callstack__agent-device/.github/workflows/macos.yml

name: macOS

on:
  pull_request:
    paths-ignore:
      - 'docs/**'
      - 'website/**'
      - 'README.md'
      - 'AGENTS.md'
      - 'CHANGELOG.md'
      - 'CONTEXT.md'
      - 'CONTRIBUTING.md'
      - 'LICENSE'
      - 'SECURITY.md'
      - '.github/actions/build-docs/action.yml'
      - '.github/workflows/deploy.yml'
      - '.github/workflows/pr-preview.yml'
      - '.github/workflows/pr-preview-cleanup.yml'
  push:
    branches:
      - main

permissions:
  contents: read

concurrency:
  group: ci-${{ github.workflow }}-${{ github.ref }}
  cancel-in-progress: true

jobs:
  # One macOS lane for everything the macOS host can prove: command coverage,
  # the runner's pure-decision XCTests, the helper, and the replay smoke. All
  # of it consumes one runner build, compiled with the unit-test flag so the
  # host XCTest run below has a bundle to execute (#1781 A7).
  #
  # Which tests that build reaches is not a list: it is everything the macOS
  # build compiles, and the simulator-only tests keep themselves out with an
  # `os(iOS)` guard (see the classification note in RunnerTests.swift). The
  # `-skip-testing:` entry is validated by `pnpm check:xctest-selection` for
  # the same reason the nightly's is: `RunnerTests/testCommand` is the
  # runner's 24-hour server entry point, and a typo re-arms the hang. The
  # lane's reporter asserts the executed count equals the reach the check
  # derives from source, so a build without the compile flag or a guard that
  # compiles a file out reads as red rather than as a smaller green.
  #
  # Kill criterion: if the host run proves flaky where the simulator run of the
  # same tests is not (a platform-branch difference this classification
  # missed), gate the offender `os(iOS)` and it returns to the simulator lanes;
  # if that happens to more than a handful, the host lane is the wrong tool and
  # goes.
  smoke-macos:
    name: Smoke Tests
    runs-on: macos-26
    timeout-minutes: 80
    env:
      AGENT_DEVICE_STATE_DIR: ${{ github.workspace }}/.tmp/agent-device-state
      AGENT_DEVICE_IOS_RUNNER_DERIVED_PATH: ${{ github.workspace }}/.tmp/macos-runner-derived
      AGENT_DEVICE_XCUITEST_INCLUDE_UNIT_TESTS: '1'
      RESULT_BUNDLE_PATH: ${{ github.workspace }}/.tmp/xctest-host/RunnerTests.xcresult
      RESULT_SUMMARY_PATH: ${{ github.workspace }}/.tmp/xctest-host/test-results-summary.json
      XCTEST_LANE: host
    steps:
      - name: Checkout
        uses: actions/checkout@de0fac2e4500dabe0009e67214ff5f5447ce83dd # v6.0.2

      - name: Setup toolchain
        uses: ./.github/actions/setup-node-pnpm

      - name: Run macOS command coverage contract
        uses: ./.github/actions/run-gate
        with: { gate: macos-coverage }

      - name: Restore and build macOS XCTest runner
        uses: ./.github/actions/setup-apple-runner-build
        with:
          derived-path: ${{ env.AGENT_DEVICE_IOS_RUNNER_DERIVED_PATH }}
          cache-key-prefix: macos-runner-prebuilt
          gate: swift-runner-macos
          xcuitest-platform: macos
          xcuitest-destination: platform=macOS,arch=arm64

      - name: Run the host-runnable runner XCTests
        run: |
          set -euo pipefail
          XCTESTRUN_PATH="$(find "$AGENT_DEVICE_IOS_RUNNER_DERIVED_PATH/Build/Products" -maxdepth 1 -name '*.xctestrun' -print -quit)"
          test -n "$XCTESTRUN_PATH"
          mkdir -p "$(dirname "$RESULT_BUNDLE_PATH")"
          xcodebuild test-without-building \
            -xctestrun "$XCTESTRUN_PATH" \
            -destination 'platform=macOS,arch=arm64' \
            -skip-testing:AgentDeviceRunnerUITests/RunnerTests/testCommand \
            -resultBundlePath "$RESULT_BUNDLE_PATH"

      # Same shape as xctest-nightly.yml: extract is best-effort, the report step asserts.
      - name: Extract the test-results summary
        if: always()
        run: |
          set -uo pipefail
          [ -d "$RESULT_BUNDLE_PATH" ] || exit 0
          xcrun xcresulttool get test-results summary --path "$RESULT_BUNDLE_PATH" --compact \
            > "$RESULT_SUMMARY_PATH" 2>/dev/null ||
            xcrun xcresulttool get test-results summary --path "$RESULT_BUNDLE_PATH" \
              > "$RESULT_SUMMARY_PATH" 2>/dev/null ||
            rm -f "$RESULT_SUMMARY_PATH"
          exit 0

      - name: Report the run and assert it executed the source-derived set
        if: always()
        run: |
          set -euo pipefail
          if [ ! -s "$RESULT_SUMMARY_PATH" ]; then
            echo 'No usable test-results summary was produced, so nothing here can show a test ran.'
            echo 'The suite step above failed before or during the run; read its log first.'
            echo '### iOS runner host XCTest lane' >> "$GITHUB_STEP_SUMMARY"
            echo 'No result-bundle summary; see the run log and the uploaded bundle.' >> "$GITHUB_STEP_SUMMARY"
            exit 1
          fi
          node --experimental-strip-types scripts/xctest-run-summary.ts

      - name: Upload result bundle
        if: always()
        uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2
        with:
          name: xctest-host-results-${{ github.run_id }}-${{ github.run_attempt }}
          path: .tmp/xctest-host
          if-no-files-found: warn

      - name: Build macOS helper
        uses: ./.github/actions/run-gate
        with: { gate: macos-helper }

      - name: Run macOS integration test
        uses: ./.github/actions/run-gate
        with:
          gate: replay-macos
          args: |
            --retries
            2
            --report-junit
            test/artifacts/replays-macos.junit.xml

      - name: Upload macOS artifacts
        if: always()
        uses: ./.github/actions/upload-agent-device-artifacts
        with:
          artifact-name: macos-artifacts
          agent-state-dir: ${{ env.AGENT_DEVICE_STATE_DIR }}
          runner-derived-path: ${{ env.AGENT_DEVICE_IOS_RUNNER_DERIVED_PATH }}