From f19414e707e68eb1d0fe669963a476c4656a259c Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Tue, 2 Nov 2021 17:59:22 -0700 Subject: arabesque import 'hit' field is 1/0, not true/false --- python/fatcat_tools/importers/arabesque.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/python/fatcat_tools/importers/arabesque.py b/python/fatcat_tools/importers/arabesque.py index ccf35446..c8f7c77c 100644 --- a/python/fatcat_tools/importers/arabesque.py +++ b/python/fatcat_tools/importers/arabesque.py @@ -62,13 +62,13 @@ class ArabesqueMatchImporter(EntityImporter): def want(self, row): if self.require_grobid and not row['postproc_status'] == "200": return False - if (row['hit'] is True + if (bool(row['hit']) is True and row['final_sha1'] and row['final_timestamp'] and row['final_timestamp'] != "-" and len(row['final_timestamp']) == 14 and row['final_mimetype'] - and row['hit'] is True + and bool(row['hit']) is True and row['identifier']): return True else: -- cgit v1.2.3