From 8abd647c59c9eb8f0fefd2b329e62b2b32bac6ea Mon Sep 17 00:00:00 2001 From: nixxo Date: Thu, 5 Nov 2020 20:52:28 +0100 Subject: [PATCH] [mailru] removed escaped braces, use urljoin, added tests --- youtube_dlc/extractor/mailru.py | 13 +++++++++++-- 1 file changed, 11 insertions(+), 2 deletions(-) diff --git a/youtube_dlc/extractor/mailru.py b/youtube_dlc/extractor/mailru.py index 6fdf70aa6..5bfe40649 100644 --- a/youtube_dlc/extractor/mailru.py +++ b/youtube_dlc/extractor/mailru.py @@ -12,6 +12,7 @@ parse_duration, remove_end, try_get, + urljoin, ) @@ -93,6 +94,14 @@ class MailRuIE(InfoExtractor): { 'url': 'https://my.mail.ru//list//sinyutin10/video/_myvideo/4.html', 'only_matching': True, + }, + { + 'url': 'https://my.mail.ru/mail/cloud-strife/video/embed/Games/2009', + 'only_matching': True, + }, + { + 'url': 'https://videoapi.my.mail.ru/videos/embed/mail/cloud-strife/Games/2009.html', + 'only_matching': True, } ] @@ -110,7 +119,7 @@ def _real_extract(self, url): webpage = self._download_webpage(url, video_id) page_config = self._parse_json(self._search_regex([ r'(?s)]+class="sp-video__page-config"[^>]*>(.+?)', - r'(?s)"video":\s*(\{.+?\}),'], + r'(?s)"video":\s*({.+?}),'], webpage, 'page config', default='{}'), video_id, fatal=False) if page_config: meta_url = page_config.get('metaUrl') or page_config.get('video', {}).get('metaUrl') or page_config.get('metadataUrl') @@ -121,7 +130,7 @@ def _real_extract(self, url): # fix meta_url if missing the host address if re.match(r'^\/\+\/', meta_url): - meta_url = 'https://my.mail.ru' + meta_url + meta_url = urljoin('https://my.mail.ru', meta_url) if meta_url: video_data = self._download_json(