distinct去重多个字段,其他字段不去重,一起输出

例如：

select AA, BB, CC
 from tableName;

要求是对AA，BB这两个字段都去重，CC是不需要去重的，最后一起输出

求解！！！

下面的内容全部是在Hive/Sparl SQL中实现的, 因为MySQL5.7这种没有窗口函数

关于distinct需要去重多个字段的问题, 下面的例子希望可以给你们带来启发, 大家一起讨论一下

distinct去重多个字段可以考虑 row_number() over(partition by 字段1[,字段2] order by 日期) 这种形式: 

原数据:

-- 原表中中的数据:

select id
       ,substring(recode_date, 1, 10) as create_date
       ,device_id
from dwd.dwd_ce_user_log
where substring(recode_date, 1, 10) = '2020-08-24'
  and device_id = '00de6cbcf0bbc87f';


id              create_date     device_id
76052968        2020-08-24      00de6cbcf0bbc87f
76052973        2020-08-24      00de6cbcf0bbc87f
76061303        2020-08-24      00de6cbcf0bbc87f
76061309        2020-08-24      00de6cbcf0bbc87f
76090080        2020-08-24      00de6cbcf0bbc87f
76129054        2020-08-24      00de6cbcf0bbc87f
76139129        2020-08-24      00de6cbcf0bbc87f
76039128        2020-08-24      00de6cbcf0bbc87f
76039129        2020-08-24      00de6cbcf0bbc87f
76042362        2020-08-24      00de6cbcf0bbc87f
76042363        2020-08-24      00de6cbcf0bbc87f

 用distinct去重多个字段

-- 用distinct去重多个字段
select distinct substring(recode_date, 1, 10) as create_date
       ,device_id
from dwd.dwd_ce_user_log
where substring(recode_date, 1, 10) = '2020-08-24'
  and device_id = '00de6cbcf0bbc87f';


create_date     device_id
2020-08-24      00de6cbcf0bbc87f

用row_number() over(partition by 字段1[,字段2] order by 字段)

-- 用row_number() over(partition by 字段1[,字段2] order by 字段)

select substring(recode_date, 1, 10) as create_date
       ,device_id
from (
       select *, row_number() over(partition by device_id, substring(recode_date,1,10) order by recode_date desc) rank
       from dwd.dwd_ce_user_log
     ) tmp
where tmp.rank =1
  and substring(recode_date, 1, 10) = '2020-08-24'
  and device_id = '00de6cbcf0bbc87f';
  

create_date     device_id
2020-08-24      00de6cbcf0bbc87f