yt_dlp/extractor/veo.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     int_or_none,
   4     mimetype2ext,
   5     str_or_none,
   6     unified_timestamp,
   7     url_or_none,
   8 )
   9
  10
  11 class VeoIE(InfoExtractor):
  12     _VALID_URL = r'https?://app\.veo\.co/matches/(?P<id>[0-9A-Za-z-_]+)'
  13
  14     _TESTS = [{
  15         'url': 'https://app.veo.co/matches/20201027-last-period/',
  16         'info_dict': {
  17             'id': '20201027-last-period',
  18             'ext': 'mp4',
  19             'title': 'Akidemy u11s v Bradford Boys u11s (Game 3)',
  20             'thumbnail': 're:https://c.veocdn.com/.+/thumbnail.jpg',
  21             'upload_date': '20201028',
  22             'timestamp': 1603847208,
  23             'duration': 1916,
  24             'view_count': int,
  25         }
  26     }, {
  27         'url': 'https://app.veo.co/matches/20220313-2022-03-13_u15m-plsjq-vs-csl/',
  28         'only_matching': True,
  29     }]
  30
  31     def _real_extract(self, url):
  32         video_id = self._match_id(url)
  33
  34         metadata = self._download_json(
  35             'https://app.veo.co/api/app/matches/%s' % video_id, video_id)
  36
  37         video_data = self._download_json(
  38             'https://app.veo.co/api/app/matches/%s/videos' % video_id, video_id, 'Downloading video data')
  39
  40         formats = []
  41         for fmt in video_data:
  42             mimetype = str_or_none(fmt.get('mime_type'))
  43             format_url = url_or_none(fmt.get('url'))
  44             # skip configuration file for panoramic video
  45             if not format_url or mimetype == 'video/mp2t':
  46                 continue
  47
  48             height = int_or_none(fmt.get('height'))
  49             render_type = str_or_none(fmt.get('render_type'))
  50             format_id = f'{render_type}-{height}p' if render_type and height else None
  51
  52             # Veo returns panoramic video information even if panoramic video is not available.
  53             # e.g. https://app.veo.co/matches/20201027-last-period/
  54             if render_type == 'panorama':
  55                 if not self._is_valid_url(format_url, video_id, format_id):
  56                     continue
  57
  58             formats.append({
  59                 'url': format_url,
  60                 'format_id': format_id,
  61                 'ext': mimetype2ext(mimetype),
  62                 'width': int_or_none(fmt.get('width')),
  63                 'height': height,
  64                 'vbr': int_or_none(fmt.get('bit_rate'), scale=1000),
  65             })
  66
  67         return {
  68             'id': video_id,
  69             'title': str_or_none(metadata.get('title')),
  70             'formats': formats,
  71             'thumbnail': url_or_none(metadata.get('thumbnail')),
  72             'timestamp': unified_timestamp(metadata.get('created')),
  73             'view_count': int_or_none(metadata.get('view_count')),
  74             'duration': int_or_none(metadata.get('duration')),
  75         }