cleanup-history.yml

name: Cleanup gh-pages History

on:
  schedule:
    # Run on the 1st of every month at 3:00 AM UTC
    - cron: '0 3 1 * *'
  workflow_dispatch:
    inputs:
      keep_days:
        description: 'Keep commits from last N days (default: 30)'
        required: false
        type: number
        default: 30
      dry_run:
        description: 'Dry run (show what would be done without pushing)'
        required: false
        type: boolean
        default: false

permissions:
  contents: write

jobs:
  cleanup:
    runs-on: ubuntu-latest
    steps:
      - name: Checkout gh-pages
        uses: actions/checkout@v4
        with:
          ref: gh-pages
          fetch-depth: 0  # Full history needed for squashing

      - name: Cleanup old history
        shell: bash
        run: |
          set -e

          KEEP_DAYS="${{ github.event.inputs.keep_days || 30 }}"
          DRY_RUN="${{ github.event.inputs.dry_run || 'false' }}"

          echo "=== gh-pages History Cleanup ==="
          echo "Keep commits from last: ${KEEP_DAYS} days"
          echo "Dry run: ${DRY_RUN}"
          echo ""

          # Get current branch info
          CURRENT_COMMITS=$(git rev-list --count HEAD)
          CURRENT_SIZE=$(du -sh .git | cut -f1)
          echo "Current state:"
          echo "  Commits: ${CURRENT_COMMITS}"
          echo "  .git size: ${CURRENT_SIZE}"
          echo ""

          # Find the cutoff date
          CUTOFF_DATE=$(date -d "-${KEEP_DAYS} days" +%Y-%m-%d 2>/dev/null || date -v-${KEEP_DAYS}d +%Y-%m-%d)
          echo "Cutoff date: ${CUTOFF_DATE}"

          # Find commits to keep (after cutoff date)
          COMMITS_TO_KEEP=$(git rev-list --after="${CUTOFF_DATE}" HEAD | wc -l | tr -d ' ')
          echo "Commits after cutoff: ${COMMITS_TO_KEEP}"

          if [ "${COMMITS_TO_KEEP}" -eq 0 ]; then
            echo "No recent commits found, keeping at least the latest commit"
            COMMITS_TO_KEEP=1
          fi

          # If we have very few commits, nothing to clean
          if [ "${CURRENT_COMMITS}" -le "${COMMITS_TO_KEEP}" ]; then
            echo ""
            echo "Nothing to clean up - all commits are within retention period"
            exit 0
          fi

          COMMITS_TO_SQUASH=$((CURRENT_COMMITS - COMMITS_TO_KEEP))
          echo "Commits to squash: ${COMMITS_TO_SQUASH}"
          echo ""

          # Show what will be removed
          echo "=== Commits to be squashed ==="
          git log --oneline --skip=${COMMITS_TO_KEEP} -20 || true
          if [ "${COMMITS_TO_SQUASH}" -gt 20 ]; then
            echo "... and $((COMMITS_TO_SQUASH - 20)) more"
          fi
          echo ""

          if [ "${DRY_RUN}" = "true" ]; then
            echo "=== DRY RUN - No changes made ==="
            exit 0
          fi

          # Configure git
          git config user.name "github-actions[bot]"
          git config user.email "github-actions[bot]@users.noreply.github.com"

          # Create a new orphan branch with squashed history
          echo "=== Squashing history ==="

          # Get the commit hash to start from (first commit to keep)
          FIRST_KEPT_COMMIT=$(git rev-list --after="${CUTOFF_DATE}" HEAD | tail -1)

          if [ -z "${FIRST_KEPT_COMMIT}" ]; then
            FIRST_KEPT_COMMIT=$(git rev-parse HEAD)
          fi

          # Create orphan branch from the state before our retention period
          # with all current files
          git checkout --orphan temp-cleanup

          # Stage all current files
          git add -A

          # Create initial squashed commit
          git commit -m "chore: squash history (commits before ${CUTOFF_DATE})

          This commit contains the squashed history of gh-pages branch.
          Original commits before ${CUTOFF_DATE} have been combined to reduce repository size.

          Squashed ${COMMITS_TO_SQUASH} commits.

          🤖 Generated by cleanup-history workflow"

          # Now cherry-pick the recent commits on top
          if [ "${COMMITS_TO_KEEP}" -gt 0 ]; then
            echo "Cherry-picking ${COMMITS_TO_KEEP} recent commits..."
            RECENT_COMMITS=$(git rev-list --reverse --after="${CUTOFF_DATE}" gh-pages)

            for commit in ${RECENT_COMMITS}; do
              # Skip if this would result in empty commit
              git cherry-pick --allow-empty "${commit}" || {
                echo "  Skipping commit ${commit} (likely empty or conflict)"
                git cherry-pick --abort 2>/dev/null || true
              }
            done
          fi

          # Force update gh-pages
          git branch -D gh-pages
          git branch -m gh-pages

          # Garbage collect
          echo ""
          echo "=== Running garbage collection ==="
          git reflog expire --expire=now --all
          git gc --prune=now --aggressive

          NEW_SIZE=$(du -sh .git | cut -f1)
          NEW_COMMITS=$(git rev-list --count HEAD)

          echo ""
          echo "=== Cleanup complete ==="
          echo "Before: ${CURRENT_COMMITS} commits, ${CURRENT_SIZE}"
          echo "After:  ${NEW_COMMITS} commits, ${NEW_SIZE}"

      - name: Push cleaned history
        if: ${{ github.event.inputs.dry_run != 'true' }}
        run: |
          git push origin gh-pages --force

      - name: Summary
        run: |
          if [ "${{ github.event.inputs.dry_run }}" = "true" ]; then
            echo "## Dry Run Complete" >> $GITHUB_STEP_SUMMARY
            echo "No changes were pushed." >> $GITHUB_STEP_SUMMARY
          else
            echo "## History Cleanup Complete" >> $GITHUB_STEP_SUMMARY
            echo "The gh-pages branch history has been squashed." >> $GITHUB_STEP_SUMMARY
            echo "" >> $GITHUB_STEP_SUMMARY
            echo "Old commits have been combined into a single commit to reduce repository size." >> $GITHUB_STEP_SUMMARY
          fi