yt_dlp/extractor/daum.py

   1 # coding: utf-8
   2
   3 from __future__ import unicode_literals
   4
   5 import itertools
   6
   7 from .common import InfoExtractor
   8 from ..compat import (
   9     compat_urllib_parse_unquote,
  10 )
  11 from ..utils import parse_qs
  12
  13
  14 class DaumBaseIE(InfoExtractor):
  15     _KAKAO_EMBED_BASE = 'http://tv.kakao.com/embed/player/cliplink/'
  16
  17
  18 class DaumIE(DaumBaseIE):
  19     _VALID_URL = r'https?://(?:(?:m\.)?tvpot\.daum\.net/v/|videofarm\.daum\.net/controller/player/VodPlayer\.swf\?vid=)(?P<id>[^?#&]+)'
  20     IE_NAME = 'daum.net'
  21
  22     _TESTS = [{
  23         'url': 'http://tvpot.daum.net/v/vab4dyeDBysyBssyukBUjBz',
  24         'info_dict': {
  25             'id': 'vab4dyeDBysyBssyukBUjBz',
  26             'ext': 'mp4',
  27             'title': '마크 헌트 vs 안토니오 실바',
  28             'description': 'Mark Hunt vs Antonio Silva',
  29             'upload_date': '20131217',
  30             'thumbnail': r're:^https?://.*\.(?:jpg|png)',
  31             'duration': 2117,
  32             'view_count': int,
  33             'comment_count': int,
  34             'uploader_id': 186139,
  35             'uploader': '콘간지',
  36             'timestamp': 1387310323,
  37         },
  38     }, {
  39         'url': 'http://m.tvpot.daum.net/v/65139429',
  40         'info_dict': {
  41             'id': '65139429',
  42             'ext': 'mp4',
  43             'title': '1297회, \'아빠 아들로 태어나길 잘 했어\' 민수, 감동의 눈물[아빠 어디가] 20150118',
  44             'description': 'md5:79794514261164ff27e36a21ad229fc5',
  45             'upload_date': '20150118',
  46             'thumbnail': r're:^https?://.*\.(?:jpg|png)',
  47             'duration': 154,
  48             'view_count': int,
  49             'comment_count': int,
  50             'uploader': 'MBC 예능',
  51             'uploader_id': 132251,
  52             'timestamp': 1421604228,
  53         },
  54     }, {
  55         'url': 'http://tvpot.daum.net/v/07dXWRka62Y%24',
  56         'only_matching': True,
  57     }, {
  58         'url': 'http://videofarm.daum.net/controller/player/VodPlayer.swf?vid=vwIpVpCQsT8%24&ref=',
  59         'info_dict': {
  60             'id': 'vwIpVpCQsT8$',
  61             'ext': 'flv',
  62             'title': '01-Korean War ( Trouble on the horizon )',
  63             'description': 'Korean War 01\r\nTrouble on the horizon\r\n전쟁의 먹구름',
  64             'upload_date': '20080223',
  65             'thumbnail': r're:^https?://.*\.(?:jpg|png)',
  66             'duration': 249,
  67             'view_count': int,
  68             'comment_count': int,
  69             'uploader': '까칠한 墮落始祖 황비홍님의',
  70             'uploader_id': 560824,
  71             'timestamp': 1203770745,
  72         },
  73     }, {
  74         # Requires dte_type=WEB (#9972)
  75         'url': 'http://tvpot.daum.net/v/s3794Uf1NZeZ1qMpGpeqeRU',
  76         'md5': 'a8917742069a4dd442516b86e7d66529',
  77         'info_dict': {
  78             'id': 's3794Uf1NZeZ1qMpGpeqeRU',
  79             'ext': 'mp4',
  80             'title': '러블리즈 - Destiny (나의 지구) (Lovelyz - Destiny)',
  81             'description': '러블리즈 - Destiny (나의 지구) (Lovelyz - Destiny)\r\n\r\n[쇼! 음악중심] 20160611, 507회',
  82             'upload_date': '20170129',
  83             'uploader': '쇼! 음악중심',
  84             'uploader_id': 2653210,
  85             'timestamp': 1485684628,
  86         },
  87     }]
  88
  89     def _real_extract(self, url):
  90         video_id = compat_urllib_parse_unquote(self._match_id(url))
