Skip to main content
Version: 3.x

Import action

Overview​

Filament v3.1 introduced a prebuilt action that is able to import rows from a CSV. When the trigger button is clicked, a modal asks the user for a file. Once they upload one, they are able to map each column in the CSV to a real column in the database. If any rows fail validation, they will be compiled into a downloadable CSV for the user to review after the rest of the rows have been imported. Users can also download an example CSV file containing all the columns that can be imported.

This feature uses job batches and database notifications, so you need to publish those migrations from Laravel. Also, you need to publish the migrations for tables that Filament uses to store information about imports:

php artisan queue:batches-table
php artisan notifications:table
php artisan vendor:publish --tag=filament-actions-migrations

php artisan migrate

If you're using PostgreSQL, make sure that the data column in the notifications migration is using json(): $table->json('data').

If you're using UUIDs for your User model, make sure that your notifiable column in the notifications migration is using uuidMorphs(): $table->uuidMorphs('notifiable').

You may use the ImportAction like so:

use App\Filament\Imports\ProductImporter;
use Filament\Actions\ImportAction;

ImportAction::make()
    ->importer(ProductImporter::class)

If you want to add this action to the header of a table instead, you can use Filament\Tables\Actions\ImportAction:

use App\Filament\Imports\ProductImporter;
use Filament\Tables\Actions\ImportAction;
use Filament\Tables\Table;

public function table(Table $table): Table
{
    return $table
        ->headerActions([
            ImportAction::make()
                ->importer(ProductImporter::class)
        ]);
}

The "importer" class needs to be created to tell Filament how to import each row of the CSV.

If you have more than one ImportAction in the same place, you should give each a unique name in the make() method:

ImportAction::make('importProducts')
    ->importer(ProductImporter::class)

ImportAction::make('importBrands')
    ->importer(BrandImporter::class)

Creating an importer​

To create an importer class for a model, you may use the make:filament-importer command, passing the name of a model:

php artisan make:filament-importer Product

This will create a new class in the app/Filament/Imports directory. You now need to define the columns that can be imported.

Automatically generating importer columns​

If you'd like to save time, Filament can automatically generate the columns for you, based on your model's database columns, using --generate:

php artisan make:filament-importer Product --generate

Defining importer columns​

To define the columns that can be imported, you need to override the getColumns() method on your importer class, returning an array of ImportColumn objects:

use Filament\Actions\Imports\ImportColumn;

public function getColumns(): array
{
    return [
        ImportColumn::make('name')
            ->requiredMapping()
            ->rules(['required', 'max:255']),
        ImportColumn::make('sku')
            ->label('SKU')
            ->requiredMapping()
            ->rules(['required', 'max:32']),
        ImportColumn::make('price')
            ->numeric()
            ->rules(['numeric', 'min:0']),
    ];
}

Customizing the label of an import column​

The label for each column will be generated automatically from its name, but you can override it by calling the label() method:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('sku')
    ->label('SKU')

Requiring an importer column to be mapped to a CSV column​

You can call the requiredMapping() method to make a column required to be mapped to a column in the CSV. Columns that are required in the database should be required to be mapped:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('sku')
    ->requiredMapping()

If you require a column in the database, you also need to make sure that it has a rules(['required']) validation rule.

Validating CSV data​

You can call the rules() method to add validation rules to a column. These rules will check the data in each row from the CSV before it is saved to the database:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('sku')
    ->rules(['required', 'max:32'])

Any rows that do not pass validation will not be imported. Instead, they will be compiled into a new CSV of "failed rows", which the user can download after the import has finished. The user will be shown a list of validation errors for each row that failed.

Casting state​

Before validation, data from the CSV can be cast. This is useful for converting strings into the correct data type, otherwise validation may fail. For example, if you have a price column in your CSV, you may want to cast it to a float:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('price')
    ->castStateUsing(function (string $state): ?float {
        if (blank($state)) {
            return null;
        }
        
        $state = preg_replace('/[^0-9.]/', '', $state);
        $state = floatval($state);
    
        return round($state, precision: 2);
    })

In this example, we pass in a function that is used to cast the $state. This function removes any non-numeric characters from the string, casts it to a float, and rounds it to two decimal places.

