Question

我有一个任务表，我查询并尝试按截止日期对结果进行分组，并将每组返回的任务数量限制为10个。

     id    |     name         | duedate
---------------------------------------------
    101    |      task1       | 2015-03-05
    102    |      task2       |
    104    |      task3       | 2015-03-06
    105    |      task4       | 2015-03-07
    106    |      task5       | 
    107    |      task6       | 2015-02-28
    108    |      task7       | 2015-02-27
    109    |      task8       | 2015-02-05
    111    |      task9       |

我尝试将任务分为4组，并将每组限制为10个任务：

Duedate =今天
Duedate =今天 - 7天
Duedate =今天+ 7天
Duedate = null

我使用postgres窗口功能来实现这一目标。问题是我无法使用＆＃34;分区＆＃34;对于团体。

到目前为止，我已经设法提出这个：

select * from ( select id,name,duedate,row_number()  
  OVER (PARTITION BY duedate order by duedate)  as rownum from tasks
  ) tmp where rownum < 10;

这不会起作用，因为它为每个人都创造了一个小组。

是否可以不使用多个查询？

Answer 1

您可以在case：

中使用partition by表达式

select *
from (select id, name, duedate,
             row_number()  over (partition by (case when duedate = CURRENT_DATE then 1
                                                    when duedate >= CURRENT_DATE - interval '7 day' and duedate < CURRENT_DATE then 2
                                                    when duedate <= CURRENT_DATE + interval '7 day' and duedate > CURRENT_DATE then 3
                                               end)
                                 order by duedate) as rownum
      from tasks
      where duedate >= CURRENT_DATE - interval '7 day' and
            duedate <= CURRENT_DATE + interval '7 day'
     ) tmp
where rownum < 10;

Answer 2

如果您的目的是形成代表

的四个小组

今天
一周前和今天（今天除外）之间的项目
自今天起一周内（今天除外）的物品
缺少duedate

并忽略所有其他日期，然后这样的事情会起作用：

select *,
  -- I left the case expression in to show what it does.
  case 
    when Duedate = current_date 
    then 1
    when Duedate >= current_date - interval '7 days' 
     and Duedate < current_date 
    then 2
    when Duedate <= current_date + interval '7 days' 
     and Duedate > current_date 
    then 3
  end as grouping
from ( 
  select 
    id,
    name,
    duedate,
    row_number() OVER (PARTITION BY 
      case 
        when Duedate = current_date then 1
        when Duedate >= current_date - interval '7 days' 
                       and Duedate < current_date 
                       then 2
        when Duedate <= current_date + interval '7 days' 
                       and Duedate > current_date 
                       then 3
      end 
      order by duedate
  ) as rownum 
  from tasks
  where (Duedate >= current_date - interval '7 days'
    and Duedate <= current_date + interval '7 days') 
     or Duedate is null

) tmp where rownum < 10;

Sample SQL Fiddle（略微修改日期以获得一些结果）

这会产生类似于此的结果：

|  id |   name |                 duedate | rownum | grouping |
|-----|--------|-------------------------|--------|----------|
| 112 | task10 | April, 04 2015 00:00:00 |      1 |        1 |
| 113 | task11 | April, 04 2015 00:00:00 |      2 |        1 |
| 107 |  task6 | March, 28 2015 00:00:00 |      1 |        2 |
| 108 |  task7 | March, 29 2015 00:00:00 |      2 |        2 |
| 104 |  task3 | April, 06 2015 00:00:00 |      1 |        3 |
| 105 |  task4 | April, 07 2015 00:00:00 |      2 |        3 |
| 101 |  task1 | April, 11 2015 00:00:00 |      3 |        3 |
| 102 |  task2 |                  (null) |      1 |   (null) |
| 106 |  task5 |                  (null) |      2 |   (null) |
| 111 |  task9 |                  (null) |      3 |   (null) |

按日期限制

2 个答案: