SQL sub select if exists - sql

Question

I am using SQL Server 2012. I have two tables to hold orders for products. Order which has a received date and OrderItem which has a price and order id fk.
I am trying to write a query get all orders within a date range, group them by the date, and then sum all the order items price to get the total of all orders for that date.
I have this working. Now I want to add another column to select the difference between the total price for that day, and 7 days ago. If there are no orders 7 days ago then the column should be null.
So at the moment I have the below query:
select cast(o.ReceivedDate as date) as OrderDate,
coalesce(count(orderItems.orderId), 0) as Orders,
coalesce(sum(orderItems.Price), 0) as Price
from [Order] o
left outer join (
select o.Id as orderId, sum(ot.Price) as Price
from OrderItem ot
join [Order] o on ot.OrderId = o.Id
where o.ReceivedDate >= #DateFrom and o.ReceivedDate <= #DateTo
group by o.Id
) as orderItems on o.Id = orderItems.orderId
where o.ReceivedDate >= #DateFrom and o.ReceivedDate <= #DateTo
group by cast(o.ReceivedDate as date)
order by cast(o.ReceivedDate as date) desc
So how can I add my other column to this query? I need to do something like:
//pseudo
if o.RecievedDate - 7 exists then orderItems.Price - Price from 7 days ago else null
But I am not sure how to do this? I have created a sqlfiddle to help explain http://sqlfiddle.com/#!6/8b837/1
So from my sample data what I want to achieve is results like this:
| ORDERDATE | ORDERS | PRICE | DIFF7DAYS |
---------------------------------------------
| 2013-01-25 | 3 | 38 | 28 |
| 2013-01-24 | 1 | 12 | null |
| 2013-01-23 | 1 | 10 | null |
| 2013-01-22 | 1 | 33 | null |
| 2013-01-18 | 1 | 10 | null |
| 2013-01-10 | 1 | 3 | -43 |
| 2013-01-08 | 2 | 11 | null |
| 2013-01-04 | 1 | 1 | null |
| 2013-01-03 | 3 | 46 | null |
As you can see, the 25th has a order 7 days ago so the difference is shown. The 24th doesn't so null is displayed.
Any help would be much appreciated.

Answer 1

Not sure why you are using a left outer join between [Orders] table and the subquery as there cannot be orders without order items (in general):
To get your results you could do it in a simplified version as below using a CTE
SQL-FIDDLE-DEMO
;with cte as (
select convert(date,o.ReceivedDate) orderDate,
count(distinct o.Id) as Orders,
coalesce(sum(ot.Price),0) as Price
from OrderItem ot
join [Order] o on ot.OrderId = o.Id
where o.ReceivedDate >= #DateFrom and o.ReceivedDate <= #DateTo
group by convert(date,o.ReceivedDate)
)
select c1.orderDate, c1.Orders, c1.Price, c1.Price-c2.Price DIFF7DAYS
from cte c1 left join cte c2 on dateadd(day,-7,c1.orderdate) = c2.orderdate
order by c1.orderdate desc
| ORDERDATE | ORDERS | PRICE | DIFF7DAYS |
-------------------------------------------
| 2013-01-25 | 3 | 38 | 28 |
| 2013-01-24 | 1 | 12 | (null) |
| 2013-01-23 | 1 | 10 | (null) |
| 2013-01-22 | 1 | 33 | (null) |
| 2013-01-18 | 1 | 10 | (null) |
| 2013-01-10 | 1 | 3 | -43 |
| 2013-01-08 | 2 | 11 | (null) |
| 2013-01-04 | 1 | 1 | (null) |
| 2013-01-03 | 3 | 46 | (null) |

Answer 2

Use a temp table and join it on the datediff.
DECLARE #DateFrom datetime
SET #DateFrom = '2012-12-02'
DECLARE #DateTo datetime
SET #DateTo = '2013-03-13'
CREATE TABLE #temp ( orderdate date, orders int, price money)
INSERT INTO #temp
SELECT cast(o.ReceivedDate AS date) AS OrderDate,
coalesce(count(orderItems.orderId), 0) AS Orders,
coalesce(sum(orderItems.Price), 0) AS Price
FROM [Order] o
LEFT OUTER JOIN (
SELECT o.Id AS orderId, sum(ot.Price) AS Price
FROM OrderItem ot
JOIN [Order] o ON ot.OrderId = o.Id
WHERE o.ReceivedDate >= #DateFrom AND o.ReceivedDate <= #DateTo
GROUP BY o.Id
) AS orderItems ON o.Id = orderItems.orderId
WHERE o.ReceivedDate >= #DateFrom AND o.ReceivedDate <= #DateTo
GROUP BY cast(o.ReceivedDate AS date)
SELECT t1.orderdate, t1.orders, t1.price,
t1.price - t2.price AS diff7days
FROM #temp t1 LEFT JOIN #temp t2
ON datediff(DAY, t2.orderdate, t1.orderdate) = 7
ORDER BY t1.orderdate DESC
http://sqlfiddle.com/#!6/8b837/34