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