php - Why don’t the new class-based Laravel Eloquent accessors apply with `->pluck('name')` as they did w

admin管理员组

文章数量:1129201

I’m working on a Laravel (and FilamentPHP) project and noticed a discrepancy between the old “magic” accessor method and the new class-based Attribute accessor introduced in Laravel 8.40+. Specifically, when I use:

Model::query()->pluck('name', 'id')

• The old magic accessor (getNameAttribute()) applied ucwords() to the plucked values.
• The new class-based accessor (protected function name(): Attribute) does not apply the transformation when plucking.

Here’s a simplified example of my model:

<?php

namespace App\Models;

use Illuminate\Database\Eloquent\Casts\Attribute;
use Illuminate\Database\Eloquent\Builder;
use Illuminate\Database\Eloquent\Model;
use App\Enums\SomeEnumStatusState;

class MyModel extends Model
{
    // OLD Magic Accessor (works with pluck)
    // public function getNameAttribute($value)
    // {
    //     return ucwords(strtolower($value));
    // }

    // NEW Class-based Accessor (not applying with pluck)
    protected function name(): Attribute
    {
        return Attribute::make(
            get: fn ($value) => ucwords(strtolower($value))
        );
    }

    public function scopeOpen(Builder $query): Builder
    {
        return $query->where('state', SomeEnumStatusState::OPEN);
    }

    protected function casts(): array
    {
        return [
            'state' => SomeEnumStatusState::class,
        ];
    }
}

And in my FilamentPHP Resource:

Forms\Components\Select::make('model_id')
    ->label('Status')
    ->relationship('modelRelationship', 'name')
    ->options(MyModel::query()->pluck('name', 'id')),

My questions are:

1. Why does Model::query()->pluck('name', 'id') return the raw database values when using the new class-based name(): Attribute but returns transformed values under the old getNameAttribute() method?

2. Is this the intended behavior, or am I missing something about how pluck() interacts with the newer Attribute accessors?

3. What is the recommended or “best practice” way to retrieve transformed attribute values (especially with FilamentPHP) under the new accessor approach?

Any insights, explanations, or code examples showing how to ensure pluck() respects the class-based accessor would be greatly appreciated!

I’m working on a Laravel (and FilamentPHP) project and noticed a discrepancy between the old “magic” accessor method and the new class-based Attribute accessor introduced in Laravel 8.40+. Specifically, when I use:

Model::query()->pluck('name', 'id')

• The old magic accessor (getNameAttribute()) applied ucwords() to the plucked values.
• The new class-based accessor (protected function name(): Attribute) does not apply the transformation when plucking.

Here’s a simplified example of my model:

<?php

namespace App\Models;

use Illuminate\Database\Eloquent\Casts\Attribute;
use Illuminate\Database\Eloquent\Builder;
use Illuminate\Database\Eloquent\Model;
use App\Enums\SomeEnumStatusState;

class MyModel extends Model
{
    // OLD Magic Accessor (works with pluck)
    // public function getNameAttribute($value)
    // {
    //     return ucwords(strtolower($value));
    // }

    // NEW Class-based Accessor (not applying with pluck)
    protected function name(): Attribute
    {
        return Attribute::make(
            get: fn ($value) => ucwords(strtolower($value))
        );
    }

    public function scopeOpen(Builder $query): Builder
    {
        return $query->where('state', SomeEnumStatusState::OPEN);
    }

    protected function casts(): array
    {
        return [
            'state' => SomeEnumStatusState::class,
        ];
    }
}

And in my FilamentPHP Resource:

Forms\Components\Select::make('model_id')
    ->label('Status')
    ->relationship('modelRelationship', 'name')
    ->options(MyModel::query()->pluck('name', 'id')),

My questions are:

1. Why does Model::query()->pluck('name', 'id') return the raw database values when using the new class-based name(): Attribute but returns transformed values under the old getNameAttribute() method?

2. Is this the intended behavior, or am I missing something about how pluck() interacts with the newer Attribute accessors?

3. What is the recommended or “best practice” way to retrieve transformed attribute values (especially with FilamentPHP) under the new accessor approach?

Any insights, explanations, or code examples showing how to ensure pluck() respects the class-based accessor would be greatly appreciated!

• php
• laravel
• eloquent
• laravel-filament
• filamentphp

Share Improve this question Follow asked Jan 8 at 10:38 Deon OkonkwoDeon Okonkwo 16522 silver badges1010 bronze badges 2

• I think it makes more sense that plucking from a query returns the raw DB values, since it does not create models but a generic array – kris gjika Commented Jan 8 at 11:14
• you can likely get what you want via MyModel::query()->select('name', 'id')->get()->pluck('name', 'id') – kris gjika Commented Jan 8 at 11:16

Add a comment  | 

2 Answers 2

Sorted by: Reset to default 1

Looking into the vendor files for this, the pluck method respects mutators as follows:

        // If the model has a mutator for the requested column, we will spin through
        // the results and mutate the values so that the mutated version of these
        // columns are returned as you would expect from these Eloquent models.
        if (! $this->model->hasGetMutator($column) &&
            ! $this->model->hasCast($column) &&
            ! in_array($column, $this->model->getDates())) {
            return $results;
        }

Taking a dive deeper into the hasGetMutator method we can see that this will only check for method names using the previous magic methods.

    public function hasGetMutator($key)
    {
        return method_exists($this, 'get'.Str::studly($key).'Attribute');
    }

One way i've found to get around this would be to specify the attribute on the $appends array (relevant docs) on the model and then ->get() before ->pluck(...) like so:

Model::query()->get()->pluck('name', 'id')

AFAIK, pluck doesn't access your model but just gets the columns (and only those) directly from the database. This has a significant performance advantage, since much less data is retrieved from the database, and Laravel doesn't need to construct the model objects. The disadvantage however, is that you get only the literal database values which may or may not suffice your needs.

Hence when using pluck on the query, your model isn't constructed/booted and thus your accessor isn't used. If you want to retrieve those attribute values, you can do something like

MyModel::all()->map(fn (MyModel $record) => [$record->id, $record->name])

This retrieves all records of this model from the database, constructs all the models and puts it in a collection and then gets the id and name.

本文标签：