# Changelog

All notable changes to finrod are documented here.

Format: [Keep a Changelog](https://keepachangelog.com/en/1.1.0/).
Versioning: [Semantic Versioning](https://semver.org/spec/v2.0.0.html).

## [Unreleased]

## [0.5.0] - 2026-07-05

### Changed
- **All browser dependencies are now vendored into the image — nothing loads from a CDN at runtime.** A new `frontend/` esbuild step (run in a Node stage of the Dockerfile) bundles CodeMirror into a single self-contained `static/vendor/codemirror.js` and copies cropperjs + Tesseract.js (engine, wasm core, and the Dutch `nld` language pack) into `static/vendor/`. The page imports only from `/static/vendor/…`. Google Fonts (Material Symbols, Cinzel) is the one remaining external `<link>`; it degrades gracefully and isn't in the failing path.

### Fixed
- Empty language and model dropdowns whenever esm.sh was degraded. The editor's top-level `import`s pulled `@codemirror/*` from esm.sh, whose transitive graph contained an **unpinned** `@codemirror/view@^6.27.0` range URL. During esm.sh's storage outage that range returned an error body as `text/plain`, so the browser blocked the module for a disallowed MIME type — and a single failed top-level import aborts the whole `<script type="module">`, so `fillLangs()`/`loadModels()` never ran despite a healthy backend. Vendoring removes every runtime CDN import, eliminating the failure mode entirely.

### Removed
- Vim editing mode (`@replit/codemirror-vim`) and its `Vim.mapCommand` H/L bindings. The document-level Shift-H/L focus cycling across the dropdowns/buttons is retained. One fewer dependency to vendor.

## [0.3.0] - 2026-06-09

### Added
- "Scan image" button next to the input pane: opens the phone's rear camera (or a file picker on desktop), runs OCR locally in the browser via Tesseract.js (loaded from esm.sh on first use), and replaces the editor contents with the recognised text. Language follows the "From" dropdown via an ISO 639-2 mapping; "Auto-detect" loads a common European set (`eng+por+deu+nld`). Progress status shows beneath the input ("Loading OCR engine…" → "OCR 40%" → "OCR done in 3.2s"). Zero image-size cost — same CDN pattern as CodeMirror.
- Images are downscaled to 1600 px on the long edge before OCR. A 12 MP phone photo (~4000 px) drops to ~1.3 MP, making Tesseract 4-8× faster with no real accuracy loss for printed text. Smaller-than-1600 px inputs are passed through unchanged.
- Adaptive thresholding (Bradley's method with integral images) binarises the downscaled image before OCR — drops hallucinated text from shadows, gradients, and surface texture on phone photos. Output is pure black-on-white, much closer to the scanned-page input Tesseract was trained on.
- Per-word confidence filter (≥ 60) drops the low-confidence noise Tesseract still emits. Re-assembles lines from kept words; status line reports how many words were dropped.
- Crop & rotate modal between image pick and OCR. After choosing an image, a modal pops up with cropperjs's drag-handle crop UI and Rotate left / Rotate right / Reset buttons. EXIF orientation is honoured automatically. Cancel / `Esc` / clicking the backdrop aborts; `Enter` accepts. Library lazy-loaded from esm.sh (no bundling).

### Changed
- "Scan image" button replaced with a floating action button at the bottom-right of the viewport, using a Material Symbols Outlined camera glyph (`photo_camera`) for consistent rendering across platforms (was a small text button in the input pane header).
- "Translate" inline button replaced with a primary FAB using the `translate` Material Symbols icon, stacked directly under the camera FAB at the bottom-right corner (closest to the thumb on phones). The camera FAB is now the *secondary* in the stack, just above. Both 56 × 56 px, same accent fill. The translate FAB gains a pulsing-ring `busy` animation during a running translation; `goBtn.textContent` swap is gone (was meaningless for an icon-only button). `body` got `~9 rem` of bottom padding so content scrolls clear of the stack.
- FAB stack rides above the on-screen keyboard on mobile via `interactive-widget=resizes-content` in the viewport meta tag (handled by Chrome on Android).
- Layout: From / To dropdowns now occupy 2 columns at the top (was 3 columns including Model).
- Model selector moved into a slide-out side drawer (hamburger menu in the top-left). Drawer opens from the left with a backdrop, closes via the X button, backdrop click, or `Esc`. Honours Android safe-area insets.
- Page header (`h1 finrod` + tagline) removed from the main view to recover vertical space. The "Finrod" wordmark now lives only inside the side drawer, set in Cinzel (classical-inscription serif, loaded from Google Fonts) alongside the Star of Fëanor mark.
- Default source language is now Dutch (was Auto-detect). Target stays English.
- Crop modal:
  - No default crop box. User drags on the image to draw the region; falls back to full image if they confirm without drawing.
  - Buttons are now Material Symbols icons: `rotate_left`, `rotate_right`, `invert_colors` (new — see below), `check`. Cancel kept as text. The `check` (Use this) button is larger and more prominent. The Reset button was removed (rarely useful with no default box).
  - Inner padding bumped to 2 rem so cropperjs's corner handles never sit on the right/bottom edges where Android's edge-swipe back gesture fires. iOS safe-area insets honoured on the backdrop and FAB.
  - Header text is now a hint ("Drag on the image to select the text region") instead of a title.

