Question

I have two tables containing three columns containing the same types of information(id, attributes, and amounts). I would like to combine them in a union and join that union with another table containing a list of id’s to filter them, then sum common attributes.

table_a                     table_b                   table_ids
id   attr    amount         id   attr     amount      uid     a_id
1    'atr1'  10             1    'atr2'   4           'id1'   1
1    'atr2'  5              2    'atr3'   2           'id2'   2
2    'atr3'  7              2    'atr1'   11
3    'atr1'  8

I tried

SELECT table_ids.uid, t.attr, SUM(t.amount)
FROM table_ids
JOIN (SELECT CONCAT('a', id) AS comb_id, attr, amount FROM table_a
    UNION ALL
    SELECT CONCAT('b', id) AS comb_id, attr, amount FROM table_b
    ) t
ON CONCAT('a', table_ids.a_id) = t.comb_id
GROUP BY t.attr

but am only getting the attribute for last matching a_id and the number of matching rows times the last matching row amount for the sum.

(There will be another table with more id’s filtering table_b attributes but I haven’t even gotten one filtering table to work yet.)

Answer 1

Hard to be 100% sure what you are aiming for but something like this may fit the bill

SELECT
    joinedData.uid,
    joinedData.attr,
    SUM(joinedData.amount) AS amount
FROM (
    (
        SELECT
            table_ids.uid,
            table_a.attr,
            SUM(table_a.amount) AS amount
        FROM table_a
        INNER JOIN table_ids
        ON table_ids.a_id = table_a.id
        GROUP BY table_ids.uid, table_a.attr
    )
    UNION
    (
        SELECT
            table_ids.uid,
            table_b.attr,
            SUM(table_b.amount) AS amount
        FROM table_b
        INNER JOIN table_ids
        ON table_ids.b_id = table_b.id
        GROUP BY table_ids.uid, table_b.attr
    )
) AS joinedData
GROUP BY joinedData.uid, joinedData.attr

Have of course assumed that table_b will join id onto table_ids.b_id, change it as required.