Handle empty dataframes in TableViz

TableViz fails to display empty dataframes returning an error like: "None of [Index(['project', 'count'], dtype='object')] are in the [columns]" The behavior has been observed while testing 0.36.0rc3 with Druid datasources. issue: apache#9468
elukey · Apr 6, 2020 · 19a4425 · 19a4425
1 parent c0807c1
commit 19a4425
Showing 1 changed file with 12 additions and 11 deletions.
diff --git a/superset/viz.py b/superset/viz.py
@@ -625,17 +625,18 @@ def get_data(self, df: pd.DataFrame) -> VizData:
             self.form_data.get("percent_metrics") or []
         )
 
-        df = pd.concat(
-            [
-                df[non_percent_metric_columns],
-                (
-                    df[percent_metric_columns]
-                    .div(df[percent_metric_columns].sum())
-                    .add_prefix("%")
-                ),
-            ],
-            axis=1,
-        )
+        if not df.empty:
+            df = pd.concat(
+                [
+                    df[non_percent_metric_columns],
+                    (
+                        df[percent_metric_columns]
+                        .div(df[percent_metric_columns].sum())
+                        .add_prefix("%")
+                    ),
+                ],
+                axis=1,
+            )
 
         data = self.handle_js_int_overflow(
             dict(records=df.to_dict(orient="records"), columns=list(df.columns))