Game Localizer

简体中文 | English

Game Localizer is a localization pipeline for game text, not merely a machine-translation tool. It brings resource scanning, translation memory, model-assisted translation, quality assurance, human revision, artifact building, and publishing together in one traceable workflow.

The project was open-sourced in August 2026, but the workflow itself grew out of several years of maintaining Chinese localization for Mir Tankov. Practices validated through recurring real game updates were generalized into this reusable framework. Localization resources and release metadata are maintained separately in tanki-i18n-metadata, keeping redistributable framework code distinct from game-specific content and release data.

Project status

Game Localizer is an actively developed pre-1.0 open-source project. The current repository already includes cross-platform CI, automated regression coverage, full-history secret scanning, bilingual documentation, an example project, and explicit QualityGate / human-authorization boundaries around release and publishing workflows.

The next major design track is M8 Agentic Workflows and the Tauri Client. It is intentionally documented as a proposal before implementation so agent permissions, deterministic validation, security boundaries, and human approval requirements remain reviewable.

• Contributing
• Security policy
• Changelog
• Issue tracker

Key features

• Supports Gettext PO/MO, ParaTranz JSON, and Paradox YAML.
• Manages SQLite translation memory with stable coordinates and source fingerprints.
• Supports OpenAI-compatible translation models, concurrency, rate limiting, and tokenizers.
• Checks placeholders, terminology, residual source-language text, filtering, and normalization.
• Provides auditable human revision, failure recovery, and incremental rebuilds.
• Builds preview artifacts and QualityGate-protected release artifacts.
• Publishes to local directories, GitHub Releases, Cloudflare R2, and Alibaba Cloud OSS.

Documentation

The complete usage documentation now lives in one documentation site. This README remains a project entry point so parallel instructions do not drift apart.

• Documentation site
• Documentation source
• First-time setup
• Usage guide
• project.yaml reference
• TM and SQLite

Shortest path to a running dashboard

Python 3.10 or later is required. Windows PowerShell:

python -m venv .venv
.\.venv\Scripts\Activate.ps1
python -m pip install -e ".[all]"
localizer dashboard projects/example/project.yaml --host 127.0.0.1 --port 8080

Then open http://127.0.0.1:8080. The documentation site is authoritative for project configuration, credentials, CLI workflows, TM migration, and publishing.

Development

python -X utf8 -m unittest discover -s tests

The documentation site uses the open-source Material for MkDocs. Preview it locally with:

python -m pip install -e ".[docs]"
mkdocs serve

Current limitations

• Community-platform workflows currently have a configuration model and offline synchronization components, but no online API client.
• The web dashboard targets local observability and targeted single-user revision, not multi-user approval workflows.
• Remote publishing requires the relevant optional dependencies, environment credentials, and explicit governance configuration.
• M8 Agentic Workflows and the Tauri Client is currently a design proposal.

License

This project is free software licensed under the GNU General Public License v3.0 or later (SPDX: GPL-3.0-or-later). Third-party dependencies and external resources retain their respective licenses. See LICENSE for the complete terms.