From d47f46e17e8611d6bad81b1cae3cc076385a6283 Mon Sep 17 00:00:00 2001 From: The Hatsune Daishi Date: Sat, 18 Sep 2021 14:55:17 +0900 Subject: [PATCH] [damtomo] Add extractor (#992) Authored by: nao20010128nao --- yt_dlp/extractor/damtomo.py | 113 +++++++++++++++++++++++++++++++++ yt_dlp/extractor/extractors.py | 4 ++ 2 files changed, 117 insertions(+) create mode 100644 yt_dlp/extractor/damtomo.py diff --git a/yt_dlp/extractor/damtomo.py b/yt_dlp/extractor/damtomo.py new file mode 100644 index 000000000..456cd35a4 --- /dev/null +++ b/yt_dlp/extractor/damtomo.py @@ -0,0 +1,113 @@ +# coding: utf-8 +from __future__ import unicode_literals + +import re + +from .common import InfoExtractor +from ..utils import ExtractorError, clean_html, int_or_none, try_get, unified_strdate +from ..compat import compat_str + + +class DamtomoBaseIE(InfoExtractor): + def _real_extract(self, url): + video_id = self._match_id(url) + webpage, handle = self._download_webpage_handle(self._WEBPAGE_URL_TMPL % video_id, video_id, encoding='sjis') + + if handle.url == 'https://www.clubdam.com/sorry/': + raise ExtractorError('You are rate-limited. Try again later.', expected=True) + if '

予期せぬエラーが発生しました。

' in webpage: + raise ExtractorError('There is an error on server-side. Try again later.', expected=True) + + description = self._search_regex(r'(?m)
\s*

\s*([^<]*?)\s*

', webpage, 'description', default=None) + uploader_id = self._search_regex(r'(?P.+?)', webpage)} + + # since videos do not have title, give the name of song instead + data_dict['user_name'] = re.sub(r'\s*さん\s*$', '', data_dict['user_name']) + title = data_dict.get('song_title') + + stream_tree = self._download_xml( + self._DKML_XML_URL % video_id, video_id, note='Requesting stream information', encoding='sjis', + # doing this has no problem since there is no character outside ASCII, + # and never likely to happen in the future + transform_source=lambda x: re.sub(r'\s*encoding="[^"]+?"', '', x)) + m3u8_url = try_get(stream_tree, lambda x: x.find( + './/d:streamingUrl', {'d': self._DKML_XML_NS}).text.strip(), compat_str) + if not m3u8_url: + raise ExtractorError('Failed to obtain m3u8 URL') + formats = self._extract_m3u8_formats(m3u8_url, video_id, ext='mp4') + self._sort_formats(formats) + + return { + 'id': video_id, + 'title': title, + 'uploader_id': uploader_id, + 'description': description, + 'uploader': data_dict.get('user_name'), + 'upload_date': unified_strdate(self._search_regex(r'(\d{4}/\d{2}/\d{2})', data_dict.get('date'), 'upload_date', default=None)), + 'view_count': int_or_none(self._search_regex(r'(\d+)', data_dict['audience'], 'view_count', default=None)), + 'like_count': int_or_none(self._search_regex(r'(\d+)', data_dict['nice'], 'like_count', default=None)), + 'track': title, + 'artist': data_dict.get('song_artist'), + 'formats': formats, + } + + +class DamtomoVideoIE(DamtomoBaseIE): + IE_NAME = 'damtomo:video' + _VALID_URL = r'https?://(?:www\.)?clubdam\.com/app/damtomo/(?:SP/)?karaokeMovie/StreamingDkm\.do\?karaokeMovieId=(?P\d+)' + _WEBPAGE_URL_TMPL = 'https://www.clubdam.com/app/damtomo/karaokeMovie/StreamingDkm.do?karaokeMovieId=%s' + _DKML_XML_URL = 'https://www.clubdam.com/app/damtomo/karaokeMovie/GetStreamingDkmUrlXML.do?movieSelectFlg=2&karaokeMovieId=%s' + _DKML_XML_NS = 'https://www.clubdam.com/app/damtomo/karaokeMovie/GetStreamingDkmUrlXML' + _TESTS = [{ + 'url': 'https://www.clubdam.com/app/damtomo/karaokeMovie/StreamingDkm.do?karaokeMovieId=2414316', + 'info_dict': { + 'id': '2414316', + 'title': 'Get Wild', + 'uploader': 'Kドロン', + 'uploader_id': 'ODk5NTQwMzQ', + 'track': 'Get Wild', + 'artist': 'TM NETWORK(TMN)', + 'upload_date': '20201226', + } + }] + + +class DamtomoRecordIE(DamtomoBaseIE): + IE_NAME = 'damtomo:record' + _VALID_URL = r'https?://(?:www\.)?clubdam\.com/app/damtomo/(?:SP/)?karaokePost/StreamingKrk\.do\?karaokeContributeId=(?P\d+)' + _WEBPAGE_URL_TMPL = 'https://www.clubdam.com/app/damtomo/karaokePost/StreamingKrk.do?karaokeContributeId=%s' + _DKML_XML_URL = 'https://www.clubdam.com/app/damtomo/karaokePost/GetStreamingKrkUrlXML.do?karaokeContributeId=%s' + _DKML_XML_NS = 'https://www.clubdam.com/app/damtomo/karaokePost/GetStreamingKrkUrlXML' + _TESTS = [{ + 'url': 'https://www.clubdam.com/app/damtomo/karaokePost/StreamingKrk.do?karaokeContributeId=27376862', + 'info_dict': { + 'id': '27376862', + 'title': 'イカSUMMER [良音]', + 'description': None, + 'uploader': 'NANA', + 'uploader_id': 'MzAyMDExNTY', + 'upload_date': '20210721', + 'view_count': 4, + 'like_count': 1, + 'track': 'イカSUMMER [良音]', + 'artist': 'ORANGE RANGE', + } + }, { + 'url': 'https://www.clubdam.com/app/damtomo/karaokePost/StreamingKrk.do?karaokeContributeId=27489418', + 'info_dict': { + 'id': '27489418', + 'title': '心みだれて〜say it with flowers〜(生音)', + 'uploader_id': 'NjI1MjI2MjU', + 'description': 'やっぱりキーを下げて正解だった感じ。リベンジ成功ということで。', + 'uploader': '箱の「中の人」', + 'upload_date': '20210815', + 'view_count': 5, + 'like_count': 3, + 'track': '心みだれて〜say it with flowers〜(生音)', + 'artist': '小林明子', + } + }] diff --git a/yt_dlp/extractor/extractors.py b/yt_dlp/extractor/extractors.py index 802907bd9..3dd56c65a 100644 --- a/yt_dlp/extractor/extractors.py +++ b/yt_dlp/extractor/extractors.py @@ -310,6 +310,10 @@ DailymotionPlaylistIE, DailymotionUserIE, ) +from .damtomo import ( + DamtomoRecordIE, + DamtomoVideoIE, +) from .daum import ( DaumIE, DaumClipIE,