X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fdaum.py;h=13709557754afc133c2654e202d52c32d5d85d66;hb=HEAD;hp=cf158c42d6a04be6fb3c060a2d3f6187aa06f85a;hpb=33cc74eeebe9355f38145d0d6ddc84e5a2dc1af8;p=youtube-dl diff --git a/youtube_dl/extractor/daum.py b/youtube_dl/extractor/daum.py index cf158c42d..137095577 100644 --- a/youtube_dl/extractor/daum.py +++ b/youtube_dl/extractor/daum.py @@ -1,24 +1,22 @@ -# encoding: utf-8 +# coding: utf-8 from __future__ import unicode_literals -import re import itertools from .common import InfoExtractor from ..compat import ( - compat_urllib_parse, + compat_parse_qs, compat_urllib_parse_unquote, + compat_urlparse, ) -from ..utils import ( - int_or_none, - str_to_int, - xpath_text, - unescapeHTML, -) -class DaumIE(InfoExtractor): +class DaumBaseIE(InfoExtractor): + _KAKAO_EMBED_BASE = 'http://tv.kakao.com/embed/player/cliplink/' + + +class DaumIE(DaumBaseIE): _VALID_URL = r'https?://(?:(?:m\.)?tvpot\.daum\.net/v/|videofarm\.daum\.net/controller/player/VodPlayer\.swf\?vid=)(?P[^?#&]+)' IE_NAME = 'daum.net' @@ -30,23 +28,29 @@ class DaumIE(InfoExtractor): 'title': '마크 헌트 vs 안토니오 실바', 'description': 'Mark Hunt vs Antonio Silva', 'upload_date': '20131217', - 'thumbnail': 're:^https?://.*\.(?:jpg|png)', + 'thumbnail': r're:^https?://.*\.(?:jpg|png)', 'duration': 2117, 'view_count': int, 'comment_count': int, + 'uploader_id': 186139, + 'uploader': '콘간지', + 'timestamp': 1387310323, }, }, { 'url': 'http://m.tvpot.daum.net/v/65139429', 'info_dict': { 'id': '65139429', 'ext': 'mp4', - 'title': 'md5:a100d65d09cec246d8aa9bde7de45aed', + 'title': '1297회, \'아빠 아들로 태어나길 잘 했어\' 민수, 감동의 눈물[아빠 어디가] 20150118', 'description': 'md5:79794514261164ff27e36a21ad229fc5', - 'upload_date': '20150604', - 'thumbnail': 're:^https?://.*\.(?:jpg|png)', + 'upload_date': '20150118', + 'thumbnail': r're:^https?://.*\.(?:jpg|png)', 'duration': 154, 'view_count': int, 'comment_count': int, + 'uploader': 'MBC 예능', + 'uploader_id': 132251, + 'timestamp': 1421604228, }, }, { 'url': 'http://tvpot.daum.net/v/07dXWRka62Y%24', @@ -57,66 +61,44 @@ class DaumIE(InfoExtractor): 'id': 'vwIpVpCQsT8$', 'ext': 'flv', 'title': '01-Korean War ( Trouble on the horizon )', - 'description': '\nKorean War 01\nTrouble on the horizon\n전쟁의 먹구름', + 'description': 'Korean War 01\r\nTrouble on the horizon\r\n전쟁의 먹구름', 'upload_date': '20080223', - 'thumbnail': 're:^https?://.*\.(?:jpg|png)', + 'thumbnail': r're:^https?://.*\.(?:jpg|png)', 'duration': 249, 'view_count': int, 'comment_count': int, + 'uploader': '까칠한 墮落始祖 황비홍님의', + 'uploader_id': 560824, + 'timestamp': 1203770745, + }, + }, { + # Requires dte_type=WEB (#9972) + 'url': 'http://tvpot.daum.net/v/s3794Uf1NZeZ1qMpGpeqeRU', + 'md5': 'a8917742069a4dd442516b86e7d66529', + 'info_dict': { + 'id': 's3794Uf1NZeZ1qMpGpeqeRU', + 'ext': 'mp4', + 'title': '러블리즈 - Destiny (나의 지구) (Lovelyz - Destiny)', + 'description': '러블리즈 - Destiny (나의 지구) (Lovelyz - Destiny)\r\n\r\n[쇼! 음악중심] 20160611, 507회', + 'upload_date': '20170129', + 'uploader': '쇼! 