Martin Dvorak 756a6257a2 Chat mode, phase 1: grounded streaming chat over your notes

Entry points (no sixth tab): Ask in the Library header (desktop button,
phone icon), Ask about this deck in the deck menu, Ask a question on
study/homework results (compact result travels as context), and a Chat
row in the desktop sidebar. Phones get a full-screen route with back +
New chat; the tab bar steps aside.

- Client-side retrieval: chunk scoped decks (~800 tokens), BM25 against
  the question + recent turns, top chunks under ~7k tokens, each
  labelled with deck + section. Notes come from the deck's stored
  sourceText (already kept for Regenerate); decks without it fall back
  to generated content and the UI says so.
- Stateless server: /api/chat validates against the CHAT_MODELS
  allowlist, trims the history to ~6k tokens, caps context at ~9k and
  streams token/thinking/done/error NDJSON. /api/chat/extract reduces
  attachments to text through the existing vision step.
- UI: user bubbles + full-width assistant text, streaming with stop,
  auto-scroll that pauses with a jump-to-bottom button, safe markdown
  (React nodes, no HTML injection; KaTeX dynamically imported only when
  a message has math), Thinking row for reasoning, copy/regenerate/
  edit-and-resend, model + context pickers (last model remembered),
  composer with visualViewport keyboard handling, safe-area, 16px
  textarea, Enter-to-send on desktop.
- Usage-limit failures keep the draft and show the shared message with
  Settings/Ollama actions; stopping maps the reader abort to a clean
  stopped state; offline disables sending with a reason.
- Settings: Keep notes text for chat toggle; deck menu: Delete notes
  text. RATE_LIMIT_CHAT + CHAT_MODELS in .env.example and README.
- Tests: chat suite (63 checks, stub Ollama) wired into run-all.
  Full set: 12/12 suites, 487 checks. SW -> freezmo-v32.

Also: tests/model-compare.mjs (diagnostic) benchmarks two models on the
same notes (latency, repair rounds, activity types) so the gpt-oss vs
gemma default can be decided on numbers; defaults unchanged.

2026-10-11 18:32:11 +00:00

3.5 KiB

Raw Blame History

Browser suites

End-to-end checks for Freezmo, driven by Playwright against a running production build. Everything lives in this repo (nothing in /tmp), so a reboot cannot wipe the harness.

Requirements

The app is running on http://127.0.0.1:3207 (systemctl --user start freezmo).
A Chromium build. The suites auto-detect the newest Playwright Chromium in ~/.cache/ms-playwright/; override with CHROME=/path/to/chrome.
npm install (dev deps: playwright-core, axe-core).

Running

npm run build && systemctl --user restart freezmo   # fresh build first
npm run test:e2e                 # every suite, in order
npm run test:e2e -- phase1       # only suites whose name contains "phase1"

Each suite prints PASS/FAIL lines and exits non-zero on failure; the runner prints a per-suite summary at the end.

Suites

Suite	Covers
`regression.mjs`	Library, deck card + menu, study feed to results, homework entry, desktop shell + settings/health panel.
`bubbles-test.mjs`	Ice-scene bubbles: stacking/layout contract, keyframes only transform+opacity, rise/sway/scale/fade phases, per-breakpoint counts (5/3/2), phone pacing, reduced motion, tab-hidden parking.
`mobile-test.mjs`	Five-tab bar at 360/390 × EN/CS × dark/light × reduced motion (labels, targets, no hamburger/drawer), the library ⋯ sheet (Import JSON / New folder), the settings sheet (touch targets, key flow, drag-to-close, backdrop), axe contrast audits.
`wizard-test.mjs`	Deck 3-step flow and homework flow, flat action bar, in-page mode switch, the loading lock, abort-on-leave.
`native-test.mjs`	Manifest/PWA head, offline banner + disabled generate, share target GET + POST, Back-gesture layers, undo toast, tab re-tap, haptics, draft restore.
`i18n-test.mjs`	EN/CS dictionary parity, locale switch, Czech desktop shell, Czech phone tab labels and plurals.
`freeze-loading-test.mjs`	The generation loading screen appears immediately, hides the action bar, Cancel returns with the notice.
`review-test.mjs`	Smart Review loading feedback under a stalled request, the session end-to-end, mistakes, and the health-probe cache.
`phase1-test.mjs`	Backup export/import, persistence row, nudge, encrypted sync (enable/join/push/delete + API validation), usage-limit handling (client + a stub Ollama).
`phase2-test.mjs`	Answer history + SRS from a real study run, review/mistakes, stats, exam mode, push reminders hidden without VAPID and validated with it.
`phase3-test.mjs`	Check my work (API + phone/desktop UI), follow-ups, guide step explain, turn-into-a-study-deck.
`chat-test.mjs`	Chat: streaming API (validation, allowlist, context travel, history window, thinking separation, usage-limit mapping), attachment extraction, entry points, streaming/stop, markdown + KaTeX, model/scope pickers, edit/regenerate, offline, reduced motion.

Extra instances

phase1, phase2 and phase3 start their own next start instance with a stub Ollama (or VAPID keys) on a spare port (3208/3209). Each suite checks the port is free first and kills the whole process group on exit, so ports do not leak between suites. Scratch data for the VAPID test lives in tests/.tmp/ (gitignored).

Diagnostics (not part of the run)

node tests/bubble-perf.mjs [label] measures the ice-scene frame cost with a CDP performance trace (rAF frame timing, long tasks, main-thread busy time, style recalcs) on a 4× throttled phone profile and a 2× desktop profile. It always exits 0; the numbers are for tuning the bubble count/speed.

3.5 KiB Raw Blame History Unescape Escape