yt_dlp/extractor/melonvod.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     int_or_none,
   4     urljoin,
   5 )
   6
   7
   8 class MelonVODIE(InfoExtractor):
   9     _VALID_URL = r'https?://vod\.melon\.com/video/detail2\.html?\?.*?mvId=(?P<id>[0-9]+)'
  10     _TEST = {
  11         'url': 'http://vod.melon.com/video/detail2.htm?mvId=50158734',
  12         'info_dict': {
  13             'id': '50158734',
  14             'ext': 'mp4',
  15             'title': "Jessica 'Wonderland' MV Making Film",
  16             'thumbnail': r're:^https?://.*\.jpg$',
  17             'artist': 'Jessica (제시카)',
  18             'upload_date': '20161212',
  19             'duration': 203,
  20         },
  21         'params': {
  22             'skip_download': 'm3u8 download',
  23         },
  24     }
  25
  26     def _real_extract(self, url):
  27         video_id = self._match_id(url)
  28
  29         play_info = self._download_json(
  30             'http://vod.melon.com/video/playerInfo.json', video_id,
  31             note='Downloading player info JSON', query={'mvId': video_id})
  32
  33         title = play_info['mvInfo']['MVTITLE']
  34
  35         info = self._download_json(
  36             'http://vod.melon.com/delivery/streamingInfo.json', video_id,
  37             note='Downloading streaming info JSON',
  38             query={
  39                 'contsId': video_id,
  40                 'contsType': 'VIDEO',
  41             })
  42
  43         stream_info = info['streamingInfo']
  44
  45         formats = self._extract_m3u8_formats(
  46             stream_info['encUrl'], video_id, 'mp4', m3u8_id='hls')
  47
  48         artist_list = play_info.get('artistList')
  49         artist = None
  50         if isinstance(artist_list, list):
  51             artist = ', '.join(
  52                 [a['ARTISTNAMEWEBLIST']
  53                  for a in artist_list if a.get('ARTISTNAMEWEBLIST')])
  54
  55         thumbnail = urljoin(info.get('staticDomain'), stream_info.get('imgPath'))
  56
  57         duration = int_or_none(stream_info.get('playTime'))
  58         upload_date = stream_info.get('mvSvcOpenDt', '')[:8] or None
  59
  60         return {
  61             'id': video_id,
  62             'title': title,
  63             'artist': artist,
  64             'thumbnail': thumbnail,
  65             'upload_date': upload_date,
  66             'duration': duration,
  67             'formats': formats,
  68         }