From cd566e0b44cfa7cb110b60158aa029189e2d03ff Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Thu, 18 Apr 2019 11:14:39 -0700 Subject: fix missing postproc bug --- arabesque.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/arabesque.py b/arabesque.py index b5b2224..cf8cf96 100755 --- a/arabesque.py +++ b/arabesque.py @@ -515,8 +515,8 @@ def forward(seed_id_file, map_db, output_db): print("MISSING url: {}".format(raw_line.strip())) # need to insert *something* in this case... initial_domain = urllib3.util.parse_url(seed_url).host - c.execute("INSERT INTO crawl_result VALUES (?,?,?,?,?,?,?,?,?,?,?,?)", - (seed_url, identifier, initial_domain, None, None, None, None, None, None, None, None, False)) + c.execute("INSERT INTO crawl_result VALUES (?,?,?,?,?,?,?,?,?,?,?,?,?)", + (seed_url, identifier, initial_domain, None, None, None, None, None, None, None, None, False, None)) counts['map-url-missing'] += 1 continue row = first_row -- cgit v1.2.3