yt_dlp/extractor/damtomo.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..compat import compat_str
   5 from ..utils import ExtractorError, clean_html, int_or_none, try_get, unified_strdate
   6
   7
   8 class DamtomoBaseIE(InfoExtractor):
   9     def _real_extract(self, url):
  10         video_id = self._match_id(url)
  11         webpage, handle = self._download_webpage_handle(self._WEBPAGE_URL_TMPL % video_id, video_id, encoding='sjis')
  12
  13         if handle.url == 'https://www.clubdam.com/sorry/':
  14             raise ExtractorError('You are rate-limited. Try again later.', expected=True)
  15         if '<h2>予期せぬエラーが発生しました。</h2>' in webpage:
  16             raise ExtractorError('There is an error on server-side. Try again later.', expected=True)
  17
  18         description = self._search_regex(r'(?m)<div id="public_comment">\s*<p>\s*([^<]*?)\s*</p>', webpage, 'description', default=None)
  19         uploader_id = self._search_regex(r'<a href="https://www\.clubdam\.com/app/damtomo/member/info/Profile\.do\?damtomoId=([^"]+)"', webpage, 'uploader_id', default=None)
  20
  21         data_dict = {
  22             mobj.group('class'): re.sub(r'\s+', ' ', clean_html(mobj.group('value')))
  23             for mobj in re.finditer(r'(?s)<(p|div)\s+class="(?P<class>[^" ]+?)">(?P<value>.+?)</\1>', webpage)}
  24
  25         # since videos do not have title, give the name of song instead
  26         data_dict['user_name'] = re.sub(r'\s*さん\s*$', '', data_dict['user_name'])
  27         title = data_dict.get('song_title')
  28
  29         stream_tree = self._download_xml(
  30             self._DKML_XML_URL % video_id, video_id, note='Requesting stream information', encoding='sjis',
  31             # doing this has no problem since there is no character outside ASCII,
  32             # and never likely to happen in the future
  33             transform_source=lambda x: re.sub(r'\s*encoding="[^"]+?"', '', x))
  34         m3u8_url = try_get(stream_tree, lambda x: x.find(
  35             './/d:streamingUrl', {'d': self._DKML_XML_NS}).text.strip(), compat_str)
  36         if not m3u8_url:
  37             raise ExtractorError('Failed to obtain m3u8 URL')
  38         formats = self._extract_m3u8_formats(m3u8_url, video_id, ext='mp4')
  39
  40         return {
  41             'id': video_id,
  42             'title': title,
  43             'uploader_id': uploader_id,
  44             'description': description,
  45             'uploader': data_dict.get('user_name'),
  46             'upload_date': unified_strdate(self._search_regex(r'(\d{4}/\d{2}/\d{2})', data_dict.get('date'), 'upload_date', default=None)),
  47             'view_count': int_or_none(self._search_regex(r'(\d+)', data_dict['audience'], 'view_count', default=None)),
  48             'like_count': int_or_none(self._search_regex(r'(\d+)', data_dict['nice'], 'like_count', default=None)),
  49             'track': title,
  50             'artist': data_dict.get('song_artist'),
  51             'formats': formats,
  52         }
  53
  54
  55 class DamtomoVideoIE(DamtomoBaseIE):
  56     IE_NAME = 'damtomo:video'
  57     _VALID_URL = r'https?://(?:www\.)?clubdam\.com/app/damtomo/(?:SP/)?karaokeMovie/StreamingDkm\.do\?karaokeMovieId=(?P<id>\d+)'
  58     _WEBPAGE_URL_TMPL = 'https://www.clubdam.com/app/damtomo/karaokeMovie/StreamingDkm.do?karaokeMovieId=%s'
  59     _DKML_XML_URL = 'https://www.clubdam.com/app/damtomo/karaokeMovie/GetStreamingDkmUrlXML.do?movieSelectFlg=2&karaokeMovieId=%s'
  60     _DKML_XML_NS = 'https://www.clubdam.com/app/damtomo/karaokeMovie/GetStreamingDkmUrlXML'
  61     _TESTS = [{
  62         'url': 'https://www.clubdam.com/app/damtomo/karaokeMovie/StreamingDkm.do?karaokeMovieId=2414316',
  63         'info_dict': {
  64             'id': '2414316',
  65             'title': 'Get Wild',
  66             'uploader': 'Ｋドロン',
  67             'uploader_id': 'ODk5NTQwMzQ',
  68             'track': 'Get Wild',
  69             'artist': 'TM NETWORK(TMN)',
  70             'upload_date': '20201226',
  71         }
  72     }]
  73
  74
  75 class DamtomoRecordIE(DamtomoBaseIE):
  76     IE_NAME = 'damtomo:record'
  77     _VALID_URL = r'https?://(?:www\.)?clubdam\.com/app/damtomo/(?:SP/)?karaokePost/StreamingKrk\.do\?karaokeContributeId=(?P<id>\d+)'
  78     _WEBPAGE_URL_TMPL = 'https://www.clubdam.com/app/damtomo/karaokePost/StreamingKrk.do?karaokeContributeId=%s'
  79     _DKML_XML_URL = 'https://www.clubdam.com/app/damtomo/karaokePost/GetStreamingKrkUrlXML.do?karaokeContributeId=%s'
  80     _DKML_XML_NS = 'https://www.clubdam.com/app/damtomo/karaokePost/GetStreamingKrkUrlXML'
  81     _TESTS = [{
  82         'url': 'https://www.clubdam.com/app/damtomo/karaokePost/StreamingKrk.do?karaokeContributeId=27376862',
  83         'info_dict': {
  84             'id': '27376862',
  85             'title': 'イカSUMMER [良音]',
  86             'uploader': 'ＮＡＮＡ',
  87             'uploader_id': 'MzAyMDExNTY',
  88             'upload_date': '20210721',
  89             'view_count': 4,
  90             'like_count': 1,
  91             'track': 'イカSUMMER [良音]',
  92             'artist': 'ORANGE RANGE',
  93         }
  94     }, {
  95         'url': 'https://www.clubdam.com/app/damtomo/karaokePost/StreamingKrk.do?karaokeContributeId=27489418',
  96         'info_dict': {
  97             'id': '27489418',
  98             'title': '心みだれて〜say it with flowers〜(生音)',
  99             'uploader_id': 'NjI1MjI2MjU',
 100             'description': 'やっぱりキーを下げて正解だった感じ。リベンジ成功ということで。',
 101             'uploader': '箱の「中の人」',
 102             'upload_date': '20210815',
 103             'view_count': 5,
 104             'like_count': 3,
 105             'track': '心みだれて〜say it with flowers〜(生音)',
 106             'artist': '小林明子',
 107         }
 108     }]