hive中使用自定义函数(UDF)实现分析函数row_number的功能

之前部门实现row_number是使用的transform，我觉得用UDF实现后，平时的使用会更方便，免去了transform相对繁琐的语法。

   用到的测试表为：

hive> desc row_number_test;
OK
id1    int
id2    string
age    int
score   double
name   string

 

hive> select * from row_number_test;
OK
2      t04    25     60.0   youlia
1      t01    20     85.0   liujiannan
1      t02    24     70.0   zengqiu
2      t03    30     88.0   hongqu
2      t03    27     70.0   yongqi
1      t02    19     75.0   wangdong
1      t02    24     70.0   zengqiu

 

使用时要先在子查询中进行分区与排序，比如oracle中这样一句SQL：

select row_number() over (partition by id1 order by age desc)from row_number_test;

转换为hive语句应该是：

select row_number(id1) from  --partition by的字段传到row_number函数中去

    (select *from row_number_test distribute by id1 sort by id1,age desc) a;

 

如果partition by 两个字段：

select row_number() over (partition by id1,id2 orderby score) from row_number_test;

转换为hive语句应该是：

select row_number(id1,id2)  --partition by的字段传到row_number函数中去

    from(select * from row_number_test distribute by id1,id2 sort byid1,id2,score) a;

 

展示一下查询结果：

1.

select id1,id2,age,score,name,row_number(id1) rn from (select *from row_number_test distribute by id1 sort by id1,age desc) a;

 

OK
2      t03    30     88.0   hongqu         1
2      t03    27     70.0   yongqi         2
2      t04    25     60.0   youlia         3
1      t02    24     70.0   zengqiu        1
1      t02    24     70.0   zengqiu        2
1      t01    20     85.0   liujiannan     3
1      t02    19     75.0   wangdong       4

 

2.

select id1,id2,age,score,name,row_number(id1,id2) rn from(select * from row_number_test distribute by id1,id2 sortby id1,id2,score) a;

 

OK
2      t04    25     60.0   youlia         1
1      t02    24     70.0   zengqiu        1
2      t03    27     70.0   yongqi         1
1      t02    24     70.0    zengqiu       2
1      t02    19     75.0    wangdong      3
1      t01    20     85.0   liujiannan     1
2      t03    30     88.0    hongqu        2

 

下面是代码，只实现了接收1个参数和2个参数的evaluator方法，参数再多的照搬代码就可以了，代码仅供参考：

package org.rowincrement;

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 RowIncrement 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("RowIncrement takes only one argument");
      }
      if (!args[0].getTypeName().equals("int")) {
       throw new UDFArgumentException("RowIncrement 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