yt_dlp/extractor/zaiko.py

   1 import base64
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     ExtractorError,
   6     extract_attributes,
   7     int_or_none,
   8     str_or_none,
   9     traverse_obj,
  10     try_call,
  11     unescapeHTML,
  12     url_basename,
  13     url_or_none,
  14 )
  15
  16
  17 class ZaikoBaseIE(InfoExtractor):
  18     def _download_real_webpage(self, url, video_id):
  19         webpage, urlh = self._download_webpage_handle(url, video_id)
  20         final_url = urlh.url
  21         if 'zaiko.io/login' in final_url:
  22             self.raise_login_required()
  23         elif '/_buy/' in final_url:
  24             raise ExtractorError('Your account does not have tickets to this event', expected=True)
  25         return webpage
  26
  27     def _parse_vue_element_attr(self, name, string, video_id):
  28         page_elem = self._search_regex(rf'(<{name}[^>]+>)', string, name)
  29         attrs = {}
  30         for key, value in extract_attributes(page_elem).items():
  31             if key.startswith(':'):
  32                 attrs[key[1:]] = self._parse_json(
  33                     value, video_id, transform_source=unescapeHTML, fatal=False)
  34         return attrs
  35
  36
  37 class ZaikoIE(ZaikoBaseIE):
  38     _VALID_URL = r'https?://(?:[\w-]+\.)?zaiko\.io/event/(?P<id>\d+)/stream(?:/\d+)+'
  39     _TESTS = [{
  40         'url': 'https://zaiko.io/event/324868/stream/20571/20571',
  41         'info_dict': {
  42             'id': '324868',
  43             'ext': 'mp4',
  44             'title': 'ZAIKO STREAMING TEST',
  45             'alt_title': '[VOD] ZAIKO STREAMING TEST_20210603(Do Not Delete)',
  46             'uploader_id': '454',
  47             'uploader': 'ZAIKO ZERO',
  48             'release_timestamp': 1583809200,
  49             'thumbnail': r're:^https://[\w.-]+/\w+/\w+',
  50             'thumbnails': 'maxcount:2',
  51             'release_date': '20200310',
  52             'categories': ['Tech House'],
  53             'live_status': 'was_live',
  54         },
  55         'params': {'skip_download': 'm3u8'},
  56         'skip': 'Your account does not have tickets to this event',
  57     }]
  58
  59     def _real_extract(self, url):
  60         video_id = self._match_id(url)
  61
  62         webpage = self._download_real_webpage(url, video_id)
  63         stream_meta = self._parse_vue_element_attr('stream-page', webpage, video_id)
  64
  65         player_page = self._download_webpage(
  66             stream_meta['stream-access']['video_source'], video_id,
  67             'Downloading player page', headers={'referer': 'https://zaiko.io/'})
  68         player_meta = self._parse_vue_element_attr('player', player_page, video_id)
  69         status = traverse_obj(player_meta, ('initial_event_info', 'status', {str}))
  70         live_status, msg, expected = {
  71             'vod': ('was_live', 'No VOD stream URL was found', False),
  72             'archiving': ('post_live', 'Event VOD is still being processed', True),
  73             'deleting': ('post_live', 'This event has ended', True),
  74             'deleted': ('post_live', 'This event has ended', True),
  75             'error': ('post_live', 'This event has ended', True),
  76             'disconnected': ('post_live', 'Stream has been disconnected', True),
  77             'live_to_disconnected': ('post_live', 'Stream has been disconnected', True),
  78             'live': ('is_live', 'No livestream URL found was found', False),
  79             'waiting': ('is_upcoming', 'Live event has not yet started', True),
  80             'cancelled': ('not_live', 'Event has been cancelled', True),
  81         }.get(status) or ('not_live', f'Unknown event status "{status}"', False)
  82
  83         stream_url = traverse_obj(player_meta, ('initial_event_info', 'endpoint', {url_or_none}))
  84         formats = self._extract_m3u8_formats(
  85             stream_url, video_id, live=True, fatal=False) if stream_url else []
  86         if not formats:
  87             self.raise_no_formats(msg, expected=expected)
  88
  89         thumbnail_urls = [
  90             traverse_obj(player_meta, ('initial_event_info', 'poster_url')),
  91             self._og_search_thumbnail(self._download_webpage(
  92                 f'https://zaiko.io/event/{video_id}', video_id, 'Downloading event page', fatal=False) or ''),
  93         ]
  94
  95         return {
  96             'id': video_id,
  97             'formats': formats,
  98             'live_status': live_status,
  99             **traverse_obj(stream_meta, {
 100                 'title': ('event', 'name', {str}),
 101                 'uploader': ('profile', 'name', {str}),
 102                 'uploader_id': ('profile', 'id', {str_or_none}),
 103                 'release_timestamp': ('stream', 'start', 'timestamp', {int_or_none}),
 104                 'categories': ('event', 'genres', ..., {lambda x: x or None}),
 105             }),
 106             **traverse_obj(player_meta, ('initial_event_info', {
 107                 'alt_title': ('title', {str}),
 108             })),
 109             'thumbnails': [{'url': url, 'id': url_basename(url)} for url in thumbnail_urls if url_or_none(url)]
 110         }
 111
 112
 113 class ZaikoETicketIE(ZaikoBaseIE):
 114     _VALID_URL = r'https?://(?:www.)?zaiko\.io/account/eticket/(?P<id>[\w=-]{49})'
 115     _TESTS = [{
 116         'url': 'https://zaiko.io/account/eticket/TZjMwMzQ2Y2EzMXwyMDIzMDYwNzEyMTMyNXw1MDViOWU2Mw==',
 117         'playlist_count': 1,
 118         'info_dict': {
 119             'id': 'f30346ca31-20230607121325-505b9e63',
 120             'title': 'ZAIKO STREAMING TEST',
 121             'thumbnail': 'https://media.zkocdn.net/pf_1/1_3wdyjcjyupseatkwid34u',
 122         },
 123         'skip': 'Only available with the ticketholding account',
 124     }]
 125
 126     def _real_extract(self, url):
 127         ticket_id = self._match_id(url)
 128         ticket_id = try_call(
 129             lambda: base64.urlsafe_b64decode(ticket_id[1:]).decode().replace('|', '-')) or ticket_id
 130
 131         webpage = self._download_real_webpage(url, ticket_id)
 132         eticket = self._parse_vue_element_attr('eticket', webpage, ticket_id)
 133
 134         return self.playlist_result(
 135             [self.url_result(stream, ZaikoIE) for stream in traverse_obj(eticket, ('streams', ..., 'url'))],
 136             ticket_id, **traverse_obj(eticket, ('ticket-details', {
 137                 'title': 'event_name',
 138                 'thumbnail': 'event_img_url',
 139             })))