diff options
| author | Bryan Newbold <bnewbold@robocracy.org> | 2018-09-25 15:00:05 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@robocracy.org> | 2018-09-25 15:00:05 -0700 |
| commit | c1d0fea36ff88cd8c01fbf5fc7d871ac08ffef70 (patch) | |
| tree | 9818951b70849b19f3f8f55504bb3a0ab955dc3c | |
| parent | 747281b45ddc35312c6978e0a38a3145abc8883a (diff) | |
| download | fatcat-c1d0fea36ff88cd8c01fbf5fc7d871ac08ffef70.tar.gz fatcat-c1d0fea36ff88cd8c01fbf5fc7d871ac08ffef70.zip | |
more crossref_importer fixes (ugh)
| -rw-r--r-- | python/fatcat/crossref_importer.py | 13 |
1 files changed, 11 insertions, 2 deletions
diff --git a/python/fatcat/crossref_importer.py b/python/fatcat/crossref_importer.py index 8e179b69..10efe59c 100644 --- a/python/fatcat/crossref_importer.py +++ b/python/fatcat/crossref_importer.py @@ -195,11 +195,20 @@ class FatcatCrossrefImporter(FatcatImporter): if max(len(contribs), len(refs), len(abstracts)) > 750: return None + # release date parsing is amazingly complex release_date = obj['issued']['date-parts'][0] - if len(release_date) == 3: + if not release_date or not release_date[0]: + # got some NoneType, even though at least year is supposed to be set + release_date = None + elif len(release_date) == 3: release_date = datetime.datetime(year=release_date[0], month=release_date[1], day=release_date[2]) else: + # only the year is actually required; mangle to first day for date + # (TODO: something better?) release_date = datetime.datetime(year=release_date[0], month=1, day=1) + # convert to string ISO datetime format (if not null) + if release_date: + release_date = release_date.isoformat() + "Z" re = fatcat_client.ReleaseEntity( work_id=None, @@ -216,7 +225,7 @@ class FatcatCrossrefImporter(FatcatImporter): pmid=extids['pmid'], pmcid=extids['pmcid'], wikidata_qid=extids['wikidata_qid'], - release_date=release_date.isoformat() + "Z", + release_date=release_date, issue=obj.get('issue'), volume=obj.get('volume'), pages=obj.get('page'), |