Infection: RFC: leave only one element in the non empty returned array

[maks-rafalko]

For the following method

public function getSomeCollection(): array
{
    // ...

    return $collection;
}

we can return a mutated collection with only one element instead of N.

- return $collection;
+ return \count($collection) > 1 ? [array_keys($collection)[0] => array_values($collection)[0]] : $collection;

Why [array_keys($c)[0] => array_values($c)[0]]?

Because array can be indexed by string keys, by IDs, so not just by indexes 0..N.

With such code

$collection = [33 => $userObject1, 44 => $userObject2];

[array_keys($collection)[0] => array_values($collection)[0]] will be evaluated to [33 => $userObject1];

and

$collection = ['abc'=> 1, 'def' => 2];

will be evaluated to ['abc'=> 1];

What do you think?

[BackEndTea]

I think this could be a decent mutator, but it may be too complex of a mutation for it to make sense to users.

If i see the following diff, then i don't really know what is being mutated. So far, all diffs make it pretty clear what the 'new' behaviour of the code is.

- return $collection;
+ return \count($collection) > 1 ? [array_keys($collection)[0] => array_values($collection)[0]] : $collection;

So personally i'd say no to this mutator.

[localheinz]

How about using array_slice() then?

[maks-rafalko]

array_slice() with preserve_keys=true should do the job, thank you @localheinz 👍

array_slice(['abc'=> 1, 'def' => 2], 0, 1, true) // ['abc' => 1]

array_slice([33 => $userObject1, 44 => $userObject2], 0, 1, true) // [33 => $userObject1]

So the diff will be:

- return $collection;
+ return \count($collection) > 1 ? array_slice($collection, 0, 1, true) : $collection;