fix: guard potential null derefs and remove dead branches (#42358)

Co-authored-by: Enzo Martellucci <52219496+EnxDev@users.noreply.github.com>

.github/actions/setup-backend/action.yml

@@ -5,6 +5,10 @@ inputs:
     description: 'Python version to set up. Accepts a version number, "current", or "next".'
     required: true
     default: 'current'
+  cache:
+    description: 'Cache dependencies. Options: pip'
+    required: false
+    default: 'pip'
   requirements-type:
     description: 'Type of requirements to install. Options: base, development, default'
     required: false
@@ -39,24 +43,17 @@ runs:
       uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
       with:
         python-version: ${{ steps.set-python-version.outputs.python-version }}
-    - name: Install uv
-      if: inputs.install-superset == 'true'
-      uses: astral-sh/setup-uv@c771a70e6277c0a99b617c7a806ffedaca235ff9 # v9.0.0
-      with:
-        python-version: ${{ steps.set-python-version.outputs.python-version }}
-        enable-cache: true
-    - name: Install apt packages
-      if: inputs.install-superset == 'true'
-      uses: awalsh128/cache-apt-pkgs-action@553a35bb8ebd9fcabcb1c9451aa4c98e1b4ca8a9 # v1.6.3
-      with:
-        packages: libldap2-dev libsasl2-dev
-        version: 1.0
+        cache: ${{ inputs.cache }}
     - name: Install dependencies
       env:
         INPUT_INSTALL_SUPERSET: ${{ inputs.install-superset }}
         INPUT_REQUIREMENTS_TYPE: ${{ inputs.requirements-type }}
       run: |
         if [ "$INPUT_INSTALL_SUPERSET" = "true" ]; then
+          sudo apt-get update && sudo apt-get -y install libldap2-dev libsasl2-dev
+
+          pip install --upgrade pip setuptools wheel uv
+
           if [ "$INPUT_REQUIREMENTS_TYPE" = "dev" ]; then
             uv pip install --system -r requirements/development.txt
           elif [ "$INPUT_REQUIREMENTS_TYPE" = "base" ]; then

.github/workflows/bump-python-package.yml

@@ -45,10 +45,7 @@ jobs:
           python-version: "3.11"
 
       - name: Install uv
-        uses: astral-sh/setup-uv@c771a70e6277c0a99b617c7a806ffedaca235ff9 # v9.0.0
-        with:
-          python-version: "3.11"
-          enable-cache: true
+        run: pip install uv
 
       - name: supersetbot bump-python -p "${{ github.event.inputs.package }}"
         env:

.github/workflows/superset-docs-deploy.yml

@@ -76,10 +76,7 @@ jobs:
           distribution: "zulu"
           java-version: "21"
       - name: Install Graphviz
-        uses: awalsh128/cache-apt-pkgs-action@553a35bb8ebd9fcabcb1c9451aa4c98e1b4ca8a9 # v1.6.3
-        with:
-          packages: graphviz
-          version: 1.0
+        run: sudo apt-get install -y graphviz
       - name: Compute Entity Relationship diagram (ERD)
         env:
           SUPERSET_SECRET_KEY: not-a-secret

.github/workflows/superset-translations.yml

@@ -78,10 +78,7 @@ jobs:
 
       - name: Install gettext tools
         if: steps.check.outputs.python == 'true' || steps.check.outputs.frontend == 'true'
-        uses: awalsh128/cache-apt-pkgs-action@553a35bb8ebd9fcabcb1c9451aa4c98e1b4ca8a9 # v1.6.3
-        with:
-          packages: gettext
-          version: 1.0
+        run: sudo apt-get update && sudo apt-get install -y gettext
 
       # Fetch the base ref so we can compare PR-introduced regressions
       # against a fair baseline (also runs babel_update against the base

superset/common/query_context_factory.py

@@ -318,7 +318,7 @@ class QueryContextFactory:  # pylint: disable=too-few-public-methods
             # another temporal filter. A new filter based on the value of
             # the granularity will be added later in the code.
             # In practice, this is replacing the previous default temporal filter.
-            if is_adhoc_column(filter_to_remove):  # type: ignore
+            if filter_to_remove and is_adhoc_column(filter_to_remove):  # type: ignore
                 filter_to_remove = filter_to_remove.get("sqlExpression")
 
             if filter_to_remove:

