- Replace native confirm() in the recording list with an inline
two-step confirm row; hide the player via visibility:hidden
while open so the row height stays stable.
- Hard-redirect recording delete back to /web/cases/{id}/recordings
instead of relying on the Referer header (which silently fell
back to /web/cases when stripped).
- Switch analyze/reset to a hidden return_to form field so the
source page (case detail vs. case list) is preserved reliably,
with same-origin /web/ prefix validation.
- Tighten delete_recording_test on the concrete Location header;
adjust analyze/reset redirect expectations to the new
case-detail fallback.
Adds a new POST endpoint for web-based oneliner overrides and updates
the UI to allow manual editing. The person icon replaces the pencil icon
for doctor-authored titles, signifying manual authorship rather than
editability.
Introduces inline editing for oneliner text within the case list. This
feature allows users to directly modify oneliner entries in the UI, with
changes being optimistically applied locally and then asynchronously
sent to the server.
Includes:
- New `OnelinerEditState` to manage the editing buffer and focus.
- `SaveResult` struct to communicate the outcome of background save
operations.
- UI logic to toggle between read and edit modes for oneliner entries.
- Handling of keyboard inputs (Enter, Esc) and button clicks for saving
or canceling edits.
- Toast banner to display save errors to the user.
- Asynchronous `put_oneliner` calls for background server updates.
- Local storage update for optimistic UI updates.
Introduces `set_oneliner_local` to `CaseStore` for optimistic UI
updates.
Also adds `case_update` module with `put_oneliner` to handle server-side
persistence.
Introduces `OnelinerLocks` to serialize read-modify-write operations on
`oneliner.json`. This prevents race conditions between the manual
override API (`PUT /api/cases/{case_id}/oneliner`) and the transcription
worker's auto-regeneration process.
The manual override now acquires a lock specific to the `case_dir`
before writing. The transcription worker's `update_oneliner` function
also acquires the same lock around its post-LLM re-read-and-write
sequence. This ensures that a doctor's manual edit always takes
precedence, even if it arrives while an auto-regeneration is in
progress.
This change also includes:
- A new `routes::oneliner_override` module for the PUT handler.
- Validation for the oneliner text length and emptiness.
- Integration tests for the override functionality, covering happy path,
error conditions, early latching, race conditions, and parallel PUTs.
This commit introduces a new `Manual` variant to the `OnelinerState`
enum, allowing for manual overrides of the oneliner text. This change
affects the client and server components to handle and display this new
state.
A `OnelinerOverrideRequest` struct is also added for API requests to set
manual oneliners. New tests are included to ensure the `Manual` variant
serializes and deserializes correctly.
The `compute_oneliner_display` function in
`server/src/routes/user_web.rs` is updated to treat `Manual` states the
same as `Ready` states for display purposes. The
`cases_needing_oneliner_in` function in
`server/src/transcribe/recovery.rs` is updated to not retry cases that
are in a `Manual` state.
Tap opens the Wear OS system input picker (voice/keyboard/handwriting),
pinned to de-DE for medical German vocabulary. Doctor-authored oneliners
latch a manual flag that blocks the simulated LLM burst from overwriting
them; a subsequent manual edit still wins.
Tap on a case in the list (or Tile continue-case) now opens a new
CaseDetailScreen showing date + oneliner + record EdgeButton. Only
hitting record enters RecordingScreen, which is itself trimmed to
just a live mm:ss counter and a stop EdgeButton. Swipe-right on the
recording screen discards the in-flight recording (no confirmation).
"New case" button still bypasses the detail step.
State machine: Recording now counts elapsedSeconds (up) instead of
secondsLeft. New OnStopTap and OnDiscardTap events, new reducer
transitions. Safety cap at 300 s.
ViewModel: startRecordingFlow split into a cancellable ticker job
plus a separate finalizeAndUpload run in applicationScope so the
upload survives the pop that follows a clean stop. AtomicBoolean
finalizationStarted guards against double-finalization between the
stop, discard, and auto-stop paths.
