MYSQL match multiple rows across same table with multiple ids - mysql

Question

I'm working out a way to process orders where I would have to satisfy the order from multiple vendors, and I want to know which 2 vendors could supply the given order.
The product_ids in my order are 10,20,30,40,50
SELECT *
FROM vendors
WHERE product_id IN (10,20,30,40,50)
gives me all vendors who have at least 1 of the products
vendor_id | product_id
1234 10
1234 20
1234 30
1234 40
1235 10
1235 40
1236 20
1236 30
1236 40
1237 50
9876 10
9876 20
9876 30
9876 40
9877 10
9877 40
9877 50
9878 10
9878 20
9878 30
9878 50
After several crazy subqueries (too long to post here) I can get to a table that shows me what each vendor is missing, and my thought is to then JOIN to that table based on those missing items and then show the vendors who have the refined list.
result should look like (comma separated or not doesn't matter, just easier to read)
vender_1 | product_id | missing_product_id | vendor_2
1234 10,20,30,40 50 9876,9878
1235 10,40 20,30,50 9878
1236 20,30,40 10,50 9877
1237 50 10,20,30,40 1234
or
vender_1 | product_id | missing_product_id | vendor_2
1234 10
1234 20
1234 30
1234 40
1234 50 9876
1234 50 9878
etc...

Answer 1

You want group by and having:
SELECT v.vendor_id
FROM vendors v -- Shouldn't this be called vendorProducts ?
WHERE v.product_id IN (10, 20, 30, 40, 50)
GROUP BY v.vendor_id
HAVING COUNT(DISTINCT v.product_id) = 5;
For multiple vendors, you can extend the above logic. The idea is to join the table together to get a list of pairs of vendors and all the products that they together have. Then, do the same logic as above:
SELECT v.vendor_id1, v.vendor_id2
FROM (SELECT DISTINCT v1.vendor_id as vendor_id1, v2.vendor_id as vendor_id2,
(CASE WHEN n.n = 1 THEN v1.product_id ELSE v2.product_id END) as product_id
FROM vendors v1 JOIN
vendors v2
ON v1.product_id <> v2.product_id AND
v1.vendor_id < v2.vendor_id CROSS JOIN
(SELECT 1 as n UNION ALL SELECT 2) n
UNION ALL
-- Then include the singletons, just in case
SELECT v.vendor_id, NULL, v.product_id
FROM vendors v
) v
WHERE v.product_id IN (10, 20, 30, 40, 50)
GROUP BY v.vendor_id1, v.vendor_id2
HAVING COUNT(DISTINCT v.product_id) = 5;
You can actually do the product filtering in the subquery -- to make the query more efficient. As for making this more general, the "5" is the number of items. I don't know how the ultimate query is being constructed.
EDIT II:
This is a hard problem with a lot of data. Here is another approach that might work better if you have lots of products and few vendors:
select v1.*, v2.*
from (select vendor_id,
max(product_id = 1) as p1,
max(product_id = 2) as p2,
max(product_id = 3) as p3,
max(product_id = 4) as p4,
max(product_id = 5) as p5,
from vendors
where product_id in (1, 2, 3, 4, 5)
group by vendor_id
) v1 join
(select vendor_id,
max(product_id = 1) as p1,
max(product_id = 2) as p2,
max(product_id = 3) as p3,
max(product_id = 4) as p4,
max(product_id = 5) as p5,
from vendors
where product_id in (1, 2, 3, 4, 5)
group by vendor_id
) v2
on (v1.p1 + v2.p1) > 0 and
(v1.p2 + v2.p2) > 0 and
(v1.p3 + v2.p3) > 0 and
(v1.p4 + v2.p4) > 0 and
(v1.p5 + v2.p5) > 0;
Note: If one vendor has all the products, then it will appear paired with every other vendor.