# SPDX-FileCopyrightText: Copyright (c) 2024-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: Apache-2.0

name: Release Pipeline

on:
  workflow_dispatch:
    inputs:
      commit_sha:
        description: 'Git commit SHA whose post-merge CI images to publish (full 40-char SHA).'
        required: true
        type: string
      rc_number:
        description: 'RC number (e.g., 0 for rc0). Leave empty to auto-increment.'
        required: false
        type: string

# Note: workflow_dispatch can only be triggered from release/* branches
# This is enforced in the prepare-release job via branch validation

permissions:
  contents: write

env:
  REGISTRY_IMAGE: ai-dynamo/dynamo

jobs:
  # ============================================================================
  # GATE: Version Extraction
  # ============================================================================

  prepare-release:
    name: Prepare Release
    runs-on: prod-default-small-v2
    outputs:
      version: ${{ steps.extract.outputs.version }}
      commit_sha: ${{ steps.extract.outputs.commit_sha }}
    steps:
      - name: Extract version and validate inputs
        id: extract
        env:
          COMMIT_SHA: ${{ github.event.inputs.commit_sha }}
          BRANCH_NAME: ${{ github.ref_name }}
        run: |
          set -euo pipefail

          if ! [[ "${COMMIT_SHA}" =~ ^[0-9a-f]{40}$ ]]; then
            echo "Error: commit_sha must be a full 40-character hex SHA (got: '${COMMIT_SHA}')"
            exit 1
          fi

          if [[ ! "$BRANCH_NAME" =~ ^release/[0-9]+\.[0-9]+\.[0-9]+ ]]; then
            echo "Error: workflow_dispatch must be triggered from a release/* branch"
            echo "Current branch: $BRANCH_NAME"
            exit 1
          fi
          VERSION="${BRANCH_NAME#release/}"

          echo "version=${VERSION}" >> $GITHUB_OUTPUT
          echo "commit_sha=${COMMIT_SHA}" >> $GITHUB_OUTPUT
          echo "Detected version: ${VERSION}"
          echo "Source commit SHA: ${COMMIT_SHA}"

  # ============================================================================
  # NGC PUBLISH: RC tag, crane copy to NGC, Helm chart push
  # Sources images from ECR using SHA-based tags produced by post-merge CI.
  # ============================================================================

  release-publish:
    name: Tag RC & Publish to NGC
    needs: [prepare-release]
    if: needs.prepare-release.result == 'success'
    runs-on: prod-builder-amd-v1 # TODO: needs to identify the correct runner here, definitely should not be prod-builder-amd-v1
    environment: automated-release
    env:
      VERSION: ${{ needs.prepare-release.outputs.version }}
      COMMIT_SHA: ${{ needs.prepare-release.outputs.commit_sha }}
      REGISTRY_IMAGE: ai-dynamo/dynamo
      AWS_DEFAULT_REGION: ${{ secrets.AWS_DEFAULT_REGION }}
    steps:
      - name: Checkout at source commit
        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd  # v6.0.2
        # fetch-depth: 0 + fetch-tags required here — git tag -l enumerates existing RC tags
        with:
          ref: ${{ needs.prepare-release.outputs.commit_sha }}
          fetch-depth: 0
          fetch-tags: true
          token: ${{ secrets.RC_GITHUB_PAT }}

      - name: Determine next RC tag
        id: rc_tag
        env:
          INPUT_RC_NUMBER: ${{ github.event.inputs.rc_number }}
        run: |
          set -euo pipefail

          if [ -n "${INPUT_RC_NUMBER}" ]; then
            if ! [[ "${INPUT_RC_NUMBER}" =~ ^[0-9]+$ ]]; then
              echo "Error: rc_number must be a non-negative integer (got: ${INPUT_RC_NUMBER})"
              exit 1
            fi
            NEXT_RC="${INPUT_RC_NUMBER}"
            echo "Using provided RC number: ${NEXT_RC}"
          else
            echo "No RC number provided. Auto-incrementing..."
            RC_PATTERN="v${VERSION}-rc"
            EXISTING_RCS=$(git tag -l "${RC_PATTERN}*" | grep -E "^v${VERSION}-rc[0-9]+$" | sort -V || true)

            if [ -z "$EXISTING_RCS" ]; then
              NEXT_RC=0
              echo "No existing RC tags found. Starting with rc0."
            else
              LAST_RC=$(echo "$EXISTING_RCS" | tail -1)
              LAST_RC_NUM=${LAST_RC#v${VERSION}-rc}
              NEXT_RC=$((LAST_RC_NUM + 1))
              echo "Found existing RC tags:"
              echo "$EXISTING_RCS"
              echo "Last RC: ${LAST_RC}, Next RC number: ${NEXT_RC}"
            fi
          fi

