Datra
July 28, 2026 · View on GitHub
한국어 | English
Datra is a game data management system that uses C# Source Generators to automatically generate serialization code for CSV, JSON, and YAML data. Works seamlessly in both Unity and .NET environments.
Features
- Multiple Formats: CSV, JSON, YAML with auto-detection
- Zero Boilerplate: Source Generators eliminate manual serialization code
- Type Safety: Compile-time validation with strong typing
- Data References: Type-safe
DataRef<T>for cross-table references - Unity Integration: Built-in Editor window with Table/Form views
- Localization: Multi-language support with
LocaleRef - Advanced Types: Nested structs, polymorphic JSON, arrays, enums
Quick Start
1. Define Your Data Model
using Datra.Attributes;
using Datra.Interfaces;
[TableData("Characters.csv")]
public partial class CharacterData : ITableData<string>
{
public string Id { get; set; }
public string Name { get; set; }
public int Level { get; set; }
public int Health { get; set; }
}
[SingleData("GameConfig.json")]
public partial class GameConfigData
{
public string GameName { get; set; }
public int MaxLevel { get; set; }
}
2. Configure the Data Context
[assembly: DatraConfiguration("GameData",
Namespace = "MyGame.Generated" // Required
)]
3. Load and Use Data
var provider = new FileRawDataProvider("path/to/data");
var context = new GameDataContext(provider, new DataLoaderFactory());
await context.LoadAllAsync();
var hero = context.Character.GetById("hero_001");
var config = context.GameConfig.Get();
Installation
Unity
Add to Packages/manifest.json:
{
"dependencies": {
"com.penspanic.datra": "https://github.com/penspanic/Datra.git?path=Datra",
"com.penspanic.datra.editor": "https://github.com/penspanic/Datra.git?path=Datra.Editor",
"com.penspanic.datra.unity": "https://github.com/penspanic/Datra.git?path=Datra.Unity"
}
}
.NET
Add project references:
<ProjectReference Include="path/to/Datra/Datra.csproj" />
<ProjectReference Include="path/to/Datra.Generators/Datra.Generators.csproj"
OutputItemType="Analyzer"
ReferenceOutputAssembly="false" />
Trimming (Blazor WebAssembly, self-contained .NET)
Datra is marked IsTrimmable, so an application that publishes trimmed and
never calls Datra has it removed entirely. Nothing to configure.
YamlDotNet, which Datra depends on, carries no such marking, so the linker
copies it whole and it stays on the wire even with no references left pointing
at it. An application that reads no YAML can say so:
<PropertyGroup>
<DatraTrimYaml>true</DatraTrimYaml>
</PropertyGroup>
Off by default, and it has to be: Datra's YAML path is reflection-driven, so a trimmed build that does read YAML fails rather than running slow. JSON and CSV are unaffected either way.
Measured on a Blazor WebAssembly game that reads only JSON: Datra (347 KB) and YamlDotNet (293 KB) leave the output, and with AOT the native blob drops a further 547 KB brotli.
The source generator's DatraEmitYamlSerializers property is also reachable
from a consuming project as of 0.3.5 — before that the package shipped no
build/ props declaring it compiler-visible, so setting it did nothing.
Unity Editor
The Unity Editor window provides:
- Table View / Form View for data editing
- Real-time change tracking with save/revert
- Collection editors for List, Dictionary, arrays
- DataRef selectors with dropdown pickers
- Localization panel for multi-language editing
Open via: Window > Datra > Data Editor
Project Structure
Datra/
├── Datra/ # Core runtime library
├── Datra.Generators/ # Source Generator (compile-time)
├── Datra.Analyzers/ # Roslyn Analyzers
├── Datra.Editor/ # Shared editor utilities
├── Datra.Unity/ # Unity packages (Runtime, Editor, Addressables)
├── Datra.Tests/ # Unit tests
└── Datra.SampleData/ # Sample data models
Documentation
- Features Guide - Detailed feature documentation
- Unity Guide - Unity integration guide
- Developer Guide - Internal development guide
License
MIT License - see LICENSE