### Added
- Camera viewfinder modal on mobile (via `navigator.mediaDevices.getUserMedia` with `facingMode: environment`) — replaces the `<input type=file capture>` flow. Tapping the shutter goes directly to the crop modal, skipping the OS-level "retake / use" confirm step. Falls back to the file picker if camera access is denied or unavailable; a "Pick from file" icon (`photo_library`) inside the camera modal also routes to the picker on demand.
- Invert-colours toggle (`invert_colors`) in the crop modal. Tap to flip light-on-dark text into the dark-on-light polarity Tesseract expects. The grayscale buffer is inverted before adaptive thresholding so the binarised output is correct.
- Mobile-first modal layout (≤ 640 px): the crop and camera modals now fill the viewport (no backdrop padding, no border-radius), buttons hug the screen edges in the footer, and the camera video occupies the entire body. Body still has ~20 px horizontal padding so cropper's corner handles stay outside Android's edge-swipe back-gesture zone. iOS notch / home-bar insets honoured.

### Fixed
- "Hot: …" status no longer claims "unloads in X" when the server's `keep_alive` is `-1` (Ollama returns an absurd year-4001 `expires_at`). Threshold: any expiry >7 days drops the suffix and just shows the model name.

## [0.2.0] - 2026-06-09

### Added
- Vim bindings in the input via CodeMirror 6 + `@replit/codemirror-vim` (loaded from esm.sh, no build step). Supports `i`/`Esc`, `hjkl`, word/line motions, `dd`/`yy`/`p` (vim's internal register), undo/redo, `/` search, `:` command line. Auto-disabled on touch devices (detected via `(pointer: coarse)`) where modal editing fights the virtual keyboard.
- `Shift-H` / `Shift-L` cycle focus across the page controls (source → target → model → editor → button) like `Shift-Tab` / `Tab`. Also overridden in vim normal mode (replacing vim's viewport-top/bottom default) so the same shortcut works to leave the editor.
- On page load, the model dropdown defaults to whichever model Ollama already has hot — skips the 5-15 s disk-reload pain on the first translation. After load it follows the user's manual selection.

### Changed
- Editor font-size pinned to `1rem` to match the output pane (CodeMirror 6 ships smaller by default).

## [0.1.0] - 2026-06-09

First release. Stateless translator UI that proxies streaming requests to
Ollama on radagast.

### Added
- FastAPI backend with three endpoints:
  - `GET /api/models` — filtered (`gemma,aya` by default) list from Ollama's `/api/tags`
  - `POST /api/translate` — streams Ollama's NDJSON `/api/generate` response back to the browser
  - `GET /api/loaded` — proxies Ollama's `/api/ps` for the currently-loaded model
  - `GET /healthz` — for Kubernetes probes
- Single-page UI:
  - Source / target / model dropdowns (30+ languages, "Auto-detect" source)
  - Token-by-token streaming into the output pane with a blinking caret
  - Live status line: spinner while waiting for first token, pulsing dot during streaming, elapsed time and token count
  - Glowing border on the output pane during streaming
  - Loaded-model indicator (`● ` prefix in dropdown, "Hot: … — unloads in Xm" status line); refreshes on page load, on translation completion, and 500 ms after typing pauses
  - ⌘/Ctrl-Enter to translate from the input
  - Dark mode, inline SVG favicon (Star of Fëanor)
- Dockerfile: two-stage `uv` build, slim Python 3.13 runtime
- `build.sh`: multi-arch (amd64 + arm64) manual push to Scaleway, tags `:latest`, `:<git-short-sha>`, and `:<pyproject-version>`
- Cluster deploy via the `valinor` repo at `apps/finrod/` (bjw-s app-template, internal-only ingress at `finrod.jpnadas.xyz`)

### Fixed
- Lock against public PyPI explicitly — the workstation's global `uv` config had pointed `uv.lock` sources at the dexterenergy private artifact registry, which 401'd inside Docker