yt_dlp/extractor/tv2hu.py

   1 # encoding: utf-8
   2 from .common import InfoExtractor
   3 from ..utils import (
   4     traverse_obj,
   5     UnsupportedError,
   6 )
   7
   8
   9 class TV2HuIE(InfoExtractor):
  10     IE_NAME = 'tv2play.hu'
  11     _VALID_URL = r'https?://(?:www\.)?tv2play\.hu/(?!szalag/)(?P<id>[^#&?]+)'
  12     _TESTS = [{
  13         'url': 'https://tv2play.hu/mintaapak/mintaapak_213_epizod_resz',
  14         'info_dict': {
  15             'id': '249240',
  16             'ext': 'mp4',
  17             'title': 'Mintaapák - 213. epizód',
  18             'series': 'Mintaapák',
  19             'duration': 2164,
  20             'description': 'md5:7350147e75485a59598e806c47967b07',
  21             'thumbnail': r're:^https?://.*\.jpg$',
  22             'release_date': '20210825',
  23             'episode_number': 213,
  24         },
  25         'params': {
  26             'skip_download': True,
  27         },
  28     }, {
  29         'url': 'https://tv2play.hu/taxi_2',
  30         'md5': '585e58e2e090f34603804bb2c48e98d8',
  31         'info_dict': {
  32             'id': '199363',
  33             'ext': 'mp4',
  34             'title': 'Taxi 2',
  35             'series': 'Taxi 2',
  36             'duration': 5087,
  37             'description': 'md5:47762155dc9a50241797ded101b1b08c',
  38             'thumbnail': r're:^https?://.*\.jpg$',
  39             'release_date': '20210118',
  40         },
  41         'params': {
  42             'skip_download': True,
  43         },
  44     }]
  45
  46     def _real_extract(self, url):
  47         id = self._match_id(url)
  48         json_data = self._download_json(f'https://tv2play.hu/api/search/{id}', id)
  49
  50         if json_data['contentType'] == 'showpage':
  51             ribbon_ids = traverse_obj(json_data, ('pages', ..., 'tabs', ..., 'ribbonIds'), get_all=False, expected_type=list)
  52             entries = [self.url_result(f'https://tv2play.hu/szalag/{ribbon_id}',
  53                                        ie=TV2HuSeriesIE.ie_key(), video_id=ribbon_id) for ribbon_id in ribbon_ids]
  54             return self.playlist_result(entries, playlist_id=id)
  55         elif json_data['contentType'] != 'video':
  56             raise UnsupportedError(url)
  57
  58         video_id = str(json_data['id'])
  59         player_id = json_data.get('playerId')
  60         series_json = json_data.get('seriesInfo', {})
  61
  62         video_json_url = self._download_json(f'https://tv2play.hu/api/streaming-url?playerId={player_id}', video_id)['url']
  63         video_json = self._download_json(video_json_url, video_id)
  64         m3u8_url = self._proto_relative_url(traverse_obj(video_json, ('bitrates', 'hls')))
  65         formats, subtitles = self._extract_m3u8_formats_and_subtitles(m3u8_url, video_id)
  66
  67         return {
  68             'id': video_id,
  69             'title': json_data['title'],
  70             'series': json_data.get('seriesTitle'),
  71             'duration': json_data.get('length'),
  72             'description': json_data.get('description'),
  73             'thumbnail': 'https://tv2play.hu' + json_data.get('thumbnailUrl'),
  74             'release_date': json_data.get('uploadedAt').replace('.', ''),
  75             'season_number': series_json.get('seasonNr'),
  76             'episode_number': series_json.get('episodeNr'),
  77             'formats': formats,
  78             'subtitles': subtitles,
  79         }
  80
  81
  82 class TV2HuSeriesIE(InfoExtractor):
  83     IE_NAME = 'tv2playseries.hu'
  84     _VALID_URL = r'https?://(?:www\.)?tv2play\.hu/szalag/(?P<id>[^#&?]+)'
  85
  86     _TESTS = [{
  87         'url': 'https://tv2play.hu/szalag/59?rendezes=nepszeruseg',
  88         'playlist_mincount': 284,
  89         'info_dict': {
  90             'id': '59',
  91         }
  92     }]
  93
  94     def _real_extract(self, url):
  95         id = self._match_id(url)
  96         json_data = self._download_json(f'https://tv2play.hu/api/ribbons/{id}/0?size=100000', id)
  97         entries = []
  98         for card in json_data.get('cards', []):
  99             video_id = card.get('slug')
 100             if video_id:
 101                 entries.append(self.url_result(f'https://tv2play.hu/{video_id}',
 102                                                ie=TV2HuIE.ie_key(), video_id=video_id))
 103
 104         return self.playlist_result(entries, playlist_id=id)