make sure empty strings cannot be added to doi and pmid (#735)

Author: jdkent

store/neurostore/models/data.py

@@ -178,6 +178,8 @@ class BaseStudy(BaseMixin, db.Model):
     __table_args__ = (
         db.CheckConstraint(level.in_(["group", "meta"])),
         db.UniqueConstraint("doi", "pmid", name="doi_pmid"),
+        db.CheckConstraint("pmid ~ '^(?=.*\\S).+$' OR name IS NULL"),
+        db.CheckConstraint("doi ~ '^(?=.*\\S).+$' OR name IS NULL"),
         sa.Index("ix_base_study___ts_vector__", _ts_vector, postgresql_using="gin"),
     )
 
@@ -280,6 +282,8 @@ class Study(BaseMixin, db.Model):
 
     __table_args__ = (
         db.CheckConstraint(level.in_(["group", "meta"])),
+        db.CheckConstraint("pmid ~ '^(?=.*\\S).+$' OR name IS NULL"),
+        db.CheckConstraint("doi ~ '^(?=.*\\S).+$' OR name IS NULL"),
         sa.Index("ix_study___ts_vector__", _ts_vector, postgresql_using="gin"),
     )
 

store/neurostore/resources/data.py

@@ -457,8 +457,13 @@ def post(self):
         )
         hashed_results = {(bs.doi or "") + (bs.pmid or ""): bs for bs in results}
         for study_data in data:
-            lookup_hash = study_data.get("doi", "") + study_data.get("pmid", "")
-            record = hashed_results.get(lookup_hash)
+            doi = study_data.get("doi", "") or ""
+            pmid = study_data.get("pmid", "") or ""
+            lookup_hash = doi + pmid
+            if lookup_hash == "" or lookup_hash.isspace():
+                record = None
+            else:
+                record = hashed_results.get(lookup_hash)
             if record is None:
                 with db.session.no_autoflush:
                     record = self.__class__.update_or_create(study_data, flush=False)

store/neurostore/schemas/data.py

@@ -331,6 +331,15 @@ class Meta:
             "level",
         )
 
+    @pre_load
+    def check_nulls(self, data, **kwargs):
+        """ensure data is not empty string or whitespace"""
+        for attr in ["pmid", "pmcid", "doi"]:
+            val = data.get(attr, None)
+            if val is not None and (val == "" or val.isspace()):
+                data[attr] = None
+        return data
+
 
 class StudySchema(BaseDataSchema):
     metadata = fields.Dict(attribute="metadata_", dump_only=True)
@@ -378,6 +387,15 @@ class Meta:
             "level",
         )
 
+    @pre_load
+    def check_nulls(self, data, **kwargs):
+        """ensure data is not empty string or whitespace"""
+        for attr in ["pmid", "pmcid", "doi"]:
+            val = data.get(attr, None)
+            if val is not None and (val == "" or val.isspace()):
+                data[attr] = None
+        return data
+
 
 class StudysetSchema(BaseDataSchema):
     # serialize

store/neurostore/tests/api/test_base_studies.py

@@ -24,6 +24,11 @@ def test_post_list_of_studies(auth_client, ingest_neuroquery):
         {
             "name": "another new name",
         },
+        {
+            "doi": "",
+            "pmid": "",
+            "name": "no ids",
+        }
     ]
 
     result = auth_client.post("/api/base-studies/", data=test_input)

store/neurostore/tests/api/test_studies.py

@@ -126,6 +126,8 @@ def test_clone_studies_with_data(auth_client, ingest_neurosynth, session):
     study_data["analyses"].append(
         {"name": "new analysis", "points": second_analysis_points}
     )
+    study_data["pmid"] = ""
+    study_data["doi"] = ""
 
     resp = auth_client.post(
         f"/api/studies/?source_id={study_entry.id}",
@@ -138,6 +140,8 @@ def test_clone_studies_with_data(auth_client, ingest_neurosynth, session):
     assert data["analyses"][0]["points"][0]["coordinates"] == [0, 0, 0]
     assert data["analyses"][1]["points"][0]["coordinates"] == [0, 0, 0]
     assert "new analysis" in [a["name"] for a in data["analyses"]]
+    assert data["pmid"] is None
+    assert data["doi"] is None
 
 
 def test_private_studies(user_data, auth_clients, session):