Implement offline cache coordination in BLoC
epic-annual-impact-summary-orchestration-task-006 — Wire the summary offline cache into WrappedSummaryBloc so that every successful LoadSummary or ChangePeriod call persists the result to the cache. On network failure the BLoC reads the latest cached summary and emits WrappedSummaryOffline with the stale data and a staleness timestamp so the screen can display an offline banner.
Acceptance Criteria
Technical Requirements
Execution Context
Tier 2 - 518 tasks
Can start after Tier 1 completes
Implementation Notes
Define a `SummaryCache` abstract class with `read(userId, period)` and `write(userId, period, CachedSummary)` methods. Implement with Hive (encrypted box) or flutter_secure_storage for JSON payloads. In the `_fetchAndEmit` helper: after a successful network response, call `unawaited(_cache.write(...))` so the write does not delay UI. On catch, call `final cached = await _cache.read(userId, period)` and branch on null vs expired.
Define `_isExpired(CachedSummary c) => DateTime.now().difference(c.cachedAt).inDays > 30`. Clear the entire user cache namespace on logout — hook into an auth state listener or expose a `clearUserCache(userId)` method. For testing, pass a `FakeSummaryCache` that records calls.
Testing Requirements
Unit tests with bloc_test and mock cache service. Test cases: (1) emits Loaded and verifies cache.write was called with correct key and payload, (2) emits Offline with correct cachedAt when network fails and cache hit, (3) emits Error(NO_DATA) when network fails and cache miss, (4) emits Error when cache entry is older than 30 days, (5) cache write failure does not change Loaded emission, (6) cache is keyed separately for different periods. Integration test: write to real Hive box, simulate offline, verify WrappedSummaryOffline contains identical slides to what was written.
If the device transitions between online and offline states while the user is mid-session in the wrapped screen, the BLoC may emit conflicting state transitions (loaded → error → offline) that cause visual flickering or an inconsistent UI state such as showing the offline banner over an already-loaded summary.
Mitigation & Contingency
Mitigation: Implement a connectivity stream listener in the BLoC that only triggers a state re-evaluation when transitioning from online to offline, not on every connectivity event. Once a summary is in the Loaded state, the BLoC should not transition to error/offline unless the user explicitly requests a refresh. Store the last-loaded data in BLoC state so it survives connectivity changes.
Contingency: If state flickering is observed in testing, add a minimum 3-second debounce on connectivity state changes before the BLoC reacts, and display a non-blocking top banner rather than replacing the entire screen state.
The push notification deep-link to the wrapped-summary-screen must work correctly whether the app is in the foreground, background, or terminated state. Handling all three app launch states on both iOS and Android is a common source of edge-case bugs, particularly when authentication state must be restored before the deep link can be resolved.
Mitigation & Contingency
Mitigation: Implement deep-link handling through the existing notification-deep-link-handler component which already manages app-state-aware routing. Define the wrapped-summary route in the navigation config early in the epic so the router is ready before notification dispatch is wired. Test all three app states (foreground, background, terminated) explicitly in the QA checklist.
Contingency: If terminated-state deep-linking fails on specific platforms, fall back to launching the app to the home screen with an in-app notification banner prompting the user to open their summary, rather than direct deep-link navigation.
The wrapped-summary-screen manages a large number of AnimationController instances (one or more per slide) via the wrapped-animation-controller. If disposal is not triggered correctly when the user exits mid-flow (e.g., via system back gesture or deep-link away), memory leaks will accumulate across session navigation.
Mitigation & Contingency
Mitigation: Implement screen disposal via Flutter's dispose() lifecycle method calling a single wrapped-animation-controller.disposeAll() method that iterates the named controller registry. Write a test that navigates to the screen, starts animations, then navigates away and verifies no active AnimationController listeners remain using Flutter's test binding.
Contingency: If disposal bugs are detected in production via memory profiling, patch by converting all AnimationControllers to use AutomaticKeepAliveClientMixin false and wrap each slide in a widget that disposes its own controller when removed from the widget tree.