py-metrics

PyMetrics

PyMetrics is a self-hostable platform that turns any Python repository into 50+ software-metric columns and ships them as CSV/Parquet-ready tables for dashboards, research, or downstream ML.

{ width="140" }

What it does

• Accepts a GitHub URL, clones it in a sandbox, and produces per-file, per-class, and per-function metrics.
• Combines Radon (raw, CC, MI, Halstead), PyDriller (git history), and Pylint (optional) into a single table.
• Streams live progress over Server-Sent Events; results land in SQLite.
• Exports CSV / ZIP artifacts under descriptive repo/ref/date folders.
• Explains metric definitions and source references directly from the dashboard.

Why it exists

Metrics pipelines are everywhere and yet nowhere: every lab re-implements “clone → measure → aggregate” in a notebook that no one can reproduce. PyMetrics makes the pipeline a reproducible service and ships a curated dataset pipeline so readers can reproduce and publish a CC-BY-4.0 corpus.

Getting started

• Install — docker compose up or native Python.
• Usage — analyze your first repo via UI or REST.
• Metrics reference — every metric, its units, and references.
• Metrik sözlüğü TR — metriklerin Türkçe, sıralı ve detaylı açıklaması.
• Dataset — the dataset pipeline and how to rebuild a release.
• Architecture — how the pieces fit together.

Citing

Both the analyzer and the dataset are citable. See CITATION.cff at the repository root, or copy the BibTeX entry from the dataset landing page on Kaggle.

This site is open source. Improve this page.