背景
Hive中有非常多的内置函数,但是根据我们线上业务的不同,单单使用Hive内置函数已经满足不了我们的需求了。这时候就需要我们去开发自定义的函数(如:自定义UDF函数),为了像使用Hive内置函数一样方便,我们需要将自定义好的UDF函数编译到Hive源码中。这篇文章详细记录了编译Hive源码来支持自定义的UDF函数。
环境
-
开发工具:ideaIU-2019.3.1.exe
-
本地环境:Windows 10
-
编译环境:CentOS Linux release 7.2.1511 (Core)
-
Maven版本:apache-maven-3.6.3-bin.tar.gz
-
Hive版本:hive-1.1.0-cdh5.16.2.tar.gz
-
Hive源码:hive-1.1.0-cdh5.16.2-src.tar.gz
-
JDK版本:java version "1.8.0_192"
自定义UDF函数
pom文件pom.xml
<?xml version="1.0" encoding="UTF-8"?>
<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
<modelVersion>4.0.0</modelVersion>
<groupId>com.haohaodata.bigdata</groupId>
<artifactId>hadoop0001</artifactId>
<version>1.0-SNAPSHOT</version>
<name>hadoop0001</name>
<url>http://www.example.com</url>
<repositories>
<repository>
<id>cloudera</id>
<url>https://repository.cloudera.com/artifactory/cloudera-repos/</url>
</repository>
</repositories>
<properties>
<project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
<maven.compiler.source>1.8</maven.compiler.source>
<maven.compiler.target>1.8</maven.compiler.target>
<hadoop.version>2.6.0-cdh5.16.2</hadoop.version>
<hive.version>1.1.0-cdh5.16.2</hive.version>
</properties>
<dependencies>
<dependency>
<groupId>junit</groupId>
<artifactId>junit</artifactId>
<version>4.11</version>
<scope>test</scope>
</dependency>
<!-- 添加hive的依赖 -->
<dependency>
<groupId>org.apache.hive</groupId>
<artifactId>hive-exec</artifactId>
<version>${hive.version}</version>
</dependency>
</dependencies>
<build>
<pluginManagement>
<plugins>
<plugin>
<artifactId>maven-clean-plugin</artifactId>
<version>3.1.0</version>
</plugin>
<plugin>
<artifactId>maven-resources-plugin</artifactId>
<version>3.0.2</version>
</plugin>
<plugin>
<artifactId>maven-compiler-plugin</artifactId>
<version>3.8.0</version>
</plugin>
<plugin>
<artifactId>maven-surefire-plugin</artifactId>
<version>2.22.1</version>
</plugin>
<plugin>
<artifactId>maven-jar-plugin</artifactId>
<version>3.0.2</version>
</plugin>
<plugin>
<artifactId>maven-install-plugin</artifactId>
<version>2.5.2</version>
</plugin>
<plugin>
<artifactId>maven-deploy-plugin</artifactId>
<version>2.8.2</version>
</plugin>
<plugin>
<artifactId>maven-site-plugin</artifactId>
<version>3.7.1</version>
</plugin>
<plugin>
<artifactId>maven-project-info-reports-plugin</artifactId>
<version>3.0.0</version>
</plugin>
</plugins>
</pluginManagement>
</build>
</project>
UDF函数UDFhaohao.java
package com.haohaodata.bigdata.hadoop.hive;
import org.apache.hadoop.hive.ql.exec.UDF;
/**
* 自定义UDF函数
* 需求:你要老婆不
*/
public class UDFhaohao extends UDF {
public String evaluate() {
String send_wife = new String("KO,你要老婆不要? 只要你开金口,我马上给你送来……");
return send_wife;
}
public String evaluate(String name) {
String send_wife = new String(name + ",你要老婆不要? 只要你开金口,我马上给你送来……");
return send_wife;
}
// 本地测试
public static void main(String[] args) {
UDFhaohao udFhaohao = new UDFhaohao();
String r1 = udFhaohao.evaluate();
String r2 = udFhaohao.evaluate("hager");
System.out.println(r1);
System.out.println(r2);
}
}
源码修改
将下载好的源码包上传到服务器然后解压
tar -zxvf hive-1.1.0-cdh5.16.2-src.tar.gz -C ~/sourcecode/
修改UDFhaohao.java文件中的package
package com.haohaodata.bigdata.hadoop.hive;
// 改为:
package org.apache.hadoop.hive.ql.udf;
将自定义的UDFhaohao.java放到源码中
~/sourcecode/hive-1.1.0-cdh5.16.2/ql/src/java/org/apache/hadoop/hive/ql/udf/UDFhaohao.java
修改源码文件FunctionRegistry.java
vi ~/sourcecode/hive-1.1.0-cdh5.16.2/ql/src/java/org/apache/hadoop/hive/ql/exec/FunctionRegistry.java
添加:
import org.apache.hadoop.hive.ql.udf.UDFhaohao;
static {
system.registerUDF("UDFhaohao", UDFhaohao.class, false);
……
}
编译源码
开始编译:
cd ~/sourcecode/hive-1.1.0-cdh5.16.2/
mvn clean package -DskipTests -Phadoop-2 -Pdist
编译完成(第一次编译+网速慢 所以编译时间有点长):
[INFO] --- build-helper-maven-plugin:1.8:attach-artifact (attach-jdbc-driver) @ hive-packaging ---
[INFO] ------------------------------------------------------------------------
