StreamVault is a TV-first free IPTV player for Android TV built with Kotlin, Jetpack Compose, Room, Hilt, and Media3.
Kotlin iptv-player
fix(settings): restore title case for sync option labels test(playback): add catch-up eligibility boundary test test(epg): enhance guide preview pane to render program time range with en dash fix(epg): update EPG override descriptor separator to use a bullet test(settings): add test for sync option labels to ensure title case docs(audit): document Phase 5 residual findings implementation plan
David Nashasha977b154d033
diff --git a/app/src/test/java/com/streamvault/app/settings/AppSettingsUpdateAdapterTest.kt b/app/src/test/java/com/streamvault/app/settings/AppSettingsUpdateAdapterTest.ktindex 538e0be3..ddeaaf96 100644--- a/app/src/test/java/com/streamvault/app/settings/AppSettingsUpdateAdapterTest.kt+++ b/app/src/test/java/com/streamvault/app/settings/AppSettingsUpdateAdapterTest.kt@@ -1,7 +1,9 @@package com.streamvault.app.settingsimport com.google.common.truth.Truth.assertThat+import com.streamvault.app.BuildConfigimport com.streamvault.app.update.AppUpdateDownloadState+import com.streamvault.app.update.AppUpdateChannelimport com.streamvault.app.update.AppUpdateInstallerimport com.streamvault.app.update.GitHubReleaseCheckerimport kotlinx.coroutines.flow.MutableStateFlow@@ -17,11 +19,20 @@ class AppSettingsUpdateAdapterTest {val updateInstaller = mock<AppUpdateInstaller>()whenever(updateInstaller.downloadState).thenReturn(MutableStateFlow(AppUpdateDownloadState()))val adapter = AppSettingsUpdateAdapter(releaseChecker, updateInstaller)+ val remoteVersionName = if (AppUpdateChannel.fromBuildConfig(+ BuildConfig.APP_UPDATE_CHANNEL,+ BuildConfig.VERSION_NAME+ ) == AppUpdateChannel.Beta+ ) {+ "999.0.0-beta"+ } else {+ "999.0.0"+ }assertThat(adapter.isRemoteVersionNewer(remoteVersionCode = Int.MAX_VALUE,- remoteVersionName = "999.0.0",+ remoteVersionName = remoteVersionName,remotePublishedAt = null)).isTrue()diff --git a/docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md b/docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.mddeleted file mode 100644index 5b0996f8..00000000--- a/docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md+++ /dev/null@@ -1,333 +0,0 @@-# Compose Reduction Phase 5: Settings and Live Regression Audit--Date started: 2026-08-30-Date completed: 2026-08-31-Status: Audit complete; three isolated regression fixes verified and integrated, with documented residual risks--## Scope and comparison points--This audit compares the two most recent Phase 5 slices with their recorded-pre-extraction implementations:--| Slice | Historical baseline | First implementation commit | Target |-| --- | --- | --- | --- |-| Settings | `e12f816f` | `1c0e6ae6` | frozen working-tree snapshot based on `bb52375d` |-| Live/EPG | `ea349d83` | `712751b7` | frozen working-tree snapshot based on `bb52375d` |-| Combined Settings + Live | `e12f816f` | both slices | frozen working-tree snapshot based on `bb52375d` |--The frozen target includes non-document tracked patch hash-`29be114106b2c471531e25caa6fd1a54027d2b3b` and untracked manifest SHA-256-`46ebf25851b43bd6ecdb9daff3d1aa68fd2644c0313104c0fdcb36b4ad699cc9`.-The untracked files at freeze time were:--- `feature/provider/src/test/java/com/streamvault/feature/provider/navigation/ProviderGraphNavigationTest.kt`-- `feature/settings/src/test/java/com/streamvault/feature/settings/presentation/SettingsProviderSummaryFormattingTest.kt`--The external audit root is-`C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993`.-The captured patch SHA-256 is-`2f3d882d3e6aa9e234583d295c4f043132ee5d635aef8bcbf068d5af0582383`.-Reconstruction verification produced the same tracked Git-diff hash-`29be114106b2c471531e25caa6fd1a54027d2b3b` and untracked manifest SHA-256-`46ebf25851b43bd6ecdb9daff3d1aa68fd2644c0313104c0fdcb36b4ad699cc9`-in the target worktree. Both historical worktrees were clean after creation.--| Worktree | Path | State |-| --- | --- | --- |-| Settings baseline | `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\settings-baseline` | detached `e12f816f`, clean |-| Live baseline | `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\live-baseline` | detached `ea349d83`, clean |-| Integrated audit target | `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\target` | branch `audit/phase5-settings-live-integrated`; frozen `bb52375d` snapshot plus verified fixes through `3f4152d9` |--The accepted audit design and executable plan are:--- `docs/superpowers/specs/2026-08-30-phase5-settings-live-regression-audit-design.md`-- `docs/superpowers/plans/2026-08-30-phase5-settings-live-regression-audit.md`--## Isolation rule--The primary `feature/improveCompose` checkout is shared with another agent.-The audit will not change its branch, stage or commit files, edit code, build,-or install from it. Only Markdown evidence is written there. Historical-builds, target reconstruction, tests, screenshots, fixes, and fix commits are-performed in uniquely named external worktrees.--## Finding ledger--The two examples supplied in the request are known, already-fixed control-findings for this review. They are recorded to ensure the audit recognizes the-original symptom and the applied correction; they are not pending validation-tasks:--| Supplied example | Audit finding | Fixed behavior recorded |-| --- | --- | --- |-| Live TV category title was black instead of white | `LIVE-UI-001` | The category sidebar title explicitly retains the light/white `OnSurface` color. |-| Settings → Add Providers stayed in the Add Provider window after completion | `SET-NAV-001` | Completion returns through the existing Settings route via Back-first navigation with startup fallback. |--| ID | Slice | Classification | Severity | Summary | Evidence and status |-| --- | --- | --- | --- | --- | --- |-| `LIVE-UI-001` | Live | Confirmed regression, fixed in frozen patch | Major | The Live category sidebar title inherited host content color after extraction instead of retaining the original light/white `OnSurface` title. | Introduced by `9214b14b`. Baseline source explicitly used `OnSurface`; extracted header omitted it. The frozen patch restores it. The focused connected pixel test passed even under deliberately black `LocalContentColor`. |-| `SET-NAV-001` | Settings/provider entry | Pre-existing defect, fixed in frozen patch | Major | Completing Add Provider from Settings remained on the provider setup route instead of returning to the prior screen. | The behavior exists before Settings baseline `e12f816f`, so it is not an extraction regression. The supplied/frozen patch uses Back first with startup fallback, and the focused ProviderGraph navigation test passes. This is recorded as fixed; no additional production-shell replay is required for this known fix. |-| `SET-UPD-001` | Settings | Confirmed regression, fixed and verified | Major | A persisted newer app release was always mapped as `isUpdateAvailable = false` while applying a preference snapshot. | Introduced by `20a3f3f5`. Red test failed because the mapper exposed no comparator; fix `c2286afc` injects `appUpdatePort::isRemoteVersionNewer`. Focused and full Settings suites pass. Both historical