brick_sqlite 1.0.0
brick_sqlite: ^1.0.0 copied to clipboard

SQLite connector for Brick, a data persistence library for Flutter

Brick SQLite Provider #

Local storage for Flutter apps using Brick.

Supported Query Configuration #

providerArgs #

The following map exactly to their SQLite keywords. The values will be inserted into a SQLite statement without being prepared.

  • collate
  • having
  • groupBy
  • limit
  • offset
  • orderBy

As the values are directly inserted, use the field name:

//given this field
@Sqlite(name: 'last_name')
final String lastName;

  where: [Where.exact('lastName', 'Mustermann')],
  providerArgs: {'orderBy': 'lastName ASC'},

where: #

All fields and associations are supported. All Compare values are also supported without additional configuration.

Models #

Intelligent Migrations #

Whenever a new field(s) is added or removed from a connected model, Brick can automatically generate a migration for SQLite. For even less friction, running the watcher while actively developing will create the migration on model save:

flutter pub run build_runner watch

💡 While Brick guesses right most of the time, the migration should still be reviewed after it's created (for example, a DropTable might just be a RenameTable or maybe onDeleteCascade needs to be set for a InsertForeignKey).

Fields #

Maps can be serialized, but they must be digestible by jsonEncode.

@Sqlite(columnType:) #

In some exceptional circumstances, low-level manipulation of Brick's automatic schema creation is necessary.

⚠️ This is an advanced feature. In nearly every case, you can trust Brick's determination of your field. If you're frequently using this option, consider your greater architecture and use of Brick.

@Sqlite(columnType: Column.blob)
final Uint8List image;

⚠️ Because this feature overrides Brick assumptions about the column type, the field will be inserted (toSqlite) as is and returned as is from deserialization (fromSqlite). @Sqlite(fromGenerator:) and @Sqlite(toGenerator:) are required if Brick does not know how to serialize the field.

@Sqlite(name:) #

SQLite column names can be named per field except with associations. Using name: is strongly discouraged as Brick's naming consistency is reliable and easily managed through migrations.

@Sqlite(name: "full_name")
final String lastName;

@Sqlite(unique:) #

This fields are marked UNIQUE in SQLite and are useful for external identifiers. An error will throw if a non-unique value is inserted.

@Sqlite(unique: true)
final String lastName;

Updating Associations #

If your instance fields are mutable (i.e. non final), Brick will reconcile associations after saving the instance.

class MyModel extends SqliteModel {
  /// if this field were `final List<AssociationModel> associations`
  /// calling `associations.clear()` and then `upsert(instance)` **would not**
  /// change the value of associations.
  List<AssociationModel> associations;


final instance = MyModel([AssociationModel()]);
instance.associations.length // => 1
await provider.upsert<MyModel>(instance);
final instanceFromSqlite = await provider.get<MyModel>(
  query: Query.where('primaryKey', instance.primaryKey, limit1: true)
instanceFromSqlite.associations.length // => 0

Unsupported Field Types #

The following are not serialized to SQLite. However, unsupported types can still be accessed in the model as non-final fields.

  • Nested List<> e.g. <List<List<int>>>
  • Many-to-many associations

Multiplatform Support #

Brick SQLite can be used when developing for Windows, MacOS, and Linux platforms. The following is not required for iOS and Android development except in a test environment..

  1. Add sqflite_common packages to your pubspec. If you're stubbing SQLite responses for testing, the packages only need to be added under dev_dependencies:.

     sqflite_common: any
     sqflite_common_ffi: any
  2. Use the SQLite FFI database factory when initializing your provider:

     import 'package:sqflite_common/sqlite_api.dart';
     import 'package:sqflite_common_ffi/sqflite_ffi.dart';
       sqliteProvider: SqliteProvider(
         databaseFactory: databaseFactoryFfi,
  3. Make sure FFI is initialized when starting your app or running unit tests:

     void main() {

Testing #

SQLite providers should use sqlite_ffi as described in multiplatform support:

import 'package:sqflite_common/sqlite_api.dart';
import 'package:sqflite_common_ffi/sqflite_ffi.dart';

void main() {

  final provider = SqliteProvider(
    databaseFactory: databaseFactoryFfi,
    modelDictionary: dictionary,

  setUpAll(() async {
    await provider.migrate(myMigrations);
    // upsert any expected data
    await provider.upsert<DemoModel>(DemoModel('Guy'));


Can I specify a different table name? #

Table names, association column names, and primary key column names are managed by the package. They are currently unchangeable.

Memory Cache Provider #

The Memory Cache Provider is a key-value store that functions based on the SQLite primary keys to optimize the SQLite provider queries. This is especially effective for low-level associations. The provider only caches models it's aware of:

// app models: `User`, `Hat`
// `User` is never stored in memory

It is not recommended to use this provider with parent models that have child associations, as those children may be updated in the future without notifying the parent.


Why can't I declare a model argument? #

Due to an open analyzer bug, a custom model cannot be passed to the repository as a type argument.

pub points



SQLite connector for Brick, a data persistence library for Flutter

Repository (GitHub)
View/report issues


API reference




brick_core, brick_sqlite_abstract, logging, meta, sqflite, synchronized


Packages that depend on brick_sqlite