Display: FLAG_KEEP_SCREEN_ON on the recording screen via
view.keepScreenOn. DisposableEffect.onDispose doubles as the
discard trigger for swipe-right / back / host destruction;
discardIfStillRecording() is a no-op after a clean stop.
CaseDetailScreen lays out three vertical thirds: date/time at top
(Heute/Gestern/dd.MM.yy + HH:mm:ss), oneliner in the middle,
EdgeButton at the bottom. formatTime extracted to TimeFormat.kt
so list and detail never drift apart.
Introduces a `minutesAgo` helper function and uses it in `CaseStoreStub`
to populate `CaseEntry` objects with relative timestamps for recent
activities. Also updates `formatTime` in `CaseListScreen` to display
"Gerade eben", "Vor 1 Minute", or "Vor X Minuten" for entries younger
than an hour. This provides a more user-friendly display for recent
cases.
Refactor `seedDemoData` to use `java.time` for more accurate date and
time generation, including specific times for today, yesterday, and
older dates.
Update `formatTime` in `CaseListScreen` to leverage `java.time` for a
more robust and localized date and time formatting, supporting "today",
"yesterday", and older date formats.
This ensures that the `ANDROID_SERIAL` environment variable is exported
when
`ADB_SERIAL` is set. This is crucial because Android Gradle Plugin's
`android.injected.device.serial` property can silently fall back to
selecting
all attached devices if the serial doesn't match exactly. Exporting
`ANDROID_SERIAL` provides a deeper level of filtering for spawned `adb`
binaries, maintaining the intended device selection.
Improve the visual presentation of individual case entries by adjusting
styling and text handling. This includes setting a minimum height for
rows, applying consistent padding, and enabling text truncation with an
ellipsis for longer oneliner descriptions.
Drop the legacy "(none)" target fallback that quietly picked the first
attached device. Every device-bound subcommand now requires a `watch` or
`emulator` prefix (or DOCTATE_TARGET env var).
The emulator path never auto-picks a running emulator: even with exactly
one running, the menu shows so the user can boot a different AVD
instead. ADB_SERIAL / WEAR_AVD still bypass the menu for scripted runs.
Menu results now travel through a global RESOLVED_SERIAL/PICKED instead
of stdout+$(), so `die` inside the resolvers aborts the main shell
directly rather than getting swallowed by a command-substitution subshell.
This commit introduces support for Wear OS Tiles and Complications,
enabling users to view case information and launch the app directly from
their watch face.
Key changes include:
- **New Complication Service:** `DoctateComplicationService.kt` provides
a simple complication that displays "Doctate" and launches the app's
case list upon tap.
- **New Tile Service:** `DoctateTileService.kt` serves a dynamic tile
showing the current case's one-liner. It includes tappable areas to
navigate to the case list, create a new case, or continue an existing
one.
- **UI Navigation:** `AppNav.kt` and related files (`CaseListScreen.kt`,
`MainActivity.kt`, `NavCommand.kt`) set up the navigation structure
for the Wear OS app, handling intents from the Tile and Complication.
- **Data Structures:** `CaseEntry.kt` defines the minimal data structure
for case information displayed on the watch.
- **In-Memory Data Store:** `CaseStoreStub.kt` acts as a temporary,
in-memory store for case data, ensuring consistency across the
activity, Tile, and Complication.
- **Dependency Updates:** Added necessary Wear OS libraries to
`build.gradle.kts` and `libs.versions.toml`.
- **AndroidManifest:** Configured the `MainActivity` to handle
`singleTask` launch mode and added necessary intent filters for
complications.
- **Demo Data:** Seeded `CaseStoreStub` with demo data for initial
testing and visualization.
The project plan has been updated to reflect the successful testing of
the Pixel Watch 2 hardware. The plan now includes details about
ADB-over-WiFi pairing and testing against a local development server. It
also mentions the successful manual verification of the end-to-end flow
on the real device.
Further testing, including LTE mode, doze mode, and long-term foreground
service behavior, is still pending.
These files mutate on every IDE click and leak per-developer state
(last deployed device, recent activity) into commits. A single root
`.idea/` ignore replaces the scattered per-file rules in
watch/wearos/.gitignore.
