Get the duplicate entries from table mysql - mysql

Question

I have the table structure as shown below. The database is MariaDB.
+-----------+----------+--------------+-----------------+
| id_object | name | value_double | value_timestamp |
+-----------+----------+--------------+-----------------+
| 1 | price | 1589 | null |
| 1 | payment | 1590 | null |
| 1 | date | null | 2012-04-17 |
| 2 | price | 1589 | null |
| 2 | payment | 1590 | null |
| 2 | date | null | 2012-04-17 |
| 3 | price | 1589 | null |
| 3 | payment | 1590 | null |
| 3 | date | null | 2012-09-25 |
| ... | ... | ... | .. |
+-----------+----------+--------------+-----------------+
1) I need to get the duplicates by three entries: price & payment & date;
For example: the record with id_object=2 is duplicate because price, payment and date are the same as values of the record with id_object=1. Record with id_object = 3 is not the duplicate because the date is different (2012-09-25 != 2012-04-17)
2) I should remove the duplicates except one copy of them.
I thought to do three select operations and join each select on id_object. I can get the duplicates by one entry (price | payment | date). I faced the problem doing the joins
SELECT `id_object`,`name`,{P.`value_double` | P.`value_timestamp`}
FROM record P
INNER JOIN(
SELECT {value_double | value_timestamp}
FROM record
WHERE name = {required_entry}
GROUP BY {value_double | value_timestamp}
HAVING COUNT(id_object) > 1
)temp ON {P.value_double = temp.value_double | P.value_timestamp = temp.value_timestamp}
WHERE name = {required_entry}
Can someone help and show the pure (better) solution?

Answer 1

Though less efficient than certain alternatives, I find an approach along these lines easier to read...
SELECT MIN(id_object) id_object
, price
, payment
, date
FROM
( SELECT id_object
, MAX(CASE WHEN name = 'price' THEN value_double END) price
, MAX(CASE WHEN name = 'payment' THEN value_double END) payment
, MAX(CASE WHEN name = 'date' THEN value_timestamp END) date
FROM eav
GROUP
BY id_object
) x
GROUP
BY price
, payment
, date;

Answer 2

I would just group_concat() the values together and do the test that way:
select t.*
from t join
(select min(id_object) id_object
from (select id_object,
group_concat(name, ':', coalesce(value_double, ''), ':', coalesce(value_timestamp, '') order by name) pairs
from t
where name in ('price', 'payment', 'date')
group by id_object
) tt
group by pairs
) tt
on t.id_object = tt.id_object;
To actually delete the ones that are not the minimum id for each group of related values:
delete t
from t left join
(select min(id) as id
from (select id, group_concat(name, ':', coalesce(value_double, ''), ':', coalesce(value_timestamp, '' order by name) as pairs,
from t
where name in ('price', 'payment', 'date')
group by id
) tt
group by pairs
) tt
on t.id = tt.id
where tt.id is null;