AgentSkillsCN

Run Reviewer Tests

运行评审测试

SKILL.md

run-reviewer-tests

Run UI tests specifically for the SegmentEditorAdaptiveBrushReviewer module.

Usage

code
/run-reviewer-tests [--exit] [suite]

Where:

  • •--exit - Exit Slicer after tests complete (for automated iteration)
  • •suite is one of:
    • •reviewer (default) - Run all reviewer tests
    • •reviewer_unit - Run unit tests only (SequenceRecorder, ViewGroupManager, SceneViewBookmarks)
    • •reviewer_ui - Run UI tests only (slice navigation, bookmarks, playback, visualization, rating, keyboard)
    • •reviewer_integration - Run integration tests only (full workflow)

What This Skill Does

  1. •Reads SLICER_PATH from .env file
  2. •Launches Slicer with the test runner script
  3. •Executes Reviewer test cases (10 tests total)
  4. •Captures screenshots at each UI interaction
  5. •Reports pass/fail with detailed assertions

Test Categories

Unit Tests (reviewer_unit) - 3 tests

TestDescription
unit_sequence_recorderSequenceRecorder start/stop/step/navigation
unit_view_group_managerViewGroupManager linking, slice offsets
unit_scene_view_bookmarksSceneViewBookmarks add/restore/delete

UI Tests (reviewer_ui) - 6 tests

TestDescription
ui_reviewer_slice_navigationSlider, nav buttons, boundaries, linking
ui_reviewer_bookmarksAdd/restore/delete buttons, combo box
ui_reviewer_workflow_playbackRecording start/stop, playback navigation
ui_reviewer_visualizationLayout buttons, view modes, toggles
ui_reviewer_ratingRating buttons 1-4, notes, save, trial nav
ui_reviewer_keyboardAll keyboard shortcuts

Integration Tests (reviewer_integration) - 1 test

TestDescription
integration_reviewer_workflowFull 9-phase workflow test

Execution Steps

Step 1: Read Slicer Path from .env

bash
grep SLICER_PATH .env | cut -d= -f2

Step 2: Launch Slicer with Test Runner

For automated iteration (faster):

bash
"$SLICER_PATH" --python-script scripts/run_tests.py --exit reviewer_unit

For interactive follow-up:

bash
"$SLICER_PATH" --python-script scripts/run_tests.py reviewer

Step 3: Check Results

Read the output folder for results:

code
test_runs/<timestamp>_reviewer/
├── results.json          # Pass/fail for each test
├── metadata.json         # Summary statistics
├── screenshots/
│   ├── manifest.json     # Screenshot descriptions
│   └── *.png             # Captured UI states
└── logs/
    ├── test_run.log      # Test execution log
    └── slicer_session.log

Step 4: Analyze Failures

For each failed test:

  1. •Read the error message in results.json
  2. •Check screenshots for visual context
  3. •Search logs for tracebacks

Expected Bugs

The plan identified these likely bugs to find:

  1. •

    'list' object has no attribute 'get' in _on_run_selected (line ~1509)

    • •Cause: ResultsLoader returns list for test format runs
    • •Fix: Add type check before .get() call
  2. •

    Bidirectional sync race conditions

    • •Cause: Slider updates triggering callbacks during programmatic changes
    • •Fix: Use guard flag or disconnect signals during update
  3. •

    Bookmark restoration with changed scene

    • •Cause: SceneView references deleted nodes
    • •Fix: Validate nodes exist before restore
  4. •

    Metrics with empty segments

    • •Cause: Division by zero or empty labelmap
    • •Fix: Check segment voxel count > 0
  5. •

    Recording without loaded segmentation

    • •Cause: Start recording clicked before loading data
    • •Fix: Disable button until segmentation loaded
  6. •

    Keyboard shortcuts when dialogs open

    • •Cause: Shortcuts fire even during modal dialogs
    • •Fix: Check for modal state in event filter

Recommended Workflow

  1. •

    Start with unit tests (fastest feedback):

    code
    /run-reviewer-tests --exit reviewer_unit
  2. •

    Fix any unit test failures before moving to UI tests

  3. •

    Run UI tests:

    code
    /run-reviewer-tests --exit reviewer_ui
  4. •

    Review screenshots for UI issues

  5. •

    Run full integration:

    code
    /run-reviewer-tests --exit reviewer_integration

Follow-up Skills

After running tests:

  • •/review-test-results - Analyze failures and screenshots
  • •Use reviewer-bug-fixer agent - Fix identified bugs
  • •Use ui-improver agent - Address UI issues from screenshots