旋转数据并将单个列放入具有相关行的多列中

Posted 2023-01-18

我在Amazon Redshift（PostgreSQL）中有如下数据：

acct | id_cd |用户名

acc1 | 123 |查尔姆acc1 | 123 |史密斯acc1 | 123 |埃哈蒙acc1 | 123 |列宁acc2 | 234 |查尔姆acc2 | 234 |瑞蒂斯acc3 | 345 |贾巴尔维acc3 | 345 |双重acc3 | 345 |吉利acc4 | 345 | rshant

而且我需要它看起来像这样。

acct | id_cd |用户名1 |用户名2 |用户名3

acc1 | 123 | kchalm |史密斯|埃哈蒙acc1 | 123 |列宁| |acc2 | 234 | kchalm | rtheis |acc3 | 345 | jbalvi | twoods |吉利acc4 | 345 | rshant | |

我已经尝试了以下成功的尝试。为了创建3个用户名列，我使用带有WHERE标准的联接将数据从视图中拉出所需的次数。但是，我不禁感到比现在可以自动化该过程的方法更好的方法。有什么建议吗？

    SELECT
    ut.acct,
    ut.id_code,
    ut.username,
    ROW_NUMBER() OVER(PARTITION BY ut.account_code
ORDER BY
    ut.username DESC) AS row_num
FROM
    orig_user_table ut ;

select
    a.acct
    ,a.id_cd
    ,a.username1
    ,b.username2
FROM
    (
    SELECT
        acct,
        id_cd,
        username AS username1
    FROM
        vw_user_table_nums
    WHERE
        row_num = 1) a
LEFT JOIN (
    SELECT
        acct,
        username AS username2
    FROM
        vw_user_table_nums
    WHERE
        row_num = 2) b ON
    a.acct = b.acct
    ...

答案

从现有子查询开始，一种方法使用条件聚合而不是联接：

select
    acct,
    id_code,
    max(case when rn = 1 then username end) username1,
    max(case when rn = 2 then username end) username2,
    max(case when rn = 3 then username end) username3
from (
    select
        ut.*,
        row_number() over(partition by ut.account_code order by ut.username desc) as rn
    from orig_user_table ut 
) t
group by acct, id_code
order by acct, id_code;