Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/melonvod.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     int_or_none,
   7     urljoin,
   8 )
   9
  10
  11 class MelonVODIE(InfoExtractor):
  12     _VALID_URL = r'https?://vod\.melon\.com/video/detail2\.html?\?.*?mvId=(?P<id>[0-9]+)'
  13     _TEST = {
  14         'url': 'http://vod.melon.com/video/detail2.htm?mvId=50158734',
  15         'info_dict': {
  16             'id': '50158734',
  17             'ext': 'mp4',
  18             'title': "Jessica 'Wonderland' MV Making Film",
  19             'thumbnail': r're:^https?://.*\.jpg$',
  20             'artist': 'Jessica (제시카)',
  21             'upload_date': '20161212',
  22             'duration': 203,
  23         },
  24         'params': {
  25             'skip_download': 'm3u8 download',
  26         }
  27     }
  28
  29     def _real_extract(self, url):
  30         video_id = self._match_id(url)
  31
  32         play_info = self._download_json(
  33             'http://vod.melon.com/video/playerInfo.json', video_id,
  34             note='Downloading player info JSON', query={'mvId': video_id})
  35
  36         title = play_info['mvInfo']['MVTITLE']
  37
  38         info = self._download_json(
  39             'http://vod.melon.com/delivery/streamingInfo.json', video_id,
  40             note='Downloading streaming info JSON',
  41             query={
  42                 'contsId': video_id,
  43                 'contsType': 'VIDEO',
  44             })
  45
  46         stream_info = info['streamingInfo']
  47
  48         formats = self._extract_m3u8_formats(
  49             stream_info['encUrl'], video_id, 'mp4', m3u8_id='hls')
  50         self._sort_formats(formats)
  51
  52         artist_list = play_info.get('artistList')
  53         artist = None
  54         if isinstance(artist_list, list):
  55             artist = ', '.join(
  56                 [a['ARTISTNAMEWEBLIST']
  57                  for a in artist_list if a.get('ARTISTNAMEWEBLIST')])
  58
  59         thumbnail = urljoin(info.get('staticDomain'), stream_info.get('imgPath'))
  60
  61         duration = int_or_none(stream_info.get('playTime'))
  62         upload_date = stream_info.get('mvSvcOpenDt', '')[:8] or None
  63
  64         return {
  65             'id': video_id,
  66             'title': title,
  67             'artist': artist,
  68             'thumbnail': thumbnail,
  69             'upload_date': upload_date,
  70             'duration': duration,
  71             'formats': formats
  72         }