yt_dlp/extractor/daum.py

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