MySQL query difficulty - mysql

Question

I would like your help with the following:
query 1:
SELECT stock.ST_CODE as xCODE, (sum(ordercont.QTY) / 12) as x12WeeksAvg
FROM orders
LEFT JOIN ordercont ON ordercont.O_REF = orders.O_REF
LEFT JOIN stock ON stock.PD_CODE = ordercont.PD_CODE
WHERE stock.ST_CODE IS NOT NULL
AND orders.O_CUSTDATE BETWEEN DATE_SUB( (DATE_ADD(CURDATE(), INTERVAL - WEEKDAY(CURDATE()) DAY)) , INTERVAL 84 DAY)
AND DATE_SUB(DATE(NOW()), INTERVAL DAYOFWEEK(NOW())-1 DAY)
GROUP BY stock.ST_CODE
query 2:
SELECT stock.ST_CODE as xCODE, (sum(ordercont.QTY) / 4) as x4WeeksAvg
FROM orders LEFT JOIN ordercont ON ordercont.O_REF = orders.O_REF
LEFT JOIN stock ON stock.PD_CODE = ordercont.PD_CODE
WHERE stock.ST_CODE IS NOT NULL
AND orders.O_CUSTDATE BETWEEN DATE_SUB( (DATE_ADD(CURDATE(), INTERVAL - WEEKDAY(CURDATE()) DAY)) , INTERVAL 28 DAY)
AND DATE_SUB(DATE(NOW()), INTERVAL DAYOFWEEK(NOW())-1 DAY)
GROUP BY stock.ST_CODE
Is there any way to have the result like this (the queries returns different number of results):
------------------------------------
| xCode | x12WeeksAvg | x4WeeksAvg |
------------------------------------
| ... | ... | ... |
------------------------------------
Thank you!!

Answer 1

Split the BETWEEN condition
orders.O_CUSTDATE BETWEEN DATE_SUB( (DATE_ADD(CURDATE(), INTERVAL - WEEKDAY(CURDATE()) DAY)) , INTERVAL 84 DAY)
AND DATE_SUB(DATE(NOW()), INTERVAL DAYOFWEEK(NOW())-1 DAY)
into two parts
AND orders.O_CUSTDATE >= DATE_SUB( (DATE_ADD(CURDATE(), INTERVAL - WEEKDAY(CURDATE()) DAY)) , INTERVAL 84 DAY)
AND orders.O_CUSTDATE <= DATE_SUB(DATE(NOW()), INTERVAL DAYOFWEEK(NOW())-1 DAY)
and rewrite it to
AND orders.O_CUSTDATE >= CURDATE() - INTERVAL (WEEKDAY(NOW()) + 84) DAY
AND orders.O_CUSTDATE <= CURDATE() - INTERVAL (DAYOFWEEK(NOW()) - 1) DAY
to get shorter and more readable code. Do the same for the second query.
Now the two queries differ only in two expressions
(sum(ordercont.QTY) / 12) as x12WeeksAvg
(sum(ordercont.QTY) / 4) as x4WeeksAvg
orders.O_CUSTDATE >= CURDATE() - INTERVAL (WEEKDAY(NOW()) + 84) DAY
orders.O_CUSTDATE >= CURDATE() - INTERVAL (WEEKDAY(NOW()) + 28) DAY
Given
<X>: orders.O_CUSTDATE >= CURDATE() - INTERVAL (WEEKDAY(NOW()) + 84) DAY
and
<Y>: orders.O_CUSTDATE >= CURDATE() - INTERVAL (WEEKDAY(NOW()) + 28) DAY
you can use conditional aggregation to combine the two queries:
SELECT stock.ST_CODE as xCODE,
(sum(CASE WHEN <X> THEN ordercont.QTY ELSE 0 END) / 12) as x12WeeksAvg,
(sum(CASE WHEN <Y> THEN ordercont.QTY ELSE 0 END) / 4) as x4WeeksAvg
FROM orders
LEFT JOIN ordercont ON ordercont.O_REF = orders.O_REF
LEFT JOIN stock ON stock.PD_CODE = ordercont.PD_CODE
WHERE stock.ST_CODE IS NOT NULL
AND orders.O_CUSTDATE <= CURDATE() - INTERVAL (DAYOFWEEK(NOW()) - 1) DAY
GROUP BY stock.ST_CODE
Replace <X> and <Y> accordingly.
Update
To improve the performance you can add
AND orders.O_CUSTDATE >= CURDATE() - INTERVAL (WEEKDAY(NOW()) + 84) DAY
to the WHERE clause.
Also note that the condition stock.ST_CODE IS NOT NULL will convert your LEFT JOINs to INNER JOINs. So you could as well just write JOIN.

Answer 2

You can subquery them like this to get your desired output:
SELECT A.xCODE,
COALESCE(B.x12WeeksAvg, 'N/A') as x12WeeksAvg,
COALESCE(C.x4WeeksAvg, 'N/A') as x4WeeksAvg
FROM (SELECT DISTINCT ST_CODE as xCODE FROM stock) A
LEFT JOIN (
SELECT stock.ST_CODE as xCODE, (sum(ordercont.QTY) / 12) as x12WeeksAvg
FROM orders
LEFT JOIN ordercont ON ordercont.O_REF = orders.O_REF
LEFT JOIN stock ON stock.PD_CODE = ordercont.PD_CODE
WHERE stock.ST_CODE IS NOT NULL
AND orders.O_CUSTDATE BETWEEN DATE_SUB( (DATE_ADD(CURDATE(), INTERVAL - WEEKDAY(CURDATE()) DAY)) , INTERVAL 84 DAY) AND DATE_SUB(DATE(NOW()), INTERVAL DAYOFWEEK(NOW())-1 DAY)
GROUP BY stock.ST_CODE
) B ON A.xCODE = B.xCODE
LEFT JOIN (
SELECT stock.ST_CODE as xCODE, (sum(ordercont.QTY) / 4) as x4WeeksAvg
FROM orders
LEFT JOIN ordercont ON ordercont.O_REF = orders.O_REF
LEFT JOIN stock ON stock.PD_CODE = ordercont.PD_CODE
WHERE stock.ST_CODE IS NOT NULL
AND orders.O_CUSTDATE BETWEEN DATE_SUB( (DATE_ADD(CURDATE(), INTERVAL - WEEKDAY(CURDATE()) DAY)) , INTERVAL 28 DAY) AND DATE_SUB(DATE(NOW()), INTERVAL DAYOFWEEK(NOW())-1 DAY)
GROUP BY stock.ST_CODE
) C ON A.xCODE = C.xCODE
WHERE COALESCE(B.x12WeeksAvg, C.x4WeeksAvg) IS NOT NULL