FIRST aggregate function which I can use with HAVING clause

问题

I have a weird requirement which I need to use inside my Stored Procedure in SQL Server 2008 R2.

I need a FIRST aggregate function which returns the first element of a sequence and I will use that with HAVING clause.

Let me give you an example:

DECLARE @fooTable AS TABLE(
    ID INT,
    CategoryName NVARCHAR(100),
    Name NVARCHAR(100),
    MinAllow INT,
    Price DECIMAL(18,2)
);

INSERT INTO @fooTable  VALUES(1, 'Cat1', 'Product1', 2, 112.2);
INSERT INTO @fooTable  VALUES(2, 'Cat2', 'Product2', 4, 12.34);
INSERT INTO @fooTable  VALUES(3, 'Cat1', 'Product3', 5, 233.32);
INSERT INTO @fooTable  VALUES(4, 'Cat3', 'Product4', 4, 12.43);
INSERT INTO @fooTable  VALUES(5, 'Cat3', 'Product5', 1, 13.00);

DECLARE @minAllowParam AS INT = 3;

SELECT ft.CategoryName, SUM(ft.Price) FROM @fooTable ft
GROUP BY ft.CategoryName;

As you see, we have a table and some dummy values. Inside the SELECT query, we group the categories together and sum the price of the products up.

This query returns the following result:

CategoryName     TotalPrice
---------------- ----------------
Cat1              345.52
Cat2              12.34
Cat3              25.43

What I need here is something like this:

SELECT ft.CategoryName, SUM(ft.Price) FROM @fooTable ft
GROUP BY ft.CategoryName
HAVING GetFIRST(MinAllow) >= @minAllowParam;

In that our case with a query something like this, we should be able to select following results:

INSERT INTO @fooTable  VALUES(2, 'Cat2', 'Product2', 4, 12.34);
INSERT INTO @fooTable  VALUES(4, 'Cat3', 'Product4', 4, 12.43);
INSERT INTO @fooTable  VALUES(5, 'Cat3', 'Product5', 1, 13.00);

As the INSERT INTO @fooTable VALUES(1, 'Cat1', 'Product1', 2, 112.2); record is the first element of a sequence and has the value of 2 for MinAllow column, Cat1 should be out of scope here. On the other hand, INSERT INTO @fooTable VALUES(5, 'Cat3', 'Product5', 1, 13.00); record has the value of 1 for MinAllow column but is the second element of the sequence. So, Cat3 is safe and can be selected.

Note: MIN or MAX is not what I am looking for!

I know that this example logically does not make sense but I have a situation which it totally does and hard to explain here on the other hand.

Any thoughts?

Edit:

I am assuming that I can achieve what I want here by creating a CLR User-Defined Aggregate Function but I do not want to do this if there is any other choice

回答1:

How about

SELECT f1.CategoryName, SUM(f1.Price) 
FROM @fooTable AS f1
INNER JOIN (
    SELECT MinAllow, CategoryName
    FROM (
         SELECT MinAllow, CategoryName, ROW_NUMBER() OVER (PARTITION BY CategoryName ORDER BY ID) AS m
         FROM @fooTable
    ) AS f
    WHERE m = 1
) AS f2 ON f1.CategoryName = f2.CategoryName
WHERE f2.MinAllow >= @minAllowParam
GROUP BY f1.CategoryName

I know not a very elegant query. Maybe I can tweak it a little if I work on it a little longer!

Edit: Ok the inner most subquery should be unnecessary. This should also work:

SELECT f1.CategoryName, SUM(f1.Price) 
FROM @fooTable AS f1
INNER JOIN (
    SELECT MinAllow, CategoryName, ROW_NUMBER() OVER (PARTITION BY CategoryName ORDER BY ID) AS m
    FROM @fooTable
) AS f2 ON f1.CategoryName = f2.CategoryName
WHERE f2.m = 1 AND f2.MinAllow >= @minAllowParam
GROUP BY f1.CategoryName

回答2:

UPDATE: a readable query:

SELECT ft.CategoryName, SUM(ft.Price) 
FROM fooTable ft
    cross apply
    (
       select top 1 MinAllow
         from fooTable a
        where a.CategoryName = ft.CategoryName
        order by ID
    ) a
where a.MinAllow >= @minAllowParam
GROUP BY ft.CategoryName;

You might filter categories having first (by id?) MinAllow >= @minAllowParam:

...
inner join 
(
   select 
   -- Add columns you might need
     CategoryName,
     Price
   from
     fooTable
   inner join
   (
     -- First ID in category
     select
       min(id) id
     from
       fooTable
     group by
       CategoryName
   ) firstID
   -- Back to all columns
     ON fooTable.ID = firstID.ID
   -- but only if category sequence starts properly
    AND fooTable.MinAllow >= @minAllowParam
) a
-- Allow MinAllow categories only
  ON fooTable.CategoryName = a.CategoryName

回答3:

Have a look at ROW_NUMBER() with partitioning:

http://msdn.microsoft.com/en-us/library/ms186734.aspx

回答4:

There are two ways that I know to achieve FIRST aggregate function:

ROW_NUMBER()

(WHERE ROW_NUMBER_COLUMN = 1)

http://msdn.microsoft.com/en-us/library/ms186734.aspx

AND

SELECT ...., (SELECT TOP 1 FROM ... WHERE (outer table join) ORDER BY SOMETHING) AS [FIRST]
FROM ...

来源：https://stackoverflow.com/questions/10120145/first-aggregate-function-which-i-can-use-with-having-clause