From 0cd655b946ebfbc3ed0bf1a2a1c1d9b94adca135 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Wed, 14 Jul 2021 16:38:21 -0700 Subject: more blocked-cookie patterns; fix old typo --- python/sandcrawler/ingest.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'python') diff --git a/python/sandcrawler/ingest.py b/python/sandcrawler/ingest.py index b610ab4..290bebc 100644 --- a/python/sandcrawler/ingest.py +++ b/python/sandcrawler/ingest.py @@ -523,7 +523,7 @@ class IngestFileWorker(SandcrawlerWorker): # check for popular cookie blocking URL patterns. On successful SPN # crawls, shouldn't see these redirect URLs - if '/cookieAbsent' in next_url or 'cookieSet=1' in next_url: + if '/cookieAbsent' in next_url or 'cookieSet=1' in next_url or 'error=cookies_not_supported' in next_url: result['status'] = 'blocked-cookie' return result @@ -571,7 +571,7 @@ class IngestFileWorker(SandcrawlerWorker): result['status'] = resource.status return result - if resource.terminal_url and ('/cookieAbsent' in next_url or 'cookieSet=1' in resource.terminal_url): + if resource.terminal_url and ('/cookieAbsent' in resource.terminal_url or 'cookieSet=1' in resource.terminal_url): result['status'] = 'blocked-cookie' return result -- cgit v1.2.3