sql - 在 PostgreSQL 中将重叠的间隔分割成较小的、接触的间隔

Question

我需要有关查询的帮助，该查询应按以下方式分割一些重叠的日期间隔。

例如，如果我们有下表

----------------------------------------------------------------------
|   column_name    |    value    |    start_date    |    end_date    |
----------------------------------------------------------------------
|     column1      |    value1   |    03-09-2020    |   26-09-2020   |
----------------------------------------------------------------------
|     column1      |    value2   |    07-09-2020    |   20-09-2020   |
----------------------------------------------------------------------

我在提出查询以生成以下输出时遇到问题:

----------------------------------------------------------------------
|   column_name    |    value    |    start_date    |    end_date    |
----------------------------------------------------------------------
|     column1      |    value1   |    03-09-2020    |   07-09-2020   |
----------------------------------------------------------------------
|     column1      |    value2   |    07-09-2020    |   20-09-2020   |
----------------------------------------------------------------------
|     column1      |    value1   |    20-09-2020    |   26-09-2020   |
----------------------------------------------------------------------

Answer 1

一种方法是获取所有日期并使用横向连接来计算该日期的值:

with dates as (
      select start_date as date
      from t
      union 
      select end_date
      from t
     )
select tt.value, d.date as start_date, lead(d.date) over (order by d.date) as end_date
from dates d left join lateral
     (select t.*
      from t
      where start_date <= d.date and
            end_date > d.date
      order by start_date desc  -- this gets the most recent overlapping value
      limit 1
     ) tt
     on true
order by d.date;

Here是一个数据库<> fiddle 。

请注意，这会添加一个额外的行，其中包含最后的“无值”句点。如果您不需要，可以轻松过滤掉它。