I have开发者_Python百科 two columns
OrderId OrderStatus
120 1
120 2
121 1
123 1
123 2
I want to retrieve only the row with OrderId 121 because it has had an OrderStatus of 1 but it does not have an OrderStatus of 2.
Because you don't mention there being a 'StatusDate
' or anything like that, I am going to take
I want to only bring back the row with OrderId 121 because it has had an OrderStatus of 1 BUT IT HAS NEVER HAD AN OrderStatus of 2.
to mean
I want to only bring back OrderId 121, because there is a row with OrderId 121 and OrderStatus 1, but there is no row with OrderId 121 and OrderStatus 2
With SQL 2005 and later, EXCEPT
makes this extremely concise:
SELECT OrderId FROM [Order] WHERE OrderStatus = 1
EXCEPT
SELECT OrderId FROM [Order] WHERE OrderStatus = 2
EXCEPT
returns distinct values so there is no need for any further DISTINCT
.
You can use a self-join, searching for OrderStatus = 1 from the left and missed join with OrderStatus = 2 from the right:
declare @t table(OrderID int, OrderStatus int)
insert into @t values (120, 1)
insert into @t values (120, 2)
insert into @t values (121, 1)
insert into @t values (123, 1)
insert into @t values (123, 2)
select t1.*
from @t
t1 left join
(select * from @t where OrderStatus = 2) as t2 on t2.OrderID = t1.OrderID
where
t1.OrderStatus = 1 and
t2.OrderID is null
If your order statuses are only 1 and 2, and the orders must be a status of 1 at some point before becoming a status of 2, you can search for the orders with a maximum order status value of 1:
select distinct orderid from orders
group by orderid
having max(orderstatus) = 1;
Demo: http://www.sqlize.com/2k3C2SqMH2
Or if it's not quite that simple, we can be more explicit about not allowing an orderstatus of 2 to ever have happened by using a not exists
clause:
select distinct orderid
from orders o
where not exists (
select * from orders o2
where o2.orderid = o.orderid
and o2.orderstatus = 2
);
Demo: http://www.sqlize.com/o6fSvWmvaj
精彩评论