Grouping and aggregation

Home >> Examples >> Grouping and aggregation

Build reports — totals per category, user activity summaries, lists bucketed by status — by composing groupBy, map, and fold. Replace multi-pass code with one declarative pipeline.

Functions used

The scenario

A list of payment records. We want three reports from the same data:

  1. Total revenue per currency.
  2. Count of payments per status, per customer.
  3. Payments split into “refunded” vs “settled”.
$payments = [
    ['customer' => 'A', 'currency' => 'GBP', 'amount' => 100, 'status' => 'settled'],
    ['customer' => 'A', 'currency' => 'EUR', 'amount' => 50,  'status' => 'settled'],
    ['customer' => 'B', 'currency' => 'GBP', 'amount' => 75,  'status' => 'settled'],
    ['customer' => 'A', 'currency' => 'GBP', 'amount' => 25,  'status' => 'refunded'],
    ['customer' => 'B', 'currency' => 'EUR', 'amount' => 40,  'status' => 'settled'],
];

1. Revenue per currency — groupBy + sumWhere

use PinkCrab\FunctionConstructors\GeneralFunctions as F;
use PinkCrab\FunctionConstructors\Arrays as A;

$byCurrency = A\groupBy(F\getProperty('currency'))($payments);
// ['GBP' => [payment, payment, payment], 'EUR' => [payment, payment]]

$totalInGroup = A\sumWhere(F\getProperty('amount'));

$revenueByCurrency = array_map($totalInGroup, $byCurrency);
// ['GBP' => 200, 'EUR' => 90]

groupBy buckets, sumWhere does “map to amount + sum” in one pass per bucket.

2. Payment counts per status, per customer — nested groupBy

$byCustomer = A\groupBy(F\getProperty('customer'))($payments);

$statusCountsFor = fn(array $rows) =>
    array_map('count', A\groupBy(F\getProperty('status'))($rows));

$summary = array_map($statusCountsFor, $byCustomer);
/*
[
    'A' => ['settled' => 2, 'refunded' => 1],
    'B' => ['settled' => 2],
]
*/

$statusCountsFor is a reusable callable — “for a list of payments, return count per status”. Apply it to every customer’s bucket.

3. Refunded vs settled — partition

[$settled, $refunded] = A\partition(
    fn($p) => $p['status'] === 'refunded'
)($payments);

// $settled  — truthy bucket (index 1)
// $refunded — falsy bucket (index 0)

Wait — partition puts truthy matches at index 1, falsy at 0. So unpacking is [$falsy, $truthy]. Rename to match:

[$notRefunded, $refunded] = A\partition(
    fn($p) => $p['status'] === 'refunded'
)($payments);

One traversal, two buckets.

Custom aggregator — fold

For anything sumWhere doesn’t cover, drop to fold:

$stats = A\fold(
    fn($acc, $p) => [
        'count' => $acc['count'] + 1,
        'total' => $acc['total'] + $p['amount'],
        'max'   => max($acc['max'], $p['amount']),
    ],
    ['count' => 0, 'total' => 0, 'max' => 0]
);

print_r($stats($payments));
// ['count' => 5, 'total' => 290, 'max' => 100]

The pattern

Compose these and you’ve replaced every “two foreach loops and a running total” pattern.