2.Hive UDF

2.Hive UDF

2.1 Udf

UDF：进来一个出去一个，row mapping。是row级别操作开发

2.2 原始数据

创建销售工资表

hive> create table  if not exists sale12(id bigint, sal float, sale float ,mdate string)

     row format delimited fields terminated by ',';

本地数据

[hadoop@h201 hhh]$ cat sale12.txt

10001,5000,100500,20200202

10002,6000,600000,20200202

10003,5000,50000,20200202

10004,5000,280000,20200202

10005,8000,980000,20200202

10001,5000,205800,20200102

10002,6000,250000,20200102

导入数据

hive> load data local inpath '/home/hadoop/hhh/sale12.txt' into table sale12;

2.3 开发

开发流程

1. 集成UDF类
2. 重写evaluate方法

package com.hzr;
import org.apache.hadoop.hive.ql.exec.UDF;
public class myudf extends UDF {
    public float evaluate(float a, float b) {
        if (b>100000 && b<=200000) {
            return (float) (a+b*0.05);
        } else if(b>200000 && b<=500000){
            return (float)(a+b*0.08);
        } else if(b>500000){
            return (float)(a+b*0.15);
        }else {
               return a;
            }
    }
}

2.4 生成jar包并执行

 

 

2.4.2拷贝到Linux上

[hadoop@h201 ~]$ cp /ff/hivejar/udf1.jar /home/hadoop/qq/

2.4.3 执行udf

添加jar包到hive中

hive> add jar /home/hadoop/qq/udf1.jar;

创建UDF函数 udf123

hive> create temporary function udf123 as 'com.hzr.myudf';

显示字段名

hive> set hive.cli.print.header=true;

应用UDF函数

hive> select udf123(sal,sale) as gongzi from sale12;

结果：

gongzi

10025.0

96000.0

5000.0

27400.0

155000.0

21464.0

26000.0