spark学习--求DataFrema的交集、差集、并集

package test0816

import org.apache.spark.sql.SparkSession

/**
  * @author WGY
  */
object Test extends App {
  private val spark: SparkSession = SparkSession.builder().master("local[2]").appName("mysql").getOrCreate()
  //先声明两个DF
  val df1 = spark.createDataFrame(
    List((1, "zhangsan"), (2, "lisi"), (3, "wangwu"))).toDF("id", "name")
  df1.show()

  val df2 = spark.createDataFrame(
    List((1, "zhangsan"), (2, "lisi"), (4, "zhaoliu"))).toDF("id", "name")
  df2.show()
  
  //差集
  println("差集")
  df1.except(df2).show()//取df1中有df2中没有的
  df2.except(df1).show()//取df2中有df1中没有的
  //交集
  println("交集")
  df1.intersect(df2).show()
  //并集
  println("并集")
  df1.union(df2).show()
  
  spark.close()
}