[INFO] Reactor Summary for Hive 1.1.0-cdh5.16.2:
[INFO]
[INFO] Hive ............................................... SUCCESS [22:21 min]
[INFO] Hive Classifications ............................... SUCCESS [06:41 min]
[INFO] Hive Shims Common .................................. SUCCESS [10:35 min]
[INFO] Hive Shims 0.23 .................................... SUCCESS [05:04 min]
[INFO] Hive Shims Scheduler ............................... SUCCESS [ 0.457 s]
[INFO] Hive Shims ......................................... SUCCESS [ 0.213 s]
[INFO] Hive Common ........................................ SUCCESS [08:48 min]
[INFO] Hive Serde ......................................... SUCCESS [02:12 min]
[INFO] Hive Metastore ..................................... SUCCESS [08:44 min]
[INFO] Hive Ant Utilities ................................. SUCCESS [ 44.108 s]
[INFO] Spark Remote Client ................................ SUCCESS [ 01:03 h]
[INFO] Hive Query Language ................................ SUCCESS [13:26 min]
[INFO] Hive Service ....................................... SUCCESS [15:31 min]
[INFO] Hive Accumulo Handler .............................. SUCCESS [03:36 min]
[INFO] Hive JDBC .......................................... SUCCESS [ 38.016 s]
[INFO] Hive Beeline ....................................... SUCCESS [07:42 min]
[INFO] Hive CLI ........................................... SUCCESS [ 0.887 s]
[INFO] Hive Contrib ....................................... SUCCESS [ 0.796 s]
[INFO] Hive HBase Handler ................................. SUCCESS [06:00 min]
[INFO] Hive HCatalog ...................................... SUCCESS [03:16 min]
[INFO] Hive HCatalog Core ................................. SUCCESS [ 3.891 s]
[INFO] Hive HCatalog Pig Adapter .......................... SUCCESS [ 30.240 s]
[INFO] Hive HCatalog Server Extensions .................... SUCCESS [02:04 min]
[INFO] Hive HCatalog Webhcat Java Client .................. SUCCESS [ 0.799 s]
[INFO] Hive HCatalog Webhcat .............................. SUCCESS [04:54 min]
[INFO] Hive HCatalog Streaming ............................ SUCCESS [ 0.752 s]
[INFO] Hive HWI ........................................... SUCCESS [ 0.628 s]
[INFO] Hive ODBC .......................................... SUCCESS [ 0.480 s]
[INFO] Hive Shims Aggregator .............................. SUCCESS [ 0.119 s]
[INFO] Hive TestUtils ..................................... SUCCESS [ 0.150 s]
[INFO] Hive Packaging ..................................... SUCCESS [02:49 min]
[INFO] ------------------------------------------------------------------------
[INFO] BUILD SUCCESS
[INFO] ------------------------------------------------------------------------
[INFO] Total time: 03:09 h
[INFO] Finished at: 2020-01-24T01:11:56+08:00
[INFO] ------------------------------------------------------------------------
部署Hive
部署方式
- 使用编译好的jar包重新部署Hive
# 编译好的jar包路径 ll ~/sourcecode/hive-1.1.0-cdh5.16.2/packaging/target/apache-hive-1.1.0-cdh5.16.2-bin.tar.gz
- 将编译好的hive-exec-1.1.0-cdh5.16.2.jar替换掉Hive集群中的
# hive-exec-1.1.0-cdh5.16.2.jar所在路径 ll ~/sourcecode/hive-1.1.0-cdh5.16.2/packaging/target/apache-hive-1.1.0-cdh5.16.2-bin/apache-hive-1.1.0-cdh5.16.2-bin/lib/hive-exec-1.1.0-cdh5.16.2.jar
这里使用第二中部署方式
# 备份正在使用的hive-exec-1.1.0-cdh5.16.2.jar
cd ~/app/hive/lib/
mv hive-exec-1.1.0-cdh5.16.2.jar hive-exec-1.1.0-cdh5.16.2.jar.bak
# 将我们自己打的jar包放到这里
cp ~/sourcecode/hive-1.1.0-cdh5.16.2/packaging/target/apache-hive-1.1.0-cdh5.16.2-bin/apache-hive-1.1.0-cdh5.16.2-bin/lib/hive-exec-1.1.0-cdh5.16.2.jar ./
使用自定义UDF
# 查看我们自定义的udf
hive (default)> show functions;
# 测试
hive (default)> select udfhaohao();
OK
_c0
KO,你要老婆不要? 只要你开金口,我马上给你送来……
Time taken: 0.45 seconds, Fetched: 1 row(s)
hive (default)> select udfhaohao("KO");
OK
_c0
KO,你要老婆不要? 只要你开金口,我马上给你送来……
Time taken: 0.042 seconds, Fetched: 1 row(s)
hive (default)>