Feature Synth

Python bindings for the Rust-based feature synthesizer from the Data Pilot project.

Installation

pip install feature_synth

Usage

from feature_synth import PyColumnSchema, PyDataType, PyLogicalType, make_custom_df

# Define schema
schema = [
    PyColumnSchema(
        id="0", 
        name="age", 
        dtype=PyDataType.Int32, 
        logical_type=PyLogicalType.Number
    ),
    PyColumnSchema(
        id="1", 
        name="active", 
        dtype=PyDataType.Boolean, 
        logical_type=PyLogicalType.Boolean
    ),
]

# Generate DataFrame
df = make_custom_df(schema, rows=100)
print(df)

Features

• Type-safe data generation with Rust performance
• Polars integration for efficient data manipulation
• Full type hints for IDE support
• Multiple data types including numbers, booleans, dates, strings
• Logical types for semantic data generation

This package uses PyO3 and maturin to expose Rust code to Python via a native extension.

🚀 Quick Start (Developers)

🧱 Prerequisites

• Python 3.8+
• Rust toolchain (cargo, rustc, etc.)
• maturin (install with pip install maturin)
• Optional: use virtualenv or pyenv for isolation

🔧 Setup for Local Development

1. Clone the repo and enter the directory:

   git clone https://github.com/dvreed77/feature-synth.git
   cd feature-synth
   

2. Create a virtual environment (recommended):

   python -m venv .venv
   source .venv/bin/activate
   

3. Install dependencies & build the native extension:

   pip install maturin
   maturin develop
   

4. Test in Python:

   import feature_synth
   print(feature_synth.add(2, 3))  # should output 5
   

🛠 Project Structure

feature_synth/
├── Cargo.toml         # Rust crate configuration
├── pyproject.toml     # Python metadata (PEP 621)
├── src/
│   └── lib.rs         # Rust code exposed to Python
└── README.md          # You are here

🔄 Rebuilding after Rust Code Changes

Just re-run:

maturin develop

In your Python REPL, restart or use:

import importlib
import feature_synth
importlib.reload(feature_synth)

🐍 Packaging & Publishing

To build a wheel:

maturin build

To publish to PyPI:

maturin publish --username __token__ --password pypi-<your-token>

🧪 Testing

You can test your Rust code as usual with:

cargo test

Or test Python usage by writing Python unit tests in a separate test script.

📄 License

MIT