# Quit + Screenshot Shortcuts

**Date:** 2026-09-23 09:08 UTC  
**Task:** Add Q to quit, S/Shift+S to save screenshot with/without captions, update About dialog shortcuts

## Changes Made

### Rust backend
- **`src-tauri/src/commands/export.rs`** — Added `save_screenshot` Tauri command. Accepts output directory, video title, timestamp label, and PNG byte data. Uses the existing `generate_output_path` and `expand_tilde_path` from `clip_exporter` for consistent filename generation. Creates output directory if it doesn't exist.
- **`src-tauri/src/lib.rs`** — Registered `export::save_screenshot` in the invoke handler.

### Frontend bindings
- **`src/lib/bindings/export.ts`** — Added `saveScreenshot()` function that invokes the new Rust command.

### Screenshot utility
- **`src/lib/utils/screenshot.ts`** (new file) — Canvas-based screenshot capture:
  - Draws the current video frame onto an offscreen canvas at native resolution (`videoWidth` x `videoHeight`)
  - Optionally composites captions using Canvas 2D text APIs, matching the CSS caption style (font, color, background, outline/shadow, word highlighting, position)
  - Scales caption font size proportionally from preview to native resolution
  - Exports as PNG blob, converts to byte array, sends to Rust backend
  - Filename format: `{videoTitle} - {totalMs}ms-screenshot.png`

### VideoPlayer
- **`src/lib/components/VideoPlayer.svelte`** — Added `export function getParsedCues()` to expose parsed VTT cues to the parent component.

### App keybindings
- **`src/App.svelte`** — Added:
  - `Q` — Quit app via `exit(0)` from `@tauri-apps/plugin-process`
  - `S` — Screenshot with captions
  - `Shift+S` — Screenshot without captions
  - `bind:this={videoPlayerRef}` on VideoPlayer for accessing `getParsedCues()`
  - `captureScreenshot(withCaptions)` async function that orchestrates the capture

### About dialog
- **`src/lib/components/AboutDialog.svelte`** — Added Q, S, and Shift+S to the global shortcuts list.

## Verification
- `svelte-check`: 0 errors, 0 warnings
- `vitest run`: 69 tests passed across 10 test files
- `cargo check`: compiles clean

## Lessons Learned
- Canvas 2D text rendering requires explicit scaling when compositing at native resolution vs. CSS preview resolution. Using `videoHeight / clientHeight` as the scale factor ensures caption size matches visually.
- The `toBlob` → `arrayBuffer` → `Array.from(Uint8Array)` pipeline is necessary because Tauri's IPC expects `Vec<u8>` on the Rust side, which maps to `number[]` in TypeScript invoke calls.

## Follow-up Items
- Visual QA recommended to verify caption rendering fidelity on screenshots matches the preview exactly (especially text wrapping and word-level highlighting).
- Could add a brief toast notification on successful screenshot save (currently only logs to console).