Files
ComfyUI_frontend/browser_tests
jaeone94 82bea29dda fix: defer node auto-pan until drag starts (#12654)
## Summary

Fix a Vue node drag edge case where holding the partially off-screen
Advanced inputs button could continuously auto-pan the canvas even
though the pointer had not moved into an actual drag.

Linear:
[FE-938](https://linear.app/comfyorg/issue/FE-938/holding-partially-off-screen-advanced-inputs-causes-continuous)

## Changes

- **What**: Move Vue node auto-pan initialization from `startDrag()` to
`handleDrag()`, so auto-pan starts only after the pointer interaction
has become a real drag.
- **What**: Keep the existing auto-pan behavior during active drags by
creating the controller on the first `handleDrag()` call, updating its
pointer position on later drag frames, and preserving the existing
`onPan` position adjustments.
- **What**: Add unit coverage for the important drag lifecycle
invariants: no auto-pan on pointerdown/startDrag, auto-pan starts on
handleDrag, the same controller is reused across handleDrag calls, and
cleanup still stops auto-pan on endDrag.
- **What**: Add a Playwright regression that places the Advanced inputs
button partially beyond the visible canvas edge, holds the pointer down
without moving, and verifies the canvas offset stays stable.
- **What**: Add `data-testid="advanced-inputs-button"` to the Advanced
inputs footer button variants so the regression test does not depend on
translated button text.
- **Breaking**: None.
- **Dependencies**: None.

## Root Cause

`useNodeDrag.startDrag()` created and started `AutoPanController`
immediately on pointerdown. When the Advanced inputs button was partly
outside the canvas bounds, a stationary pointer near the visible canvas
edge was enough for auto-pan to begin, even before any drag movement
occurred.

The pointer interaction layer already distinguishes press/hold from real
dragging before calling `handleDrag()`. Deferring auto-pan to
`handleDrag()` aligns auto-pan startup with that drag threshold and
prevents a plain hold from panning the canvas.

## Review Focus

- Auto-pan should not start from `startDrag()`/pointerdown alone.
- Auto-pan should still start promptly once `handleDrag()` runs for an
actual drag.
- Repeated `handleDrag()` calls should reuse the existing controller
rather than recreate it.
- Existing `onPan` behavior should continue to update drag start
positions, selected node start positions, selected groups, and node
positions during active drags.
- The E2E intentionally asserts the canvas offset, not node bounds,
because the reported bug is unintended canvas auto-pan while the pointer
is stationary.

## Red-Green Verification

- `a00b5d2fb test: add failing advanced button hold pan regression`:
adds the Playwright regression and test id plumbing. This was verified
red against the pre-fix production code.
- `5c207ae28 fix: defer node auto-pan until drag starts`: adds the
production fix and unit coverage. The same regression is verified green
with the fix.

## Validation

- `pnpm format`
- `pnpm lint`
- `pnpm typecheck`
- `pnpm typecheck:browser`
- `pnpm test:unit`
- `pnpm test:unit
src/renderer/extensions/vueNodes/layout/useNodeDrag.test.ts`
- `PLAYWRIGHT_SETUP_API_URL=http://127.0.0.1:8188 pnpm
test:browser:local
browser_tests/tests/vueNodes/interactions/node/move.spec.ts --grep
"should not pan while holding the Advanced button without dragging"`
- Pre-push hook: `pnpm knip --cache`

## Screenshots (Before / After)

Before 


https://github.com/user-attachments/assets/6080de2d-e2da-4b38-a1ed-1f1f88548c2d

After 


https://github.com/user-attachments/assets/f331271a-9ea1-41ec-92cb-974bc57be56b
2026-06-05 05:26:18 +00:00
..
2026-01-27 17:59:19 -08:00

Playwright Testing for ComfyUI_frontend

This document outlines the setup, usage, and common patterns for Playwright browser tests in the ComfyUI_frontend project.

Prerequisites

CRITICAL: Start ComfyUI backend with --multi-user flag:

python main.py --multi-user

Without this flag, parallel tests will conflict and fail randomly.

