fix(ci): streamline Playwright configuration and remove preflight setup test

2026-02-04 23:34:48 +00:00
parent 7c81c7e3de
commit 22619326de
4 changed files with 35 additions and 237 deletions
@@ -9,7 +9,6 @@ applyTo: '**'
 - **Locators**: Prioritize user-facing, role-based locators (`getByRole`, `getByLabel`, `getByText`, etc.) for resilience and accessibility. Use `test.step()` to group interactions and improve test readability and reporting.
 - **Assertions**: Use auto-retrying web-first assertions. These assertions start with the `await` keyword (e.g., `await expect(locator).toHaveText()`). Avoid `expect(locator).toBeVisible()` unless specifically testing for visibility changes.
 - **Timeouts**: Rely on Playwright's built-in auto-waiting mechanisms. Avoid hard-coded waits or increased default timeouts.
- **Switch/Toggle Components**: Use helper functions from `tests/utils/ui-helpers.ts` (`clickSwitch`, `expectSwitchState`, `toggleSwitch`) for reliable interactions. Never use `{ force: true }` or direct clicks on hidden inputs.
 - **Clarity**: Use descriptive test and step titles that clearly state the intent. Add comments only to explain complex logic or non-obvious interactions.


@@ -30,123 +29,6 @@ applyTo: '**'
 - **Element Counts**: Use `toHaveCount` to assert the number of elements found by a locator.
 - **Text Content**: Use `toHaveText` for exact text matches and `toContainText` for partial matches.
 - **Navigation**: Use `toHaveURL` to verify the page URL after an action.
- **Switch States**: Use `expectSwitchState(locator, boolean)` to verify toggle states. This is more reliable than `toBeChecked()` directly.
-
-### Switch/Toggle Interaction Patterns
-
-Switch components use a hidden `<input>` with styled siblings, requiring special handling:
-
-```typescript
-import { clickSwitch, expectSwitchState, toggleSwitch } from './utils/ui-helpers';
-
-// ✅ RECOMMENDED: Click switch with helper
-const aclSwitch = page.getByRole('switch', { name: /acl/i });
-await clickSwitch(aclSwitch);
-
-// ✅ RECOMMENDED: Assert switch state
-await expectSwitchState(aclSwitch, true); // Checked
-
-// ✅ RECOMMENDED: Toggle and verify state change
-const newState = await toggleSwitch(aclSwitch);
-console.log(`Switch is now ${newState ? 'enabled' : 'disabled'}`);
-
-// ❌ AVOID: Direct click on hidden input
-await aclSwitch.click(); // May fail in WebKit/Firefox
-
-// ❌ AVOID: Force clicking (anti-pattern)
-await aclSwitch.click({ force: true }); // Bypasses real user behavior
-
-// ❌ AVOID: Hard-coded waits
-await page.waitForTimeout(500); // Non-deterministic, slows tests
-```
-
-**When to Use**:
- Settings pages with enable/disable toggles
- Security dashboard module switches (CrowdSec, ACL, WAF, Rate Limiting)
- Access lists and configuration toggles
- Any UI component using the `Switch` primitive from shadcn/ui
-
-**References**:
- [Helper Implementation](../../tests/utils/ui-helpers.ts)
- [QA Report](../../docs/reports/qa_report.md)
-
-### Testing Scope: E2E vs Integration
-
-**CRITICAL:** Playwright E2E tests verify **UI/UX functionality** on the Charon management interface (port 8080). They should NOT test middleware enforcement behavior.
-
-#### What E2E Tests SHOULD Cover
-
-✅ **User Interface Interactions:**
- Form submissions and validation
- Navigation and routing
- Visual state changes (toggles, badges, status indicators)
- Authentication flows (login, logout, session management)
- CRUD operations via the management API
- Responsive design (mobile vs desktop layouts)
- Accessibility (ARIA labels, keyboard navigation)
-
-✅ **Example E2E Assertions:**
-```typescript
-// GOOD: Testing UI state
-await expect(aclToggle).toBeChecked();
-await expect(statusBadge).toHaveText('Active');
-await expect(page).toHaveURL('/proxy-hosts');
-
-// GOOD: Testing API responses in management interface
-const response = await request.post('/api/v1/proxy-hosts', { data: hostConfig });
-expect(response.ok()).toBeTruthy();
-```
-
-#### What E2E Tests should NOT Cover
-
-❌ **Middleware Enforcement Behavior:**
- Rate limiting blocking requests (429 responses)
- ACL denying access based on IP rules (403 responses)
- WAF blocking malicious payloads (SQL injection, XSS)
- CrowdSec IP bans
-
-❌ **Example Wrong E2E Assertions:**
-```typescript
-// BAD: Testing middleware behavior (rate limiting)
-for (let i = 0; i < 6; i++) {
-  await request.post('/api/v1/emergency/reset');
-}
-expect(response.status()).toBe(429); // ❌ This tests Caddy middleware
-
-// BAD: Testing WAF blocking
-await request.post('/api/v1/data', { data: "'; DROP TABLE users--" });
-expect(response.status()).toBe(403); // ❌ This tests Coraza WAF
-```
-
-#### Integration Tests for Middleware
-
-Middleware enforcement is verified by **integration tests** in `backend/integration/`:
-
- `cerberus_integration_test.go` - Overall security suite behavior
- `coraza_integration_test.go` - WAF blocking (SQL injection, XSS)
- `crowdsec_integration_test.go` - IP reputation and bans
- `rate_limit_integration_test.go` - Request throttling
-
-These tests run in Docker Compose with full Caddy+Cerberus stack and are executed in separate CI workflows.
-
-#### When to Skip Tests
-
-Use `test.skip()` for tests that require middleware enforcement:
-
-```typescript
-test('should rate limit after 5 attempts', async ({ request }) => {
-  test.skip(
-    true,
-    'Rate limiting enforced via Cerberus middleware (port 80). Verified in integration tests (backend/integration/).'
-  );
-  // Test body...
-});
-```
-
-**Skip Reason Template:**
-```
-"[Behavior] enforced via Cerberus middleware (port 80). Verified in integration tests (backend/integration/)."
-```


