# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements.  See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership.  The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License.  You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied.  See the License for the
# specific language governing permissions and limitations
# under the License.

name: CI

on:
  pull_request:
    branches: [main]
  push:
    branches: [main]
  workflow_dispatch:

concurrency:
  group: ci-${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: ${{ github.event_name == 'pull_request' }}

permissions:
  contents: read

jobs:
  # Path planning and install-free repository contracts stay on the cheap lane
  # that every pull request must run. The outputs select the separate heavy lane
  # without teaching the workflow a second copy of the path rules.
  plan:
    runs-on: ubuntu-latest
    timeout-minutes: 45
    outputs:
      asf_source: ${{ steps.plan.outputs.asf_source }}
      astryx_surface: ${{ steps.plan.outputs.astryx_surface }}
      cli_package: ${{ steps.plan.outputs.cli_package }}
      code: ${{ steps.plan.outputs.code }}
      e2e: ${{ steps.plan.outputs.e2e }}
      heavy: ${{ steps.plan.outputs.heavy }}
      release_contract: ${{ steps.plan.outputs.release_contract }}
      runtime_host: ${{ steps.plan.outputs.runtime_host }}
      runtime_sandbox: ${{ steps.plan.outputs.runtime_sandbox }}
      storage_stress: ${{ steps.plan.outputs.storage_stress }}
      storybook: ${{ steps.plan.outputs.storybook }}
      standard_workspaces: ${{ steps.plan.outputs.standard_workspaces }}
    steps:
      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
        with:
          fetch-depth: 0
          persist-credentials: false

      - id: plan
        name: Select affected test surfaces
        env:
          BASE_SHA: ${{ github.event_name == 'push' && github.event.before || github.event.pull_request.base.sha }}
          HEAD_SHA: ${{ github.event_name == 'push' && github.sha || github.event.pull_request.head.sha }}
        run: |
          # PR checks may predate later main changes, so the planning lane also
          # validates the exact merged delta. Dispatches and unavailable history
          # fail safe to every surface.
          if [[ "${{ github.event_name }}" == "workflow_dispatch" ]] || [[ "$BASE_SHA" =~ ^0+$ ]] || ! git cat-file -e "${BASE_SHA}^{commit}"; then
            node scripts/ci-test-plan.mjs --full >> "$GITHUB_OUTPUT"
          else
            node scripts/ci-test-plan.mjs --base "$BASE_SHA" --head "$HEAD_SHA" >> "$GITHUB_OUTPUT"
          fi

      - name: Test CI planner
        run: node --test --test-concurrency=1 scripts/ci-test-plan.test.mjs scripts/verify-windows-harness.test.mjs

      # Pure Node like the planner test, and the labelling workflow imports this
      # module directly, so a tier or exclusion change is caught here rather
      # than by mislabelling live pull requests.
      - name: Test PR effort classification
        run: node --test --test-concurrency=1 scripts/pr-effort.test.mjs

      # The scheduled lifecycle workflow imports this pure policy module. Keep
      # its time boundaries and exemptions deterministic before it can write.
      - name: Test issue and PR lifecycle policy
        run: node --test --test-concurrency=1 scripts/issue-pr-lifecycle.test.mjs

      # Same shape and the same needs: a regenerate-and-diff contract that runs
      # on Node alone, so it belongs beside the planner test rather than behind
      # an install.
      - name: Check Windows test inventory
        run: npm run windows:inventory

      # Runs on the PR merge result: after a sibling protocol change lands on
      # main with the same epoch text, the silently merged tree still carries
      # the current base parent's epoch and this fails instead of shipping two
      # incompatible protocols under one number (#3313).
      - name: Guard the protocol compatibility epoch
        if: github.event_name == 'pull_request'
        run: node scripts/protocol-epoch-check.mjs --base 'HEAD^1'

      - name: Test the epoch guard
        run: node --test --test-concurrency=1 scripts/protocol-epoch-check.test.mjs

      - name: Test Computer Use script contracts
        run: node --test scripts/ax-tree-audit.test.mjs scripts/computer-use/lab-root.test.mjs

      - name: Test script entrypoint contracts
        run: node --test scripts/script-entrypoints.test.mjs

