Group rows in an associative array of associative arrays by column value and preserve the original first level keys

Group rows in an associative array of associative arrays by column value and preserve the original first level keys

I have an array of subarrays in the following format:

[
    'a' => ['id' => 20, 'name' => 'chimpanzee'],
    'b' => ['id' => 40, 'name' => 'meeting'],
    'c' => ['id' => 20, 'name' => 'dynasty'],
    'd' => ['id' => 50, 'name' => 'chocolate'],
    'e' => ['id' => 10, 'name' => 'bananas'],
    'f' => ['id' => 50, 'name' => 'fantasy'],
    'g' => ['id' => 50, 'name' => 'football']
]

And I would like to group it into a new array based on the id field in each subarray.

array
(
    10 => array
          (
            e => array ( id = 10, name = bananas )
          )
    20 => array
          (
            a => array ( id = 20, name = chimpanzee )
            c => array ( id = 20, name = dynasty )
          )
    40 => array
          (
            b => array ( id = 40, name = meeting )
          )
    50 => array
          (
            d => array ( id = 50, name = chocolate )
            f => array ( id = 50, name = fantasy )
            g => array ( id = 50, name = football )
          )
)

$arr = array();

foreach ($old_arr as $key => $item) {
   $arr[$item['id']][$key] = $item;
}

ksort($arr, SORT_NUMERIC);

foreach($array as $key => $value){
   $newarray[$value['id']][$key] = $value;
}

var_dump($newarray);

piece of cake ;)

The following code adapts @Tim Cooper’s code to mitigate Undefined index: id errors in the event that one of the inner arrays doesn’t contain an id:

$arr = array();

foreach($old_arr as $key => $item)
{
    if(array_key_exists('id', $item))
        $arr[$item['id']][$key] = $item;
}

ksort($arr, SORT_NUMERIC);

However, it will drop inner arrays without an id.

E.g.

$old_arr = array(
    'a' => array ( 'id' => 20, 'name' => 'chimpanzee' ),
    'b' => array ( 'id' => 40, 'name' => 'meeting' ),
    'c' => array ( 'id' => 20, 'name' => 'dynasty' ),
    'd' => array ( 'id' => 50, 'name' => 'chocolate' ),
    'e' => array ( 'id' => 10, 'name' => 'bananas' ),
    'f' => array ( 'id' => 50, 'name' => 'fantasy' ),
    'g' => array ( 'id' => 50, 'name' => 'football' ),
    'h' => array ( 'name' => 'bob' )
);

will drop the 'h' array completely.

You can also use Arrays::groupBy() from ouzo-goodies:

$groupBy = Arrays::groupBy($array, Functions::extract()->id);

print_r($groupBy);

And result:

Array
(
    [20] => Array
        (
            [0] => Array
                (
                    [id] => 20
                    [name] => chimpanzee
                )

            [1] => Array
                (
                    [id] => 20
                    [name] => dynasty
                )

        )

    [40] => Array
        (
            [0] => Array
                (
                    [id] => 40
                    [name] => meeting
                )

        )

    [50] => Array
        (
            [0] => Array
                (
                    [id] => 50
                    [name] => chocolate
                )

            [1] => Array
                (
                    [id] => 50
                    [name] => fantasy
                )

            [2] => Array
                (
                    [id] => 50
                    [name] => football
                )

        )

    [10] => Array
        (
            [0] => Array
                (
                    [id] => 10
                    [name] => bananas
                )

        )

)

And here are the docs for Arrays and Functions.

Here is a function that will take an array as the first argument and a criteria (a string or callback function) as the second argument. The function returns a new array that groups the array as asked for.

