Integration test: expiry notification UI end-to-end flow
epic-certificate-expiry-notifications-user-interface-task-012 — Write Flutter integration tests that exercise the full in-app expiry notification UI flow: Notifications tab loads and renders ExpiryNotificationBanner for a seeded expiry record; tapping the banner navigates to NotificationDetailView; peer mentor sees correct actions and coordinator sees additional actions; coordinator acknowledge-lapse action removes the banner from the Notifications tab; all WCAG 2.2 AA contrast and semantics checks pass across the full flow.
Acceptance Criteria
Technical Requirements
Execution Context
Tier 6 - 158 tasks
Can start after Tier 5 completes
Implementation Notes
Create the test at integration_test/certificate_expiry_notification_flow_test.dart. Abstract all seed/teardown into a CertificationTestFixture class to keep test bodies readable. Use a ProviderOverride or BlocOverride at the app level to inject a test-mode Supabase client pointing to the local test instance. The acknowledge-lapse action likely triggers an optimistic UI update followed by a Supabase RPC call — use tester.pumpAndSettle with a generous timeout (e.g.
10 seconds) to allow the async round-trip. For the banner-removal assertion after acknowledge-lapse, confirm both that the BLoC state has transitioned (no ExpiryNotificationBanner in tree) and that a re-query of the Notifications tab does not re-surface the banner. When running on CI, ensure the local Supabase instance is started as a service step before flutter test integration_test.
Testing Requirements
Use the Flutter integration_test package (integration_test/app_test.dart). Seed a certification record via a test helper that calls Supabase directly or uses a fake repository injected at app startup. Use IntegrationTestWidgetsFlutterBinding.ensureInitialized(). Navigate to the Notifications tab by tapping the bottom nav item.
Use tester.tap and tester.pumpAndSettle for navigation and async state resolution. For contrast checking, use the flutter_test AccessibilityGuideline with meetsGuideline(textContrastGuideline) and meetsGuideline(labeledTapTargetGuideline). Run once with a seeded peer mentor session and once with a coordinator session. Use tester.ensureSemantics() and verify the SemanticsController for each screen transition.
Include tearDown that removes the seeded certification record to prevent test pollution.
The persistent banner must remain visible across app sessions and only disappear when a specific backend condition is met (renewal or coordinator acknowledgement). If the BLoC state is not properly sourced from the notification record repository on every app launch, the banner may disappear prematurely or fail to reappear after a session restart.
Mitigation & Contingency
Mitigation: Drive the banner's visibility exclusively from a Supabase real-time subscription on the notification records table filtered by mentor_id and acknowledged_at IS NULL. Never persist banner visibility state locally. Write an integration test that restarts the BLoC and verifies the banner reappears from the database source.
Contingency: If real-time subscriptions introduce latency or connection reliability issues in offline-first scenarios, add a local cache flag that is only cleared when the repository confirms the acknowledgement write succeeded, with a cache TTL of 24 hours as a fallback.
The notification detail view must conditionally render coordinator-specific actions based on the authenticated user's role. Incorrect role resolution could expose the 'Acknowledge Lapse' action to peer mentors or hide it from coordinators, breaking the workflow and potentially allowing unauthorised state changes.
Mitigation & Contingency
Mitigation: Source the role check from the existing role_state_manager BLoC that is already authenticated against Supabase role claims. Do not rely on a local flag. The coordinator acknowledgement service backend also validates role server-side, providing defence in depth. Add widget tests that render the detail view with mentor and coordinator role fixtures and assert the presence or absence of coordinator actions.
Contingency: If a role resolution bug is found in production, immediately disable the acknowledge action via a feature flag and patch the role check in a hotfix release. The server-side validation in the coordinator acknowledgement service ensures no actual state change can occur even if the button is incorrectly rendered.