superset/common/utils/query_cache_manager.py

@@ -206,11 +206,9 @@ class QueryCacheManager:
                 )
                 query_cache.status = QueryStatus.SUCCESS
                 query_cache.is_loaded = True
-                query_cache.is_cached = cache_value is not None
+                query_cache.is_cached = True
                 query_cache.sql_rowcount = cache_value.get("sql_rowcount", None)
-                query_cache.cache_dttm = (
-                    cache_value["dttm"] if cache_value is not None else None
-                )
+                query_cache.cache_dttm = cache_value["dttm"]
                 query_cache.queried_dttm = cache_value.get(
                     "queried_dttm", cache_value.get("dttm")
                 )

superset/daos/dashboard.py

@@ -397,7 +397,8 @@ class DashboardDAO(BaseDAO[Dashboard]):
             md["color_namespace"] = data.get("color_namespace")
 
         md["expanded_slices"] = data.get("expanded_slices", {})
-        md["refresh_frequency"] = data.get("refresh_frequency", 0)
+        if "refresh_frequency" in data:
+            md["refresh_frequency"] = data["refresh_frequency"]
         md["color_scheme"] = data.get("color_scheme", "")
         md["label_colors"] = data.get("label_colors", {})
         md["shared_label_colors"] = data.get("shared_label_colors", [])

superset/db_engine_specs/databricks.py

@@ -534,6 +534,7 @@ class DatabricksDynamicBaseEngineSpec(BasicParametersMixin, DatabricksBaseEngine
         ],
     ) -> list[SupersetError]:
         errors: list[SupersetError] = []
+        connect_args: dict[str, Any] = {}
         if extra := json.loads(properties.get("extra")):  # type: ignore
             engine_params = extra.get("engine_params", {})
             connect_args = engine_params.get("connect_args", {})

superset/db_engine_specs/presto.py

@@ -1235,6 +1235,7 @@ class PrestoEngineSpec(PrestoBaseEngineSpec):
         all_columns: list[ResultSetColumnType] = []
         expanded_columns = []
         current_array_level = None
+        unnested_rows: dict[int, int] = defaultdict(int)
         while to_process:
             column, level = to_process.popleft()
             if column["column_name"] not in [
@@ -1248,7 +1249,7 @@ class PrestoEngineSpec(PrestoBaseEngineSpec):
             # added by the first. every time we change a level in the nested arrays
             # we reinitialize this.
             if level != current_array_level:
-                unnested_rows: dict[int, int] = defaultdict(int)
+                unnested_rows = defaultdict(int)
                 current_array_level = level
 
             name = column["column_name"]

tests/unit_tests/common/test_query_context_factory.py

@@ -532,6 +532,42 @@ class TestQueryContextFactory:
 
         assert query_object.columns == ["ds", "other_col"]
 
+    def test_apply_granularity_no_filter_to_remove(self):
+        """No x-axis and no temporal filters leaves the filters untouched."""
+        query_object = Mock(spec=QueryObject)
+        query_object.granularity = "P1D"
+        query_object.columns = ["other_col"]
+        query_object.post_processing = []
+        query_object.filter = [{"col": "other_col", "op": "==", "val": "value"}]
+
+        datasource = Mock()
+        datasource.columns = [{"column_name": "ds", "is_dttm": True}]
+
+        self.factory._apply_granularity(query_object, {}, datasource)
+
+        assert query_object.filter == [{"col": "other_col", "op": "==", "val": "value"}]
+
+    def test_apply_granularity_with_adhoc_temporal_filter(self):
+        """An adhoc temporal filter is matched on its SQL expression."""
+        adhoc_column = {"label": "ds_expr", "sqlExpression": "DATE(ds)"}
+        query_object = Mock(spec=QueryObject)
+        query_object.granularity = "P1D"
+        query_object.columns = ["other_col"]
+        query_object.post_processing = []
+        query_object.filter = [
+            {"col": adhoc_column, "op": "TEMPORAL_RANGE", "val": "a : b"},
+            {"col": "DATE(ds)", "op": "TEMPORAL_RANGE", "val": "a : b"},
+        ]
+
+        datasource = Mock()
+        datasource.columns = [{"column_name": "ds", "is_dttm": True}]
+
+        self.factory._apply_granularity(query_object, {}, datasource)
+
+        assert query_object.filter == [
+            {"col": adhoc_column, "op": "TEMPORAL_RANGE", "val": "a : b"}
+        ]
+
     def test_apply_filters_with_time_range(self):
         """Test _apply_filters with time_range"""
         query_object = Mock(spec=QueryObject)

tests/unit_tests/dao/dashboard_test.py

@@ -24,6 +24,7 @@ from superset.connectors.sqla.models import Database, SqlaTable
 from superset.daos.dashboard import DashboardDAO
 from superset.models.dashboard import Dashboard
 from superset.models.slice import Slice
+from superset.utils import json
 from tests.unit_tests.conftest import with_feature_flags
 