음악중심', + 'uploader_id': 2653210, + 'timestamp': 1485684628, }, }] def _real_extract(self, url): video_id = compat_urllib_parse_unquote(self._match_id(url)) - query = compat_urllib_parse.urlencode({'vid': video_id}) - movie_data = self._download_json( - 'http://videofarm.daum.net/controller/api/closed/v1_2/IntegratedMovieData.json?' + query, - video_id, 'Downloading video formats info') - - # For urls like http://m.tvpot.daum.net/v/65139429, where the video_id is really a clipid - if not movie_data.get('output_list', {}).get('output_list') and re.match(r'^\d+$', video_id): - return self.url_result('http://tvpot.daum.net/clip/ClipView.do?clipid=%s' % video_id) - - info = self._download_xml( - 'http://tvpot.daum.net/clip/ClipInfoXml.do?' + query, video_id, - 'Downloading video info') - - formats = [] - for format_el in movie_data['output_list']['output_list']: - profile = format_el['profile'] - format_query = compat_urllib_parse.urlencode({ - 'vid': video_id, - 'profile': profile, - }) - url_doc = self._download_xml( - 'http://videofarm.daum.net/controller/api/open/v1_2/MovieLocation.apixml?' + format_query, - video_id, note='Downloading video data for %s format' % profile) - format_url = url_doc.find('result/url').text - formats.append({ - 'url': format_url, - 'format_id': profile, - 'width': int_or_none(format_el.get('width')), - 'height': int_or_none(format_el.get('height')), - 'filesize': int_or_none(format_el.get('filesize')), - }) - self._sort_formats(formats) - - return { - 'id': video_id, - 'title': info.find('TITLE').text, - 'formats': formats, - 'thumbnail': xpath_text(info, 'THUMB_URL'), - 'description': xpath_text(info, 'CONTENTS'), - 'duration': int_or_none(xpath_text(info, 'DURATION')), - 'upload_date': info.find('REGDTTM').text[:8], - 'view_count': str_to_int(xpath_text(info, 'PLAY_CNT')), - 'comment_count': str_to_int(xpath_text(info, 'COMMENT_CNT')), - } + if not video_id.isdigit(): + video_id += '@my' + return self.url_result( + self._KAKAO_EMBED_BASE + video_id, 'Kakao', video_id) -class DaumClipIE(InfoExtractor): +class DaumClipIE(DaumBaseIE): _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/(?:clip/ClipView.(?:do|tv)|mypot/View.do)\?.*?clipid=(?P\d+)' IE_NAME = 'daum.net:clip' + _URL_TEMPLATE = 'http://tvpot.daum.net/clip/ClipView.do?clipid=%s' _TESTS = [{ 'url': 'http://tvpot.daum.net/clip/ClipView.do?clipid=52554690', @@ -126,33 +108,26 @@ class DaumClipIE(InfoExtractor): 'title': 'DOTA 2GETHER 시즌2 6회 - 2부', 'description': 'DOTA 2GETHER 시즌2 6회 - 2부', 'upload_date': '20130831', - 'thumbnail': 're:^https?://.*\.(?:jpg|png)', + 'thumbnail': r're:^https?://.*\.(?:jpg|png)', 'duration': 3868, 'view_count': int, + 'uploader': 'GOMeXP', + 'uploader_id': 6667, + 'timestamp': 1377911092, }, }, { 'url': 'http://m.tvpot.daum.net/clip/ClipView.tv?clipid=54999425', 'only_matching': True, }] + @classmethod + def suitable(cls, url): + return False if DaumPlaylistIE.suitable(url) or DaumUserIE.suitable(url) else super(DaumClipIE, cls).suitable(url) + def _real_extract(self, url): video_id = self._match_id(url) - clip_info = self._download_json( - 'http://tvpot.daum.net/mypot/json/GetClipInfo.do?clipid=%s' % video_id, - video_id, 'Downloading clip info')['clip_bean'] - - return { - '_type': 'url_transparent', - 'id': video_id, - 'url': 'http://tvpot.daum.net/v/%s' % clip_info['vid'], - 'title': unescapeHTML(clip_info['title']), - 'thumbnail': clip_info.get('thumb_url'), - 'description': clip_info.get('contents'), - 'duration': int_or_none(clip_info.get('duration')), - 'upload_date': clip_info.get('up_date')[:8], - 'view_count': int_or_none(clip_info.get('play_count')), - 'ie_key': 'Daum', - } + return self.url_result( + self._KAKAO_EMBED_BASE + video_id, 'Kakao', video_id) class DaumListIE(InfoExtractor): @@ -162,8 +137,8 @@ class DaumListIE(InfoExtractor): for pagenum in itertools.count(1): list_info = self._download_json( 'http://tvpot.daum.net/mypot/json/GetClipInfo.do?size=48&init=true&order=date&page=%d&%s=%s' % ( - pagenum, list_id_type, list_id), list_id,'Downloading list info - %s' % pagenum) - + pagenum, list_id_type, list_id), list_id, 'Downloading list info - %s' % pagenum) + entries.extend([ self.url_result( 'http://tvpot.daum.net/v/%s' % clip['vid']) @@ -179,10 +154,21 @@ class DaumListIE(InfoExtractor): return name, entries + def _check_clip(self, url, list_id): + query_dict = compat_parse_qs(compat_urlparse.urlparse(url).query) + if 'clipid' in query_dict: + clip_id = query_dict['clipid'][0] + if self._downloader.params.get('noplaylist'): + self.to_screen('Downloading just video %s because of --no-playlist' % clip_id) + return self.url_result(DaumClipIE._URL_TEMPLATE % clip_id, 'DaumClip') + else: + self.to_screen('Downloading playlist %s - add --no-playlist to just download video' % list_id) + class DaumPlaylistIE(DaumListIE): _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/mypot/(?:View\.do|Top\.tv)\?.*?playlistid=(?P[0-9]+)' IE_NAME = 'daum.net:playlist' + _URL_TEMPLATE = 'http://tvpot.daum.net/mypot/View.do?playlistid=%s' _TESTS = [{ 'note': 'Playlist url with clipid', @@ -207,12 +193,16 @@ class DaumPlaylistIE(DaumListIE): } }] - def _real_extract(self, url): - if DaumClipIE.suitable(url) and self._downloader.params.get('noplaylist'): - return self.url_result(url, 'DaumClip') + @classmethod + def suitable(cls, url): + return False if DaumUserIE.suitable(url) else super(DaumPlaylistIE, cls).suitable(url) + def _real_extract(self, url): list_id = self._match_id(url) - self.to_screen('Downloading playlist %s - add --no-playlist to just download video' % list_id) + + clip_result = self._check_clip(url, list_id) + if clip_result: + return clip_result name, entries = self._get_entries(list_id, 'playlistid') @@ -220,7 +210,7 @@ class DaumPlaylistIE(DaumListIE): class DaumUserIE(DaumListIE): - _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/mypot/(?:View|Top)\.do\?.*?ownerid=(?P[0-9a-zA-Z]+)' + _VALID_URL = r'https?://(?:m\.)?tvpot\.daum\.net/mypot/(?:View|Top)\.(?:do|tv)\?.*?ownerid=(?P[0-9a-zA-Z]+)' IE_NAME = 'daum.net:user' _TESTS = [{ @@ -254,17 +244,23 @@ class DaumUserIE(DaumListIE): }, { 'url': 'http://tvpot.daum.net/mypot/Top.do?ownerid=o2scDLIVbHc0', 'only_matching': True, + }, { + 'url': 'http://m.tvpot.daum.net/mypot/Top.tv?ownerid=45x1okb1If50&playlistid=3569733', + 'only_matching': True, }] def _real_extract(self, url): - if DaumClipIE.suitable(url) and self._downloader.params.get('noplaylist'): - return self.url_result(url, 'DaumClip') + list_id = self._match_id(url) - if DaumPlaylistIE.suitable(url): - return self.url_result(url, 'DaumPlaylist') + clip_result = self._check_clip(url, list_id) + if clip_result: + return clip_result + + query_dict = compat_parse_qs(compat_urlparse.urlparse(url).query) + if 'playlistid' in query_dict: + playlist_id = query_dict['playlistid'][0] + return self.url_result(DaumPlaylistIE._URL_TEMPLATE % playlist_id, 'DaumPlaylist') - list_id = self._match_id(url) - self.to_screen('Downloading playlist %s - add --no-playlist to just download video' % list_id) name, entries = self._get_entries(list_id, 'ownerid') return self.playlist_result(entries, list_id, name)