yt_dlp/extractor/cloudycdn.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     int_or_none,
   6     parse_iso8601,
   7     url_or_none,
   8     urlencode_postdata,
   9 )
  10 from ..utils.traversal import traverse_obj
  11
  12
  13 class CloudyCDNIE(InfoExtractor):
  14     _VALID_URL = r'(?:https?:)?//embed\.cloudycdn\.services/(?P<site_id>[^/?#]+)/media/(?P<id>[\w-]+)'
  15     _EMBED_REGEX = [rf'<iframe[^>]+\bsrc=[\'"](?P<url>{_VALID_URL})']
  16     _TESTS = [{
  17         'url': 'https://embed.cloudycdn.services/ltv/media/46k_d23-6000-105?',
  18         'md5': '64f72a360ca530d5ed89c77646c9eee5',
  19         'info_dict': {
  20             'id': '46k_d23-6000-105',
  21             'ext': 'mp4',
  22             'timestamp': 1700589151,
  23             'duration': 1442,
  24             'upload_date': '20231121',
  25             'title': 'D23-6000-105_cetstud',
  26             'thumbnail': 'https://store.cloudycdn.services/tmsp00060/assets/media/660858/placeholder1700589200.jpg',
  27         },
  28     }, {
  29         'url': 'https://embed.cloudycdn.services/izm/media/26e_lv-8-5-1',
  30         'md5': '798828a479151e2444d8dcfbec76e482',
  31         'info_dict': {
  32             'id': '26e_lv-8-5-1',
  33             'ext': 'mp4',
  34             'title': 'LV-8-5-1',
  35             'timestamp': 1669767167,
  36             'thumbnail': 'https://store.cloudycdn.services/tmsp00120/assets/media/488306/placeholder1679423604.jpg',
  37             'duration': 1205,
  38             'upload_date': '20221130',
  39         },
  40     }, {
  41         # Video-only m3u8 formats need manual fixup
  42         'url': 'https://embed.cloudycdn.services/ltv/media/08j_d24-6000-074',
  43         'md5': 'fc472e40f6e6238446509be411c920e2',
  44         'info_dict': {
  45             'id': '08j_d24-6000-074',
  46             'ext': 'mp4',
  47             'upload_date': '20240620',
  48             'duration': 1673,
  49             'title': 'D24-6000-074-cetstud',
  50             'timestamp': 1718902233,
  51             'thumbnail': 'https://store.cloudycdn.services/tmsp00060/assets/media/788392/placeholder1718903938.jpg',
  52         },
  53         'params': {'format': 'bv'},
  54     }]
  55     _WEBPAGE_TESTS = [{
  56         'url': 'https://www.tavaklase.lv/video/es-esmu-mina-um-2/',
  57         'md5': '63074e8e6c84ac2a01f2fb8bf03b8f43',
  58         'info_dict': {
  59             'id': 'cqd_lib-2',
  60             'ext': 'mp4',
  61             'upload_date': '20230223',
  62             'duration': 629,
  63             'thumbnail': 'https://store.cloudycdn.services/tmsp00120/assets/media/518407/placeholder1678748124.jpg',
  64             'timestamp': 1677181513,
  65             'title': 'LIB-2',
  66         },
  67     }]
  68
  69     def _real_extract(self, url):
  70         site_id, video_id = self._match_valid_url(url).group('site_id', 'id')
  71
  72         data = self._download_json(
  73             f'https://player.cloudycdn.services/player/{site_id}/media/{video_id}/',
  74             video_id, data=urlencode_postdata({
  75                 'version': '6.4.0',
  76                 'referer': url,
  77             }))
  78
  79         formats, subtitles = [], {}
  80         for m3u8_url in traverse_obj(data, ('source', 'sources', ..., 'src', {url_or_none})):
  81             fmts, subs = self._extract_m3u8_formats_and_subtitles(m3u8_url, video_id, fatal=False)
  82             for fmt in fmts:
  83                 if re.search(r'chunklist_b\d+_vo_', fmt['url']):
  84                     fmt['acodec'] = 'none'
  85             formats.extend(fmts)
  86             self._merge_subtitles(subs, target=subtitles)
  87
  88         return {
  89             'id': video_id,
  90             'formats': formats,
  91             'subtitles': subtitles,
  92             **traverse_obj(data, {
  93                 'title': ('name', {str}),
  94                 'duration': ('duration', {int_or_none}),
  95                 'timestamp': ('upload_date', {parse_iso8601}),
  96                 'thumbnail': ('source', 'poster', {url_or_none}),
  97             }),
  98         }