Real hardware has no USB and can't resolve 10.0.2.2, so we pair over
ADB-WiFi and point the app at the dev laptop's LAN URL. run.sh gains a
`watch`/`emulator` target prefix so both devices can be attached in
parallel without serial-picking roulette.
- network_security_config.xml: whitelist 192.168.178.27 for cleartext
- build.gradle.kts: resolveProp() threads Gradle CLI `-P` overrides
into BuildConfig (SERVER_URL, API_KEY) ahead of local.properties
- run.sh: adb_cmd wrapper honours ADB_SERIAL; new `devices` and
`connect [ip:port]` subcommands; `.watch_serial` caches the paired
endpoint so `./run.sh watch install` just works on subsequent runs
- .gitignore: ignore `.watch_serial`
Completes the 5-second-record-and-upload flow:
- RecordingViewModel orchestrates recorder, countdown, and upload
- MediaRecorder wrapper produces 16kHz/AAC-LC m4a (Whisper-compatible)
- Compose RecordingScreen renders the state pyramid
(Idle / RequestingPermission / Recording / Uploading / Success / Error)
- UiState + pure reducer with full JVM unit-test coverage
- run.sh "test" subcommand wraps the unit and instrumented tiers
(set SKIP_INSTRUMENTED=1 to bypass the on-device tier)
Verified end-to-end: emulator captures "Hallo, hallo", lands at
tmpdata/<user>/<case>/*.m4a, Whisper transcribes, Ollama classifies,
document.md is written — full pipeline in ~5s.
Known: MockWebServer-based UploadClientTest hangs in setUp on the
Wear OS 34 emulator (likely SELinux socket policy). Deferred — the
real-server end-to-end already validates the wire protocol.
Builds the network foundation for the Wear OS vertical PoC:
OkHttp-based UploadClient mirroring doctate-client-core's wire protocol
(multipart /api/upload, X-API-Key, transient/terminal classification),
BuildConfig-backed Settings sourced from local.properties, CaseId factory,
DoctateApp service locator, MockWebServer-based UploadClientTest for
wire-format assertions, plus the manifest/permissions/network-security-
config plumbing for emulator-loopback cleartext.
MainActivity is still the placeholder UI; audio recording, ViewModel,
and recording screen come in follow-up commits.
Adds the deviceManager.xml file, which configures device sorting within
the IDE, to the `.idea` directory. This ensures consistent device
management across different developer environments.
Inner-loop tooling so the dev cycle does not require Android Studio.
Subcommands: build, install, start, stop, logcat, shot, clean
(default = install + start). Auto-detects JAVA_HOME (bundled JBR) and
adb. Boots an emulator on demand when no device is attached: interactive
menu of available AVDs, with WEAR_AVD=<name> env-var bypass for
non-interactive use. Waits for sys.boot_completed (180s timeout) before
installing.
Initial skeleton for the Pixel Watch dictation client (Plan Phase 5b).
Generated by Android Studio's "Empty Wear App" template, kept verbatim
to defer bikeshedding until real friction appears.
Toolchain pinned via wrapper / version catalog:
AGP 9.2.0, Kotlin 2.2.10, Gradle 9.4.1, JDK 21 (auto-provisioned).
compileSdk 36, minSdk 30 — covers Pixel Watch 1/2/3.
UI starter: Wear Compose Material3 1.5.6 with TransformingLazyColumn
and EdgeButton — already close to the planned per-case full-screen
list pattern.
Manifest declares the app standalone, so the Wear OS Network Proxy
will tunnel uploads transparently when the watch lacks LTE/WiFi.
No Phone-side companion module required.
Multi-module split (:core-domain, :core-audio, :core-sync, :core-http,
:core-storage, :app-mobile) intentionally postponed — current single
:app module is the smallest thing that compiles, which gives us a
known-good baseline before restructuring.
AS's default .gitignore left as-is. A handful of .idea/ project-config
files (compiler.xml, gradle.xml, misc.xml, codeStyles, runConfigurations)
are committed by AS's design; revisit if machine-specific paths or
multi-machine drift become painful.