  91         if not video_id.isdigit():
  92             video_id += '@my'
  93         return self.url_result(
  94             self._KAKAO_EMBED_BASE + video_id, 'Kakao', video_id)
  95
  96
  97 class DaumClipIE(DaumBaseIE):
  98     _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/(?:clip/ClipView.(?:do|tv)|mypot/View.do)\?.*?clipid=(?P<id>\d+)'
  99     IE_NAME = 'daum.net:clip'
 100     _URL_TEMPLATE = 'http://tvpot.daum.net/clip/ClipView.do?clipid=%s'
 101
 102     _TESTS = [{
 103         'url': 'http://tvpot.daum.net/clip/ClipView.do?clipid=52554690',
 104         'info_dict': {
 105             'id': '52554690',
 106             'ext': 'mp4',
 107             'title': 'DOTA 2GETHER 시즌2 6회 - 2부',
 108             'description': 'DOTA 2GETHER 시즌2 6회 - 2부',
 109             'upload_date': '20130831',
 110             'thumbnail': r're:^https?://.*\.(?:jpg|png)',
 111             'duration': 3868,
 112             'view_count': int,
 113             'uploader': 'GOMeXP',
 114             'uploader_id': 6667,
 115             'timestamp': 1377911092,
 116         },
 117     }, {
 118         'url': 'http://m.tvpot.daum.net/clip/ClipView.tv?clipid=54999425',
 119         'only_matching': True,
 120     }]
 121
 122     @classmethod
 123     def suitable(cls, url):
 124         return False if DaumPlaylistIE.suitable(url) or DaumUserIE.suitable(url) else super(DaumClipIE, cls).suitable(url)
 125
 126     def _real_extract(self, url):
 127         video_id = self._match_id(url)
 128         return self.url_result(
 129             self._KAKAO_EMBED_BASE + video_id, 'Kakao', video_id)
 130
 131
 132 class DaumListIE(InfoExtractor):
 133     def _get_entries(self, list_id, list_id_type):
 134         name = None
 135         entries = []
 136         for pagenum in itertools.count(1):
 137             list_info = self._download_json(
 138                 'http://tvpot.daum.net/mypot/json/GetClipInfo.do?size=48&init=true&order=date&page=%d&%s=%s' % (
 139                     pagenum, list_id_type, list_id), list_id, 'Downloading list info - %s' % pagenum)
 140
 141             entries.extend([
 142                 self.url_result(
 143                     'http://tvpot.daum.net/v/%s' % clip['vid'])
 144                 for clip in list_info['clip_list']
 145             ])
 146
 147             if not name:
 148                 name = list_info.get('playlist_bean', {}).get('name') or \
 149                     list_info.get('potInfo', {}).get('name')
 150
 151             if not list_info.get('has_more'):
 152                 break
 153
 154         return name, entries
 155
 156     def _check_clip(self, url, list_id):
 157         query_dict = parse_qs(url)
 158         if 'clipid' in query_dict:
 159             clip_id = query_dict['clipid'][0]
 160             if self.get_param('noplaylist'):
 161                 self.to_screen('Downloading just video %s because of --no-playlist' % clip_id)
 162                 return self.url_result(DaumClipIE._URL_TEMPLATE % clip_id, 'DaumClip')
 163             else:
 164                 self.to_screen('Downloading playlist %s - add --no-playlist to just download video' % list_id)
 165
 166
 167 class DaumPlaylistIE(DaumListIE):
 168     _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/mypot/(?:View\.do|Top\.tv)\?.*?playlistid=(?P<id>[0-9]+)'
 169     IE_NAME = 'daum.net:playlist'
 170     _URL_TEMPLATE = 'http://tvpot.daum.net/mypot/View.do?playlistid=%s'
 171
 172     _TESTS = [{
 173         'note': 'Playlist url with clipid',