Setup

ComfyUI devtools

ComfyUI_devtools is included in this repository under tools/devtools/. During CI/CD, these files are automatically copied to the custom_nodes directory.
ComfyUI_devtools adds additional API endpoints and nodes to ComfyUI for browser testing.

For local development, copy the devtools files to your ComfyUI installation:

cp -r tools/devtools/* /path/to/your/ComfyUI/custom_nodes/ComfyUI_devtools/

Node.js & Playwright Prerequisites

Ensure you have the Node.js version specified in .nvmrc installed. Then, set up the Chromium test driver:

pnpm exec playwright install chromium --with-deps

Environment Configuration

Create .env from the template:

cp .env_example .env

Key settings for debugging:

# Remove Vue dev overlay that blocks UI elements
DISABLE_VUE_PLUGINS=true

# Test against dev server (recommended) or backend directly
PLAYWRIGHT_TEST_URL=http://localhost:5173  # Dev server
# PLAYWRIGHT_TEST_URL=http://localhost:8188  # Direct backend
PLAYWRIGHT_SETUP_API_URL=http://localhost:8188  # Setup/auth API when using the dev server URL above

# Path to ComfyUI for backing up user data/settings before tests
TEST_COMFYUI_DIR=/path/to/your/ComfyUI

Common Setup Issues

Release API Mocking

By default, all tests mock the release API (api.comfy.org/releases) to prevent release notification popups from interfering with test execution. This is necessary because the release notifications can appear over UI elements and block test interactions.

To test with real release data, you can disable mocking:

await comfyPage.setup({ mockReleases: false })

For tests that specifically need to test release functionality, see the example in tests/releaseNotifications.spec.ts.

Running Tests

Always use UI mode for development:

pnpm test:browser:local --ui

UI mode features:

  • Locator picker: Click the target icon, then click any element to get the exact locator code to use in your test. The code appears in the Locator tab.
  • Step debugging: Step through your test line-by-line by clicking Source tab
  • Time travel: In the Actions tab/panel, click any step to see the browser state at that moment
  • Console/Network Tabs: View logs and API calls at each step
  • Attachments Tab: View all snapshots with expected and actual images

Playwright UI Mode

For CI or headless testing:

pnpm test:browser:local                    # Run all tests
pnpm test:browser:local widget.spec.ts     # Run specific test file

Slowing the browser down for debugging

When running with --headed (or --ui), set SLOW_MO to a millisecond delay to slow every Playwright action down so you can watch what is happening. The delay only applies when PLAYWRIGHT_LOCAL is set (the default for the pnpm test:browser:local script).

SLOW_MO=250 pnpm test:browser:local --headed widget.spec.ts

Test Structure

Browser tests in this project follow a specific organization pattern:

  • Fixtures: Located in fixtures/ - These provide test setup and utilities

    • ComfyPage.ts - The main fixture for interacting with ComfyUI
    • ComfyMouse.ts - Utility for mouse interactions with the canvas
    • Components fixtures in fixtures/components/ - Page object models for UI components
  • Tests: Located in tests/ - The actual test specifications

    • Organized by functionality (e.g., widget.spec.ts, interaction.spec.ts)
    • Snapshot directories (e.g., widget.spec.ts-snapshots/) contain reference screenshots
  • Utilities: Located in utils/ - Common utility functions

    • litegraphUtils.ts - Utilities for working with LiteGraph nodes

Writing Effective Tests

When writing new tests, follow these patterns:

Test Structure

// Import the test fixture
import { comfyPageFixture as test } from '@e2e/fixtures/ComfyPage'

test.describe('Feature Name', () => {
  // Set up test environment if needed
  test.beforeEach(async ({ comfyPage }) => {
    // Common setup
  })

  test('should do something specific', async ({ comfyPage }) => {
    // Test implementation
  })
})

Leverage Existing Fixtures and Helpers

Always check for existing helpers and fixtures before implementing new ones:

  • ComfyPage: Main fixture with methods for canvas interaction and node management
  • ComfyMouse: Helper for precise mouse operations on the canvas
  • Component Fixtures: Check browser_tests/fixtures/components/ for UI component page objects (e.g. Actionbar.ts, Templates.ts, ContextMenu.ts)
  • Helper Classes: Check browser_tests/fixtures/helpers/ for domain-specific helper classes wired into ComfyPage (e.g. CanvasHelper.ts, WorkflowHelper.ts)
  • Utility Functions: Check browser_tests/fixtures/utils/ for standalone utilities (e.g. fitToView.ts, clipboardSpy.ts, builderTestUtils.ts)

Most common testing needs are already addressed by these helpers, which will make your tests more consistent and reliable.

