yt_dlp/extractor/veoh.py

   1 from __future__ import unicode_literals
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     int_or_none,
   6     parse_duration,
   7     qualities,
   8     try_get
   9 )
  10
  11
  12 class VeohIE(InfoExtractor):
  13     _VALID_URL = r'https?://(?:www\.)?veoh\.com/(?:watch|videos|embed|iphone/#_Watch)/(?P<id>(?:v|e|yapi-)[\da-zA-Z]+)'
  14
  15     _TESTS = [{
  16         'url': 'http://www.veoh.com/watch/v56314296nk7Zdmz3',
  17         'md5': '620e68e6a3cff80086df3348426c9ca3',
  18         'info_dict': {
  19             'id': 'v56314296nk7Zdmz3',
  20             'ext': 'mp4',
  21             'title': 'Straight Backs Are Stronger',
  22             'description': 'md5:203f976279939a6dc664d4001e13f5f4',
  23             'thumbnail': 're:https://fcache\\.veoh\\.com/file/f/th56314296\\.jpg(\\?.*)?',
  24             'uploader': 'LUMOback',
  25             'duration': 46,
  26             'view_count': int,
  27             'average_rating': int,
  28             'comment_count': int,
  29             'age_limit': 0,
  30             'categories': ['technology_and_gaming'],
  31             'tags': ['posture', 'posture', 'sensor', 'back', 'pain', 'wearable', 'tech', 'lumo'],
  32         },
  33     }, {
  34         'url': 'http://www.veoh.com/embed/v56314296nk7Zdmz3',
  35         'only_matching': True,
  36     }, {
  37         'url': 'http://www.veoh.com/watch/v27701988pbTc4wzN?h1=Chile+workers+cover+up+to+avoid+skin+damage',
  38         'md5': '4a6ff84b87d536a6a71e6aa6c0ad07fa',
  39         'info_dict': {
  40             'id': '27701988',
  41             'ext': 'mp4',
  42             'title': 'Chile workers cover up to avoid skin damage',
  43             'description': 'md5:2bd151625a60a32822873efc246ba20d',
  44             'uploader': 'afp-news',
  45             'duration': 123,
  46         },
  47         'skip': 'This video has been deleted.',
  48     }, {
  49         'url': 'http://www.veoh.com/watch/v69525809F6Nc4frX',
  50         'md5': '4fde7b9e33577bab2f2f8f260e30e979',
  51         'note': 'Embedded ooyala video',
  52         'info_dict': {
  53             'id': '69525809',
  54             'ext': 'mp4',
  55             'title': 'Doctors Alter Plan For Preteen\'s Weight Loss Surgery',
  56             'description': 'md5:f5a11c51f8fb51d2315bca0937526891',
  57             'uploader': 'newsy-videos',
  58         },
  59         'skip': 'This video has been deleted.',
  60     }, {
  61         'url': 'http://www.veoh.com/watch/e152215AJxZktGS',
  62         'only_matching': True,
  63     }, {
  64         'url': 'https://www.veoh.com/videos/v16374379WA437rMH',
  65         'md5': 'cceb73f3909063d64f4b93d4defca1b3',
  66         'info_dict': {
  67             'id': 'v16374379WA437rMH',
  68             'ext': 'mp4',
  69             'title': 'Phantasmagoria 2, pt. 1-3',
  70             'description': 'Phantasmagoria: a Puzzle of Flesh',
  71             'thumbnail': 're:https://fcache\\.veoh\\.com/file/f/th16374379\\.jpg(\\?.*)?',
  72             'uploader': 'davidspackage',
  73             'duration': 968,
  74             'view_count': int,
  75             'average_rating': int,
  76             'comment_count': int,
  77             'age_limit': 18,
  78             'categories': ['technology_and_gaming', 'gaming'],
  79             'tags': ['puzzle', 'of', 'flesh'],
  80         }
  81     }]
  82
  83     def _real_extract(self, url):
  84         video_id = self._match_id(url)
  85         metadata = self._download_json(
  86             'https://www.veoh.com/watch/getVideo/' + video_id,
  87             video_id)
  88         video = metadata['video']
  89         title = video['title']
  90
  91         thumbnail_url = None
  92         q = qualities(['Regular', 'HQ'])
  93         formats = []
  94         for f_id, f_url in video.get('src', {}).items():
  95             if not f_url:
  96                 continue
  97             if f_id == 'poster':
  98                 thumbnail_url = f_url
  99             else:
 100                 formats.append({
 101                     'format_id': f_id,
 102                     'quality': q(f_id),
 103                     'url': f_url,
 104                 })
 105         self._sort_formats(formats)
 106
 107         categories = metadata.get('categoryPath')
 108         if not categories:
 109             category = try_get(video, lambda x: x['category'].strip().removeprefix('category_'))
 110             categories = [category] if category else None
 111         tags = video.get('tags')
 112
 113         return {
 114             'id': video_id,
 115             'title': title,
 116             'description': video.get('description'),
 117             'thumbnail': thumbnail_url,
 118             'uploader': video.get('author', {}).get('nickname'),
 119             'duration': int_or_none(video.get('lengthBySec')) or parse_duration(video.get('length')),
 120             'view_count': int_or_none(video.get('views')),
 121             'formats': formats,
 122             'average_rating': int_or_none(video.get('rating')),
 123             'comment_count': int_or_none(video.get('numOfComments')),
 124             'age_limit': 18 if video.get('contentRatingId') == 2 else 0,
 125             'categories': categories,
 126             'tags': tags.split(', ') if tags else None,
 127         }