/**
 * Group items from an array together by some criteria or value.
 *
 * @param  $arr array The array to group items from
 * @param  $criteria string|callable The key to group by or a function the returns a key to group by.
 * @return array
 *
 */
function groupBy($arr, $criteria): array
{
    return array_reduce($arr, function($accumulator, $item) use ($criteria) {
        $key = (is_callable($criteria)) ? $criteria($item) : $item[$criteria];
        if (!array_key_exists($key, $accumulator)) {
            $accumulator[$key] = [];
        }

        array_push($accumulator[$key], $item);
        return $accumulator;
    }, []);
}

Here is the given array:

$arr = array(
    'a' => array ( 'id' => 20, 'name' => 'chimpanzee' ),
    'b' => array ( 'id' => 40, 'name' => 'meeting' ),
    'c' => array ( 'id' => 20, 'name' => 'dynasty' ),
    'd' => array ( 'id' => 50, 'name' => 'chocolate' ),
    'e' => array ( 'id' => 10, 'name' => 'bananas' ),
    'f' => array ( 'id' => 50, 'name' => 'fantasy' ),
    'g' => array ( 'id' => 50, 'name' => 'football' )
);

And examples using the function with a string and a callback function:

$q = groupBy($arr, 'id');
print_r($q);

$r = groupBy($arr, function($item) {
    return $item['id'];
});
print_r($r);

The results are the same in both examples:

Array
(
    [20] => Array
        (
            [0] => Array
                (
                    [id] => 20
                    [name] => chimpanzee
                )

            [1] => Array
                (
                    [id] => 20
                    [name] => dynasty
                )

        )

    [40] => Array
        (
            [0] => Array
                (
                    [id] => 40
                    [name] => meeting
                )

        )

    [50] => Array
        (
            [0] => Array
                (
                    [id] => 50
                    [name] => chocolate
                )

            [1] => Array
                (
                    [id] => 50
                    [name] => fantasy
                )

            [2] => Array
                (
                    [id] => 50
                    [name] => football
                )

        )

    [10] => Array
        (
            [0] => Array
                (
                    [id] => 10
                    [name] => bananas
                )

        )

)

Passing the callback is overkill in the example above, but using the callback finds its use when you pass in an array of objects, a multidimensional array, or have some arbitrary thing you want to group by.

Maybe it's worth to mention that you can also use php array_reduce function

$items = [
    ['id' => 20, 'name' => 'chimpanzee'],
    ['id' => 40, 'name' => 'meeting'],
    ['id' => 20, 'name' => 'dynasty'],
    ['id' => 50, 'name' => 'chocolate'],
    ['id' => 10, 'name' => 'bananas'],
    ['id' => 50, 'name' => 'fantasy'],
    ['id' => 50, 'name' => 'football'],
];

// Grouping
$groupedItems = array_reduce($items, function ($carry, $item) {
    $carry[$item['id']][] = $item;
    return $carry;
}, []);
// Sorting
ksort($groupedItems, SORT_NUMERIC);

print_r($groupedItems);

https://www.php.net/manual/en/function.array-reduce.php

Because of how PHP's sorting algorithm treats multidimensional arrays -- it sorts by size, then compares elements one at a time, you can actually use a key-preserving sort on the input BEFORE restructuring. In functional style programming, this means that you don't need to declare the result array as a variable.

Code: (Demo)

asort($array);
var_export(
    array_reduce(
        array_keys($array),
        function($result, $k) use ($array) {
            $result[$array[$k]['id']][$k] = $array[$k];
            return $result;
        }
    )
);

---

I must say that functional programming is not very attractive for this task because the first level keys must be preserved.

Although array_walk() is more succinct, it still requires the result array to be passed into the closure as a reference variable. (Demo)

asort($array);
$result = [];
array_walk(
    $array,
    function($row, $k) use (&$result) {
        $result[$row['id']][$k] = $row;
    }
);
var_export($result);

---

I'd probably recommend a classic loop for this task. The only thing the loop needs to do is rearrange the first and second level keys. (Demo)

asort($array);
$result = [];
foreach ($array as $k => $row) {
    $result[$row['id']][$k] = $row;
}
var_export($result);

To be completely honest, I expect that ksort() will be more efficient than pre-loop sorting, but I wanted to a viable alternative.

ReferenceURL : https://stackoverflow.com/questions/7574857/group-rows-in-an-associative-array-of-associative-arrays-by-column-value-and-pre