fix: avoid filters containing null value

superset/connectors/base/models.py

@@ -419,7 +419,9 @@ def query(self, query_obj: QueryObjectDict) -> QueryResult:
         """
         raise NotImplementedError()
 
-    def values_for_column(self, column_name: str, limit: int = 10000) -> List[Any]:
+    def values_for_column(
+        self, column_name: str, limit: int = 10000, contain_null: bool = True,
+    ) -> List[Any]:
         """Given a column, returns an iterable of distinct values
 
         This is used to populate the dropdown showing a list of

superset/connectors/druid/models.py

@@ -948,7 +948,9 @@ def metrics_and_post_aggs(
         )
         return aggs, post_aggs
 
-    def values_for_column(self, column_name: str, limit: int = 10000) -> List[Any]:
+    def values_for_column(
+        self, column_name: str, limit: int = 10000, contain_null: bool = True,
+    ) -> List[Any]:
         """Retrieve some values for the given column"""
         logger.info(
             "Getting values for columns [{}] limited to [{}]".format(column_name, limit)

superset/connectors/sqla/models.py

@@ -173,7 +173,9 @@ def query(self, query_obj: QueryObjectDict) -> QueryResult:
     def get_query_str(self, query_obj: QueryObjectDict) -> str:
         raise NotImplementedError()
 
-    def values_for_column(self, column_name: str, limit: int = 10000) -> List[Any]:
+    def values_for_column(
+        self, column_name: str, limit: int = 10000, contain_null: bool = True,
+    ) -> List[Any]:
         raise NotImplementedError()
 
 
@@ -712,7 +714,9 @@ def get_fetch_values_predicate(self) -> TextClause:
                 )
             ) from ex
 
-    def values_for_column(self, column_name: str, limit: int = 10000) -> List[Any]:
+    def values_for_column(
+        self, column_name: str, limit: int = 10000, contain_null: bool = True,
+    ) -> List[Any]:
         """Runs query against sqla to retrieve some
         sample values for the given column.
         """
@@ -728,6 +732,9 @@ def values_for_column(self, column_name: str, limit: int = 10000) -> List[Any]:
         if limit:
             qry = qry.limit(limit)
 
+        if not contain_null:
+            qry = qry.where(target_col.get_sqla_col().isnot(None))
+
         if self.fetch_values_predicate:
             qry = qry.where(self.get_fetch_values_predicate())
 

superset/views/core.py

@@ -921,7 +921,9 @@ def filter(  # pylint: disable=no-self-use
         datasource.raise_for_access()
         row_limit = apply_max_row_limit(config["FILTER_SELECT_ROW_LIMIT"])
         payload = json.dumps(
-            datasource.values_for_column(column, row_limit),
+            datasource.values_for_column(
+                column_name=column, limit=row_limit, contain_null=False,
+            ),
             default=utils.json_int_dttm_ser,
             ignore_nan=True,
         )

tests/integration_tests/sqla_models_tests.py

@@ -463,3 +463,23 @@ def test_labels_expected_on_mutated_query(self):
         db.session.delete(table)
         db.session.delete(database)
         db.session.commit()
+
+    def test_values_for_column(self):
+        table = SqlaTable(
+            table_name="test_null_in_column",
+            sql="SELECT 'foo' as foo UNION SELECT 'bar' UNION SELECT NULL",
+            database=get_example_database(),
+        )
+        TableColumn(column_name="foo", type="VARCHAR(255)", table=table)
+
+        with_null = table.values_for_column(
+            column_name="foo", limit=10000, contain_null=True
+        )
+        assert None in with_null
+        assert len(with_null) == 3
+
+        without_null = table.values_for_column(
+            column_name="foo", limit=10000, contain_null=False
+        )
+        assert None not in without_null
+        assert len(without_null) == 2