discover-and-dispatch.yml

name: Discover and Dispatch PR Fixes

on:
  schedule:
    # Run daily at midnight UTC
    - cron: '0 0 * * *'
  workflow_dispatch:
    inputs:
      target_repo:
        description: 'Specific repo to check (optional, e.g., VectorInstitute/repo-name)'
        required: false
      dry_run:
        description: 'Dry run - discover PRs but do not dispatch fixes'
        type: boolean
        default: false

permissions:
  contents: read
  actions: write

env:
  ORG_NAME: VectorInstitute

jobs:
  discover:
    runs-on: ubuntu-latest
    outputs:
      prs: ${{ steps.find-prs.outputs.prs }}

    steps:
      - name: Checkout bot repository
        uses: actions/checkout@v6

      - name: Find bot PRs with failing checks
        id: find-prs
        run: |
          # Determine which repos to check
          if [ -n "${{ github.event.inputs.target_repo }}" ]; then
            REPOS="${{ github.event.inputs.target_repo }}"
          else
            if [ -f repos-to-monitor.csv ]; then
              REPOS=$(tail -n +2 repos-to-monitor.csv | grep -v '^$' | tr '\n' ' ')
              echo "Reading repos from repos-to-monitor.csv"
            else
              echo "Error: repos-to-monitor.csv not found"
              exit 1
            fi
          fi

          PRS_TO_FIX="[]"
          BOT_AUTHORS=("app/dependabot" "pre-commit-ci[bot]")

          for REPO in $REPOS; do
            # Skip the bot repository itself
            if [ "$REPO" = "VectorInstitute/aieng-bot" ]; then
              echo "Skipping $REPO (bot repository)"
              continue
            fi

            echo "Checking $REPO for open bot PRs..."

            for BOT_AUTHOR in "${BOT_AUTHORS[@]}"; do
              # Get all open PRs from this bot author
              REPO_PRS=$(gh pr list --repo "$REPO" \
                --author "$BOT_AUTHOR" \
                --state open \
                --json number,title,url,headRefName 2>/dev/null || echo "[]")

              if [ "$REPO_PRS" != "[]" ] && [ -n "$REPO_PRS" ]; then
                BOT_PRS=$(echo "$REPO_PRS" | jq -c "[.[] | {repo: \"$REPO\", pr_number: .number, title: .title, head_ref: .headRefName}]")
                echo "Found open PRs in $REPO from $BOT_AUTHOR:"
                echo "$BOT_PRS" | jq '.'
                PRS_TO_FIX=$(echo "$PRS_TO_FIX" | jq -c ". + $BOT_PRS")
              fi
            done
          done

          PRS_TO_FIX=$(echo "$PRS_TO_FIX" | jq -c 'unique_by({repo, pr_number})')

          echo "Total PRs to fix: $(echo "$PRS_TO_FIX" | jq 'length')"
          echo "$PRS_TO_FIX" | jq '.'

          echo "prs=$(echo "$PRS_TO_FIX" | jq -c '.')" >> $GITHUB_OUTPUT
        env:
          GH_TOKEN: ${{ secrets.ORG_ACCESS_TOKEN }}

  dispatch:
    runs-on: ubuntu-latest
    needs: discover
    if: needs.discover.outputs.prs != '[]' && github.event.inputs.dry_run != 'true'

    strategy:
      matrix:
        pr: ${{ fromJson(needs.discover.outputs.prs) }}
      max-parallel: 5
      fail-fast: false

    steps:
      - name: Dispatch fix job for ${{ matrix.pr.repo }}#${{ matrix.pr.pr_number }}
        run: |
          echo "Dispatching fix job for ${{ matrix.pr.repo }}#${{ matrix.pr.pr_number }}"
          echo "Title: ${{ matrix.pr.title }}"

          gh workflow run fix-pr-agent.yml \
            --repo VectorInstitute/aieng-bot \
            --field target_repo="${{ matrix.pr.repo }}" \
            --field pr_number="${{ matrix.pr.pr_number }}" \
            --field max_retries="3"

          echo "Fix job dispatched"
        env:
          GH_TOKEN: ${{ secrets.ORG_ACCESS_TOKEN }}

  report-summary:
    runs-on: ubuntu-latest
    needs: [discover, dispatch]
    if: always()

    steps:
      - name: Generate summary
        run: |
          PR_COUNT=$(echo '${{ needs.discover.outputs.prs }}' | jq 'length')
          DRY_RUN="${{ github.event.inputs.dry_run }}"

          echo "## PR Fix Discovery Summary" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "- **Organization**: ${{ env.ORG_NAME }}" >> $GITHUB_STEP_SUMMARY
          echo "- **PRs Found**: $PR_COUNT" >> $GITHUB_STEP_SUMMARY
          echo "- **Run Time**: $(date -u +"%Y-%m-%d %H:%M:%S UTC")" >> $GITHUB_STEP_SUMMARY

          if [ "$DRY_RUN" = "true" ]; then
            echo "- **Mode**: Dry Run (no fixes dispatched)" >> $GITHUB_STEP_SUMMARY
          else
            echo "- **Mode**: Active (fixes dispatched)" >> $GITHUB_STEP_SUMMARY
          fi

          echo "" >> $GITHUB_STEP_SUMMARY

          if [ "$PR_COUNT" -gt 0 ]; then
            echo "### PRs Discovered" >> $GITHUB_STEP_SUMMARY
            echo '${{ needs.discover.outputs.prs }}' | \
              jq -r '.[] | "- [\(.repo)#\(.pr_number)](https://github.com/\(.repo)/pull/\(.pr_number)) - \(.title)"' >> $GITHUB_STEP_SUMMARY
          else
            echo "No open bot PRs found." >> $GITHUB_STEP_SUMMARY
          fi