Search comma separated string in Column t-sql - sql-server-2008

Question

In a property mgt system, I'm saving buyers based on their preferences. Say a person interested in houses which has more than 2 & less than 4. So I saved it as 2,3,4. Please see the attachment.
When searching, say someone searching the buyers who are interested in houses which has more than 2, how should i write the select statement to check the bedroom column.
If someone search buyers who are interested in houses which has more than 2 bathrooms; what could be the select statement?

Answer 1

I still think a min/max is the much better table structure, but if you can't change it, try this.
First, let's come up with some base rules. If any of these is violated, then the final answer will need modification (and will probably be more complicated).
If a string contains + anywhere in it, the maximum is infinity.
The + can only occur at the end of the string.
The list will always be comma separated integers.
The smallest number will always be first in the list.
The largest number will always be last in the list.
If all those are true, then after a LOT of work, I think I have something you can use. The basic idea was to come up with a pair of functions that will get the min/max values out of your strings. Once you have these functions, you can use them in WHERE clauses.
See this SQL Fiddle for starters. Function definitions on the left, a sample query to give you the gist of how they work on the right.
CREATE FUNCTION dbo.list_min(#list_str AS VARCHAR(MAX))
RETURNS INT
WITH RETURNS NULL ON NULL INPUT
AS BEGIN
DECLARE #comma_index INT;
SET #comma_index = CHARINDEX(',', #list_str);
DECLARE #result INT;
IF (0 < #comma_index)
SET #result = CONVERT(INT, LEFT(#list_str, #comma_index - 1));
ELSE
SET #result = CONVERT(INT, REPLACE(#list_str, '+', ''));
RETURN #result;
END;
and
CREATE FUNCTION dbo.list_max(#list_str AS VARCHAR(MAX))
RETURNS INT
WITH RETURNS NULL ON NULL INPUT
AS BEGIN
IF (#list_str LIKE '%+')
RETURN 2147483647; -- Max INT
DECLARE #comma_index INT;
SET #comma_index = CHARINDEX(',', REVERSE(#list_str));
DECLARE #result INT;
IF (0 < #comma_index)
SET #result = CONVERT(INT, RIGHT(#list_str, #comma_index - 1));
ELSE
SET #result = CONVERT(INT, #list_str);
RETURN #result;
END;
(If anyone can think of a way to get rid of that ridiculous result variable, please let me know. I was getting errors about the last statement "must be a return statement" when putting the return inside IF/ELSE, and I couldn't get a CASE syntax working.)
With these in hand, you can do queries like this:
SELECT *
FROM stuff
WHERE dbo.list_min(carspace) <= 2 AND 2 <= dbo.list_max(carspace)
which will only select your second row. (SQL Fiddle of this query.)
A third function you might find useful is one that gives you the max in the list but ignores the +. To do that, it's essentially the list_max function, but without the IF block that checks for +. The get that functionality, you might want to just remove the + check from list_max, and create another function that checks for + and calls list_max if there's no +.
I'm not sure about the performance characteristics here. I imagine they aren't great. You might want to consider some function based indexing if you have a large amount of data to search through.
Good luck. Hope this helps.

Answer 2

Wouldn't it make more sense to store a min and max and then query with <= and >=?

Answer 3

Your table design is not optimal. You can simply store the number of bedrooms as an integer. Instead of "1,2,3,4", you would be looking at just 4 in that case.
To answer the particular question though, you can do the replace trick to count the number of commas in the column as such:
SELECT * FROM myTable WHERE LEN(col) - LEN(REPLACE(col, ',','')) >= someNumber