      # Install-free like its neighbours: the gate reads one source file and
      # compares it to a hand-edited inventory, so a hook that silently widens
      # its scope to the whole tree fails here rather than in a profile (#4109).
      - name: Check the hooks scoped to the whole shell
        run: npm run check:app-shell-hooks

      - name: Test the app-shell hook gate
        run: node --test --test-concurrency=1 scripts/check-app-shell-hooks.test.mjs

      - name: Verify ASF npm preflight policy
        run: npm run check:asf-npm

      # The source-header gate has to see every file that lands, not only the
      # files an affected surface selects, so it runs unconditionally beside
      # the other install-free checks.
      - name: Check ASF source headers
        run: npm run check:asf-headers

  heavy:
    needs: plan
    if: needs.plan.outputs.heavy == 'true'
    runs-on: ubuntu-latest
    timeout-minutes: 45
    steps:
      - uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
        with:
          fetch-depth: 0
          persist-credentials: false

      - uses: actions/setup-node@820762786026740c76f36085b0efc47a31fe5020 # v7.0.0
        with:
          node-version: '24'
          cache: npm

      - name: Select the release npm toolchain
        if: needs.plan.outputs.cli_package == 'true'
        run: npm install --global --no-audit --no-fund "$(node -p 'require("./package.json").packageManager')"

      - name: Restore Electron artifact cache
        if: needs.plan.outputs.code == 'true'
        uses: actions/cache@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
        with:
          path: ~/.cache/electron
          key: electron-${{ runner.os }}-${{ hashFiles('package-lock.json') }}
          restore-keys: electron-${{ runner.os }}-

      - name: Install Linux runtime dependencies
        if: needs.plan.outputs.runtime_sandbox == 'true'
        run: sudo apt-get update && sudo apt-get install -y ripgrep bubblewrap

      # Ubuntu 24.04 hosted runners gate unprivileged user namespaces through
      # AppArmor, which otherwise makes bwrap fail while configuring loopback.
      - name: Enable bubblewrap user namespaces
        if: needs.plan.outputs.runtime_sandbox == 'true'
        run: |
          if [[ -e /proc/sys/kernel/apparmor_restrict_unprivileged_userns ]]; then
            sudo sysctl -w kernel.apparmor_restrict_unprivileged_userns=0
          fi
          if [[ -e /proc/sys/kernel/unprivileged_userns_clone ]]; then
            sudo sysctl -w kernel.unprivileged_userns_clone=1
          fi

      - name: Install dependencies
        if: needs.plan.outputs.code == 'true' || needs.plan.outputs.astryx_surface == 'true' || needs.plan.outputs.asf_source == 'true' || needs.plan.outputs.cli_package == 'true' || needs.plan.outputs.release_contract == 'true'
        run: npm ci

      # The header audit above remains install-free. The complete source gate
      # also exercises generation and therefore runs after its pinned formatter
      # dependency is installed, matching the source-candidate workflow.
      - name: Verify ASF source release mechanics
        if: needs.plan.outputs.asf_source == 'true'
        run: npm run check:asf-source

      # Parsed dependency rules and an exact legacy-debt ledger keep the
      # renderer root from absorbing new feature or Desktop ownership while
      # the existing AppShell is migrated behind stable boundaries.
      - name: Check renderer architecture
        if: needs.plan.outputs.code == 'true'
        env:
          BASE_SHA: ${{ github.event_name == 'push' && github.event.before || github.event.pull_request.base.sha }}
        run: |
          if [[ -n "$BASE_SHA" && ! "$BASE_SHA" =~ ^0+$ ]]; then
            npm run check:renderer-architecture -- --base "$BASE_SHA"
          else
            npm run check:renderer-architecture
          fi

      - name: Lint
        if: needs.plan.outputs.code == 'true'
        run: npm run lint

      - name: Check formatting
        if: needs.plan.outputs.code == 'true'
        run: npm run format:check

      # This generated artifact describes the whole renderer/UI tree, so every
      # code-validation run checks it even when the triggering diff is outside
      # an Astryx surface. Keep it before Build so stale output is reported
      # directly instead of being hidden behind an earlier compilation failure.
      - name: Astryx surface inventory
        if: needs.plan.outputs.code == 'true' || needs.plan.outputs.astryx_surface == 'true'
        run: |
          npm run astryx:surface-inventory
          npm run astryx:surface-inventory:test

