name: Check Website URLs

on:
  schedule:
    # Runs daily at 03:00 UTC
    - cron: '0 3 * * *'
  workflow_dispatch: # Allows manual triggering

permissions:
  contents: write # Needed to commit changes back to the repo

jobs:
  check-urls:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout repository
        uses: actions/checkout@v4
        with:
          # Fetch full history to be able to compare changes if needed (optional)
          fetch-depth: 0

      - name: Install jq
        run: sudo apt-get update && sudo apt-get install -y jq

      - name: Check URLs and update modflix.json
        id: check_urls
        run: |
          echo "Starting URL check process..."
          cp modflix.json modflix.json.bak # Backup original file
          updated_json=$(cat modflix.json)
          changes_made=false

          # Iterate over each key in the JSON object
          for key in $(echo "$updated_json" | jq -r 'keys_unsorted[]'); do
            original_url=$(echo "$updated_json" | jq -r --arg k "$key" '.[$k].url')
            echo "Checking key: $key, URL: $original_url"

            # Use curl to follow redirects (-L), get final URL (-w '%{url_effective}'),
            # suppress output (-s), discard body (-o /dev/null), get status code (-w '%{http_code}')
            # Set timeouts to prevent hanging
            response=$(curl -Lso /dev/null -w "%{http_code} %{url_effective}" --max-time 15 --connect-timeout 7 "$original_url")
            http_code=$(echo "$response" | awk '{print $1}')
            final_url=$(echo "$response" | awk '{print $2}')

            echo "  -> Status: $http_code, Final URL: $final_url"

            # Check if it was a redirect (3xx) and the final URL is different
            if [[ "$http_code" =~ ^3[0-9]{2}$ ]] && [[ "$original_url" != "$final_url" ]]; then
              echo "  -> Redirect detected. Original: $original_url, New Effective: $final_url"

              # Extract the scheme and host from the final URL
              new_scheme_host=$(echo "$final_url" | sed -E 's|^(https?://[^/]+).*|\1|')

              # Extract the path, query, and fragment from the original URL
              # This preserves things like trailing slashes or specific paths
              original_path_etc=$(echo "$original_url" | sed -E 's|https?://[^/]+(.*)|\1|')
              if [[ -z "$original_path_etc" ]]; then
                 # Handle cases where the original URL might just be the domain (e.g., https://example.com)
                 # Check if the original URL ended with a slash
                 if [[ "$original_url" == */ ]]; then
                    original_path_etc="/"
                 else
                     # If final URL has a path and original didn't, maybe use final's path?
                     # For now, keeping it simple: if original had no path, new one has no path unless original ended with /
                     : # Keep original_path_etc empty or "/" based on above check
                 fi
              fi


              # Construct the new URL using the new scheme/host and original path/query/fragment
              new_url="${new_scheme_host}${original_path_etc}"
              echo "  -> Constructing new URL: $new_url"

              # Update the JSON content in the variable
              updated_json=$(echo "$updated_json" | jq --arg k "$key" --arg nu "$new_url" '.[$k].url = $nu')
              changes_made=true
              echo "  -> Updated JSON for key $key"
            elif [[ "$http_code" == "200" ]]; then
              echo "  -> URL OK (200)."
            else
              echo "  -> URL check failed or returned non-redirect/non-200 status ($http_code). No changes made for this URL."
            fi
          done

          if $changes_made; then
            echo "Changes were made. Writing updated modflix.json"
            echo "$updated_json" | jq '.' > modflix.json # Write updated content back, pretty-printed
            # Set output for the next step
            echo "changes_detected=true" >> $GITHUB_OUTPUT
          else
            echo "No changes detected in URLs."
            # Restore original file to avoid timestamp changes if only formatting changed
            mv modflix.json.bak modflix.json
            echo "changes_detected=false" >> $GITHUB_OUTPUT
          fi

      - name: Commit and push changes
        if: steps.check_urls.outputs.changes_detected == 'true'
        run: |
          git config --global user.name 'github-actions[bot]'
          git config --global user.email 'github-actions[bot]@users.noreply.github.com'
          git add modflix.json
          # Check if there are staged changes before committing
          if ! git diff --staged --quiet; then
            git commit -m "chore: Update redirected URLs in modflix.json

          Automatically updated URLs based on HTTP redirects."
            git push
            echo "Changes committed and pushed."
          else
            echo "No effective changes to commit after writing file (might be formatting only)."
          fi
        env:
          GITHUB_TOKEN: ${{ secrets.GITHUB_TOKEN }} # Use the default token