SparkSQL之DSL风格的wordcount

object WordCount_DSL {
  def main(args: Array[String]): Unit = {
    //1.创建sparkSession
    val spark: SparkSession = SparkSession.builder().appName("wc").master("local[*]").getOrCreate()
    val sc: SparkContext = spark.sparkContext

    //2、读取文件
    val ds: Dataset[String] = spark.read.textFile("E:\\ideal_workspace\\spark\\day01\\words.txt")

    //3、对每一行的数据进行切割
    import spark.implicits._
    val wordDs: Dataset[String] = ds.flatMap(_.split(" "))
    //wordDs.show()

    //4.数据查询
    wordDs.groupBy("value").count().orderBy($"count".desc).show()

    spark.stop()
    sc.stop()
  }

}