Import Conventions

  • Prefer @e2e/* for imports within browser_tests/
  • Continue using @/* for imports from src/
  • Avoid introducing new deep relative imports within browser_tests/ when the alias is available

Key Testing Patterns

  1. Focus elements explicitly: Canvas-based elements often need explicit focus before interaction:

    // Click the canvas first to focus it before pressing keys
    await comfyPage.canvas.click()
    await comfyPage.page.keyboard.press('a')
    
  2. Mark canvas as dirty if needed: Some interactions need explicit canvas updates:

    // After programmatically changing node state, mark canvas dirty
    await comfyPage.page.evaluate(() => {
      window['app'].graph.setDirtyCanvas(true, true)
    })
    
  3. Use node references over coordinates: Node references from fixtures/utils/litegraphUtils.ts provide stable ways to interact with nodes:

    // Prefer this:
    const node = await comfyPage.getNodeRefsByType('LoadImage')[0]
    await node.click('title')
    
    // Over this:
    await comfyPage.canvas.click({ position: { x: 100, y: 100 } })
    
  4. Wait for canvas to render after UI interactions:

    await comfyPage.nextFrame()
    
  5. Clean up persistent server state: While most state is reset between tests, anything stored on the server persists:

    // Reset settings that affect other tests (these are stored on server)
    await comfyPage.setSetting('Comfy.ColorPalette', 'dark')
    await comfyPage.setSetting('Comfy.NodeBadge.NodeIdBadgeMode', 'None')
    
    // Clean up uploaded files if needed
    comfyPage.deleteFileAfterTest({ filename: 'image.png' })
    
  6. Prefer functional assertions over screenshots: Use screenshots only when visual verification is necessary:

    // Prefer this:
    await expect.poll(() => node.isPinned()).toBe(true)
    await expect.poll(() => node.getProperty('title')).toBe('Expected Title')
    
    // Over this - only use when needed:
    await expect(comfyPage.canvas).toHaveScreenshot('state.png')
    
  7. Use minimal test workflows: When creating test workflows, keep them as minimal as possible:

    // Include only the components needed for the test
    await comfyPage.loadWorkflow('single_ksampler')
    
  8. Debug helpers for visual debugging (remove before committing):

    ComfyPage includes temporary debug methods for troubleshooting:

    test('debug failing interaction', async ({ comfyPage }, testInfo) => {
      // Add visual markers to see click positions
      await comfyPage.debugAddMarker({ x: 100, y: 200 })
    
      // Attach screenshot with markers to test report
      await comfyPage.debugAttachScreenshot(testInfo, 'node-positions', {
        element: 'canvas',
        markers: [{ position: { x: 100, y: 200 } }]
      })
    
      // Show canvas overlay for easier debugging
      await comfyPage.debugShowCanvasOverlay()
    
      // Remember to remove debug code before committing!
    })
    

    Available debug methods:

    • debugAddMarker(position) - Red circle at position
    • debugAttachScreenshot(testInfo, name) - Attach to test report
    • debugShowCanvasOverlay() - Show canvas as overlay
    • debugGetCanvasDataURL() - Get canvas as base64

Common Patterns and Utilities

Page Object Pattern

Tests use the Page Object pattern to create abstractions over the UI:

// Using the ComfyPage fixture
test('Can toggle boolean widget', async ({ comfyPage }) => {
  await comfyPage.loadWorkflow('widgets/boolean_widget')
  const node = (await comfyPage.getFirstNodeRef())!
  const widget = await node.getWidget(0)
  await widget.click()
})

Node References

The NodeReference class provides helpers for interacting with LiteGraph nodes:

