yt_dlp/extractor/coub.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     ExtractorError,
   4     float_or_none,
   5     int_or_none,
   6     parse_iso8601,
   7     qualities,
   8 )
   9
  10
  11 class CoubIE(InfoExtractor):
  12     _VALID_URL = r'(?:coub:|https?://(?:coub\.com/(?:view|embed|coubs)/|c-cdn\.coub\.com/fb-player\.swf\?.*\bcoub(?:ID|id)=))(?P<id>[\da-z]+)'
  13
  14     _TESTS = [{
  15         'url': 'http://coub.com/view/5u5n1',
  16         'info_dict': {
  17             'id': '5u5n1',
  18             'ext': 'mp4',
  19             'title': 'The Matrix Moonwalk',
  20             'thumbnail': r're:^https?://.*\.jpg$',
  21             'duration': 4.6,
  22             'timestamp': 1428527772,
  23             'upload_date': '20150408',
  24             'uploader': 'Artyom Loskutnikov',
  25             'uploader_id': 'artyom.loskutnikov',
  26             'view_count': int,
  27             'like_count': int,
  28             'repost_count': int,
  29             'age_limit': 0,
  30         },
  31     }, {
  32         'url': 'http://c-cdn.coub.com/fb-player.swf?bot_type=vk&coubID=7w5a4',
  33         'only_matching': True,
  34     }, {
  35         'url': 'coub:5u5n1',
  36         'only_matching': True,
  37     }, {
  38         # longer video id
  39         'url': 'http://coub.com/view/237d5l5h',
  40         'only_matching': True,
  41     }]
  42
  43     def _real_extract(self, url):
  44         video_id = self._match_id(url)
  45
  46         coub = self._download_json(
  47             f'http://coub.com/api/v2/coubs/{video_id}.json', video_id)
  48
  49         if coub.get('error'):
  50             raise ExtractorError(
  51                 '{} said: {}'.format(self.IE_NAME, coub['error']), expected=True)
  52
  53         title = coub['title']
  54
  55         file_versions = coub['file_versions']
  56
  57         QUALITIES = ('low', 'med', 'high', 'higher')
  58
  59         MOBILE = 'mobile'
  60         IPHONE = 'iphone'
  61         HTML5 = 'html5'
  62
  63         SOURCE_PREFERENCE = (MOBILE, IPHONE, HTML5)
  64
  65         quality_key = qualities(QUALITIES)
  66         preference_key = qualities(SOURCE_PREFERENCE)
  67
  68         formats = []
  69
  70         for kind, items in file_versions.get(HTML5, {}).items():
  71             if kind not in ('video', 'audio'):
  72                 continue
  73             if not isinstance(items, dict):
  74                 continue
  75             for quality, item in items.items():
  76                 if not isinstance(item, dict):
  77                     continue
  78                 item_url = item.get('url')
  79                 if not item_url:
  80                     continue
  81                 formats.append({
  82                     'url': item_url,
  83                     'format_id': f'{HTML5}-{kind}-{quality}',
  84                     'filesize': int_or_none(item.get('size')),
  85                     'vcodec': 'none' if kind == 'audio' else None,
  86                     'acodec': 'none' if kind == 'video' else None,
  87                     'quality': quality_key(quality),
  88                     'source_preference': preference_key(HTML5),
  89                 })
  90
  91         iphone_url = file_versions.get(IPHONE, {}).get('url')
  92         if iphone_url:
  93             formats.append({
  94                 'url': iphone_url,
  95                 'format_id': IPHONE,
  96                 'source_preference': preference_key(IPHONE),
  97             })
  98
  99         mobile_url = file_versions.get(MOBILE, {}).get('audio_url')
 100         if mobile_url:
 101             formats.append({
 102                 'url': mobile_url,
 103                 'format_id': f'{MOBILE}-audio',
 104                 'source_preference': preference_key(MOBILE),
 105             })
 106
 107         thumbnail = coub.get('picture')
 108         duration = float_or_none(coub.get('duration'))
 109         timestamp = parse_iso8601(coub.get('published_at') or coub.get('created_at'))
 110         uploader = coub.get('channel', {}).get('title')
 111         uploader_id = coub.get('channel', {}).get('permalink')
 112
 113         view_count = int_or_none(coub.get('views_count') or coub.get('views_increase_count'))
 114         like_count = int_or_none(coub.get('likes_count'))
 115         repost_count = int_or_none(coub.get('recoubs_count'))
 116
 117         age_restricted = coub.get('age_restricted', coub.get('age_restricted_by_admin'))
 118         if age_restricted is not None:
 119             age_limit = 18 if age_restricted is True else 0
 120         else:
 121             age_limit = None
 122
 123         return {
 124             'id': video_id,
 125             'title': title,
 126             'thumbnail': thumbnail,
 127             'duration': duration,
 128             'timestamp': timestamp,
 129             'uploader': uploader,
 130             'uploader_id': uploader_id,
 131             'view_count': view_count,
 132             'like_count': like_count,
 133             'repost_count': repost_count,
 134             'age_limit': age_limit,
 135             'formats': formats,
 136         }