 174         'url': 'http://tvpot.daum.net/mypot/View.do?playlistid=6213966&clipid=73806844',
 175         'info_dict': {
 176             'id': '6213966',
 177             'title': 'Woorissica Official',
 178         },
 179         'playlist_mincount': 181
 180     }, {
 181         'note': 'Playlist url with clipid - noplaylist',
 182         'url': 'http://tvpot.daum.net/mypot/View.do?playlistid=6213966&clipid=73806844',
 183         'info_dict': {
 184             'id': '73806844',
 185             'ext': 'mp4',
 186             'title': '151017 Airport',
 187             'upload_date': '20160117',
 188         },
 189         'params': {
 190             'noplaylist': True,
 191             'skip_download': True,
 192         }
 193     }]
 194
 195     @classmethod
 196     def suitable(cls, url):
 197         return False if DaumUserIE.suitable(url) else super(DaumPlaylistIE, cls).suitable(url)
 198
 199     def _real_extract(self, url):
 200         list_id = self._match_id(url)
 201
 202         clip_result = self._check_clip(url, list_id)
 203         if clip_result:
 204             return clip_result
 205
 206         name, entries = self._get_entries(list_id, 'playlistid')
 207
 208         return self.playlist_result(entries, list_id, name)
 209
 210
 211 class DaumUserIE(DaumListIE):
 212     _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/mypot/(?:View|Top)\.(?:do|tv)\?.*?ownerid=(?P<id>[0-9a-zA-Z]+)'
 213     IE_NAME = 'daum.net:user'
 214
 215     _TESTS = [{
 216         'url': 'http://tvpot.daum.net/mypot/View.do?ownerid=o2scDLIVbHc0',
 217         'info_dict': {
 218             'id': 'o2scDLIVbHc0',
 219             'title': '마이 리틀 텔레비전',
 220         },
 221         'playlist_mincount': 213
 222     }, {
 223         'url': 'http://tvpot.daum.net/mypot/View.do?ownerid=o2scDLIVbHc0&clipid=73801156',
 224         'info_dict': {
 225             'id': '73801156',
 226             'ext': 'mp4',
 227             'title': '[미공개] 김구라, 오만석이 부릅니다 \'오케피\' - 마이 리틀 텔레비전 20160116',
 228             'upload_date': '20160117',
 229             'description': 'md5:5e91d2d6747f53575badd24bd62b9f36'
 230         },
 231         'params': {
 232             'noplaylist': True,
 233             'skip_download': True,
 234         }
 235     }, {
 236         'note': 'Playlist url has ownerid and playlistid, playlistid takes precedence',
 237         'url': 'http://tvpot.daum.net/mypot/View.do?ownerid=o2scDLIVbHc0&playlistid=6196631',
 238         'info_dict': {
 239             'id': '6196631',
 240             'title': '마이 리틀 텔레비전 - 20160109',
 241         },
 242         'playlist_count': 11
 243     }, {
 244         'url': 'http://tvpot.daum.net/mypot/Top.do?ownerid=o2scDLIVbHc0',
 245         'only_matching': True,
 246     }, {
 247         'url': 'http://m.tvpot.daum.net/mypot/Top.tv?ownerid=45x1okb1If50&playlistid=3569733',
 248         'only_matching': True,
 249     }]
 250
 251     def _real_extract(self, url):
 252         list_id = self._match_id(url)
 253
 254         clip_result = self._check_clip(url, list_id)
 255         if clip_result:
 256             return clip_result
 257
 258         query_dict = parse_qs(url)
 259         if 'playlistid' in query_dict:
 260             playlist_id = query_dict['playlistid'][0]
 261             return self.url_result(DaumPlaylistIE._URL_TEMPLATE % playlist_id, 'DaumPlaylist')
 262
 263         name, entries = self._get_entries(list_id, 'ownerid')
 264
 265         return self.playlist_result(entries, list_id, name)