// Getting node by type and interacting with it
const nodes = await comfyPage.getNodeRefsByType('LoadImage')
const loadImageNode = nodes[0]
const widget = await loadImageNode.getWidget(0)
await widget.click()

Visual Regression Testing

Tests use screenshot comparisons to verify UI state:

// Take a screenshot and compare to reference
await expect(comfyPage.canvas).toHaveScreenshot('boolean_widget_toggled.png')

Waiting for Animations

Always call nextFrame() after actions that trigger animations:

await comfyPage.canvas.click({ position: { x: 100, y: 100 } })
await comfyPage.nextFrame() // Wait for canvas to redraw

Mouse Interactions

Canvas operations use special helpers to ensure proper timing:

// Using ComfyMouse for drag and drop
await comfyMouse.dragAndDrop(
  { x: 100, y: 100 }, // From
  { x: 200, y: 200 } // To
)

// Standard ComfyPage helpers
await comfyPage.drag({ x: 100, y: 100 }, { x: 200, y: 200 })
await comfyPage.pan({ x: 200, y: 200 })
await comfyPage.zoom(-100) // Zoom in

Workflow Management

Tests use workflows stored in assets/ for consistent starting points:

// Load a test workflow
await comfyPage.loadWorkflow('single_ksampler')

// Wait for workflow to load and stabilize
await comfyPage.nextFrame()

Custom Assertions

The project includes custom Playwright assertions through comfyExpect:

// Check if a node is in a specific state
await expect(node).toBePinned()
await expect(node).toBeBypassed()
await expect(node).toBeCollapsed()

Troubleshooting Common Issues

Flaky Tests

  • Timing Issues: Always wait for animations to complete with nextFrame()
  • Coordinate Sensitivity: Canvas coordinates are viewport-relative; use node references when possible
  • Test Isolation: Tests run in parallel; avoid dependencies between tests
  • Screenshots vary: Ensure your OS and browser match the reference environment (Linux)
  • Async / await: Race conditions are a very common cause of test flakiness

Screenshot Testing

Due to variations in system font rendering, screenshot expectations are platform-specific. Please note:

  • Do not commit local screenshot expectations to the repository
  • We maintain Linux screenshot expectations as our GitHub Action runner operates in a Linux environment
  • While developing, you can generate local screenshots for your tests, but these will differ from CI-generated ones

Working with Screenshots Locally

Option 1 - Skip screenshot tests (add to playwright.config.ts):

export default defineConfig({
  grep: process.env.CI ? undefined : /^(?!.*screenshot).*$/
})

Option 2 - Generate local baselines for comparison:

pnpm test:browser:local --update-snapshots

Creating New Screenshot Baselines

For PRs from Comfy-Org/ComfyUI_frontend branches:

  1. Write test with toHaveScreenshot('filename.png')
  2. Create PR and add New Browser Test Expectation label
  3. CI will generate and commit the Linux baseline screenshots

Note: Fork PRs cannot auto-commit screenshots. A maintainer will need to commit the screenshots manually for you (don't worry, they'll do it).

Viewing Test Reports

Automated Test Deployment

The project automatically deploys Playwright test reports to Cloudflare Pages for every PR and push to main branches.

Accessing Test Reports

  • From PR comments: Click the "View Report" links for each browser
  • Direct URLs: Reports are available at https://[branch].comfyui-playwright-[browser].pages.dev (branch-specific deployments)
  • From GitHub Actions: Download artifacts from failed runs

How It Works

  1. Test execution: All browser tests run in parallel across multiple browsers

  2. Report generation: HTML reports are generated for each browser configuration

  3. Cloudflare deployment: Each browser's report deploys to its own Cloudflare Pages project with branch isolation:

    • comfyui-playwright-chromium (with branch-specific URLs)
    • comfyui-playwright-mobile-chrome (with branch-specific URLs)
    • comfyui-playwright-chromium-2x (2x scale, with branch-specific URLs)
    • comfyui-playwright-chromium-0-5x (0.5x scale, with branch-specific URLs)
  4. PR comments: GitHub automatically updates PR comments with:

    • / Test status for each browser
    • Direct links to interactive test reports
    • Real-time progress updates as tests complete

Resources