hive中生成从1到n的连续数字的UDTF方法

之前在博客中分享了个生成从1到n的连续数字的transform或map方法，最近研究了一下UDTF，发现用UDTF写出来的函数用起来更方便，语法更接近于普通的函数。而且在JAVA中对参数的个数、类型等进行校验也更容易。

 

ORACLE中生成从1到n的连续数字的方法很多，最简单的一种是：

select level from dual connect by level<=5;

     LEVEL

----------

         1

         2

         3

         4

         5

 

使用UDTF写出来的函数进行查询：

hive> select serial(5) as col1 from dual;

OK
1
2
3
4
5

 

或者使用lateral view进行查询：

hive> select T.col1 from dual lateral view serial(5) T as col1;

OK
1
2
3
4
5

 

提供一下java代码，仅供参考：

 

package com.hadoopbook.hive;

import java.util.ArrayList;
import org.apache.hadoop.hive.ql.exec.UDFArgumentException;
import org.apache.hadoop.hive.ql.metadata.HiveException;
import org.apache.hadoop.hive.ql.exec.UDFArgumentLengthException;
import org.apache.hadoop.hive.ql.udf.generic.GenericUDTF;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory;
import org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector;
import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory;

public class UDTFSerial extends GenericUDTF {
 
  Object[] result = new Object[1];
 
  @Override
  public void close() throws HiveException {
  }
 
  @Override
  public StructObjectInspector initialize(ObjectInspector[] args)
          throws UDFArgumentException {
      if (args.length != 1) {
          throw new UDFArgumentLengthException("UDTFSerial takes only one argument");
      }
      if (!args[0].getTypeName().equals("int")) {
       throw new UDFArgumentException("UDTFSerial only takes an integer as a parameter");
        }
      ArrayList fieldNames = new ArrayList();
      ArrayList fieldOIs = new ArrayList();
      fieldNames.add("col1");
      fieldOIs.add(PrimitiveObjectInspectorFactory.javaIntObjectInspector);
     
      return ObjectInspectorFactory.getStandardStructObjectInspector(fieldNames,fieldOIs);
  }
 
  @Override
  public void process(Object[] args) throws HiveException {
   try
   {
    int n = Integer.parseInt(args[0].toString());
    for (int i=0;i     {
     result[0] = i+1;
     forward(result);
    }
   }
   catch (Exception e) { 
    throw new HiveException("UDTFSerial has an exception");
   }
  }
}