and target checks used the same update release/version (`1.0.17.1`), so version drift does not explain the behavioral difference. |-| `SET-NAV-002` | Settings | Confirmed regression, fixed and verified | Major | The extracted parental-control screen received an empty top-navigation destination list. | Introduced by `0441d1e6` and retained by `579a00cc`. Fix `11f0ee6f` forwards the app-owned destinations. A composed route test and production-shell before/after screenshots verify restoration. |-| `SET-UPD-002` | Settings | Confirmed regression, already fixed | Critical | The extracted update adapter recursively called its own `isRemoteVersionNewer` override and could overflow the stack. | Introduced by `e9a44e24`; fixed in committed target by `d36005a7` using an aliased policy function. The target unit suite passed. |-| `LIVE-STATE-001` | Live | Confirmed regression, fixed and verified | Major | After a preview engine emitted `ERROR`, a later `READY`/`PLAYING` state left the engine error latched and could keep the recovered render surface hidden. | A red `IDLE -> ERROR -> READY` ViewModel test reproduced the latch. Fix `3f4152d9` tracks engine-originated errors and clears only those after recovery; the existing preparation-error persistence test still passes. |-| `LIVE-TEXT-001` | Guide | Confirmed regression, fixed in frozen patch | Minor | Extraction further corrupted an already-broken Guide preview time separator, adding invalid/mojibake code points. | Introduced by `f56d47da`. The frozen patch replaces it with a real en dash, improving on both extracted and historical text. Focused rendered/text coverage is still missing. |-| `LIVE-TEXT-002` | Guide | Pre-existing defect, open | Minor | The EPG override candidate descriptor separator remains mojibake. | The exact corruption is present in Live baseline `ea349d83`; it is not an extraction regression. |-| `SET-COPY-001` | Settings | Inconclusive copy drift | Minor | `Sync Now` and `Rebuild Index` became sentence-case `Sync now` and `Rebuild index`. | The only unexplained effective resource changes across the audited 26 resource folders. No behavior changes; product/design intent is not recorded. |-| `LIVE-POLICY-001` | Live | Inconclusive boundary drift | Minor | Archive eligibility is checked twice against wall clock instead of once. | Normally defensive/equivalent, but a request could expire between checks. Needs a shared/injected-clock boundary decision rather than a speculative fix. |-| `TEST-GOLDEN-001` | Live test infrastructure | Deterministic golden mismatch; no material UI regression observed | Minor | Six strict Live image goldens fail exact equality on the audit emulator. | Two target captures are byte-identical. All images retain 1920x1080 geometry; maximum channel delta is 1-7, mean absolute RGBA error is at most 0.0232, and no layout displacement is visible. Do not regenerate baselines until provenance/environment is resolved. |--The three additional production fixes were first developed and verified in the-isolated audit worktrees, then cherry-picked into the primary-`feature/improveCompose` branch as `713e4f6b`, `59e40f98`, and `da94af62`.-The supplied category-color and Add Provider examples were already present in-the primary snapshot and remain recorded as known fixed control findings.--## Improvements--- The frozen patch repairs pre-existing corrupted separators, arrows, minus,- cloud, and dismiss glyphs across Settings, and repairs the Japanese backup- count. These are improvements rather than extraction-regression fixes.-- The Guide preview separator fix produces a correct en dash, improving on the- already-corrupt historical baseline as well as fixing the worse extracted- value.-- French and Italian Live resource mojibake present in the historical baseline- is corrected in the frozen patch.-- The update-comparison recursion is fixed in committed target code and guarded- by a focused unit test.-- Live locale ownership is statically complete for the 25 audited locale- folders, with Vietnamese and Chinese Live ownership added during extraction.-- The extracted feature boundaries replace app coupling with explicit ports and- add targeted route, adapter, focus, remote-shortcut, preview, and stable-row- coverage. These are architectural/testability improvements, not proof of- runtime parity.--## Evidence matrix--| Area | Static review | Automated | Historical visual pair | Behavioral replay | Result |-| --- | --- | --- | --- | --- | --- |-| Settings shell, sections, dialogs, focus, Back | Static layout/order equivalent except findings | Integrated unit, boundary, assembly, and 13/13 connected tests passed | Settings entry and parental route paired | Parental navigation restored; Add Provider completion covered by focused route test | Pass |-| Settings provider/backup/parental/EPG/platform seams | Static adapter/action equivalence except `SET-*` findings | Cached-update and parental-route regression tests pass | Cached-update and parental top-nav pairs captured | Cached update action and parental destinations present | Pass |-| Live categories, channels, search, filters, favorites/reorder | Static policies/order/callbacks equivalent except title finding | 19/19 non-golden connected tests passed; strict title pixel test passed | Live baseline/target entry pair and six deterministic component candidates analyzed | Live list/category rendering exercised | Pass; exact goldens remain red |-| Guide grid, preview, options, search, recording/reminder/archive | Static policy equivalent except documented text/clock findings | Unit and boundary suites passed | Separator improvement inspected in source/resources | Credential/storage-dependent actions not replayed | Pass within available fixtures |-| Settings-to-Live persisted controls | Static cross-slice wiring reviewed | Integrated unit/boundary suites passed | Same provider fixture used | Live channels loaded after Settings checks | Pass |-| Preview/player/MultiView handoff | Static adapters/callback ordering equivalent except fixed `LIVE-STATE-001` | Recovery red/green test plus full Live unit suite passed | Not applicable | Two channels reached and retained playback with recovery | Pass |-| Two-channel long-duration Live playback | Not applicable | 61 screenshots per channel plus media/log checks | Not applicable | Both channels progressed and ended `PLAYING`, `error=null` | Pass with transient source-window recoveries documented |--## Automated verification log--### Target boundary and unit gate--The first attempt stopped during Gradle configuration because detached Git-worktrees do not copy the ignored `local.properties` file and neither-`ANDROID_HOME` nor `ANDROID_SDK_ROOT` was set. The primary checkout's existing-`local.properties` was copied into each external worktree; this supplies the-valid Android SDK path and the existing public development fixture without-changing the primary checkout. The identical Gradle command was then rerun:--```text-gradlew.bat :feature:settings:verifyFeatureSettingsBoundary \- :feature:settings:testDebugUnitTest \- :feature:live:verifyFeatureLiveBoundary \- :feature:live:testDebugUnitTest \- --console=plain --warning-mode=none-```--Result: `BUILD SUCCESSFUL in 2m 10s`; 87 actionable tasks, 37 executed and 50-from cache. Both fail-closed feature-boundary checks passed and both unit-test-suites passed. Gradle discarded the configuration-cache entry because the two-custom boundary tasks access project/script objects at execution time; this is-recorded as build-tooling debt, not a Settings or Live functional failure.