name: Deploy test site

# Builds site/dist (site/public/ plus demo pages copied out of docs/, see
# site/build.sh) and publishes it to https://cnxvinyl.lol/ on the
# buildfor.life web server (s1), the same Caddy that serves the blog.
# One-time setup (DNS, Caddy vhost, secrets, deploy key): site/README.md.
#
# Secrets (repo or user level):
#   VPS_HOST, VPS_USER, VPS_SSH_KEY   required; same server as the blog
#   VPS_PORT                          optional, default 22
#   VPS_KNOWN_HOSTS                   optional pinned host key line(s);
#                                     without it the key is scanned per run

on:
  push:
    branches: [master]
    paths:
      - "site/**"
      - "docs/**/*.html"
      - "docs/pitch/**"
      - ".gitea/workflows/deploy-site.yml"
  workflow_dispatch:

# A newer push supersedes a running deploy. Safe to cancel mid-run: rsync
# writes the non-live directory and the rename swap happens last.
concurrency:
  group: deploy-cnxvinyl-lol
  cancel-in-progress: true

jobs:
  deploy:
    runs-on: ubuntu-latest
    env:
      VPS_HOST: ${{ secrets.VPS_HOST }}
      VPS_USER: ${{ secrets.VPS_USER }}
      VPS_PORT: ${{ secrets.VPS_PORT }}
      SITE_DIR: /var/www/cnxvinyl.lol
      SITE_URL: https://cnxvinyl.lol
    steps:
      - name: Checkout
        uses: actions/checkout@v4

      - name: Build site/dist
        run: bash site/build.sh

      - name: Publish to the package registry (bkk01)
        # bkk01 serves cnxvinyl.lol from releases in the B4L package registry
        # (cnx-network-clan, docs/src/s1-migration.md): the package event
        # makes it fetch this version within seconds. Until DNS moves to
        # bkk01, s1 below stays the live server and gets the same build.
        # First, so a failing s1 step cannot hold back bkk01's release.
        # REGISTRY_USER / REGISTRY_TOKEN: organization secrets, an account
        # with a write:package token in a B4L team that may write packages.
        env:
          REGISTRY_USER: ${{ secrets.REGISTRY_USER }}
          REGISTRY_TOKEN: ${{ secrets.REGISTRY_TOKEN }}
        run: |
          if [ -z "$REGISTRY_USER" ] || [ -z "$REGISTRY_TOKEN" ]; then
            echo "::error::REGISTRY_USER and REGISTRY_TOKEN must be set as organization secrets"
            exit 1
          fi
          version="$(date -u +%Y%m%dT%H%M%S)-${GITHUB_SHA::12}"
          tar -czf site.tar.gz -C site/dist .
          curl -fsS --user "$REGISTRY_USER:$REGISTRY_TOKEN" --upload-file site.tar.gz \
            "${{ github.server_url }}/api/packages/${{ github.repository_owner }}/generic/cnxvinyl.lol/$version/site.tar.gz"
          echo "published cnxvinyl.lol $version ($(du -h site.tar.gz | cut -f1))"