Pulls three pattern groups into shared helpers so a rename or wire-format
tweak edits one file instead of 10+:
- BulkAction enum in doctate-common replaces the "close"/"analyze"/"reset"
string literals that were duplicated between the bulk handler and 3
attack/CSRF test files. FromStr preserves the exact "Unbekannte Aktion"
error shape; the handler match is now exhaustive over the enum.
- join_url helper in doctate-common absorbs 7 identical
trim_end_matches+format! call sites across client-core, client-desktop,
server/transcribe (ollama, whisper), and server/analyze (llm).
- server/tests/common/artefacts.rs re-exports ONELINER_FILENAME
(doctate-common), DOCUMENT_FILE + ANALYSIS_INPUT_FILE
(doctate-server::analyze), and CLOSE_MARKER (doctate-server::paths) so
test files reference the canonical name instead of inlining literals.
Also lifts client-desktop local duplication:
- paths.rs: project_path helper collapses 4 identical ProjectDirs chains
- main.rs: or_die helper replaces 4 eprintln!+exit(1) blocks
- app.rs: named RecordingContext struct replaces (Uuid, String) tuple
at 3 sites around the ffmpeg-flush finalization path
Verification: 397 tests pass (baseline was 389; +8 for new unit tests on
BulkAction and join_url), 0 failed, 4 ignored (unchanged). clippy clean.
Finishes the lift of shared helpers into `tests/common/`. Covered:
- health, web, oneliners_api, upload (31 tests; upload exercises the
new `multipart_upload_body` helper driven by doctate_common field
constants)
- sse_integration, sse_cleanup, transcribe, oneliner_heal_decoupled,
silent_case_empty, failed_only_case_empty_oneliner,
transient_failure_retries (24 tests)
Also applied cargo fmt across the test tree and fixed one clippy
needless_borrows_for_generic_args warning in analyze_test.
All 387 tests pass; 3 ignored (as before).
Side effect: health_test previously used a hardcoded `/tmp/doctate-test`
data path, which parallel `cargo test` runs could collide on. The
migration replaces it with the common unique-tmpdir pattern, removing
a latent flake.
Lift duplicated test-harness code (User factories, TestConfig builder,
login/CSRF flow, form/json helpers, case-directory seeding, URL path
builders) into `server/tests/common/` so future API changes touch one
file instead of every integration test.
Migrated batches: csrf_attack, auth, login, magic_link, security_headers
(28 tests); analyze, case_page, delete_recording, close_watermark,
retention_sweep (67 tests). All 95 tests still green.
Net line delta across these 10 files: +1113 / -1896 (~780 lines removed),
plus ~500 lines of new shared infrastructure under tests/common/.
Adds the csrf_token hidden field to every POST form in the web UI
(logout, bulk, purge-closed, close, reopen, analyze, reset,
delete-recording). Login stays without a token — SameSite=Strict
already blocks the relevant attack shapes and forced-login CSRF
has no impact here.
Shape:
- New askama macro partials/csrf_field.html renders the hidden
input; 3 templates import + {% call csrf::field(csrf_token) %}
inside each <form method="POST">.
- AuthenticatedWebUser carries csrf_token (cloned out of the
session once during extraction) so render handlers don't need a
second store lookup.
- 3 template structs (MyCasesTemplate, CasePageTemplate,
CaseRecordingsTemplate) gain the field; render sites pass
user.csrf_token through.
Safety-net tests:
- Each rendered page must contain the exact session csrf_token in
a hidden input — catches anyone adding a new form without the
macro.
- Happy-path round-trip: fetch page, parse token from HTML, POST
with token → 303. Catches drift between rendered and accepted
token formats.
Introduces CsrfForm<T>: a FromRequest extractor that looks up the
session's csrf_token and constant-time-compares it to a csrf_token
field in the form body. Drop-in replacement for Form<T> on every
state-changing /web/ POST handler. Missing or expired session →
redirect to /web/login; malformed body → 400; token mismatch → 403.
WebSession carries csrf_token, minted alongside the session token at
login and magic-link consume. Not rotated per request — rotation
would break multi-tab use and buys little over SameSite=Strict
cookies.
