From 1c22d7a7f30917abfd2b7495f7bd02d51cb8528a Mon Sep 17 00:00:00 2001 From: Namnamseo <0201ssw+github@gmail.com> Date: Thu, 24 Aug 2017 11:32:24 +0900 Subject: [kakao] Add extractor (closes #12298) --- youtube_dl/extractor/kakao.py | 140 ++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 140 insertions(+) create mode 100644 youtube_dl/extractor/kakao.py (limited to 'youtube_dl/extractor/kakao.py') diff --git a/youtube_dl/extractor/kakao.py b/youtube_dl/extractor/kakao.py new file mode 100644 index 000000000..0caa41e9e --- /dev/null +++ b/youtube_dl/extractor/kakao.py @@ -0,0 +1,140 @@ +# coding: utf-8 + +from __future__ import unicode_literals + +from .common import InfoExtractor +from ..utils import ( + int_or_none, + compat_str, + unified_timestamp, +) + + +class KakaoIE(InfoExtractor): + _VALID_URL = r'https?://tv.kakao.com/channel/(?P\d+)/cliplink/(?P\d+)' + IE_NAME = 'kakao.com' + + _TESTS = [{ + 'url': 'http://tv.kakao.com/channel/2671005/cliplink/301965083', + 'md5': '702b2fbdeb51ad82f5c904e8c0766340', + 'info_dict': { + 'id': '301965083', + 'ext': 'mp4', + 'title': '乃木坂46 バナナマン 「3期生紹介コーナーが始動！顔高低差GPも！」 『乃木坂工事中』', + 'uploader_id': 2671005, + 'uploader': '그랑그랑이', + 'timestamp': 1488160199, + 'upload_date': '20170227', + } + }, { + 'url': 'http://tv.kakao.com/channel/2653210/cliplink/300103180', + 'md5': 'a8917742069a4dd442516b86e7d66529', + 'info_dict': { + 'id': '300103180', + 'ext': 'mp4', + 'description': '러블리즈 - Destiny (나의 지구) (Lovelyz - Destiny)\r\n\r\n[쇼! 음악중심] 20160611, 507회', + 'title': '러블리즈 - Destiny (나의 지구) (Lovelyz - Destiny)', + 'uploader_id': 2653210, + 'uploader': '쇼 음악중심', + 'timestamp': 1485684628, + 'upload_date': '20170129', + } + }] + + def _real_extract(self, url): + video_id = self._match_id(url) + + player_url = 'http://tv.kakao.com/embed/player/cliplink/' + video_id + \ + '?service=kakao_tv&autoplay=1&profile=HIGH&wmode=transparent' + player_header = {'Referer': player_url} + + impress = self._download_json( + 'http://tv.kakao.com/api/v1/ft/cliplinks/%s/impress' % video_id, + video_id, 'Downloading video info', + query={ + 'player': 'monet_html5', + 'referer': url, + 'uuid': '', + 'service': 'kakao_tv', + 'section': '', + 'dteType': 'PC', + 'fields': 'clipLink,clip,channel,hasPlusFriend,-service,-tagList' + }, headers=player_header) + + clipLink = impress['clipLink'] + clip = clipLink['clip'] + + video_info = { + 'id': video_id, + 'title': clip['title'], + 'description': clip.get('description'), + 'uploader': clipLink.get('channel', {}).get('name'), + 'uploader_id': clipLink.get('channelId'), + 'duration': int_or_none(clip.get('duration')), + 'view_count': int_or_none(clip.get('playCount')), + 'like_count': int_or_none(clip.get('likeCount')), + 'comment_count': int_or_none(clip.get('commentCount')), + } + + tid = impress.get('tid', '') + raw = self._download_json( + 'http://tv.kakao.com/api/v1/ft/cliplinks/%s/raw' % video_id, + video_id, 'Downloading video formats info', + query={ + 'player': 'monet_html5', + 'referer': url, + 'uuid': '', + 'service': 'kakao_tv', + 'section': '', + 'tid': tid, + 'profile': 'HIGH', + 'dteType': 'PC', + }, headers=player_header, fatal=False) + + formats = [] + for fmt in raw.get('outputList', []): + try: + profile_name = fmt['profile'] + fmt_url_json = self._download_json( + 'http://tv.kakao.com/api/v1/ft/cliplinks/%s/raw/videolocation' % video_id, + video_id, 'Downloading video URL for profile %s' % profile_name, + query={ + 'service': 'kakao_tv', + 'section': '', + 'tid': tid, + 'profile': profile_name + }, headers=player_header, fatal=False) + + if fmt_url_json is None: + continue + + fmt_url = fmt_url_json['url'] + formats.append({ + 'url': fmt_url, + 'format_id': profile_name, + 'width': int_or_none(fmt.get('width')), + 'height': int_or_none(fmt.get('height')), + 'format_note': fmt.get('label'), + 'filesize': int_or_none(fmt.get('filesize')) + }) + except KeyError: + pass + + self._sort_formats(formats) + video_info['formats'] = formats + + top_thumbnail = clip.get('thumbnailUrl') + thumbs = [] + for thumb in clip.get('clipChapterThumbnailList', []): + thumbs.append({ + 'url': thumb.get('thumbnailUrl'), + 'id': compat_str(thumb.get('timeInSec')), + 'preference': -1 if thumb.get('isDefault') else 0 + }) + video_info['thumbnail'] = top_thumbnail + video_info['thumbnails'] = thumbs + + upload_date = unified_timestamp(clipLink.get('createTime')) + video_info['timestamp'] = upload_date + + return video_info -- cgit 1.4.1 From f70ddd4aebbfb0bdf2f63c1eba5b5614d2cfb70f Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Sat, 23 Sep 2017 07:25:15 +0700 Subject: [kakao] Improve (closes #14007) --- youtube_dl/extractor/kakao.py | 113 +++++++++++++++++++++++------------------- 1 file changed, 61 insertions(+), 52 deletions(-) (limited to 'youtube_dl/extractor/kakao.py') diff --git a/youtube_dl/extractor/kakao.py b/youtube_dl/extractor/kakao.py index 0caa41e9e..c9b438efc 100644 --- a/youtube_dl/extractor/kakao.py +++ b/youtube_dl/extractor/kakao.py @@ -3,16 +3,17 @@ from __future__ import unicode_literals from .common import InfoExtractor +from ..compat import compat_str from ..utils import ( int_or_none, - compat_str, unified_timestamp, + update_url_query, ) class KakaoIE(InfoExtractor): _VALID_URL = r'https?://tv.kakao.com/channel/(?P\d+)/cliplink/(?P\d+)' - IE_NAME = 'kakao.com' + _API_BASE = 'http://tv.kakao.com/api/v1/ft/cliplinks' _TESTS = [{ 'url': 'http://tv.kakao.com/channel/2671005/cliplink/301965083', @@ -44,60 +45,57 @@ class KakaoIE(InfoExtractor): def _real_extract(self, url): video_id = self._match_id(url) - player_url = 'http://tv.kakao.com/embed/player/cliplink/' + video_id + \ - '?service=kakao_tv&autoplay=1&profile=HIGH&wmode=transparent' - player_header = {'Referer': player_url} + player_header = { + 'Referer': update_url_query( + 'http://tv.kakao.com/embed/player/cliplink/%s' % video_id, { + 'service': 'kakao_tv', + 'autoplay': '1', + 'profile': 'HIGH', + 'wmode': 'transparent', + }) + } + + QUERY_COMMON = { + 'player': 'monet_html5', + 'referer': url, + 'uuid': '', + 'service': 'kakao_tv', + 'section': '', + 'dteType': 'PC', + } + query = QUERY_COMMON.copy() + query['fields'] = 'clipLink,clip,channel,hasPlusFriend,-service,-tagList' impress = self._download_json( - 'http://tv.kakao.com/api/v1/ft/cliplinks/%s/impress' % video_id, + '%s/%s/impress' % (self._API_BASE, video_id), video_id, 'Downloading video info', - query={ - 'player': 'monet_html5', - 'referer': url, - 'uuid': '', - 'service': 'kakao_tv', - 'section': '', - 'dteType': 'PC', - 'fields': 'clipLink,clip,channel,hasPlusFriend,-service,-tagList' - }, headers=player_header) - - clipLink = impress['clipLink'] - clip = clipLink['clip'] - - video_info = { - 'id': video_id, - 'title': clip['title'], - 'description': clip.get('description'), - 'uploader': clipLink.get('channel', {}).get('name'), - 