Question

I have a staff database table containing staff members, with user_no and user_name columns. I have another, department, table containing the departments which staff can be members of, with dept_no and dept_name as columns.

Because staff can be members of multiple departments, I have a third, staff_dept, table with a user_no column and a dept_no column, which are the primary keys of those other two tables. This table shows which departments each member of staff belongs to and contains one row for each user/department intersection.

I would like to have an output in the form of a spreadsheet (CSV file, whatever; I’ll be fine mangling the results into a usable form after I’ve got them) with one column for each department, and one row for each user, with an X appearing at each intersection, as defined in staff_dept.

Can I write a single SQL query which will achieve this result? or will I have to do some “real” programming (because it’s not a “real” program until you’ve nested three or four for loops, obviously) to collect and format this data?

Answer 1

This can be done with a PIVOT table (using SQL Server):

SELECT user_name, [dept1name], [dept2name], [dept3name], ...
FROM
    (SELECT s.user_name, d.dept_name,
     case when sd.user_no is not null then 'X' else '' end as matches
     from staff s
     cross join department d
     left join staff_dept sd on s.user_no = sd.user_no and d.dept_no = sd.dept_no
    ) AS s
PIVOT
(
    min(matches)
    FOR dept_name IN ([dept1name], [dept2name], [dept3name], ...)
) AS pvt
order by user_name

Demo: http://www.sqlfiddle.com/#!3/c136d/5

Edit: To generate the PIVOT query dynamically from the list of departments in the table, you would make use of dynamic SQL, i.e., generate the code into a variable and use sp_executesql helper stored procedure. Here’s an example: http://www.sqlfiddle.com/#!3/c136d/14