yt_dlp/extractor/vube.py

   1 from __future__ import unicode_literals
   2
   3
   4 from .common import InfoExtractor
   5 from ..compat import (
   6     compat_str,
   7 )
   8 from ..utils import (
   9     int_or_none,
  10 )
  11
  12
  13 class VubeIE(InfoExtractor):
  14     IE_NAME = 'vube'
  15     IE_DESC = 'Vube.com'
  16     _VALID_URL = r'https?://vube\.com/(?:[^/]+/)+(?P<id>[\da-zA-Z]{10})\b'
  17
  18     _TESTS = [
  19         {
  20             'url': 'http://vube.com/trending/William+Wei/Y8NUZ69Tf7?t=s',
  21             'md5': 'e7aabe1f8f1aa826b9e4735e1f9cee42',
  22             'info_dict': {
  23                 'id': 'Y8NUZ69Tf7',
  24                 'ext': 'mp4',
  25                 'title': 'Best Drummer Ever [HD]',
  26                 'description': 'md5:2d63c4b277b85c2277761c2cf7337d71',
  27                 'thumbnail': r're:^https?://.*\.jpg',
  28                 'uploader': 'William',
  29                 'timestamp': 1406876915,
  30                 'upload_date': '20140801',
  31                 'duration': 258.051,
  32                 'like_count': int,
  33                 'dislike_count': int,
  34                 'comment_count': int,
  35                 'categories': ['amazing', 'hd', 'best drummer ever', 'william wei', 'bucket drumming', 'street drummer', 'epic street drumming'],
  36             },
  37             'skip': 'Not accessible from Travis CI server',
  38         }, {
  39             'url': 'http://vube.com/Chiara+Grispo+Video+Channel/YL2qNPkqon',
  40             'md5': 'db7aba89d4603dadd627e9d1973946fe',
  41             'info_dict': {
  42                 'id': 'YL2qNPkqon',
  43                 'ext': 'mp4',
  44                 'title': 'Chiara Grispo - Price Tag by Jessie J',
  45                 'description': 'md5:8ea652a1f36818352428cb5134933313',
  46                 'thumbnail': r're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/102e7e63057-5ebc-4f5c-4065-6ce4ebde131f\.jpg$',
  47                 'uploader': 'Chiara.Grispo',
  48                 'timestamp': 1388743358,
  49                 'upload_date': '20140103',
  50                 'duration': 170.56,
  51                 'like_count': int,
  52                 'dislike_count': int,
  53                 'comment_count': int,
  54                 'categories': ['pop', 'music', 'cover', 'singing', 'jessie j', 'price tag', 'chiara grispo'],
  55             },
  56             'skip': 'Removed due to DMCA',
  57         },
  58         {
  59             'url': 'http://vube.com/SerainaMusic/my-7-year-old-sister-and-i-singing-alive-by-krewella/UeBhTudbfS?t=s&n=1',
  60             'md5': '5d4a52492d76f72712117ce6b0d98d08',
  61             'info_dict': {
  62                 'id': 'UeBhTudbfS',
  63                 'ext': 'mp4',
  64                 'title': 'My 7 year old Sister and I singing "Alive" by Krewella',
  65                 'description': 'md5:40bcacb97796339f1690642c21d56f4a',
  66                 'thumbnail': r're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/102265d5a9f-0f17-4f6b-5753-adf08484ee1e\.jpg$',
  67                 'uploader': 'Seraina',
  68                 'timestamp': 1396492438,
  69                 'upload_date': '20140403',
  70                 'duration': 240.107,
  71                 'like_count': int,
  72                 'dislike_count': int,
  73                 'comment_count': int,
  74                 'categories': ['seraina', 'jessica', 'krewella', 'alive'],
  75             },
  76             'skip': 'Removed due to DMCA',
  77         }, {
  78             'url': 'http://vube.com/vote/Siren+Gene/0nmsMY5vEq?n=2&t=s',
  79             'md5': '0584fc13b50f887127d9d1007589d27f',
  80             'info_dict': {
  81                 'id': '0nmsMY5vEq',
  82                 'ext': 'mp4',
  83                 'title': 'Frozen - Let It Go Cover by Siren Gene',
  84                 'description': 'My rendition of "Let It Go" originally sung by Idina Menzel.',
  85                 'thumbnail': r're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/10283ab622a-86c9-4681-51f2-30d1f65774af\.jpg$',
  86                 'uploader': 'Siren',
  87                 'timestamp': 1395448018,
  88                 'upload_date': '20140322',
  89                 'duration': 221.788,
  90                 'like_count': int,
  91                 'dislike_count': int,
  92                 'comment_count': int,
  93                 'categories': ['let it go', 'cover', 'idina menzel', 'frozen', 'singing', 'disney', 'siren gene'],
  94             },
  95             'skip': 'Removed due to DMCA',
  96         }
  97     ]
  98
  99     def _real_extract(self, url):
 100         mobj = self._match_valid_url(url)
 101         video_id = mobj.group('id')
 102
 103         video = self._download_json(
 104             'http://vube.com/t-api/v1/video/%s' % video_id, video_id, 'Downloading video JSON')
 105
 106         public_id = video['public_id']
 107
 108         formats = []
 109
 110         for media in video['media'].get('video', []) + video['media'].get('audio', []):
 111             if media['transcoding_status'] != 'processed':
 112                 continue
 113             fmt = {
 114                 'url': 'http://video.thestaticvube.com/video/%s/%s.mp4' % (media['media_resolution_id'], public_id),
 115                 'abr': int(media['audio_bitrate']),
 116                 'format_id': compat_str(media['media_resolution_id']),
 117             }
 118             vbr = int(media['video_bitrate'])
 119             if vbr:
 120                 fmt.update({
 121                     'vbr': vbr,
 122                     'height': int(media['height']),
 123                 })
 124             formats.append(fmt)
 125
 126         if not formats and video.get('vst') == 'dmca':
 127             self.raise_no_formats(
 128                 'This video has been removed in response to a complaint received under the US Digital Millennium Copyright Act.',
 129                 expected=True)
 130
 131         self._sort_formats(formats)
 132
 133         title = video['title']
 134         description = video.get('description')
 135         thumbnail = self._proto_relative_url(video.get('thumbnail_src'), scheme='http:')
 136         uploader = video.get('user_alias') or video.get('channel')
 137         timestamp = int_or_none(video.get('upload_time'))
 138         duration = video['duration']
 139         view_count = video.get('raw_view_count')
 140         like_count = video.get('total_likes')
 141         dislike_count = video.get('total_hates')
 142
 143         comments = video.get('comments')
 144         comment_count = None
 145         if comments is None:
 146             comment_data = self._download_json(
 147                 'http://vube.com/api/video/%s/comment' % video_id,
 148                 video_id, 'Downloading video comment JSON', fatal=False)
 149             if comment_data is not None:
 150                 comment_count = int_or_none(comment_data.get('total'))
 151         else:
 152             comment_count = len(comments)
 153
 154         categories = [tag['text'] for tag in video['tags']]
 155
 156         return {
 157             'id': video_id,
 158             'formats': formats,
 159             'title': title,
 160             'description': description,
 161             'thumbnail': thumbnail,
 162             'uploader': uploader,
 163             'timestamp': timestamp,
 164             'duration': duration,
 165             'view_count': view_count,
 166             'like_count': like_count,
 167             'dislike_count': dislike_count,
 168             'comment_count': comment_count,
 169             'categories': categories,
 170         }