diff.yml

on:
  workflow_dispatch:
  repository_dispatch:
    types: [release_event]

name: Diff Data

concurrency:
  group: diff-data
  cancel-in-progress: false

permissions:
  contents: read

jobs:
  diff_data:
    name: Diff and Release
    runs-on: ubuntu-latest
    # ncipollo/release-action creates a release + tag via GITHUB_TOKEN,
    # which requires contents: write (top-level default is contents: read).
    permissions:
      contents: write
    env:
      TZ: Europe/London
    steps:
      - uses: clusterflick/.github/setup@main

      - name: Get release tags
        id: tags
        run: |
          RELEASES=$(gh api repos/clusterflick/data-transformed/releases?per_page=10 \
            --jq 'sort_by(.published_at) | reverse')
          CURRENT=$(echo "$RELEASES" | jq -r '.[0].tag_name')
          PREVIOUS=$(echo "$RELEASES" | jq -r '.[1].tag_name')
          # The diff is anchored to when the current release was published rather
          # than to the wall clock, so a re-run — or a run repeated months later —
          # reproduces the same blob instead of quietly reclassifying everything
          # that has happened since as past
          CURRENT_PUBLISHED_AT=$(echo "$RELEASES" | jq -r '.[0].published_at')
          echo "current=$CURRENT" >> "$GITHUB_OUTPUT"
          echo "previous=$PREVIOUS" >> "$GITHUB_OUTPUT"
          echo "current_published_at=$CURRENT_PUBLISHED_AT" >> "$GITHUB_OUTPUT"
          echo "Current: $CURRENT ($CURRENT_PUBLISHED_AT), Previous: $PREVIOUS"

          # The very first release has nothing to be diffed against
          if [ -z "$PREVIOUS" ] || [ "$PREVIOUS" = "null" ]; then
            echo "No previous release to diff against. Skipping."
            echo "has_previous=false" >> "$GITHUB_OUTPUT"
          else
            echo "has_previous=true" >> "$GITHUB_OUTPUT"
          fi
        env:
          GH_TOKEN: ${{ secrets.PAT }}

      - name: Download current release
        if: steps.tags.outputs.has_previous == 'true'
        uses: clusterflick/release-downloader@v3
        with:
          token: ${{ secrets.PAT }}
          repository: "clusterflick/data-transformed"
          tag: ${{ steps.tags.outputs.current }}
          fileName: "*"
          out-file-path: "transformed-data/current"

      - name: Download previous release
        if: steps.tags.outputs.has_previous == 'true'
        uses: clusterflick/release-downloader@v3
        with:
          token: ${{ secrets.PAT }}
          repository: "clusterflick/data-transformed"
          tag: ${{ steps.tags.outputs.previous }}
          fileName: "*"
          out-file-path: "transformed-data/previous"

      - name: Diff Data
        if: steps.tags.outputs.has_previous == 'true'
        run: npm run diff -- ${{ steps.tags.outputs.current }} ${{ steps.tags.outputs.previous }} ${{ steps.tags.outputs.current_published_at }}

      # The diff only writes its output when something actually changed, so its
      # absence is how a no-op run reports itself.
      - name: Check for changes
        id: changes
        run: |
          if [ -f diffed-data/diffed-data.json ]; then
            echo "has_changes=true" >> "$GITHUB_OUTPUT"
          else
            echo "No changes between releases; skipping release."
            echo "has_changes=false" >> "$GITHUB_OUTPUT"
          fi

      - name: Generate Tag
        if: steps.changes.outputs.has_changes == 'true'
        run: echo "TAG=$(TZ=Europe/London date +'%Y%m%d.%H%M%S')" >> $GITHUB_ENV
      - uses: ncipollo/release-action@v1
        if: steps.changes.outputs.has_changes == 'true'
        id: release
        with:
          allowUpdates: false
          artifactErrorsFailBuild: true
          artifacts: "diffed-data/*"
          makeLatest: true
          tag: ${{ env.TAG }}
          commit: main
          body: |
            Changes between data-transformed ${{ steps.tags.outputs.previous }} and ${{ steps.tags.outputs.current }}

            Created by job ${{ github.server_url }}/${{ github.repository }}/actions/runs/${{ github.run_id }}
      - name: Output summary
        if: steps.changes.outputs.has_changes == 'true'
        run: |
          echo "🔖 New release - ${{ steps.release.outputs.html_url }}" >> $GITHUB_STEP_SUMMARY