.get('thumbnail') or video.get('thumbnail_src'),
  97             scheme='http:')
  98         uploader = data.get('user', {}).get('channel', {}).get('name') or video.get('user_alias')
  99         uploader_id = data.get('user', {}).get('name')
 100         timestamp = int_or_none(video.get('upload_time'))
 101         duration = video['duration']
 102         view_count = video.get('raw_view_count')
 103         like_count = video.get('rlikes')
 104         dislike_count = video.get('rhates')
 105
 106         comment = self._download_json(
 107             'http://vube.com/api/video/%s/comment' % video_id, video_id, 'Downloading video comment JSON')
 108
 109         comment_count = int_or_none(comment.get('total'))
 110
 111         return {
 112             'id': video_id,
 113             'formats': formats,
 114             'title': title,
 115             'description': description,
 116             'thumbnail': thumbnail,
 117             'uploader': uploader,
 118             'uploader_id': uploader_id,
 119             'timestamp': timestamp,
 120             'duration': duration,
 121             'view_count': view_count,
 122             'like_count': like_count,
 123             'dislike_count': dislike_count,
 124             'comment_count': comment_count,
 125         }