准备:先下载对应版本的源码,并源码编译一次。
1、找出对应的java文件,并对其修改
[root@hadoop001 hive-1.1.0-cdh5.7.0]# find . -name FunctionRegistry.java ./ql/src/java/org/apache/hadoop/hive/ql/exec/FunctionRegistry.java
2、修改FunctionRegistry.java文件,在文件的static块中参考其它函数的注册方法和导入自定义的java文件即可。
例如参考:Substr这个函数
import org.apache.hadoop.hive.ql.udf.UDFSubstr; import org.apache.hadoop.hive.ql.udf.HelloWord; // 添加自定义UDF的包 static { system.registerGenericUDF("concat", GenericUDFConcat.class); system.registerUDF("my_hello", HelloWord.class, false, new FunctionResource[0]); // 添加注册信息 system.registerUDF("substr", UDFSubstr.class, false); ... }
[root@hadoop001 hive-1.1.0-cdh5.7.0]# find . -name UDFSubstr.java ./ql/src/java/org/apache/hadoop/hive/ql/udf/UDFSubstr.javaHelloWord.java代码如下
package org.apache.hadoop.hive.ql.udf; import org.apache.hadoop.hive.ql.exec.Description; import org.apache.hadoop.hive.ql.exec.UDF; import org.apache.hadoop.io.Text; @Description( name = "my_hello", value = "my_hello(String)", extended = "return new Text(\"HelloWord:\" + s.toString());" ) public class HelloWord extends UDF { public Text evaluate(final Text s) { if (s == null) { return null; } return new Text("HelloWord:" + s.toString()); } public static void main(String str[]) { System.out.println("hello"); Text s = new Text("abc"); System.out.println(new HelloWord().evaluate(s).toString()); } }
4、编译 mvn clean package -Phadoop-2 -DskipTests -Pdist
5、替换lib(替换前别忘了备份、替换前别忘了备份、替换前别忘了备份),这里别忘了把原来lib中mysql驱动拷贝回来
6、进入hive检查是否存在该函数
hive> desc function my_hello; OK my_hello(String) Time taken: 0.018 seconds, Fetched: 1 row(s) OK。 hive> select my_hello(ename) from emp_dept_partition limit 3; OK HelloWord:SMITH HelloWord:JONES HelloWord:SCOTT Time taken: 0.108 seconds, Fetched: 3 row(s)【来自@若泽大数据】