Provet Cloud Discharge Note Generator

Provet is a specialized tool that automatically generates high-quality discharge notes for veterinary consultations using OpenAI's language models. The tool processes JSON consultation data to produce comprehensive, professionally formatted discharge notes that veterinarians can provide to pet owners.

✨ Features

• 🚀 Advanced LLM Integration: Leverages OpenAI's models to generate accurate, contextual discharge notes
• 📋 Template-Based Generation: Uses Jinja2 templates for consistent note formatting
• 🔄 Multiple Deployment Options: Run as CLI tool, API service, or use in development
• 🐳 Docker Support: Containerized deployment for easy scaling
• 📱 REST API: Integrate with existing veterinary practice management systems
• ⚙️ Highly Configurable: Adjust model parameters and customize system instructions

📋 Requirements

• Python 3.13 or higher (or uv which manages Python version for you)
• OpenAI API key

🚀 Quick Start

The project supports three different setup methods:

# Run the interactive setup script
./setup.sh

Choose from:

1. Development environment (uv) - For contributors and developers
2. API service (Docker with uv) - For running the API service
3. CLI only (uv) - For simple command-line usage

🛠️ Setup Options

1. Development Environment (uv)

Best for active development work:

./scripts/setup_dev.sh

Requirements:

• uv (installed by the script if needed)

This setup:

• Creates a Python 3.13 virtual environment with uv
• Installs all dependencies using uv (including development tools)
• Sets up environment for local development

2. API Service (Docker with uv)

Best for running the API service:

./scripts/setup_docker.sh

Requirements:

• Docker
• Docker Compose

This setup:

• Builds the Docker image using uv for dependency management
• Includes a test stage that runs unit tests during the build process
• Starts the API service using Docker Compose
• Exposes the API at http://localhost:8000

3. CLI Only (uv)

Best for simple command-line usage:

./scripts/setup_cli.sh

Requirements:

• uv (installed by the script if needed)

This setup:

• Creates a separate Python 3.13 virtual environment (.venv-cli) using uv
• Installs only the main required dependencies using uv sync
• Sets up the CLI command for minimal installation

💻 Usage

Command Line

After setting up with any of the options:

# Using the CLI wrapper script
./provet_cli.py data/consultation1.json

# Or as a Python module
python -m provet data/consultation1.json

This will generate a discharge note and save it to a JSON file in the solution directory.

API Service

When running with the Docker setup:

# The API is available at
http://localhost:8000

# Swagger UI documentation at
http://localhost:8000/docs

⚙️ Environment Variables

Copy .env-template to .env and add your OpenAI API key:

OPENAI_API_KEY=your_openai_api_key_here

Additional configuration options:

# LLM Configuration
OPENAI_MODEL=gpt-4o
TEMPERATURE=0.7
MAX_TOKENS=800

# Add custom instructions to the system message
CUSTOM_SYSTEM_INSTRUCTION=It is extremely important that you follow the exact format provided in the prompt template. Do not deviate from the section structure or add extra sections.

📁 Project Structure

provet/                 # Core package
├── __init__.py         # Package initialization
├── __main__.py         # Command-line entry point
├── core/               # Core functionality
│   ├── app.py          # Main application (Facade pattern)
│   ├── data_models.py  # Data models using dataclasses
│   ├── io_manager.py   # File I/O operations
│   └── llm_service.py  # Language model interaction
├── templates/          # Built-in Jinja2 templates
└── utils/              # Utility modules

api/                    # API service
├── main.py             # FastAPI application
└── __init__.py         # Package initialization

scripts/                # Setup and utility scripts
├── setup_dev.sh        # Development environment setup (uv)
├── setup_docker.sh     # Docker setup for API service (uses uv)
├── setup_cli.sh        # CLI setup with uv
├── cleanup.sh          # Clean up environments
└── verify_setup.sh     # Verify setup requirements

data/                   # Sample input data
├── consultation1.json  # Sample consultation data
└── consultation2.json  # More sample consultation data

solution/               # Generated output files

# Dependency files
pyproject.toml          # Project configuration and dependencies
Dockerfile              # Container definition
docker-compose.yml      # Container orchestration

🔄 Development Workflow

1. Initial Setup: Run ./scripts/setup_dev.sh to create the development environment
2. Activate Environment: Run source .venv/bin/activate to activate the development virtual environment (Or source .venv-cli/bin/activate if you only installed the CLI)
3. Run Tests: Run python -m pytest to execute tests
4. Make Changes: Edit code in the provet/ directory
5. Test CLI: Run python -m provet data/some_file.json to test
6. Test API: Run python -m uvicorn api.main:app --reload for local API testing

🐳 Docker Build Process

The Docker build uses a multi-stage process:

1. Builder stage: Sets up dependencies and prepares the application code
2. Test stage: Runs unit tests to ensure code quality and functionality
3. Runtime stage: Creates the final lightweight image for production use

This ensures that only images that pass tests are deployed to production.

📦 Dependency Management

The project uses uv exclusively for dependency management, with all dependencies defined in pyproject.toml.

Installing Dependencies with uv

# Install core package with all dependencies
uv sync --all-groups

# Install with specific groups
uv sync --group main  # Main dependencies only (for CLI)
uv sync --group dev   # Development dependencies
uv sync --group test  # Test dependencies
uv sync --group api   # API dependencies

🧹 Cleanup

To reset your environment:

./scripts/cleanup.sh

This will remove all local environments and stop Docker containers (data files will not be affected).

📝 License

This project is licensed under the MIT License - see the LICENSE file for details.