publish.yml

name: Publish

on:
  workflow_dispatch:
  push:
    branches:
      - "main"

concurrency:
  group: ${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

jobs:
  main:
    name: "Publish docs.ankidroid.org"
    if: github.repository == 'ankidroid/ankidroiddocs'
    runs-on: ubuntu-latest
    steps:
      - name: SSH key setup
        uses: webfactory/ssh-agent@v0.9.0
        with:
          ssh-private-key: ${{ secrets.DOCS_SSH_PRIVATE_KEY }}

      - name: SSH server key setup
        run: |
          ssh-keyscan ankidroid.org >> ~/.ssh/known_hosts
        shell: bash

      - name: Git Checkout
        uses: actions/checkout@v4

      - name: Prepare website directory
        run: |
          mkdir _site
          cp -r img _site/
          cp icons/* _site/

      - name: Install Asciidoctor
        run: |
          sudo chown -R $USER /var/lib/gems/
          sudo chown -R $USER /usr/local/bin
          gem install asciidoctor
        shell: bash

      - name: Render Docs
        run: asciidoctor ./*.asc -D _site/

      - name: Copy Docs to ankidroid.org
        run: scp -r _site/* ankidroid@ankidroid.org:/usr/share/nginx/html/
        shell: bash

      - name: Restart nginx web server
        #  - why? nginx used to fail periodically requiring a restart. But who has access? How to restart?
        # This alloww people with publish access a button-push way to force a restart if necessary,
        # and the restart is quick enough that doing it on publish is fine.
        run: ssh ankidroid@ankidroid.org /usr/bin/sudo /usr/sbin/service nginx restart