Hive如何自定义函数实现将自定义函数注册到hive源码中并重新编译hive

这篇文章主要介绍Hive如何自定义函数实现将自定义函数注册到hive源码中并重新编译hive，文中介绍的非常详细，具有一定的参考价值，感兴趣的小伙伴们一定要看完！

hive版本：
    hive-1.1.0-cdh6.7.0
1 编写UDF函数
    1.1 用户idea+Maven创建项目，pom.xml文件的配置如下：

<?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.ruozedata.bigdata</groupId>
 <artifactId>hive</artifactId>
 <version>1.0-SNAPSHOT</version>

 <name>hive</name>
 <!-- FIXME change it to the project's WEBsite -->
 <url>http://www.example.com</url>

 <properties>
   <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
   <maven.compiler.source>1.7</maven.compiler.source>
   <maven.compiler.target>1.7</maven.compiler.target>
   <hive.version>1.1.0-cdh6.7.0</hive.version>
   <hadoop.version>2.6.0-cdh6.7.0</hadoop.version>
 </properties>

 <!--Note: CDH版本一定需要添加如下依赖-->
 <repositories>
   <repository>
     <id>cloudera</id>
     <url>https://repository.cloudera.com/artifactory/cloudera-repos</url>
   </repository>
 </repositories>

 <dependencies>
   <!-- hdfs -->
   <dependency>
     <groupId>org.apache.hadoop</groupId>
     <artifactId>hadoop-common</artifactId>
     <version>${hadoop.version}</version>
   </dependency>

   <dependency>
     <!-- Hive -->
     <groupId>org.apache.hive</groupId>
     <artifactId>hive-exec</artifactId>
     <version>${hive.version}</version>
   </dependency>

   <dependency>
     <groupId>junit</groupId>
     <artifactId>junit</artifactId>
     <version>4.11</version>
     <scope>test</scope>
   </dependency>
 </dependencies>

 <build>
   <pluginManagement><!-- lock down plugins versions to avoid using Maven defaults (may be moved to parent pom) -->
     <plugins>
<plugin>
 <artifactId>maven-clean-plugin</artifactId>
 <version>3.0.0</version>
</plugin>
<!-- see http://maven.apache.org/ref/current/maven-core/default-bindings.html#Plugin_bindings_for_jar_packaging -->
<plugin>
 <artifactId>maven-resources-plugin</artifactId>
 <version>3.0.2</version>
</plugin>
<plugin>
 <artifactId>maven-compiler-plugin</artifactId>
 <version>3.7.0</version>
</plugin>
<plugin>
 <artifactId>maven-surefire-plugin</artifactId>
 <version>2.20.1</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>
     </plugins>
   </pluginManagement>
 </build>
</project>


    1.2 下载相关包：右键maven项目，maven->Reimport，就可自动下载jar包了，网络不好的话有点慢，

    1.3 创建类，右键包名：New->java Class 新建HelloUDF.java类
    package com.ruozedata.bigdata;
    import org.apache.hadoop.hive.ql.exec.Description;
    import org.apache.hadoop.hive.ql.exec.UDF;

    @Description(
    name="sayhelloudf",
    value="_FUNC_(input_str) - returns Hello : input_str  or _FUNC_(input_str,input_str2) - returns Hello : input_str:input_str2 ",
    extended = "Example:\n"+
                    " > SELECT _FUNC_('xx') FROM dual LIMIT 1;\n" +
                    "   Hello : xx "
                    )
    public class HelloUDF extends UDF {

        public String evaluate(String input){
                return "Hello:" + input;
       }

       public String evaluate(String input, String input2){
                return "Hello:" + input + " : " + input2;
       }
    }

2 下载hive-1.1.0-cdh6.7.0-src.tar.gz源码文件
     http://archive.cloudera.com/cdh6/cdh/5/hive-1.1.0-cdh6.7.0-src.tar.gz
    [hadoop@hadoop002 software]$ pwd
    /home/hadoop/software
    [hadoop@hadoop002 software]$ ll
    total 499272
    drwxrwxr-x. 17 hadoop hadoop      4096 Mar 24  2016 hadoop-2.6.0-cdh6.7.0
    -rw-rw-r--.  1 hadoop hadoop  42610549 Jul  5 15:05 hadoop-2.6.0-cdh6.7.0-src.tar.gz
    -rw-rw-r--.  1 hadoop hadoop 311585484 Feb 20 07:16 hadoop-2.6.0-cdh6.7.0.tar.gz
    -rw-rw-r--.  1 hadoop hadoop  14652104 Feb 21 05:28 hive-1.1.0-cdh6.7.0-src.tar.gz
    -rw-rw-r--.  1 hadoop hadoop 116082695 Feb 21 05:28 hive-1.1.0-cdh6.7.0.tar.gz
    -rw-rw-r--.  1 hadoop hadoop  29966286 Feb 22 02:28 sqoop-1.4.6-cdh6.7.0.tar.gz

