Use a column name to specify a predefined table - sql

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, e.g.
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!