Cross-multiplying table - sql

Question

I have this SQL code and I want to show the sum of each item on its charge slip and on their receipt:
select item_description, sum(receipt_qty) as Samp1, sum(chargeSlip_qty) as Samp2
from Items inner join Receipt_Detail on (Receipt_Detail.item_number =
Items.item_number)
inner join ChargeSlip_Detail on (ChargeSlip_Detail.item_number =
Items.item_number)
group by item_description
It produces this output:
Acetazolamide 2681 1730
Ascorbic Acid 1512 651
Paracetamol 1370 742
Silk 576 952
But it should be:
Acetazolamide 383 173
Ascorbic Acid 216 93
Paracetamol 274 106
Silk 96 238
What's wrong with my code?

Answer 1

Since you are joining tables, you might have a one-to-many relationship that is causing the problem when you then get the sum(). So you can use subqueries to get the result. This will get the sum() for the receipt and chargeslip for each item_number and then you join that back to your items table to get the final result:
select i.item_description,
r.Samp1,
c.Samp2
from Items i
inner join
(
select sum(receipt_qty) Samp1,
item_number
from Receipt_Detail
group by item_number
) r
on r.item_number = i.item_number
inner join
(
select sum(chargeSlip_qty) Samp2,
item_number
from ChargeSlip_Detail
group by item_number
) c
on c.item_number = i.item_number

Answer 2

Do the GROUP BYs first, per Item_Number, so you don't multiply out rows from Receipt_Detail and ChargeSlip_Detail. That is, you generate the SUM values per Item_Number before JOINing back to Items
select
I.item_description,
R.Samp1,
C.Samp2
from
Items I
inner join
(SELECT item_number, sum(receipt_qty) as Samp1
FROM Receipt_Detail
GROUP BY item_number
) R
on (R.item_number = I.item_number)
inner join
(SELECT item_number, sum(chargeSlip_qty) as Samp2
FROM ChargeSlip_Detail
GROUP BY item_number
) C
on (C.item_number = I.item_number)

Answer 3

A left join returns rows from the left table, and for each row in the left table, all matching rows in the right table.
So for example:
create table Customers (name varchar(50));
insert Customers values
('Tim'),
('John'),
('Spike');
create table Orders (customer_name varchar(50), product varchar(50));
insert Orders values (
('Tim', 'Guitar'),
('John', 'Drums'),
('John', 'Trumpet');
create table Addresses (customer_name varchar(50), address varchar(50));
insert Addresses values (
('Tim', 'Penny Lane 1'),
('John', 'Abbey Road 1'),
('John', 'Abbey Road 2');
Then if you run:
select c.name
, count(o.product) as Products
, count(a.address) as Addresses
from Customers c
left join Orders o on o.customer_name = c.name
left join Addresses a on a.customer_name = c.name
group by name
You get:
name Products Addresses
Tim 1 1
John 4 4
Spike 0 0
But John doesn't have 4 products!
If you run without the group by, you can see why the counts are off:
select *
from Customers c
left join Orders o on o.customer_name = c.name
left join Addresses a on a.customer_name = c.name
You get:
name customer_name product customer_name address
Tim Tim Guitar Tim Penny Lane 1
John John Drums John Abbey Road 1
John John Drums John Abbey Road 2
John John Trumpet John Abbey Road 1
John John Trumpet John Abbey Road 2
Spike NULL NULL NULL NULL
As you can see, the joins end up repeating each other. For each product, the list of addresses is repeated. That gives you the wrong counts. To solve this problem, use one of the excellent other answers:
select c.name
, o.order_count
, a.address_count
from Customers c
left join
(
select customer_name
, count(*) as order_count
from Orders
group by
customer_name
) o
on o.customer_name = c.name
left join
(
select customer_name
, count(*) as address_count
from Addresses
group by
customer_name
) a
on a.customer_name = c.name
The subqueries ensure only one row is joined per customer. The result is much better:
name order_count address_count
Tim 1 1
John 2 2
Spike NULL NULL