Cleaning temp S3 files for Pandas.read_sql_athena(ctas_approach=True)

Author: igorborgest

awswrangler/pandas.py

@@ -582,13 +582,16 @@ def _read_sql_athena_ctas(self,
         self._session.glue.delete_table_if_exists(database=database, table=name)
         manifest_path: str = f"{s3_output}/tables/{query_id}-manifest.csv"
         paths: List[str] = self._session.athena.extract_manifest_paths(path=manifest_path)
-        if not paths:
-            return pd.DataFrame()
         logger.debug(f"paths: {paths}")
-        return self.read_parquet(path=paths,
-                                 procs_cpu_bound=procs_cpu_bound,
-                                 wait_objects=True,
-                                 wait_objects_timeout=15.0)
+        if not paths:
+            df: pd.DataFrame = pd.DataFrame()
+        else:
+            df = self.read_parquet(path=paths,
+                                   procs_cpu_bound=procs_cpu_bound,
+                                   wait_objects=True,
+                                   wait_objects_timeout=15.0)
+        self._session.s3.delete_listed_objects(objects_paths=[manifest_path] + paths)
+        return df
 
     def _read_sql_athena_regular(self,
                                  sql: str,