在分区循环中获取SparkSession

Question

我正在使用Java-Spark，我试图在SparkSession（rdd）中获得foreachPartition，我正在尝试的是：'

public static void main(String[] args) {

SparkConf conf = new SparkConf().setAppName("myappName").setMaster("mymaster");

Integer duration = 2000;

JavaStreamingContext ssc = new JavaStreamingContext(conf, new Duration(duration));

//Get Kafka read message as stream
JavaInputDSteram<ConsumerRecord<Object,String>> stream = KafkaUtils.createDirectStream(ssc, LocationStrategies.PreferConsistent(),
    ConsumerStrategies.Subscribe(inputTopics, inputKafkaParams));

    stream.foreachRdd(rdd -> {

        //my logic
        rdd.foreachPartition(iterator ->{
            SparkSession sparkSession = SparkSession.builder().config(conf).getOrCreate(); // null
            Dataset<String ds = sparkSession.createDataset(myJsonList /*ArrayList<String>*/, org.apache.spark.sql.Encoders.STRING());
        );


    );

 }

但以下行：SparkSession sparkSession = SparkSession.builder().config(conf).getOrCreate(); 返回始终为null， 如何在分区循环中获取SparkSession？我只想使用SparkSQL在JSON上做一些逻辑并获取DataSet等，

我该如何解决？

*也许我输入了代码

谢谢！