Addresses the review of PR #46 (#12/#13):
- Age-retention backfill/prune loop: mark a folder complete at the
retention floor and resume from the persisted nextBeforeUid low-water
mark; loosening resumes via AccountRepository.resetBackfillProgress.
- Guard the windowed reconcile bound to the lowest positive UID so a
getUID==-1 message can't collapse it and wipe backfilled history.
- Order count-based retention by uid DESC to match the fetch window,
ending the re-fetch/re-prune churn for high-UID/old-Date messages.
- BackfillWorker chains slices while work remains.
- Extract shared effectiveRetention / isActiveNetworkUnmetered /
attachmentCacheDir helpers; remove dead deleteSyncedNotIn/getForAccount;
refresh only pre-existing rows in persistBatch; add composite index
(accountId, folder, uid) with migration + regenerated 13.json.
Adds an age-floor prune regression test. Fast gate + androidTest compile
green on JDK 21.
Follow-ups filed for below-the-cut findings: #93, #94, #95, #96.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Code-review fixes: an all-empty paragraph group (a blank line isolated by an
alignment split) emitted <p></p>, which the parser collapses — it now emits one
<br> per line so blank lines round-trip. The identical span-merge helper that
existed in both the parser and RichTextEditing is now a single shared
mergeSameValueSpans() in RichText.kt, and the private applyBlock/applyLink drop
their never-used default font resolver.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Send now scans the subject and body for "attach" and its variants
(word-bounded, case-insensitive). When the text mentions one but the
message carries no attachment, an AlertDialog asks "Need to attach
anything?" — Yes returns to composing and pulses the attach button,
No sends the message as-is, and dismissing cancels the send. (#79)
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
RichStyle becomes a sealed interface (Bold/Italic/Underline/Strikethrough +
FontFamily/FontSize/FontColor/Highlight); RichTextContent gains alignments,
images, and baseStyle channels. The HTML serializer emits merged <span style>
runs, text-align on <p>/<li> (splitting merged paragraphs at alignment
boundaries), <img src="cid:…"> over the visible [image: name] token, and a
single outer <div style> for the base style. The parser is a faithful inverse
and additionally tolerates <del>/<strike>, px font sizes, #rgb colors, and
start/end alignment synonyms; unknown CSS is ignored without dropping text.
hasFormatting() covers every new channel so ComposeViewModel.normalizedHtml()
never silently drops serialized formatting. The editor carries parameterized
style identity via string annotations (libremail:style / libremail:image), maps
alignment onto ParagraphStyle ranges, holds baseStyle in separate field state,
and RichTextEditing.toggleStyle now replaces a different value of the same kind
while styleAt() answers "current value over the selection" for pickers.
ColorSwatchRow is added for the upcoming color/highlight dialogs. No UI change.
Closes#70
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The Cc and Bcc fields now start collapsed into small left-aligned link
buttons under the To box, freeing about two field heights of vertical
space for the message body. Tapping a link expands it into the regular
input field and focuses it; a field also expands on its own when it
already carries recipients (reply-all/mailto prefill, resumed drafts)
and never re-collapses once shown, so it cannot vanish mid-edit. The
expansion state lives in the UI via rememberSaveable and survives
rotation. Moving the Bcc field also gives it the medium shape every
sibling field already had.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
saveOrDeleteDraft persists the Bcc line, but the init-block restore
never copied it back, so reopening a draft silently dropped its Bcc
recipients (and re-saving then lost them for good).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Tapping a new-mail notification only brought the app to the foreground:
the content PendingIntent was a bare launch intent shared by every
notification, and nothing on the activity side handled a message target.
Per-message notifications now carry an explicit open-message intent —
action + id extra + a per-message data URI, so each message keeps its
own PendingIntent under filterEquals instead of all collapsing onto one
FLAG_UPDATE_CURRENT entry. MainActivity parses the id on fresh launch
and in onNewIntent and hands it to the NavHost as pending state (the
pendingCompose handoff pattern) to navigate to the reader. The group
summary keeps the plain open-the-app intent.
Fixes#56
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Resolve the Room schema-version collision: main's PR #54 added MIGRATION_11_12 (folders.specialUse), colliding with this branch's v11->v12 uid/retention/backfill migration. Renumbered ours to MIGRATION_12_13 — the two migrations touch disjoint tables, so ours stacks cleanly on top — bumped the DB to version 13, kept main's 12.json as the v12 schema and regenerated 13.json, and renamed Migration11To12Test -> Migration12To13Test.
