yt_dlp/extractor/parler.py

   1 from .common import InfoExtractor
   2 from .youtube import YoutubeIE
   3 from ..utils import (
   4     clean_html,
   5     int_or_none,
   6     strip_or_none,
   7     traverse_obj,
   8     unified_timestamp,
   9     urljoin,
  10 )
  11
  12
  13 class ParlerIE(InfoExtractor):
  14     IE_DESC = 'Posts on parler.com'
  15     _VALID_URL = r'https?://parler\.com/feed/(?P<id>[0-9a-f]{8}-(?:[0-9a-f]{4}-){3}[0-9a-f]{12})'
  16     _TESTS = [
  17         {
  18             'url': 'https://parler.com/feed/df79fdba-07cc-48fe-b085-3293897520d7',
  19             'md5': '16e0f447bf186bb3cf64de5bbbf4d22d',
  20             'info_dict': {
  21                 'id': 'df79fdba-07cc-48fe-b085-3293897520d7',
  22                 'ext': 'mp4',
  23                 'thumbnail': 'https://bl-images.parler.com/videos/6ce7cdf3-a27a-4d72-bf9c-d3e17ce39a66/thumbnail.jpeg',
  24                 'title': 'Parler video #df79fdba-07cc-48fe-b085-3293897520d7',
  25                 'description': 'md5:6f220bde2df4a97cbb89ac11f1fd8197',
  26                 'timestamp': 1659785481,
  27                 'upload_date': '20220806',
  28                 'uploader': 'Tulsi Gabbard',
  29                 'uploader_id': 'TulsiGabbard',
  30                 'uploader_url': 'https://parler.com/TulsiGabbard',
  31                 'view_count': int,
  32                 'comment_count': int,
  33                 'repost_count': int,
  34             },
  35         },
  36         {
  37             'url': 'https://parler.com/feed/f23b85c1-6558-470f-b9ff-02c145f28da5',
  38             'md5': 'eaba1ff4a10fe281f5ce74e930ab2cb4',
  39             'info_dict': {
  40                 'id': 'r5vkSaz8PxQ',
  41                 'ext': 'mp4',
  42                 'live_status': 'not_live',
  43                 'comment_count': int,
  44                 'duration': 1267,
  45                 'like_count': int,
  46                 'channel_follower_count': int,
  47                 'channel_id': 'UCox6YeMSY1PQInbCtTaZj_w',
  48                 'upload_date': '20220716',
  49                 'thumbnail': 'https://i.ytimg.com/vi/r5vkSaz8PxQ/maxresdefault.jpg',
  50                 'tags': 'count:17',
  51                 'availability': 'public',
  52                 'categories': ['Entertainment'],
  53                 'playable_in_embed': True,
  54                 'channel': 'Who Knows What! With Mahesh & Friends',
  55                 'title': 'Tom MacDonald Names Reaction',
  56                 'uploader': 'Who Knows What! With Mahesh & Friends',
  57                 'uploader_id': '@maheshchookolingo',
  58                 'age_limit': 0,
  59                 'description': 'md5:33c21f0d35ae6dc2edf3007d6696baea',
  60                 'channel_url': 'https://www.youtube.com/channel/UCox6YeMSY1PQInbCtTaZj_w',
  61                 'view_count': int,
  62                 'uploader_url': 'http://www.youtube.com/@maheshchookolingo',
  63             },
  64         },
  65     ]
  66
  67     def _real_extract(self, url):
  68         video_id = self._match_id(url)
  69         data = self._download_json(f'https://api.parler.com/v0/public/parleys/{video_id}',
  70                                    video_id)['data']
  71         if data.get('link'):
  72             return self.url_result(data['link'], YoutubeIE)
  73
  74         return {
  75             'id': video_id,
  76             'title': strip_or_none(data.get('title')) or '',
  77             **traverse_obj(data, {
  78                 'url': ('video', 'videoSrc'),
  79                 'thumbnail': ('video', 'thumbnailUrl'),
  80                 'description': ('body', {clean_html}),
  81                 'timestamp': ('date_created', {unified_timestamp}),
  82                 'uploader': ('user', 'name', {strip_or_none}),
  83                 'uploader_id': ('user', 'username', {str}),
  84                 'uploader_url': ('user', 'username', {urljoin('https://parler.com/')}),
  85                 'view_count': ('views', {int_or_none}),
  86                 'comment_count': ('total_comments', {int_or_none}),
  87                 'repost_count': ('echos', {int_or_none}),
  88             }),
  89         }