Skip to content

v0.1.0 - Initial Release: Complete Architecture, Multi-Language UI & Enhanced Analysis

Choose a tag to compare

View all tags
@Xe-Persistent Xe-Persistent released this 06 Jan 13:32
· 288 commits to master since this release
v0.1.0
d9b98aa

简体中文 (Simplified Chinese)

Akagi-NG v0.1.0 是一个具有完整功能的里程碑版本。本项目旨在为在线日本麻将提供实时的 AI 辅助分析，此版本包含了完整的架构搭建、核心 AI 逻辑实现、现代化的前端 UI 以及针对 Windows 平台的发布支持。

🚀 核心功能 (Features)

  • 立直模拟推演 (Riichi Lookahead)：当玩家可以选择立直时，AI 会模拟后续局势并推荐最佳的立直舍牌。
  • 复合杠牌分析 (Multi-Kan Analysis)：支持并显示复杂的杠牌选项（如暗杠 vs 加杠 vs 大明杠），提供更细致的决策辅助。
  • 多语言 UI 支持：完整集成了 i18n，支持简体中文、繁體中文、日语和英语。
  • 实时设置系统：实现了前后端互通的设置系统，支持动态调整 AI 模型、日志级别及浏览器行为。
  • 现代化前端：
    • 构建了基于 React 的流式推荐 UI 。
    • 模块化了麻将组件，支持高亮显示与交互。
    • 增加了设置面板与错误处理机制。

🏗️ 架构与重构 (Refactoring & Architecture)

  • 后端重构：
    • 深度重构 mjai_bot，优化了 Bot 决策流。
    • 重构前端适配层 frontend_adapter，增强了异常回退逻辑。
    • 统一使用绝对导入 (akagi_ng. 命名空间)，修复了大量相对导入引发的问题。
  • 雀魂桥接 (Majsoul Bridge)：
    • 整理并迁移了雀魂相关的桥接代码与 Proto 定义。
    • 优化了 LiqiProto 的日志输出级别，减少不必要的噪音。
  • 工程化 (Project Structure)：
    • 建立了标准的目录结构，统一管理 assets（资源/Schema）、config（配置文件）、lib（二进制扩展）与 models（模型权值）目录。
    • 清理了冗余的 ensure_dir 调用与废弃的训练代码。

📦 发布与部署 (Release & Deployment)

  • 自动化构建：新增 build_release.py，支持一键打包 Windows 可执行文件。
  • 版本元数据：自动为 EXE 文件嵌入版本号与版权信息。
  • 文档完善：提供了详尽的中英文 README，包含安装指南与基于真实 UI 的运行截图。

English

Akagi-NG v0.1.0 is a milestone release offering a complete, production-ready solution for real-time Mahjong AI analysis. This release includes the full architecture setup, core AI logic implementation, a modern web-based UI, and Windows deployment support.

🚀 Key Features

  • Riichi Lookahead: Simulates hand progression to recommend the optimal tile to discard when declaring Riichi.
  • Multi-Kan Analysis: Identifies and displays multiple Kan options (Ankan/Kakan/Daiminkan) simultaneously for precise decision-making.
  • Multi-Language UI: Fully integrated i18n support for Simplified Chinese, Traditional Chinese, Japanese, and English.
  • Real-time Configuration: Implemented a synchronized settings system for dynamic adjustment of AI models, logging levels, and browser behavior.
  • Modern Web UI:
    • Built a React-based streaming recommendation interface.
    • Modularized Mahjong components with highlighting and interaction support.
    • Added a comprehensive settings panel and error handling mechanisms.

🏗️ Architecture & Refactoring

  • Backend Overhaul:
    • Deep refactoring of mjai_bot to optimize decision flows.
    • Refactored frontend_adapter with robust fallback logic.
    • Standardized absolute imports (using the akagi_ng. namespace) across the codebase.
  • Majsoul Bridge:
    • Relocated and organized Majsoul bridge files and Proto definitions.
    • Tuned logging levels in LiqiProto to reduce noise.
  • Project Structure:
    • Established a standardized directory structure, centrally managing assets (resources/schemas), config (configuration files), lib (binary extensions), and models (model weights) .
    • Cleaned up redundant directory creation calls and deprecated training code.

📦 Release & Deployment

  • Automated Build: Introduced build_release.py for one-click Windows executable packaging.
  • Version Metadata: Automated embedding of version info and copyright details into the EXE file.
  • Comprehensive Documentation: Updated English and Chinese READMEs with installation guides and screenshot galleries showcasing real-world usage.
Assets 3
Loading