Verified locally: assembleDebug, testDebugUnitTest, lintDebug, ktlint, detekt, compileDebugAndroidTestKotlin, and Migration12To13Test on an API 37 emulator all pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
CI's "Static analysis" job runs :app:ktlintCheck :app:detekt, which the
local preflight gate did not, so style violations in test/androidTest
source sets (which lintDebug skips) failed the merge gate only after
push. Add both to the /preflight skill and mirror the change in CLAUDE.md.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Body expression on the signature line (function-signature) and one
argument per wrapped line (argument-list-wrapping) in the tests added
for drawer folder de-duplication.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The androidTest compile failed ONLY in CI with "Unresolved reference 'observeAll'"
on the single line using dao.observeAll(), while every other MessageDao call in the
same file resolved, the identical observeAll().first().map{}.toSet() in
LibreMailDatabaseTest compiled fine in the same unit, and the file compiled cleanly
locally (even `clean --no-build-cache`). That points to a Kotlin incremental-compilation
artifact specific to the newly-added file, not a code error.
Replace the observeAll()-based readback with explicit getById point lookups — a clearer
per-row assertion that also sidesteps the glitch. Verified on the API 37 emulator (5/5).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The drawer rendered every standard-role folder with a generic friendly
name (e.g. "Drafts") and discarded the server name, so a Gmail account
with both a provider built-in folder and a same-named user folder showed
two identical entries (Drafts, Archive, Spam).
De-duplicate labels provider-agnostically: when 2+ folders would render
the same name, the provider's built-in special folder (identified by RFC
6154 SPECIAL-USE flags, now persisted on the folder cache) gets the
provider name appended ("Archive - Gmail"), a nested user folder gets its
parent location ("Reports (Work)"), and a top-level user folder keeps its
plain name. Only triggers on a real collision, so stock accounts are
unchanged.
Adds a `specialUse` column to the folders table (Room v11 -> v12).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The gate's "backfill and prune never interleave" guarantee was only argued
structurally: the MailBackfiller/MailPruner unit tests each build a throwaway,
uncontended gate, so the serialization is never exercised. Add MailMaintenanceGateTest:
- oneGateSerializesConcurrentCriticalSections: 50 coroutines contend on one gate;
an overlap counter must never exceed 1 (guards against a per-access mutex).
- aConcurrentPruneWaitsForAnInFlightBackfillToReleaseTheGate: a real MailBackfiller
parks inside the gate (its fetchOlderThan suspended) while a real MailPruner is
launched concurrently; the two share ONLY the gate, and the prune provably cannot
enter its critical section until the backfill releases.
Turns the defence-in-depth guarantee from argued to asserted.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The backfiller's "no message fetched twice" claim (full-history + resume tests)
previously rested on the insertNew fake de-duping by id, so a re-fetched page was
silently absorbed and `cached.size == TOTAL` could not fail on it. Count the rows
offered to insertNew BEFORE de-dupe and assert it equals TOTAL - WINDOW, so any
re-request of an already-cached page now fails the test. This isolates the real
boundary-descent guarantee and, unlike a fetchOlderThan call-count, is independent
of BACKFILL_BATCH_SIZE.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The MailPruner/MailBackfiller unit tests mock the DAO, so the queries that
actually define the device-only retention floor were never exercised against a
real database: the newest-N-by-(timestampMillis, uid) prune selection, the
strict age cutoff, the windowed reconcile that spares backfilled history, and
the lowest-uid / count / oldest floor probes the backfiller stops on.
Add an instrumented MessageDao test on an in-memory Room DB covering all of
them, including the timestamp/uid tie-break direction (a flipped ORDER BY would
locally delete the user's newest mail) and inInbox/folder/account scoping.
This closes the disjoint-sets safety argument at the SQL-boundary level, not
just the orchestration level. Verified on the API 37 emulator (5/5).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
MessageDao.observeAll() ran `SELECT * FROM messages` and returned full
MessageEntity rows — including the potentially large body/isHtml columns —
for every cached message at once. Dragging big HTML bodies through SQLite's
shared ~2 MB CursorWindow overflowed it once enough bodies were cached,
crashing with "Couldn't read row N from CursorWindow" (#51).
