chore(sql-lab): catch PyArrow deserialization error (#11201)

superset/exceptions.py

@@ -93,3 +93,7 @@ class QueryObjectValidationError(SupersetException):
 
 class DashboardImportException(SupersetException):
     pass
+
+
+class SerializationError(SupersetException):
+    pass

superset/views/core.py

@@ -72,6 +72,7 @@ from superset.databases.filters import DatabaseFilter
 from superset.exceptions import (
     CertificateException,
     DatabaseNotFound,
+    SerializationError,
     SupersetException,
     SupersetSecurityException,
     SupersetTimeoutException,
@@ -1961,7 +1962,9 @@ class Superset(BaseSupersetView):  # pylint: disable=too-many-public-methods
         return self.results_exec(key)
 
     @staticmethod
-    def results_exec(key: str) -> FlaskResponse:
+    def results_exec(  # pylint: disable=too-many-return-statements
+        key: str,
+    ) -> FlaskResponse:
         """Serves a key off of the results backend
 
         It is possible to pass the `rows` query argument to limit the number
@@ -1995,9 +1998,15 @@ class Superset(BaseSupersetView):  # pylint: disable=too-many-public-methods
             return json_errors_response([ex.error], status=403)
 
         payload = utils.zlib_decompress(blob, decode=not results_backend_use_msgpack)
-        obj = _deserialize_results_payload(
-            payload, query, cast(bool, results_backend_use_msgpack)
-        )
+        try:
+            obj = _deserialize_results_payload(
+                payload, query, cast(bool, results_backend_use_msgpack)
+            )
+        except SerializationError:
+            return json_error_response(
+                __("Data could not be deserialized. You may want to re-run the query."),
+                status=404,
+            )
 
         if "rows" in request.args:
             try:

superset/views/utils.py

@@ -33,7 +33,11 @@ import superset.models.core as models
 from superset import app, dataframe, db, is_feature_enabled, result_set
 from superset.connectors.connector_registry import ConnectorRegistry
 from superset.errors import ErrorLevel, SupersetError, SupersetErrorType
-from superset.exceptions import SupersetException, SupersetSecurityException
+from superset.exceptions import (
+    SerializationError,
+    SupersetException,
+    SupersetSecurityException,
+)
 from superset.legacy import update_time_range
 from superset.models.core import Database
 from superset.models.dashboard import Dashboard
@@ -521,7 +525,10 @@ def _deserialize_results_payload(
             ds_payload = msgpack.loads(payload, raw=False)
 
         with stats_timing("sqllab.query.results_backend_pa_deserialize", stats_logger):
-            pa_table = pa.deserialize(ds_payload["data"])
+            try:
+                pa_table = pa.deserialize(ds_payload["data"])
+            except pa.ArrowSerializationError:
+                raise SerializationError("Unable to deserialize table")
 
         df = result_set.SupersetResultSet.convert_table_to_df(pa_table)
         ds_payload["data"] = dataframe.df_to_records(df) or []