      - name: Build
        if: needs.plan.outputs.code == 'true' || needs.plan.outputs.cli_package == 'true' || needs.plan.outputs.release_contract == 'true'
        run: npm run build

      - name: Release contracts
        if: needs.plan.outputs.release_contract == 'true'
        run: npm run check:release

      - name: Typecheck
        if: needs.plan.outputs.code == 'true'
        run: npm run typecheck

      # Two drift contracts over the shipped app-icon artwork, sitting beside
      # the theme drift check for the same reason: the committed bytes are a
      # build output that nothing else re-derives, so without this a change to
      # the generator, to DEFAULT_APP_ICON, or to the packaging config can go
      # green while the artwork it names no longer matches.
      - name: App icon artwork drift
        if: needs.plan.outputs.code == 'true'
        run: node --test scripts/verify-packaged-app-icons.test.mjs scripts/generate-app-icons.test.mjs

      - name: Astryx theme drift
        if: needs.plan.outputs.code == 'true'
        run: npm run astryx:theme -- --check

      - name: Knip (apps/desktop)
        if: needs.plan.outputs.code == 'true'
        run: npx knip --workspace apps/desktop

      - name: Knip (packages/ui)
        if: needs.plan.outputs.code == 'true'
        run: npx knip --workspace packages/ui

      - name: Linux sandbox smoke
        if: needs.plan.outputs.runtime_sandbox == 'true'
        env:
          MAKA_REQUIRE_LINUX_SANDBOX_SMOKE: '1'
        run: npm exec -w @maka/runtime -- node --test dist/__tests__/linux-sandbox-smoke.test.js

      - name: Run affected standard workspace tests
        if: needs.plan.outputs.standard_workspaces != ''
        env:
          STORAGE_STRESS: ${{ needs.plan.outputs.storage_stress }}
          WORKSPACES: ${{ needs.plan.outputs.standard_workspaces }}
        run: |
          if [[ "$STORAGE_STRESS" == "true" ]]; then
            export MAKA_STORAGE_STRESS=1
          fi
          node scripts/run-workspace-tests-parallel.mjs --concurrency=3 --workspaces="$WORKSPACES"

      - name: Run live Eval egress proxy test
        if: contains(needs.plan.outputs.standard_workspaces, 'packages/eval')
        env:
          MAKA_EVAL_EGRESS_PROXY_TEST: '1'
        run: |
          docker pull python:3.12-slim
          docker build \
            --tag maka-eval-egress-proxy:12.2.3 \
            --file packages/eval/harbor/egress-proxy/Dockerfile \
            packages/eval/harbor
          npm --workspace @maka/eval run test:egress-proxy:live

      - name: Run Runtime Host tests
        if: needs.plan.outputs.runtime_host == 'true'
        run: npm --workspace @maka/runtime-host run test:dist

      - name: Ensure xvfb
        if: needs.plan.outputs.e2e == 'true'
        run: command -v xvfb-run >/dev/null 2>&1 || { sudo apt-get update && sudo apt-get install -y xvfb; }

      - name: Desktop e2e
        if: needs.plan.outputs.e2e == 'true'
        run: xvfb-run -a npm exec -w @maka/desktop -- playwright test --config e2e/playwright.config.ts

      - name: Browser WebContentsView semantic smoke
        if: needs.plan.outputs.e2e == 'true'
        # Hosted Linux runners cannot configure Electron's SUID helper. This
        # smoke loads only its loopback fixture; production stays sandboxed.
        run: xvfb-run -a npm exec --workspace @maka/desktop -- electron --no-sandbox scripts/browser-observe-act-smoke.mjs

      - name: Alignment audit
        if: needs.plan.outputs.e2e == 'true'
        run: xvfb-run -a node scripts/audit-alignment.mjs

      - name: Install Playwright Chromium
        if: needs.plan.outputs.storybook == 'true'
        run: npx playwright install --with-deps chromium

      - name: Build Storybook
        if: needs.plan.outputs.storybook == 'true'
        run: npm --workspace @maka/desktop run build-storybook

      - name: Storybook smoke
        if: needs.plan.outputs.storybook == 'true'
        run: npm --workspace @maka/desktop run smoke:storybook

