Merge pull request ClickHouse#87012 from ClickHouse/backport/25.8/86166

Backport ClickHouse#86166 to 25.8: Remove UB while parsing Iceberg partition transform expression

Author: clickhouse-gh[bot]

src/Storages/ObjectStorage/DataLakes/Iceberg/Utils.cpp

@@ -496,11 +496,15 @@ Poco::JSON::Object::Ptr getPartitionField(
     }
     else if (partition_function->name == "icebergTruncate")
     {
+        if (!param.has_value())
+            throw Exception(ErrorCodes::BAD_ARGUMENTS, "TRUNCATE function for iceberg partitioning requires one integer parameter");
         result->set(Iceberg::f_transform, fmt::format("truncate[{}]", *param));
         return result;
     }
     else if (partition_function->name == "icebergBucket")
     {
+        if (!param.has_value())
+            throw Exception(ErrorCodes::BAD_ARGUMENTS, "BUCKET function for iceberg partitioning requires one integer parameter");
         result->set(Iceberg::f_transform, fmt::format("bucket[{}]", *param));
         return result;
     }

tests/integration/test_storage_iceberg/test.py

@@ -2590,6 +2590,23 @@ def test_writes_with_compression_metadata(started_cluster, format_version, stora
     instance.query(f"INSERT INTO {TABLE_NAME} VALUES ('123', 1);", settings={"allow_experimental_insert_into_iceberg": 1, "iceberg_metadata_compression_method": "gzip"})
     assert instance.query(f"SELECT * FROM {TABLE_NAME} ORDER BY ALL") == '123\t1\n'
 
+def test_writes_create_table_bugs(started_cluster):
+    instance = started_cluster.instances["node1"]
+    TABLE_NAME = "test_writes_create_table_bugs_" + get_uuid_str()
+    TABLE_NAME_1 = "test_writes_create_table_bugs_" + get_uuid_str()
+    instance.query(
+        f"CREATE TABLE {TABLE_NAME} (c0 Int) ENGINE = IcebergLocal('/iceberg_data/default/{TABLE_NAME}/', 'CSV') AS (SELECT 1 OFFSET 1 ROW);",
+        settings={"allow_experimental_insert_into_iceberg": 1}
+    )
+
+    error = instance.query_and_get_error(
+        f"CREATE TABLE {TABLE_NAME_1} (c0 Int) ENGINE = IcebergLocal('/iceberg_data/default/{TABLE_NAME_1}/', 'CSV') PARTITION BY (icebergTruncate(c0));",
+        settings={"allow_experimental_insert_into_iceberg": 1}
+    )
+
+    assert "BAD_ARGUMENTS" in error
+    assert "LOGICAL_ERROR" not in error
+
 
 @pytest.mark.parametrize("format_version", [1, 2])
 @pytest.mark.parametrize("storage_type", ["local"])