yt_dlp/extractor/fathom.py

   1 import json
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     extract_attributes,
   6     float_or_none,
   7     get_element_html_by_id,
   8     parse_iso8601,
   9 )
  10 from ..utils.traversal import traverse_obj
  11
  12
  13 class FathomIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:www\.)?fathom\.video/share/(?P<id>[^/?#&]+)'
  15     _TESTS = [{
  16         'url': 'https://fathom.video/share/G9mkjkspnohVVZ_L5nrsoPycyWcB8y7s',
  17         'md5': '0decd5343b8f30ae268625e79a02b60f',
  18         'info_dict': {
  19             'id': '47200596',
  20             'ext': 'mp4',
  21             'title': 'eCom Inucbator - Coaching Session',
  22             'duration': 8125.380507,
  23             'timestamp': 1699048914,
  24             'upload_date': '20231103',
  25         },
  26     }, {
  27         'url': 'https://fathom.video/share/mEws3bybftHL2QLymxYEDeE21vtLxGVm',
  28         'md5': '4f5cb382126c22d1aba8a939f9c49690',
  29         'info_dict': {
  30             'id': '46812957',
  31             'ext': 'mp4',
  32             'title': 'Jon, Lawrence, Neman chat about practice',
  33             'duration': 3571.517847,
  34             'timestamp': 1698933600,
  35             'upload_date': '20231102',
  36         },
  37     }]
  38
  39     def _real_extract(self, url):
  40         display_id = self._match_id(url)
  41         webpage = self._download_webpage(url, display_id)
  42         props = traverse_obj(
  43             get_element_html_by_id('app', webpage), ({extract_attributes}, 'data-page', {json.loads}, 'props'))
  44         video_id = str(props['call']['id'])
  45
  46         return {
  47             'id': video_id,
  48             'formats': self._extract_m3u8_formats(props['call']['video_url'], video_id, 'mp4'),
  49             **traverse_obj(props, {
  50                 'title': ('head', 'title', {str}),
  51                 'duration': ('duration', {float_or_none}),
  52                 'timestamp': ('call', 'started_at', {parse_iso8601}),
  53             }),
  54         }