      - name: Update stable Rust for CLI packaging
        if: needs.plan.outputs.cli_package == 'true'
        run: rustup update stable --no-self-update

      - id: cli-rustc
        name: Resolve CLI Rust cache version
        if: needs.plan.outputs.cli_package == 'true'
        shell: bash
        run: |
          echo "version=$(rustc --version | cut -d ' ' -f 2)" >> "$GITHUB_OUTPUT"
          echo "revision=$(git rev-parse HEAD)" >> "$GITHUB_OUTPUT"
          {
            echo "KACHE_CACHE_DIR=${{ runner.temp }}/kache-cli-package"
            echo "KACHE_RUNTIME_DIR=${{ runner.temp }}/kache-cli-package-runtime"
            echo "RUSTC_WRAPPER=kache"
          } >> "$GITHUB_ENV"

      - name: Install Kache for CLI packaging
        if: needs.plan.outputs.cli_package == 'true'
        uses: taiki-e/install-action@1ed6d7be6168f6c9046541087ff549b6bc581fdf # v2
        with:
          tool: kache@0.16.0

      - id: cli-kache-cache
        name: Restore CLI Rust build cache
        if: needs.plan.outputs.cli_package == 'true'
        uses: actions/cache/restore@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
        with:
          path: ${{ runner.temp }}/kache-cli-package
          key: kache-runtime-host-peer-cli-package-v0.16.0-${{ runner.os }}-${{ runner.arch }}-rust-${{ steps.cli-rustc.outputs.version }}-${{ steps.cli-rustc.outputs.revision }}
          restore-keys: |
            kache-runtime-host-peer-cli-package-v0.16.0-${{ runner.os }}-${{ runner.arch }}-rust-${{ steps.cli-rustc.outputs.version }}-

      - name: Install cargo-deny for CLI packaging
        if: needs.plan.outputs.cli_package == 'true'
        uses: taiki-e/install-action@1ed6d7be6168f6c9046541087ff549b6bc581fdf # v2
        with:
          tool: cargo-deny@0.20.2

      - name: Build CLI release candidate
        if: needs.plan.outputs.cli_package == 'true'
        run: npm run release:cli:pack -- --allow-dirty

      - name: Report CLI Rust build cache
        if: needs.plan.outputs.cli_package == 'true'
        shell: bash
        run: kache report --format github >> "$GITHUB_STEP_SUMMARY"

      - name: Save CLI Rust build cache
        if: needs.plan.outputs.cli_package == 'true' && github.ref_name == github.event.repository.default_branch
        uses: actions/cache/save@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
        with:
          path: ${{ runner.temp }}/kache-cli-package
          key: ${{ steps.cli-kache-cache.outputs.cache-primary-key }}

      - name: Validate installed CLI release candidate
        if: needs.plan.outputs.cli_package == 'true'
        run: npm run release:cli:smoke -- packages/cli/release/*.tgz

  # Branch protection requires the `test` context on every pull request. Keep
  # this aggregation job unconditional so documentation-only plans report
  # success while selected heavy validation still propagates every failure.
  test:
    needs: [plan, heavy]
    if: always()
    runs-on: ubuntu-latest
    timeout-minutes: 45
    steps:
      - name: Report required test status
        env:
          HEAVY_RESULT: ${{ needs.heavy.result }}
          HEAVY_SELECTED: ${{ needs.plan.outputs.heavy }}
          PLAN_RESULT: ${{ needs.plan.result }}
        run: |
          if [[ "$PLAN_RESULT" != "success" ]]; then
            echo "::error::CI planning and lightweight checks ended with $PLAN_RESULT"
            exit 1
          fi

          case "$HEAVY_SELECTED" in
            true)
              if [[ "$HEAVY_RESULT" != "success" ]]; then
                echo "::error::Selected heavy CI ended with $HEAVY_RESULT"
                exit 1
              fi
              ;;
            false)
              if [[ "$HEAVY_RESULT" != "skipped" ]]; then
                echo "::error::Unselected heavy CI ended with $HEAVY_RESULT instead of skipped"
                exit 1
              fi
              ;;
            *)
              echo "::error::CI planner produced an invalid heavy selection: $HEAVY_SELECTED"
              exit 1
              ;;
          esac

          echo "Required CI passed (heavy selected: $HEAVY_SELECTED)"