SQL Join data and get rows that don't match with NULL - sql

Question

I have two tables that I want join as follows:
Table 1
Code1 | Code2 | Date(1) | Amount(1)
A | AA | 201802 | 100
A | AA | 201803 | 50
A | AA | 201804 | 30
Table 2
Code1 | Code2 | Date(2) | Amount(2)
A | AA | 201801 | 20
A | AA | 201802 | 10
A | AA | 201803 | 10
And I want the resulting table to look like this:
Result
Code1 | Code2 | Date(1) | Date(2) | Amount(1) | Amount(2)
A | AA | NULL | 201801 | NULL | 20
A | AA | 201802 | 201802 | 100 | 10
A | AA | 201803 | 201803 | 50 | 10
A | AA | 201804 | NULL | 30 | NULL
So I need to join these two tables
on table1.Code1 = table2.Code1 AND table1.Code2 = table2.Code2 AND table1.Date(1) = table2.Date(2)
But I also want the rows where the dates don't match with a null is the columns related to the non matching table (such as the row for Date(1) = 201804 in my example).
I have tried joining that two tables with left, right and outer join but I still am not successful in getting the rows with the nulls (probably because Code1 and Code2 don't exist for that particular missing row)
Maybe a cross apply could work, but I am not sure how to execute it.
I want the most efficient way in terms of performance because this is a part of a big query containing lots of data and lots of calculations.
UPDATE:
The code I used is:
Select table1.Code 1, table1.Code2, Table1.Date(1), table2.Date(2), table1.Amount(1), table2.amount(2)
FROM Table1
Full Outer Join
table2 ON
table1.Code1 = table2.Code1
AND table1.Code2 = table2.Code2
AND table1.date(1) = table2.date(2)
Which gives me the following result:
Code1 | Code2 | Date(1) | Date(2) | Amount(1) | Amount(2)
A | AA | 201802 | 201802 | 100 | 10
A | AA | 201803 | 201803 | 50 | 10
Which is missing these two rows:
A | AA | NULL | 201801 | NULL | 20
A | AA | 201804 | NULL | 30 | NULL

Answer 1

You may try this.
--sample dataset
DECLARE #tab1 as table (
Code1 varchar(10),
Code2 varchar(10),
Date1 int,
Amount1 int )
insert into #tab1
values
('A', 'AA', 201802, 100),
('A', 'AA', 201803, 50),
('A', 'AA', 201804, 30),
('B', 'AA', 201802, 100) --additional
DECLARE #tab2 as table (
Code1 varchar(10),
Code2 varchar(10),
Date2 int,
Amount2 int )
insert into #tab2
values
('A', 'AA', 201802, 100),
('A', 'AA', 201803, 50),
('A', 'AA', 201801, 30)
query
SELECT *
FROM (
select
coalesce(table1.Code1,table2.Code1) as Code1,
coalesce(table1.Code2,table2.Code2) as Code2,
table1.Date1,
table2.Date2,
table1.Amount1,
table2.amount2
FROM #tab1 as Table1
Full Outer Join #tab2 as table2 ON
table1.Code1 = table2.Code1
AND table1.Code2 = table2.Code2
AND table1.date1= table2.date2
) as t1
CROSS APPLY ( --to exclude records not matched by "Code 1 and Code 2"
SELECT top 1
Code1
FROM #tab2 as t
where t.Code1 = t1.Code1
and t.Code2 = t1.Code2
) as c
ORDER BY t1.Date1
or like this:
select
coalesce(table1.Code1,table2.Code1) as Code1,
coalesce(table1.Code2,table2.Code2) as Code2,
table1.Date1,
table2.Date2,
table1.Amount1,
table2.amount2
FROM #tab1 as Table1
Full Outer Join #tab2 as table2 ON
table1.Code1 = table2.Code1
AND table1.Code2 = table2.Code2
AND table1.date1= table2.date2
where exists (select null --to exclude records not matched by "Code 1 and Code 2"
from #tab2 as t2
where coalesce(table1.Code1,table2.Code1) = t2.Code1
and coalesce(table1.Code2,table2.Code2) = t2.Code2)
ORDER BY table1.Date1

Answer 2

My suggested solution involves a full join and another join to a derived table that contains all the combinations of code1 and code2 that exists in both tables, using the intersect operator.
First, create and populate sample data (Please save us this step in your future questions):
DECLARE #T1 AS TABLE
(
Code1 char(1),
Code2 char(2),
Date1 char(6),
Amount1 int
)
DECLARE #T2 AS TABLE
(
Code1 char(1),
Code2 char(2),
Date2 char(6),
Amount2 int
)
INSERT INTO #T1 (Code1, Code2, Date1, Amount1) VALUES
('A', 'AA', '201802', 100)
,('A', 'AA', '201803', 50)
,('A', 'AA', '201804', 30)
,('B', 'AA', '201802', 30); -- Note: Added to the original sample data
INSERT INTO #T2 (Code1, Code2, Date2, Amount2) VALUES
('A', 'AA', '201801', 20)
,('A', 'AA', '201802', 10)
,('A', 'AA', '201803', 10)
,('A', 'AB', '201802', 10); -- Note: Added to the original sample data
The query:
SELECT ISNULL(T1.Code1, T2.Code1) As Code1,
ISNULL(T1.Code2, T2.Code2) As Code2,
Date1, Date2, Amount1, Amount2
FROM #T1 As T1
FULL JOIN #T2 As T2
ON T1.Code1 = T2.Code1
AND T1.Code2 = T2.Code2
AND T1.Date1 = T2.Date2
-- Remove this next join if you want to get rows where codes don't match
JOIN (
SELECT Code1, Code2
FROM #T1
INTERSECT
SELECT Code1, Code2
FROM #T2
) As CommonCodes
ON CommonCodes.Code1 = ISNULL(T1.Code1, T2.Code1)
AND CommonCodes.Code2 = ISNULL(T1.Code2, T2.Code2)
ORDER BY Date1
Results:
Code1 Code2 Date1 Date2 Amount1 Amount2
A AA NULL 201801 NULL 20
A AA 201802 201802 100 10
A AA 201803 201803 50 10
A AA 201804 NULL 30 NULL
You can see a live demo on rextester.

Answer 3

Your updated query should work if you ISNULL the CodeX columns.
declare #t1 table (Code1 varchar(4), Code2 varchar(4), Date1 date, Amount1 int)
declare #t2 table (Code1 varchar(4), Code2 varchar(4), Date2 date, Amount2 int)
insert into #t1
values
('A', 'AA', '2018-02-01', 100 ),
('A', 'AA', '2018-03-01', 50 ),
('A', 'AA', '2018-04-01', 30 )
insert into #t2
values
('A', 'AA', '2018-01-01', 20 ),
('A', 'AA', '2018-02-01', 10 ),
('A', 'AA', '2018-03-01', 10 )
SELECT
code1
,code2
,date1
,date2
,amount1
,amount2
FROM (
SELECT code1, code2 FROM #t1
INTERSECT
SELECT code1, code2 FROM #t2
) t0
CROSS APPLY (
SELECT
date1, date2, amount1, amount2
FROM #t1 t1
FULL OUTER JOIN #t2 t2 ON t1.Code1 = t2.Code1 and t1.Code2 = t2.Code2 and date1 = date2
WHERE
t0.code1 = isnull(t1.Code1, t2.code1)
and t0.code2 = isnull(t1.Code2, t2.code2)
) tt
ORDER BY
date1, date2