Left Join INT on Varchar - sql-server

Question

Im writing a report to show what features our clients want when building there home.
I want to left join 2 tables however the way the data is stored its making it difficult for me to do the join.
Table 1 tbl_Main_Holding has a field called Requirements and the data is stored as a varchar and can have multiple values like 1,4,7
1 = "Eco-Build"
4 = "Conservatory"
7 = "Basement"
Table 2 [tbl_Features] has the fields ID (INT) and Description (Varchar)
SELECT * FROM dbo.tbl_Main_Holding AS rm
LEFT JOIN [dbo].[tbl_Features] AS f
ON rm.Requirements = f.id
The join below wont work as i would need to convert the varchar to INT
However that's not my problem my problem is how do i show the results of clients that have selected multiple feature, how dopes this left join work?
Im using SQL Server 2008 and the data for both tables are store as so.

Answer 1

Step 1 is to go and find the person that designed this table structure (even if it is you) then whack them round the head with a stick.
Step 2 is to redesign the tables, a junction table is what is required here, not stuffing multiple integers into a single varchar column. For good measure at the end of step two you should hit the original designer with a stick again.
CREATE TABLE tbl_Main_Holding_Requirements
(
MainHoldingID INT NOT NULL, --FK TO `tbl_main_Holding`
FeatureID INT NOT NULL -- FK TO Require `tbl_Features`
);
Now, each requirement represents a row in this table, rather than a new item on your list, so your join is now simple:
SELECT *
FROM dbo.tbl_Main_Holding AS rm
LEFT JOIN dbo.tbl_Main_Holding_Requirements AS r
ON r.MainHoldingID = rm.ID
LEFT JOIN [dbo].[tbl_Features] AS f
ON f.ID = r.FeatureID;
If you need to bring this back up to a comma delimited list, then you can do it in the presentation layer, or with SQL-Server's XML Extensions:
SELECT *,
Features = STUFF(f.Features.value('.', 'NVARCHAR(MAX)'), 1, 1, '')
FROM dbo.tbl_Main_Holding AS rm
OUTER APPLY
( SELECT CONCAT(',', f.Description)
FROM dbo.tbl_Main_Holding_Requirements AS r
INNER JOIN [dbo].[tbl_Features] AS f
ON f.ID = r.FeatureID
WHERE r.MainHoldingID = rm.ID
FOR XML PATH(''), TYPE
) f (Features);
If step two is not possible, then you can get around this using LIKE:
SELECT *
FROM dbo.tbl_Main_Holding AS rm
LEFT JOIN [dbo].[tbl_Features] AS f
ON ',' + rm.Requirements + ',' LIKE '%,' + CONVERT(VARCHAR(10), f.ID) + ',%';
Once again, if the features need to be reduced back to a single row, then you can use XML extensions again:
SELECT *,
Features = STUFF(f.Features.value('.', 'NVARCHAR(MAX)'), 1, 1, '')
FROM dbo.tbl_Main_Holding AS rm
OUTER APPLY
( SELECT CONCAT(',', f.Description)
FROM [dbo].[tbl_Features] AS f
WHERE ',' + rm.Requirements + ',' LIKE '%,' + CONVERT(VARCHAR(10), f.ID) + ',%'
FOR XML PATH(''), TYPE
) f (Features);
Another option is to split the comma separated values into a list using some kind of Split function, but as the testing in this article shows, if you don't need to access the individual values from the list, it is more efficient to just use LIKE.

Answer 2

As I wrote in my comment, please read Is storing a delimited list in a database column really that bad?
You really should normalize your database to avoid these things.
Now, assuming you can't change the database schema, there is a simple trick with like that you can use:
SELECT * FROM dbo.tbl_Main_Holding AS rm
LEFT JOIN [dbo].[tbl_Features] AS f
ON ',' + rm.Requirements +',' LIKE '%,' + CAST(f.id as varchar(10)) + ',%'
Note that I've added a comma before and after the rm.Requirements column and also before and after the f.id column.