自定义UDAF函数

Java 实现自定义UDAF函数，代码如下：

package com.**;

import org.apache.spark.sql.Row;
import org.apache.spark.sql.expressions.MutableAggregationBuffer;
import org.apache.spark.sql.expressions.UserDefinedAggregateFunction;
import org.apache.spark.sql.types.DataType;
import org.apache.spark.sql.types.DataTypes;
import org.apache.spark.sql.types.StructType;

public class SelfConcatFunction extends UserDefinedAggregateFunction {
    @Override
    public StructType inputSchema() {
        return new StructType().add("SELFCONCAT",DataTypes.StringType);
    }

    @Override
    public StructType bufferSchema() {
        return new StructType().add("concat_str",DataTypes.StringType);
    }

    @Override
    public DataType dataType() {
        return DataTypes.StringType;
    }

    @Override
    public boolean deterministic() {
        return true;
    }

    @Override
    public void initialize(MutableAggregationBuffer mutableAggregationBuffer) {
        mutableAggregationBuffer.update(0,null);
    }

    @Override
    public void update(MutableAggregationBuffer mutableAggregationBuffer, Row row) {
        if(mutableAggregationBuffer.getString(0)!=null){
            if(row.getString(0)!=null){
                String tmp = mutableAggregationBuffer.getString(0)+"|"+row.getString(0);                
                if(!mutableAggregationBuffer.getString(0).contains(row.getString(0))){
                    mutableAggregationBuffer.update(0,tmp);
                }
            }
        }else{
            if(row.getString(0)!=null){
                mutableAggregationBuffer.update(0,row.getString(0));
            }
        }
    }

    @Override
    public void merge(MutableAggregationBuffer mutableAggregationBuffer, Row row) {
        if(mutableAggregationBuffer.getString(0)!=null){
            if(row.getString(0)!=null){
                String tmp=mutableAggregationBuffer.getString(0)+"|"+row.getString(0);
                if(!mutableAggregationBuffer.getString(0).contains(row.getString(0))){
                    mutableAggregationBuffer.update(0,tmp);
                }
            }
        }else{
            if(row.getString(0)!=null){
                mutableAggregationBuffer.update(0,row.getString(0));
            }
        }
    }

    @Override
    public Object evaluate(Row row) {
        if(row.getString(0)!=null){
            return "|"+row.getString(0)+"|";
        }
        else{
            return null;
        }

    }
}

SelfConcatFunction concatFunc= new SelfConcatFunction();//引用自定义函数
sparkSession.udf().register("CONCAT_BUCKET", concatFunc);//注册到sparkSession

在聚合SQL中使用该注册的函数： CONCAT_BUCKET

select CONCAT_BUCKET(AAA) as combine_aaa,bbb from tableName group by bbb;

后来发现，hive内置函数也可以实现相同功能，SQL如下：

select concat_ws('|',collect_set(AAA)) as combine_aaa,bbb from tableNamegroup by bbb;