fix(b2b): return blank learner names as null

src/ol_analytics_api/tenants/b2b_dashboard/learner_queries.py

@@ -45,6 +45,10 @@
     " END"
 )
 
+# Upstream stores "" rather than NULL for learners who never set a name. Null
+# blank names so they sort with the missing ones instead of before every name.
+_BLANK_AS_NULL_NAME = "NULLIF(TRIM(full_name), '')"
+
 # The four branches of _COMPLETION_STATUS. Mutually exclusive, so these buckets
 # never overlap; a row with withheld outcomes falls into none of them.
 _STATUSES = ("not_started", "in_progress", "passed", "certified")
@@ -118,7 +122,8 @@ def learner_progress(filters: ProgressFilters) -> ProgressQuery:
     if not filters.include_inactive:
         scope.append("enrollment_is_active = TRUE")
     records = (
-        "SELECT user_pk, courserun_pk, user_global_id AS learner_id, email, full_name,"  # noqa: S608
+        "SELECT user_pk, courserun_pk, user_global_id AS learner_id, email,"  # noqa: S608
+        f" {_BLANK_AS_NULL_NAME} AS full_name,"
         " courserun_readable_id, courserun_title, courserun_start_on, courserun_end_on,"
         " enrollment_created_on AS enrolled_on, enrollment_is_active, enrollment_mode,"
         f" {_COMPLETION_STATUS} AS completion_status, is_passing, grade_value AS grade,"
@@ -149,8 +154,8 @@ def learner_progress(filters: ProgressFilters) -> ProgressQuery:
     where = f" WHERE {' AND '.join(predicates)}" if predicates else ""
 
     direction = "DESC" if filters.descending else "ASC"
-    # Nulls last either way (full_name is often null), then a unique tie-break
-    # so LIMIT/OFFSET paging is deterministic.
+    # Nulls last either way (full_name is often null, and `records` nulls blank
+    # ones), then a unique tie-break so LIMIT/OFFSET paging is deterministic.
     order_by = (
         f"{filters.sort.value} IS NULL, {filters.sort.value} {direction}, user_pk, courserun_pk"
     )

src/ol_analytics_api/tenants/b2b_learner_records/queries.py

@@ -109,6 +109,10 @@ def _outcomes_shared() -> str:
     "chatbot_interactions",
 )
 
+# Upstream stores "" rather than NULL for learners who never set a name. Null
+# blank names so they sort with the missing ones instead of before every name.
+_BLANK_AS_NULL_NAME = "NULLIF(TRIM(full_name), '')"
+
 
 @dataclass(frozen=True)
 class RecordFilters:
@@ -248,7 +252,8 @@ def enrollments(schema: str, filters: RecordFilters) -> RecordQuery:
         scope.append("courserun_readable_id = %s")
         scope_params.append(filters.courserun_id)
     records = (
-        "SELECT user_pk, user_global_id AS learner_id, email, full_name,"  # noqa: S608
+        "SELECT user_pk, user_global_id AS learner_id, email,"  # noqa: S608
+        f" {_BLANK_AS_NULL_NAME} AS full_name,"
         " sso_organization_id AS organization_id, contract_id,"
         " b2b_contract_name AS contract_name, courserun_readable_id AS courserun_id,"
         " courserun_title, courserun_start_on, courserun_end_on,"
@@ -305,7 +310,8 @@ def learners(schema: str, filters: RecordFilters) -> RecordQuery:
     schema = validate_sql_identifier(schema)
     if filters.contract_id is None and not filters.include_inactive:
         records = (
-            "SELECT user_pk, user_global_id AS learner_id, email, full_name,"  # noqa: S608
+            "SELECT user_pk, user_global_id AS learner_id, email,"  # noqa: S608
+            f" {_BLANK_AS_NULL_NAME} AS full_name,"
             " sso_organization_id AS organization_id, organization_name, membership_source,"
             " is_organization_manager, first_enrolled_on, last_enrolled_on, courses_enrolled,"
             " courses_passed, courses_certified,"
@@ -376,7 +382,8 @@ def _recomputed_learners(schema: str, filters: RecordFilters) -> tuple[str, list
 
     enrollment_rollup = (
         "SELECT user_pk, MAX(user_global_id) AS user_global_id, MAX(email) AS email,"  # noqa: S608
-        " MAX(full_name) AS full_name, MAX(sso_organization_id) AS sso_organization_id,"
+        f" MAX({_BLANK_AS_NULL_NAME}) AS full_name,"
+        " MAX(sso_organization_id) AS sso_organization_id,"
         " MAX(organization_name) AS organization_name,"
         f" MIN({enrolled_on}) AS first_enrolled_on,"
         f" MAX({enrolled_on}) AS last_enrolled_on,"
@@ -400,7 +407,8 @@ def _recomputed_learners(schema: str, filters: RecordFilters) -> tuple[str, list
     records = (
         "SELECT COALESCE(l.user_pk, e.user_pk) AS user_pk,"  # noqa: S608
         " COALESCE(l.user_global_id, e.user_global_id) AS learner_id,"
-        " COALESCE(l.email, e.email) AS email, COALESCE(l.full_name, e.full_name) AS full_name,"
+        " COALESCE(l.email, e.email) AS email,"
+        " COALESCE(NULLIF(TRIM(l.full_name), ''), e.full_name) AS full_name,"
         " COALESCE(l.sso_organization_id, e.sso_organization_id) AS organization_id,"
         " COALESCE(l.organization_name, e.organization_name) AS organization_name,"
         " CASE WHEN l.membership_source IN ('roster', 'both') AND e.courses_enrolled > 0"

tests/test_dashboard_learner_progress.py

@@ -252,6 +252,12 @@ async def test_sort_puts_nulls_last_with_a_unique_tie_break(app):
     )
 
 
+async def test_blank_names_read_as_null_so_they_sort_last(app):
+    pool = _FakePool()
+    await _get(app, pool)
+    assert "NULLIF(TRIM(full_name), '') AS full_name," in pool.page_call()[0]
+
+
 async def test_unknown_sort_key_is_rejected(app):
     response = await _get(app, _FakePool(), params={"sort": "grade"})
     assert response.status_code == 422

tests/test_learner_records.py

@@ -433,6 +433,17 @@ async def test_include_inactive_learners_keep_every_roster_member(app, monkeypat
     assert params == (ORG_ID, ORG_ID, 100, 0)
 
 
+@pytest.mark.parametrize(
+    "path",
+    ["enrollments", "learners", "learners?contract_id=42", "learners?include_inactive=true"],
+)
+async def test_blank_names_read_as_null(app, monkeypatch, path):
+    pool = _FakePool()
+    await _get(app, f"/organizations/{ORG_ID}/{path}", _partner_header(ORG_ID), pool, monkeypatch)
+    query, _ = pool.page_call()
+    assert "NULLIF(TRIM(full_name), '')" in query
+
+
 async def test_recomputed_learners_report_the_staler_view(app, monkeypatch):
     older = datetime.datetime(2026, 8, 12, 6, 0)  # noqa: DTZ001
     pool = _FakePool(as_of_by_mv={queries.ENROLLMENT_MV: older})