          RC_TAG="v${VERSION}-rc${NEXT_RC}"

          # Normalize version for Helm SemVer: X.Y.Z.suffix → X.Y.Z-suffix
          SEMVER_VERSION=$(echo "${VERSION}" | sed -E 's/^([0-9]+\.[0-9]+\.[0-9]+)\.(.+)/\1-\2/')
          HELM_CHART_VERSION="${SEMVER_VERSION}-rc${NEXT_RC}"

          echo "rc_tag=${RC_TAG}" >> $GITHUB_OUTPUT
          echo "rc_number=${NEXT_RC}" >> $GITHUB_OUTPUT
          echo "ngc_version_tag=${VERSION}rc${NEXT_RC}" >> $GITHUB_OUTPUT
          echo "helm_chart_version=${HELM_CHART_VERSION}" >> $GITHUB_OUTPUT
          echo "Will create tag: ${RC_TAG}"
          echo "Helm chart version: ${HELM_CHART_VERSION}"

      - name: Create RC tag
        env:
          RC_TAG: ${{ steps.rc_tag.outputs.rc_tag }}
        run: |
          set -euo pipefail
          git config user.name "github-actions[bot]"
          git config user.email "github-actions[bot]@users.noreply.github.com"
          git tag -a "${RC_TAG}" -m "Release candidate ${RC_TAG}"
          git push origin "${RC_TAG}"
          echo "Created and pushed tag: ${RC_TAG}"

      - name: Setup crane
        env:
          CRANE_VERSION: v0.20.2
        run: |
          curl -sL "https://github.com/google/go-containerregistry/releases/download/${CRANE_VERSION}/go-containerregistry_Linux_x86_64.tar.gz" \
            | tar -xzf - crane
          sudo mv crane /usr/local/bin/
          crane version

      - name: Login to ECR
        run: |
          ACCOUNT_ID="$(aws sts get-caller-identity --query Account --output text)"
          ECR_HOSTNAME="${ACCOUNT_ID}.dkr.ecr.${AWS_DEFAULT_REGION}.amazonaws.com"
          aws ecr get-login-password --region ${AWS_DEFAULT_REGION} | docker login --username AWS --password-stdin "${ECR_HOSTNAME}"

      - name: Login to NGC
        env:
          NGC_TOKEN: ${{ secrets.NGC_PUBLISH_TOKEN }}
          NGC_USERNAME: ${{ secrets.NGC_PUBLISH_USERNAME }}
        run: |
          echo "${NGC_TOKEN}" | docker login nvcr.io -u "${NGC_USERNAME}" --password-stdin
          echo "${NGC_TOKEN}" | crane auth login nvcr.io -u "${NGC_USERNAME}" --password-stdin

      - name: Copy images to NGC
        id: copy_images
        env:
          NGC_REGISTRY: nvcr.io
          NGC_ORG: ${{ secrets.NGC_PUBLISH_ORG }}
          NGC_VERSION_TAG: ${{ steps.rc_tag.outputs.ngc_version_tag }}
        run: |
          set -euo pipefail

          SUCCESSFUL_COPIES=()
          FAILED_COPIES=()

