Group multidimensional array data based on two column values and sum values of one column in each group

Question

I have an array which is created as a combination of two database queries from two separate databases, it looks similar to:

$arr1 = [
    ['part' => '1', 'address' => 'aaa', 'type' => '1', 'count' => 5],
    ['part' => '1', 'address' => 'bbb', 'type' => '1', 'count' => 5],
    ['part' => '1', 'address' => 'ccc', 'type' => '1', 'count' => 5],
    ['part' => '2', 'address' => 'aaa', 'type' => '1', 'count' => 5],
    ['part' => '2', 'address' => 'bbb', 'type' => '1', 'count' => 5],
    ['part' => '2', 'address' => 'ccc', 'type' => '2', 'count' => 5]
];

I am looking for a way to group this array by part and type values. I also need to know the total of the count values as they are grouped.

The results would be something like:

$arr2 = [
    ['part' => '1', 'type' => '1', 'count' => 15],
    ['part' => '2', 'type' => '1', 'count' => 10],
    ['part' => '2', 'type' => '2', 'count' => 5]
];

but I just can't see how to do this. I have seen a few examples of grouping by a single key/value, but not by multiple values at once.

Answer 1

This function should do the job.

function groupByPartAndType($input) {
  $output = Array();

  foreach($input as $value) {
    $output_element = &$output[$value['part'] . "_" . $value['type']];
    $output_element['part'] = $value['part'];
    $output_element['type'] = $value['type'];
    !isset($output_element['count']) && $output_element['count'] = 0;
    $output_element['count'] += $value['count'];
  }

  return array_values($output);
}

If both databases are on the same database server you would be able to do this using SQLs GROUP BY feature.

Answer 2

The following:

$arr2 = array();
foreach ($arr1 as $a) {

  unset($a['address']);
  $key = $a['type'] . '-' . $a['part'];

  if (isset($arr2[$key])) {
    $arr2[$key]['count'] += $a['count'];
  } else {
    $arr2[$key] = $a;
  }

}
$arr2 = array_values($arr2);

Would output

array
  0 => 
    array
      'part' => string '1' (length=1)
      'type' => string '1' (length=1)
      'count' => int 15
  1 => 
    array
      'part' => string '2' (length=1)
      'type' => string '1' (length=1)
      'count' => int 10
  2 => 
    array
      'part' => string '2' (length=1)
      'type' => string '2' (length=1)
      'count' => int 5

Answer 3

Something like

 $newarr=array();
 foreach ( $arr as $Key => $Value ) {
 $newarr[$Value[part]][]=$arr[$key];
 }


 foreach ( $newarr[part] as $Key => $Value ) {
 ...
 }

Answer 4

Full answer for multi-keys arrays grouping is

// *    $arr - associative multi keys data array
// *    $group_by_fields - array of fields to group by
// *    $sum_by_fields - array of fields to calculate sum in group

function array_group_by($arr, $group_by_fields = false, $sum_by_fields = false) {
    if ( empty($group_by_fields) ) return; // * nothing to group

    $fld_count = 'grp:count'; // * field for count of grouped records in each record group

    // * format sum by
    if (!empty($sum_by_fields) && !is_array($sum_by_fields)) {
        $sum_by_fields = array($sum_by_fields);
    }

    // * protected  from collecting
    $fields_collected = array();

    // * do
    $out = array();
    foreach($arr as $value) {
        $newval = array();
        $key = '';
        foreach ($group_by_fields as $field) {
            $key .= $value[$field].'_';
            $newval[$field] = $value[$field];
            unset($value[$field]);
        }
        // * format key
        $key = substr($key,0,-1);

        // * count
        if (isset($out[$key])) { // * record already exists
            $out[$key][$fld_count]++;
        } else {
            $out[$key] = $newval;
            $out[$key][$fld_count]=1;
        }
        $newval = $out[$key];

        // * sum by
        if (!empty($sum_by_fields)) {
            foreach ($sum_by_fields as $sum_field) {
                if (!isset($newval[$sum_field])) $newval[$sum_field] = 0;
                $newval[$sum_field] += $value[$sum_field];
                unset($value[$sum_field]);
            }
        }

        // * collect differencies
        if (!empty($value))
            foreach ($value as $field=>$v) if (!is_null($v)) {
                if (!is_array($v)) {
                    $newval[$field][$v] = $v;
                } else $newval[$field][join('_', $v)] = $v; // * array values 
            }

        $out[$key] = $newval;
    }
    return array_values($out);
}

Answer 5

If this task was necessary in one of my projects, I would craft a snippet that would not need reference variables or any iterated function calls.

Inside the loop, declare the composite temporary key as a variable (since it is used more than once). Push the new row into the result array using the composite key as the temporary first-level key.

Use the null coalescing operator to use the pre-existing count for a given group or zero if the group has not yet been encountered. Then add the new count value to the previously accumulated count .

This technique will unconditionally overwrite the encountered group every time it is repeated. In doing so, the data will be updated with the correct part , type , and count values throughout the iterative process.

When the loop finishes, re-index the result array by calling array_values() .

Code: ( Demo )

$arr1 = [
    ['part' => '1', 'address' => 'aaa', 'type' => '1', 'count' => 5],
    ['part' => '1', 'address' => 'bbb', 'type' => '1', 'count' => 5],
    ['part' => '1', 'address' => 'ccc', 'type' => '1', 'count' => 5],
    ['part' => '2', 'address' => 'aaa', 'type' => '1', 'count' => 5],
    ['part' => '2', 'address' => 'bbb', 'type' => '1', 'count' => 5],
    ['part' => '2', 'address' => 'ccc', 'type' => '2', 'count' => 5]
];

$result = [];
foreach ($arr1 as $row) {
    $compositeKey = $row['part'] . '-' . $row['type'];
    $result[$compositeKey] = [
        'part' => $row['part'],
        'type' => $row['type'],
        'count' => ($result[$compositeKey]['count'] ?? 0) + $row['count']
    ];
}
var_export(array_values($result));

Output:

array (
  0 => 
  array (
    'part' => '1',
    'type' => '1',
    'count' => 15,
  ),
  1 => 
  array (
    'part' => '2',
    'type' => '1',
    'count' => 10,
  ),
  2 => 
  array (
    'part' => '2',
    'type' => '2',
    'count' => 5,
  ),
)

ps Ideally, this task probably could/should be performed in the sql but we don't have the details to provide any specific guidance.