I want to remove the need for a nested query if I can from my query below, but I'm struggling to work out how.

This is the schema:

SET ANSI_NULLS ON
GO
SET QUOTED_IDENTIFIER ON
GO

CREATE TABLE [dbo].[expiration]
(
    [batch_number] [int] NOT NULL,
    [fruit_number] [int] NOT NULL,
    [store_number] [int] NOT NULL,
    [expiration_date] [date] NULL
) ON [PRIMARY]

CREATE TABLE [dbo].[fruits]
(
    [fruit_number] [int] NOT NULL,
    [fruit_name] [nvarchar](50) NOT NULL
) ON [PRIMARY]

This is the data:

INSERT INTO [dbo].[expiration] ([batch_number], [fruit_number], [store_number], [expiration_date]) 
VALUES (1, 3, 4, CAST(N'2021-11-25' AS Date))

INSERT INTO [dbo].[expiration] ([batch_number], [fruit_number], [store_number], [expiration_date]) 
VALUES (1, 2, 2, CAST(N'2021-11-22' AS Date))

INSERT INTO [dbo].[expiration] ([batch_number], [fruit_number], [store_number], [expiration_date]) 
VALUES (1, 5, 3, CAST(N'2021-11-30' AS Date))

INSERT INTO [dbo].[expiration] ([batch_number], [fruit_number], [store_number], [expiration_date]) 
VALUES (2, 2, 7, NULL)

INSERT INTO [dbo].[expiration] ([batch_number], [fruit_number], [store_number], [expiration_date]) 
VALUES (2, 3, 2, CAST(N'2021-12-12' AS Date))

INSERT INTO [dbo].[expiration] ([batch_number], [fruit_number], [store_number], [expiration_date]) 
VALUES (1, 1, 5, NULL)

INSERT INTO [dbo].[expiration] ([batch_number], [fruit_number], [store_number], [expiration_date]) 
VALUES (2, 1, 6, CAST(N'2021-11-28' AS Date))

INSERT INTO [dbo].[fruits] ([fruit_number], [fruit_name]) 
VALUES (1, N'banana')

INSERT INTO [dbo].[fruits] ([fruit_number], [fruit_name]) 
VALUES (2, N'apple')

INSERT INTO [dbo].[fruits] ([fruit_number], [fruit_name]) 
VALUES (3, N'pear')

INSERT INTO [dbo].[fruits] ([fruit_number], [fruit_name]) 
VALUES (4, N'peach')

INSERT INTO [dbo].[fruits] ([fruit_number], [fruit_name]) 
VALUES (5, N'strawberry')

And this is my query:

SELECT
    fruit_number, 
    MAX(expirationDate) as expirationDate
FROM
    (SELECT
        f.fruit_number,
        CASE
            WHEN e.expiration_date is NULL AND e.fruit_number IS NOT NULL THEN 1
            ELSE 0
        END AS expirationDate
    FROM
        expiration AS e
    FULL OUTER JOIN 
        fruits AS f ON f.fruit_number = e.fruit_number
    WHERE
        f.fruit_number IS NOT NULL) t
GROUP BY
    fruit_number
ORDER BY
    fruit_number

It produces this result set:

The resultset is what I'm after, but it's ugly having the nested query. Is it possible to do this without the nested query? An online query analyser (https://www.eversql.com/sql-query-optimizer/) said to move the sub-query into a temp table and query against that, but isn't just doing the same thing in more steps?

CodePudding user response：

Give this a try, I believe it gets you what you wanted:

SELECT t1.fruit_number
, CASE WHEN MIN(ISNULL(expiration_date, '1/1/1900')) = CAST('1/1/1900' as date) and t2.fruit_number IS NOT NULL THEN 1 ELSE 0 END expirationDate
FROM fruits t1 
LEFT JOIN expiration t2 on t1.fruit_number = t2.fruit_number
GROUP BY t1.fruit_number, t2.fruit_number
ORDER BY t1.fruit_number

CodePudding user response：

The first change I would make is with your joins. It makes no sense to use a FULL OUTER JOIN then put in a where clause that says f.fruit_number IS NOT NULL. This means that every row must have a record in fruits, so your query would make more sense as SELECT .. FROM fruits AS f LEFT JOIN expiration AS e ON e.fruit_number = f.fruit_number.

You can also remove the subquery by placing your case expression directly within the MAX function:

SELECT  f.fruit_number, 
        f.fruit_name,
        expirationDate = MAX(CASE WHEN e.expiration_date IS NULL 
                                    AND e.fruit_number IS NOT NULL THEN 1 ELSE 0 END)
FROM    dbo.fruits AS f
        LEFT JOIN dbo.expiration AS e
            ON e.fruit_number = f.fruit_number
GROUP BY f.fruit_number, f.fruit_name
ORDER BY f.fruit_number;

Example on db<>fiddle