Use LLMs to assist in turning ordinary Python into Cython-oriented source, either as classic .pyx/.pxd output or pure-Python-with-Cython-annotations.
The project is still in active reconstruction. The current focus is a reliable single-purpose harness with a modern CLI, prompt tuning, and pyproject.toml driven configuration.
Generate paired .pyx and .pxd files from Python source.
uv run recython convert .\examples\src_multiple_regression\multiple_regression .\tmp\classic --style classic --exclude __init__Generate Python files intended for Cython's pure-Python mode.
uv run recython convert .\examples\src_multiple_regression\multiple_regression .\tmp\pure --style pure --exclude __init__Inspect the bundled templates before tuning or replacing them.
uv run recython prompts list
uv run recython prompts show pure- Compile Python-heavy modules for speed-sensitive paths.
- Compile for distribution and mild source obfuscation.
- Prepare code for tighter integration with C and C++ ecosystems.
The medium-term goal is a focused harness with two primary scenarios:
- Initial cythonization for a module or package.
- Maintenance cythonization for code that has drifted after upstream Python changes.
That foundation should support later TUI and Tkinter GUI frontends without forking core logic.
The modernization roadmap lives in docs/modernization_plan.md.