MySQL join two tables with group by in each - mysql

Question

I have two mysql tables with part numbers and qty's. I want to sum each tables qty sum(qty) ... group by partNumber Then join the two tables on the part number.
Sometimes table A will have part numbers that table b does not and vice versa. Below is an image of what I am expecting.
I've tried something like this, but this returns a row for each table and I want it to return 1 combined row
SELECT *, null as macroQty, sum(qty) as cardinalQty
FROM parts.cardinal where fileinfoid IN
(select cardinalFiles from parts.reports where fileinfoid = 418)
GROUP BY partNumber UNION ALL
SELECT *, sum(qty) as macroQty, null as cardinalQty
FROM parts.macro where fileinfoid IN
(select macroFiles from parts.reports where fileinfoid = 418 )
GROUP BY partNumber
I also tried wrapping it in an outer select and grouping by the part number from the outer select like this, but this results in the second inner select being null always
SELECT * FROM (
SELECT *, null as macroQty, sum(qty) as cardinalQty
FROM parts.cardinal where fileinfoid IN
(select cardinalFiles from parts.reports where fileinfoid = 418)
GROUP BY partNumber UNION ALL
SELECT *, sum(qty) as macroQty, null as cardinalQty
FROM parts.macro where fileinfoid IN
(select macroFiles from parts.reports where fileinfoid = 418 )
GROUP BY partNumber
) combined GROUP BY combined.partNumber

Answer 1

One approach would be to identify unique part numbers across the 2 tables (using a UNION with it's applied distinct) and then use correlated sub queries to get the sums. For example
drop table if exists a,b;
create table a(id int,val int);
create table b(id int,val int);
insert into a values(1,10),(1,10),(3,10),(4,10);
insert into b values (2,10),(4,10),(4,10);
select (select sum(a.val) from a where a.id = s.id) aval,
(select sum(b.val) from b where b.id = s.id) bval,
s.id partno
from
(
select id from a
union select id from b
) s
order by s.id;
+------+------+--------+
| aval | bval | partno |
+------+------+--------+
| 20 | NULL | 1 |
| NULL | 10 | 2 |
| 10 | NULL | 3 |
| 10 | 20 | 4 |
+------+------+--------+
4 rows in set (0.00 sec)

Answer 2

I would phrase this as a join between two subqueries which each find the sum in their respective tables. However, since each table does not necessarily contain all part numbers, and in fact there may be part numbers unique to each table, we will have to use a full outer join approach.
SELECT
t1.partNumber,
t1.cardinalQty,
COALECSE(t2.macroQty, 0) AS macroQty
FROM
(
SELECT partNumber, SUM(qty) AS cardinalQty
FROM cardinal
GROUP BY partNumber
) t1
LEFT JOIN
(
SELECT partNumber, SUM(qty) AS macroQty
FROM macro
GROUP BY partNumber
) t2
ON t1.partNumber = t2.partNumber
UNION ALL
SELECT
t2.partNumber,
0 AS cardinalQty,
t2.macroQty
FROM
(
SELECT partNumber, SUM(qty) AS cardinalQty
FROM cardinal
GROUP BY partNumber
) t1
RIGHT JOIN
(
SELECT partNumber, SUM(qty) AS macroQty
FROM macro
GROUP BY partNumber
) t2
ON t1.partNumber = t2.partNumber
WHERE t1.partNumber IS NULL;
Keep in mind that under normal conditions, in a well designed database, you should rarely encounter a situation which requires using a full outer join. Actually, a full outer join screams out that there is a design problem. In this case, you don't have a single parts table containing all part numbers. That table should exist, so unless you enjoy big ugly queries, you should create a parts table where the partNumber is a primary key.