yt_dlp/extractor/veoh.py

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