quatalog-scraper/.github/workflows/scraper.yml

name: Update data
on:
  workflow_dispatch:
  schedule:
    - cron: '0 * * * *'

jobs:
  scrape-data:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout scraping repo
        uses: actions/checkout@v3
        with:
          path: quatalog-scraping
        
      - name: Clone QuACS data
        uses: actions/checkout@v3
        with:
          repository: quacs/quacs-data
          path: quacs-data

      - name: Clone Quatalog data
        uses: actions/checkout@v3
        with:
          repository: quatalog/quatalog-data
          path: quatalog-data
      
      - name: Run scraper
        run: |
          cd quatalog-scraping
          find .
          pwd
          # Usage: CourseOfferingsScraper <data_directory> <terms_offered_file> <prerequisites_file> <list_of_terms_file>
          ./bin/CourseOfferingScraper \
            ./quacs-data/semester_data \
            ./quatalog-data/terms_offered.json \
            ./quatalog-data/prerequisites.json \
            ./quatalog-data/terms_list.json
                      
      - name: Copy `catalog.json` from QuACS data
        run: |
          CURRENT_TERM="$(grep terms_list.json -e 'current_term' | cut -f4 -d '"')"
          cp "./quacs-data/semester_data/$CURRENT_TERM/catalog.json" ./quatalog-data/catalog.json
                    
      - name: Push new data
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }}
        run: |
          cd ./quatalog-data
          git config --global user.name "Quatalog Updater"
          git config --global user.email "github_actions@quatalog.com"
          git config --unset-all http.https://github.com/.extraheader
          git add terms_offered.json prerequisites.json terms_list.json catalog.json
          git commit -m "$(date)" || exit 0
          git push