Filter Crossref type and fix release_date naming in oa web workflow

The-Academic-Observatory · Jun 15, 2023 · ba0e490 · ba0e490
1 parent 88c0ea7
commit ba0e490
Show file tree

Hide file tree

Showing 2 changed files with 4 additions and 3 deletions.
diff --git a/academic_observatory_workflows/database/sql/create_aggregate.sql.jinja2 b/academic_observatory_workflows/database/sql/create_aggregate.sql.jinja2
@@ -1017,6 +1017,7 @@ SELECT
 
 FROM `{{ project_id }}.{{ dataset_id }}.doi{{ snapshot_date.strftime('%Y%m%d') }}` as dois, UNNEST(dois.affiliations.{{ aggregation_field }}) as aggregrate
 WHERE aggregrate.identifier IS NOT NULL
+AND dois.crossref.type NOT IN ("dataset")
 GROUP BY aggregrate.identifier, crossref.{{ group_by_time_field }}
 )
 

diff --git a/academic_observatory_workflows/workflows/oa_web_workflow.py b/academic_observatory_workflows/workflows/oa_web_workflow.py
@@ -697,7 +697,7 @@ def build_datasets(self, release: OaWebRelease, **kwargs):
             )
             for version in versions
         ]
-        last_updated = zenodo_versions[0].snapshot_date.format("D MMMM YYYY")
+        last_updated = zenodo_versions[0].release_date.format("D MMMM YYYY")
         country_stats = make_entity_stats(countries)
         institution_stats = make_entity_stats(institutions)
         stats = Stats(START_YEAR, END_YEAR, last_updated, zenodo_versions, country_stats, institution_stats)
@@ -960,11 +960,11 @@ def to_dict(self) -> Dict:
 
 @dataclasses.dataclass
 class ZenodoVersion:
-    snapshot_date: pendulum.DateTime
+    release_date: pendulum.DateTime
     download_url: str
 
     def to_dict(self) -> Dict:
-        return {"snapshot_date": self.snapshot_date.strftime("%Y-%m-%d"), "download_url": self.download_url}
+        return {"release_date": self.release_date.strftime("%Y-%m-%d"), "download_url": self.download_url}
 
 
 @dataclasses.dataclass