📡 GitHub Trending Rank Radar
English | 中文
Independently collected trend rankings of open-source GitHub projects — fully open source, self-collected data.
🌐 Live: https://zhuikan.com/ (mirror: https://awesome-market.github.io/trending-rank-radar/)
Eight boards
| Board | Criteria |
|---|
| New | Created within the last 90 days, sorted by current stars (≥20★) |
| Daily / Weekly / Monthly Growth | Absolute star growth over the window, desc (growth > 0, current ≥100★) |
| Daily / Weekly / Monthly Rate | Relative rate = growth ÷ star base at window start × 100 (base ≥50) |
| All-Time | Current total stars, desc |
- Growth is computed from daily star snapshot diffs, collected twice a day at 10:00 / 22:00 Beijing time
- Rate boards and growth boards are listed separately (the eight-board layout)
- When history is too short, the earliest available snapshot is used as the base and the UI marks the actual window (e.g.
10-15 ~ 11-08 · 24d)
Replay by day / month / year
- Day: the eight-board snapshot for any historical date
- Month: month-end boards + cumulative monthly growth + daily Top10 digest
- Year: annual growth board + monthly board entries
UI features
- Chinese / English: one-click
EN/中 toggle in the top-right corner, persisted locally
- Dark / light theme: manual ☀/☾ toggle, falls back to system preference, flash-free loading
- Wide rows: owner + repo name + avatar + language/star/fork/watch/PR/issue/release meta + adaptive description + inline 30-day sparkline
- Podium: staggered gold/silver/bronze cards for the top three; empty states show accumulation progress and a fallback board preview
- Desktop-first: designed for ≥1440px screens, usable degradation on smaller ones
Architecture
GitHub Trending / Search API / GraphQL API
│ GitHub Actions, twice a day
▼
data/snapshots/*.json ──► scripts/rank.py ──► site/data/*.json
(the git repo is the DB) (boards precomputed) (pure static, Pages-hosted)
- Zero-dependency Python 3: collection and ranking use only the standard library (
urllib + html.parser + json)
- Pure static frontend: vanilla HTML/CSS/JS, no build tooling, served directly by GitHub Pages
- Data as archive: every snapshot is committed to git — the repo itself is a downloadable historical database
Methodology notes
- The New / All-Time boards rank within the tracked set (~5,000 repos, maintained via Trending + Search API + growth signals), not all of GitHub
- Rate = growth ÷ star base at window start, floored to an integer percentage
issues counts come from the GitHub API (REST semantics include PRs; GraphQL reports issues / pullRequests separately)- Data is for reference only — not advice of any kind
Run locally
# collect a snapshot (needs GITHUB_TOKEN)
GITHUB_TOKEN=xxx python3 scripts/collect.py
# rebuild boards & site data
python3 scripts/rank.py --site-base http://localhost:8000
# preview
cd site && python3 -m http.server 8000
Repo layout
├── docs/ # design docs (data pipeline / ranking / frontend / automation)
├── scripts/ # collect.py · rank.py · gh.py (zero-dependency)
├── data/ # snapshots & metadata (git is the archive)
├── site/ # generated static site (Pages deploy source)
└── .github/workflows/ # scheduled collection + Pages deploy
License
Apache-2.0
