$王大少

导航

hive的自定义函数

1)Hive 自带一些函数,比如:max/min等,但是数量有限,自己可以通过自定义UDF来方便的扩展。

2)Hive提供的内置函数无法满足你的业务处理需要时,此时就可以考虑使用用户自定义函数(UDFuser-defined function)。

3)根据用户自定义函数类别分为以下三种:

1UDFUser-Defined-Function

一进一出

2UDAFUser-Defined Aggregation Function

聚集函数,多进一出

类似于:count/max/min

3UDTFUser-Defined Table-Generating Functions

一进多出

如lateral view explore()

4)官方文档地址

https://cwiki.apache.org/confluence/display/Hive/HivePlugins

5)编程步骤:

1继承org.apache.hadoop.hive.ql.UDF

2需要实现evaluate函数;evaluate函数支持重载;

6)注意事项

1UDF必须要有返回类型,可以返回null,但是返回类型不能为void

2UDF中常用Text/LongWritable等类型,不推荐使用java类型

 

 

简单UDF示例

第一步创建maven  java 工程,导入jar

<repositories>
    <repository>
        <id>cloudera</id>
 <url>https://repository.cloudera.com/artifactory/cloudera-repos/</url>
    </repository>
</repositories>
<dependencies>
    <dependency>
        <groupId>org.apache.hadoop</groupId>
        <artifactId>hadoop-common</artifactId>
        <version>2.6.0-cdh5.14.0</version>
    </dependency>
    <dependency>
        <groupId>org.apache.hive</groupId>
        <artifactId>hive-exec</artifactId>
        <version>1.1.0-cdh5.14.0</version>
    </dependency>
</dependencies>
<build>
<plugins>
    <plugin>
        <groupId>org.apache.maven.plugins</groupId>
        <artifactId>maven-compiler-plugin</artifactId>
        <version>3.0</version>
        <configuration>
            <source>1.8</source>
            <target>1.8</target>
            <encoding>UTF-8</encoding>
        </configuration>
    </plugin>
     <plugin>
         <groupId>org.apache.maven.plugins</groupId>
         <artifactId>maven-shade-plugin</artifactId>
         <version>2.2</version>
         <executions>
             <execution>
                 <phase>package</phase>
                 <goals>
                     <goal>shade</goal>
                 </goals>
                 <configuration>
                     <filters>
                         <filter>
                             <artifact>*:*</artifact>
                             <excludes>
                                 <exclude>META-INF/*.SF</exclude>
                                 <exclude>META-INF/*.DSA</exclude>
                                 <exclude>META-INF/*/RSA</exclude>
                             </excludes>
                         </filter>
                     </filters>
                 </configuration>
             </execution>
         </executions>
     </plugin>
</plugins>
</build>

 

第二步:开发java继承UDF,写evaluate 方法

public class ItcastUDF extends UDF {
    public Text evaluate(final Text s) {
        if (null == s) {
            return null;
        }
        //返回大写字母
        return new Text(s.toString().toUpperCase());

    }
}

第三步:将我们的项目打包,并上传到hivelib目录下
第四步:添加我们的jar

重命名我们的jar包名称

cd /export/servers/hive-1.1.0-cdh5.14.0/lib

mv original-day_06_hive_udf-1.0-SNAPSHOT.jar udf.jar

 

hive的客户端添加我们的jar

add jar /export/servers/hive-1.1.0-cdh5.14.0/lib/udf.jar;

第五步:设置函数与我们的自定义函数关联

create temporary function tolowercase as 'cn.itcast.udf.ItcastUDF';

第六步使用自定义函数

select tolowercase('abc');

 

posted on 2020-04-01 13:35  $王大少  阅读(434)  评论(0编辑  收藏  举报