From 11dcced98bd842b41896d6e8f7edc711434ea12c Mon Sep 17 00:00:00 2001 From: 0x4ea5 <0x4ea5@gmail.com> Date: Sun, 25 Sep 2022 19:01:51 -0700 Subject: [PATCH] added support to bilibili collectiondetail api --- src/you_get/extractors/bilibili.py | 22 +++++++++++++++++++++- 1 file changed, 21 insertions(+), 1 deletion(-) diff --git a/src/you_get/extractors/bilibili.py b/src/you_get/extractors/bilibili.py index 1a13b61c..6d34c2c4 100644 --- a/src/you_get/extractors/bilibili.py +++ b/src/you_get/extractors/bilibili.py @@ -115,11 +115,15 @@ class Bilibili(VideoExtractor): @staticmethod def bilibili_space_channel_api(mid, cid, pn=1, ps=100): return 'https://api.bilibili.com/x/space/channel/video?mid=%s&cid=%s&pn=%s&ps=%s&order=0&jsonp=jsonp' % (mid, cid, pn, ps) + + @staticmethod + def bilibili_space_collection_api(mid, cid, pn=1, ps=30): + return 'https://api.bilibili.com/x/polymer/space/seasons_archives_list?mid=%s&season_id=%s&sort_reverse=false&page_num=%s&page_size=%s' % (mid, cid, pn, ps) @staticmethod def bilibili_series_archives_api(mid, sid, pn=1, ps=100): return 'https://api.bilibili.com/x/series/archives?mid=%s&series_id=%s&pn=%s&ps=%s&only_normal=true&sort=asc&jsonp=jsonp' % (mid, sid, pn, ps) - + @staticmethod def bilibili_space_favlist_api(fid, pn=1, ps=20): return 'https://api.bilibili.com/x/v3/fav/resource/list?media_id=%s&pn=%s&ps=%s&order=mtime&type=0&tid=0&jsonp=jsonp' % (fid, pn, ps) @@ -628,6 +632,8 @@ class Bilibili(VideoExtractor): sort = 'space_channel' elif re.match(r'https?://space\.?bilibili\.com/(\d+)/channel/seriesdetail\?.*sid=(\d+)', self.url): sort = 'space_channel_series' + elif re.match(r'https?://space\.?bilibili\.com/(\d+)/channel/collectiondetail\?.*sid=(\d+)', self.url): + sort = 'space_channel_collection' elif re.match(r'https?://space\.?bilibili\.com/(\d+)/favlist\?.*fid=(\d+)', self.url): sort = 'space_favlist' elif re.match(r'https?://space\.?bilibili\.com/(\d+)/video', self.url): @@ -752,6 +758,20 @@ class Bilibili(VideoExtractor): url = 'https://www.bilibili.com/video/av%s' % video['aid'] self.__class__().download_playlist_by_url(url, **kwargs) + elif sort == 'space_channel_collection': + m = re.match(r'https?://space\.?bilibili\.com/(\d+)/channel/collectiondetail\?.*sid=(\d+)', self.url) + mid, sid = m.group(1), m.group(2) + api_url = self.bilibili_space_collection_api(mid, sid) + api_content = get_content(api_url, headers=self.bilibili_headers(referer=self.url)) + archives_info = json.loads(api_content) + # TBD: channel of more than 100 videos + + epn, i = len(archives_info['data']['archives']), 0 + for video in archives_info['data']['archives']: + i += 1; log.w('Extracting %s of %s videos ...' % (i, epn)) + url = 'https://www.bilibili.com/video/av%s' % video['aid'] + self.__class__().download_playlist_by_url(url, **kwargs) + elif sort == 'space_favlist': m = re.match(r'https?://space\.?bilibili\.com/(\d+)/favlist\?.*fid=(\d+)', self.url) vmid, fid = m.group(1), m.group(2)