这是俺小伙伴写的： select dtm,id_,num_,sum(num_) over(partition by dtm,id) as total_num from tableA lateral view explode(split(id , ’,’)) split_id as id_ lateral view explode(split(num , ’,’)) split_num as num_ 但是不知道两个lateral view explode连接的结果是否能一一对应，于是我修改了一下 总体思路是先拆id列，每个id都会对应上num数组，此时编号，标记此id是id数组的第几个，再拆num数组，再以id分组编号，标记此num是num数组的第几位，当第一个编号=第二个编号，说明数组位置相同一一对应，不知道是否正确，欢迎各位小伙伴批评指正~ 这是俺修改后的： select dim,id1 as id,num1 as num,sum(num) over(partition by id1) as total_num from( select dim,id1,rank1,num1,rank() over(partition by id1) as rank2 from( select dtm,id1,num,rank() over(order by (select 0)) as rank1 from tableA lateral view explode(split(id , ’,’)) split_id as id1 ) lateral view explode(split(num , ’,’)) split_num as num1 ) where rank1=rank2