-The complete successful log is-`C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\logs\target-boundary-unit-rerun.log`.--### Target check, Android-test compilation, and debug assembly--```text-gradlew.bat :feature:settings:check :feature:live:check \- :feature:settings:compileDebugAndroidTestKotlin \- :feature:live:compileDebugAndroidTestKotlin \- :app:assembleDebug --console=plain --warning-mode=none-```--Result: `BUILD SUCCESSFUL in 5m 11s`; 426 actionable tasks, 224 executed,-118 from cache, and 84 up to date. Both feature lint/check gates, both-Android-test compilations, and the app debug APK completed successfully. The-same custom boundary-task configuration-cache warnings were emitted and the-cache entry was discarded. Complete log:-`C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\logs\target-check-androidtest-assemble.log`.--### Historical APK buildability--- Settings rollback `e12f816f`: `:app:assembleDebug` passed in 3m 31s;- 154 actionable tasks, 100 executed and 54 from cache.-- Live rollback `ea349d83`: `:app:assembleDebug` passed in 6m 57s;- 176 actionable tasks, 92 executed and 84 from cache.--The historical baseline APKs are therefore available for paired emulator-replay. Complete logs are `settings-baseline-assemble-debug.log` and-`live-baseline-assemble-debug.log` under the external audit `logs` directory.--### Target connected Settings suite--The audit emulator was recorded before execution as AOSP TV x86,-1920x1080, density 320, locale `en-US`, font scale 1.0, with all animation-scales at 1.0. No app process or foreground activity was present at capture.--```text-gradlew.bat :feature:settings:connectedDebugAndroidTest --console=plain-```--Result: all 12 tests passed with no failures or skips; `BUILD SUCCESSFUL in-1m 13s`. This covers the feature's connected Settings behavior suite, but it-does not compose the app-owned parental route or prove the full Settings to-Provider Setup and back journey. Complete logs are `emulator-pre-audit-config.txt`-and `target-settings-connected.log` under the external audit `logs` directory.--### Target connected Live suite and golden diagnosis--```text-gradlew.bat :feature:live:connectedDebugAndroidTest --console=plain-```--The suite executed 25 tests. Nineteen behavioral tests passed, including the-new `categorySidebarHeader_titleUsesLightOnSurfaceText` regression test. Six-strict exact-pixel cases in `LivePresentationGoldenTest` failed. Recording the-same six fixtures twice on the unchanged emulator produced byte-identical-candidate PNGs, establishing deterministic output. Comparing candidates to-checked-in expected images produced:--| Golden | Differing pixels | Image percent | Maximum channel delta | Mean absolute RGBA error |-| --- | ---: | ---: | ---: | ---: |-| `live_category_row` | 453 | 0.021846% | 1 | 0.000078366 |-| `live_channel_row_surface` | 6,151 | 0.296634% | 6 | 0.002638889 |-| `live_channel_surfaces` | 95,789 | 4.619454% | 7 | 0.019952980 |-| `live_reorder_top_bar` | 143,727 | 6.931279% | 2 | 0.023128376 |-| `live_selection_chips` | 1,745 | 0.084153% | 1 | 0.000274764 |-| `live_source_switcher` | 1,199 | 0.057822% | 1 | 0.000169753 |--Every pair is 1920x1080. The amplified diffs follow the same shapes and text-with sub-channel color rounding rather than displaced geometry; no material-visual change is observable. The strict suite therefore remains red, but the-evidence does not justify changing production UI or accepting regenerated-baselines. Candidate PNGs, raw/amplified diffs, and-`golden-diff-metrics.json` are under the external audit-`visual/live-target` directory. Complete execution logs are-`target-live-connected.log`, `target-live-golden-record.log`, and-`target-live-golden-record-run2.log`.--### Red/green regression fixes and integrated gate--Each newly changed behavior was first isolated in its own external worktree,-demonstrated red, fixed, run focused green, then run through its full feature-suite before integration:--| Finding | Red evidence | Fix | Green evidence |-| --- | --- | --- | --- |-| `SET-UPD-001` | `red-cached-update.log`: test could not supply a remote-version comparator because production omitted the dependency | `c2286afc` | `green-cached-update-focused.log`; `green-cached-update-settings-suite.log` |-| `SET-NAV-002` | `red-parental-nav.log`: composed route API did not provide navigation destinations | `11f0ee6f` | `green-parental-nav-focused.log`; `green-parental-nav-settings-suite.log` |-| `LIVE-STATE-001` | `red-live-preview-recovery.log`: expected cleared error after `READY`, observed `test-message` | `3f4152d9` | `green-live-preview-recovery-focused.log`; `green-live-preview-recovery-suite.log` |--The integrated target then passed both feature unit suites, both fail-closed-boundary tasks, and `:app:assembleDebug`; see-`logs\integrated-boundary-unit-assemble.log`. Settings connected tests passed-13/13 (`logs\integrated-settings-connected.log`). Live connected tests again-passed all 19 behavioral tests and failed only the same six deterministic-exact-pixel goldens (`logs\integrated-live-connected.log`). `graphify update .`-completed in the external target after these code changes: 1,225 files,-13,302 nodes, 23,516 edges, and 423 communities.--A final 2026-08-31 rerun produced the same outcome: unit/boundary/assembly-`BUILD SUCCESSFUL`, Settings connected 13/13, and Live connected 25 executed-with exactly the same six golden failures and no behavioral-test failures.--## Production-shell visual and behavior comparison--The emulator comparisons used the historical APKs and the integrated audit APK-at 1920x1080, density 320, `en-US`, font scale 1.0, and animation scales 1.0.--- Historical Settings parental controls showed the complete top destination- bar. The pre-fix target showed only the StreamVault label. The integrated- target again shows Home, Live TV, Movies, Series, Downloads, Guide, Search,- Plugins, and Settings. Evidence:- `visual\manual-settings-baseline\parental-with-topnav.png`,- `visual\manual-target\parental-empty-topnav.png`, and- `visual\manual-integrated-target\parental-topnav-restored.png`.-- With the same update release/version available in both comparisons, the- pre-fix extracted Settings screen displayed the cached newer release but- reported `Up to date` and omitted the download action. The integrated target- displays `Download update` and the latest-release action. Evidence:- `visual\manual-live-baseline\settings-about-cached-update.png` and- `visual\manual-integrated-target\settings-about-cached-update.png`.-- Historical and integrated Live entry screens retain the same structure,- placement, colors, and white Categories title. Evidence:- `visual\manual-live-baseline\live-entry.png` and- `visual\manual-integrated-target\live-entry.png`.