Replace it with observeSummaries(), a body-less column projection into a new
lightweight MessageSummary POJO. The list never renders or searches the body,
and the reader already loads it lazily per-message via getById when a message
is opened, so nothing else needs it.
Add a regression test that reads back rows whose bodies exceed the window.
Closes#51
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
AGP 9's consistent resolution shares the runtime serialization version with the
androidTest classpath, where androidx.savedstate pins it to 1.7.3. Room 2.8's
schema-bundle serializers are compiled against >= 1.8.0, so MigrationTestHelper
threw AbstractMethodError on GeneratedSerializer.typeParametersSerializers(),
failing every E2E job. Import the serialization BOM as a platform to force 1.8.1.
Verified on-device (API 37): Migration9To10Test fails unpatched, passes patched.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Replace the fixed 50-message-per-folder header cap with a background,
resumable full-history backfill, and add a user-configurable device-only
retention limit that prunes local mail beyond it without ever deleting from
the server.
- ImapClient.fetchOlderThan pages a folder backwards in bounded batches,
locating the boundary by binary search over message numbers (O(log n) tiny
UID fetches, memory bounded to one batch).
- MailBackfiller + BackfillWorker page each synced folder newest→oldest,
persisting a per-folder boundary in a new backfill_progress table so a run
interrupted by process death / network loss resumes exactly where it stopped.
Runs off the sync mutex, so foreground sync / pull-to-refresh stay responsive.
- MailSyncer now reconciles server deletions only within the recent UID window
(deleteSyncedInWindowNotIn) instead of wiping everything outside the recent
50, so backfilled history survives each foreground sync. A materialized
messages.uid column powers the windowed reconcile and backfill boundary.
- Body/attachment prefetch still honours FetchPolicy (headers first).
- Per-account count/age overrides (nullable) with a global default; 0 = keep
everything (the default, matching #12).
- MailPruner + PruneWorker delete local rows beyond the limit (cascading
attachment rows + on-disk cache), never issuing a server delete. Deletes are
chunked under SQLite's 999-parameter limit.
- Precedence with backfill: backfill pauses (does not complete) at the
retention floor and both jobs share a maintenance mutex, so they never
contend; foreground fetch is also capped by the count so it can't re-download
what pruning just trimmed.
- Settings UI for the global default and per-account override, with copy making
clear it is device-only, not the server.
Room schema v9→v10 (migration + exported schema + MigrationTestHelper test).
GreenMail tests prove the backfill caches >50 and resumes after interruption;
pruning tests cover count/age limits and never touch the server.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add a guided, F-Droid-safe onboarding step and an Advanced Settings recovery
row that let users move LibreMail to "Unrestricted" battery usage, so IMAP
IDLE push (IdleService) and periodic WorkManager sync aren't throttled or
killed by Doze. Deep-links to the system app-details screen rather than the
restricted REQUEST_IGNORE_BATTERY_OPTIMIZATIONS dialog, so it needs no new
permission and is safe on Play (#17) and F-Droid (#16).
- BatteryPromptDecision: pure, unit-tested gate (supported && !unrestricted && !handled)
- BatteryOptimizationManager: reads isIgnoringBatteryOptimizations, builds the deep-link intent
- Onboarding step shown after the first account is added; skipped when already
unrestricted or already handled; re-checks status on resume
- Advanced Settings row shows current status and re-opens the system screen
- battery_prompt_handled flag persisted in the settings DataStore (kept out of AppSettings)
- Unit tests for the decision + view model; Espresso E2E for the step
Closes#49
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Resolve LibreMailApp.kt: union the compose function params so mailto prefill
(pendingCompose/onComposeHandled) coexists with onboarding start-gating
(appViewModel) and the crash dialog (startupViewModel); keep LaunchedEffect +
getValue/remember imports. Verified locally: assembleDebug + testDebugUnitTest +
lintDebug + ktlintCheck + detekt.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Resolve conflicts from #40/#41/#44:
- build.gradle.kts: keep DEBUG_REPORT_ENDPOINT field + val, take #40's
outlookRedirectScheme (gmailRedirectScheme was deleted).
