平时很少用到map类型，这次用到了记录一下。
举个栗子：

select 
    mid,
    tags['电影'] as movie,
    tags['音乐'] as music
from(
select 
    mid,str_to_map(concat_ws(',',collect_list(concat(key,":",value)))) as tags 
from (
 select
    mid,
    key,
    concat_ws('-',collect_list(value))as value
from
(
    select 1 as mid,'电影' as key,'惊悚' as value
    union ALL
    select 1 as mid,'音乐' as key,'摇滚' as value 
    union ALL
    select 1 as mid,'电影' as key,'喜剧' as value 
) a
group by mid,key)t
group by mid) tt

构造数据：

运行结果：

适用场景：行转列，且mid的key个数可能不一样的情况。