aboutsummaryrefslogtreecommitdiffstats
path: root/python/fatcat_tools
diff options
context:
space:
mode:
authorbnewbold <bnewbold@archive.org>2020-03-23 16:37:08 +0000
committerMartin Czygan <martin@archive.org>2020-03-23 16:37:08 +0000
commit8af9df9fff925c90f2bfb52c4a2b2ea918b4eda2 (patch)
tree8d1e9a900279ab0e151e488f63273ac4f9b702fe /python/fatcat_tools
parent4bcef62ecd98f2719fc4d1cef35394b0bad5cb2b (diff)
downloadfatcat-8af9df9fff925c90f2bfb52c4a2b2ea918b4eda2.tar.gz
fatcat-8af9df9fff925c90f2bfb52c4a2b2ea918b4eda2.zip
datacite: add year sanity restrictions
Example of entities with bogus years: https://fatcat.wiki/release/search?q=doi_registrar%3Adatacite+year%3A%3E2100 We can do a clean-up task, but first need to prevent creation of new bad metadata.
Diffstat (limited to 'python/fatcat_tools')
-rw-r--r--python/fatcat_tools/importers/datacite.py7
1 files changed, 7 insertions, 0 deletions
diff --git a/python/fatcat_tools/importers/datacite.py b/python/fatcat_tools/importers/datacite.py
index 4e382348..db4709c2 100644
--- a/python/fatcat_tools/importers/datacite.py
+++ b/python/fatcat_tools/importers/datacite.py
@@ -222,6 +222,7 @@ class DataciteImporter(EntityImporter):
self.read_issn_map_file(issn_map_file)
self.debug = debug
self.insert_log_file = insert_log_file
+ self.this_year = datetime.datetime.now().year
print('datacite with debug={}'.format(self.debug), file=sys.stderr)
@@ -311,6 +312,12 @@ class DataciteImporter(EntityImporter):
release_date, release_month, release_year = parse_datacite_dates(
attributes.get('dates', []))
+ # block bogus far-future years/dates
+ if release_year is not None and (release_year > (self.this_year + 5) or release_year < 1000):
+ release_date = None
+ release_month = None
+ release_year = None
+
# Some records do not use the "dates" field (e.g. micropub), but:
# "attributes.published" or "attributes.publicationYear"
if not any((release_date, release_month, release_year)):