When the Spark engine runs a mapping that reads from an Azure SQL data warehouse source, control characters are being added to the data, resulting in inaccurate mapping results.