'uploader_id': clipLink.get('channelId'), - 'duration': int_or_none(clip.get('duration')), - 'view_count': int_or_none(clip.get('playCount')), - 'like_count': int_or_none(clip.get('likeCount')), - 'comment_count': int_or_none(clip.get('commentCount')), - } + query=query, headers=player_header) + + clip_link = impress['clipLink'] + clip = clip_link['clip'] + + title = clip.get('title') or clip_link.get('displayTitle') tid = impress.get('tid', '') + + query = QUERY_COMMON.copy() + query.update({ + 'tid': tid, + 'profile': 'HIGH', + }) raw = self._download_json( - 'http://tv.kakao.com/api/v1/ft/cliplinks/%s/raw' % video_id, + '%s/%s/raw' % (self._API_BASE, video_id), video_id, 'Downloading video formats info', - query={ - 'player': 'monet_html5', - 'referer': url, - 'uuid': '', - 'service': 'kakao_tv', - 'section': '', - 'tid': tid, - 'profile': 'HIGH', - 'dteType': 'PC', - }, headers=player_header, fatal=False) + query=query, headers=player_header) formats = [] for fmt in raw.get('outputList', []): try: profile_name = fmt['profile'] fmt_url_json = self._download_json( - 'http://tv.kakao.com/api/v1/ft/cliplinks/%s/raw/videolocation' % video_id, - video_id, 'Downloading video URL for profile %s' % profile_name, + '%s/%s/raw/videolocation' % (self._API_BASE, video_id), + video_id, + 'Downloading video URL for profile %s' % profile_name, query={ 'service': 'kakao_tv', 'section': '', @@ -119,11 +117,8 @@ class KakaoIE(InfoExtractor): }) except KeyError: pass - self._sort_formats(formats) - video_info['formats'] = formats - top_thumbnail = clip.get('thumbnailUrl') thumbs = [] for thumb in clip.get('clipChapterThumbnailList', []): thumbs.append({ @@ -131,10 +126,24 @@ class KakaoIE(InfoExtractor): 'id': compat_str(thumb.get('timeInSec')), 'preference': -1 if thumb.get('isDefault') else 0 }) - video_info['thumbnail'] = top_thumbnail - video_info['thumbnails'] = thumbs - - upload_date = unified_timestamp(clipLink.get('createTime')) - video_info['timestamp'] = upload_date + top_thumbnail = clip.get('thumbnailUrl') + if top_thumbnail: + thumbs.append({ + 'url': top_thumbnail, + 'preference': 10, + }) - return video_info + return { + 'id': video_id, + 'title': title, + 'description': clip.get('description'), + 'uploader': clip_link.get('channel', {}).get('name'), + 'uploader_id': clip_link.get('channelId'), + 'thumbnails': thumbs, + 'timestamp': unified_timestamp(clip_link.get('createTime')), + 'duration': int_or_none(clip.get('duration')), + 'view_count': int_or_none(clip.get('playCount')), + 'like_count': int_or_none(clip.get('likeCount')), + 'comment_count': int_or_none(clip.get('commentCount')), + 'formats': formats, + } -- cgit 1.4.1 From 4d182955a2bbe95eb4b931f82204f402ffd5552d Mon Sep 17 00:00:00 2001 From: Sergey M․ Date: Sun, 24 Sep 2017 00:19:27 +0700 Subject: [kakao] Fix _VALID_URL --- youtube_dl/extractor/kakao.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'youtube_dl/extractor/kakao.py') diff --git a/youtube_dl/extractor/kakao.py b/youtube_dl/extractor/kakao.py index c9b438efc..7fa140b0c 100644 --- a/youtube_dl/extractor/kakao.py +++ b/youtube_dl/extractor/kakao.py @@ -12,7 +12,7 @@ from ..utils import ( class KakaoIE(InfoExtractor): - _VALID_URL = r'https?://tv.kakao.com/channel/(?P\d+)/cliplink/(?P\d+)' + _VALID_URL = r'https?://tv\.kakao\.com/channel/(?P\d+)/cliplink/(?P\d+)' _API_BASE = 'http://tv.kakao.com/api/v1/ft/cliplinks' _TESTS = [{ -- cgit 1.4.1