run.yaml

name: Build R Packages

on:
  workflow_dispatch:
  push:
    branches:
      - 'build/**'
    paths:
      - 'logs/dispatched-packages.txt'
      - 'logs/successful-packages.txt'

concurrency:
  group: build-r-packages-${{ github.ref }}
  cancel-in-progress: false

jobs:
  build:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v6
        with:
          token: ${{ secrets.PAT }}
          persist-credentials: true

      - name: Update to latest
        run: |
          git pull origin ${GITHUB_REF#refs/heads/}
          git reset --hard origin/${GITHUB_REF#refs/heads/}

      - name: Save kubeconfig
        shell: bash
        run: mkdir -p ~/.kube && echo "${{ secrets.KUBECONFIG }}" > ~/.kube/config
      
      - name: Install Kubectl
        run: |
          curl -LO 'https://dl.k8s.io/release/v1.34.1/bin/linux/amd64/kubectl' && \
          chmod +x ./kubectl && \
          mv ./kubectl /usr/local/bin/kubectl && \
          kubectl version

      - name: Check for completed cycle
        id: cycle_check
        run: |
          # Check if PACKAGES file exists (indicates completed cycle)
          if [[ -f "PACKAGES" ]]; then
            echo "Cycle already completed (PACKAGES file exists). Skipping."
            echo "skip=true" >> $GITHUB_OUTPUT
          else
            echo "skip=false" >> $GITHUB_OUTPUT
          fi

      - name: Manage build lifecycle
        if: steps.cycle_check.outputs.skip != 'true'
        env:
          RCLONE_CONF: ${{secrets.RCLONE_CONF}}
        id: build
        run: |
          set -euxo pipefail

          # Use branch name as the build identifier
          BRANCH_NAME=${GITHUB_REF#refs/heads/}
          BUILD_ID=$(echo "${BRANCH_NAME}" | sed 's|build/||' | tr '/' '-')
          
          echo "Build ID: ${BUILD_ID}"
          echo "Branch: ${BRANCH_NAME}"
          
          # Check if this is first run (no biocdeps.json yet)
          if [[ ! -s "biocdeps.json" ]]; then
            echo "First run - setting up Kubernetes resources..."
            
            CONTAINER=$(cat CONTAINER_BASE_IMAGE.bioc)
            
            # Setup Kubernetes resources - this handles running deps_json.R via init container
            bash ./.github/scripts/setup_k8s.sh "nfs-client" "500Gi" "${BUILD_ID}"
          else
            echo "Continuing existing build..."
            CONTAINER=$(cat CONTAINER_BASE_IMAGE.bioc)
            
            # Handle completed jobs
            bash ./.github/scripts/handle_k8s_jobs.sh \
              "${BUILD_ID}" \
              "logs/successful-packages.txt" \
              "logs/failed-packages.txt"
          fi

          # Ensure log directory exists before any operations
          mkdir -p logs
          touch logs/dispatched-packages.txt

          # Find and dispatch packages
          python ./.github/scripts/find_ready_pkgs.py \
            "biocdeps.json" \
            "ready_packages.txt" \
            "logs/dispatched-packages.txt" \
            "logs/successful-packages.txt" \
            "remaining-packages.json"
          
          # Calculate truncated build ID for PVC name
          BUILD_ID_CLEAN=$(echo "${BUILD_ID}" | tr '[:upper:]' '[:lower:]' | tr '.' '-' | tr -cd '[:alnum:]-')
          VERSION_SUFFIX="${BUILD_ID_CLEAN: -3}"
          DATE_PART=$(echo "$BUILD_ID_CLEAN" | cut -d'-' -f1-3 | tr -d '-')
          CONTAINER_HINT=$(echo "$BUILD_ID_CLEAN" | sed 's/^[0-9-]*-[0-9]*-//' | sed 's/-[^-]*$//' | cut -c1-4)
          BUILD_ID_SHORT="${DATE_PART}-${CONTAINER_HINT}-${VERSION_SUFFIX}"
          PVC_NAME="pvc-${BUILD_ID_SHORT}"
          
          # Dispatch jobs with xargs and delay
          cat "ready_packages.txt" | xargs -i bash -c "bash ./.github/scripts/dispatch_k8s_job.sh {} ${CONTAINER} ${PVC_NAME} ${BUILD_ID} && sleep 1"

          # Track if we had any new activity
          git add .
          ACTIVITY_CHECK=$(git diff --staged --name-status | grep -q "logs/.*packages.txt" && echo "true" || echo "false")
          echo "had_activity=${ACTIVITY_CHECK}" >> $GITHUB_OUTPUT
          
          # Get counter values
          NULL_PUSH_COUNT=$(cat "null_push_counter" 2>/dev/null || echo "0")
          RESET_ATTEMPTS=$(cat "reset_attempts_counter" 2>/dev/null || echo "0")
          