          ACCOUNT_ID="$(aws sts get-caller-identity --query Account --output text)"
          ECR_HOSTNAME="${ACCOUNT_ID}.dkr.ecr.${AWS_DEFAULT_REGION}.amazonaws.com"

          echo "========================================"
          echo "Copying images from ECR to NGC (registry-to-registry)"
          echo "Source commit SHA: ${COMMIT_SHA}"
          echo "NGC Version Tag: ${NGC_VERSION_TAG}"
          echo "========================================"

          copy_image() {
            local SRC="$1" DST="$2" LABEL="$3"
            echo "----------------------------------------"
            echo "Copying: ${LABEL}"
            # crane copy preserves multi-arch manifest lists by default (no --platform needed)
            if crane copy "${SRC}" "${DST}"; then
              echo "  Copied: ${LABEL}"
              SUCCESSFUL_COPIES+=("${LABEL}")
              return 0
            else
              echo "  Warning: Failed to copy ${LABEL}, skipping..."
              FAILED_COPIES+=("${LABEL}")
              return 1
            fi
          }

          # ---- CUDA 12 runtime images (vllm and sglang) ----
          echo ""
          echo "=== CUDA 12 Runtime Images (vllm, sglang) ==="
          CUDA12_FRAMEWORKS=("vllm" "sglang")
          for FRAMEWORK in "${CUDA12_FRAMEWORKS[@]}"; do
            NGC_NAME="${FRAMEWORK}-runtime"
            SOURCE="${ECR_HOSTNAME}/${REGISTRY_IMAGE}:${COMMIT_SHA}-${FRAMEWORK}-runtime-cuda12"
            TARGET="${NGC_REGISTRY}/${NGC_ORG}/ai-dynamo/${NGC_NAME}:${NGC_VERSION_TAG}"
            copy_image "${SOURCE}" "${TARGET}" "${NGC_NAME}:${NGC_VERSION_TAG}"
          done

          # ---- CUDA 13 runtime images (vllm, sglang, trtllm) ----
          echo ""
          echo "=== CUDA 13 Runtime Images (vllm, sglang, trtllm) ==="
          CUDA13_FRAMEWORKS=("vllm" "sglang" "trtllm")
          for FRAMEWORK in "${CUDA13_FRAMEWORKS[@]}"; do
            if [ "${FRAMEWORK}" = "trtllm" ]; then
              NGC_NAME="tensorrtllm-runtime"
            else
              NGC_NAME="${FRAMEWORK}-runtime"
            fi

            SOURCE="${ECR_HOSTNAME}/${REGISTRY_IMAGE}:${COMMIT_SHA}-${FRAMEWORK}-runtime-cuda13"
            TARGET="${NGC_REGISTRY}/${NGC_ORG}/ai-dynamo/${NGC_NAME}:${NGC_VERSION_TAG}-cuda13"
            copy_image "${SOURCE}" "${TARGET}" "${NGC_NAME}:${NGC_VERSION_TAG}-cuda13"
          done

          # ---- EFA runtime images (amd64 only, no multi-arch manifest needed) ----
          echo ""
          echo "=== EFA Runtime Images ==="

          # vllm EFA (CUDA 12, amd64 only)
          SOURCE="${ECR_HOSTNAME}/${REGISTRY_IMAGE}:${COMMIT_SHA}-vllm-runtime-efa-cuda12"
          TARGET="${NGC_REGISTRY}/${NGC_ORG}/ai-dynamo/vllm-runtime:${NGC_VERSION_TAG}-efa"
          copy_image "${SOURCE}" "${TARGET}" "vllm-runtime:${NGC_VERSION_TAG}-efa"

          # trtllm EFA (CUDA 13, amd64 only)
          SOURCE="${ECR_HOSTNAME}/${REGISTRY_IMAGE}:${COMMIT_SHA}-trtllm-runtime-efa-cuda13"
          TARGET="${NGC_REGISTRY}/${NGC_ORG}/ai-dynamo/tensorrtllm-runtime:${NGC_VERSION_TAG}-efa"
          copy_image "${SOURCE}" "${TARGET}" "tensorrtllm-runtime:${NGC_VERSION_TAG}-efa"

          # ---- Frontend image (already multi-arch from build-frontend-image workflow) ----
          echo ""
          echo "=== Frontend Image ==="
          SOURCE="${ECR_HOSTNAME}/${REGISTRY_IMAGE}:${COMMIT_SHA}-frontend"
          TARGET="${NGC_REGISTRY}/${NGC_ORG}/ai-dynamo/dynamo-frontend:${NGC_VERSION_TAG}"
          copy_image "${SOURCE}" "${TARGET}" "dynamo-frontend:${NGC_VERSION_TAG}"

