run_all_frameworks.yml

name: Run All Frameworks

on: pull_request


jobs:

  detect_changes:  # Figure out which frameworks need to be evaluated based on which files are changed.
    name: Detect Changes
    runs-on: ubuntu-latest
    outputs:
      frameworks: ${{ steps.find-required-tests.outputs.frameworks }}
      tasks: ${{ steps.find-required-tests.outputs.tasks }}
      benchmark: ${{ steps.find-required-tests.outputs.benchmark }}
      skip_baseline: ${{ steps.find-required-tests.outputs.skip_baseline }}
      skip_evaluation: ${{ steps.find-required-tests.outputs.skip_evaluation }}
    steps:
      - uses: actions/checkout@v3
      - name: pull base branch
        run: |
          git fetch --unshallow origin $GITHUB_BASE_REF
          git branch --track $GITHUB_BASE_REF origin/$GITHUB_BASE_REF
      - id: find-required-tests
        name: Detect Common Changes  # detect if any changes occurred that should trigger all frameworks.
        run: |
          changed_files=$(git diff --name-only $GITHUB_BASE_REF...HEAD)

          shopt -s globstar
          common_files="amlb/** "\
          "resources/** "\
          "frameworks/shared/** "\
          ".github/workflows/run_all_frameworks.yml "\
          ".github/runbenchmark/action.yml "\
          "runbenchmark.py "\
          "requirements.txt"
          
          echo Common files: $common_files
          echo Changed files: $changed_files

          is_common=1
          for f in $changed_files
          do
            if echo $common_files | grep -e $f >> /dev/null;
            then
                echo "File $f detected as common file"
                is_common=0
                break
            fi
          done

          # Indicates which jobs should be executed (0) or not (1)
          skip_evaluation=0
          if [[ $is_common -eq 0 ]];
          then
            FRAMEWORKS='["autogluon", "autosklearn", "gama", "h2oautoml", "mlplanweka", "mlr3automl", "naiveautoml", "tpot", "tunedrandomforest"]'
            TASKS='["iris", "kc2", "cholesterol"]'
            BENCHMARK='["test"]'
          else
            TASKS='["APSFailure", "bioresponse", "dresses-sales", "eucalyptus", "internet-advertisements", "kc1", "micro-mass"]'
            BENCHMARK='["validation"]'
            changed_frameworks=$(git diff --name-only HEAD..$GITHUB_BASE_REF | grep -o -i -P 'frameworks/(?!shared).*/' | uniq | sed -e 's/frameworks//' -e 's/\///g')
            if [ ! -z "$changed_frameworks" ];
            then
              json_array=[
              for framework in $changed_frameworks; do json_array=$json_array\"$framework\",; done
              FRAMEWORKS=${json_array::-1}]  #remove trailing comma and add closing bracket
            else
              # No changes to common files or frameworks - must be e.g. docs. No need to run tests.
              skip_evaluation=1
              FRAMEWORKS=[]
            fi
          fi

          echo Building matrix for frameworks: $FRAMEWORKS

          echo "::set-output name=frameworks::$FRAMEWORKS"
          echo "::set-output name=tasks::$TASKS"
          echo "::set-output name=benchmark::$BENCHMARK"
          echo "::set-output name=skip_baseline::$is_common"
          echo "::set-output name=skip_evaluations::$skip_evaluation"

  baseline:
    name: ${{ matrix.framework }}/${{ matrix.task }}
    runs-on: ubuntu-latest
    needs: detect_changes
    if: needs.detect_changes.outputs.skip_baseline == 0
    strategy:
      matrix:
        framework: [constantpredictor, randomforest]
        task: [iris, kc2, cholesterol]
      fail-fast:  false
    steps:
      - uses: actions/checkout@v3
      - name: Setup Python 3.9
        uses: actions/setup-python@v4
        with:
          python-version: '3.9'
      - name: Create venv
        run: python -m venv venv
      - uses: actions/cache@v3
        id: cache
        with:
          path: /home/runner/work/automlbenchmark/automlbenchmark/venv/lib/python3.9/site-packages
          key: pip-v3-${{ hashFiles('**/requirements.txt') }}
          restore-keys: |
            pip-v3-
      - name: Install Requirements
        if: steps.cache.outputs.cache-hit != 'true'
        run: |
          source venv/bin/activate
          python -m pip install --upgrade pip
          pip install -r requirements.txt
          pip show openml
      - name: Run constantpredictor on openml iris
        run: |
          source venv/bin/activate
          python runbenchmark.py ${{ matrix.framework }} -t ${{ matrix.task }} -f 0 -e

  run_frameworks:
    name: ${{ matrix.framework }}/${{ matrix.task }}
    runs-on: ubuntu-latest
    needs:
     - baseline
     - detect_changes
    if: ${{ success() }} || ${{ cancelled() }}

    strategy:
      matrix:
        python-version: [3.9]
        framework: ${{ fromJson(needs.detect_changes.outputs.frameworks) }}
        task: ${{ fromJson(needs.detect_changes.outputs.tasks) }}
        benchmark: ${{ fromJson(needs.detect_changes.outputs.benchmark) }}
      fail-fast:  true  # not sure about this one, but considering the big workload it might be nicer

    steps:
    - uses: actions/checkout@v3
    - name: Setup Python ${{ matrix.python-version }}
      uses: actions/setup-python@v4
      with:
        python-version: ${{ matrix.python-version }}
    - name: Create venv
      run: python -m venv venv
    - uses: actions/cache@v3
      id: cache
      with:
        path: /home/runner/work/automlbenchmark/automlbenchmark/venv/lib/python3.9/site-packages
        key: pip-v3-${{ hashFiles('**/requirements.txt') }}
        restore-keys: |
          pip-v3-
    - name: Install Requirements
      if: steps.cache.outputs.cache-hit != 'true'
      run: |
        source venv/bin/activate
        python -m pip install --upgrade pip
        python -m pip install -r requirements.txt
    - name: Run ${{ matrix.framework }} on ${{ matrix.task }}
      run: |
        source venv/bin/activate
        python runbenchmark.py ${{ matrix.framework }} ${{ matrix.benchmark }} test -f 0 -t ${{ matrix.task }} -e
      env:
        GITHUB_PAT: ${{ secrets.PUBLIC_ACCESS_GITHUB_PAT }}