public class Tex extends GenericUDTF {

     /**

      * 对传入的参数进行初始化

      * 判断参数个数/类型

      * 初始化表结构

      */

     @Override

     public StructObjectInspector initialize(ObjectInspector[] argOIs) throws UDFArgumentException {

         if (argOIs.length != 1) {

             throw new UDFArgumentLengthException("actuly only one argument is expected");

         }

         if (argOIs[0].getCategory() != ObjectInspector.Category.PRIMITIVE) {

             throw new UDFArgumentTypeException(0, "type of String is expected but " + argOIs[0].getTypeName() + "is passed");

         }

         //初始化表结构

         //创建数组列表存储表字段

         List<String> fieldNames = new LinkedList<String>();

         List<ObjectInspector> fieldIOs = new LinkedList<ObjectInspector>();

         //表字段

         fieldNames.add("name");

         fieldNames.add("age");

         //表字段数据类型

         fieldIOs.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);

         fieldIOs.add(PrimitiveObjectInspectorFactory.javaStringObjectInspector);

         //将表结构两部分聚合在一起

         return ObjectInspectorFactory.getStandardStructObjectInspector(fieldNames, fieldIOs);

     }

     /**

      * 对数据处理的代码

      * 如果是多列的话,可以将每一行的数据存入数组中,然后将数组传入forward,

      * forward每调用一次都会产生一行数据

      */

     @Override

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

         String str = args[0].toString();

         String[] splited = str.split(";");

         for (int i = 0; i < splited.length; i++) {

             try {

                 String[] res = splited[i].split(",");

                 forward(res);

             } catch (Exception e) {

                 continue;

             }

         }

     }

     //方法调用完毕时关闭方法

     @Override

     public void close() throws HiveException {

     }

 }