Question

I have a table structure that contains a identifier column and a column that contains a deliminated string. What I would like to achieve is to insert the deliminated string into a new table as individual records for each of the values in the split deliminated string.

My table structure for the source table is as follows:

CREATE TABLE tablea(personID VARCHAR(8), delimStr VARCHAR(100))

Some sample data:

INSERT INTO tablea (personID, delimStr) VALUES ('A001','Monday, Tuesday')
INSERT INTO tablea (personID, delimStr) VALUES ('A002','Monday, Tuesday, Wednesday')
INSERT INTO tablea (personID, delimStr) VALUES ('A003','Monday')

My destination table is as follows:

CREATE TABLE tableb(personID VARCHAR(8), dayName VARCHAR(10))

I am attempting to create a Stored Procedure to undertake the insert, my SP so far looks like:

CREATE PROCEDURE getTKWorkingDays
  @pos integer = 1
  , @previous_pos integer = 0
  AS 
    BEGIN 
    DECLARE @value varchar(50)
            , @string varchar(100)
            , @ttk varchar(8)   
    WHILE @pos > 0 
      BEGIN 
        SELECT  @ttk = personID
                , @string = delimStr
                FROM    dbo.tablea
        SET @pos = CHARINDEX(',', @string, @previous_pos + 1) 
          IF @pos > 0 
            BEGIN       
              SET @value = SUBSTRING(@string, @previous_pos + 1, @pos - @previous_pos - 1)  
              INSERT  INTO dbo.tableb ( personID, dayName ) VALUES  ( @ttk, @value )        
              SET @previous_pos = @pos  
            END     
      END
      IF @previous_pos < LEN(@string) 
        BEGIN   
          SET @value = SUBSTRING(@string, @previous_pos + 1, LEN(@string))  
          INSERT  INTO dbo.tableb ( tkinit, dayName ) VALUES  ( @ttk, @value ) 
        END 
END

The data that was inserted (only 1 records out of the 170 or so in the original table which after spliting the deliminated string should result in about 600 or so records in the new table), was incorrect.

What I am expecting to see using the sample data above is:

personID    dayName
A001        Monday
A001        Tuesday
A002        Monday
A002        Tuesday
A002        Wednesday
A003        Monday

Is anyone able to point out any resources or identify where I am going wrong, and how to make this query work?

The Database is MS SQL Server 2000.

I thank you in advance for any assistance you are able to provide.

Matt

Answer 1

Well your SELECT statement which gets the “next” person doesn’t have a WHERE clause, so I’m not sure how SQL Server will know to move to the next person. If this is a one-time task, why not use a cursor?

CREATE TABLE #n(n INT PRIMARY KEY);

INSERT #n(n) SELECT TOP 100 number     FROM [master].dbo.spt_values 
WHERE number > 0    GROUP BY number    ORDER BY number;

DECLARE 
    @PersonID VARCHAR(8),  @delimStr VARCHAR(100), 
    @str VARCHAR(100),     @c CHAR(1);

DECLARE c CURSOR LOCAL FORWARD_ONLY STATIC READ_ONLY
    FOR SELECT PersonID, delimStr FROM dbo.tablea;

OPEN c;

FETCH NEXT FROM c INTO @PersonID, @delimStr;

SET @c = ',';

WHILE @@FETCH_STATUS = 0
BEGIN
    SELECT @delimStr = @c + @delimStr + @c;

    -- INSERT dbo.tableb(tkinit, [dayName])
    SELECT @PersonID, LTRIM(SUBSTRING(@delimStr, n+1, CHARINDEX(@c, @delimStr, n+1)-n-1))
        FROM #n AS n
        WHERE n.n <= LEN(@delimStr) - 1
        AND SUBSTRING(@delimStr, n.n, 1) = @c;

    FETCH NEXT FROM c INTO @PersonID, @delimStr;
END

CLOSE c;
DEALLOCATE c;

DROP TABLE #n;

If you create a permanent numbers table (with more than 100 rows, obviously) you can use it for many purposes. You could create a split function that allows you to do the above without a cursor (well, without an explicit cursor). But this would probably work best later, when you finally get off of SQL Server 2000. Newer versions of SQL Server have much more flexible and extensible ways of performing splitting and joining.