Hive:如何消除重复的子串

Hive: how to eliminate the duplicated substrings

蜂巢 table:

  create table T (i int, s string);

  insert into T values
  (1, "a1&b2"),
  (1, "b2&c3"),
  (2, "c1&d2"),
  (2, "c1");

s 列包含由 &

分隔的值

所需的输出应按第 1 列分组并连接 s 列,但只有一个唯一的子字符串值(由 & 分隔):

   i    grouped_s
   --   -------------
   1    a1&b2&c3
   2    c1&d2

这是我的尝试:

SELECT i,
concat_ws('&',
 collect_set(
  split(concat_ws('&' , collect_set(s)), "&" )
 )
)
as grouped_s
FROM T group by i;

我知道了:

FAILED: SemanticException [Error 10128]: Line 6:24 Not yet supported place for UDAF 'collect_set' 

我也想在不使用嵌套 SQL 的情况下实现它。

使用lateral view + explode:

select t.i, concat_ws('&',collect_set(e.val)) as grouped_s
  from T t 
       lateral view outer explode(split(t.s,'&')) e as val
 group by t.i; 

结果:

t.i grouped_s
1   a1&b2&c3
2   c1&d2