diff options
author | Bryan Newbold <bnewbold@robocracy.org> | 2019-01-28 21:21:42 -0800 |
---|---|---|
committer | Bryan Newbold <bnewbold@robocracy.org> | 2019-01-28 21:21:42 -0800 |
commit | d15bb03e80cdb9041f8464cca5b040d6e766efa0 (patch) | |
tree | 4f6c79b5de2ea121e2d13b02f51b277a314c8e75 | |
parent | 95a60dc56d956c1267eb39d7044a77655ec48ce6 (diff) | |
download | fatcat-d15bb03e80cdb9041f8464cca5b040d6e766efa0.tar.gz fatcat-d15bb03e80cdb9041f8464cca5b040d6e766efa0.zip |
fixes to transform code
-rw-r--r-- | python/fatcat_tools/transforms.py | 20 |
1 files changed, 11 insertions, 9 deletions
diff --git a/python/fatcat_tools/transforms.py b/python/fatcat_tools/transforms.py index 9cb557ad..fc079382 100644 --- a/python/fatcat_tools/transforms.py +++ b/python/fatcat_tools/transforms.py @@ -30,6 +30,7 @@ def entity_from_json(json_str, entity_type, api_client=None): return api_client.deserialize(thing, entity_type) def check_kbart(year, archive): + print(archive) if not archive or not archive.get('year_spans'): return None for span in archive['year_spans']: @@ -93,13 +94,14 @@ def release_to_elasticsearch(entity): in_ia_sim = False in_shadow = False + release_year = release.release_year if release.release_date: # .isoformat() results in, eg, '2010-10-22' (YYYY-MM-DD) t['release_date'] = release.release_date.isoformat() - if release.release_year is None: - t['release_year'] = release.release_date.year - if release.release_year is not None: - t['release_year'] = release.release_year + if not release_year: + release_year = release.release_date.year + if release_year: + t['release_year'] = release_year t['any_abstract'] = len(release.abstracts) > 0 t['ref_count'] = len(release.refs or []) @@ -118,15 +120,15 @@ def release_to_elasticsearch(entity): t['container_type'] = container.container_type if container.extra: c_extra = container.extra - if c_extra.get('kbart') and release.year: - in_jstor = check_kbart(release.year, c_extra['kbart'].get('jstor')) + if c_extra.get('kbart') and release_year: + in_jstor = check_kbart(release_year, c_extra['kbart'].get('jstor')) in_kbart = in_jstor for archive in ('portico', 'lockss', 'clockss'): - in_kbart = in_kbart or check_kbart(release.year, c_extra['kbart'].get(archive)) + in_kbart = in_kbart or check_kbart(release_year, c_extra['kbart'].get(archive)) if c_extra.get('ia'): - if c_extra['ia'].get('sim') and release.year: - in_ia_sim = check_kbart(release, c_extra['ia']['sim'].get('year_spans')) + if c_extra['ia'].get('sim') and release_year: + in_ia_sim = check_kbart(release_year, c_extra['ia']['sim']) if c_extra['ia'].get('longtail_oa'): is_longtail_oa = True if c_extra.get('sherpa_romeo'): |