 ## Example Test Structure
@@ -194,11 +76,6 @@ test.describe('Movie Search Feature', () => {
 4. **Validate**: Ensure tests pass consistently and cover the intended functionality
 5. **Report**: Provide feedback on test results and any issues discovered

-### Execution Constraints
-
- **No Truncation**: Never pipe Playwright test output through `head`, `tail`, or other truncating commands. Playwright runs interactively and requires user input to quit when piped, causing the command to hang indefinitely.
- **Full Output**: Always capture the complete test output to analyze failures accurately.
-
 ## Quality Checklist

 Before finalizing tests, ensure:
@@ -46,8 +46,8 @@ env:
  REGISTRY: ghcr.io
  IMAGE_NAME: ${{ github.repository_owner }}/charon
  PLAYWRIGHT_COVERAGE: ${{ vars.PLAYWRIGHT_COVERAGE || '0' }}
-  DEBUG: 'charon:*,charon-test:*'
-  PLAYWRIGHT_DEBUG: '1'
+  # Standard Playwright runner debugging - shows test execution flow
+  DEBUG: 'pw:test'
  CI_LOG_LEVEL: 'verbose'

 concurrency:
@@ -215,36 +215,19 @@ jobs:
      - name: Run Chromium tests
        timeout-minutes: 4
        run: |
-          echo "═══════════════════════════════════════════="
+          echo "════════════════════════════════════════════"
          echo "Chromium E2E Tests"
          echo "Start Time: $(date -u +'%Y-%m-%dT%H:%M:%SZ')"
          echo "════════════════════════════════════════════"

          echo "Node: $(node -v)"
          echo "NPM:  $(npm -v)"
-          echo "Kernel: $(uname -a)"
-          echo "Memory (free -m):"
-          free -m || true

          SHARD_START=$(date +%s)
          echo "SHARD_START=$SHARD_START" >> $GITHUB_ENV

-          # Enable verbose Playwright output to diagnose browser launch hang
-          # pw:browser* = browser connection and launch
-          # pw:protocol = CDP protocol messages
-          # pw:channel = IPC between Playwright and browser
-          # pw:api = Playwright runner API calls (helps when hang occurs before browser launch)
-          export DEBUG=pw:api,pw:browser*,pw:protocol,pw:channel
-
-          echo "════════════════════════════════════════════"
-          echo "Preflight: listing tests (discovery only)"
-          echo "Time: $(date -u +'%Y-%m-%dT%H:%M:%SZ')"
-          npx playwright test --project=chromium --list
-          echo "════════════════════════════════════════════"
-          echo "Running tests"
-          echo "Time: $(date -u +'%Y-%m-%dT%H:%M:%SZ')"
-
          npx playwright test --project=chromium
+
          echo "════════════════════════════════════════════"
        env:
          PLAYWRIGHT_BASE_URL: http://127.0.0.1:8080
@@ -410,12 +393,6 @@ jobs:
          SHARD_START=$(date +%s)
          echo "SHARD_START=$SHARD_START" >> $GITHUB_ENV

-          # Enable verbose Playwright output to diagnose browser launch hang
-          # pw:browser* = browser connection and launch
-          # pw:protocol = CDP protocol messages
-          # pw:channel = IPC between Playwright and browser
-          export DEBUG=pw:browser*,pw:protocol,pw:channel
-
          npx playwright test --project=firefox

          SHARD_END=$(date +%s)
@@ -424,7 +401,6 @@ jobs:
          echo "════════════════════════════════════════════"
          echo "Firefox Tests Complete | Duration: ${SHARD_DURATION}s"
          echo "════════════════════════════════════════════"
-          echo "════════════════════════════════════════════"
        env:
          PLAYWRIGHT_BASE_URL: http://127.0.0.1:8080
          CI: true
@@ -589,12 +565,6 @@ jobs:
          SHARD_START=$(date +%s)
          echo "SHARD_START=$SHARD_START" >> $GITHUB_ENV

-          # Enable verbose Playwright output to diagnose browser launch hang
-          # pw:browser* = browser connection and launch
-          # pw:protocol = CDP protocol messages
-          # pw:channel = IPC between Playwright and browser
-          export DEBUG=pw:browser*,pw:protocol,pw:channel
-
          npx playwright test --project=webkit

          SHARD_END=$(date +%s)