-- The historical Settings provider summary visibly contained mojibake between- fields; the frozen/integrated target renders proper bullets. This is an- improvement, not a regression. Evidence:- `visual\manual-settings-baseline\settings-entry.png` and- `visual\manual-target\settings-entry.png`.-- The Add Provider completion issue is a supplied, already-fixed control- finding (`SET-NAV-001`), covered by the focused ProviderGraph navigation test;- it is not an outstanding production-shell validation task.--## Long-duration Live playback--The integrated target was exercised on two public-fixture HLS channels. Each-capture loop requested a screenshot, then slept two seconds. PNG encoding made-the observed screenshot-start interval longer than two seconds, so both runs-exceeded the required historical stuck window.--| Channel | Screenshots | Observed duration/cadence | Unique hashes | Final media session | Log result |-| --- | ---: | --- | ---: | --- | --- |-| `00s Replay` | 61 | 190.3 s; mean start interval 3.171 s | 59 | `PLAYING`, `error=null` | HLS prepare and first-frame success; two non-consecutive duplicate pairs; no fatal, stuck, MPEG-TS, malformed-source, or fallback marker |-| `3ABN Dare To Dream Network` | 61 | 317.1 s; mean start interval 5.286 s | 61 | `PLAYING`, `error=null` | HLS prepare and first-frame success; all hashes unique; no fatal, stuck, MPEG-TS, malformed-source, or fallback marker |--Both streams encountered transient `BehindLiveWindow` source-window errors.-The player retried as HLS, produced another first frame, and ended healthy. This-is successful recovery evidence, not a claim that the public source emitted no-transient errors. Screenshots, timing, hashes, final media sessions, and logs-are under `playback\channel1-00s-replay` and-`playback\channel2-3abn-dare-to-dream` in the external audit root.--## Static equivalence summary--Rename-aware source/history comparison supports equivalence, except for the-ledger findings, in these areas:--- Settings route patterns/arguments, section order, rail dimensions/colors,- focus requester, persistence/action ordering, platform adapters, backup and- recording bridges, and effective localized resources.-- Live/Guide route patterns and sentinels, request mapping, category/channel- list order and identity, search/filters/hidden/favorites/reorder policies,- Back/D-pad/shortcut/long-press policy, preview/fullscreen/MultiView callback- ordering, and Guide grid/action policy.--Static equivalence is supplemented by the runtime evidence above. Platform/SAF-behavior, credentialed provider refresh, recording/reminder/archive side-effects, process restoration, and exact pixels across device configurations-remain outside the evidence obtained by this audit.--## Current limitations--- The primary checkout is intentionally live and may change after the frozen- target snapshot. Later drift will be reported rather than silently folded- into the comparison.-- Historical commits may expose tasks or fixtures that differ from the target;- unavailable historical gates will be recorded exactly.-- Credentialed providers, external accounts, recording storage, archive- windows, notification permission, and physical hardware remain evidence- limitations unless suitable fixtures are already available.--## Recommended disposition--1. The three audit fixes are now integrated in primary as `713e4f6b`,- `59e40f98`, and `da94af62` (source commits `c2286afc`, `11f0ee6f`, and- `3f4152d9`).-2. Keep the six Live goldens open as test-infrastructure work; do not regenerate- them from this emulator without first identifying the baseline renderer.-3. Decide product intent for `SET-COPY-001` and clock semantics for- `LIVE-POLICY-001`; neither warrants a speculative audit fix.-4. Fix pre-existing `LIVE-TEXT-002` separately so it is not misclassified as- Phase 5 extraction fallout.diff --git a/docs/superpowers/plans/2026-08-30-phase5-settings-live-regression-audit.md b/docs/superpowers/plans/2026-08-30-phase5-settings-live-regression-audit.mddeleted file mode 100644index f567d1a5..00000000--- a/docs/superpowers/plans/2026-08-30-phase5-settings-live-regression-audit.md+++ /dev/null@@ -1,283 +0,0 @@-# Phase 5 Settings and Live Regression Audit Implementation Plan--> **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.--**Goal:** Compare the Phase 5 Settings and Live extractions with their historical rollback commits, document every observable difference, and fix confirmed regressions in an isolated worktree.--**Architecture:** Three detached external worktrees provide the Settings baseline, Live baseline, and frozen target. Static slice reviews run independently; all builds and emulator work are serialized. The active checkout receives Markdown evidence only, while target code fixes remain committed in the isolated audit worktree.--**Tech Stack:** Git worktrees, Gradle/Android instrumentation, ADB API 36 TV emulator, Compose UI tests and goldens, PowerShell, Graphify.--**Spec:** `docs/superpowers/specs/2026-08-30-phase5-settings-live-regression-audit-design.md`--## Global Constraints--- Do not change branches, stage, commit, build, install, or edit code in the primary checkout.-- Only Markdown audit documentation may be written in the primary checkout.-- Settings baseline is `e12f816fd12ac688a6df6391b5740d33d8a3ada0`.-- Live baseline is `ea349d83dceb8baa45ef1a18b7a45fa295a264f5`.-- Frozen target is `bb52375d04e72e95a7dd70b0c79e2aab91c2121a` plus tracked patch hash `29be114106b2c471531e25caa6fd1a54027d2b3b` and untracked manifest SHA-256 `46ebf25851b43bd6ecdb9daff3d1aa68fd2644c0313104c0fdcb36b4ad699cc9`.-- Emulator installs, connected tests, fixture mutations, and screenshot captures run serially.-- Post-extraction goldens are supporting evidence, not the historical visual oracle.-- Fixes use a failing reproducer first and remain committed only in the isolated target worktree.-- Live playback validation covers at least two channels with the required 2-second, 45/61-frame protocol.-- Run `graphify update .` in the target worktree after code changes.-------### Task 1: Freeze and isolate the audit inputs--**Files:**-- Modify: `docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md`-- Create outside primary checkout: `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\target-working-tree.patch`-- Create outside primary checkout: `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\untracked-manifest.txt`--**Interfaces:**-- Consumes: the primary checkout read-only Git state.-- Produces: three detached worktree paths and a reproducible target fingerprint.--- [ ] **Step 1: Record the primary checkout state without modifying it**--Run `git status --short --branch`, `git rev-parse HEAD`, and the tracked/untracked fingerprint commands from the design. Copy the two approved untracked test files into the temporary audit root.--- [ ] **Step 2: Capture the non-document target patch outside the checkout**--Run `git diff --binary -- . ':(exclude)docs/**'` and write its output to `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\target-working-tree.patch`. Record its SHA-256 and verify the source checkout status is unchanged.