SQL Shifting 2 columns a row down with rollover - sql

Question

I am currently trying to manipulate a table in SQL Server.
The following table is an example:
Id | PrimaryId | PrimaryMail | SecondaryId | SecondaryMail
1. 1. email1#something. 5. email2#something
2. 2. email3#something. 6. email4#something
3. 3. email5#something. 7. email6#something
4. 4. email7#something. 8. email8#something
I want it to become the following:
Id | PrimaryId | PrimaryMail | SecondaryId | SecondaryMail
1. 1. email1#something. 6. email4#something
2. 2. email3#something. 7. email6#something
3. 3. email5#something. 8. email8#something
4. 4. email7#something. 5. email2#something
I found a similar question on here but the person did not want to use joins and the actual thing I am after is the rollover mechanic that makes either the first row become the last row or the last row the first row depending on what direction you shift the columns.
I already have the following code that makes the items in the columns go up by 1 row. But the last row currently stays the same and I lose the data from the first row.
--Calculate the height of the table
DECLARE #CouplesCount AS INT
SELECT #CouplesCount = COUNT(*) FROM Couples
--Shift the Secondary users down one row
UPDATE c1
SET c1.SecondaryId = c2.SecondaryId, c1.SecondaryMail = c2.SecondaryMail
FROM Couples c1 join
Couples c2
on c2.Id = c1.Id + 1
WHERE c1.CouplesId <= #CouplesCount - 1
SELECT * FROM Couples
Thank in advance to anyone that can help me.
p.s. I cannot use MySQL code

Answer 1

If the version of SQL Server that you use is SQL Server 2012 or later, you can use window functions LEAD and FIRST_VALUE.
LEAD returns the values from the "next" row (as defined by the ORDER BY clause). It will return NULL for the last row, because there is no "next" row there.
So, for the last row we need to return the first row, which can be done by the FIRST_VALUE.
Sample data
DECLARE #Couples TABLE (
Id int, PrimaryId int, PrimaryMail nvarchar(100),
SecondaryId int, SecondaryMail nvarchar(100));
INSERT INTO #Couples VALUES
(1, 1, 'email1#something', 5, 'email2#something'),
(2, 2, 'email3#something', 6, 'email4#something'),
(3, 3, 'email5#something', 7, 'email6#something'),
(4, 4, 'email7#something', 8, 'email8#something');
Query
SELECT
C.Id
,C.PrimaryId
,C.PrimaryMail
,ISNULL(LEAD(C.SecondaryId) OVER (ORDER BY C.Id),
FIRST_VALUE(C.SecondaryId) OVER (ORDER BY C.Id))
AS SecondaryId
,ISNULL(LEAD(C.SecondaryMail) OVER (ORDER BY C.Id),
FIRST_VALUE(C.SecondaryMail) OVER (ORDER BY C.Id))
AS SecondaryMail
FROM
#Couples AS C
;
Result
+----+-----------+------------------+-------------+------------------+
| Id | PrimaryId | PrimaryMail | SecondaryId | SecondaryMail |
+----+-----------+------------------+-------------+------------------+
| 1 | 1 | email1#something | 6 | email4#something |
| 2 | 2 | email3#something | 7 | email6#something |
| 3 | 3 | email5#something | 8 | email8#something |
| 4 | 4 | email7#something | 5 | email2#something |
+----+-----------+------------------+-------------+------------------+
Update
If you need to actually change the table itself, it is easy to do as well, just wrap the query in a CTE and update it.
WITH
CTE
AS
(
SELECT
C.Id
,C.PrimaryId
,C.PrimaryMail
,C.SecondaryId
,C.SecondaryMail
,ISNULL(LEAD(C.SecondaryId) OVER (ORDER BY C.Id),
FIRST_VALUE(C.SecondaryId) OVER (ORDER BY C.Id))
AS NewSecondaryId
,ISNULL(LEAD(C.SecondaryMail) OVER (ORDER BY C.Id),
FIRST_VALUE(C.SecondaryMail) OVER (ORDER BY C.Id))
AS NewSecondaryMail
FROM
#Couples AS C
)
UPDATE CTE
SET
SecondaryId = NewSecondaryId
,SecondaryMail = NewSecondaryMail
;
SELECT * FROM #Couples ORDER BY Id;