- LibreMailApp.kt: function takes BOTH appViewModel (start-dest gating, #44)
and startupViewModel (crash dialog, #42); use renamed AccountPickerScreen.
- SettingsScreen.kt: keep both the Diagnostics (#42) and Backup (#41) sections.
Verified locally: assembleDebug + testDebugUnitTest + lintDebug + ktlintCheck + detekt.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Real cause of the API 29-36 E2E timeout (the earlier 5s->15s bump didn't help,
proving it wasn't slowness): AppPasswordSetupScreen is a scrolling Column and the
"Test and add" button sits below the fold on the short default matrix emulator, so
the positional performClick was a silent no-op -> no add -> no navigation -> the
add-another wait never resolved. It passed on API 37 only because that job uses a
taller pixel_2 AVD. performScrollTo() each field + the button before interacting,
matching the existing pattern in SettingsScreenTest. Verified compileDebugAndroid
TestKotlin + ktlintCheck on JDK 21.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
OnboardingFlowTest passed on the API-37 job but timed out (ComposeTimeoutException
after 5000ms) across the animation-disabled API 29-36 matrix, at the single
async-gated transition: click -> viewModelScope coroutine -> addImapAccount ->
DONE -> LaunchedEffect -> navigate -> AddAnother render. The flow is correct
(green on API 37; ManualSetupScreenTest proves the add-callback path); the 5s cap
was just too tight for that compound step on slower matrix emulators. waitUntil
returns as soon as the text appears, so the happy path is unaffected.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The #32/#33 change added a required onReportProblem parameter to SettingsScreen
but left the existing instrumented SettingsScreenTest calling the old signature,
so :app:compileDebugAndroidTestKotlin failed in every E2E job (the local fast
gate never compiles the androidTest variant). Pass onReportProblem = {} in the
test. Verified with :app:compileDebugAndroidTestKotlin on JDK 21.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Bring README in line with the post-batch shipped state (issue #20, folding in
#31's README reconciliation):
- Rewrite the status blurb and feature list to cover the onboarding flow, rich
compose (HTML + multipart/alternative + signatures), full-history backfill
with a device-only retention cap, opt-in app lock, mailto/default-app, and
opt-in local debug reporting.
- Remove the Gmail OAuth setup section and the "no stored passwords for Gmail"
claim; Gmail/Yahoo/iCloud are now app-password IMAP/SMTP vendors.
- Add an "Accounts and onboarding" section (Outlook OAuth; Gmail/Yahoo/iCloud
app password with vendor app-password pages + Gmail 2SV note; Other IMAP/SMTP)
and keep the Outlook OAuth setup section.
- Add a "Privacy and data flow" note: opt-in cache encryption, local
user-initiated debug reporting (no hosted pipeline), and opt-in Android Backup.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Bring rich composition to LibreMail (issues #36, #37, #38, and #23).
#36 HTML editor + toolbar
- New pure, JVM-testable rich-text model (`richtext/`): RichTextContent with
inline styles + links and block markers ("• ", "N. ", "> "), serializing to a
narrow email-safe HTML subset and back (fromHtml is a faithful inverse).
- Rich editor in ComposeScreen with a bold/italic/underline, bulleted/numbered
list, block-quote, and link toolbar, backed by AnnotatedString. Unformatted
text stays plaintext-only (null HTML) so it feels unchanged and is accessible.
- #23: rounded corners on the compose fields/body via MaterialTheme.shapes.
#37 multipart/alternative + reply/forward quoting
- SmtpSender builds multipart/alternative (text/plain + text/html), nested in
multipart/mixed when there are attachments; GraphSender sends HTML content.
- HtmlToText produces a readable text/plain fallback; ReplyBuilder quotes HTML
originals as clean blockquotes (tags stripped) without corruption.
- HTML body persists/restores through drafts and the outbox (new nullable
bodyHtml columns; Room v9->v10 migration + schema).
#38 signatures
- New signatures table (multiple per account, one default) + repository/DAO;
migration backfills the existing per-account signature as the default.
- Rich signatures reuse the #36 editor; a Signatures management screen (list,
add/edit/delete, set default) is linked from per-account settings.
- The account's default signature auto-inserts on new compose / reply / forward
(honoring the enable toggle), placed above the quote, and stays editable.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Jason Ross
JMR-devandClaude Opus 4.8