--- [ ] **Step 3: Create three detached external worktrees**--Use `git worktree add --detach` to create `settings-baseline` at `e12f816f`, `live-baseline` at `ea349d83`, and `target` at `bb52375d` beneath `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993`. Apply the frozen patch and copied untracked tests only in `target`.--- [ ] **Step 4: Verify target reconstruction**--Compare `git diff --binary` and the untracked manifest in the target worktree with the frozen hashes. Record exact paths and hashes in the audit report.--### Task 2: Audit the Settings slice statically--**Files:**-- Modify: `docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md`-- Create outside primary checkout: `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\settings-static-review.md`--**Interfaces:**-- Consumes: Settings rollback and frozen target source.-- Produces: a path-by-path Settings risk inventory and candidate findings.--- [ ] **Step 1: Inventory changed Settings ownership and adapters**--Diff `e12f816f..target` for Settings source/resources/tests, app Settings adapters, navigation registration, provider entry/completion, backup preview, and shared UI dependencies.--- [ ] **Step 2: Review observable contracts**--Check route arguments, callback order, dialog dismissal, Back/focus behavior, list ordering, defaults, persistence, locale text, colors, spacing, and platform handoffs against the pre-extraction implementation.--- [ ] **Step 3: Trace candidate changes to commits**--Use targeted `git log -p`, `git blame`, and parent comparisons to identify the earliest commit that introduced each candidate difference.--- [ ] **Step 4: Write the Settings review artifact**--Record every reviewed area, candidate finding, evidence path, and areas proven equivalent. Add confirmed or inconclusive findings to the main audit report.--### Task 3: Audit the Live slice statically--**Files:**-- Modify: `docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md`-- Create outside primary checkout: `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\live-static-review.md`--**Interfaces:**-- Consumes: Live rollback and frozen target source.-- Produces: a path-by-path Live/EPG risk inventory and candidate findings.--- [ ] **Step 1: Inventory changed Live/EPG ownership and adapters**--Diff `ea349d83..target` for Home, Live, EPG, app route/request mapping, preview handoff, surface refresh, MultiView composition, resources, and tests.--- [ ] **Step 2: Review observable contracts**--Check state defaults, ordering/identity, focus and remote dispatch, preview lifecycle/callback order, colors/layout/text, dialogs, search/filter/reorder, Guide navigation, and player/MultiView handoff.--- [ ] **Step 3: Trace candidate changes to commits**--Identify the earliest introducing commit for every observable difference, including the category-title color inheritance regression.--- [ ] **Step 4: Write the Live review artifact**--Record every reviewed area, candidate finding, evidence path, and areas proven equivalent. Add confirmed or inconclusive findings to the main audit report.--### Task 4: Audit cross-slice contracts--**Files:**-- Modify: `docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md`-- Create outside primary checkout: `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\cross-slice-review.md`--**Interfaces:**-- Consumes: both static reviews and the Graphify architecture map.-- Produces: cross-slice journey and dependency findings.--- [ ] **Step 1: Trace Settings-to-Live persisted controls**--Review quick filters, source-switcher visibility, category modes/sorting, parental controls, EPG settings, and provider selection from Settings mutation through Live/EPG observation.--- [ ] **Step 2: Trace provider and playback composition seams**--Review Add Provider completion, backup preview, player request mapping, preview handoff, surface refresh, and MultiView planner callbacks across app/feature boundaries.--- [ ] **Step 3: Record cross-slice findings and test gaps**--Classify differences and identify the exact unit/connected/manual evidence needed for each unresolved contract.--### Task 5: Run automated structural and behavioral gates--**Files:**-- Modify: `docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md`-- Create outside primary checkout: `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\logs\*.log`--**Interfaces:**-- Consumes: clean baseline and reconstructed target worktrees.-- Produces: exact command/result evidence without emulator state overlap.--- [ ] **Step 1: Run target Settings gates**--Run feature boundary, unit, lint/check, Android-test compilation, and focused app adapter/navigation tests. Save complete logs outside the primary checkout.--- [ ] **Step 2: Run target Live gates**--Run feature boundary, unit, lint/check, Android-test compilation, focused app adapters, playback handoff, and debug assembly. Save complete logs.--- [ ] **Step 3: Run historical buildability checks**--Build each historical debug target with the tasks available at that commit. Record missing tasks and unrelated baseline failures exactly.--- [ ] **Step 4: Run connected suites serially**--Run Settings and Live connected tests one suite at a time on `emulator-5554`; record XML paths, counts, failures, and skips.--### Task 6: Capture and compare historical visual states--**Files:**-- Modify: `docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md`-- Create outside primary checkout: `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\visual\` with `settings-baseline`, `settings-target`, `live-baseline`, `live-target`, and `diffs` children.--**Interfaces:**-- Consumes: baseline/target APKs and identical emulator fixture/configuration.-- Produces: paired 1920x1080 screenshots and visual classifications.--- [ ] **Step 1: Record and normalize emulator configuration**--Capture resolution, density, locale, layout direction, font scale, animation scales, package version, and fixture-reset procedure before installing either build.--- [ ] **Step 2: Capture Settings pairs**--Replay the same Settings navigation, sections, dialogs, focus states, backup preview, provider management, parental controls, and error/empty states against baseline and target.--- [ ] **Step 3: Capture Live/EPG pairs**--Replay equivalent drawer, list, selected/focused, search/filter, hidden/favorite/reorder, preview, Guide, option/dialog, empty/error, and MultiView-entry states.--- [ ] **Step 4: Compute and review diffs**--Record dimensions and SHA-256 hashes, generate pixel/perceptual comparisons, manually inspect every nonzero difference, and classify it in the findings ledger.--### Task 7: Replay behavioral parity journeys--**Files:**-- Modify: `docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md`-- Create outside primary checkout: `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\behavior\*.md`--**Interfaces:**-- Consumes: the normalized baseline/target emulator state.-- Produces: step-by-step route, focus, callback, persistence, and side-effect comparisons.--- [ ] **Step 1: Replay Settings journeys**--Cover D-pad section navigation, dialog opening/dismissal, Back/focus restoration, provider add completion, preference persistence, backup safe paths, update check, and parental/EPG controls.