在提取数据库中重复行的查询结果中仅选择较新的记录

Posted 2023-02-16

【中文标题】在提取数据库中重复行的查询结果中仅选择较新的记录

【英文标题】：Select only newer records among the result of a query that extracts duplicate rows in a db

【发布时间】：2021-03-20 08:27:19

【问题描述】：

我有一个数据库，我可以在其中有多个具有相同字段“帮助”的行（这是一种相同项目的历史变更日志）。 通过以下查询，我将提取具有相同帮助的行，按日期降序排列：

select aid,data_ril,specie,id from scu.censimento t1
where (select count(*) from scu.censimento t2
where t1.aid = t2.aid) > 1
order by aid, data_ril desc

这应该是与父层相关的子表（由仅显示最新援助记录的 postgres 物化视图制作），以下是我得到的结果：

    +------+------------+--------+------+
    | aid  | data_ril   | specie | id   |
    +------+------------+--------+------+
    | 349  | 2020-06-18 | 35     | 349  |
    +------+------------+--------+------+
    | 349  | 2020-06-17 | 35     | 2004 |
    +------+------------+--------+------+
    | 700  | 2020-08-07 | 58     | 700  |
    +------+------------+--------+------+
    | 700  | 2020-07-06 | 58     | 2006 |
    +------+------------+--------+------+
    | 700  | 2020-05-02 | 15     | 1956 |
    +------+------------+--------+------+
    | 1316 | 2020-09-02 | 1      | 1316 |
    +------+------------+--------+------+
    | 1316 | 2020-08-27 | 1      | 2005 |
    +------+------------+--------+------+
    | 1317 | 2020-09-02 | 2      | 1317 |
    +------+------------+--------+------+
    | 1317 | 2020-08-27 | 2      | 1996 |
    +------+------------+--------+------+
    | 1481 | 2020-12-03 | 21     | 2112 |
    +------+------------+--------+------+
    | 1481 | 2020-09-08 | 49     | 1481 |
    +------+------------+--------+------+
    | 1492 | 2020-09-28 | 6      | 1492 |
    +------+------------+--------+------+
    | 1492 | 2020-09-08 | 6      | 1999 |
    +------+------------+--------+------+
    | 1688 | 2020-11-03 | 72     | 1688 |
    +------+------------+--------+------+
    | 1688 | 2020-10-08 | 72     | 2000 |
    +------+------------+--------+------+

我想知道修改上述查询的 SQL 语法，以便显示所有重复行，但日期最新的行除外，这样我就可以拥有如下表：

+------+------------+--------+------+
| aid  | data_ril   | specie | id   |
+------+------------+--------+------+
| 349  | 2020-06-17 | 35     | 2004 |
+------+------------+--------+------+
| 700  | 2020-07-06 | 58     | 2006 |
+------+------------+--------+------+
| 700  | 2020-05-02 | 15     | 1956 |
+------+------------+--------+------+
| 1316 | 2020-08-27 | 1      | 2005 |
+------+------------+--------+------+
| 1317 | 2020-08-27 | 2      | 1996 |
+------+------------+--------+------+
| 1481 | 2020-09-08 | 49     | 1481 |
+------+------------+--------+------+
| 1492 | 2020-09-08 | 6      | 1999 |
+------+------------+--------+------+
| 1688 | 2020-10-08 | 72     | 2000 |
+------+------------+--------+------+

提前致谢。

【参考方案1】：

您可以使用窗口函数来做到这一点。这个想法是通过降序data_ril对具有相同aid的记录进行排名，然后过滤掉每个组的顶部记录。

select aid, data_ril, specie, id
from (
    select t.*, 
        row_number() over(partition by aid order by data_ril desc) rn
    from mytable t 
) t
where rn > 1
order by aid, data_ril