Sql 检索行,直到其数量总和小于@quantity
我有一张桌子: 数据标识、名称、数量 我想编写一个存储过程来获取@quantity并检索它们的数量总和大于或等于@quantity的最小行 例如:Sql 检索行,直到其数量总和小于@quantity,sql,tsql,stored-procedures,Sql,Tsql,Stored Procedures,我有一张桌子: 数据标识、名称、数量 我想编写一个存储过程来获取@quantity并检索它们的数量总和大于或等于@quantity的最小行 例如: id name quantity 1 a 3 2 b 1 3 c 7 4 d 2 对于@quantity=5,良好的结果可以是: 1, a, 3 2, b, 1 3, c, 7 或 或 等等。结果中的数量总和必须大于或等于@quantity,但我不需要额外
id name quantity
1 a 3
2 b 1
3 c 7
4 d 2
对于@quantity=5,良好的结果可以是:
1, a, 3
2, b, 1
3, c, 7
或
或
等等。结果中的数量总和必须大于或等于@quantity,但我不需要额外的行。编辑:添加解释
declare @goal int
set @goal = 5
;with numbered as
(
-- this first query gives a sequential number to all records
-- the biggest quantities are number 1,2,3 etc to the smallest quantities
-- this ordering is because the minimum number of rows is most quickly
-- achieved by combining the biggest quantities
select id, name, quantity,
row_number() over (order by quantity desc) rn
from data
), cte as
(
-- this is a common table expression
-- it starts off with the record with the biggest quantity
select id, name, quantity, rn, sumsofar = quantity
from numbered
where rn = 1
union all
-- this is the recursive part
-- "CTE" references the previous record in the loop
select b.*, sumsofar + b.quantity
from numbered b
inner join cte a on b.rn = a.rn+1 -- add the next record (as ordered)
-- continue the loop only if the goal is not yet reached
where a.sumsofar < @goal
)
-- this finally selects all the records that have been collected to satisfy the @goal
select * from cte
-- NOTE: if @goal cannot be reached even with all records added up, this just
-- returns all records
这仅适用于SQL Server 2005+,因为它使用公共表表达式
这应该比下面的效果更好,但下面的也适用于2000年,或者你可以在2005/8年比较这两种方法,找出最适合你的方法
declare @goal int set @goal = 5
select c.*
from data c inner join
(
select top 1 * from
(
-- find the pivotal a.id on which the goal is reached
select a.id, a.quantity, sum(b.quantity) s
from data a
-- this works on the b records on quantity desc, and id asc
inner join data b
on b.quantity < a.quantity
or (b.quantity=a.quantity and b.id > a.id)
group by a.id, a.quantity
having sum(b.quantity) >= @goal
union all
-- if ALL records together cannot satisfy the GOAL
-- we add a id=-1 record to mark this fact
select -1, null, null
) oneonly
-- find the min required, or return the fake -1 ID
ORDER BY case when id=-1 then 1 else 0 end, s ASC, id ASC
) d
on c.quantity > d.quantity
or (c.quantity=d.quantity and c.id <= d.id)
or d.id = -1 -- this matches all rows in C
请原谅我,但在中间例子中,数量之和等于5——那么这怎么可能呢?目前,问题的主体和标题似乎也不一致。如果您有一个特定的问题,请askI修改此代码以用于我的项目。但我有点问题。我需要对“金额”求和,哪种类型是小数点18,2。但是当执行时,我遇到了这样的错误:“递归查询cte的SumSoFar列中的锚和递归部分之间的类型不匹配。”对不起,我的英语水平很高
declare @goal int
set @goal = 5
;with numbered as
(
-- this first query gives a sequential number to all records
-- the biggest quantities are number 1,2,3 etc to the smallest quantities
-- this ordering is because the minimum number of rows is most quickly
-- achieved by combining the biggest quantities
select id, name, quantity,
row_number() over (order by quantity desc) rn
from data
), cte as
(
-- this is a common table expression
-- it starts off with the record with the biggest quantity
select id, name, quantity, rn, sumsofar = quantity
from numbered
where rn = 1
union all
-- this is the recursive part
-- "CTE" references the previous record in the loop
select b.*, sumsofar + b.quantity
from numbered b
inner join cte a on b.rn = a.rn+1 -- add the next record (as ordered)
-- continue the loop only if the goal is not yet reached
where a.sumsofar < @goal
)
-- this finally selects all the records that have been collected to satisfy the @goal
select * from cte
-- NOTE: if @goal cannot be reached even with all records added up, this just
-- returns all records
declare @goal int set @goal = 5
select c.*
from data c inner join
(
select top 1 * from
(
-- find the pivotal a.id on which the goal is reached
select a.id, a.quantity, sum(b.quantity) s
from data a
-- this works on the b records on quantity desc, and id asc
inner join data b
on b.quantity < a.quantity
or (b.quantity=a.quantity and b.id > a.id)
group by a.id, a.quantity
having sum(b.quantity) >= @goal
union all
-- if ALL records together cannot satisfy the GOAL
-- we add a id=-1 record to mark this fact
select -1, null, null
) oneonly
-- find the min required, or return the fake -1 ID
ORDER BY case when id=-1 then 1 else 0 end, s ASC, id ASC
) d
on c.quantity > d.quantity
or (c.quantity=d.quantity and c.id <= d.id)
or d.id = -1 -- this matches all rows in C