0Pricing
Flutter Mobile Development · レッスン

riverpod_generatorと@riverpodによるコード生成

riverpod_generatorのアノテーションを使い、ボイラープレートなしで型安全なプロバイダーを生成します。

「riverpod_generatorと@riverpodによるコード生成」はCoddyKit上の無料Flutter Mobile Developmentレッスンです。 これはレッスン2/4です。 下記で完全なレッスンを無料で読むことができます。その後、ブラウザ内の組み込みコードエディタと24時間対応のAIチューターでハンズオン演習できます。 これはFlutter Mobile Development学習パスの一部であり、ウェブとCoddyKitアプリ全体で進捗が同期されます。 Flutter Mobile Developmentコースには全4レッスンが含まれています。

このレッスンの一部はまだ翻訳されておらず、英語で表示されています。

Why Code Generation?

Before Riverpod 2.0, you picked the right provider type by hand: Provider, StateProvider, FutureProvider, StreamProvider, NotifierProvider, and so on. Choosing wrong meant rewrites.

The riverpod_generator package flips this around. You write a plain function or class and add the @riverpod annotation. The generator inspects your return type and produces the correct, fully type-safe provider for you.

  • Less boilerplate — no manual provider declarations.
  • Type-safe parameters — pass arguments without .family gymnastics.
  • Auto-disposed by default — generated providers behave like autoDispose.

Adding the Dependencies

Code generation needs both runtime and dev-time packages. riverpod_annotation ships the @riverpod annotation you use in source. riverpod_generator and build_runner run the build step that emits the .g.dart files.

A typical pubspec.yaml for a Flutter app looks like this.

dependencies:
  flutter:
    sdk: flutter
  flutter_riverpod: ^2.5.1
  riverpod_annotation: ^2.3.5

dev_dependencies:
  build_runner: ^2.4.11
  riverpod_generator: ^2.4.0
  custom_lint: ^0.6.4
  riverpod_lint: ^2.3.10

Your First Generated Provider

The smallest generated provider is a top-level function annotated with @riverpod. The first parameter is always a Ref object; the return type decides everything.

Because this function returns a plain String synchronously, the generator emits a read-only provider exposing that value. You consume it via ref.watch(helloWorldProvider) exactly like a hand-written Provider<String>.

Note the two required pieces: the part directive and the // ignore_for_file comment is optional — but the part 'file.g.dart'; is mandatory.

import 'package:riverpod_annotation/riverpod_annotation.dart';

part 'hello.g.dart';

@riverpod
String helloWorld(Ref ref) {
  return 'Hello, Riverpod 2.0';
}

Running the Generator

The annotation alone does nothing until build_runner generates the companion .g.dart file. Run it from the project root.

  • One-off build: generates once and exits. --delete-conflicting-outputs clears stale generated files.
  • Watch mode: regenerates automatically every time you save a source file — ideal during active development.

After it finishes, the helloWorldProvider symbol becomes available for import.

# Generate once
dart run build_runner build --delete-conflicting-outputs

# Or watch and rebuild on save
dart run build_runner watch --delete-conflicting-outputs

Return Type Drives the Provider

The generator reads your return type and silently picks the matching provider kind. This is the core convenience of code generation: you never name a provider type again.

  • Return T → synchronous provider (like Provider<T>).
  • Return Future<T> → async provider exposing AsyncValue<T> (like FutureProvider).
  • Return Stream<T> → stream provider exposing AsyncValue<T> (like StreamProvider).

Below, simply changing the signature to Future turns it into an async provider — no other change needed.

import 'package:riverpod_annotation/riverpod_annotation.dart';

part 'user.g.dart';

@riverpod
Future<String> userName(Ref ref) async {
  await Future<void>.delayed(const Duration(seconds: 1));
  return 'Ada Lovelace';
}

Passing Parameters (No More .family)

With hand-written providers, parameterizing meant .family and a tuple-like single argument. The generator lets you add normal function parameters after ref, and they become strongly typed provider arguments.

Here messageProvider takes an int id. You call it as ref.watch(messageProvider(42)). Multiple parameters and named/optional parameters all work.

import 'package:riverpod_annotation/riverpod_annotation.dart';

part 'message.g.dart';

@riverpod
Future<String> message(Ref ref, int id) async {
  final repo = ref.watch(messageRepositoryProvider);
  return repo.fetchById(id);
}

// Usage in a widget:
// final msg = ref.watch(messageProvider(42));

Stateful Logic: The Notifier Class

For mutable state with methods, annotate a class that extends the generated base class _$ClassName. You override build() to return the initial state; the generator wires up a NotifierProvider for you.

Inside methods you mutate state, and listeners rebuild automatically. This replaces the old Notifier + manual NotifierProvider declaration with a single annotated class.

import 'package:riverpod_annotation/riverpod_annotation.dart';

part 'counter.g.dart';

@riverpod
class Counter extends _$Counter {
  @override
  int build() => 0;

  void increment() => state++;
  void reset() => state = 0;
}

Async Notifiers

If your build() returns a Future, the generator produces an AsyncNotifier. The exposed state is an AsyncValue<T> that automatically tracks loading, data, and error states.