          # Note: Reset failed logic removed - handle failures manually or via separate workflow if needed

          # Backup changed files to temp directory
          mkdir -p /tmp/restore
          
          # Stage all changes to identify modified files
          git add .
          
          # Copy changed files to restore directory
          git diff --staged --name-only | xargs -r -I{} bash -c 'mkdir -p "/tmp/restore/$(dirname "{}")" && cp "{}" "/tmp/restore/{}"'
          
          echo "Backup complete - copied changed files"

      - name: Commit and push results
        if: steps.cycle_check.outputs.skip != 'true'
        id: repo_changes
        run: |
          BRANCH_NAME=${GITHUB_REF#refs/heads/}
          git pull origin ${BRANCH_NAME} && git reset --hard origin/${BRANCH_NAME} || git reset --hard origin/${BRANCH_NAME}
          cp -r /tmp/restore/* ./ || true
          git config --global user.name 'GitHub Actions'
          git config --global user.email 'actions@github.com'
          git add . || true
          
          # Check if there are actual changes to commit
          if git diff --staged --quiet; then
            echo "No changes to commit"
            # If no changes, increment the null push counter
            NULL_PUSH_COUNT=$(cat "null_push_counter" 2>/dev/null || echo "0")
            NEW_COUNT=$((NULL_PUSH_COUNT + 1))
            echo "${NEW_COUNT}" > "null_push_counter"
            
            # Commit just the counter change
            git add "null_push_counter"
            git commit -m "Increment null push counter to ${NEW_COUNT}"
            git push || (git pull --rebase && git push)
            echo "pushed=false" >> $GITHUB_OUTPUT
          else
            # Check for new successful packages
            NEW_SUCCESS=$(git diff --staged -- "logs/successful-packages.txt" | grep -c "^\+" | grep -v "^\+\+\+" || echo "0")
            
            # Always reset null push counter on meaningful changes
            echo "0" > "null_push_counter"
            git add "null_push_counter"
            
            # Only reset attempts counter if we have new successful packages
            if [ "${NEW_SUCCESS}" -gt 0 ]; then
              echo "New successful packages detected (${NEW_SUCCESS}), resetting attempts counter"
              echo "0" > "reset_attempts_counter"
              git add "reset_attempts_counter"
            fi
            
            # Commit all changes
            git commit -m "Build state update: ${{ github.run_id }}"
            git push || (git pull --rebase && git push)
            echo "pushed=true" >> $GITHUB_OUTPUT
          fi

      - name: Check if cycle is ready to finish
        if: steps.cycle_check.outputs.skip != 'true'
        id: check_complete
        run: |
          if bash ./.github/scripts/check_cycle_complete.sh > /tmp/cycle-check-output.txt 2>&1; then
            echo "notify=yes" >> $GITHUB_OUTPUT
            
            # Get container name for notification
            CONTAINER=$(cat CONTAINER_BASE_IMAGE.bioc)
            
            # Get stats for notification
            SUCCESSFUL=$(wc -l < logs/successful-packages.txt 2>/dev/null || echo 0)
            FAILED=$(wc -l < logs/failed-packages.txt 2>/dev/null || echo 0)
            # Count only blocked packages (remaining packages that are not failed)
            REMAINING=$(python3 -c "import json; remaining = json.load(open('remaining-packages.json')); failed = set(line.strip() for line in open('logs/failed-packages.txt') if line.strip()); blocked = [pkg for pkg in remaining if pkg not in failed]; print(len(blocked))" 2>/dev/null || echo 0)
            
            # Create simple single-line notification message (avoid quotes to prevent JSON escaping issues)
            MSG="Build cycle ready for ${CONTAINER}: ${SUCCESSFUL} successful, ${FAILED} failed, ${REMAINING} blocked. Run finish_cycle: https://github.com/${{ github.repository }}/actions/workflows/finish_cycle.yaml"
            
            # Replace any double quotes with single quotes to avoid JSON issues
            MSG="${MSG//\"/\'}"
            
            echo "SLACK_MESSAGE=${MSG}" >> "$GITHUB_ENV"
          else
            echo "notify=no" >> $GITHUB_OUTPUT
            cat /tmp/cycle-check-output.txt || true
          fi

      - name: Notify slack channel
        uses: slackapi/slack-github-action@v2.1.1
        with:
          method: chat.postMessage
          token: ${{ secrets.SLACK_BOT_TOKEN }}
          payload: |
            {
              "channel": "${{ secrets.SLACK_CHANNEL_ID }}",
              "text": "${{ env.SLACK_MESSAGE }}"
            }
        env:
          SLACK_BOT_TOKEN: ${{ secrets.SLACK_BOT_TOKEN }}
          SLACK_CHANNEL_ID: ${{ secrets.SLACK_CHANNEL_ID }}
        if: env.SLACK_BOT_TOKEN != null && env.SLACK_CHANNEL_ID != null && steps.check_complete.outputs.notify == 'yes'
        continue-on-error: true