refresh-data.yaml

# Keeps the two numbers on the homepage that come from GitHub moving on their own:
# the contributor map, regenerated here and committed, and the project star counts,
# which are looked up while Vercel builds the site.
#
# They need different things. The map is a committed snapshot because a new profile
# location has to be matched by a person in tools/locations.json before it can be a
# dot, so this job regenerates it and pushes. The stars are already fetched on every
# build and only want a build to happen — which a push gives, and the Vercel deploy
# hook at the end covers on the days there is nothing to push.
name: Refresh Data

on:
  schedule:
    # Daily, at an odd minute past the hour. Everything scheduled on the hour is
    # queued together, so asking for :00 is asking to be run late.
    - cron: "23 4 * * *"
  # So that a city added to tools/locations.json reaches the map today rather than
  # tomorrow.
  workflow_dispatch:

# data/contributors.json is the only file written, by the one step that commits.
permissions:
  contents: write

# Two of these pushing the same file would be one of them rebasing or failing.
concurrency:
  group: refresh-data

jobs:
  refresh:
    runs-on: ubuntu-latest
    env:
      # Empty unless the secret exists, which is what the last step tests: the
      # rebuild is an extra, and the map must refresh with or without it. `secrets`
      # cannot be read from a step's `if`, so it is lifted to the job here.
      VERCEL_DEPLOY_HOOK: ${{ secrets.VERCEL_DEPLOY_HOOK }}
    steps:
      - uses: actions/checkout@v4

      - uses: actions/setup-node@v4
        with:
          # Pinned rather than `lts/*`: the tools are run here and nowhere else,
          # and a runner quietly moving to a new major is not something a daily
          # unattended job should discover on its own.
          #
          # It is not the version Vercel builds with — Vercel is configured in its
          # own dashboard, not in this repository — so the two can drift. Nothing
          # here depends on them matching: this job runs the generator, and the
          # build only reads the JSON it commits.
          node-version: "24.3.0"

      - name: Regenerate the contributor map
        shell: bash
        env:
          # The default token, which is all this needs: public repositories, read
          # only. `make contributors` falls back to `gh auth token` when it is
          # unset, and there is no `gh` login on a runner.
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        run: |
          make contributors | tee /tmp/refresh.log

          # The locations the table could not place, onto the run's summary page.
          # They are the one thing here that needs a person: each is a line to add
          # to tools/locations.json, and until then it is a contributor with no dot.
          # A log nobody opens would be no notice at all.
          if grep -q "not in tools/locations.json" /tmp/refresh.log; then
            {
              echo "### Locations missing from \`tools/locations.json\`"
              echo
              echo '```'
              sed -n '/not in tools\/locations.json/,$p' /tmp/refresh.log
              echo '```'
            } >> "$GITHUB_STEP_SUMMARY"
          fi

      - name: Commit the new snapshot
        id: commit
        shell: bash
        run: |
          # tools/contributors.mjs leaves the file alone when only its `generated`
          # date would have moved, so a quiet day is a clean tree here rather than
          # a commit that changes one line and means nothing.
          if git diff --quiet -- data/contributors.json; then
            echo "no change to commit"
            exit 0
          fi

          git config user.name "github-actions[bot]"
          git config user.email "41898282+github-actions[bot]@users.noreply.github.com"
          git commit -m "chore: refresh the contributor map" -- data/contributors.json
          git push
          # Read by the next step: this push is itself a Vercel deploy.
          echo "pushed=true" >> "$GITHUB_OUTPUT"

      - name: Ask Vercel to rebuild
        # Only when nothing was pushed. A push already rebuilds the site, star
        # counts and all; this is for the ordinary day where the map has not moved
        # but a repository has gained stars and nothing would otherwise go and look.
        if: steps.commit.outputs.pushed != 'true' && env.VERCEL_DEPLOY_HOOK != ''
        shell: bash
        run: curl --fail --silent --show-error -X POST -d '{}' "$VERCEL_DEPLOY_HOOK"