How to organise rows in a union - sql

Question

Having a little issue with my results where I am trying to create a fixture list for each team, playing home and away. I’ve almost got it working but the problem I have is displayed in the following results:
I have 2 teams playing each other in week 1 home and away and the same two teams playing each other home and away in week for.
What this should display when I insert the data and read it is week 1, team 4 is at home playing team 18 who are way, and then the reverse fixture is played in week 4.
In other words row 2 and 3 are incorrect and should only show rows 1 and 4. What do I need to change in the code below to get this to work?
INSERT INTO dbo.Fixture (WeekNumber, HomeTeamID, AwayTeamID, FixtureDate, LeagueID)
SELECT
ROW_NUMBER() OVER (PARTITION BY h.teamID ORDER BY h.TeamID, a.TeamID, h.LeagueID) AS WeekNumber,
h.TeamID,
a.TeamID,
DATEADD(day,(ROW_NUMBER() OVER (ORDER BY h.LeagueID)-1)*7,#StartFixtureWeek) AS FixtureWeek,
h.LeagueID
FROM dbo.Team h
CROSS JOIN dbo.Team a
WHERE h.TeamID <> a.TeamID
AND h.LeagueID = a.LeagueID
UNION
SELECT
ROW_NUMBER() OVER (PARTITION BY a.teamID ORDER BY h.TeamID, a.TeamID, h.LeagueID) AS WeekNumber,
h.TeamID,
a.TeamID,
DATEADD(day,(ROW_NUMBER() OVER (ORDER BY a.LeagueID)-1)*7,#StartFixtureWeek) AS FixtureWeek,
h.LeagueID
FROM dbo.Team h
CROSS JOIN dbo.Team a
WHERE h.TeamID <> a.TeamID
AND h.LeagueID = a.LeagueID
select * from dbo.Fixture
WHERE (HomeTeamID = 4 AND AwayTeamID = 18) OR (HomeTeamID = 18 AND AwayTeamID = 4)
UPDATE:
Below is an explanation and design of the desired output:
WeekNumber HomeTeamID AwayTeamID FixtureWeek LeagueID
1 1 4 10-06-2016 1
2 1 3 17-06-2016 1
3 1 2 24-06-2016 1
4 4 1 30-06-2016 1
5 3 1 06-07-2016 1
6 2 1 13-07-2016 1
1 2 3 10-06-2016 1
2 2 4 17-06-2016 1
3 3 4 24-06-2016 1
4 3 2 30-06-2016 1
5 4 2 06-07-2016 1
6 4 3 13-07-2016 1
1 5 8 10-06-2016 2
2 5 7 17-06-2016 2
3 5 6 24-06-2016 2
4 8 5 30-06-2016 2
5 7 5 06-07-2016 2
6 6 5 13-07-2016 2
1 6 7 10-06-2016 2
2 6 8 17-06-2016 2
3 7 8 24-06-2016 2
4 7 6 30-06-2016 2
5 8 6 06-07-2016 2
6 8 7 13-07-2016 2
Ok so have two leagues (LeagueID 1 and LeagueID 2)
In League 1 there are 4 teams (TeamID) – 1, 2, 3, 4 – They play each other home and away but they can’t play two games within the same week.
In League 2 there are 4 teams (TeamID) – 5, 6, 7, 8 – They play each other home and away but they can’t play two games within the same week.
Both Leagues start on the same day and add 7 days for every game (or in other words every week)
The output doesn’t show it but preferable if each team can play their home game one week and then the next week play away, then play home, then away etc. But if we can get the above output to display first and then fiddle about to order them home and away then that’s fine.

Answer 1

CROSS JOIN with h.TeamID <> a.TeamID generates all pairs, i.e. (1,2) and (2,1). I think it is easier if we separate them, so WHERE filter would become h.TeamID > a.TeamID or h.TeamID < a.TeamID.
Week numbers should be assigned for each League, so it should be partitioned by LeagueID. Then it is a matter of sorting teams in some way, and then in opposite direction.
In the result below you can see how a pattern of team pairs for 6 weeks repeat in two opposite directions.
Most likely it is possible to achieve the same or similar result with one CROSS JOIN that uses <> without UNION and a more complicated expression in ROW_NUMBER.
Side note: don't use UNION if you don't really need it. Use UNION ALL.
Side note 2: it doesn't matter if you use CROSS JOIN or INNER JOIN here. Optimizer would generate identical execution plans. I think CROSS JOIN here is more readable, shows more clearly the intention to generate a Cartesian product - set of all pairs.
Sample data
DECLARE #T TABLE (TeamID int, LeagueID int);
INSERT INTO #T (TeamID, LeagueID) VALUES
(1, 1),
(2, 1),
(3, 1),
(4, 1),
(5, 2),
(6, 2),
(7, 2),
(8, 2);
Query
SELECT
h.TeamID AS HomeTeamID
,a.TeamID AS AwayTeamID
,h.LeagueID
,ROW_NUMBER() OVER
(PARTITION BY h.LeagueID ORDER BY h.TeamID, a.TeamID) AS WeekNumber
,1 AS SortOrder
FROM
#T AS h
CROSS JOIN #T AS a
WHERE
h.LeagueID = a.LeagueID
AND h.TeamID < a.TeamID
UNION ALL
SELECT
h.TeamID AS HomeTeamID
,a.TeamID AS AwayTeamID
,h.LeagueID
,ROW_NUMBER() OVER
(PARTITION BY h.LeagueID ORDER BY a.TeamID DESC, h.TeamID DESC) AS WeekNumber
,2 AS SortOrder
FROM
#T AS h
CROSS JOIN #T AS a
WHERE
h.LeagueID = a.LeagueID
AND h.TeamID > a.TeamID
ORDER BY
LeagueID
,SortOrder
,WeekNumber
;
Result
+------------+------------+----------+------------+-----------+
| HomeTeamID | AwayTeamID | LeagueID | WeekNumber | SortOrder |
+------------+------------+----------+------------+-----------+
| 1 | 2 | 1 | 1 | 1 |
| 1 | 3 | 1 | 2 | 1 |
| 1 | 4 | 1 | 3 | 1 |
| 2 | 3 | 1 | 4 | 1 |
| 2 | 4 | 1 | 5 | 1 |
| 3 | 4 | 1 | 6 | 1 |
| 4 | 3 | 1 | 1 | 2 |
| 4 | 2 | 1 | 2 | 2 |
| 3 | 2 | 1 | 3 | 2 |
| 4 | 1 | 1 | 4 | 2 |
| 3 | 1 | 1 | 5 | 2 |
| 2 | 1 | 1 | 6 | 2 |
| 5 | 6 | 2 | 1 | 1 |
| 5 | 7 | 2 | 2 | 1 |
| 5 | 8 | 2 | 3 | 1 |
| 6 | 7 | 2 | 4 | 1 |
| 6 | 8 | 2 | 5 | 1 |
| 7 | 8 | 2 | 6 | 1 |
| 8 | 7 | 2 | 1 | 2 |
| 8 | 6 | 2 | 2 | 2 |
| 7 | 6 | 2 | 3 | 2 |
| 8 | 5 | 2 | 4 | 2 |
| 7 | 5 | 2 | 5 | 2 |
| 6 | 5 | 2 | 6 | 2 |
+------------+------------+----------+------------+-----------+