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 (
   7     int_or_none,
   8     compat_str,
   9 )
  10
  11
  12 class VubeIE(InfoExtractor):
  13     IE_NAME = 'vube'
  14     IE_DESC = 'Vube.com'
  15     _VALID_URL = r'http://vube\.com/(?:[^/]+/)+(?P<id>[\da-zA-Z]{10})\b'
  16
  17     _TESTS = [
  18         {
  19             'url': 'http://vube.com/Chiara+Grispo+Video+Channel/YL2qNPkqon',
  20             'md5': 'db7aba89d4603dadd627e9d1973946fe',
  21             'info_dict': {
  22                 'id': 'YL2qNPkqon',
  23                 'ext': 'mp4',
  24                 'title': 'Chiara Grispo - Price Tag by Jessie J',
  25                 'description': 'md5:8ea652a1f36818352428cb5134933313',
  26                 'thumbnail': 're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/102e7e63057-5ebc-4f5c-4065-6ce4ebde131f\.jpg$',
  27                 'uploader': 'Chiara.Grispo',
  28                 'timestamp': 1388743358,
  29                 'upload_date': '20140103',
  30                 'duration': 170.56,
  31                 'like_count': int,
  32                 'dislike_count': int,
  33                 'comment_count': int,
  34                 'categories': ['pop', 'music', 'cover', 'singing', 'jessie j', 'price tag', 'chiara grispo'],
  35             }
  36         },
  37         {
  38             'url': 'http://vube.com/SerainaMusic/my-7-year-old-sister-and-i-singing-alive-by-krewella/UeBhTudbfS?t=s&n=1',
  39             'md5': '5d4a52492d76f72712117ce6b0d98d08',
  40             'info_dict': {
  41                 'id': 'UeBhTudbfS',
  42                 'ext': 'mp4',
  43                 'title': 'My 7 year old Sister and I singing "Alive" by Krewella',
  44                 'description': 'md5:40bcacb97796339f1690642c21d56f4a',
  45                 'thumbnail': 're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/102265d5a9f-0f17-4f6b-5753-adf08484ee1e\.jpg$',
  46                 'uploader': 'Seraina',
  47                 'timestamp': 1396492438,
  48                 'upload_date': '20140403',
  49                 'duration': 240.107,
  50                 'like_count': int,
  51                 'dislike_count': int,
  52                 'comment_count': int,
  53                 'categories': ['seraina', 'jessica', 'krewella', 'alive'],
  54             }
  55         }, {
  56             'url': 'http://vube.com/vote/Siren+Gene/0nmsMY5vEq?n=2&t=s',
  57             'md5': '0584fc13b50f887127d9d1007589d27f',
  58             'info_dict': {
  59                 'id': '0nmsMY5vEq',
  60                 'ext': 'mp4',
  61                 'title': 'Frozen - Let It Go Cover by Siren Gene',
  62                 'description': 'My rendition of "Let It Go" originally sung by Idina Menzel.',
  63                 'thumbnail': 're:^http://frame\.thestaticvube\.com/snap/[0-9x]+/10283ab622a-86c9-4681-51f2-30d1f65774af\.jpg$',
  64                 'uploader': 'Siren',
  65                 'timestamp': 1395448018,
  66                 'upload_date': '20140322',
  67                 'duration': 221.788,
  68                 'like_count': int,
  69                 'dislike_count': int,
  70                 'comment_count': int,
  71                 'categories': ['let it go', 'cover', 'idina menzel', 'frozen', 'singing', 'disney', 'siren gene'],
  72             }
  73         }
  74     ]
  75
  76     def _real_extract(self, url):
  77         mobj = re.match(self._VALID_URL, url)
  78         video_id = mobj.group('id')
  79
  80         video = self._download_json(
  81             'http://vube.com/t-api/v1/video/%s' % video_id, video_id, 'Downloading video JSON')
  82
  83         public_id = video['public_id']
  84
  85         formats = []
  86
  87         for media in video['media'].get('video', []) + video['media'].get('audio', []):
  88             if media['transcoding_status'] != 'processed':
  89                 continue
  90             fmt = {
  91                 'url': 'http://video.thestaticvube.com/video/%s/%s.mp4' % (media['media_resolution_id'], public_id),
  92                 'abr': int(media['audio_bitrate']),
  93                 'format_id': compat_str(media['media_resolution_id']),
  94             }
  95             vbr = int(media['video_bitrate'])
  96             if vbr:
  97                 fmt.update({
  98                     'vbr': vbr,
  99                     'height': int(media['height']),
 100                 })
 101             formats.append(fmt)
 102
 103         self._sort_formats(formats)
 104
 105         title = video['title']
 106         description = video.get('description')
 107         thumbnail = self._proto_relative_url(video.get('thumbnail_src'), scheme='http:')
 108         uploader = video.get('user_alias') or video.get('channel')
 109         timestamp = int_or_none(video.get('upload_time'))
 110         duration = video['duration']
 111         view_count = video.get('raw_view_count')
 112         like_count = video.get('total_likes')
 113         dislike_count = video.get('total_hates')
 114
 115         comments = video.get('comments')
 116         comment_count = None
 117         if comments is None:
 118             comment_data = self._download_json(
 119                 'http://vube.com/api/video/%s/comment' % video_id,
 120                 video_id, 'Downloading video comment JSON', fatal=False)
 121             if comment_data is not None:
 122                 comment_count = int_or_none(comment_data.get('total'))
 123         else:
 124             comment_count = len(comments)
 125
 126         categories = [tag['text'] for tag in video['tags']]
 127
 128         return {
 129             'id': video_id,
 130             'formats': formats,
 131             'title': title,
 132             'description': description,
 133             'thumbnail': thumbnail,
 134             'uploader': uploader,
 135             'timestamp': timestamp,
 136             'duration': duration,
 137             'view_count': view_count,
 138             'like_count': like_count,
 139             'dislike_count': dislike_count,
 140             'comment_count': comment_count,
 141             'categories': categories,
 142         }