我发现很难将Parquet地板文件装入Hive桌。我正在amazon emr cluster和spark上进行数据处理。但是我需要读取输出的Parquet文件来验证我的转换。我有以下模式的Parquet文件：

root
 |-- ATTR_YEAR: long (nullable = true)
 |-- afil: struct (nullable = true)
 |    |-- clm: struct (nullable = true)
 |    |    |-- amb: struct (nullable = true)
 |    |    |    |-- L: string (nullable = true)
 |    |    |    |-- cdTransRsn: string (nullable = true)
 |    |    |    |-- dist: struct (nullable = true)
 |    |    |    |    |-- T: string (nullable = true)
 |    |    |    |    |-- content: double (nullable = true)
 |    |    |    |-- dscStrchPurp: string (nullable = true)
 |    |    |-- amt: struct (nullable = true)
 |    |    |    |-- L: string (nullable = true)
 |    |    |    |-- T: string (nullable = true)
 |    |    |    |-- content: double (nullable = true)
 |    |    |-- amtTotChrg: double (nullable = true)
 |    |    |-- cdAccState: string (nullable = true)
 |    |    |-- cdCause: string (nullable = true)

如何使用这种模式创建配置单元外部表，并将Parquet文件加载到该配置单元表中进行分析？