Please note: if a column is not required by validation, and it is empty, it will not be cast.

Filament also ships with some built-in casting methods:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('price')
    ->numeric() // Casts the state to a float.

ImportColumn::make('price')
    ->numeric(decimalPlaces: 2) // Casts the state to a float, and rounds it to 2 decimal places.

ImportColumn::make('quantity')
    ->integer() // Casts the state to an integer.

ImportColumn::make('is_visible')
    ->boolean() // Casts the state to a boolean.

Mutating the state after it has been cast​

If you're using a built-in casting method or array cast, you can mutate the state after it has been cast by passing a function to the castStateUsing() method:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('price')
    ->numeric()
    ->castStateUsing(function (float $state): ?float {
        if (blank($state)) {
            return null;
        }
    
        return round($state * 100);
    })

You can even access the original state before it was cast, by defining an $originalState argument in the function:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('price')
    ->numeric()
    ->castStateUsing(function (float $state, mixed $originalState): ?float {
        // ...
    })

Importing relationships​

You may use the relationship() method to import a relationship. At the moment, only BelongsTo relationships are supported. For example, if you have a category column in your CSV, you may want to import the category relationship:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('author')
    ->relationship()

In this example, the author column in the CSV will be mapped to the author_id column in the database. The CSV should contain the primary keys of authors, usually id.

If the column has a value, but the author cannot be found, the import will fail validation. Filament automatically adds validation to all relationship columns, to ensure that the relationship is not empty when it is required.

Customizing the relationship import resolution​

If you want to find a related record using a different column, you can pass the column name as resolveUsing:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('author')
    ->relationship(resolveUsing: 'email')

You can pass in multiple columns to resolveUsing, and they will be used to find the author, in an "or" fashion. For example, if you pass in ['email', 'username'], the record can be found by either their email or username:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('author')
    ->relationship(resolveUsing: ['email', 'username'])

You can also customize the resolution process, by passing in a function to resolveUsing, which should return a record to associate with the relationship:

use App\Models\Author;
use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('author')
    ->relationship(resolveUsing: function (array $state): ?Author {
        return Author::query()
            ->where('email', $state)
            ->orWhere('username', $state)
            ->first();
    })

You could even use this function to dynamically determine which columns to use to resolve the record:

use App\Models\Author;
use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('author')
    ->relationship(resolveUsing: function (array $state): ?Author {
        if (filter_var($state, FILTER_VALIDATE_EMAIL)) {
            return 'email';
        }
    
        return 'username';
    })

Handling multiple values in a single column as an array​

You may use the array() method to cast the values in a column to an array. It accepts a delimiter as its first argument, which is used to split the values in the column into an array. For example, if you have a documentation_urls column in your CSV, you may want to cast it to an array of URLs:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('documentation_urls')
    ->array(',')

In this example, we pass in a comma as the delimiter, so the values in the column will be split by commas, and cast to an array.

Casting each item in an array​

If you want to cast each item in the array to a different data type, you can chain the built-in casting methods:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('customer_ratings')
    ->array(',')
    ->integer() // Casts each item in the array to an integer.

Validating each item in an array​

If you want to validate each item in the array, you can chain the nestedRecursiveRules() method:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('customer_ratings')
    ->array(',')
    ->integer()
    ->rules(['array'])
    ->nestedRecursiveRules(['integer', 'min:1', 'max:5'])

Customizing how a column is filled into a record​

If you want to customize how column state is filled into a record, you can pass a function to the fillRecordUsing() method:

use App\Models\Product;

ImportColumn::make('sku')
    ->fillRecordUsing(function (Product $record, string $state): void {
        $record->sku = strtoupper($state);
    })

Adding helper text below the import column​

Sometimes, you may wish to provide extra information for the user before validation. You can do this by adding helperText() to a column, which gets displayed below the mapping select:

use Filament\Forms\Components\TextInput;

ImportColumn::make('skus')
    ->array(',')
    ->helperText('A comma-separated list of SKUs.')