--- [ ] **Step 2: Replay Live journeys**--Cover category/source/channel selection, search, quick filters, hidden items, favorites/reorder, preview replacement/fullscreen/return, Guide navigation/options/search, and MultiView planning.--- [ ] **Step 3: Replay cross-slice persistence journeys**--Change Settings-owned Live preferences, force-stop/relaunch, and verify the corresponding Live/EPG state before restoring fixture state.--- [ ] **Step 4: Classify every observed behavior difference**--Record regression, improvement, equivalent, pre-existing, or inconclusive status with exact evidence.--### Task 8: Validate long-duration Live playback--**Files:**-- Modify: `docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md`-- Create outside primary checkout: `C:\Users\david\AppData\Local\Temp\streamvault-phase5-regression-audit-20260830-4e19dbd570eb4dd2bb945daa94aa2993\live-playback\channel-1\` and `channel-2\`.--**Interfaces:**-- Consumes: target APK, seeded live provider, and serialized emulator access.-- Produces: two-channel screenshot/hash/media-session/log evidence.--- [ ] **Step 1: Select and record two available HLS channels**--Record channel names and stream type without exposing credentials or sensitive URLs.--- [ ] **Step 2: Capture each channel at two-second cadence**--Prefer 61 screenshots over roughly two minutes per channel; never accept fewer than 45 over approximately 90 seconds.--- [ ] **Step 3: Validate progression and player health**--Count unique hashes, capture media session state, save sanitized logcat, and scan for HLS prepare/read/first-frame/recovery plus fatal/stuck/MPEG-TS markers.--- [ ] **Step 4: Record pass/fail per channel**--Do not infer a general Live pass from one healthy channel.--### Task 9: Fix confirmed regressions in the target worktree--**Files:**-- Modify in target worktree only: exact production/test files named by confirmed findings.-- Modify: `docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md`--**Interfaces:**-- Consumes: confirmed regression findings with deterministic reproducers.-- Produces: isolated fix commits and focused passing tests.--- [ ] **Step 1: Write or verify a failing focused test per regression**--Run each reproducer before the production fix and save the failure output. If an existing frozen-target change already contains the fix, verify the test fails against its immediate pre-fix form in a disposable worktree or parent comparison.--- [ ] **Step 2: Implement the smallest behavior-preserving fix**--Do not bundle redesign, unrelated cleanup, or acceptance of changed behavior into the fix.--- [ ] **Step 3: Run focused and neighboring tests**--Run the direct reproducer plus the relevant feature/app contract suite and record exact results.--- [ ] **Step 4: Commit and review each fix in isolation**--Commit only in the target worktree. Record commit SHA, test evidence, reviewer verdict, and integration notes in the report.--### Task 10: Final verification and report closure--**Files:**-- Modify: `docs/COMPOSE_REDUCTION_PHASE5_SETTINGS_LIVE_REGRESSION_AUDIT.md`-- Modify: `docs/COMPOSE_REDUCTION_AND_UI_ARCHITECTURE_PLAN.md`--**Interfaces:**-- Consumes: all static, automated, visual, behavioral, playback, and fix evidence.-- Produces: the final audit report and an explicit active-worktree integrity statement.--- [ ] **Step 1: Run final focused suites in the target worktree**--Re-run Settings/Live boundaries, unit/check gates, connected suites, app debug assembly, and all regression reproducers. Run `graphify update .` after code changes.--- [ ] **Step 2: Perform a final read-only primary checkout drift check**--Compare current HEAD/status with the frozen target. Record later changes separately; do not overwrite or incorporate them silently.--- [ ] **Step 3: Complete the findings ledger**--Ensure every finding has classification, severity, evidence, fix/test status, and remaining limitation. List improvements separately and retain inconclusive gates as open.--- [ ] **Step 4: Verify checkout integrity and hand off isolated commits**--Confirm the primary checkout has no audit-caused non-Markdown changes. Report isolated fix SHAs and paths; do not merge, cherry-pick, push, or remove another agent's worktree.diff --git a/docs/superpowers/specs/2026-08-30-phase5-settings-live-regression-audit-design.md b/docs/superpowers/specs/2026-08-30-phase5-settings-live-regression-audit-design.mddeleted file mode 100644index 675003f9..00000000--- a/docs/superpowers/specs/2026-08-30-phase5-settings-live-regression-audit-design.md+++ /dev/null@@ -1,153 +0,0 @@-# Phase 5 Settings and Live Regression Audit Design--Date: 2026-08-30-Status: Approved for execution--## Context--Phase 5 mechanically extracted `:feature:settings` and `:feature:live` from-`:app`. The accepted extraction designs require behavioral and visual parity,-but the existing connected suites and post-extraction goldens do not by-themselves prove parity with the pre-extraction application. Confirmed-regressions have already included an inherited black Live category title and-provider completion leaving the Add Provider route open.--The primary checkout is actively used by another agent. This audit may write-Markdown there, but must not change its branch, stage or commit files, edit-code, build from it, install an APK from it, or otherwise disturb it. All code-inspection that needs generated artifacts, builds, tests, fixes, or commits-will occur in isolated external worktrees.--## Goal--Produce an evidence-backed regression audit for the Phase 5 Settings and Live-slices, classify every observed difference as a regression, intentional-improvement, pre-existing issue, equivalent change, or inconclusive result,-and fix confirmed regressions in an isolated audit worktree with focused tests.--## Comparison points--- Settings rollback: `e12f816fd12ac688a6df6391b5740d33d8a3ada0`.- This is the recorded inventory/baseline commit and the parent of the first- Settings implementation commit, `1c0e6ae6`.-- Live rollback: `ea349d83dceb8baa45ef1a18b7a45fa295a264f5`.- This is the rollback point recorded in the Live report and the parent of the- first Live implementation commit, `712751b7`.-- Combined rollback: the Settings rollback, used once to detect interaction- drift across both extracted slices.-- Frozen target: `bb52375d04e72e95a7dd70b0c79e2aab91c2121a`- plus the non-document working-tree snapshot captured at audit start.- The tracked patch hash is `29be114106b2c471531e25caa6fd1a54027d2b3b`;- the untracked-file manifest SHA-256 is- `46ebf25851b43bd6ecdb9daff3d1aa68fd2644c0313104c0fdcb36b4ad699cc9`.--The target is frozen because the primary checkout may continue changing while-the audit runs. A final read-only drift check will report changes made after-the freeze; it will not silently redefine the target.--## Isolation and concurrency--Create uniquely named detached worktrees beneath Windows temporary storage:--- Settings historical worktree at the Settings rollback.-- Live historical worktree at the Live rollback.-- Target audit worktree at the frozen target, with the captured tracked patch- and two captured untracked test files applied.--Read-only Settings, Live, and cross-slice history reviews may run in parallel.