Question

I have the following tables in a SQL Server 2000 database:

Master

MasterID | Details   | [other fields]
=====================================
PK (int) | Free text | ...

LogTable

LogID    | MasterID | UserID    | LogDate    | LogText
==========================================================
PK (int) | FK (int) | VarChar(2)| Date stamp | Free text

There may be many Log entries for each master record.

I have a query which extracts the most recent three associated Log entries for each Master row as shown below. Note that appropriate conversion and formatting is performed to achieve the LogData concatenation (omitted for clarity):

SELECT 
    M.MasterID, M.Details, L.LogDate + L.UserID + L.LogText AS LogData 
FROM
    MasterTable M 
INNER JOIN 
    LogTable L ON M.MasterID = L.MasterID 
    AND L.LogID IN (SELECT TOP 3 LogID FROM LogTable 
                    WHERE MasterID = M. MasterID ORDER BY LogDate DESC)

This produces output like this:

MasterID | Details | LogData
========================================================
1        | First   | 05/11/2012 AB Called Client  
2        | Second  | 08/11/2012 CD Client Visit  
2        | Second  | 07/11/2012 CD Called Client  
2        | Second  | 05/11/2012 AB Called Client  

What I need to achieve is showing the data from the second table as columns in the output, all reported against each single master record, thus avoiding repeated data. Like so:

MasterID | Details | LogData1                    | LogData2                    | LogData3
===========================================================================================================
1        | First   | 05/11/2012 AB Called Client | (null)                      | (null)  
2        | Second  | 08/11/2012 CD Client Visit  | 07/11/2012 CD Called Client | 05/11/2012 AB Called Client  

Note that in the real world requirement, this solution will be part of flattening 5 tables with the output consisting of approx 20,000 rows and 90 columns of data.

Thanks in advance.

Answer 1

I’m going to post this, just to show it can be done, but HIGHLY SUGGEST, not do it through SQL. Should be done through the UI that’s displaying to be more dynamic on your columns. Even then, I would design this differently.

-- create master table
DECLARE @MasterTable TABLE (
    [MasterID] [int] IDENTITY (1, 1) NOT NULL ,
    [Details] [varchar] (50) ,
    [AdditionalField_1] [varchar] (50) ,
    [AdditionalField_n] [varchar] (50) 
)
-- create log table
DECLARE @LogTable TABLE (
    [LogID] [int] IDENTITY (1, 1) NOT NULL ,
    [MasterID] [int] NULL ,
    [UserID] [varchar] (2) ,
    [LogDate] [datetime] NULL ,
    [LogText] [varchar] (50) 
) 
-- insert into master table
INSERT INTO @MasterTable  (Details)
        VALUES ('First')
INSERT INTO @MasterTable  (Details)
        VALUES ('Second') 
-- insert into log table
INSERT INTO @LogTable  (MasterID, UserID, LogDate, LogText)
        VALUES (1, 'AB', '05/11/2012', 'Called Client')   
INSERT INTO @LogTable  (MasterID, UserID, LogDate, LogText)
        VALUES (2, 'AB', '05/11/2012', 'Called Client')   
INSERT INTO @LogTable  (MasterID, UserID, LogDate, LogText)
        VALUES (2, 'CD', '07/11/2012', 'Called Client')   
INSERT INTO @LogTable  (MasterID, UserID, LogDate, LogText)
        VALUES (2, 'CD', '08/11/2012', 'Client Visit') 
-- create table to display data
DECLARE @MyTemp TABLE (MasterID INT, Details VARCHAR(50), LogData1 VARCHAR(50), LogData2 VARCHAR(50), LogData3 VARCHAR(50))
INSERT INTO @MyTemp SELECT MasterID, Details, NULL, NULL, NULL FROM @MasterTable
-- create vars
DECLARE @ID INT, @NewID INT, @MasterID INT, @NewValue VARCHAR(100)
SET @ID = 0
-- loop through data
WHILE @ID >-1
BEGIN
    -- clear vars
    SELECT @NewID = NULL, @MasterID = NULL, @NewValue = NULL
    -- get first record
    SELECT TOP 1    
        @NewValue = CONVERT(VARCHAR(10), LogDate, 103)+ ' ' + UserID + ': ' + LogText
    ,   @MasterID=MasterID
    ,   @NewID=LogID 
    FROM @LogTable WHERE LogID>@ID
    -- if no data, exit loop
    IF @NewID IS NULL
        BREAK
    -- update record based on valuds in fields
    UPDATE m 
        SET @ID = @NewID
        ,   LogData1 = (CASE WHEN m.LogData1 IS NULL THEN @NewValue ELSE m.LogData1 END)
        ,   LogData2 = (CASE WHEN m.LogData1 IS NOT NULL THEN 
                            (CASE WHEN m.LogData2 IS NULL THEN @NewValue ELSE m.LogData2 END)
                        ELSE m.LogData2 END)
        ,   LogData3 = (CASE WHEN m.LogData1 IS NOT NULL THEN 
                            (CASE WHEN m.LogData2 IS NOT NULL THEN 
                                (CASE WHEN m.LogData3 IS NULL THEN @NewValue ELSE m.LogData3 END)
                             ELSE m.LogData3 END)
                        ELSE m.LogData3 END)
    FROM @MyTemp m
    WHERE m.MasterID=@MasterID
END
--display all data
SELECT * FROM @MyTemp