Updating existing records when importing​

When generating an importer class, you will see this resolveRecord() method:

use App\Models\Product;

public function resolveRecord(): ?Product
{
    // return Product::firstOrNew([
    //     // Update existing records, matching them by `$this->data['column_name']`
    //     'email' => $this->data['email'],
    // ]);

    return new Product();
}

This method is called for each row in the CSV, and is responsible for returning a model instance that will be filled with the data from the CSV, and saved to the database. By default, it will create a new record for each row. However, you can customize this behavior to update existing records instead. For example, you might want to update a product if it already exists, and create a new one if it doesn't. To do this, you can uncomment the firstOrNew() line, and pass the column name that you want to match on. For a product, we might want to match on the sku column:

use App\Models\Product;

public function resolveRecord(): ?Product
{
    return Product::firstOrNew([
        'sku' => $this->data['sku'],
    ]);
}

Updating existing records when importing only​

If you want to write an importer that only updates existing records, and does not create new ones, you can return null if no record is found:

use App\Models\Product;

public function resolveRecord(): ?Product
{
    return Product::query()
        ->where('sku', $this->data['sku'])
        ->first();
}

If you'd like to fail the import row if no record is found, you can throw a RowImportFailedException with a message:

use App\Models\Product;
use Filament\Actions\Imports\Exceptions\RowImportFailedException;

public function resolveRecord(): ?Product
{
    $product = Product::query()
        ->where('sku', $this->data['sku'])
        ->first();

    if (! $product) {
        throw new RowImportFailedException("No product found with SKU [{$this->data['sku']}].");
    }

    return $product;
}

When the import is completed, the user will be able to download a CSV of failed rows, which will contain the error messages.

Ignoring blank state for an import column​

By default, if a column in the CSV is blank, and mapped by the user, and it's not required by validation, the column will be imported as null in the database. If you'd like to ignore blank state, and use the existing value in the database instead, you can call the ignoreBlankState() method:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('price')
    ->ignoreBlankState()

Using import options​

The import action can render extra form components that the user can interact with when importing a CSV. This can be useful to allow the user to customize the behavior of the importer. For instance, you might want a user to be able to choose whether to update existing records when importing, or only create new ones. To do this, you can return options form components from the getOptionsFormComponents() method on your importer class:

use Filament\Forms\Components\Checkbox;

public static function getOptionsFormComponents(): array
{
    return [
        Checkbox::make('updateExisting')
            ->label('Update existing records'),
    ];
}

Alternatively, you can pass a set of static options to the importer through the options() method on the action:

use Filament\Actions\ImportAction;

ImportAction::make()
    ->importer(ProductImporter::class)
    ->options([
        'updateExisting' => true,
    ])

Now, you can access the data from these options inside the importer class, by calling $this->options. For example, you might want to use it inside resolveRecord() to update an existing product:

use App\Models\Product;

public function resolveRecord(): ?Product
{
    if ($this->options['updateExisting'] ?? false) {
        return Product::firstOrNew([
            'sku' => $this->data['sku'],
        ]);
    }

    return new Product();
}

Improving import column mapping guesses​

By default, Filament will attempt to "guess" which columns in the CSV match which columns in the database, to save the user time. It does this by attempting to find different combinations of the column name, with spaces, -, _, all cases insensitively. However, if you'd like to improve the guesses, you can call the guess() method with more examples of the column name that could be present in the CSV:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('sku')
    ->guess(['id', 'number', 'stock-keeping unit'])

Providing example CSV data​

Before the user uploads a CSV, they have an option to download an example CSV file, containing all the available columns that can be imported. This is useful, as it allows the user to import this file directly into their spreadsheet software, and fill it out.