Handlers: bulk, purge-closed, reset, close, reopen, analyze,
delete-recording, logout all now require CsrfForm<_>. Login stays
unprotected (SameSite=Strict alone is sufficient — forced-login CSRF
has no impact on this codebase). /api/... is header-auth, exempt.
Constant-time compare via subtle::ConstantTimeEq avoids timing
oracles on the token.
Template work is NOT in this commit — browser forms still post
without csrf_token, so the live web UI will 403 until Schritt 6
(templates render the hidden field).
Tests: all 12 red specs in csrf_attack_test.rs now green, #[ignore]
removed. Integration tests that POSTed to protected endpoints
switched to a new create_router_and_session_store entrypoint which
returns a handle to the store so tests can read csrf_token for the
active session.
Attaches a SetResponseHeaderLayer stack to create_router_with_state
(not main.rs) so tests observe the same response shape as production.
`if_not_present` mode so per-route overrides (magic.rs already sets
its own Referrer-Policy) are preserved.
Sets: X-Content-Type-Options, X-Frame-Options, Referrer-Policy,
Content-Security-Policy, Permissions-Policy. HSTS is deliberately
omitted until TLS termination is in place — a cached max-age on a
plain-HTTP deployment is irreversible.
CSP uses 'unsafe-inline' for script/style since templates contain
inline scripts; revisit if any user input ever renders unescaped.
Removes #[ignore] from the 10 attack-confirming header tests; two
regression anchors (no-HSTS, no-duplicated magic header) were already
green.
Add details about the `purge-closed` endpoint, including its admin-only
nature and confirmation requirement.
Clarify the admin-only checks for bulk actions, emphasizing that they
are now performed at the entry handler level.
Refine the explanation of the case list display, particularly the
`open_count/total_count` grouping and how closed cases are handled.
Introduce the concept of analysis preview in the case list, explaining
its implementation using CSS `line-clamp` and avoiding server-side
truncation.
Enhance the description of admin
Move admin checks from individual bulk actions to the main handler for
`bulk.rs` and `case_actions.rs`. This consolidates the authorization
logic for these sensitive operations.
Additionally, refine the HTML template to conditionally render bulk
action UI elements and the purge form only when the user is an admin.
This ensures that non-admin users do not see or have access to these
administrative functions.
Update tests to reflect these changes, ensuring that non-admin users are
correctly rejected for these actions and that the UI is properly hidden.
Replaces `analysis_preview` with `analysis_html` to display the full
markdown-rendered document content instead of just a plain-text preview.
This allows for richer content presentation and enables client-side
expansion for detailed views.
The HTML template has been updated to include a new `.analysis`
container that handles the expand/collapse functionality using
JavaScript. This approach avoids server-side truncation and relies on
CSS for presentation.
Introduce a dedicated case-link class for clickable case titles.
Enhance the line1 div to use flexbox for better alignment of time,
title, and recordings count.
Add a separate link for recordings, improving the structure and user
experience of the case list.
Introduce `count_closed_by_date` and `most_recent_date_label` to
efficiently count closed cases for display purposes when `show_closed`
is false.
Modify `group_by_utc_date` to accept a `closed_extra` map, allowing it
to correctly calculate both open and total case counts per day.
Update the `my_cases.html` template to display the new
`open_count/total_count` format for case groups.
Introduce a `preview_lines` setting in `users.toml` to control the
number of visible lines for the analysis preview in the case list. This
feature extracts the first paragraph of the `document.md` file, strips
markdown formatting, and displays it, capped by the configured
`preview_lines`. The actual line clamping is handled client-side via CSS
`line-clamp`.
The /api/oneliners time window is now read per-user from users.toml
(window_hours, default 72h). Clients no longer carry a window:
client.toml oneliner_window_hours, SyncConfig.window_hours, the
?hours=N query param, and the render_case_list cutoff filter are
gone. ETag suffix keeps the effective hours so an admin edit to
users.toml invalidates client caches on the next request.
OnelinersResponse.window_hours stays in the wire format, but now
exists solely to anchor client reconciliation.
- my_cases.html: `.status-badge.open` from red (#e24a4a) to blue (#4a90e2),
matching case_page.html (which was already blue) and the reopen-btn
accent. Resolves the cross-template inconsistency and frees red for
the Fehler-Badge.
