name: Build Hugo

on:
  push:
    branches: [ "main" ]

env:
  REGISTRY: tea.ahmza.com
  IMAGE_NAME: ${{ gitea.repository }}

jobs:
  build:
    runs-on: ubuntu-latest

    permissions:
      contents: read
      packages: write
      attestations: write
      id-token: write

    steps:
      - name: Checkout Repo
        uses: actions/checkout@v4

      - name: Login to Gitea Container Registry
        uses: docker/login-action@v3
        with:
          registry: ${{ env.REGISTRY }}
          username: ${{ gitea.repository_owner }}
          password: ${{ secrets.GITEATOKEN }}

      - name: Set up QEMU
        uses: docker/setup-qemu-action@v3

      - name: Set up Docker Buildx
        uses: docker/setup-buildx-action@v3

      - name: Build and push
        uses: docker/build-push-action@v6
        with:
          context: .
          platforms: linux/arm64, linux/amd64
          push: true
          build-args: |
            BUILDKIT_CONTEXT_KEEP_GIT_DIR=true
          tags: |
            ${{ env.REGISTRY }}/${{ gitea.repository_owner }}/blog-ahmza-com:latest
            ${{ env.REGISTRY }}/${{ gitea.repository_owner }}/blog-ahmza-com:${{ gitea.event.head_commit.id }}
          # Registry-backed build cache. The GitHub-hosted 'gha' cache is not
          # reachable from self-hosted Gitea runners, so store the cache as a
          # dedicated tag in the same registry instead.
          cache-from: type=registry,ref=${{ env.REGISTRY }}/${{ gitea.repository_owner }}/blog-ahmza-com:buildcache
          cache-to: type=registry,ref=${{ env.REGISTRY }}/${{ gitea.repository_owner }}/blog-ahmza-com:buildcache,mode=max
  deploy:
    runs-on: ubuntu-latest
    needs: build

    steps:
      - name: Pre-flight - verify VPN prerequisites
        # NetBird brings up a WireGuard interface, which needs a TUN device and
        # NET_ADMIN inside the job container. Without them, `netbird up` hangs
        # ~50s and dies with a cryptic "DeadlineExceeded". Check up front so the
        # failure is obvious and points at the fix (runner config.yaml).
        run: |
          echo "::group::NetBird prerequisites"
          if [ -e /dev/net/tun ]; then
            echo "OK: /dev/net/tun present"
          else
            echo "::error title=Missing TUN device::/dev/net/tun not found. Add --device=/dev/net/tun to the runner's config.yaml container.options."
            exit 1
          fi
          if command -v capsh >/dev/null 2>&1; then
            capsh --print | grep -q 'cap_net_admin' \
              && echo "OK: NET_ADMIN present" \
              || echo "::warning title=NET_ADMIN missing::netbird up will likely time out (DeadlineExceeded) configuring wt0. Add --cap-add=NET_ADMIN to the runner's config.yaml container.options."
          else
            echo "note: capsh unavailable, skipping NET_ADMIN check"
          fi
          echo "::endgroup::"

      - name: Netbird Connect
        id: netbird
        timeout-minutes: 3
        uses: Alemiz112/netbird-connect@v1
        with:
          setup-key: ${{ secrets.NETBIRD_SETUP_KEY }}
          hostname: 'gitea-actions-blog'
          management-url: ${{ secrets.NETBIRD_URL }}

      - name: NetBird status
        # Surface the actual peer/route state in the log and hard-assert that
        # the management connection is up before attempting any deploys.
        run: |
          echo "::group::netbird status"
          sudo netbird status --detail || true
          echo "::endgroup::"
          sudo netbird status | grep -q 'Management: Connected' \
            && echo "NetBird management connected." \
            || { echo "::error title=NetBird not connected::Management not connected after netbird up."; exit 1; }

      - name: Deploy to RPIC-Cluster
        # RPIC cluster is temporarily out of service. Don't let its failure
        # block the Gimli deploy or fail the whole job. Remove this line once
        # the cluster is back.
        continue-on-error: true
        uses: appleboy/ssh-action@v1
        with:
          host: ${{ secrets.RPIC_HOST }}
          username: ${{ secrets.SSH_USER }}
          key: ${{ secrets.RPIC_KEY }}
          port: ${{ secrets.SSH_PORT }}
          script: 'kubectl rollout restart deployment blog -n blog-ahmza-com'

      - name: Deploy to Gimli
        uses: appleboy/ssh-action@v1
        with:
          host: ${{ secrets.GIMLI_HOST }}
          username: ${{ secrets.SSH_USER }}
          key: ${{ secrets.GIMLI_KEY }}
          port: ${{ secrets.SSH_PORT }}
          # Pull first, then reconcile in place with `up -d`. Avoid `compose down`:
          # it removes caddy_cf and the default network, which severs the SSH
          # session's own path before `up` can restore it (intermittent
          # "remote command exited without exit status"). `up -d` only recreates
          # containers whose image/config changed and never tears the network out
          # from under the connection. --force-recreate ensures the new :latest
          # image is picked up even when the tag digest changed.
          script: 'docker compose pull && docker compose up -d --force-recreate --remove-orphans'

      - name: NetBird diagnostics & disconnect
        # Runs even on failure: captures final state for debugging and tears the
        # peer down so stale 'gitea-actions-blog' peers don't pile up.
        if: always()
        run: |
          echo "::group::final netbird status"
          sudo netbird status --detail || echo "netbird unavailable"
          echo "::endgroup::"
          sudo netbird down || true