You can also add an example row to the CSV, to show the user what the data should look like. To fill in this example row, you can pass in an example column value to the example() method:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('sku')
    ->example('ABC123')

By default, the name of the column is used in the header of the example CSV. You can customize the header per-column using exampleHeader():

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('sku')
    ->exampleHeader('SKU')

Using a custom user model​

By default, the imports table has a user_id column. That column is constrained to the users table:

$table->foreignId('user_id')->constrained()->cascadeOnDelete();

In the Import model, the user() relationship is defined as a BelongsTo relationship to the App\Models\User model. If the App\Models\User model does not exist, or you want to use a different one, you can bind a new Authenticatable model to the container in a service provider's register() method:

use App\Models\Admin;
use Illuminate\Contracts\Auth\Authenticatable;

$this->app->bind(Authenticatable::class, Admin::class);

If your authenticatable model uses a different table to users, you should pass that table name to constrained():

$table->foreignId('user_id')->constrained('admins')->cascadeOnDelete();

Using a polymorphic user relationship​

If you want to associate imports with multiple user models, you can use a polymorphic MorphTo relationship instead. To do this, you need to replace the user_id column in the imports table:

$table->morphs('user');

Then, in a service provider's boot() method, you should call Import::polymorphicUserRelationship() to swap the user() relationship on the Import model to a MorphTo relationship:

use Filament\Actions\Imports\Models\Import;

Import::polymorphicUserRelationship();

Limiting the maximum number of rows that can be imported​

To prevent server overload, you may wish to limit the maximum number of rows that can be imported from one CSV file. You can do this by calling the maxRows() method on the action:

ImportAction::make()
    ->importer(ProductImporter::class)
    ->maxRows(100000)

Changing the import chunk size​

Filament will chunk the CSV, and process each chunk in a different queued job. By default, chunks are 100 rows at a time. You can change this by calling the chunkSize() method on the action:

ImportAction::make()
    ->importer(ProductImporter::class)
    ->chunkSize(250)

If you are encountering memory or timeout issues when importing large CSV files, you may wish to reduce the chunk size.

Changing the CSV delimiter​

The default delimiter for CSVs is the comma (,). If your import uses a different delimiter, you may call the csvDelimiter() method on the action, passing a new one:

ImportAction::make()
    ->importer(ProductImporter::class)
    ->csvDelimiter(';')

You can only specify a single character, otherwise an exception will be thrown.

Changing the column header offset​

If your column headers are not on the first row of the CSV, you can call the headerOffset() method on the action, passing the number of rows to skip:

ImportAction::make()
    ->importer(ProductImporter::class)
    ->headerOffset(5)

Customizing the import job​

The default job for processing imports is Filament\Actions\Imports\Jobs\ImportCsv. If you want to extend this class and override any of its methods, you may replace the original class in the register() method of a service provider:

use App\Jobs\ImportCsv;
use Filament\Actions\Imports\Jobs\ImportCsv as BaseImportCsv;

$this->app->bind(BaseImportCsv::class, ImportCsv::class);

Or, you can pass the new job class to the job() method on the action, to customize the job for a specific import:

use App\Jobs\ImportCsv;

ImportAction::make()
    ->importer(ProductImporter::class)
    ->job(ImportCsv::class)

Customizing the import queue and connection​

By default, the import system will use the default queue and connection. If you'd like to customize the queue used for jobs of a certain importer, you may override the getJobQueue() method in your importer class:

public function getJobQueue(): ?string
{
    return 'imports';
}

You can also customize the connection used for jobs of a certain importer, by overriding the getJobConnection() method in your importer class:

public function getJobConnection(): ?string
{
    return 'sqs';
}

Customizing the import job middleware​

By default, the import system will only process one job at a time from each import. This is to prevent the server from being overloaded, and other jobs from being delayed by large imports. That functionality is defined in the WithoutOverlapping middleware on the importer class:

public function getJobMiddleware(): array
{
    return [
        (new WithoutOverlapping("import{$this->import->getKey()}"))->expireAfter(600),
    ];
}

If you'd like to customize the middleware that is applied to jobs of a certain importer, you may override this method in your importer class. You can read more about job middleware in the Laravel docs.

