Select values based on other value within joined tables SQL - mysql

Question

I would like to ask for help for an SQL request that give me values from two tables.
As an example I have one Table orders und one table processing.
I would like to make an report of the orders and the state of processing.
table orders
id | status | div
-------------------
1 | wating_r | div1
2 | closed | div2
3 | closed | div3
-
table processing:
id | order_id | type | date
----------------------------------------
1 | 2 | send_request | 15.01.15
2 | 2 | send_invoice | 30.01.15
3 | 1 | send_request | 01.02.15
4 | 3 | send_request2 | 10.02.15
5 | 3 | send_invoice | 15.02.15
what I would like to get:
order_id | status | date_request | date_request2 | date_invoice
--------------------------------------------------------------------------------
1 | waiting_r | 01.02.15 | NULL | NULL
2 | closed | 15.01.15 | NULL | 30.01.15
3 | closed | NULL | 10.02.15 | 15.02.15
my solution:
select orders.id as order_id, orders.status, IF(processing.type='send_invoice',date_format(processing.date, '%Y-%m-%d'), NULL) as date_invoice, IF(processing.type='send_request',date_format(processing.date, '%Y-%m-%d'), NULL) as date_request, IF(processing.type='send_request2',date_format(processing.date, '%Y-%m-%d'), NULL) as date_request2
from orders
inner join processing on orders.id = processing.order_id
where
case
when orders.status='closed' then processing.type='send_invoice'
when orders.status='waiting_r' then processing.type='send_request'
when orders.status='waiting_2'then processing.type='send_request2'
end
This works fine but with this IF statements I doesn't become the dates from the requests when an invoice was sent - I only get the date of the invoice.
Instead of the case request I tried the following but in this case I have more than one line for every order. When I tried to "group by" I have mixed data.
where
processing.type in ('send_invoice', 'send_request', 'completion_request_send')

Answer 1

You need to left-join the second table to the first three times, like so.
SELECT o.id AS order_id, o.status,
p1.date AS date_request,
p2.date AS date_request2,
p3.date AS date_invoice
FROM orders o
LEFT JOIN processing p1 ON o.id = p1.order_id AND p1.type='send_request'
LEFT JOIN processing p2 ON o.id = p2.order_id AND p2.type='send_request2'
LEFT JOIN processing p3 ON o.id = p3.order_id AND p3.type='send_invoice'
ORDER BY 1,2
This left-join with an id-matching criterion and the specific type choice pulls out the rows you need for each column. Left, as opposed to inner, join, allows the missing values to be shown as null.
Here it is, working. http://sqlfiddle.com/#!9/b8c74/5/0
This is a typical pattern for joining a key/value table where the (id/key) pairs are unique.
Edit Unfortunately it generates duplicate result set rows in situations where there's a duplicate key for a particular value. To deal with that, it's necessary to deduplicate the key/value table (processing) in this case.
This subquery will do that, taking the latest date value.
SELECT type, order_id, MAX(date) AS date
FROM processing
GROUP BY type, order_id
Then you have to use that subquery in the main query. This is where it would be good if MySQL had common table expressions. But it doesn't so things get kind of verbose.
SELECT o.id AS order_id, o.status,
p1.date AS date_request,
p2.date AS date_request2,
p3.date AS date_invoice
FROM orders o
LEFT JOIN (
SELECT type, order_id, MAX(date) AS date
FROM processing
GROUP BY type, order_id
) p1 ON o.id = p1.order_id AND p1.type='send_request'
LEFT JOIN (
SELECT type, order_id, MAX(date) AS date
FROM processing
GROUP BY type, order_id
) p2 ON o.id = p2.order_id AND p2.type='send_request2'
LEFT JOIN (
SELECT type, order_id, MAX(date) AS date
FROM processing
GROUP BY type, order_id
) p3 ON o.id = p3.order_id AND p3.type='send_invoice'
ORDER BY 1,2