          # ---- Operator image (multi-arch manifest already built by post-merge operator-build) ----
          echo ""
          echo "=== Operator Image ==="
          OPERATOR_SOURCE="${ECR_HOSTNAME}/${REGISTRY_IMAGE}:${COMMIT_SHA}-operator"
          OPERATOR_TARGET="${NGC_REGISTRY}/${NGC_ORG}/ai-dynamo/kubernetes-operator:${NGC_VERSION_TAG}"
          copy_image "${OPERATOR_SOURCE}" "${OPERATOR_TARGET}" "kubernetes-operator:${NGC_VERSION_TAG}"

          # ---- Planner image (CPU-only, multi-arch from post-merge planner-pipeline) ----
          echo ""
          echo "=== Planner Image ==="
          SOURCE="${ECR_HOSTNAME}/${REGISTRY_IMAGE}:${COMMIT_SHA}-dynamo-planner"
          TARGET="${NGC_REGISTRY}/${NGC_ORG}/ai-dynamo/dynamo-planner:${NGC_VERSION_TAG}"
          copy_image "${SOURCE}" "${TARGET}" "dynamo-planner:${NGC_VERSION_TAG}"
          # ---- Snapshot image ----
          echo ""
          echo "=== Snapshot Image ==="
          SNAPSHOT_SOURCE="${ECR_HOSTNAME}/${REGISTRY_IMAGE}:${COMMIT_SHA}-snapshot-agent"
          SNAPSHOT_TARGET="${NGC_REGISTRY}/${NGC_ORG}/ai-dynamo/snapshot-agent:${NGC_VERSION_TAG}"
          copy_image "${SNAPSHOT_SOURCE}" "${SNAPSHOT_TARGET}" "snapshot-agent:${NGC_VERSION_TAG}"

          # ---- Summary ----
          echo "successful_count=${#SUCCESSFUL_COPIES[@]}" >> $GITHUB_OUTPUT
          echo "failed_count=${#FAILED_COPIES[@]}" >> $GITHUB_OUTPUT

          printf '%s\n' "${SUCCESSFUL_COPIES[@]}" > /tmp/successful_copies.txt
          printf '%s\n' "${FAILED_COPIES[@]}" > /tmp/failed_copies.txt 2>/dev/null || true

          echo "========================================"
          echo "NGC Publishing Summary:"
          echo "  Successful: ${#SUCCESSFUL_COPIES[@]}"
          echo "  Failed: ${#FAILED_COPIES[@]}"
          echo "========================================"

          if [ ${#SUCCESSFUL_COPIES[@]} -eq 0 ]; then
            echo "ERROR: No images were successfully copied to NGC!"
            exit 1
          fi

      - name: Package and push Helm charts to NGC
        env:
          NGC_HELM_REPO: https://helm.ngc.nvidia.com/${{ secrets.NGC_PUBLISH_ORG }}/ai-dynamo
          NGC_TOKEN: ${{ secrets.NGC_PUBLISH_TOKEN }}
          NGC_USERNAME: ${{ secrets.NGC_PUBLISH_USERNAME }}
          HELM_CHART_VERSION: ${{ steps.rc_tag.outputs.helm_chart_version }}
        run: |
          set -euo pipefail

          REPO_ALIAS="ngc-staging-dynamo"

          helm plugin install https://github.com/chartmuseum/helm-push || true

          echo "${NGC_TOKEN}" | helm repo add "${REPO_ALIAS}" \
            --username="${NGC_USERNAME}" \
            --password-stdin \
            "${NGC_HELM_REPO}" > /dev/null 2>&1

          helm repo add nats https://nats-io.github.io/k8s/helm/charts/ || true
          helm repo add bitnami https://charts.bitnami.com/bitnami || true

          echo "" >> $GITHUB_STEP_SUMMARY
          echo "### Helm Charts" >> $GITHUB_STEP_SUMMARY