Customizing the import job retries​

By default, the import system will retry a job for 24 hours. This is to allow for temporary issues, such as the database being unavailable, to be resolved. That functionality is defined in the getJobRetryUntil() method on the importer class:

use Carbon\CarbonInterface;

public function getJobRetryUntil(): ?CarbonInterface
{
    return now()->addDay();
}

If you'd like to customize the retry time for jobs of a certain importer, you may override this method in your importer class. You can read more about job retries in the Laravel docs.

Customizing the import job tags​

By default, the import system will tag each job with the ID of the import. This is to allow you to easily find all jobs related to a certain import. That functionality is defined in the getJobTags() method on the importer class:

public function getJobTags(): array
{
    return ["import{$this->import->getKey()}"];
}

If you'd like to customize the tags that are applied to jobs of a certain importer, you may override this method in your importer class.

Customizing the import job batch name​

By default, the import system doesn't define any name for the job batches. If you'd like to customize the name that is applied to job batches of a certain importer, you may override the getJobBatchName() method in your importer class:

public function getJobBatchName(): ?string
{
    return 'product-import';
}

Customizing import validation messages​

The import system will automatically validate the CSV file before it is imported. If there are any errors, the user will be shown a list of them, and the import will not be processed. If you'd like to override any default validation messages, you may do so by overriding the getValidationMessages() method on your importer class:

public function getValidationMessages(): array
{
    return [
        'name.required' => 'The name column must not be empty.',
    ];
}

To learn more about customizing validation messages, read the Laravel docs.

Customizing import validation attributes​

When columns fail validation, their label is used in the error message. To customize the label used in field error messages, use the validationAttribute() method:

use Filament\Actions\Imports\ImportColumn;

ImportColumn::make('name')
    ->validationAttribute('full name')

Lifecycle hooks​

Hooks may be used to execute code at various points within an importer's lifecycle, like before a record is saved. To set up a hook, create a protected method on the importer class with the name of the hook:

protected function beforeSave(): void
{
    // ...
}

In this example, the code in the beforeSave() method will be called before the validated data from the CSV is saved to the database.

There are several available hooks for importers:

use Filament\Actions\Imports\Importer;

class ProductImporter extends Importer
{
    // ...

    protected function beforeValidate(): void
    {
        // Runs before the CSV data for a row is validated.
    }

    protected function afterValidate(): void
    {
        // Runs after the CSV data for a row is validated.
    }

    protected function beforeFill(): void
    {
        // Runs before the validated CSV data for a row is filled into a model instance.
    }

    protected function afterFill(): void
    {
        // Runs after the validated CSV data for a row is filled into a model instance.
    }

    protected function beforeSave(): void
    {
        // Runs before a record is saved to the database.
    }

    protected function beforeCreate(): void
    {
        // Similar to `beforeSave()`, but only runs when creating a new record.
    }

    protected function beforeUpdate(): void
    {
        // Similar to `beforeSave()`, but only runs when updating an existing record.
    }

    protected function afterSave(): void
    {
        // Runs after a record is saved to the database.
    }
    
    protected function afterCreate(): void
    {
        // Similar to `afterSave()`, but only runs when creating a new record.
    }
    
    protected function afterUpdate(): void
    {
        // Similar to `afterSave()`, but only runs when updating an existing record.
    }
}

Inside these hooks, you can access the current row's data using $this->data. You can also access the original row of data from the CSV, before it was cast or mapped, using $this->originalData.

The current record (if it exists yet) is accessible in $this->record, and the import form options using $this->options.

Authorization​

By default, only the user who started the import may access the failure CSV file that gets generated if part of an import fails. If you'd like to customize the authorization logic, you may create an ImportPolicy class, and register it in your AuthServiceProvider:

use App\Policies\ImportPolicy;
use Filament\Actions\Imports\Models\Import;

protected $policies = [
    Import::class => ImportPolicy::class,
];

The view() method of the policy will be used to authorize access to the failure CSV file.

Please note that if you define a policy, the existing logic of ensuring only the user who started the import can access the failure CSV file will be removed. You will need to add that logic to your policy if you want to keep it:

use App\Models\User;
use Filament\Actions\Imports\Models\Import;

public function view(User $user, Import $import): bool
{
    return $import->user()->is($user);
}