如何获得每个状态的第一行？

Posted 2023-03-31

【中文标题】如何获得每个状态的第一行？

【英文标题】：How to get first row of each status?

【发布时间】：2021-08-23 19:33:25

【问题描述】：

我想获取每个 id 的每个状态的第一行。

每个状态可以有多行。所以我想根据之前的状态来获取每个状态的第一次出现。

例如info_required 首先出现在第 2 行，然后在第 4 行变为另一个状态 pending，然后在第 6 行再次变为 info_required。 同样，状态 pending 首先在第 4 行，然后在第 8 行，因为在第 4 行之后状态发生了变化，它需要在结果集中。

因此我想在下面获取第 1、2、4、6 和 8 行。

WITH t1 AS (
SELECT 1 AS row, 'A' AS id, 'created' AS status, '2021-05-18 18:30:00'::timestamp AS created_at UNION ALL
SELECT 2 AS row, 'A' AS id, 'info_required' AS status, '2021-05-19 11:30:00'::timestamp AS created_at UNION ALL
SELECT 3 AS row, 'A' AS id, 'info_required' AS status, '2021-05-19 12:00:00'::timestamp AS created_at UNION ALL
SELECT 4 AS row, 'A' AS id, 'pending' AS status, '2021-05-19 12:30:00'::timestamp AS created_at UNION ALL
SELECT 5 AS row, 'A' AS id, 'pending' AS status, '2021-05-20 13:30:00'::timestamp AS created_at UNION ALL
SELECT 6 AS row, 'A' AS id, 'info_required' AS status, '2021-05-20 14:30:00'::timestamp AS created_at UNION ALL
SELECT 7 AS row, 'A' AS id, 'info_required' AS status, '2021-05-20 15:30:00'::timestamp AS created_at UNION ALL
SELECT 8 AS row, 'A' AS id, 'pending' AS status, '2021-05-20 16:30:00'::timestamp AS created_at
    )
SELECT *
FROM t1

【参考方案1】：

使用CONDITIONAL_CHANGE_EVENT

WITH cte AS (
  SELECT *, CONDITIONAL_CHANGE_EVENT(status) over (partition by id 
                                                   order by created_at) AS cce
  FROM t1
)
SELECT *
FROM cte
QUALIFY ROW_NUMBER() OVER(PARTITION BY id, cce ORDER BY created_at) = 1;

---

数据准备：

CREATE TABLE t1 AS 
WITH t1 AS (
SELECT 1 AS row_, 'A' AS id, 'created' AS status, '2021-05-18 18:30:00'::timestamp AS created_at UNION ALL
SELECT 2 AS row_, 'A' AS id, 'info_required' AS status, '2021-05-19 11:30:00'::timestamp AS created_at UNION ALL
SELECT 3 AS row_, 'A' AS id, 'info_required' AS status, '2021-05-19 12:00:00'::timestamp AS created_at UNION ALL
SELECT 4 AS row_, 'A' AS id, 'pending' AS status, '2021-05-19 12:30:00'::timestamp AS created_at UNION ALL
SELECT 5 AS row_, 'A' AS id, 'pending' AS status, '2021-05-20 13:30:00'::timestamp AS created_at UNION ALL
SELECT 6 AS row_, 'A' AS id, 'info_required' AS status, '2021-05-20 14:30:00'::timestamp AS created_at UNION ALL
SELECT 7 AS row_, 'A' AS id, 'info_required' AS status, '2021-05-20 15:30:00'::timestamp AS created_at UNION ALL
SELECT 8 AS row_, 'A' AS id, 'pending' AS status, '2021-05-20 16:30:00'::timestamp AS created_at
)
SELECT *
FROM t1;

Cte部分：

SELECT *, CONDITIONAL_CHANGE_EVENT(status) over (partition by id 
                                              order by created_at) AS cce
FROM t1;

【参考方案2】：

您可以使用lag() 和qualify()：

select t.*
from t
qualify lag(status) over (partition by id order by created_at) is distinct from status;