yt_dlp/extractor/utreon.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     dict_get,
   7     int_or_none,
   8     str_or_none,
   9     try_get,
  10     unified_strdate,
  11     url_or_none,
  12 )
  13
  14
  15 class UtreonIE(InfoExtractor):
  16     _VALID_URL = r'https?://(?:www\.)?utreon.com/v/(?P<id>[a-zA-Z0-9_-]+)'
  17     _TESTS = [{
  18         'url': 'https://utreon.com/v/z_I7ikQbuDw',
  19         'info_dict': {
  20             'id': 'z_I7ikQbuDw',
  21             'ext': 'mp4',
  22             'title': 'Freedom Friday meditation - Rising in the wind',
  23             'description': 'md5:a9bf15a42434a062fe313b938343ad1b',
  24             'uploader': 'Heather Dawn Elemental Health',
  25             'thumbnail': 'https://data-1.utreon.com/v/MG/M2/NT/z_I7ikQbuDw/z_I7ikQbuDw_preview.jpg',
  26             'release_date': '20210723',
  27         }
  28     }, {
  29         'url': 'https://utreon.com/v/jerJw5EOOVU',
  30         'info_dict': {
  31             'id': 'jerJw5EOOVU',
  32             'ext': 'mp4',
  33             'title': 'When I\'m alone, I love to reflect in peace, to make my dreams come true... [Quotes and Poems]',
  34             'description': 'md5:61ee6c2da98be51b04b969ca80273aaa',
  35             'uploader': 'Frases e Poemas Quotes and Poems',
  36             'thumbnail': 'https://data-1.utreon.com/v/Mz/Zh/ND/jerJw5EOOVU/jerJw5EOOVU_89af85470a4b16eededde7f8674c96d9_cover.jpg',
  37             'release_date': '20210723',
  38         }
  39     }, {
  40         'url': 'https://utreon.com/v/C4ZxXhYBBmE',
  41         'info_dict': {
  42             'id': 'C4ZxXhYBBmE',
  43             'ext': 'mp4',
  44             'title': 'Biden’s Capital Gains Tax Rate to Test World’s Highest',
  45             'description': 'md5:fb5a6c2e506f013cc76f133f673bc5c8',
  46             'uploader': 'Nomad Capitalist',
  47             'thumbnail': 'https://data-1.utreon.com/v/ZD/k1/Mj/C4ZxXhYBBmE/C4ZxXhYBBmE_628342076198c9c06dd6b2c665978584_cover.jpg',
  48             'release_date': '20210723',
  49         }
  50     }, {
  51         'url': 'https://utreon.com/v/Y-stEH-FBm8',
  52         'info_dict': {
  53             'id': 'Y-stEH-FBm8',
  54             'ext': 'mp4',
  55             'title': 'Creeper-Chan Pranks Steve! 💚 [MINECRAFT ANIME]',
  56             'description': 'md5:7a48450b0d761b96dec194be0c5ecb5f',
  57             'uploader': 'Merryweather Comics',
  58             'thumbnail': 'https://data-1.utreon.com/v/MT/E4/Zj/Y-stEH-FBm8/Y-stEH-FBm8_5290676a41a4a1096db133b09f54f77b_cover.jpg',
  59             'release_date': '20210718',
  60         }},
  61     ]
  62
  63     def _real_extract(self, url):
  64         video_id = self._match_id(url)
  65         json_data = self._download_json(
  66             'https://api.utreon.com/v1/videos/' + video_id,
  67             video_id)
  68         videos_json = json_data['videos']
  69         formats = [{
  70             'url': format_url,
  71             'format_id': format_key.split('_')[1],
  72             'height': int(format_key.split('_')[1][:-1]),
  73         } for format_key, format_url in videos_json.items() if url_or_none(format_url)]
  74         self._sort_formats(formats)
  75         thumbnail = url_or_none(dict_get(json_data, ('cover_image_url', 'preview_image_url')))
  76         return {
  77             'id': video_id,
  78             'title': json_data['title'],
  79             'formats': formats,
  80             'description': str_or_none(json_data.get('description')),
  81             'duration': int_or_none(json_data.get('duration')),
  82             'uploader': str_or_none(try_get(json_data, lambda x: x['channel']['title'])),
  83             'thumbnail': thumbnail,
  84             'release_date': unified_strdate(json_data.get('published_datetime')),
  85         }