README.md

---
title: Github Repo Stats
emoji: 🩻
colorFrom: indigo
colorTo: yellow
sdk: gradio
sdk_version: 6.9.0
app_file: app.py
pinned: false
short_description: Visualize lines of code by language for a GitHub repo
---

# Github Repo Stats 🩻

A simple app ([deployed on HF spaces](https://huggingface.co/spaces/sukiboo/github-repo-stats)) to visualize lines of code by language and other basic stats for any public GitHub repo.

<img width="900" height="974" alt="demo" src="https://github.com/user-attachments/assets/f0cf4f76-eefe-46b6-a8da-92ddd8282ad6" />

## Why Make It

I was kinda annoyed that it doesn't exist already, or at least not in the way I imagined.

The issue with GitHub's language breakdown is that it approximates language contribution based on file size, not actual lines of code -- especially noticeable with Jupyter notebooks.
This app fixes that by counting actual lines.


## How It Works

1. Fetches the full file tree via the GitHub API
2. Filters out binary files, vendored directories (`node_modules`, `vendor`, `dist`, etc.), and non-code files
3. Downloads each source file from `raw.githubusercontent.com` concurrently, largest files first so the live breakdown stabilizes early instead of jumping when a big file lands late
4. Counts newlines in each file, mapping file extensions to languages
5. Pulls weekly commit counts over the repo's full lifetime via `/stats/contributors` and renders them as a block-character histogram below the language chart


## Running Locally

```bash
pip install -r requirements.txt
python app.py
```

Optionally set a `GITHUB_TOKEN` environment variable to avoid GitHub API rate limits and to reliably fetch the commit history:

```bash
export GITHUB_TOKEN=ghp_...
python app.py
```

## LICENSE

MIT <3