基于sparksql collect_list的udf定义踩坑

多条collect_list，然后将collect_list的结果concat起来，最初使用的是concat_ws()，但是发现超过4个collect_list就会报错，

select concat_ws("|",
collect_list(concat_ws(',',n_cgi_1,ltencrsrp_1)),
collect_list(concat_ws(',',n_cgi_2,ltencrsrp_2)),
collect_list(concat_ws(',',n_cgi_3,ltencrsrp_3)),
collect_list(concat_ws(',',n_cgi_4,ltencrsrp_4)),
collect_list(concat_ws(',',n_cgi_5,ltencrsrp_5)),
collect_list(concat_ws(',',n_cgi_6,ltencrsrp_6))) as result
from test group by id,name;

于是考虑自定义UDF函数。

collect_list函数在hive中返回值类型为array<T>，对应java的arrayList<T>,但是在写spark UDF时候报错：

Spark java.lang.ClassCastException: scala.collection.mutable.WrappedArray$ofRef cannot be cast to java.util.ArrayList

选择了一个不够牛逼但是很方便的方式，样例代码如下：

package com.kong.test.UDF;

import org.apache.spark.sql.api.java.UDF5;
import scala.collection.mutable.WrappedArray;

public class TestArray implements UDF5<WrappedArray<String>, WrappedArray<String>, WrappedArray<String>, WrappedArray<String>, WrappedArray<String>, String>  {
	private static final long serialVersionUID = 1L;

	//将array中的元素取出来，并以|隔开
	public String call(WrappedArray<String> t1, WrappedArray<String> t2, WrappedArray<String> t3,
			WrappedArray<String> t4, WrappedArray<String> t5) throws Exception {
		StringBuffer sb = new StringBuffer();

		for (int i = 0; i < t1.length(); i++) {
			String ele = t1.apply(i);
			System.out.println(ele);
			if(!"".equals(ele)){
				sb.append(ele).append("|");
			}
		}

		for (int i = 0; i < t2.length(); i++) {
			String ele = t2.apply(i);
			System.out.println(ele);
			if(!"".equals(ele)){
				sb.append(ele).append("|");
			}
		};
		for (int i = 0; i < t3.length(); i++) {
			String ele = t3.apply(i);
			System.out.println(ele);
			if(!"".equals(ele)){
				sb.append(ele).append("|");
			}
		};
		for (int i = 0; i < t4.length(); i++) {
			String ele = t4.apply(i);
			System.out.println(ele);
			if(!"".equals(ele)){
				sb.append(ele).append("|");
			}
		};
		for (int i = 0; i < t5.length(); i++) {
			String ele = t5.apply(i);
			System.out.println(ele);
			if(!"".equals(ele)){
				sb.append(ele).append("|");
			}
		};
		System.out.println(sb.toString());

		String res = sb.toString();
		String res1 = res.substring(0, res.length()-1);
		return res1;
	}

}