yt_dlp/extractor/ridehome.py

   1 from .art19 import Art19IE
   2 from .common import InfoExtractor
   3 from ..utils import extract_attributes, get_elements_html_by_class
   4 from ..utils.traversal import traverse_obj
   5
   6
   7 class RideHomeIE(InfoExtractor):
   8     _VALID_URL = r'https?://(?:www\.)?ridehome\.info/show/[\w-]+/(?P<id>[\w-]+)/?(?:$|[?#])'
   9     _TESTS = [{
  10         'url': 'https://www.ridehome.info/show/techmeme-ride-home/thu-1228-will-2024-be-the-year-apple-gets-serious-about-gaming-on-macs/',
  11         'info_dict': {
  12             'id': 'thu-1228-will-2024-be-the-year-apple-gets-serious-about-gaming-on-macs',
  13         },
  14         'playlist_count': 1,
  15         'playlist': [{
  16             'md5': 'c84ea3cc96950a9ab86fe540f3edc588',
  17             'info_dict': {
  18                 'id': '540e5493-9fe6-4c14-a488-dc508d8794b2',
  19                 'ext': 'mp3',
  20                 'title': 'Thu. 12/28 – Will 2024 Be The Year Apple Gets Serious About Gaming On Macs?',
  21                 'description': 'md5:9dba86ae9b5047a8150eceddeeb629c2',
  22                 'series': 'Techmeme Ride Home',
  23                 'series_id': '3c30e8f4-ab48-415b-9421-1ae06cd4058b',
  24                 'upload_date': '20231228',
  25                 'timestamp': 1703780995,
  26                 'modified_date': '20231230',
  27                 'episode_id': '540e5493-9fe6-4c14-a488-dc508d8794b2',
  28                 'modified_timestamp': 1703912404,
  29                 'release_date': '20231228',
  30                 'release_timestamp': 1703782800,
  31                 'duration': 1000.1502,
  32                 'thumbnail': r're:^https?://content\.production\.cdn\.art19\.com/images/.*\.jpeg$',
  33             },
  34         }],
  35     }, {
  36         'url': 'https://www.ridehome.info/show/techmeme-ride-home/portfolio-profile-sensel-with-ilyarosenberg/',
  37         'info_dict': {
  38             'id': 'portfolio-profile-sensel-with-ilyarosenberg',
  39         },
  40         'playlist_count': 1,
  41         'playlist': [{
  42             'md5': 'bf9d6efad221008ce71aea09d5533cf6',
  43             'info_dict': {
  44                 'id': '6beed803-b1ef-4536-9fef-c23cf6b4dcac',
  45                 'ext': 'mp3',
  46                 'title': '(Portfolio Profile) Sensel - With @IlyaRosenberg',
  47                 'description': 'md5:e1e4a970bce04290e0ba6f030b0125db',
  48                 'series': 'Techmeme Ride Home',
  49                 'series_id': '3c30e8f4-ab48-415b-9421-1ae06cd4058b',
  50                 'upload_date': '20220108',
  51                 'timestamp': 1641656064,
  52                 'modified_date': '20230418',
  53                 'episode_id': '6beed803-b1ef-4536-9fef-c23cf6b4dcac',
  54                 'modified_timestamp': 1681843318,
  55                 'release_date': '20220108',
  56                 'release_timestamp': 1641672000,
  57                 'duration': 2789.38122,
  58                 'thumbnail': r're:^https?://content\.production\.cdn\.art19\.com/images/.*\.jpeg$',
  59             },
  60         }],
  61     }, {
  62         'url': 'https://www.ridehome.info/show/spacecasts/big-tech-news-apples-macbook-pro-event/',
  63         'info_dict': {
  64             'id': 'big-tech-news-apples-macbook-pro-event',
  65         },
  66         'playlist_count': 1,
  67         'playlist': [{
  68             'md5': 'b1428530c6e03904a8271e978007fc05',
  69             'info_dict': {
  70                 'id': 'f4780044-6c4b-4ce0-8215-8a86cc66bff7',
  71                 'ext': 'mp3',
  72                 'title': 'md5:e6c05d44d59b6577a4145ac339de5040',
  73                 'description': 'md5:14152f7228c8a301a77e3d6bc891b145',
  74                 'series': 'SpaceCasts',
  75                 'series_id': '8e3e837d-7fe0-4a23-8e11-894917e07e17',
  76                 'upload_date': '20211026',
  77                 'timestamp': 1635271450,
  78                 'modified_date': '20230502',
  79                 'episode_id': 'f4780044-6c4b-4ce0-8215-8a86cc66bff7',
  80                 'modified_timestamp': 1683057500,
  81                 'release_date': '20211026',
  82                 'release_timestamp': 1635272124,
  83                 'duration': 2266.30531,
  84                 'thumbnail': r're:^https?://content\.production\.cdn\.art19\.com/images/.*\.jpeg$',
  85             },
  86         }],
  87     }]
  88
  89     def _real_extract(self, url):
  90         article_id = self._match_id(url)
  91         webpage = self._download_webpage(url, article_id)
  92
  93         urls = traverse_obj(
  94             get_elements_html_by_class('iframeContainer', webpage),
  95             (..., {extract_attributes}, lambda k, v: k == 'data-src' and Art19IE.suitable(v)))
  96         return self.playlist_from_matches(urls, article_id, ie=Art19IE)