title	Bunsho Assistant API
emoji	🛡️
colorFrom	blue
colorTo	blue
sdk	docker
app_port	8000

🛡️ Bunsho Assistant

Bunsho Assistant is an advanced, AI-powered legal document analyzer designed specifically for processing complex Japanese bureaucratic and legal documents. It automatically extracts text via OCR, analyzes potential risks using Google Gemini's massive context window, and provides detailed, structured reports in both English and Japanese.

---

🌟 Live Demo

Don't want to deal with local setup and massive ML models? Use our free, cloud-hosted version!

• Frontend [APP LINK]: https://bunsho-assistant-flax.vercel.app
• Backend API: https://ramvicky2004-bunsho-assistant-api.hf.space

---

✨ Features

• Advanced OCR Pipeline: Utilizes PaddleOCR and Tesseract to accurately extract Japanese text from dense PDFs and images.
• Deep Legal Analysis: Powered by Google Gemini 2.5 Flash, the system flags dangerous clauses, financial obligations, hidden risks, and cancellation penalties.
• Global Language Toggle: Instantly switch between English and Japanese UI/Analysis with a seamless global state architecture.
• Interactive Document Chat: Ask specific questions about your uploaded document directly to the AI.
• Printable PDF Reports: Generate clean, structured PDF reports detailing the risk score and findings.

---

🏗️ Architecture

This project was built to prevent Out-Of-Memory (OOM) crashes caused by heavy Machine Learning models, utilizing a robust split-stack architecture:

1. Frontend (Vercel): A Next.js application styled with TailwindCSS and Lucide-React icons, providing a dynamic, glassmorphism-inspired user interface.
2. Backend (Hugging Face Spaces): A FastAPI server running in a custom Docker container. Hosted on Hugging Face Spaces to provide the required 16GB of RAM needed to comfortably run PaddleOCR without crashing.

---

💻 Local Development Setup

If you wish to run this heavy architecture locally, follow these precise steps.

Prerequisites

• Python 3.12+
• Node.js 18+
• Tesseract OCR installed on your machine (tesseract-ocr-jpn required).

1. Backend Setup (FastAPI + AI)

cd backend

# Create a virtual environment
python -m venv venv
source venv/bin/activate  # On Windows use: venv\Scripts\activate

# Install heavy ML dependencies
pip install -r requirements.txt

# Environment Variables
# Create a .env file in the backend directory:
# GEMINI_API_KEY=your_google_ai_studio_key_here

# Start the API
uvicorn backend.main:app --reload

The backend will run on http://127.0.0.1:8000.

2. Frontend Setup (Next.js)

cd frontend

# Install UI dependencies
npm install

# Start the development server
npm run dev

The frontend will run on http://localhost:3000.

---

🐳 Docker Deployment

The repository includes a highly optimized Linux Dockerfile at the root directory specifically designed for Hugging Face Spaces. It automatically installs system-level dependencies (libgl1, poppler-utils, tesseract-ocr-jpn) before booting the FastAPI server.

To deploy on Hugging Face Spaces:

1. Create a new Docker Space.
2. Set hardware to Free (2 vCPU, 16GB RAM).
3. Push this repository to the Space.
4. Add GEMINI_API_KEY to the Space Secrets.

---

Built with ❤️ to navigate Japanese bureaucracy safely and intelligently.