 
@@ -117,3 +118,53 @@ def test_set_dash_metadata_preserves_soft_deleted_members(
     )
     # And the position slot kept its UUID rather than being nulled.
     assert positions["CHART-trashed"]["meta"]["uuid"] == str(trashed_chart.uuid)
+
+
+def test_set_dash_metadata_preserves_refresh_frequency(session: Session) -> None:
+    """set_dash_metadata must not reset refresh_frequency when absent from data.
+
+    Regression test for #42116: ``data.get("refresh_frequency", 0)`` would
+    unconditionally overwrite the existing value with 0 whenever the caller
+    did not include ``refresh_frequency`` in the data dict.
+    """
+    Dashboard.metadata.create_all(session.get_bind())
+
+    dashboard = Dashboard(
+        dashboard_title="refresh_test_dash",
+        json_metadata=json.dumps({"refresh_frequency": 30}),
+    )
+    db.session.add(dashboard)
+    db.session.flush()
+
+    # Simulate a save that does NOT include refresh_frequency
+    # (e.g. changing only the title via the PropertiesModal).
+    DashboardDAO.set_dash_metadata(dashboard, {"color_scheme": "superset"})
+
+    md = json.loads(dashboard.json_metadata)
+    assert md["refresh_frequency"] == 30, (
+        "refresh_frequency should be preserved when not present in data"
+    )
+
+
+def test_set_dash_metadata_updates_refresh_frequency_when_present(
+    session: Session,
+) -> None:
+    """set_dash_metadata must update refresh_frequency when it IS in data."""
+    Dashboard.metadata.create_all(session.get_bind())
+
+    dashboard = Dashboard(
+        dashboard_title="refresh_test_dash_2",
+        json_metadata=json.dumps({"refresh_frequency": 30}),
+    )
+    db.session.add(dashboard)
+    db.session.flush()
+
+    # Simulate a save that explicitly sets refresh_frequency to 0.
+    DashboardDAO.set_dash_metadata(
+        dashboard, {"refresh_frequency": 0, "color_scheme": "superset"}
+    )
+
+    md = json.loads(dashboard.json_metadata)
+    assert md["refresh_frequency"] == 0, (
+        "refresh_frequency should be updated when present in data"
+    )

tests/unit_tests/result_set_test.py

@@ -571,3 +571,53 @@ def test_stringify_values_non_serializable_dict_falls_back_to_str() -> None:
     # Must not raise — falls back to str()
     result = stringify_values(data)
     assert result[0] == str({"key": _Unserializable()})
+
+
+def test_empty_result_set_preserves_column_metadata() -> None:
+    """
+    Test that column metadata is preserved when query returns zero rows.
+
+    When a query returns no data but has a valid cursor description, the
+    column names and types from cursor_description should be preserved
+    in the result set. This allows downstream consumers (like the UI)
+    to display column headers even for empty result sets.
+    """
+    data: DbapiResult = []
+    description = [
+        ("id", "int", None, None, None, None, True),
+        ("name", "varchar", None, None, None, None, True),
+        ("created_at", "timestamp", None, None, None, None, True),
+    ]
+
+    result_set = SupersetResultSet(
+        data,
+        description,  # type: ignore
+        BaseEngineSpec,
+    )
+
+    # Verify column count
+    assert len(result_set.columns) == 3
+
+    # Verify column names are preserved
+    column_names = [col["column_name"] for col in result_set.columns]
+    assert column_names == ["id", "name", "created_at"]
+
+    assert result_set.columns[0]["type"] == BaseEngineSpec.get_datatype(
+        description[0][1]
+    )
+    assert result_set.columns[1]["type"] == BaseEngineSpec.get_datatype(
+        description[1][1]
+    )
+    assert result_set.columns[2]["type"] == BaseEngineSpec.get_datatype(
+        description[2][1]
+    )
+
+    # Verify the PyArrow table has the correct schema
+    assert result_set.table.num_rows == 0
+    assert len(result_set.table.column_names) == 3
+    assert list(result_set.table.column_names) == ["id", "name", "created_at"]
+
+    # Verify DataFrame conversion works
+    df = result_set.to_pandas_df()
+    assert len(df) == 0
+    assert list(map(str, df.columns)) == ["id", "name", "created_at"]