-Builds and emulator journeys are serialized. Only one process may install an-APK, mutate emulator state, capture device evidence, or run connected tests at-a time.--No audit commit is merged, cherry-picked, pushed, or applied to the primary-checkout. Fix commits remain in the target audit worktree and are reported for-later integration.--## Audit method--### Static differential review--For each slice, compare its rollback against the frozen target and inspect:--- route arguments, destination mapping, Back behavior, and dialog dismissal;-- callback order, exactly-once side effects, persistence, and error paths;-- focus request order, D-pad/Enter/long-press behavior, and focus restoration;-- colors, dimensions, placement, ordering, conditional visibility, text,- plurals, placeholders, and locale resources;-- state defaults, stable keys/content types, list ordering, and saveable state;-- app-to-feature adapters and cross-slice contracts.--The cross-slice review covers Settings controls that affect Live behavior,-provider completion entered through Settings, backup preview composition,-Live player/MultiView handoff, and shared resources or UI primitives.--### Automated evidence--Run the focused unit, boundary, lint, Android-test compilation, connected, and-app integration gates named in the two accepted extraction designs. Historical-commits use the tasks that existed at those commits; missing historical tasks-are recorded rather than treated as failures.--Every confirmed regression fixed during the audit receives the narrowest-practical reproducer test before production changes. Existing tests are not-rewritten merely to accept changed behavior.--### Visual parity--Use the same API 36 `Television_1080p` emulator at 1920x1080 for baseline and-target captures. Record locale, layout direction, density, font scale,-animation scales, theme, package, fixture state, and capture command. Reset to-the same documented fixture state between builds.--Capture Settings navigation and every reachable section/dialog state, plus-Live category/channel drawers, source switching, search, quick filters,-hidden items, favorites/reorder, preview, Guide grid/hero/options/search,-empty/error states, and MultiView entry. Compare historical and target PNGs by-dimensions, hashes, pixel/perceptual difference, and manual review. A-post-extraction golden is supporting evidence only, not a historical oracle.--### Behavioral parity--Replay equivalent D-pad, Enter, Back, long-press, text-input, persistence, and-navigation journeys against baseline and target. Record the focused node or-route at every transition and verify callback/side-effect counts where the UI-does not expose enough evidence.--### Live playback--Validate at least two Live channels. Prefer 61 screenshots per channel at a-two-second cadence, never fewer than 45 over roughly 90 seconds. A pass needs-changing hashes through the full window, media session `PLAYING` with-`error=null`, sanitized HLS prepare/read/first-frame or intended recovery-evidence, and no fatal error, stuck timeout, or unintended MPEG-TS fallback.--## Finding classification--Each finding records an ID, slice, comparison, screen/flow, type, severity,-expected and observed behavior, reproduction, suspected introduction point,-evidence paths, classification, fix commit, test, and final status.--Allowed classifications are:--- Regression: unintended loss of prior behavior or visual parity.-- Improvement: intentional or demonstrably better behavior with no contract- violation; the exact benefit and compatibility reasoning are recorded.-- Equivalent: implementation changed but observable behavior did not.-- Pre-existing: reproducible at the applicable rollback point.-- Inconclusive: blocked by fixture, hardware, external service, or unstable- evidence; never promoted to a pass.--## Completion criteria--- Both slices have complete static review inventories.-- The applicable focused automated gates are recorded with exact commands and- results.-- Baseline/target visual and behavior matrices identify passed, failed, and- unavailable states without collapsing unavailable into passed.-- Confirmed regressions have isolated fixes and focused verification, or an- explicit reason they remain open.-- Two-channel long-duration Live evidence satisfies the project protocol.-- The audit report contains all findings, improvements, limitations, target- drift, worktree fix commits, and remaining gates.-- The primary checkout has no non-Markdown audit changes.-diff --git a/domain/src/test/java/com/streamvault/domain/model/AppHomeDashboardShelfTest.kt b/domain/src/test/java/com/streamvault/domain/model/AppHomeDashboardShelfTest.ktindex 0d502647..470660a1 100644--- a/domain/src/test/java/com/streamvault/domain/model/AppHomeDashboardShelfTest.kt+++ b/domain/src/test/java/com/streamvault/domain/model/AppHomeDashboardShelfTest.kt@@ -12,6 +12,7 @@ class AppHomeDashboardShelfTest {AppHomeDashboardShelf.RECENT_CHANNELS,AppHomeDashboardShelf.LIVE_SHORTCUTS,AppHomeDashboardShelf.CONTINUE_WATCHING,+ AppHomeDashboardShelf.PINNED_CATEGORIES,AppHomeDashboardShelf.RECENT_MOVIES,AppHomeDashboardShelf.RECENT_SERIES).inOrder()diff --git a/domain/src/test/java/com/streamvault/domain/playback/ArchivePlaybackTest.kt b/domain/src/test/java/com/streamvault/domain/playback/ArchivePlaybackTest.ktindex 7a04f319..5d65503b 100644--- a/domain/src/test/java/com/streamvault/domain/playback/ArchivePlaybackTest.kt+++ b/domain/src/test/java/com/streamvault/domain/playback/ArchivePlaybackTest.kt@@ -51,6 +51,28 @@ class ArchivePlaybackTest {assertThat(channel.isArchivePlayable(program, now)).isTrue()}+ @Test+ fun `catch-up eligibility includes exact completion and window-start boundaries`() {+ val now = 3 * 86_400_000L+ val channel = Channel(+ id = 42L,+ name = "News",+ providerId = 7L,+ catchUpSupported = true,+ catchUpDays = 2,+ streamUrl = "xtream://7/live/42",+ streamId = 42L+ )+ val program = Program(+ channelId = "news",+ title = "Boundary Bulletin",+ startTime = now - (2 * 86_400_000L),+ endTime = now+ )++ assertThat(channel.isArchivePlayable(program, now)).isTrue()+ }+@Testfun `provider archive flag keeps replay enabled with replay metadata`() {val channel = Channel(diff --git a/feature/live/src/androidTest/assets/goldens/live_category_row.png b/feature/live/src/androidTest/assets/goldens/live_category_row.pngindex e5b8cad7..9c50cff2 100644Binary files a/feature/live/src/androidTest/assets/goldens/live_category_row.png and b/feature/live/src/androidTest/assets/goldens/live_category_row.png differdiff --git a/feature/live/src/androidTest/assets/goldens/live_channel_row_surface.png b/feature/live/src/androidTest/assets/goldens/live_channel_row_surface.pngindex f9d3e37b..52597bb2 100644Binary files a/feature/live/src/androidTest/assets/goldens/live_channel_row_surface.png and b/feature/live/src/androidTest/assets/goldens/live_channel_row_surface.png differdiff --git a/feature/live/src/androidTest/assets/goldens/live_channel_surfaces.png b/feature/live/src/androidTest/assets/goldens/live_channel_surfaces.pngindex e6cfc97b..08e0e13e 100644Binary files a/feature/live/src/androidTest/assets/goldens/live_channel_surfaces.png and b/feature/live/src/androidTest/assets/goldens/live_channel_surfaces.png