update-data-json.yml

on:
  workflow_dispatch:
  schedule:
    - cron: "30 10,14,19 * * *"

name: Check for metadata changes

jobs:
  check_data:
    runs-on: ubuntu-latest
    outputs:
      COMMIT_CHANGES: ${{ steps.rscript.outputs.COMMIT_CHANGES}}
      POST_BSKY: ${{ steps.rscript.outputs.POST_BSKY}}
      POST_TEXT: ${{ steps.rscript.outputs.POST_TEXT}}
    steps:
      - uses: actions/checkout@v6
    
      # Install R and dependencies
      - uses: r-lib/actions/setup-r@v2
      - uses: r-lib/actions/setup-r-dependencies@v2
        with:
          packages: |
            censusapi

      # Run R script to grab metadata and compare against saved .csv
      - name: Run update-metadata.R
        id: rscript
        run: |
          Rscript scripts/update-metadata.R

      - name: List key vars
        run: |
          echo "Did endpoints.csv change: ${{env.UPDATED_DATA}}"
          echo "Should changes be committed: ${{env.COMMIT_CHANGES}}"
          echo "Should it post on bsky (if changes are committed): ${{env.POST_BSKY}}"
          echo "What should the post text be: ${{env.POST_TEXT}}"

      # Commit using message text prepared in R
      # If anyone is adapting this code make sure to change the user.name and user.email to your own!
      - name: Commit changes
        if: ${{ env.COMMIT_CHANGES == 'true' }} 
        run: |
          # Configure git user
          git config user.name "hrecht[bot]"
          git config user.email "9753760+hrecht[bot]@users.noreply.github.com"

          # Commit files
          git add .

          git commit -m "${{env.COMMIT_MESSAGE}}" || exit 0
          git push

      # Pause a little before deploying to github pages
      # Otherwise it's happening too quickly to work
      - name: Build pause
        if: ${{ env.COMMIT_CHANGES == 'true' }} 
        run: |
          echo "Pause before building site"
          sleep 45

  # Build the page only if changes were committed
  build_site:
   needs: check_data
   if: ${{ needs.check_data.outputs.COMMIT_CHANGES == 'true' }} 
   uses: ./.github/workflows/deploy-site.yml
   secrets: inherit

  # Post to Bluesky based on R output, after site built
  post_bluesky:
    runs-on: ubuntu-latest
    needs: [check_data, build_site]
    if: ${{ needs.check_data.outputs.POST_BSKY == 'true' && needs.check_data.outputs.COMMIT_CHANGES == 'true' }} 
    steps:
        - name: Set up end of post text
          run: |
            echo
            POST_END=$(cat << EOF
             
            See the full tracker: https://www.hrecht.com/census-api-datasets/
            EOF
            )
            echo "POST_END<<EOF" >> $GITHUB_ENV
            echo "$POST_END" >> $GITHUB_ENV
            echo "EOF" >> $GITHUB_ENV

        # Pause a little before posting to give time for page to build
        - name: Bluesky pause
          run: |
            echo "Pause before posting to Bluesky"
            echo "Post text: ${{needs.check_data.outputs.POST_TEXT}}"
            sleep 30
  
        - name: Send post
          uses: myConsciousness/bluesky-post@v5
          with:
              text: ${{ needs.check_data.outputs.POST_TEXT }} ${{ env.POST_END }}
              link-preview-url: "https://www.hrecht.com/census-api-datasets/"
              identifier: ${{ secrets.BLUESKY_IDENTIFIER }}
              password: ${{ secrets.BLUESKY_PASSWORD }}