create_arche_ttl.yml

name: Dump Baserow tables and create ARCHE turtle files

on:
  workflow_dispatch:
    inputs:
      database_id:
        type: number
        description: '(Required) Database ID as number'
        required: true
        default: 0
      project_name:
        type: string
        description: '(Required) Project Name as string'
        required: true
        default: ''
      resource_files_url:
        type: string
        description: '(Optional) Latest Release of Resource Files as URL'
        required: false
        default: ''
      primary_data_directory:
        type: string
        description: '(Optional) Primary Data Directory Path as string'
        required: false
        default: ''
      primary_file_format:
        type: string
        description: '(Optional) Primary File Format as string'
        required: false
        default: 'xml'
      resource_files_url_2:
        type: string
        description: '(Optional) Latest Release of Additional Resource Files as URL'
        required: false
        default: ''
      secondary_data_directory:
        type: string
        description: '(Optional) Secondary Data Directory Path as string'
        required: false
        default: ''
      secondary_file_format:
        type: string
        description: '(Optional) Secondary File Format as string'
        required: false
        default: 'xml'
      

jobs:
  update_baserow:
    name: Dump Baserow tables and create ARCHE turtle files
    runs-on: ubuntu-latest
    env:
      PRIMARY_FILE_FORMAT: ${{ github.event.inputs.primary_file_format }}
      LATEST_RELEASE: ${{ github.event.inputs.resource_files_url }}
      PRIMARY_DATA_DIRECTORY: ${{ github.event.inputs.primary_data_directory }}
      ADDITIONAL_FILE_FORMAT: ${{ github.event.inputs.secondary_file_format }}
      ADDITIONAL_RESOURCES: ${{ github.event.inputs.resource_files_url_2 }}
      ADDITIONAL_DATA_DIRECTORY: ${{ github.event.inputs.secondary_data_directory }}
      PROJECT_NAME: ${{ github.event.inputs.project_name }}
      BASEROW_TOKEN: ${{ secrets.BASEROW_TOKEN }}
      BASEROW_DB_ID: ${{ github.event.inputs.database_id }}
      BASEROW_USER: ${{ secrets.BASEROW_USER }}
      BASEROW_PW: ${{ secrets.BASEROW_PW }}
    steps:
      - name: Check out source repository
        uses: actions/checkout@v3
      - name: Set up Python environment
        uses: actions/setup-python@v4
        with:
          python-version: "3.10"
      - name: Install dependencies
        run: |
          python -m pip install --upgrade pip
          pip install -r requirements.txt
      - name: Baserow Dump and TEI/XML Download
        run: |
          python scripts/baserow_dump.py
      - name: Adding attributes to TEI root node
        run: |
          ./shellscripts/add_attributes.sh
          ./shellscripts/denormalize.sh
      - name: Run ant to add xml-model schema
        run: |
          apt-get update && apt-get install openjdk-11-jre-headless ant -y --no-install-recommend
          ./shellscripts/dl_saxon.sh
          ant
      - name: Creating ARCHE constants ttl file
        run: |
          python scripts/arche_constants.py
      # - name: cache AV database
      #   id: avdb
      #   uses: actions/cache@v3
      #   with:
      #     path: ~/.cvdupdate
      #     key: constant
      # - name: refresh AV database
      #   run: |
      #     python3 -m pip install --user cvdupdate && cvd update
      # - name: Verify Binaries with repo-filechecker
      #   run: |
      #     ./shellscripts/filechecker.sh
      # - name: Verify RDF Metadata with arche-metadata-crawler
      #   run: |
      #     composer require "acdh-oeaw/arche-metadata-crawler:^0.5.0"
      #     ./shellscripts/crawler.sh
      - name: Commit data
        uses: stefanzweifel/git-auto-commit-action@v4
        with:
          commit_message: Downloaded/Updated Data