Query with columns calculated at runtime - sql-server

Question

I need some help with a query in SQL Server 2012.
My customer has a (sort of) internal wiki where his employees save documents; these documents are stored in a table in the database (let's call it Documentation, with columns ID, Title, Text and Tags) and accessed via a web frontend.
Up to now every employee had access to all the documents, now my customer now wants his employees to access only to specific procedures. There are no specific criteria such as role or user type, he wants to decide who sees what on an individual basis (from now on John will see documents 1,2 and 4 while Janet will see only documents 3,4 and 5 and so on). Don't ask me why...
My task is to prepare a big table in the frontend where every line is a document and the employees are in the column; for every document there is a checkbox for every employee, to indicate whether that user can access the document, something similar to this:
The problem is that the number of documents is not fixed, nor the number of employees. I need to find a query to extract this data without knowing the number of columns, it must be done dynamically at runtime.
I have access to the User Table of course, so I know the number and the name of the employees. As for the authorizations I thought I could use a new table with just 3 columns: ID, ID_Document and ID_User.
I googled and searched but I could not find an appropriate answer. I tried using a Pivot Table but it does not seem right to me, I do not have to do any aggregation to the data.
Can anybody help me?

Answer 1

I'm going to we assume you have 5 tables in total. I'm going to call these Document, Tag, Employee and then DocumentTag and DocumentEmployee. To therefore get the solution you are after you need 2 different types of aggregation, string aggregation for the tags, and pivoting for the employees.
--Base tables
CREATE TABLE dbo.Document (DocumentID int, DocumentTitle nvarchar(50));
CREATE TABLE dbo.Employee (EmployeeID int, EmployeeName nvarchar(50));
CREATE TABLE dbo.Tag (TagID int, TagName nvarchar(50));
GO
--Link tables
CREATE TABLE dbo.DocumentTag (DocumentID int, TagID int);
CREATE TABLE dbo.DocumentEmployee (DocumentID int, EmployeeID int);
GO
--Sample data
INSERT INTO dbo.Document
VALUES(2,N'Important Doc'),(3,N'New Doc');
INSERT INTO dbo.Employee
VALUES(1,N'John'),
(2,N'Mary'),
(3,N'Patricia'),
(4,N'Paul');
INSERT INTO dbo.Tag
VALUES(1,N'Classified'),
(2,N'Finance'),
(3,N'Warehouse');
GO
--Link Data
INSERT INTO dbo.DocumentTag
VALUES(1,1),(1,2),(2,3);
INSERT INTO dbo.DocumentEmployee
VALUES(1,1),(1,2),(1,3),(2,2),(2,4);
If didn't need a dynamic pivot, then your SQL would look somethuing like this:
SELECT D.DocumentID,
D.DocumentTitle,
STUFF((SELECT N' ' + T.TagName
FROM dbo.DocumentTag DT
JOIN dbo.Tag T ON DT.TagID = T.TagID
WHERE DT.DocumentID = D.DocumentID
FOR XML PATH(N''),TYPE).value('.','nvarchar(MAX)'),1,1,N'') AS Tags,
MAX(CASE E.EmployeeName WHEN N'John' THEN 1 ELSE 0 END) AS John,
MAX(CASE E.EmployeeName WHEN N'Mary' THEN 1 ELSE 0 END) AS Mary,
MAX(CASE E.EmployeeName WHEN N'Patricia' THEN 1 ELSE 0 END) AS Patricia,
MAX(CASE E.EmployeeName WHEN N'Paul' THEN 1 ELSE 0 END) AS Paul
FROM dbo.Document D
JOIN dbo.DocumentEmployee DE ON D.DocumentID = DE.EmployeeID
JOIN dbo.Employee E ON DE.EmployeeID = E.EmployeeID
GROUP BY D.DocumentID,
D.DocumentTitle;
As you need the dataset to scale as you add employees, however, you need dynamic SQL to do this. So to achieve this using the above solution you can do something like this:
DECLARE #SQL nvarchar(MAX),
#CRLF nchar(2) = NCHAR(13) + NCHAR(10);
SET #SQL = N'SELECT D.DocumentID,' + #CRLF +
N' D.DocumentTitle,' + #CRLF +
N' STUFF((SELECT N'' '' + T.TagName' + #CRLF +
N' FROM dbo.DocumentTag DT' + #CRLF +
N' JOIN dbo.Tag T ON DT.TagID = T.TagID' + #CRLF +
N' WHERE DT.DocumentID = D.DocumentID' + #CRLF +
N' FOR XML PATH(N''''),TYPE).value(''.'',''nvarchar(MAX)''),1,1,N'''') AS Tags,' + #CRLF +
STUFF((SELECT N',' + #CRLF +
N' MAX(CASE E.EmployeeName WHEN N' + QUOTENAME(E.EmployeeName,'''') + N' THEN 1 ELSE 0 END) AS ' + QUOTENAME(E.EmployeeName)
FROM dbo.Employee E
ORDER BY E.EmployeeID ASC
FOR XML PATH(''),TYPE).value('.','nvarchar(MAX)'),1,3,N'') + #CRLF +
N'FROM dbo.Document D' + #CRLF +
N' JOIN dbo.DocumentEmployee DE ON D.DocumentID = DE.EmployeeID' + #CRLF +
N' JOIN dbo.Employee E ON DE.EmployeeID = E.EmployeeID' + #CRLF +
N'GROUP BY D.DocumentID,' + #CRLF +
N' D.DocumentTitle;';
--PRINT #SQL; --YOur debugging friend
EXEC sys.sp_executesql #SQL;
DB<>Fiddle