TextGrab

Native macOS utility for extracting text from anywhere on the screen.

Tagline: Instant screen-to-text for macOS.

---

textgrab_vid.mp4

Core Workflow

1. Press the global TextGrab shortcut (default: ⌘⇧2)
2. Select a region of the screen
3. TextGrab captures that region
4. Apple Vision performs local OCR
5. Recognized text is copied to the clipboard

Shortcut → Select → Capture → OCR → Clipboard

Features

Capture & OCR

• Global shortcut — Configurable system-wide hotkey for instant capture
• Saved area capture — Secondary shortcut (⌘⌥2 by default) to re-capture the last selected region without re-selecting
• Multi-monitor support — Works across all connected displays
• Three extraction modes:
  • Text — General purpose text extraction with language correction (Fast or Accurate, Fast by default)
  • Code — Optimized for code snippets: rebuilds indentation from layout, joins OCR-split punctuation, always uses Accurate recognition
  • Table — Extracts tabular data as GitHub-flavored markdown: space-aligned and pipe-delimited columns, wrapped-cell and hyphen rejoining, always uses Accurate recognition
• Smart recognition policy — Text mode is user-configurable (Fast/Accurate); Code and Table always run Accurate automatically since Fast misses isolated digits and small cells
• OCR robustness — Auto-retry chain for tricky captures: Accurate retry and image inversion for white text on dark backgrounds (dark-mode IDEs, dark theme apps)
• Pipeline timeouts — Capture, OCR, and AI steps are race-guarded so a hung system call can never leave the app stuck
• Multi-language OCR — Add custom languages (e.g., fr-FR, ja-JP, zh-Hans)

Clipboard & History

• Clipboard history — Optional history with configurable max items (10–200) that survives app relaunches (stored locally, bounded, wiped when disabled)
• One-click recopy — Click any history item to copy it again
• Rich-text tables — Markdown tables are also placed on the clipboard as HTML so pasting into Notes/Pages/Word preserves the table

Text-to-Speech

• Speak captured text — Built-in TTS with play/stop controls in the menu bar

Apple Intelligence (macOS 26+)

• On-device correction — Uses Apple Intelligence to fix OCR errors before copying (runs entirely on-device, no cloud)
• Summarize — Generate a concise summary of captured text with one click
• Compact — Create a condensed version preserving key information
• Privacy-first — All processing happens locally; no data leaves your Mac

Menu Bar Interface

• Polished popover with material cards, a prominent Capture button, and hover highlights
• Custom extraction mode selector (Text/Code/Table) with live switching
• Real-time processing status and inline error reporting
• Selectable last-capture preview, window nudged away from the screen edge

Auto-Updates

• GitHub Releases based — Checks the latest release once a day on launch (opt-out toggle)
• Check Now — Manual check in Settings with live status and semantic version comparison
• Download & install — Downloads the DMG and opens it for drag-to-install; no signing infrastructure required
• Update notifications — Notifies when a newer version is available

Localization

• 5 languages — English (source), Hindi (हिन्दी), French (Français), German (Deutsch), Spanish (Español)
• The UI follows the system language automatically via String Catalogs

Privacy & Permissions

• 100% local processing — No data sent to external servers
• Permission management — Built-in UI for Screen Recording
• Relaunch helper — One-click app restart after granting permissions (spawns a fresh instance)

Settings

• General — Shortcuts, notifications, history
• OCR — Mode, recognition level, language correction, Apple Intelligence, languages
• Permissions — Grant/check Screen Recording
• Advanced — Updates, version info, user guide, links to source/privacy

Goals

• Native Swift/macOS implementation
• Fast screen-to-text workflow
• Local OCR; no cloud service required for the core path
• Menu-bar-first UX
• Multi-monitor support
• Minimal external dependencies
• Strong privacy defaults

Tech Stack

• Language: Swift 5.9+
• UI: SwiftUI + AppKit
• Frameworks: Vision, ScreenCaptureKit, CoreGraphics, CoreImage, Foundation, Carbon (for global shortcuts), UserNotifications, AVFoundation, ServiceManagement, FoundationModels (Apple Intelligence, macOS 26+)
• Architecture: MVVM with @ObservableObject/@StateObject
• Testing: XCTest (unit), XCUITest (UI)
• External dependencies: None

Building

# Debug build
xcodebuild -project TextGrab.xcodeproj -scheme TextGrab -configuration Debug build

# Release archive + DMG (outputs build/TextGrab-1.2.0.dmg)
./scripts/build-release.sh

# Run tests (12 unit + 2 UI)
xcodebuild -project TextGrab.xcodeproj -scheme TextGrab -configuration Debug test

Installation (macOS Gatekeeper)

Since TextGrab is open-source and not notarized, macOS will block it on first run.

Option 1 (Easiest): Right-click TextGrab.app → Select "Open" → Click "Open"

Option 2: Build from source — ./scripts/build-release.sh

Option 3 (Terminal):

xattr -d com.apple.quarantine /Applications/TextGrab.app

Known Issue: Screen Recording Permission After Updates

Locally-built DMGs use ad-hoc signing. Each build gets a unique signature, so macOS treats updates as new apps and requires re-granting Screen Recording permission.

Workaround: In System Settings → Privacy & Security → Screen Recording, toggle TextGrab off then on after updating.

License

MIT — see LICENSE

Security

See SECURITY.md for vulnerability reporting.