如何在结果集中使用 2 个分组列进行 PIVOT？

Posted 2023-02-15

【中文标题】如何在结果集中使用 2 个分组列进行 PIVOT？

【英文标题】：How to PIVOT with 2 grouping columns in the result set?

【发布时间】：2020-12-09 05:24:54

【问题描述】：

我有一个查询输出以下内容：

ApptDate    Truck_ID   Item    Qty
'8-19-20'   TruckA     ItemA   100
'8-19-20'   TruckB     ItemA   200
'8-20-20'   TruckC     ItemB   300
'8-20-20'   TruckD     ItemB   400
...

我需要 PIVOT 以便它返回：

Item    Truck_ID    Day1    Day2 ... Day14
ItemA   TruckA      100     0        0
ItemA   TruckB      200     0        0
ItemB   TruckC      0       300      0
ItemB   TruckD      0       400      0

我试过了，但是报错了：

消息 8114，第 16 级，状态 1，第 413 行 将数据类型 nvarchar 转换为日期时间时出错。 消息 473，第 16 级，状态 1，第 413 行 PIVOT 运算符中提供了不正确的值“Day1”。

select
item, truck_id, Day1, Day2, Day3, Day4, Day5, Day6, Day7, Day8, Day9, Day10, Day11, Day12, Day13, Day14 
from(
select

    ds.ApptDate
    , c.truck_id
    , c.item 
    , sum(c.qty) qty

from

    maintable c with(nolock)
    inner join secondtable ds with(nolock) on c.truck_id = ds.truckid and ds.type = 'O'

where

    ds.apptdate between cast(getdate() as date) and dateadd(day, 14, cast(getdate() as date))
    and coalesce(ds.CancelTruck, 0) <> 1
    and ds.Status <> '5'

group by

    c.truck_id
    , c.item
    , ds.ApptDate

) sourcetable

pivot
(
sum(qty)
for apptdate in ([Day1], [Day2], [Day3], [Day4], [Day5], [Day6], [Day7], [Day8], [Day9], [Day10], [Day11], [Day12], [Day13], [Day14])

) as pivottable

【参考方案1】：

由于您期望列数固定，我们不一定需要动态 SQL。一种选择使用条件聚合......以及大量重复输入：

select
    item,
    truck_id,
    sum(case when appt_date = cast(getdate() as date)                    then qty else 0 end) day0,
    sum(case when appt_date = dateadd(day, -1 , cast(getdate() as date)) then qty else 0 end) day1,
    sum(case when appt_date = dateadd(day, -2 , cast(getdate() as date)) then qty else 0 end) day2,
    ...
    sum(case when appt_date = dateadd(day, -14, cast(getdate() as date)) then qty else 0 end) day14
from ( -- your current query here --) t
group by item, truck_id
    

【讨论】：

大声笑我希望使用 PIVOT 找到一个奇特的解决方案。不过还是谢谢！！

【参考方案2】：

这种方法在最小日期和 AppDate 上使用 datediff。

;with 
min_dt_cte(min_dt) as (select min(cast(AppDate as date)) from MyTable),
pvt_dt_cte(ApptDate, Truck_ID, Item, Qty, DayNum) as (
    select t.*, datediff(d, mdc.min_dt, cast(AppDate as date))
    from min_dt_cte mdc
         cross join
         MyTable t)
select
  pdc.Item, pdc.Truck_ID, 
  iif(pdc.DayNum=1, Qty, 0) Day1,
  iif(pdc.DayNum=2, Qty, 0) Day2,
  ...
  iif(pdc.DayNum=14, Qty, 0) Day14
from
  pvt_dt_cte pdc;