differdiff --git a/feature/live/src/androidTest/assets/goldens/live_reorder_top_bar.png b/feature/live/src/androidTest/assets/goldens/live_reorder_top_bar.pngindex 69c89ee0..1c6e1719 100644Binary files a/feature/live/src/androidTest/assets/goldens/live_reorder_top_bar.png and b/feature/live/src/androidTest/assets/goldens/live_reorder_top_bar.png differdiff --git a/feature/live/src/androidTest/assets/goldens/live_selection_chips.png b/feature/live/src/androidTest/assets/goldens/live_selection_chips.pngindex affc365d..bd178719 100644Binary files a/feature/live/src/androidTest/assets/goldens/live_selection_chips.png and b/feature/live/src/androidTest/assets/goldens/live_selection_chips.png differdiff --git a/feature/live/src/androidTest/assets/goldens/live_source_switcher.png b/feature/live/src/androidTest/assets/goldens/live_source_switcher.pngindex e94c463b..163e6809 100644Binary files a/feature/live/src/androidTest/assets/goldens/live_source_switcher.png and b/feature/live/src/androidTest/assets/goldens/live_source_switcher.png differdiff --git a/feature/live/src/androidTest/java/com/streamvault/feature/live/epg/EpgScreenBehaviorTest.kt b/feature/live/src/androidTest/java/com/streamvault/feature/live/epg/EpgScreenBehaviorTest.ktindex 20f5a82a..43c8776f 100644--- a/feature/live/src/androidTest/java/com/streamvault/feature/live/epg/EpgScreenBehaviorTest.kt+++ b/feature/live/src/androidTest/java/com/streamvault/feature/live/epg/EpgScreenBehaviorTest.kt@@ -1,6 +1,7 @@package com.streamvault.feature.live.epgimport androidx.compose.foundation.rememberScrollState+import androidx.compose.runtime.CompositionLocalProviderimport androidx.compose.ui.semantics.SemanticsActionsimport androidx.compose.ui.test.assertIsDisplayedimport androidx.compose.ui.test.junit4.createComposeRule@@ -10,6 +11,7 @@ import androidx.compose.ui.unit.dpimport androidx.test.ext.junit.runners.AndroidJUnit4import com.google.common.truth.Truth.assertThatimport com.streamvault.core.ui.theme.StreamVaultTheme+import com.streamvault.domain.model.AppTimeFormatimport com.streamvault.domain.model.Channelimport com.streamvault.domain.model.Programimport com.streamvault.feature.live.presentation.epg.GuideDensity@@ -20,6 +22,7 @@ import com.streamvault.feature.live.presentation.epg.LiveGuidePreviewLabelsimport com.streamvault.feature.live.presentation.epg.LiveGuidePreviewPaneimport com.streamvault.feature.live.presentation.epg.LiveGuideToolbarLabelsimport com.streamvault.feature.live.presentation.epg.LiveGuideToolbarRow+import com.streamvault.feature.live.presentation.time.LocalLiveTimeFormatimport org.junit.Ruleimport org.junit.Testimport org.junit.runner.RunWith@@ -122,6 +125,38 @@ class EpgScreenBehaviorTest {composeRule.onNodeWithText("No schedule").assertIsDisplayed()}+ @Test+ fun guidePreviewPane_rendersProgramTimeRangeWithEnDash() {+ val now = System.currentTimeMillis()+ val program = Program(+ channelId = "world-news",+ title = "Evening Bulletin",+ startTime = now - 60_000L,+ endTime = now + 60_000L+ )++ composeRule.setContent {+ StreamVaultTheme {+ CompositionLocalProvider(LocalLiveTimeFormat provides AppTimeFormat.TWENTY_FOUR_HOUR) {+ LiveGuidePreviewPane(+ previewPlayerEngine = null,+ isPreviewLoading = false,+ focusedChannel = testChannel(21L, "World News"),+ focusedProgram = program,+ labels = LiveGuidePreviewLabels(+ title = "Preview",+ placeholderTitle = "Select a channel",+ noSchedule = "No schedule"+ )+ )+ }+ }+ }++ composeRule.onNodeWithText("Evening Bulletin").assertIsDisplayed()+ composeRule.onNodeWithText(" – ", substring = true).assertIsDisplayed()+ }+@Testfun guideGridRow_preservesChannelProgramIdentityAndCallbacks() {val now = System.currentTimeMillis()diff --git a/feature/live/src/main/java/com/streamvault/feature/live/presentation/epg/LiveGuideEpgOverrideDialog.kt b/feature/live/src/main/java/com/streamvault/feature/live/presentation/epg/LiveGuideEpgOverrideDialog.ktindex 98116aaa..502dacc5 100644--- a/feature/live/src/main/java/com/streamvault/feature/live/presentation/epg/LiveGuideEpgOverrideDialog.kt+++ b/feature/live/src/main/java/com/streamvault/feature/live/presentation/epg/LiveGuideEpgOverrideDialog.kt@@ -60,7 +60,7 @@ data class LiveGuideEpgOverrideLabels(val cancel: String)-private const val DESCRIPTOR_SEPARATOR = " \u05D2\u20AC\u00A2 "+private const val DESCRIPTOR_SEPARATOR = " • "internal fun liveGuideOverrideSummary(mapping: ChannelEpgMapping?,diff --git a/feature/live/src/test/java/com/streamvault/feature/live/presentation/epg/LiveGuideEpgOverrideTest.kt b/feature/live/src/test/java/com/streamvault/feature/live/presentation/epg/LiveGuideEpgOverrideTest.ktindex 47760c89..82a3201e 100644--- a/feature/live/src/test/java/com/streamvault/feature/live/presentation/epg/LiveGuideEpgOverrideTest.kt+++ b/feature/live/src/test/java/com/streamvault/feature/live/presentation/epg/LiveGuideEpgOverrideTest.kt@@ -51,6 +51,6 @@ class LiveGuideEpgOverrideTest {currentProviderFormat = "Provider: %1\$s",currentExternalFormat = "External: %1\$s")- ).startsWith("Manual: News")+ ).isEqualTo("Manual: News • External • news.xml")}}diff --git a/feature/settings/src/androidTest/java/com/streamvault/feature/settings/presentation/SettingsSyncLabelsTest.kt b/feature/settings/src/androidTest/java/com/streamvault/feature/settings/presentation/SettingsSyncLabelsTest.ktnew file mode 100644index 00000000..a7a3437d--- /dev/null+++ b/feature/settings/src/androidTest/java/com/streamvault/feature/settings/presentation/SettingsSyncLabelsTest.kt@@ -0,0 +1,22 @@+package com.streamvault.feature.settings.presentation++import androidx.test.ext.junit.runners.AndroidJUnit4+import androidx.test.platform.app.InstrumentationRegistry+import com.google.common.truth.Truth.assertThat+import com.streamvault.feature.settings.R+import org.junit.Test+import org.junit.runner.RunWith++@RunWith(AndroidJUnit4::class)+class SettingsSyncLabelsTest {++ @Test+ fun sync_option_labels_use_title_case() {+ val context = InstrumentationRegistry.getInstrumentation().targetContext++ assertThat(context.getString(R.string.settings_sync_option_sync_now))+ .isEqualTo("Sync Now")+ assertThat(context.getString(R.string.settings_sync_option_rebuild_index))+ .isEqualTo("Rebuild Index")+ }+}diff --git a/feature/settings/src/main/res/values/strings.xml b/feature/settings/src/main/res/values/strings.xmlindex 41d89edb..c4d531c3 100644--- a/feature/settings/src/main/res/values/strings.xml+++ b/feature/settings/src/main/res/values/strings.xml@@ -445,8 +445,8 @@<string name="settings_update_status_permission_required">Install permission required</string><string name="settings_update_available_message">StreamVault %1$s is available.</string><string name="settings_update_current_message">You already have the latest StreamVault release.</string>- <string name="settings_sync_option_sync_now">Sync now</string>- <string name="settings_sync_option_rebuild_index">Rebuild index</string>+ <string name="settings_sync_option_sync_now">Sync Now</string>+ <string name="settings_sync_option_rebuild_index">Rebuild Index</string><string name="settings_sync_option_tv">Live TV</string><string name="settings_sync_option_movies">Movies</string><string name="settings_sync_option_series">Series</string>