    解压到当前目录：
    [hadoop@hadoop002 software]$ tar -zxvf hive-1.1.0-cdh6.7.0-src.tar.gz
    [hadoop@hadoop002 software]$ ll
    total 503368
    drwxrwxr-x. 17 hadoop hadoop      4096 Mar 24  2016 hadoop-2.6.0-cdh6.7.0
    -rw-rw-r--.  1 hadoop hadoop  42610549 Jul  5 15:05 hadoop-2.6.0-cdh6.7.0-src.tar.gz
    -rw-rw-r--.  1 hadoop hadoop 311585484 Feb 20 07:16 hadoop-2.6.0-cdh6.7.0.tar.gz
    drwxrwxr-x. 32 hadoop hadoop      4096 Jul  5 15:15 hive-1.1.0-cdh6.7.0
    -rw-rw-r--.  1 hadoop hadoop  14652104 Feb 21 05:28 hive-1.1.0-cdh6.7.0-src.tar.gz
    -rw-rw-r--.  1 hadoop hadoop 116082695 Feb 21 05:28 hive-1.1.0-cdh6.7.0.tar.gz
    -rw-rw-r--.  1 hadoop hadoop  29966286 Feb 22 02:28 sqoop-1.4.6-cdh6.7.0.tar.gz


3 将HelloUDF.java类添加到源码
    3.1 将HelloUDF.java类上传到/home/hadoop/software/hive-1.1.0-cdh6.7.0/ql/src/java/org/apache/hadoop/hive/ql/udf 目录。
    3.2 修改HelloUDF.java类的包：将 package com.ruozedata.bigdata; 修改为package org.apache.hadoop.hive.ql.udf;
    3.3 修改FunctionReGIStry.java类文件，对HelloUDF.java类进行注册：
        3.3.1 引入HelloUDF.java类文件的路径：
                    import org.apache.hadoop.hive.ql.udf.HelloUDF;
 
        3.3.2 在 static 块中添加:system.registerUDF("helloudf", HelloUDF.class, false);
                如下：
                        static {
                                   system.registerGenericUDF("concat", GenericUDFConcat.class);
                                   system.registerUDF("helloudf", HelloUDF.class, false);
                                   .
                                   .
                                   .
                               }


4 编译 hive-1.1.0-cdh6.7.0
    4.1 进入hive-1.1.0-cdh6.7.0目录进行源码编译：
    [hadoop@hadoop002 software]$ cd hive-1.1.0-cdh6.7.0
    [hadoop@hadoop002 hive-1.1.0-cdh6.7.0]$ mvn clean package -DskipTests -Phadoop-2 -Pdist

    出现如下日志，表示编译成功
    [INFO] --- build-helper-maven-plugin:1.8:attach-artifact (attach-jdbc-driver) @ hive-packaging ---
    [INFO] ------------------------------------------------------------------------
    [INFO] Reactor Summary:
    [INFO] 
    [INFO] Hive ............................................... SUCCESS [  9.558 s]
    [INFO] Hive Shims Common .................................. SUCCESS [ 14.763 s]
    [INFO] Hive Shims 0.23 .................................... SUCCESS [  7.488 s]
    [INFO] Hive Shims Scheduler ............................... SUCCESS [  4.449 s]
    [INFO] Hive Shims ......................................... SUCCESS [  2.784 s]
    [INFO] Hive Common ........................................ SUCCESS [ 14.109 s]
    [INFO] Hive Serde ......................................... SUCCESS [ 12.877 s]
    [INFO] Hive Metastore ..................................... SUCCESS [ 39.906 s]
    [INFO] Hive Ant Utilities ................................. SUCCESS [  3.377 s]
    [INFO] spark Remote Client ................................ SUCCESS [ 13.578 s]
    [INFO] Hive Query Language ................................ SUCCESS [02:22 min]
    [INFO] Hive Service ....................................... SUCCESS [01:14 min]
    [INFO] Hive Accumulo Handler .............................. SUCCESS [ 36.593 s]
    [INFO] Hive JDBC .......................................... SUCCESS [01:31 min]
    [INFO] Hive Beeline ....................................... SUCCESS [  8.103 s]
    [INFO] Hive CLI ........................................... SUCCESS [  8.419 s]
    [INFO] Hive Contrib ....................................... SUCCESS [  7.041 s]
    [INFO] Hive HBase Handler ................................. SUCCESS [01:29 min]
    [INFO] Hive HCatalog ...................................... SUCCESS [ 31.376 s]
    [INFO] Hive HCatalog Core ................................. SUCCESS [ 13.366 s]
    [INFO] Hive HCatalog Pig Adapter .......................... SUCCESS [ 10.937 s]
    [INFO] Hive HCatalog Server Extensions .................... SUCCESS [ 32.887 s]
    [INFO] Hive HCatalog Webhcat Java Client .................. SUCCESS [  6.679 s]
    [INFO] Hive HCatalog Webhcat .............................. SUCCESS [ 41.266 s]
    [INFO] Hive HCatalog Streaming ............................ SUCCESS [  6.445 s]
    [INFO] Hive HWI ........................................... SUCCESS [  5.169 s]
    [INFO] Hive ODBC .......................................... SUCCESS [  3.305 s]
    [INFO] Hive Shims Aggregator .............................. SUCCESS [  1.139 s]
    [INFO] Hive TestUtils ..................................... SUCCESS [  1.869 s]
    [INFO] Hive Packaging ..................................... SUCCESS [01:19 min]
    [INFO] ------------------------------------------------------------------------
    [INFO] BUILD SUCCESS
    [INFO] ------------------------------------------------------------------------
    [INFO] Total time: 13:37 min
    [INFO] Finished at: 2018-07-05T16:02:01+08:00
    [INFO] Final Memory: 158M/494M
    [INFO] ------------------------------------------------------------------------

    4.2  编译成功后会生成一个tar包：/home/hadoop/software/hive-1.1.0-cdh6.7.0/packaging/target/apache-hive-1.1.0-cdh6.7.0-bin.tar.gz  
       
5  部署hive-1.1.0-cdh6.7.0
    5.1 直接将编译好的apache-hive-1.1.0-cdh6.7.0-bin.tar.gz进行安装，见我前面写的一篇博客：hive-1.1.0-cdh6.7.0 安装 http://blog.itpub.net/29609890/viewspace-2155488/
    5.2 替换hive-exec-1.1.0-cdh6.7.0.jar包。 在编译好的文件中找到hive-exec-1.1.0-cdh6.7.0.jar包，去替换现在使用的hadoop-2.6.0-cdh6.7.0的hive-exec-1.1.0-cdh6.7.0.jar包
       编译好的hive-exec-1.1.0-cdh6.7.0.jar包路径：/home/hadoop/software/hive-1.1.0-cdh6.7.0/packaging/target/apache-hive-1.1.0-cdh6.7.0-bin/apache-hive-1.1.0-cdh6.7.0-bin/lib/
       现在使用的hadoop-2.6.0-cdh6.7.0的hive-exec-1.1.0-cdh6.7.0.jar包路径：/home/hadoop/app/hive-1.1.0-cdh6.7.0/lib/hive-exec-1.1.0-cdh6.7.0.jar
        切换到现在使用的hive目录：
        [hadoop@hadoop002 ~]$ cd ~/app/hive-1.1.0-cdh6.7.0/
        [hadoop@hadoop002 hive-1.1.0-cdh6.7.0]$ pwd
        /home/hadoop/app/hive-1.1.0-cdh6.7.0
        [hadoop@hadoop002 hive-1.1.0-cdh6.7.0]$ 
        备份现在使用的jar包：
        /home/hadoop/app/hive-1.1.0-cdh6.7.0/lib/
         [hadoop@hadoop002 hive-1.1.0-cdh6.7.0]# mv hive-exec-1.1.0-cdh6.7.0.jar hive-exec-1.1.0-cdh6.7.0.jar.bak
        拷贝到原hive 部署位置：
        [hadoop@hadoop002 hive-1.1.0-cdh6.7.0]# cp /home/hadoop/software/hive-1.1.0-cdh6.7.0/packaging/target/apache-hive-1.1.0-cdh6.7.0-bin/apache-hive-1.1.0-cdh6.7.0-bin/lib/hive-exec-1.1.0-cdh6.7.0.jar  ./
        查看hive-exec-1.1.0-cdh6.7.0.jar包
        [hadoop@hadoop002 lib]$ ll hive-exec-1.1.0-cdh6.7.0.*
        -rw-rw-r--. 1 hadoop hadoop 19274963 Jul  5 17:44 hive-exec-1.1.0-cdh6.7.0.jar
        -rw-r--r--. 1 hadoop hadoop 19272159 Mar 24  2016 hive-exec-1.1.0-cdh6.7.0.jar.bak


6 重启hive进行验证：
     hive> desc function sayhelloudf;
    OK
    sayhelloudf(input_str) - returns Hello : input_str  or sayhelloudf(input_str,input_str2) - returns Hello : input_str:input_str2 
    Time taken: 0.031 seconds, Fetched: 1 row(s)
    hive> select sayhelloudf('zhangsan') from ruozedata.dual;
    OK
    Hello:zhangsan
    Time taken: 2.531 seconds, Fetched: 1 row(s)
    hive> select sayhelloudf('zhangsan','lisi') from ruozedata.dual;
    OK
    Hello:zhangsan : lisi
    Time taken: 0.397 seconds, Fetched: 1 row(s)
    hive> 

以上是“Hive如何自定义函数实现将自定义函数注册到hive源码中并重新编译hive”这篇文章的所有内容，感谢各位的阅读！希望分享的内容对大家有帮助，更多相关知识，欢迎关注编程网精选频道！