使用列名指定预定义表

Question

I'm trying to use set up a database for a school project, and I'm using triggers to set up referential integrity for one table. I have a table, Addresses, which stores the address for People, Studios, and Directors. Then I have a table called Address Reference. This table points to the Address table, and it has a two fields, the ReferenceID and the TableName to show which table and row this address is for. I have a Constraint so TableName will always be valid.

I'm trying to set up a trigger to make sure any rows inserted are valid, which I can do, I'm just trying to improve it. My code would look like this:

SELECT * 
    FROM inserted 
        WHERE ReferenceID IN 
            (SELECT PersonID 
                FROM inserted.TableName)

However I found I needed to use dynamic sql. So I was thinking something like this:

SELECT * 
    FROM inserted 
        WHERE ReferenceID IN 
            (EXEC('SELECT PersonID FROM' + inserted.TableName))

Which didn't work, even when I removed the exec.

I'm doing this in SQL Server Management Studio With SQL Server 11.0.3128

Let me know if you need any more information. I've looked around, and I haven't found any answers to this question that work.

Answer 1

This is a poor way to maintain referential integrity. There are a number of ways you could approach this.

The first would be to have an address table, then multiple tables to contain the links, eg

CREATE TABLE StudioAddress
(       StudioID INT NOT NULL,
        AddressID INT NOT NULL,
    CONSTRAINT PK_StudioAddress__StudioID_AddressID PRIMARY KEY (StudioID, AddressID),
    CONSTRAINT FK_StudioAddress__StudioID FOREIGN KEY (StudioID) REFERENCES Studio (StudioID),
    CONSTRAINT FK_StudioAddress__AddressID FOREIGN KEY (AddressID) REFERENCES Address (AddressID)
);

This maintains your referenctial integrity without needing triggers, and still caters for a 1 to many relationship.

Another option would be to have 3 nullable columns in your address table (StudioID, PersonID, DirectorID), each with a foreign key to the relevant table, you can the add a check constraint to ensure only one of the 3 fields is populated (if this is required).

I much prefer the first option though, it is much cleaner, and also allows for the same address to be used for multiple things.

---

ADENDUM

If this has to be done using triggers, then I think you would need to use something like this:

IF EXISTS(  SELECT  1
            FROM    inserted i
            WHERE   NOT EXISTS
                    (   SELECT  1
                        FROM    People p
                        WHERE   p.PersonID = i.ReferenceID
                        AND     i.TableName = 'People'
                        UNION ALL
                        SELECT  1
                        FROM    Studios s
                        WHERE   s.StudioID = i.ReferenceID
                        AND     i.TableName = 'Studios'
                        UNION ALL
                        SELECT  1
                        FROM    Directors d
                        WHERE   d.DirectorID = i.ReferenceID
                        AND     i.TableName = 'Directors'
                    )
        )
        BEGIN
            ROLLBACK TRANSACTION;
            RAISERROR('Referential integrity error', 16, 1);
        END

This essentially checks that for all inserted/updated rows a record exists with the relevant ID in the relevant table.

I still stand by my earlier answer though, that this is a terrible approach, and I would question any syllabus this is on!