Spark自定义函数UDF UDAF UDTF编写

一、Spark自定义函数UDF

import org.apache.spark.sql.{Row, SparkSession}
import org.apache.spark.{SparkConf, SparkContext}

object UDFDemo {
//创建样例类
  case class Hobbies(name: String, hobbies: String)

  def main(args: Array[String]): Unit = {
//建立连接
    val conf = new SparkConf().setAppName("Function").setMaster("local[*]")
    val spark = SparkSession.builder().config(conf).getOrCreate()
    val sc = spark.sparkContext

//需要手动导入一个隐式转换，否则RDD无法转换成DF
    import spark.implicits._
//导入文件
    val rdd = sc.textFile("in/hobbies.txt")
    //根据文件情况进行分割，然后分组，转换为DataFrame
    val hobbyDF = rdd.map(_.split(" "))
      .map(x => Hobbies(x(0), x(1))).toDF()

    //注册为临时表
    hobbyDF.createOrReplaceTempView("hobby")
    //注册自定义函数
    spark.udf.register("hobby_num", (x: String) => x.split(",").size )
    //.show(false)字段不隐藏
    spark.sql("select name,hobbies,hobby_num(hobbies) as hobbyNum from hobby").show(false)

  }

}

---

运行结果如下

+-----+----------------------+--------+
|name |hobbies               |hobbyNum|
+-----+----------------------+--------+
|alice|jogging,Coding,cooking|3       |
|lina |travel,dance          |2       |
+-----+----------------------+--------+

---

文档 hobbies.txt内容

alice jogging,Coding,cooking
lina travel,dance

### 二、Spark自定义函数UDAF