book-production.yml

name: book-production
run-name: book-production for ${{ inputs.doi }}

on:
  workflow_dispatch:
    inputs:
      doi:
        description: 'DOI of book (last four digits only)'
        required: true
        type: number
      has_pdf:
        description: 'Run PDF workflow'
        default: true
        required: true
        type: boolean
      has_epub:
        description: 'Run EPUB workflow'
        default: true
        required: true
        type: boolean
      run_archive:
        description: 'Run URL archiving'
        default: true
        required: true
        type: boolean
      copyright_page:
        description: 'Copyright page number (default is 4, first page is 0)'
        default: 4
        required: true
        type: number

env:
  WORK_DIR: /ebook_automation/
  OUT_DIR: /ebook_automation/output/
  WORKFLOW_URL: ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}

jobs:

  download-pdf:
    runs-on: ubuntu-latest
    if: ${{ inputs.has_pdf || inputs.run_archive }}
    steps:
      - name: Configure AWS credentials
        uses: aws-actions/configure-aws-credentials@v4
        with:
          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
          aws-region: ${{ secrets.AWS_REGION }}
      - name: Copy PDF from S3
        run: |
          aws s3 cp ${{ secrets.S3_URL }}obp.${{ inputs.doi }}.pdf .
      - name: Save PDF
        uses: actions/upload-artifact@v4
        with:
          name: original-pdf
          path: obp.${{ inputs.doi }}.pdf
          retention-days: 10
          if-no-files-found: error
          overwrite: true

  download-epub:
    runs-on: ubuntu-latest
    if: ${{ inputs.has_epub }}
    steps:
      - name: Configure AWS credentials
        uses: aws-actions/configure-aws-credentials@v4
        with:
          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
          aws-region: ${{ secrets.AWS_REGION }}
      - name: Copy EPUB from S3
        run: |
          aws s3 cp ${{ secrets.S3_URL }}obp.${{ inputs.doi }}.epub .
      - name: Save EPUB
        uses: actions/upload-artifact@v4
        with:
          name: original-epub
          path: obp.${{ inputs.doi }}.epub
          retention-days: 10
          if-no-files-found: error
          overwrite: true

  archive-pdf-urls:
    needs: download-pdf
    runs-on: ubuntu-latest
    if: ${{ inputs.run_archive }}
    steps:
      - name: Install archive-pdf-urls
        # https://github.com/thoth-pub/archive-pdf-urls
        run: cargo install archive-pdf-urls
      - name: Retrieve PDF
        uses: actions/download-artifact@v4
        with:
          name: original-pdf
      - name: Run archive-pdf-urls Rust script
        run: archive-pdf-urls obp.${{ inputs.doi }}.pdf --exclude https://doi.org/10.11647/OBP.${{ inputs.doi }}\*

  chapter-splitter:
    needs: download-pdf
    runs-on: ubuntu-latest
    if: ${{ inputs.has_pdf }}
    container:
      image: openbookpublishers/chapter-splitter
    steps:
      - name: Retrieve Data
        uses: actions/download-artifact@v4
        with:
          name: original-pdf
          path: ${{ env.WORK_DIR }}
      - name: Create Output Folder
        run: mkdir ${{ env.OUT_DIR }}
      - name: Run Python Command
        working-directory: ${{ env.WORK_DIR }}
        run: |
          python3 main.py \
          --input-file obp.${{ inputs.doi }}.pdf \
          --output-folder ${{ env.OUT_DIR }} \
          --write-urls \
          10.11647/obp.${{ inputs.doi }}
        env:
          THOTH_EMAIL: ${{ secrets.THOTH_EMAIL }}
          THOTH_PWD: ${{ secrets.THOTH_PASSWORD }}
          COPYRIGHT_PAGE: ${{ inputs.copyright_page }}
      - name: Save Data
        uses: actions/upload-artifact@v4
        with:
          name: chapters
          path: ${{ env.OUT_DIR }}
          retention-days: 1
          if-no-files-found: error
          overwrite: true

  epublius:
    needs: download-epub
    runs-on: ubuntu-latest
    container:
      image: openbookpublishers/epublius
    steps:
      - name: Retrieve Data
        uses: actions/download-artifact@v4
        with:
          name: original-epub
          path: ${{ env.WORK_DIR }}
      - name: Create Output Folder
        run: mkdir ${{ env.OUT_DIR }}
      - name: Run Python Command
        working-directory: ${{ env.WORK_DIR }}
        run: |
          ./thoth_wrapper.py obp.${{ inputs.doi }}.epub \
                             --doi 10.11647/obp.${{ inputs.doi }}
        env:
          MATHJAX: 'False'
          PRIVACYPOLICY_URL: 'https://www.openbookpublishers.com/policies/privacy/'
          OUTDIR: ${{ env.OUT_DIR }}
          THOTH_EMAIL: ${{ secrets.THOTH_EMAIL }}
          THOTH_PWD: ${{ secrets.THOTH_PASSWORD }}
      - name: Save Data
        uses: actions/upload-artifact@v4
        with:
          name: HTML
          path: ${{ env.OUT_DIR }}/10.11647/
          retention-days: 1
          if-no-files-found: error
          overwrite: true

