yt_dlp/extractor/yinyuetai.py

   1 from .common import InfoExtractor
   2 from ..utils import ExtractorError
   3
   4
   5 class YinYueTaiIE(InfoExtractor):
   6     IE_NAME = 'yinyuetai:video'
   7     IE_DESC = '音悦Tai'
   8     _VALID_URL = r'https?://v\.yinyuetai\.com/video(?:/h5)?/(?P<id>[0-9]+)'
   9     _TESTS = [{
  10         'url': 'http://v.yinyuetai.com/video/2322376',
  11         'md5': '6e3abe28d38e3a54b591f9f040595ce0',
  12         'info_dict': {
  13             'id': '2322376',
  14             'ext': 'mp4',
  15             'title': '少女时代_PARTY_Music Video Teaser',
  16             'creator': '少女时代',
  17             'duration': 25,
  18             'thumbnail': r're:^https?://.*\.jpg$',
  19         },
  20     }, {
  21         'url': 'http://v.yinyuetai.com/video/h5/2322376',
  22         'only_matching': True,
  23     }]
  24
  25     def _real_extract(self, url):
  26         video_id = self._match_id(url)
  27
  28         info = self._download_json(
  29             'http://ext.yinyuetai.com/main/get-h-mv-info?json=true&videoId=%s' % video_id, video_id,
  30             'Downloading mv info')['videoInfo']['coreVideoInfo']
  31
  32         if info['error']:
  33             raise ExtractorError(info['errorMsg'], expected=True)
  34
  35         formats = [{
  36             'url': format_info['videoUrl'],
  37             'format_id': format_info['qualityLevel'],
  38             'format': format_info.get('qualityLevelName'),
  39             'filesize': format_info.get('fileSize'),
  40             # though URLs ends with .flv, the downloaded files are in fact mp4
  41             'ext': 'mp4',
  42             'tbr': format_info.get('bitrate'),
  43         } for format_info in info['videoUrlModels']]
  44
  45         return {
  46             'id': video_id,
  47             'title': info['videoName'],
  48             'thumbnail': info.get('bigHeadImage'),
  49             'creator': info.get('artistNames'),
  50             'duration': info.get('duration'),
  51             'formats': formats,
  52         }