yt_dlp/extractor/idolplus.py

   1 from .common import InfoExtractor
   2 from ..utils import traverse_obj, try_call, url_or_none
   3
   4
   5 class IdolPlusIE(InfoExtractor):
   6     _VALID_URL = r'https?://(?:www\.)?idolplus\.com/z[us]/(?:concert/|contents/?\?(?:[^#]+&)?albumId=)(?P<id>\w+)'
   7     _TESTS = [{
   8         'url': 'https://idolplus.com/zs/contents?albumId=M012077298PPV00',
   9         'md5': '2ace3f4661c943a2f7e79f0b88cea1e7',
  10         'info_dict': {
  11             'id': 'M012077298PPV00',
  12             'ext': 'mp4',
  13             'title': '[MultiCam] Aegyo on Top of Aegyo (IZ*ONE EATING TRIP)',
  14             'release_date': '20200707',
  15             'formats': 'count:65',
  16         },
  17         'params': {'format': '532-KIM_MINJU'},
  18     }, {
  19         'url': 'https://idolplus.com/zs/contents?albumId=M01232H058PPV00&catId=E9TX5',
  20         'info_dict': {
  21             'id': 'M01232H058PPV00',
  22             'ext': 'mp4',
  23             'title': 'YENA (CIRCLE CHART MUSIC AWARDS 2022 RED CARPET)',
  24             'release_date': '20230218',
  25             'formats': 'count:5',
  26         },
  27         'params': {'skip_download': 'm3u8'},
  28     }, {
  29         # live stream
  30         'url': 'https://idolplus.com/zu/contents?albumId=M012323174PPV00',
  31         'info_dict': {
  32             'id': 'M012323174PPV00',
  33             'ext': 'mp4',
  34             'title': 'Hanteo Music Awards 2022 DAY2',
  35             'release_date': '20230211',
  36             'formats': 'count:5',
  37         },
  38         'params': {'skip_download': 'm3u8'},
  39     }, {
  40         'url': 'https://idolplus.com/zs/concert/M012323039PPV00',
  41         'info_dict': {
  42             'id': 'M012323039PPV00',
  43             'ext': 'mp4',
  44             'title': 'CIRCLE CHART MUSIC AWARDS 2022',
  45             'release_date': '20230218',
  46             'formats': 'count:5',
  47         },
  48         'params': {'skip_download': 'm3u8'},
  49     }]
  50
  51     def _real_extract(self, url):
  52         video_id = self._match_id(url)
  53         data_list = traverse_obj(self._download_json(
  54             'https://idolplus.com/api/zs/viewdata/ruleset/build', video_id,
  55             headers={'App_type': 'web', 'Country_Code': 'KR'}, query={
  56                 'rulesetId': 'contents',
  57                 'albumId': video_id,
  58                 'distribute': 'PRD',
  59                 'loggedIn': 'false',
  60                 'region': 'zs',
  61                 'countryGroup': '00010',
  62                 'lang': 'en',
  63                 'saId': '999999999998',
  64             }), ('data', 'viewData', ...))
  65
  66         player_data = {}
  67         while data_list:
  68             player_data = data_list.pop()
  69             if traverse_obj(player_data, 'type') == 'player':
  70                 break
  71             elif traverse_obj(player_data, ('dataList', ...)):
  72                 data_list += player_data['dataList']
  73
  74         formats = self._extract_m3u8_formats(traverse_obj(player_data, (
  75             'vodPlayerList', 'vodProfile', 0, 'vodServer', 0, 'video_url', {url_or_none})), video_id)
  76
  77         subtitles = {}
  78         for caption in traverse_obj(player_data, ('vodPlayerList', 'caption')) or []:
  79             subtitles.setdefault(caption.get('lang') or 'und', []).append({
  80                 'url': caption.get('smi_url'),
  81                 'ext': 'vtt',
  82             })
  83
  84         # Add member multicams as alternative formats
  85         if (traverse_obj(player_data, ('detail', 'has_cuesheet')) == 'Y'
  86                 and traverse_obj(player_data, ('detail', 'is_omni_member')) == 'Y'):
  87             cuesheet = traverse_obj(self._download_json(
  88                 'https://idolplus.com/gapi/contents/v1.0/content/cuesheet', video_id,
  89                 'Downloading JSON metadata for member multicams',
  90                 headers={'App_type': 'web', 'Country_Code': 'KR'}, query={
  91                     'ALBUM_ID': video_id,
  92                     'COUNTRY_GRP': '00010',
  93                     'LANG': 'en',
  94                     'SA_ID': '999999999998',
  95                     'COUNTRY_CODE': 'KR',
  96                 }), ('data', 'cuesheet_item', 0))
  97
  98             for member in traverse_obj(cuesheet, ('members', ...)):
  99                 index = try_call(lambda: int(member['omni_view_index']) - 1)
 100                 member_video_url = traverse_obj(cuesheet, ('omni_view', index, 'cdn_url', 0, 'url', {url_or_none}))
 101                 if not member_video_url:
 102                     continue
 103                 member_formats = self._extract_m3u8_formats(
 104                     member_video_url, video_id, note=f'Downloading m3u8 for multicam {member["name"]}')
 105                 for mf in member_formats:
 106                     mf['format_id'] = f'{mf["format_id"]}-{member["name"].replace(" ", "_")}'
 107                 formats.extend(member_formats)
 108
 109         return {
 110             'id': video_id,
 111             'title': traverse_obj(player_data, ('detail', 'albumName')),
 112             'formats': formats,
 113             'subtitles': subtitles,
 114             'release_date': traverse_obj(player_data, ('detail', 'broadcastDate')),
 115         }