我在选择 Hive 中的结构数组时遇到问题

Question

我已阅读网站上的所有示例，但没有一个适合我。我添加了 Brickhouse UDFS：
create temporary function collect_struct as 'brickhouse.udf.collect.CollectUDAF' 和

CREATE TEMPORARY FUNCTION collect AS 'brickhouse.udf.collect.CollectUDAF'

我使用了来自Hive SELECT statement to create an ARRAY of STRUCTS 和
Hive select data into an array of structs 的示例

我的 SQL：

create temporary function collect as 'brickhouse.udf.collect.CollectUDAF';

CREATE TABLE test ( <br>
house_id bigint,<br>
first_name string,<br>
last_name string<br>
);

INSERT INTO test VALUES<br>
(1,"Bob","jones"),<br>
(1,"Jenny","jones"),<br>
(2,"Sally","johnson"),<br>
(3,"John","smith"),<br>
(3,"Barb","smith");

select house_id<br>
      ,collect(named_struct("first_name", first_name, "last_name", last_name))<br>
from test<br>
group by house_id;

我总是遇到同样的错误：

失败：ClassCastException org.apache.hadoop.hive.serde2.objectinspector.StandardStructObjectInspector 不能转换为 org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector

有什么想法吗？

Answer 1

我使用的是 Hive 版本 1.1.0，并且能够创建结构数组，而无需 Brickhouse jar 文件的 UDF。试试这个。

SELECT house_id ,
       collect_list(named_struct("first_name", first_name, "last_name", last_name))
FROM house
GROUP BY house_id;

1   [{"first_name":"Bob","last_name":"jones"},{"first_name":"Jenny","last_name":"jones"}]   
2   [{"first_name":"Sally","last_name":"johnson"}]  
3   [{"first_name":"John","last_name":"smith"},{"first_name":"Barb","last_name":"smith"}]