hive udtf使用及explode

原文:http://blog.linezing.com/2011/03/hive%E4%B8%ADudtf%E7%BC%96%E5%86%99%E5%92%8C%E4%BD%BF%E7%94%A8

 

1. UDTF介绍

UDTF(User-Defined Table-Generating Functions)  用来解决 输入一行输出多行(On-to-many maping) 的需求。

2. 编写自己需要的UDTF

• 继承org.apache.hadoop.hive.ql.udf.generic.GenericUDTF。
• 实现initialize, process, close三个方法
• UDTF首先会调用initialize方法，此方法返回UDTF的返回行的信息（返回个数，类型）。初始化完成后，会调用process方法，对传入的参数进行处理，可以通过forword()方法把结果返回。最后close()方法调用，对需要清理的方法进行清理。

下面是我写的一个用来切分”key:value;key:value;”这种字符串，返回结果为key, value两个字段。供参考：

   1: import java.util.ArrayList;

 

   2:

 

   3: import org.apache.hadoop.hive.ql.udf.generic.GenericUDTF;

 

   4: import org.apache.hadoop.hive.ql.exec.UDFArgumentException;

 

   5: import org.apache.hadoop.hive.ql.exec.UDFArgumentLengthException;

 

   6: import org.apache.hadoop.hive.ql.metadata.HiveException;

 

   7: import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector;

 

   8: import org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory;

 

   9: import org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector;

 

  10: import org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory;

 

  11:

 

  12: public class ExplodeMap extends GenericUDTF{

 

  13:

 

  14:     @Override

 

  15:     public void close() throws HiveException {

 

  16:         // TODO Auto-generated method stub    

 

  17:     }

 

  18:

 

  19:     @Override

 

  20:     public StructObjectInspector initialize(ObjectInspector[] args)

 

  21:             throws UDFArgumentException {

 

  22:         if (args.length != 1) {

 

  23:             throw new UDFArgumentLengthException("ExplodeMap takes only one argument");

 

  24:         }

 

  25:         if (args[0].getCategory() != ObjectInspector.Category.PRIMITIVE) {

 

  26:             throw new UDFArgumentException("ExplodeMap takes string as a parameter");

 

  27:         }

 

  28:

 

  29:         ArrayList<String> fieldNames = new ArrayList<String>();

 

  30:         ArrayList<ObjectInspector> fieldOIs = new ArrayList<ObjectInspector>();

 

  31:         fieldNames.add("col1");

 

  32:         fieldOIs.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);

 

  33:         fieldNames.add("col2");

 

  34:         fieldOIs.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);

 

  35:

 

  36:         return ObjectInspectorFactory.getStandardStructObjectInspector(fieldNames,fieldOIs);

 

  37:     }

 

  38:

 

  39:     @Override

 

  40:     public void process(Object[] args) throws HiveException {

 

  41:         String input = args[0].toString();

 

  42:         String[] test = input.split(";");

 

  43:         for(int i=0; i<test.length; i++) {

 

  44:             try {

 

  45:                 String[] result = test[i].split(":");

 

  46:                 forward(result);

 

  47:             } catch (Exception e) {

 

  48:                 continue;

 

  49:             }

 

  50:         }

 

  51:     }

 

  52: }

 

3. 使用方法

UDTF有两种使用方法，一种直接放到select后面，一种和lateral view一起使用。

1：直接select中使用：select explode_map(properties) as (col1,col2) from src;

• 不可以添加其他字段使用：select a, explode_map(properties) as (col1,col2) from src
• 不可以嵌套调用：select explode_map(explode_map(properties)) from src
• 不可以和group by/cluster by/distribute by/sort by一起使用：select explode_map(properties) as (col1,col2) from src group by col1, col2

2：和lateral view一起使用：select src.id, mytable.col1, mytable.col2 from src lateral view explode_map(properties) mytable as col1, col2;

• 此方法更为方便日常使用。执行过程相当于单独执行了两次抽取，然后union到一个表里。

4. 参考文档

 

http://wiki.apache.org/hadoop/Hive/LanguageManual/UDF

http://wiki.apache.org/hadoop/Hive/DeveloperGuide/UDTF

http://www.slideshare.net/pauly1/userdefined-table-generating-functions