SQL split values with same ID from 2 tables - sql-server

Question

I have 2 tables (that's retrieving data from a main table). Example:
Table1
id GroupX Source GroupNum Amount
-------------------------------------------------------
1 23 School SH001 1 700
2 23 Bank BA001 2 300
3 23 Music MU001 3 500
4 23 School SH999 1 900
Table2
id GroupNum SourceAmt
----------------------------------
1 23 1 700
2 23 2 100
3 23 3 500
4 23 1 900
My dilemma is with the query I'm using. It's returning additional rows for split values(notice in table 2 "GroupNum" has a split value of 700 and 900)
My results should be
id GroupX Source GroupNum Amount SourceAmt
-----------------------------------------------------------------
1 23 School SH001 1 700 700
2 23 Bank BA001 2 300 100
3 23 Music MU001 3 500 500
4 23 School SH999 1 900 900
But instead I get this
id GroupX Source GroupNum Amount SourceAmt
-----------------------------------------------------------------
1 23 School SH001 1 700 700
2 23 School SH001 1 700 900
3 23 Bank BA001 2 300 100
4 23 Music MU001 3 500 500
5 23 School SH999 1 900 900
6 23 School SH999 1 900 700
Here's my query:
SELECT
t1.id,
t1.GroupX,
t1.Source,
t1.GroupNum,
t1.Amount,
t2.SourceAmt
FROM
table1 as t1
INNER JOIN
table2 as t2 ON t1.id = t2.id
AND t1.GroupNum = t2.GroupNum
WHERE
t1.id = 23
I've tried using Distinct as well. Assistance will be greatly appreciated.

Answer 1

If I understand correctly, you would like to join Table1 and Table2 such that the id, GroupNum, and amounts align. If this is indeed the case, then you'll need to join on the amounts as well, e.g.,:
Select t1.id, t1.Group, t1.Source, t1.GroupNum, t1.Amount, t2.SourceAmt
From table1 as t1 INNER JOIN
table2 as t2
ON t1.id = t2.id AND t1.GroupNum = t2.GroupNum AND t1.Amount = t2.SourceAmt
where id = 23
If this is not what you want, or you would prefer not to join using the amounts (e.g., you cannot guarantee that you won't see the same amount more than once), then you're in a bit of a conundrum; you'll note that (id, GroupNum) tuples are not unique in either table, and thus your join is not one-to-one. You may want to include Source in table2 or otherwise provide a transactionId in table1 that maps onto a unique ID column in table2.

Answer 2

You need an additional join key. There is no obvious candidate, except perhaps for amount -- but I'm not sure that is what you intend. SQL tables represent unordered sets, so there is no concept of matching by "line number".
You can assign a row number using row_number(). The following will do the match, but you need to specify the ordering column:
Select t1.id, t1.Group, t1.Source, t1.GroupNum, t1.Amount, t2.SourceAmt
From (select t1.*,
row_number() over (partition by t1.id order by ?) as seqnum
from table1 t1
) t1 inner join
(select t2.*
row_number() over (partition by t1.id order by ?) as seqnum
from table2 t2
) t2
on t1.id = t2.id and t1.GroupNum = t2.GroupNum and
t1.seqnum = t2.seqnum
where id = 23 ;
The ? is for the ordering column in each table.

Answer 3

I would choose a different approach from simple INNER JOIN, simply because you can do limited things with that result set (here's the result set from that inner join)
I would do multiple joins.
First I would LEFT JOIN with the default condition + where table1.[Amount] = table2.[SourceAmt]. This will give me set where [Amount] and [SourceAmt] are equal
After this I INNER JOIN with the default condition to get the Amounts that doesn't match
Here's the query
with t1 as
(
select 23 as [id], 'School' as [Group], 'SH001' as [Source], 1 as [GroupNum], 700 as [Amount]
union all
select 23, 'Bank', 'BA001', 2, 300
union all
select 23, 'Music', 'MU001', 3, 500
union all
select 23, 'School', 'SH999', 1, 900
),
t2 as
(
select 23 as [id], 1 as [GroupNum], 700 as [SourceAmt]
union all
select 23, 2, 100
union all
select 23, 3, 500
union all
select 23, 1, 900
)
select t1.*, a.*, b.*
from t1
left join t2 as a on
t1.[id] = a.[id]
and t1.[GroupNum] = a.[GroupNum]
and t1.[Amount] = a.[SourceAmt]
inner join t2 as b on
t1.[id] = b.[id]
and t1.[GroupNum] = b.[GroupNum]
where t1.[id] = 23
And here's the result set, which you can inspect
Now, I use this result as my pre-result actually, and do a little trick with CASE and [takeIt] column, here's the final query
with t1 as
(
select 23 as [id], 'School' as [Group], 'SH001' as [Source], 1 as [GroupNum], 700 as [Amount]
union all
select 23, 'Bank', 'BA001', 2, 300
union all
select 23, 'Music', 'MU001', 3, 500
union all
select 23, 'School', 'SH999', 1, 900
),
t2 as
(
select 23 as [id], 1 as [GroupNum], 700 as [SourceAmt]
union all
select 23, 2, 100
union all
select 23, 3, 500
union all
select 23, 1, 900
),
res as
(
select t1.[id],
t1.[Group],
t1.[Source],
t1.[GroupNum],
t1.[Amount],
isnull(a.[SourceAmt], b.[SourceAmt]) as [SourceAmt],
case when a.[SourceAmt] is null or a.[SourceAmt] = b.[SourceAmt] then 1
else 0
end as [takeIt]
from t1
left join t2 as a on
t1.[id] = a.[id]
and t1.[GroupNum] = a.[GroupNum]
and t1.[Amount] = a.[SourceAmt]
inner join t2 as b on
t1.[id] = b.[id]
and t1.[GroupNum] = b.[GroupNum]
where t1.[id] = 23
)
select [id],
[Group],
[Source],
[GroupNum],
[Amount],
[SourceAmt]
from res
where [takeIt] = 1