将数据集<Row>转换为 JavaRDD<Row> 然后转换为数据帧时的运行时异常

Question

我正试图使用下面的代码将一个索引列添加到数据集，该代码将其转换为JavaPairRDD。

// ds is a Dataset<Row>
JavaPairRDD<Row, Long> indexedRDD = ds.toJavaRDD()
.zipWithIndex();
// Now I am converting JavaPairRDD to JavaRDD as below.
JavaRDD<Row> rowRDD = indexedRDD
.map(tuple -> RowFactory.create(tuple._1(),tuple._2().intValue()));
// I am converting the RDD back to dataframe and it doesnt work.
Dataset<Row> authDf = session
.createDataFrame(rowRDD, ds.schema().add("ID", DataTypes.IntegerType));
// Below is the ds schema(Before adding the ID column).
ds.schema()
root
|-- user: short (nullable = true)
|-- score: long (nullable = true)
|-- programType: string (nullable = true)
|-- source: string (nullable = true)
|-- item: string (nullable = true)
|-- playType: string (nullable = true)
|-- userf: integer (nullable = true)

上面的代码抛出以下错误消息：

**Job aborted due to stage failure: Task 0 in stage 21.0 failed 4 
times, most  recent failure: Lost task 0.3 in stage 21.0 (TID 658, 
sl73caehdn0406.visa.com, executor 1):
java.lang.RuntimeException: 
Error while encoding: java.lang.RuntimeException: 
org.apache.spark.sql.catalyst.expressions.GenericRowWithSchema is not 
a valid external type for schema of smallint**

Answer 1

您在第二条语句中创建的元组由两列组成：一列是对象(由初始数据集中的所有colmn组成(，另一列是整数。第二个元组列进入类型为long的第二个结果列。第一个元组列进入第一个结果列，这是一个类型短的对象，即GenericRowWithSchema，这会导致错误。

您应该使用7个参数来创建RowFactory.create((，每个参数对应一个结果列。