博客

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