Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:116568
MIME-Version: 1.0
References: <CAL8UiQVrhOOivTL4LX+j-UXe=Scn_JwJ=zQwxga+K9kjWa2weA@mail.gmail.com>
In-Reply-To: <CAL8UiQVrhOOivTL4LX+j-UXe=Scn_JwJ=zQwxga+K9kjWa2weA@mail.gmail.com>
Date: Sun, 5 Dec 2021 07:50:41 +0700
Message-ID: <CAG4FTXwASqT6ZBzv8Bxd7Wct5ONg0Z8TbpJ0PD1Oy85QRBtcoQ@mail.gmail.com>
To: Hassan Ahmed <7snovic@gmail.com>
Cc: php internals <internals@lists.php.net>
Content-Type: text/plain; charset="UTF-8"
Subject: Re: [PHP-DEV] RFC [Discussion] array_column results grouping
From: the.liquid.metal@gmail.com (Hendra Gunawan)

>
> Hello, this is a proposal to upgrade the functionality of
> `array_column` to return an indexed array for all of the matched data
> instead of a single result.
>
> the RFC had been created here
> https://wiki.php.net/rfc/array_column_results_grouping and the PR is
> in here https://github.com/php/php-src/pull/7698.
>
> However, I am not sure, if we need to add a fourth parameter to the
> function to change its behavior or we can add it as a new function,
> something like `array_column_group` or any other name.
>

Hai Hassan.

It seems that your proposal is just save no more than 3 lines compared with
`foreach`. Moreover, `foreach` can do better than your version.

    foreach ($persons as $p) {
        $groups[$p["education"]][$p["sex"]][MORE_SUBGRUP][] = $p;
    }


output:

    $groups = [
        "elementary" => [
            "female" => [
                MORE_SUBGRUP_1 => [
                    0 => [...],
                    1 => [...]
                ],
                MORE_SUBGRUP_2 => [
                    0 => [...],
                    1 => [...]
                ],
                ...
            ],
            "male" => [
                MORE_SUBGRUP_1 => [
                    0 => [...],
                    1 => [...]
                ],
                MORE_SUBGRUP_2 => [
                    0 => [...],
                    1 => [...]
                ],
                ...
            ],
        ],
        "highschool" => [
            ...
        ],
        ...
    ];

my suggestions are:
- don't limit to array => involve object as well
- offer reducer callable as an optional last param.

reducer callable it will transform 2 steps algo (populating the groups and then
reducing) to become a single step. but if it is not provided, the function
returns the member of the group (same as example above).


old way:

    // step-1
    $group = [];
    foreach ($persons as $p) {/*...*/}

    // step-2
    foreach ($group as &$eduGroup) {
        foreach ($eduGroup as &$sexGroup) {
            $sumWeight = 0;
            foreach ($sexGroup as $item) {
                $sumWeight += $item->weight;
            }
            $sexGroup = $sumWeight / count($sexGroup);
        }
    }

output:

    $groups = [
        "elementary" => [
            "female" => 40,
            "male" => 50,
        ],
        "highschool" => [
            "female" => 60,
            "male" => 65,
        ],
        ...
    ];

new way:

    array_column_group(
        $group,
        ["education", "sex", MORE_SUBGRUP],
        fn ($item, $i, $cumm) => ($item->weight + ($cumm * $i)) / ($i + 1)
    );


the signature:

    // array_column_group signature
    array_column_group(
        array|object $array,
        int|string|array $index_key,
        callable $reducer = null
    )

    // callable signature
    fn (array|object $item, int|string $index, mixed $cummulative): mixed

note:
- $index_key can be in the form `["education", ...]` or `"education"`.
- or even `[0, 1, ...]` or `0`.

So virtually, `array_column_group` is a hybrid between `array_column` and
`array_walk` (or other function with a funny name which I don't remember).

Best Regards
Hendra Gunawan.