Skip to content

Latest commit

 

History

History
76 lines (58 loc) · 3.27 KB

README.md

File metadata and controls

76 lines (58 loc) · 3.27 KB

pythonbackport website

Static landing page for the pythonbackport GitHub org.

The page is fully data-driven. Every byte of repository metadata on the rendered page comes directly from the live GitHub API at build time. There is no on-disk cache of repo data, and there is no config file — just one Python script, one HTML template, and a generated index.html.

GitHub API ──► scripts/build.py ──► index.html
                      ▲
                      │
                 templates/index.template.html
                 (HTML shell with {{projects}} placeholder)

Files

File Role Edit?
scripts/build.py Fetches /orgs/<org> + /orgs/<org>/repos, renders HTML. rarely
scripts/make_logo.py Turns pythonbackportlogo.png into the transparent logo.png. rarely
templates/index.template.html Static HTML shell with {{projects}}, {{site_title}}, {{site_tagline}}. yes
index.html Generated. Do not edit by hand. no
logo.png Generated from pythonbackportlogo.png. Do not edit by hand. no
.github/workflows/refresh.yml Daily cron + on-push regeneration, auto-commits the rebuilt index. rarely

Local

python scripts/build.py            # full build against live API
python scripts/build.py --check    # CI guard: fail if index.html is stale
python scripts/build.py --stdout    # print HTML to stdout
python scripts/build.py --org foo  # fetch a different org
python scripts/make_logo.py        # regenerate logo.png from the artwork

GH_TOKEN (or GITHUB_TOKEN) raises the API limit from 60/h to 5,000/h.

Branding

pythonbackportlogo.png is the source artwork: a flat blue-and-yellow mark on an opaque white background. The site is dark themed, so the raw file would render as a bright white rectangle. scripts/make_logo.py removes the white backdrop (recovering true per-pixel alpha by un-mixing each shape against its brand colour), crops the transparent margin, and writes logo.png. Both the standard library and the favicon use that generated file.

Classification

The script buckets each fetched repo into one of two sections based on the live API response — no lookup tables:

  • Backports — every public repo that isn't the discussion hub.
  • Community — repos whose name, description or topics contain discussion, discussions, or forum.

The repo that hosts this very website is auto-detected and excluded by name match (website, .github, anything equal to the org login) or by description / topics containing website or github-pages.

Community link

GitHub Discussions is not enabled for this org — the org-level page answers 404 and the community repo is archived — so the community call-to-action points at the Discord invite instead. The invite is the DISCORD_INVITE constant in scripts/build.py, which means the daily refresh workflow can never silently revert it to a dead GitHub Discussions link.

Automatic refresh

.github/workflows/refresh.yml runs daily at 06:13 UTC and rebuilds index.html from the live GitHub API. If anything changed, it commits the updated file back to the master branch, and the Pages deploy workflow (.github/workflows/static.yml) publishes the new build.