yt_dlp/extractor/tvopengr.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     determine_ext,
   4     get_elements_text_and_html_by_attribute,
   5     scale_thumbnails_to_max_format_width,
   6 )
   7
   8
   9 class TVOpenGrBaseIE(InfoExtractor):
  10     def _return_canonical_url(self, url, video_id):
  11         webpage = self._download_webpage(url, video_id)
  12         canonical_url = self._og_search_url(webpage)
  13         title = self._og_search_title(webpage)
  14         return self.url_result(canonical_url, ie=TVOpenGrWatchIE.ie_key(), video_id=video_id, video_title=title)
  15
  16
  17 class TVOpenGrWatchIE(TVOpenGrBaseIE):
  18     IE_NAME = 'tvopengr:watch'
  19     IE_DESC = 'tvopen.gr (and ethnos.gr) videos'
  20     _VALID_URL = r'https?://(?P<netloc>(?:www\.)?(?:tvopen|ethnos)\.gr)/watch/(?P<id>\d+)/(?P<slug>[^/]+)'
  21     _API_ENDPOINT = 'https://www.tvopen.gr/templates/data/player'
  22
  23     _TESTS = [{
  24         'url': 'https://www.ethnos.gr/watch/101009/nikoskaprabelosdenexoymekanenanasthenhsemethmethmetallaxhomikron',
  25         'md5': '8728570e3a72e0f8d9475ba94859fdc1',
  26         'info_dict': {
  27             'id': '101009',
  28             'title': 'md5:51f68773dcb6c70498cd326f45fefdf0',
  29             'display_id': 'nikoskaprabelosdenexoymekanenanasthenhsemethmethmetallaxhomikron',
  30             'description': 'md5:78fff49f18fb3effe41b070e5c7685d6',
  31             'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/d573ba71-ec5f-43c6-b4cb-d181f327d3a8.jpg',
  32             'ext': 'mp4',
  33             'upload_date': '20220109',
  34             'timestamp': 1641686400,
  35         },
  36     }, {
  37         'url': 'https://www.tvopen.gr/watch/100979/se28099agapaomenalla7cepeisodio267cmhthrargiapashskakias',
  38         'md5': '38f98a1be0c577db4ea2d1b1c0770c48',
  39         'info_dict': {
  40             'id': '100979',
  41             'title': 'md5:e021f3001e16088ee40fa79b20df305b',
  42             'display_id': 'se28099agapaomenalla7cepeisodio267cmhthrargiapashskakias',
  43             'description': 'md5:ba17db53954134eb8d625d199e2919fb',
  44             'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/9bb71cf1-21da-43a9-9d65-367950fde4e3.jpg',
  45             'ext': 'mp4',
  46             'upload_date': '20220108',
  47             'timestamp': 1641600000,
  48         },
  49     }]
  50
  51     def _extract_formats_and_subs(self, response, video_id):
  52         formats, subs = [], {}
  53         for format_id, format_url in response.items():
  54             if format_id not in ('stream', 'httpstream', 'mpegdash'):
  55                 continue
  56             ext = determine_ext(format_url)
  57             if ext == 'm3u8':
  58                 formats_, subs_ = self._extract_m3u8_formats_and_subtitles(
  59                     format_url, video_id, 'mp4', m3u8_id=format_id,
  60                     fatal=False)
  61             elif ext == 'mpd':
  62                 formats_, subs_ = self._extract_mpd_formats_and_subtitles(
  63                     format_url, video_id, 'mp4', fatal=False)
  64             else:
  65                 formats.append({
  66                     'url': format_url,
  67                     'format_id': format_id,
  68                 })
  69                 continue
  70             formats.extend(formats_)
  71             self._merge_subtitles(subs_, target=subs)
  72         return formats, subs
  73
  74     def _real_extract(self, url):
  75         netloc, video_id, display_id = self._match_valid_url(url).group('netloc', 'id', 'slug')
  76         if netloc.find('tvopen.gr') == -1:
  77             return self._return_canonical_url(url, video_id)
  78         webpage = self._download_webpage(url, video_id)
  79         info = self._search_json_ld(webpage, video_id, expected_type='VideoObject')
  80         info['formats'], info['subtitles'] = self._extract_formats_and_subs(
  81             self._download_json(self._API_ENDPOINT, video_id, query={'cid': video_id}),
  82             video_id)
  83         info['thumbnails'] = scale_thumbnails_to_max_format_width(
  84             info['formats'], info['thumbnails'], r'(?<=/imgHandler/)\d+')
  85         description, _html = next(get_elements_text_and_html_by_attribute('class', 'description', webpage))
  86         if description and _html.startswith('<span '):
  87             info['description'] = description
  88         info['id'] = video_id
  89         info['display_id'] = display_id
  90         return info
  91
  92
  93 class TVOpenGrEmbedIE(TVOpenGrBaseIE):
  94     IE_NAME = 'tvopengr:embed'
  95     IE_DESC = 'tvopen.gr embedded videos'
  96     _VALID_URL = r'(?:https?:)?//(?:www\.|cdn\.|)(?:tvopen|ethnos).gr/embed/(?P<id>\d+)'
  97     _EMBED_REGEX = [rf'''<iframe[^>]+?src=(?P<_q1>["'])(?P<url>{_VALID_URL})(?P=_q1)''']
  98
  99     _TESTS = [{
 100         'url': 'https://cdn.ethnos.gr/embed/100963',
 101         'md5': '2da147881f45571d81662d94d086628b',
 102         'info_dict': {
 103             'id': '100963',
 104             'display_id': 'koronoiosapotoysdieythyntestonsxoleionselftestgiaosoysdenbrhkan',
 105             'title': 'md5:2c71876fadf0cda6043da0da5fca2936',
 106             'description': 'md5:17482b4432e5ed30eccd93b05d6ea509',
 107             'thumbnail': 'https://opentv-static.siliconweb.com/imgHandler/1920/5804e07f-799a-4247-a696-33842c94ca37.jpg',
 108             'ext': 'mp4',
 109             'upload_date': '20220108',
 110             'timestamp': 1641600000,
 111         },
 112     }]
 113
 114     def _real_extract(self, url):
 115         video_id = self._match_id(url)
 116         return self._return_canonical_url(url, video_id)