Dart Object Graphs

DOGs (Dart Object Graphs) is a serialization and object-mapping library for Dart that turns your classes into introspectable, immutable, and easily validated data models across multiple formats.

Documentation | Discord | Pub Package

@serializable
class Person with Dataclass<Person> {

  @LengthRange(max: 128)
  final String name;

  @Minimum(18)
  final int age;

  @SizeRange(max: 16)
  @Regex("((_)?[a-z]+[A-Za-z0-9]*)+")
  final Set<String>? tags;

  Person(this.name, this.age, this.tags);

}

• 🐦 Concise Define immutable dataclasses with a simple annotation and mixin. toString, equals, hashCode, and builders are generated automatically. No equatable, no copyWith, no additional boilerplate.
• 🚀 Fast Optimized serialization and deserialization with performance on par or better than alternatives. JSON support is built in with zero extra dependencies.
• 🧩 Extensible Replace or extend any part of the pipeline. Add new formats, validators, projections, or field converters. Works with custom Flutter types through the integration package.
• 📦 Adaptive Multiple formats supported out of the box: JSON, YAML, TOML, CBOR. Schema generation enables dynamic use cases like server-driven forms.
• 📚 Documented Comprehensive documentation and examples
• ⚡ Developer-friendly All generated code goes into a single dogs.g.dart. Models remain clean and free of part directives. Builders and nullable copy methods are exposed centrally.

Format Support

• JSON (included in dogs_core)
• YAML
• TOML
• CBOR

Core Package

• Json
  The core package comes with json support out of the box and requires no additional dependencies.

• Builders & Nullable Copy
  We provide a nullable copy method that also works with null values, as well as autogenerated builders, similar to built_value. Both let you create immutable objects with a simple and fluent api.

• Structures
  Every serializable class has an introspectable generated DogStructure that provides runtime access to the class's fields, annotations, factories and metadata. Together with the extensible core system, you can build powerful reflection-like functionality without the need for custom code generation or mirrors.

• Dataclasses
  Using our dataclass mixin automatically implements toString, equals and hashCode for your serializable classes. This removes the requirement for a package like equatable or ungodly amounts of boilerplate code, that is hard to maintain and distracting.

• Projections
  Dogs offers a powerful projection api, which enables you to transform your objects into other objects, without having to write additional boilerplate code with brittle copyWith invocations and hard coded key names.

• Polymorphism
  Dogs supports polymorphic serialization, which enables you to serialize and deserialize objects of different types. You can even use interfaces and abstract classes as types, including Object.

• Validation
  Using the validation api, you can easily validate your objects, without having the struggle of writing your own validation logic for most common use cases. Translation of messages and custom validators are supported as well.

• JsonSchema-like Schema (preview)
  Dogs offers a schema generation api, which allows you to dynamically export and parse a subset of json-schema. This enables advanced dynamic uses cases such as partially server-driven forms when used in combination with the flutter integration.

Package on pub.dev
Documentation

Flutter

Use our flutter integration to easily create forms for any serializable object, without having to write boilerplate code. The databinding system supports all primitive field types and can easily be extended to support custom field types and layouts. This package also provides converters for commonly used flutter types, like Color, Offset, Size, Rect, etc.

Package on pub.dev
Documentation

✨ Non-Intrusive Code Generation ✨

Dogs takes a different approach to generated code. Instead of scattering part files across your models, all generated output lives in a single dogs.g.dart file at the root of your lib folder. This makes development smoother and far less error-prone. No more missing part statements or fragile imports.

Even better: your model definitions stay clean. You never need to reference generated code inside them. The only time you’ll interact with generated code directly is when using builders, which are neatly exported through the central dogs.g.dart.

This design keeps Dogs non-intrusive and future-proof. While part files may be added later for flexibility, you will never be required to reference generated code from within your model definition files.