testing.yml

name: Python testing

on: [push, workflow_dispatch]

jobs:
  test:
    env:
      # Have the venv be in the project folder so it is easier to cache
      PIPENV_VENV_IN_PROJECT: true
    runs-on: ${{ matrix.os }}
    strategy:
      fail-fast: false
      matrix:
        os: [ubuntu-latest, windows-latest, macOS-latest]
        python_version: [3.9]
    defaults:
      run:
        shell: bash
    steps:
    - uses: actions/checkout@v3
    - name: Set up Python
      uses: actions/setup-python@v4
      with:
        python-version: ${{ matrix.python_version }}
    - name: Install pipenv
      run: |
        python -m pip install --upgrade pip
        pip install pipenv
    # Cache the venv between runs. Duplicated from linting.yaml
    - name: Cache virtualenv
      id: cache-venv
      uses: actions/cache@v3
      env:
        CACHE_NAME: CACHE_VIRTUALENV
      with:
        path: .venv
        # The cache action does not as of now (Oct. 2020) support clearing the cache.
        # If the cache becomes stale for any reason one can clear it by changing
        # the value of the secret `PIPENV_CACHE_VERSION` in the repository settings
        key: >
          cache_version_${{ secrets.PIPENV_CACHE_VERSION}}-${{ runner.os }}-${{ env.CACHE_NAME }}-python${{ matrix.python_version }}-${{ hashFiles('Pipfile.lock') }}
        # Fallbacks to partially restore cache
        restore-keys: |
          cache_version_${{ secrets.PIPENV_CACHE_VERSION}}-${{ runner.os }}-${{ env.CACHE_NAME }}-python${{ matrix.python_version }}-

    - name: Install dependencies
      if: steps.cache-venv.outputs.cache-hit != 'true'
      run: |
        # Use copies instead of symlinks to be more portable
        python -m venv --copies .venv
        # Install dependencies into the newly created venv
        pipenv install --deploy --dev

    - name: Allow reading and writing PDFs (linux only)
      # Edit imagemagick's policy file to allow reading and writing PDFs
      # This is needed to create Nablad instances, which is done during testing
      # Also install ghostscript, which imagemagick uses to read/write PDFs
      if: runner.os == 'Linux'
      run: |
        configure_path=$(convert -list configure | sed -n -e 's/CONFIGURE_PATH[[:space:]]*//p' || true)
        temp_file="$(sudo mktemp)"

        # sed -i is a gnu extension. Beware
        sudo sed -i -e 's/<policy domain="coder" rights=".*" pattern="PDF" \/>/<policy domain="coder" rights="read | write" pattern="PDF" \/>/w '"$temp_file" "$configure_path/policy.xml"
        if [ ! -s "$temp_file" ]; then
          echo 'Failed to edit policy file to allow reading/writing PDFs!'
          echo 'Below are the contents of `policy.xml`'
          echo 'For PDF processing to work it should be edited to have a line like:'
          echo '  "<policy domain="coder" rights="read | write" pattern="PDF" \/>"'
          echo
          cat "$configure_path/policy.xml"
          exit 1
        fi
        sudo apt-get update
        sudo apt-get install ghostscript

    - name: Exclude tests that need ImageMagick (non-linux)
      # ImageMagick does not come pre-installed on mac- and windows runners, so we skip
      # tests that need it on those hosts. This is done by excluding a specific test tag,
      # which is set on the test that needs ImageMagick
      if: runner.os != 'Linux'
      run: |
        echo 'exclude_tag=needs_imagemagick' >> $GITHUB_ENV

    - name: Run tests
      # Tests are run in parallel, which may cause issues if the tests aren't propertly isolated
      run: |
        pipenv run python manage.py test --parallel --exclude-tag="$exclude_tag"