- Both templates: `.status-badge.fehler` from deep red (#c00) to the
established system red (#e24a4a), already used by `.bulk-bar.purge`
and `.delete-btn:hover`.
- Badge label capitalised: "fehler" → "Fehler" (user-facing string
only; CSS class stays `.fehler`).
Before, every Whisper error (5xx, timeout, Minerva down, corrupt audio,
4xx) renamed `<ts>.m4a` to `<ts>.m4a.failed` uniformly, turning transient
outages into permanent sackgassen that only a manual admin reset could
undo. Cases whose every recording was `.m4a.failed` stuck without a
persisted oneliner state; the UI masked them via a fallback in
`compute_oneliner_display`.
The core insight: a recoverable error is not an error. Transient Whisper
failures now leave the audio as plain `.m4a` so the existing page-load
heal (`enqueue_pending_for_user`) re-enqueues it on the next refresh —
which is exactly what happens when Minerva comes back. No new sidecar,
no retry count, no scheduler.
Changes:
- `WhisperError::is_transient` classifies `Http`, 5xx, 408, 429 as
transient; `Io` and other 4xx as permanent.
- Transcribe worker: transient → info log + continue (audio stays .m4a);
permanent → `mark_failed` as before.
- `has_any_transcript` counts `.m4a.failed` as terminal, so
`update_oneliner` runs for failed-only cases and settles
`OnelinerState::Empty` (analogous to the silent-only fix in 4531f85).
- New verdrängender `fehler`-Badge (#c00) in case list and detail when
at least one `.m4a.failed` exists; recording-level message shortened
to plain "Transkription fehlgeschlagen".
Tests:
- New integration: transient 503 leaves `.m4a` intact, heal recovers.
- New integration: `.m4a.failed`-only case settles to
`OnelinerState::Empty` without calling Ollama.
- New unit: `is_transient` table test across relevant status codes.
- New unit: `has_any_transcript` returns true for `.m4a.failed`-only
case and false for pending-only case.
- Existing worker test retargeted from 500 to 400 and renamed; added
companion `worker_leaves_m4a_intact_on_transient_whisper_error`.
Introduce `TranscriptState { Pending, Silent, Content }` in
doctate-common as the canonical three-way state of a recording's
`.transcript.txt` sidecar. Previously each call-site projected the
raw `Option<String>` / `.exists()` onto its own 2-state view and the
projections disagreed: the UI treated a 0-byte silent transcript like
`Content` while the oneliner worker treated it like `Pending`,
leaving silent-only cases stuck on "generiere Titel …" forever with
no persisted oneliner.json.
`update_oneliner` now settles a silent-only case to
`OnelinerState::Empty` when no `Content` transcript exists and no
recording is still `Pending`, so the UI resolves to "unbenannt" and
recovery treats it as terminal.
Single reader: `paths::read_transcript_state`. All call-sites
(scan_recordings, compute_oneliner_display, has_pending_recordings,
all_transcripts_joined, read_recordings, enqueue_pending_for_user,
scan_m4as, cases_needing_oneliner_in, case_recordings.html) go
through the same typed abstraction and must handle `Silent` via
exhaustive match.
Regression tests:
- silent_case_empty_test: heal path settles silent-only case to
Empty without calling Ollama
- case_page_silent_only_shows_empty_not_generating: UI renders
"unbenannt", not "generiere Titel …"
heal_orphans_if_idle called regenerate_missing_oneliners_for_user
inline on the request task, blocking the browser for up to N×60s
when orphans existed (observed with 16 cases after manual cleanup).
Now: CAS on new OnelinerHealBusy flag + tokio::spawn + BusyGuard
reset-on-drop. Handler returns immediately; the existing
OnelinerUpdated SSE event delivers results via the usual reload.
Startup recovery shares the flag so a concurrent page-load during
boot cannot fire a parallel regen loop against Ollama.
Regression test uses wiremock + timeout(150ms) + .expect(5) to
verify both non-blocking return and dedup of parallel invocations.
Brummel