二、示例 复制代码 代码如下: -------- Create table stock (Id int not null primary key, articleno varchar(20) not null, rcvdate datetime not null, qty int not null, unitprice money not null ) go ---- insert stock select 1,"10561122","2011-1-1",15,10 union select 2,"10561122","2011-2-2",25,12 union select 3,"10561122","2011-3-3",35,15 union select 4,"10561122","2011-4-4",45,20 union select 5,"10561122","2011-5-5",55,10 union select 6,"10561122","2011-6-6",65,30 union select 7,"10561122","2011-7-7",75,17 union select 8,"10561122","2011-8-8",110,8
---成本视图 create view costLIFO as select unitprice from stock where rcvdate= (select MAX(rcvdate) from stock) go create view costFIFO as select sum(unitprice*qty)/SUM(qty) as unitprice from stock
go -----找出满足订单的、足够存货的最近日期。如果运气好的话,某一天的库存数量正好与订单要求的数字完全一样 -----就可以将总成本作为答案返回。如果订单止的数量比库存的多,什么也不返回。如果某一天的库存数量比订单数量多 ---则看一下当前的单价,乘以多出来的数量,并减去它。 ---下面这些查询和视图只是告诉我们库存商品的库存价值,注意,这些查询与视图并没有实际从库存中向外发货。 create view LIFO as select s1.rcvdate,s1.unitprice,sum(s2.qty) as qty,sum(s2.qty*s2.unitprice) as totalcost from stock s1 ,stock s2 where s2.rcvdate>=s1.rcvdate group by s1.rcvdate,s1.unitprice
go select (totalcost-((qty-300)*unitprice )) as cost from lifo as l where rcvdate=(select max(rcvdate) from lifo as l2 where qty>=300) go
create view FIFO as select s1.rcvdate,s1.unitprice,sum(s2.qty) as qty,sum(s2.qty*s2.unitprice) as totalcost from stock s1 ,stock s2 where s2.rcvdate<=s1.rcvdate group by s1.rcvdate,s1.unitprice
go select (totalcost-((qty-300)*unitprice )) as cost from fifo as l where rcvdate=(select min(rcvdate) from lifo as l2 where qty>=300) -------- go ----- -----在发货之后,实时更新库存表 create view CurrStock as select s1.rcvdate,SUM(case when s2.rcvdate>s1.rcvdate then s2.qty else 0 end) as PrvQty ,SUM(case when s2.rcvdate<=s1.rcvdate then s2.qty else 0 end) as CurrQty from stock s1 ,stock s2 where s2.rcvdate<=s1.rcvdate group by s1.rcvdate,s1.unitprice go create proc RemoveQty @orderqty int as if(@orderqty>0) begin update stock set qty =case when @orderqty>=(select currqty from CurrStock as c where c.rcvdate=stock.rcvdate) then 0 when @orderqty<(select prvqty from CurrStock c2 where c2.rcvdate=stock.rcvdate) then stock.qty else (select currqty from CurrStock as c3 where c3.rcvdate=stock.rcvdate) -@orderqty end end -- delete from stock where qty=0 --- go exec RemoveQty 20 go ---------------
三、使用“贪婪算法”进行订单配货
复制代码 代码如下: -------还有一个问题,如何使用空间最小或最大的仓库中的货物来满足订单,假设仓库不是顺序排列,你可以按钮希望的顺序任意选择满足订单。 ---使用最小的仓库可以为订单的装卸工人带来最小的工作量,使用最大的仓库,则可以在仓库中清理出更多的空间 -------例如:对于这组数据,你可以使用(1,2,3,4,5,6,7)号仓库也可以使用(5,6,7,8)号仓库中的货物来满足订单的需求。 ----这个就是装箱问题,它属于NP完全系统问题。对于一般情况来说,这种问题很难解决,因为要尝试所有的组合情况,而且如果数据量大的话, ----计算机也很难很快处理。 ---所以有了“贪婪算法”,这个算法算出来的常常是近乎最优的。这个算法的核心就是“咬最大的一口”直到达到或超越目标。 --- --1. 第一个技巧,要在表中插入一些空的哑仓库,如果你最多需要n次挑选,则增加n-1个哑仓库 insert stock select -1,"10561122","1900-1-1",0,0 union select -2,"10561122","1900-1-1",0,0 --select -3,"1900-1-1",0,0 ---- go create view pickcombos as select distinct (w1.qty+w2.qty+w3.qty) as totalpick ,case when w1.id<0 then 0 else w1.id end as bin1 ,w1.qty as qty1, case when w2.id<0 then 0 else w2.id end as bin2,w2.qty as qty2 ,case when w3.id<0 then 0 else w3.id end as bin3 ,w3.qty as qty3 from stock w1,stock w2, stock w3 where w1.id not in (w2.id,w3.id) and w2.id not in (w1.id,w3.id) and w1.qty>=w2.qty and w2.qty>=w3.qty ---- ---1.使用存储过程来找出满足或接近某一数量的挑选组合 -------- go create proc OverPick @pickqty int as if(@pickqty>0) begin select @pickqty,totalpick,bin1,qty1,bin2,qty2,bin3,qty3 from pickcombos where totalpick=(select MIN(totalpick) from pickcombos where totalpick>=@pickqty) end go exec OverPick 180
---------- select * from stock drop table stock drop view lifo drop view fifo drop view costfifo drop view costlifo drop view CurrStock drop proc OverPick drop proc RemoveQty drop view pickcombos