hive创建udf函数流程

article/2025/9/27 0:05:08

1.编写udf函数

引入pom文件

<dependencies>
<dependency>


<groupId>org.apache.commons</groupId>
<artifactId>commons-jexl3</artifactId>
<version>3.1</version>
</dependency>
<dependency> 
<groupId>org.apache.hive</groupId>
<artifactId>hive-exec</artifactId>
<version>3.1.2</version>
<scope>provided</scope>
</dependency>
</dependencies>

先确定好你要写什么函数比如我要写一个计算字符串表达式。

开始继承hive的udf接口，有很多小伙伴这个时候就喜欢看别人是怎么写的，这个时候就是体现个人差距的时候了，如何不看别人文档自己写呢？比如没网的条件下？

抄别人的为啥不直接抄hive的呢？想想hive什么udf函数最简单，lower/upper。照着抄就行。

public class StringCal extends GenericUDF 实现三个方法

initialize 初始化校验参数的

evaluate 真正执行的方法

getDisplayString: desc function时打印的话


import org.apache.commons.jexl3.JexlBuilder;
import org.apache.commons.jexl3.JexlEngine;
import org.apache.commons.jexl3.JexlExpression;
import org.apache.hadoop.hive.ql.exec.Description;
import org.apache.hadoop.hive.ql.exec.UDFArgumentException;
import org.apache.hadoop.hive.ql.exec.UDFArgumentLengthException;
import org.apache.hadoop.hive.ql.metadata.HiveException;
import org.apache.hadoop.hive.ql.udf.generic.GenericUDF;
import org.apache.hadoop.hive.ql.udf.generic.GenericUDFUtils;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorConverter;
import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory;
import org.apache.hadoop.hive.serde2.typeinfo.BaseCharTypeInfo;
import org.apache.hadoop.hive.serde2.typeinfo.TypeInfoFactory;
//有时间的这里写下，免得后面自己都不知道是干嘛的了。
@Description(name = "StringCal",value = "_FUNC_(str) - Returns str with calculate result",extended = "Example:\n"+ "  > SELECT _FUNC_('1+(-1+2.0-3.0+(4.0-5.0))+3.1-4.1+2*3+1.1*4') FROM src LIMIT 1;\n" + "  '-7.2'")
public class StringCal extends GenericUDF {private transient PrimitiveObjectInspector argumentOI;private transient PrimitiveObjectInspectorConverter.StringConverter stringConverter;private transient PrimitiveObjectInspector.PrimitiveCategory returnType = PrimitiveObjectInspector.PrimitiveCategory.STRING;private transient GenericUDFUtils.StringHelper returnHelper;//这里一大串校验，校验是不是普通类型啥的，校验是字符串还是啥，哪那么多事，反正照着抄就行，不写也没啥你自己定义的函数，别人也不会用。@Overridepublic ObjectInspector initialize(ObjectInspector[] arguments) throws UDFArgumentException {if (arguments.length != 1) {throw new UDFArgumentLengthException("StringCal requires 1 argument, got " + arguments.length);}if (arguments[0].getCategory() != ObjectInspector.Category.PRIMITIVE) {throw new UDFArgumentException("StringCal only takes primitive types, got " + argumentOI.getTypeName());}argumentOI = (PrimitiveObjectInspector) arguments[0];stringConverter = new PrimitiveObjectInspectorConverter.StringConverter(argumentOI);PrimitiveObjectInspector.PrimitiveCategory inputType = argumentOI.getPrimitiveCategory();ObjectInspector outputOI = null;BaseCharTypeInfo typeInfo;switch (inputType) {case CHAR:// return type should have same length as the input.returnType = inputType;typeInfo = TypeInfoFactory.getCharTypeInfo(GenericUDFUtils.StringHelper.getFixedStringSizeForType(argumentOI));outputOI = PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(typeInfo);break;case VARCHAR:// return type should have same length as the input.returnType = inputType;typeInfo = TypeInfoFactory.getVarcharTypeInfo(GenericUDFUtils.StringHelper.getFixedStringSizeForType(argumentOI));outputOI = PrimitiveObjectInspectorFactory.getPrimitiveWritableObjectInspector(typeInfo);break;default:returnType = PrimitiveObjectInspector.PrimitiveCategory.STRING;outputOI = PrimitiveObjectInspectorFactory.writableStringObjectInspector;break;}returnHelper = new GenericUDFUtils.StringHelper(returnType);return outputOI;}@Overridepublic Object evaluate(DeferredObject[] arguments) throws HiveException {String val = null;if (arguments[0] != null) {val = (String) stringConverter.convert(arguments[0].get());}if (val == null) {return null;}
//就这里是我自己写的 其他的都是抄的lowerUdf的。
//        String expressionString = "1+(-1+2.0-3.0+(4.0-5.0))+3.1-4.1+2*3+1.1*4";JexlEngine jexlEngine = new JexlBuilder().create();JexlExpression jexlExpression = jexlEngine.createExpression(val);Object evaluate = jexlExpression.evaluate(null);return returnHelper.setReturnValue(evaluate.toString());}@Overridepublic String getDisplayString(String[] children) {return null;}
}

然后打包成一个jar，上传的hdfs

我嫌弃打的包不好听就直接改了个名字。

然后创建函数

create function default.stringCal as 'com.chenchi.s2.function.StringCal' using jar 'hdfs:///user/hive/function/stringcalculate.jar';

建议大家用这种方式，这种方式你可以随时替换jar，使得函数都是最新的。

这个是创建函数指定用哪个jar

或者

add jar hdfs:///user/hive/function/dw_ce_analysis.jar;

create function defalut.stringcal as 'com.chenchi.s2.function.StringCal';

这个是先加载jar然后根据jar创建函数。 add jar 后list jar可以看到是在/tmp目录

注意啊这里加上数据库的名字，否则退出会话就没了。

这里我创建了两次。就是因为没加数据库的名字，反正建议各位加下。