如何每月为每个 id 创建一行？

Question

我想每月为每个 id 创建一行，直到end_date月份。

例如，第一个客户id开始于 10 月并于 11 月结束。所以我想为客户活跃的每个月获取两行。 除此之外，我想创建一个列来标记它是否在该月处于活动状态。

| id | start_date |  end_date  |
|----|------------|------------|
| a  | 2021-10-02 | 2021-11-15 |
| b  | 2021-11-13 | 2021-11-30 |
| c  | 2021-11-16 |            |

当没有end_date时，意味着它仍然处于活动状态，它必须一直到当前月份。

示例数据：

WITH t1 AS (
SELECT 'a' AS id, '2021-10-02'::date AS start_date, '2021-11-15'::date AS end_date UNION ALL
SELECT 'b' AS id, '2021-11-13'::date AS start_date, '2021-11-30'::date AS end_date UNION ALL
SELECT 'c' AS id, '2021-11-16'::date AS start_date, NULL::date AS end_date
    )

预期结果：

| id | start_date |  end_date  |   months   | is_active |
|----|------------|------------|------------|-----------|
| a  | 2021-10-02 | 2021-11-15 | 2021-10-01 | TRUE      |
| a  | 2021-10-02 | 2021-11-15 | 2021-11-01 | FALSE     |
| b  | 2021-11-13 | 2021-11-30 | 2021-11-01 | FALSE     |
| c  | 2021-11-16 |            | 2021-11-01 | TRUE      |
| c  | 2021-11-16 |            | 2021-12-01 | TRUE      |
| c  | 2021-11-16 |            | 2022-01-01 | TRUE      |

我怎样才能在雪花中实现这一点？

Answer 1

所以如果你有一个范围，你将需要一些跨越时间的东西来加入，这是可以使用生成器的地方，我将把它放入 CTE 中。 我还将使用 ROW_NUMBER() 生成月份步骤序列，以确保没有间隙。 200需要进行硬编码，因此请输入一个涵盖足够数据以满足您需要的值，或者将其弹出到表中。

WITH months AS (
    SELECT 
        ROW_NUMBER() OVER (ORDER BY NULL) - 1 as rn
    FROM TABLE(generator(rowcount => 200))
)

接下来我们要截断start_date并找到 end_date 之后的月数，并将其加入我们的范围

), range_prep AS (
    SELECT id,
        start_date,
        end_date,
        date_trunc(month, start_date) as start_month,
        datediff(month, start_month, coalesce(end_date, CURRENT_DATE())) as month_count
    FROM data
)

将它们连接在一起，并做：

SELECT id,
    r.start_date,
    r.end_date,
    dateadd(month, m.rn, r.start_month) as months,
    (r.end_date is null) OR (date_trunc(month, r.end_date) > months) AS is_active
FROM range_prep as r
JOIN months as m
    ON m.rn <= r.month_count
ORDER BY 1,2;

将所有数据与 CTE 放在一起，以获得我们拥有的data ：

WITH data AS (
    SELECT id, 
        to_date(start_date) as start_date, 
        to_date(end_date) as end_date  
    FROM VALUES 
        ('a','2021-10-02','2021-11-15'),
        ('b','2021-11-13','2021-11-30'),
        ('c','2021-11-16',null)
    v( id, start_date, end_date)
), months AS (
    SELECT 
        ROW_NUMBER() OVER (ORDER BY NULL) - 1 as rn
    FROM TABLE(generator(rowcount => 200))
), range_prep AS (
    SELECT id,
        start_date,
        end_date,
        date_trunc(month, start_date) as start_month,
        datediff(month, start_month, coalesce(end_date, CURRENT_DATE())) as month_count
    FROM data
)
SELECT id,
    r.start_date,
    r.end_date,
    dateadd(month, m.rn, r.start_month) as months,
    (r.end_date is null) OR (date_trunc(month, r.end_date) > months) AS is_active
FROM range_prep as r
JOIN months as m
    ON m.rn <= r.month_count
ORDER BY 1,2;

给出：

ID	开始日期	结束日期	几个月	活跃
一个	2021-10-02	2021-11-15	2021-10-01	真的
一个	2021-10-02	2021-11-15	2021-11-01	错误的
b	2021-11-13	2021-11-30	2021-11-01	错误的
c	2021-11-16		2021-11-01	真的
c	2021-11-16		2021-12-01	真的
c	2021-11-16		2022-01-01	真的