yt_dlp/extractor/arnes.py

   1 import urllib.parse
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     float_or_none,
   6     format_field,
   7     int_or_none,
   8     parse_iso8601,
   9     remove_start,
  10 )
  11
  12
  13 class ArnesIE(InfoExtractor):
  14     IE_NAME = 'video.arnes.si'
  15     IE_DESC = 'Arnes Video'
  16     _VALID_URL = r'https?://video\.arnes\.si/(?:[a-z]{2}/)?(?:watch|embed|api/(?:asset|public/video))/(?P<id>[0-9a-zA-Z]{12})'
  17     _TESTS = [{
  18         'url': 'https://video.arnes.si/watch/a1qrWTOQfVoU?t=10',
  19         'md5': '4d0f4d0a03571b33e1efac25fd4a065d',
  20         'info_dict': {
  21             'id': 'a1qrWTOQfVoU',
  22             'ext': 'mp4',
  23             'title': 'Linearna neodvisnost, definicija',
  24             'description': 'Linearna neodvisnost, definicija',
  25             'license': 'PRIVATE',
  26             'creator': 'Polona Oblak',
  27             'timestamp': 1585063725,
  28             'upload_date': '20200324',
  29             'channel': 'Polona Oblak',
  30             'channel_id': 'q6pc04hw24cj',
  31             'channel_url': 'https://video.arnes.si/?channel=q6pc04hw24cj',
  32             'duration': 596.75,
  33             'view_count': int,
  34             'tags': ['linearna_algebra'],
  35             'start_time': 10,
  36         },
  37     }, {
  38         'url': 'https://video.arnes.si/api/asset/s1YjnV7hadlC/play.mp4',
  39         'only_matching': True,
  40     }, {
  41         'url': 'https://video.arnes.si/embed/s1YjnV7hadlC',
  42         'only_matching': True,
  43     }, {
  44         'url': 'https://video.arnes.si/en/watch/s1YjnV7hadlC',
  45         'only_matching': True,
  46     }, {
  47         'url': 'https://video.arnes.si/embed/s1YjnV7hadlC?t=123&hideRelated=1',
  48         'only_matching': True,
  49     }, {
  50         'url': 'https://video.arnes.si/api/public/video/s1YjnV7hadlC',
  51         'only_matching': True,
  52     }]
  53     _BASE_URL = 'https://video.arnes.si'
  54
  55     def _real_extract(self, url):
  56         video_id = self._match_id(url)
  57
  58         video = self._download_json(
  59             self._BASE_URL + '/api/public/video/' + video_id, video_id)['data']
  60         title = video['title']
  61
  62         formats = []
  63         for media in (video.get('media') or []):
  64             media_url = media.get('url')
  65             if not media_url:
  66                 continue
  67             formats.append({
  68                 'url': self._BASE_URL + media_url,
  69                 'format_id': remove_start(media.get('format'), 'FORMAT_'),
  70                 'format_note': media.get('formatTranslation'),
  71                 'width': int_or_none(media.get('width')),
  72                 'height': int_or_none(media.get('height')),
  73             })
  74
  75         channel = video.get('channel') or {}
  76         channel_id = channel.get('url')
  77         thumbnail = video.get('thumbnailUrl')
  78
  79         return {
  80             'id': video_id,
  81             'title': title,
  82             'formats': formats,
  83             'thumbnail': self._BASE_URL + thumbnail,
  84             'description': video.get('description'),
  85             'license': video.get('license'),
  86             'creator': video.get('author'),
  87             'timestamp': parse_iso8601(video.get('creationTime')),
  88             'channel': channel.get('name'),
  89             'channel_id': channel_id,
  90             'channel_url': format_field(channel_id, None, f'{self._BASE_URL}/?channel=%s'),
  91             'duration': float_or_none(video.get('duration'), 1000),
  92             'view_count': int_or_none(video.get('views')),
  93             'tags': video.get('hashtags'),
  94             'start_time': int_or_none(urllib.parse.parse_qs(
  95                 urllib.parse.urlparse(url).query).get('t', [None])[0]),
  96         }