add is_oa (is_openaccess) column to base-studies (#1199)

Author: jdkent

store/backend/neurostore/models/data.py

@@ -189,6 +189,7 @@ class BaseStudy(BaseMixin, db.Model):
     public = db.Column(db.Boolean, default=True)
     level = db.Column(db.String)
     metadata_ = db.Column(JSONB)
+    is_oa = db.Column(db.Boolean, default=None, nullable=True)
     has_coordinates = db.Column(db.Boolean, default=False, nullable=False)
     has_images = db.Column(db.Boolean, default=False, nullable=False)
     user_id = db.Column(db.Text, db.ForeignKey("users.external_id"), index=True)

store/backend/neurostore/openapi

@@ -964,6 +964,7 @@ class BaseStudiesView(ObjectView, ListView):
         "flat": fields.Boolean(load_default=False),
         "info": fields.Boolean(load_default=False),
         "data_type": fields.String(load_default=None),
+        "is_oa": fields.Boolean(load_default=None, allow_none=True),
         "feature_filter": fields.List(fields.String(load_default=None)),
         "pipeline_config": fields.List(fields.String(load_default=None)),
         "feature_display": fields.List(fields.String(load_default=None)),
@@ -1187,6 +1188,11 @@ def view_search(self, q, args):
                         self._model.has_images.is_(True),
                     ),
                 )
+        is_oa = args.get("is_oa", None)
+        if is_oa is not None and not isinstance(is_oa, bool):
+            abort_validation("is_oa must be a boolean.")
+        if is_oa is not None:
+            q = q.filter(self._model.is_oa.is_(is_oa))
         # filter by year range
         year_min = args.get("year_min")
         year_max = args.get("year_max")

store/backend/neurostore/resources/data.py

@@ -462,6 +462,7 @@ class BaseStudySchema(BaseDataSchema):
     authors = fields.String(allow_none=True)
     year = fields.Integer(allow_none=True)
     level = fields.String(allow_none=True)
+    is_oa = fields.Boolean(allow_none=True)
     versions = StringOrNested("StudySchema", many=True)
     features = fields.Method("get_features")
     ace_fulltext = fields.String(load_only=True, allow_none=True)

store/backend/neurostore/schemas/data.py

@@ -420,6 +420,64 @@ def test_has_coordinates_images(auth_client, session):
     assert base_study_2.has_images is False
 
 
+def test_filter_by_is_oa(auth_client, session):
+    base_true = BaseStudy(
+        name="Open Access Study",
+        doi="10.1234/oa-study",
+        pmid="111111",
+        is_oa=True,
+        public=True,
+        level="group",
+    )
+    base_false = BaseStudy(
+        name="Closed Access Study",
+        doi="10.1234/closed-study",
+        pmid="222222",
+        is_oa=False,
+        public=True,
+        level="group",
+    )
+    base_unknown = BaseStudy(
+        name="Unknown Access Study",
+        doi="10.1234/unknown-study",
+        pmid="333333",
+        is_oa=None,
+        public=True,
+        level="group",
+    )
+    session.add_all([base_true, base_false, base_unknown])
+    session.commit()
+
+    assert session.query(BaseStudy).count() == 3
+
+    all_resp = auth_client.get("/api/base-studies/")
+    assert all_resp.status_code == 200
+    all_names = {result["name"] for result in all_resp.json()["results"]}
+    assert {
+        "Open Access Study",
+        "Closed Access Study",
+        "Unknown Access Study",
+    }.issubset(all_names)
+
+    true_resp = auth_client.get("/api/base-studies/?is_oa=true")
+    assert true_resp.status_code == 200
+    true_results = true_resp.json()["results"]
+    assert all(result["is_oa"] is True for result in true_results)
+    true_names = {result["name"] for result in true_results}
+    assert "Open Access Study" in true_names
+    assert "Closed Access Study" not in true_names
+    assert "Unknown Access Study" not in true_names
+
+    false_resp = auth_client.get("/api/base-studies/?is_oa=false")
+    assert false_resp.status_code == 200
+    false_results = false_resp.json()["results"]
+    assert all(result["is_oa"] is False for result in false_results)
+    false_names = {result["name"] for result in false_results}
+    assert "Closed Access Study" in false_names
+    assert "Open Access Study" not in false_names
+    assert "Unknown Access Study" not in false_names
+
+
 def test_config_and_feature_filters(auth_client, ingest_demographic_features, session):
     """Test filtering by both config args and feature results with version specification"""
     # Test combined feature and config filtering