      - name: Set up SSH
        env:
          VPS_SSH_KEY: ${{ secrets.VPS_SSH_KEY }}
          VPS_KNOWN_HOSTS: ${{ secrets.VPS_KNOWN_HOSTS }}
        run: |
          if [ -z "$VPS_HOST" ] || [ -z "$VPS_USER" ] || [ -z "$VPS_SSH_KEY" ]; then
            echo "::error::VPS_HOST, VPS_USER and VPS_SSH_KEY must be set as Actions secrets (see site/README.md)"
            exit 1
          fi
          port=${VPS_PORT:-22}
          # ssh refuses a config that is group/world-writable or owned by
          # another user ("Bad owner or permissions on ~/.ssh/config"), which
          # a runner with a loose umask or a reused HOME produces: create every
          # file here private and fresh.
          umask 077
          mkdir -p ~/.ssh && chmod 700 ~/.ssh
          rm -f ~/.ssh/config
          printf '%s\n' "$VPS_SSH_KEY" | tr -d '\r' > ~/.ssh/deploy_key
          chmod 600 ~/.ssh/deploy_key
          # Fail here, not at the server, when the secret is not a usable key,
          # and log its fingerprint (public) so it can be compared with
          # `ssh-keygen -lf ~deploy/.ssh/authorized_keys` on the server.
          if ! ssh-keygen -y -P "" -f ~/.ssh/deploy_key > ~/.ssh/deploy_key.pub 2>/dev/null; then
            echo "::error::VPS_SSH_KEY does not load as an unencrypted private key; paste the whole OpenSSH-format key including the BEGIN/END lines"
            exit 1
          fi
          echo "key in VPS_SSH_KEY: $(ssh-keygen -lf ~/.ssh/deploy_key.pub)"
          if [ -n "$VPS_KNOWN_HOSTS" ]; then
            printf '%s\n' "$VPS_KNOWN_HOSTS" >> ~/.ssh/known_hosts
          else
            for i in 1 2 3; do
              ssh-keyscan -T 10 -p "$port" "$VPS_HOST" > ~/.ssh/keyscan.out && [ -s ~/.ssh/keyscan.out ] && break
              echo "ssh-keyscan attempt $i failed; retrying in 15 s" >&2
              sleep 15
            done
            if [ ! -s ~/.ssh/keyscan.out ]; then
              echo "::error::ssh-keyscan could not reach $VPS_HOST:$port (server down or runner IP blocked?)"
              exit 1
            fi
            cat ~/.ssh/keyscan.out >> ~/.ssh/known_hosts
          fi
          {
            echo "Host webhost"
            echo "  HostName $VPS_HOST"
            echo "  User $VPS_USER"
            echo "  Port $port"
            echo "  IdentityFile ~/.ssh/deploy_key"
            echo "  IdentitiesOnly yes"
            echo "  BatchMode yes"
          } > ~/.ssh/config
          chmod 600 ~/.ssh/config ~/.ssh/known_hosts

      - name: Ensure rsync is available
        run: command -v rsync || (sudo apt-get update && sudo apt-get install -y rsync) || (apt-get update && apt-get install -y rsync)

      - name: Deploy (rsync to the idle copy, then rename swap)
        # Two persistent directories alternate as live and previous, as for
        # the blog: rsync only ever writes the one that is not live, and the
        # swap is three same-filesystem renames, so Caddy never serves a
        # half-written tree. The previous build stays on disk for rollback.
        run: |
          # SC2029: the paths are meant to expand on the runner side
          # shellcheck disable=SC2029
          ssh webhost "mkdir -p '$SITE_DIR' '$SITE_DIR-prev' && rm -rf '$SITE_DIR-tmp'"
          rsync -az --delete site/dist/ "webhost:$SITE_DIR-prev/"
          # shellcheck disable=SC2029
          ssh webhost "mv '$SITE_DIR' '$SITE_DIR-tmp' && mv '$SITE_DIR-prev' '$SITE_DIR' && mv '$SITE_DIR-tmp' '$SITE_DIR-prev'"
          # shellcheck disable=SC2029
          live=$(ssh webhost "cat '$SITE_DIR/version.txt'")
          echo "live on server: $live"
          [ "$live" = "$(cat site/dist/version.txt)" ] || { echo "::error::server holds $live, expected $(cat site/dist/version.txt)"; exit 1; }

      - name: Check the public URL
        # Warning only: DNS or a basic_auth login can make this
        # fail while the files on the server are already correct.
        run: |
          expected=$(cat site/dist/version.txt)
          got=$(curl -sS --max-time 20 "$SITE_URL/version.txt?nocache=$expected" || true)
          if [ "$got" = "$expected" ]; then
            echo "$SITE_URL serves $expected"
          else
            echo "::warning::$SITE_URL/version.txt returned '${got:0:80}', expected $expected"
          fi