How to NULL the repeating column data? - mysql

Question

Is there a way to NULL the values that are known to be the same in the rest of the dataset?
mysql> SELECT
-> `p1`.`id`,
-> `p1`.`name`,
-> `pp1`.`name` `product_property_name`,
-> `pp1`.`value` `product_property_name`
-> FROM
-> `product` `p1`
-> INNER JOIN
-> `product_property` `pp1`
-> ON
-> `p1`.`id` = `pp1`.`product_id`;
+----+------+-----------------------+-----------------------+
| id | name | product_property_name | product_property_name |
+----+------+-----------------------+-----------------------+
| 1 | Tar | foo | bar |
| 1 | Tar | foo1 | bar1 |
| 1 | Tar | foo2 | bar2 |
| 2 | Qaz | too | doo |
| 2 | Qaz | too1 | doo1 |
+----+------+-----------------------+-----------------------+
5 rows in set (0.00 sec)
In this case product is returned more than once because of INNER JOIN with product_property. I only need the first row of every product to group the results.
Therefore, the desired output:
+----+------+-----------------------+-----------------------+
| id | name | product_property_name | product_property_name |
+----+------+-----------------------+-----------------------+
| 1 | Tar | foo | bar |
| 1 | NULL | foo1 | bar1 |
| 1 | NULL | foo2 | bar2 |
| 2 | Qaz | too | doo |
| 2 | NULL | too1 | doo1 |
+----+------+-----------------------+-----------------------+
This would allow to dramatically cut the memory utilisation, esp. when grouping large datasets.

Answer 1

You can do what you want using variables:
SELECT `p1`.`id`,
(case when name <> #prevname then `p1`.`name` end) as name,
`pp1`.`name` as `product_property_name`,
`pp1`.`value` as `product_property_name`,
#prevname := name
FROM `product` `p1` INNER JOIN
`product_property` `pp1` cross join
(select #prevname := '') const
ON `p1`.`id` = `pp1`.`product_id`;
Although this works in practice, it is not guaranteed to work, because MySQL does not guarantee the ordering of the evaluation of columns in a select statement (so in theory the #prevname := name could happen first).
However, you might consider using group_concat() instead, and dispensing with the multiple rows per property:
SELECT `p1`.`id`,
`p1`.`name`,
group_concat(pp1.name, ', ') as product_property_names,
group_concat(pp1.value, ', ') as product_property_values
FROM `product` `p1` INNER JOIN
`product_property` `pp1`
ON `p1`.`id` = `pp1`.`product_id`
group by p1.id, p1.name;
Or, as a single pair:
SELECT `p1`.`id`,
`p1`.`name`,
group_concat(concat(pp1.name, '=', pp1.value), '; ') as product_property_pairs
FROM `product` `p1` INNER JOIN
`product_property` `pp1`
ON `p1`.`id` = `pp1`.`product_id`
group by p1.id, p1.name;
The latter two should reduce the space more than the first solution.