Select Query for Repeated Records in SQLite - sql

Question

This problem is a generalization of this question. Rather than finding all the games with specific players playing against others, I want to be able to find all the games where the same players played against each other.
Here is sample data:
1,ChrisEveret,1
1,BillieJeanKing,1
1,RogerFederer,0
1,TomasMuster,0
2,RogerFederer,1
2,SallieMae,1
2,NovakDjokovic,0
2,JimCourier,0
3,ChrisEveret,0
3,BillieJeanKing,0
3,RogerFederer,1
3,TomasMuster,1
The desired output is
1,ChrisEveret,1
1,BillieJeanKing,1
1,RogerFederer,0
1,TomasMuster,0
3,ChrisEveret,0
3,BillieJeanKing,0
3,RogerFederer,1
3,TomasMuster,1
The actual data has only about two thousand rows, so performance is not a concern. I have come up with the following remarkably convoluted and inexact partial solution:
CREATE TABLE sets (gameid int, player text ,winloss int);
.import data.csv sets
select * from sets where gameid in
(select gameid from (select gameid,mo from
(select gameid,mo,count(*) from
(select gameid,group_concat(player) as mo from
(select gameid,player from sets order by gameid,player)
group by gameid)
group by gameid)
where mo in
(select mo from (select gameid,mo,count(*) from
(select gameid,group_concat(player) as mo from
(select gameid,player from sets order by gameid,player)
group by gameid)
group by mo
having count(*)>1))));
This returns all matches where the same four people played together, but not necessarily those in which the teams were the same. I do not know if there is a solution to this problem that does not involve using group_concat(). That is the only way I was able to make even this limited progress on it, however. I also am not sure that the method used to order the group_concat results for aggregation will always work.

Answer 1

SQLite does not guarantee the ordering using group_concat() -- and there is no way to control it. So you have to use more cumbersome methods.
You can get the pairs of games with the same player using:
with s as (
select s.*, count(*) over (partition by gameid) as num_players
from sets s
)
select s1.gameid, s2.gameid
from s s1 join
s s2
on s1.player = s2.player and s1.num_players = s2.num_players
group by s1.gameid = s2.gameid
having count(*) = max(s1.num_players);
You can then use this logic if you want to get the players in each game (or just use group_concat() for that).
EDIT:
Window functions were introduced in SQLite version 3.28. In earlier versions, try this:
with s as (
select s.*, ss.num_players
from sets s join
(select gameid, count(*) as num_players
from sets s
group by gameid
) ss
on ss.gameid = s.gameid
)
select s1.gameid, s2.gameid
from s s1 join
s s2
on s1.player = s2.player and s1.num_players = s2.num_players
group by s1.gameid = s2.gameid
having count(*) = max(s1.num_players);
Here is a db<>fiddle that shows all pairs of games that have the same players (note that this includes each team to itself).