如何根据条件（SQL）从 1 条记录创建更多记录

Question

我正在尝试根据某些条件从 1 条记录创建更多记录。 例如：如果用户购买了包月套餐，该表将创建每周的记录，直到日期结束。

从这个表：

|Date Purchased| Date Ended| Package|
|2019-1-1      | 2019-2-1  | 1Month |

对此：

|Date Purchased| Date Ended| Package|
|2019-1-1      | 2019-2-1  | 1Month |
|2019-1-8      | 2019-2-1  | 1Month |
|2019-1-15     | 2019-2-1  | 1Month |
|2019-1-23     | 2019-2-1  | 1Month |

关于如何在 SQL 中执行此操作的任何想法？ 谢谢

Answer 1

您可以使用GENERATE_DATE_ARRAY()和UNNEST() ：

SELECT dte, t.ended, t.package
FROM (select t,
             GENERATE_DATE_ARRAY(t.purchased, t.ended, interval 1 week) as dtes
      from (select date('2019-01-01') as purchased, date('2019-02-01') as ended,  '1MOnth' as package
           ) t
     ) t CROSS JOIN
     UNNEST(t.dtes) dte

Answer 2

我相信您的要求是以某种方式使两个日期之间的所有星期都变得如此。 试试这个链接：

如何获取SQL Server中两个日期之间所有星期的开始日期和结束日期？

Answer 3

以下是 BigQuery 标准 SQL

#standardSQL
SELECT * EXCEPT(day) REPLACE(day AS date_purchased)
FROM `project.dataset.table`, UNNEST((
  SELECT CASE package 
    WHEN 'Week' THEN GENERATE_DATE_ARRAY(date_purchased, date_ended, INTERVAL 1 DAY)
    WHEN 'Month' THEN GENERATE_DATE_ARRAY(date_purchased, date_ended, INTERVAL 1 WEEK)
    WHEN 'Quarter' THEN GENERATE_DATE_ARRAY(date_purchased, date_ended, INTERVAL 1 MONTH)
    WHEN 'Year' THEN GENERATE_DATE_ARRAY(date_purchased, date_ended, INTERVAL 1 QUARTER)
    ELSE GENERATE_DATE_ARRAY(date_purchased, date_ended, INTERVAL 1 MONTH)
  END  
)) day   

您可以测试，使用虚拟数据在上面播放，如下例所示

#standardSQL
WITH `project.dataset.table` AS (
  SELECT DATE '2019-01-01' date_purchased, DATE '2019-02-01' date_ended, 'Month' package UNION ALL
  SELECT '2019-01-01', '2019-03-31', 'Quarter' UNION ALL
  SELECT '2019-01-01', '2020-12-31', 'Year' 
)
SELECT * EXCEPT(day) REPLACE(day AS date_purchased)
FROM `project.dataset.table`, UNNEST((
  SELECT CASE package 
    WHEN 'Week' THEN GENERATE_DATE_ARRAY(date_purchased, date_ended, INTERVAL 1 DAY)
    WHEN 'Month' THEN GENERATE_DATE_ARRAY(date_purchased, date_ended, INTERVAL 1 WEEK)
    WHEN 'Quarter' THEN GENERATE_DATE_ARRAY(date_purchased, date_ended, INTERVAL 1 MONTH)
    WHEN 'Year' THEN GENERATE_DATE_ARRAY(date_purchased, date_ended, INTERVAL 1 QUARTER)
    ELSE GENERATE_DATE_ARRAY(date_purchased, date_ended, INTERVAL 1 MONTH)
  END  
)) day   

结果

Row date_purchased  date_ended  package  
1   2019-01-01      2019-02-01  Month    
2   2019-01-08      2019-02-01  Month    
3   2019-01-15      2019-02-01  Month    
4   2019-01-22      2019-02-01  Month    
5   2019-01-29      2019-02-01  Month    
6   2019-01-01      2019-03-31  Quarter  
7   2019-02-01      2019-03-31  Quarter  
8   2019-03-01      2019-03-31  Quarter  
9   2019-01-01      2020-12-31  Year     
10  2019-04-01      2020-12-31  Year     
11  2019-07-01      2020-12-31  Year     
12  2019-10-01      2020-12-31  Year     
13  2020-01-01      2020-12-31  Year     
14  2020-04-01      2020-12-31  Year     
15  2020-07-01      2020-12-31  Year     
16  2020-10-01      2020-12-31  Year