diff options
Diffstat (limited to 'python/fatcat_tools/importers/datacite.py')
| -rw-r--r-- | python/fatcat_tools/importers/datacite.py | 10 |
1 files changed, 9 insertions, 1 deletions
diff --git a/python/fatcat_tools/importers/datacite.py b/python/fatcat_tools/importers/datacite.py index 9250fc5e..81f00876 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)): @@ -714,7 +721,8 @@ class DataciteImporter(EntityImporter): name_scheme = nid.get('nameIdentifierScheme', '') or '' if not name_scheme.lower() == "orcid": continue - orcid = nid.get('nameIdentifier', '').replace('https://orcid.org/', '') + orcid = nid.get('nameIdentifier') or '' + orcid = orcid.replace('https://orcid.org/', '') if not orcid: continue creator_id = self.lookup_orcid(orcid) |