To update state after an async action, assign AsyncValue.guard(...) to state — it runs your async code and captures success or error without manual try/catch.

import 'package:riverpod_annotation/riverpod_annotation.dart';

part 'todos.g.dart';

@riverpod
class Todos extends _$Todos {
  @override
  Future<List<String>> build() async {
    return ref.watch(todoRepositoryProvider).fetchAll();
  }

  Future<void> add(String title) async {
    state = const AsyncLoading();
    state = await AsyncValue.guard(() async {
      await ref.read(todoRepositoryProvider).create(title);
      return ref.read(todoRepositoryProvider).fetchAll();
    });
  }
}

Consuming Generated Providers

Generated providers are consumed exactly like manual ones — the generated symbol is <name>Provider for functions, or <ClassName>Provider for Notifier classes.

  • ref.watch(counterProvider) → the current state value.
  • ref.read(counterProvider.notifier) → the Notifier instance, to call methods like increment().
  • For async providers, watch returns an AsyncValue you handle with .when(...).
class CounterView extends ConsumerWidget {
  const CounterView({super.key});

  @override
  Widget build(BuildContext context, WidgetRef ref) {
    final count = ref.watch(counterProvider);
    return Column(
      children: [
        Text('Count: $count'),
        ElevatedButton(
          onPressed: () => ref.read(counterProvider.notifier).increment(),
          child: const Text('Add'),
        ),
      ],
    );
  }
}

Keep-Alive and Dependencies

Generated providers are auto-disposed by default — they drop their state when no longer watched. Two annotation options give you control:

  • @Riverpod(keepAlive: true) — keeps the provider alive even with no listeners (use for app-wide singletons like a Dio client).
  • @Riverpod(dependencies: [...]) — declares scoped overrides for provider scoping. Most apps don't need this.

The capitalized @Riverpod(...) form is just the configurable version of the lowercase @riverpod shorthand.

import 'package:dio/dio.dart';
import 'package:riverpod_annotation/riverpod_annotation.dart';

part 'http.g.dart';

@Riverpod(keepAlive: true)
Dio dio(Ref ref) {
  return Dio(BaseOptions(baseUrl: 'https://api.example.com'));
}

Pure Dart: Why the Logic Is Testable

A big payoff of code generation is that your provider bodies are plain Dart functions and classes — easy to reason about and unit test. Below is a standalone illustration of the same state++ mutation logic a generated Notifier would run, with no Flutter or Riverpod imports needed.

This kind of pure logic is exactly what you keep inside a generated @riverpod class so it stays trivial to test.

class Counter {
  int state = 0;
  void increment() => state++;
  void reset() => state = 0;
}

void main() {
  final counter = Counter();
  counter.increment();
  counter.increment();
  counter.increment();
  print('After 3 increments: ${counter.state}');
  counter.reset();
  print('After reset: ${counter.state}');
}

Quick Check

You annotate a function that returns Future<List<Product>> with @riverpod. What kind of provider does riverpod_generator emit, and how do you consume it in a widget?

Recap

You learned how riverpod_generator removes provider boilerplate:

  • Add riverpod_annotation (runtime) plus riverpod_generator and build_runner (dev), and a part '<file>.g.dart'; directive.
  • Annotate a function for read-only/derived values, or a class extending _$Name for stateful Notifiers.
  • The return type chooses the provider: T → sync, Future<T> → async (AsyncValue), Stream<T> → stream.
  • Add normal parameters after ref instead of .family.
  • Run dart run build_runner watch to regenerate on save.
  • Providers are auto-disposed by default; use @Riverpod(keepAlive: true) for app-wide singletons.

よくある質問

「riverpod_generatorと@riverpodによるコード生成」レッスンは無料ですか？

はい。「riverpod_generatorと@riverpodによるコード生成」の完全なテキストはこのウェブで無料で読めます。インタラクティブに演習し（組み込みコードエディタと24時間対応のAIチューター）、Flutter Mobile Developmentコースの残りをアンロックするには、CoddyKit PROにアップグレードしてください。 Flutter Mobile Developmentコースには全4レッスンが含まれています。

「riverpod_generatorと@riverpodによるコード生成」で何を学びますか？

riverpod_generatorのアノテーションを使い、ボイラープレートなしで型安全なプロバイダーを生成します。 ブラウザで直接実行するハンズオンコードでFlutter Mobile Developmentを演習し、24時間対応のAIチューターがレッスンを進める中での質問に答えます。

Flutter Mobile Developmentを始めるのに経験は必要ですか？

事前経験は必要ありません。CoddyKitのFlutter Mobile Developmentは初級者から上級者向けに構成されているため、ここから始めるか最初から始めて、自分のペースで進むことができます。 これはレッスン2/4です。

「riverpod_generatorと@riverpodによるコード生成」レッスンにはどのくらい時間がかかりますか？

ほとんどのCoddyKitレッスンは約5～10分かかります。各レッスンはコンパクトでインタラクティブなので、着実に進歩し、ウェブとアプリ全体で正確に前回の場所から再開できます。

このFlutter Mobile Developmentレッスンでコードを書いて実行できますか？

はい。すべてのFlutter Mobile Developmentレッスンに組み込みコードエディタが含まれているため、ブラウザでリアルコードを書いて実行し、即座のAIフィードバックを取得できます。ローカル設定は不要です。

このコースのすべてのレッスン

  1. ProviderからRiverpodへ：レガシー状態管理の移行
  2. riverpod_generatorと@riverpodによるコード生成
  3. AsyncNotifierとFutureProviderのデータパイプライン
  4. プロバイダーのスコープ、オーバーライド、ProviderObserver
← Flutter Mobile Developmentに戻る