TRex - Copy the Uncopyable
macOS / Utilitaires
TRex makes OCR easy and accessible on a Mac.
But what is OCR anyway? Imagine you have a PDF file or a Web page where you can't select the text, image with text on it, or even a YouTube video. Forget retyping it manually; with TRex, you can extract text from anywhere, and it is as easy as taking a screenshot.
TRex can help you to copy text from:
- A PDF with non-selectable text
- A screenshot
- YouTube video
- Zoom call screen sharing
- ... and more!
If you can see it on your screen - TRex can copy it.
Every feature in TRex works offline, no internet is needed:
- Text recognition (OCR)
- Read QR Code and barcodes
- Handy macOS menu bar app, with option to hide it
- Configurable global shortcuts
- Configurable recognition language
- URL scheme support trex:\\capture
Quoi de neuf dans la dernière version ?
New Features
LLM Integration
- AI-powered OCR: Use GPT-4o, Claude, Ollama, or Apple Intelligence for vision-based text recognition
- Smart post-processing: Automatically format, correct, or transform captured text with LLM
- Separate provider configuration: Choose different providers for OCR vs post-processing
- On-device processing: Apple Intelligence runs locally with no API calls (macOS 15.1+)
- Network-aware fallback: Automatically falls back to built-in OCR when offline
Multi-Region Capture
- Batch selection: Select multiple screen regions in a single capture session
- Combined results: OCR results from all regions are merged together
- Optional LLM processing: Post-process combined text with AI
- Up to 50 regions: Capture complex layouts in one go
Watch Mode
- Continuous monitoring: Monitor a screen region and automatically capture when content changes
- Interactive region selection: Draw regions with a visual overlay
- Floating control panel: Start, pause, and reselect regions on the fly
- Multiple output modes: Append to clipboard, write to file, or stream notifications
- Smart detection: SHA256-based change detection with configurable polling interval
- URL scheme support: Trigger watch mode via `trex://watch`
Capture History
- Persistent history: All OCR captures saved with thumbnails
- Quick access: Last 5 captures available in menu bar submenu
- History window: Browse all captures via `trex://showHistory`
- Configurable storage: Set max entries (1–10,000)
- One-click copy: Quickly copy any previous capture to clipboard
Table Detection
- Vision-based detection: Uses macOS 26+ document recognition for table extraction
- LLM fallback: Falls back to AI-based table detection on older systems
- Multiple formats: Export tables as Markdown, CSV, or plain text
- Menu bar toggle: Quick access to table detection settings
Bug Fixes
CLI Improvements
- Fixed CLI not exiting after stdin EOF (#61)
- Added back Intel Support
- Improved temp file handling with UUID-based paths
Language Recognition
- Added Turkish (tr-TR) to Vision OCR fallback language sets (#62)
- Fixed language code normalization to handle underscore format (#65)
UI Fixes
- Fixed Quick Setup content overflow with scrollable language grid (#63)
- Added confirmation dialog when hiding menu bar icon with recovery instructions (#60)
