test_suite.yml

name: CORE Test Suite Validation
on:
  workflow_dispatch:

env:
  DATASET_SIZE_THRESHOLD: 1000000000
  PANDAS_DIFF: False
  DASK_DIFF: False

jobs:
  validate:
    runs-on: ubuntu-latest
    steps:
      - name: Get Engine Changes
        uses: actions/checkout@v3

      - name: Get Test Suite Repository
        uses: actions/checkout@v3
        with:
          repository: cdisc-org/CORE_Test_Suite
          path: CORE_Test_Suite
          token: ${{ secrets.ACCESS_TOKEN }}

      - name: Set up Python
        uses: actions/setup-python@v4
        with:
          python-version: "3.10"

      - name: Install dependencies
        run: |
          pip install -r requirements.txt
      - name: Update rules cache
        env:
          CDISC_LIBRARY_API_KEY: ${{ secrets.CDISC_LIBRARY_API_KEY }}
        run: |
          python core.py update-cache
      - name: Parse rule list and run validation
        run: |
          rules=$(cat CORE_Test_Suite/rulelist/testrulelist.txt | sed 's/^/-r /' | tr '\n' ' ')
          python core.py validate -s sdtmig -v 3-3 $rules -d CORE_Test_Suite/data -ct sdtmct-2020-03-27 -o CORE_Test_Suite/pandas-results
      - name: Pandas comparison script
        run: |
          cd CORE_Test_Suite
          result=$(python comparison.py pandas-results.xlsx CORE-Report.xlsx pandas_comparison.xlsx)
          echo "PANDAS_DIFF=$result" >> $GITHUB_ENV
          echo "Pandas comparison result: $result"
      - name: Run validation with Dask
        env:
          DATASET_SIZE_THRESHOLD: 0
        run: |
          rules=$(cat CORE_Test_Suite/rulelist/testrulelist.txt | sed 's/^/-r /' | tr '\n' ' ')
          python core.py validate -s sdtmig -v 3-3 $rules -d CORE_Test_Suite/data -ct sdtmct-2020-03-27 -o CORE_Test_Suite/dask-results
      - name: Dask comparison script
        run: |
          cd CORE_Test_Suite
          result=$(python comparison.py dask-results.xlsx DASK-Report.xlsx dask_comparison.xlsx)
          echo "DASK_DIFF=$result" >> $GITHUB_ENV
          echo "Dask comparison result: $result"
          cd ..
      - name: Upload test results
        if: always()
        uses: actions/upload-artifact@v4
        with:
          name: comparison-results
          path: |
            CORE_Test_Suite/pandas_comparison.xlsx
            CORE_Test_Suite/pandas-results.xlsx
            CORE_Test_Suite/dask_comparison.xlsx
            CORE_Test_Suite/dask-results.xlsx
          if-no-files-found: warn
      - name: Check for differences
        if: always()
        run: |
          if [[ "${{ env.PANDAS_DIFF }}" == "true" || "${{ env.DASK_DIFF }}" == "true" ]]; then
            echo "Differences found in either Pandas or Dask comparison"
            exit 1
          elif [[ "${{ env.PANDAS_DIFF }}" == "false" && "${{ env.DASK_DIFF }}" == "false" ]]; then
            echo "No differences found in either comparison"
            exit 0
          else
            echo "Issue with comparison script"
            exit 1
          fi