  upload-chapters:
    needs: chapter-splitter
    runs-on: ubuntu-latest
    env:
      CHAPTERS_DIR: ./chapters
    steps:
      - name: Retrieve Data
        uses: actions/download-artifact@v4
        with:
          name: chapters
          path: ${{ env.CHAPTERS_DIR }}
      - name: Configure AWS credentials
        uses: aws-actions/configure-aws-credentials@v4
        with:
          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
          aws-region: ${{ secrets.AWS_REGION }}
      - name: Copy Chapter PDFs to S3
        run: |
          aws s3 cp ${{ env.CHAPTERS_DIR }} ${{ secrets.S3_URL }} --recursive

  upload-html:
    needs: epublius
    runs-on: ubuntu-latest
    steps:
      - name: Retrieve Data
        uses: actions/download-artifact@v4
        with:
          name: HTML
      - name: Configure AWS credentials
        uses: aws-actions/configure-aws-credentials@v4
        with:
          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}
          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
          aws-region: ${{ secrets.AWS_REGION }}
      - name: Copy HTML Edition to S3
        run: |
          aws s3 cp ./obp.${{ inputs.doi }}/ ${{ secrets.S3_URL }}obp.${{ inputs.doi }}/ --recursive

  cit-ex-obp-loader:
    needs: upload-html
    runs-on: ubuntu-latest
    container:
      image: openbookpublishers/cit-ex-obp-loader
    env:
      WORK_DIR: /home/obp/cit-ex/
    steps:
      - name: Run Python Command
        working-directory: ${{ env.WORK_DIR }}
        run: |
          python3 obp-loader.py 10.11647/obp.${{ inputs.doi }}
        env:
          THOTH_EMAIL: ${{ secrets.THOTH_EMAIL }}
          THOTH_PWD: ${{ secrets.THOTH_PASSWORD }}

  status:
    runs-on: ubuntu-latest
    needs: [upload-chapters, cit-ex-obp-loader]
    if: always()
    steps:
      - name: Set up job status summary
        id: generate_summary
        run: |
          get_status_emoji() {
            local status=$1
            if [ "$status" == "success" ]; then
              echo ":large_green_circle:"
            elif [ "$status" == "failure" ]; then
              echo ":red_circle:"
            else
              echo ":white_circle:"
            fi
          }

          overall_status() {
            local statuses=("$@")
            local overall="success"
            local all_skipped_or_cancelled=true

            for status in "${statuses[@]}"; do
              if [ "$status" == "failure" ]; then
                overall="failure"
                all_skipped_or_cancelled=false
                break
              elif [ "$status" == "success" ]; then
                all_skipped_or_cancelled=false
              fi
            done

            if [ "$all_skipped_or_cancelled" = true ]; then
              overall="skipped"
            fi

            echo $overall
          }

          PDF_WORKFLOW_STATUS=$(overall_status "${{ needs.download-pdf.result }}" "${{ needs.chapter-splitter.result }}" "${{ needs.upload-chapters.result }}")
          EPUB_WORKFLOW_STATUS=$(overall_status "${{ needs.download-epub.result }}" "${{ needs.epublius.result }}" "${{ needs.upload-html.result }}" "${{ needs.cit-ex-obp-loader.result }}")

          echo "| Workflow | Status |   |" >> summary.md
          echo "| -------- | ------ | - |" >> summary.md
          echo "| PDF Workflow | $PDF_WORKFLOW_STATUS | $(get_status_emoji $PDF_WORKFLOW_STATUS) |" >> summary.md
          echo "| EPUB Workflow | $EPUB_WORKFLOW_STATUS | $(get_status_emoji $EPUB_WORKFLOW_STATUS) |" >> summary.md

      - name: Write summary to GITHUB_STEP_SUMMARY
        run: cat summary.md >> $GITHUB_STEP_SUMMARY

      - name: Read summary into environment variable
        id: summary
        run: |
          echo "SUMMARY<<EOF" >> $GITHUB_ENV
          cat summary.md >> $GITHUB_ENV
          echo "EOF" >> $GITHUB_ENV

      - name: Send notification to Mattermost
        uses: mattermost/action-mattermost-notify@master
        with:
          MATTERMOST_WEBHOOK_URL: ${{ secrets.MM_WEBHOOK_URL }}
          TEXT: |
            #### Book production workflow for ${{ inputs.doi }}
            @channel please review the [workflow run](${{ env.WORKFLOW_URL }}).

            ${{ env.SUMMARY }}

  report-pdf-archiving-failure:
    runs-on: ubuntu-latest
    needs: archive-pdf-urls
    if: failure()
    steps:
      - uses: mattermost/action-mattermost-notify@master
        with:
          MATTERMOST_WEBHOOK_URL: ${{ secrets.MM_WEBHOOK_URL }}
          TEXT: |
            #### :red_circle: Archive PDF URLs for ${{inputs.doi}} has failed
            @channel please review the [workflow run](${{ env.WORKFLOW_URL }})