          PLATFORM_CHART_DIR="deploy/helm/charts/platform"
          CHART_NAME=$(awk '/^name:/ {print $2}' "${PLATFORM_CHART_DIR}/Chart.yaml")
          pushd "${PLATFORM_CHART_DIR}"
          helm dep build .
          popd

          echo "Packaging ${CHART_NAME} with version ${HELM_CHART_VERSION}..."
          helm package \
            --version "${HELM_CHART_VERSION}" \
            --app-version "${HELM_CHART_VERSION}" \
            "${PLATFORM_CHART_DIR}"

          CHART_FILE="${CHART_NAME}-${HELM_CHART_VERSION}.tgz"
          echo "Pushing ${CHART_FILE} to NGC Helm registry..."
          helm cm-push "${CHART_FILE}" "${REPO_ALIAS}"
          echo "- \`${CHART_NAME}:${HELM_CHART_VERSION}\` pushed to NGC Helm registry" >> $GITHUB_STEP_SUMMARY

          helm repo remove "${REPO_ALIAS}"

      - name: Create release summary
        env:
          RC_TAG: ${{ steps.rc_tag.outputs.rc_tag }}
          NGC_VERSION_TAG: ${{ steps.rc_tag.outputs.ngc_version_tag }}
          HELM_CHART_VERSION: ${{ steps.rc_tag.outputs.helm_chart_version }}
          SUCCESSFUL_COUNT: ${{ steps.copy_images.outputs.successful_count }}
          FAILED_COUNT: ${{ steps.copy_images.outputs.failed_count }}
        run: |
          echo "## Release Summary" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "| Property | Value |" >> $GITHUB_STEP_SUMMARY
          echo "|----------|-------|" >> $GITHUB_STEP_SUMMARY
          echo "| Version | ${VERSION} |" >> $GITHUB_STEP_SUMMARY
          echo "| Git Tag | ${RC_TAG} |" >> $GITHUB_STEP_SUMMARY
          echo "| NGC Version Tag | ${NGC_VERSION_TAG} |" >> $GITHUB_STEP_SUMMARY
          echo "| Source Commit SHA | ${COMMIT_SHA} |" >> $GITHUB_STEP_SUMMARY
          echo "| Branch | ${{ github.ref_name }} |" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "### NGC Publishing Results" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "- **Successful copies**: ${SUCCESSFUL_COUNT}" >> $GITHUB_STEP_SUMMARY
          echo "- **Failed copies**: ${FAILED_COUNT}" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "### Expected Images" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "Runtime images (CUDA 12 - default):" >> $GITHUB_STEP_SUMMARY
          echo "- \`vllm-runtime:${NGC_VERSION_TAG}\`" >> $GITHUB_STEP_SUMMARY
          echo "- \`sglang-runtime:${NGC_VERSION_TAG}\`" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "Runtime images (CUDA 13):" >> $GITHUB_STEP_SUMMARY
          echo "- \`vllm-runtime:${NGC_VERSION_TAG}-cuda13\`" >> $GITHUB_STEP_SUMMARY
          echo "- \`sglang-runtime:${NGC_VERSION_TAG}-cuda13\`" >> $GITHUB_STEP_SUMMARY
          echo "- \`tensorrtllm-runtime:${NGC_VERSION_TAG}-cuda13\`" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "EFA runtime images (amd64 only):" >> $GITHUB_STEP_SUMMARY
          echo "- \`vllm-runtime:${NGC_VERSION_TAG}-efa\`" >> $GITHUB_STEP_SUMMARY
          echo "- \`tensorrtllm-runtime:${NGC_VERSION_TAG}-efa\`" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "Operator image:" >> $GITHUB_STEP_SUMMARY
          echo "- \`kubernetes-operator:${NGC_VERSION_TAG}\`" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "Planner image:" >> $GITHUB_STEP_SUMMARY
          echo "- \`dynamo-planner:${NGC_VERSION_TAG}\` (multi-arch)" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "Frontend images:" >> $GITHUB_STEP_SUMMARY
          echo "- \`dynamo-frontend:${NGC_VERSION_TAG}\` (multi-arch)" >> $GITHUB_STEP_SUMMARY
          echo "" >> $GITHUB_STEP_SUMMARY
          echo "Helm chart:" >> $GITHUB_STEP_SUMMARY
          echo "- \`dynamo-platform:${HELM_CHART_VERSION}\` (pushed to NGC Helm registry)" >> $GITHUB_STEP_SUMMARY