Fetch extra data for large number of IDs - php

Question

I have a MySQL table containing information about objects. It looks like this:
+--------------+----+--------+-------+
| auto_incr_id | id | type | value |
+--------------+----+--------+-------+
| 1 | 1 | length | 105 |
| 2 | 1 | weight | 17 |
| 3 | 1 | price | 104 |
| 4 | 2 | length | 111 |
| 5 | 2 | weight | 18 |
| 6 | 2 | price | 87 |
+--------------+----+--------+-------+
I want to fetch the first x objects, sorted by length:
{
"id": 2,
"length": 111,
"weight": 18,
"price": 87
},
{
"id": 1,
"length": 105,
"weight": 17,
"price": 104
}
Here's what I do to achieve this:
First, I fetch the ids:
$type = "length";
$stmt = $mysqli->prepare("SELECT id FROM table WHERE type=? ORDER BY value DESC LIMIT ?");
$stmt->bind_param('si', $type, $limit);
$stmt->execute();
$result = $stmt->get_result();
$result = $result->fetch_all(MYSQLI_ASSOC);
Next, I get the other values for each object that was fetched:
$i = 0;
while ($i < count($result)) {
$stmt = $mysqli->prepare("SELECT type, value FROM table WHERE id=?");
$stmt->bind_param('i', $result[$i]['id']);
$stmt->execute();
$result_2 = $stmt->get_result();
$fetch = $result_2->fetch_all(MYSQLI_ASSOC);
$j = 0;
while ($j < count($fetch))
{
$result[$i][$fetch[$j]['type']] = $fetch[$j]['value'];
$j++;
}
}
This works great to fetch for example the top 5 ($limit = 5), but now I have a use case where I want to have this information for the top 10k or more. This take too much time, probably because it needs to go through the while loop and execute a statement for every found ID.
So I'm looking for a way to reduce the execution time getting this data. Any ideas?

Answer 1

Your issue appears to be a lack of understanding about table joins. There is a wrong and a right way to do this. The wrong way is very easy to understand, so I will show it so it makes sense...
select id, type, value from table where id in (select id from table where type='length');
That obviously gets the type/value combination from every id that has a value for type='length'. But, you aren't doing it ALL in the database. You still have to sort and group everything outside the database. Let's assume you only have type='length', 'weight' and 'price'. I can create three tables on the fly for that, creating attributes of length, weight, and price...
select l.id, length, weight, price from
(select id, value as length from table where type='length') l
left outer join
(select id, value as weight from table where type='weight') w
on l.id=w.id
left outer join
(select id, value as price from table where type='price') p
on l.id=p.id
order by length
Now, you will get each attribute in a row, one row per id. The id field is not guaranteed to be unique here. If you have the same ID with more than one length, it will show up more than once in the results. But, I hope you can see how I took the obvious subquery and turned it into a join to do all the work inside the database.
NOTE: I fixed the ambiguity error and added "left outer" to include rows where weight and price do not exist.

Answer 2

Well at least you can always use a subquery
SELECT id, type, value FROM table
WHERE id IN (SELECT id FROM table WHERE type=? ORDER BY value DESC LIMIT ?")
order by id