遍历数据，计算每个月最后三天的总和

Question

DB-Fiddle

CREATE TABLE sales (
    id SERIAL PRIMARY KEY,
    event_date DATE,
    sales_volume INT
);

INSERT INTO sales
(event_date, sales_volume)
VALUES 
('2023-01-27', '900'),
('2023-02-28', '500'),
('2023-01-29', '100'),
('2023-01-30', '600'),
('2023-01-31', '450'),
  
('2023-02-25', '300'),
('2023-02-26', '250'),
('2023-02-27', '845'),
('2023-02-28', '520'), 

('2023-03-26', '750'),
('2023-03-27', '750'),
('2023-03-28', '625'),
('2023-03-29', '885'),
('2023-03-30', '120'),
('2023-03-31', '400');

---

预期结果：

Row	event_time	sales_volume
1	1	1150
2	2	2115
3	3	1405

---

我想要迭代通过表格计算和sales_volume 的每一个月.
目前，我能够使用此查询实现此目的：

SELECT
DATE_PART('month', s.event_date) AS event_time,
SUM(s.sales_volume) AS sales_volume
FROM sales s
WHERE s.event_date BETWEEN '2023-01-29' AND '2023-01-31'
OR s.event_date BETWEEN '2023-02-26' AND '2023-02-28'
OR s.event_date BETWEEN '2023-03-29' AND '2023-03-31'
GROUP BY 1
ORDER BY 1;

但是，我的原始数据库比示例大得多，所以我需要很多这样的 OR 条件。

我需要如何修改查询以便它在不使用 OR 条件的情况下遍历表？

Answer 1

demo:db<>fiddle

SELECT
    row_number,
    SUM(sales_volume)
FROM (
    SELECT
        *,
        row_number() OVER (
            PARTITION BY date_trunc('month', event_date) 
            ORDER BY event_date DESC
        )
    FROM sales 
) s
WHERE row_number <= 3
GROUP BY row_number