From d1b61870127fdee53147ac4c249af34a35562dbe Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Sergey=20M=E2=80=A4?= Date: Mon, 26 Feb 2018 21:50:51 +0700 Subject: [PATCH] [npo] Validate and filter format URLs (closes #15709) --- youtube_dl/extractor/npo.py | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/youtube_dl/extractor/npo.py b/youtube_dl/extractor/npo.py index f756331f8..eafce4b84 100644 --- a/youtube_dl/extractor/npo.py +++ b/youtube_dl/extractor/npo.py @@ -195,6 +195,10 @@ def _get_info(self, video_id): formats = [] urls = set() + def is_legal_url(format_url): + return format_url and format_url not in urls and re.match( + r'^(?:https?:)?//', format_url) + QUALITY_LABELS = ('Laag', 'Normaal', 'Hoog') QUALITY_FORMATS = ('adaptive', 'wmv_sb', 'h264_sb', 'wmv_bb', 'h264_bb', 'wvc1_std', 'h264_std') @@ -208,7 +212,7 @@ def _get_info(self, video_id): })['items'][0] for num, item in enumerate(items): item_url = item.get('url') - if not item_url or item_url in urls: + if not is_legal_url(item_url): continue urls.add(item_url) format_id = self._search_regex( @@ -279,7 +283,7 @@ def add_format_url(format_url): if not is_live: for num, stream in enumerate(metadata.get('streams', [])): stream_url = stream.get('url') - if not stream_url or stream_url in urls: + if not is_legal_url(stream_url): continue urls.add(stream_url) # smooth streaming is not supported