Textify and Latexify every part of your screen

A OCR toolbox integrated in your mac

Preview

Dark	Light
...	...

Features

• 🔍 Advanced OCR Capabilities

  • Text extraction from any part of your screen
  • LaTeX equation recognition and conversion with Nougat model
  • Support for multiple OCRS models and Nougat models.

• ⚡ Quick Access

  • Menubar integration for easy access
  • Global keyboard shortcuts for instant capture
  • Custom shortcuts for specific OCR configurations

• 🎨 User Interface

  • Clean and modern interface in the menubar
  • Dark and light theme support
  • Hoverlay panel (spotlight style) to choose model to use for OCR

• 🛠️ Customization

  • Configurable OCR settings
  • Customizable keyboard shortcuts
  • Auto-start option
  • Sound feedback toggle

• 📋 Clipboard Integration

  • Automatic clipboard copying
  • Quick copy buttons for results

• 📝 History Management

  • Screenshot history tracking
  • Search through past OCR results
  • Option to disable history for privacy

• 🔒 Privacy and Security

  • Local processing of screenshots
  • Optional history disable feature

Installation

Getting Started

Prerequisites

• Bun - Fast all-in-one JavaScript runtime & package manager
• uv - Fast Python package installer
• Rust - For Tauri's backend

Development Setup

1. Install the project dependencies for javascript

bun install

2. Python Sidecar Setup:

# Navigate to src-python directory and create a virtual environment
cd src-python
# Create a virtual environment
uv venv
# Activate the virtual environment
source .venv/bin/activate
# Sync the dependencies
uv sync

The Python sidecar (ocr_mlx) is packaged using box-packager, which:

• Takes the entry point defined in pyproject.toml (ocr_mlx.endpoint:main)
• Bundles all dependencies into a single executable
• Places the executable in src-tauri/binaries with platform-specific naming
• Uses PyApp to bootstrap the Python environment at runtime

3. Build the Python Sidecar:

# This command will:
# 1. Package the Python app using box-packager
# 2. Copy the executable to src-tauri/binaries
# 3. Name it appropriately for your platform
bun run python:package:build

# Initialize the packaged environment
bun run python:package:reset

4. Start Development Server:

# This will:
# 1. Start the Svelte dev server
# 2. Launch the Tauri window
# 3. Initialize the Python sidecar for OCR
bun run tauri dev

Building for Production

# Build the complete application:
# 1. Checks if Python sidecar needs rebuilding
# 2. Rebuilds sidecar if needed
# 3. Builds the Tauri application
bun run tauri build

The built application will be available in src-tauri/target/release.

Development Workflow

Frontend Development (SvelteKit + Tauri)

• src/ - SvelteKit frontend code
• src-tauri/ - Tauri backend code
• Hot reload enabled for both frontend and Rust changes

Python Sidecar Development

• src-python/ - Python OCR service code
• Development mode: bun run python:package:dev
• API development:

  # Generate OpenAPI specs from Python service
  bun run python:package:generate-openapi
  
  # Generate TypeScript client from OpenAPI specs
  bun run python:package:generate-client

Asset Generation

• bun run icon:generate - Generate app icons from SVG
• bun run icon:generate-tray - Generate animated tray icons

Development Tools

• SvelteKit 5 with Runes for reactive frontend
• Tauri v2 for native capabilities
• Tailwind CSS + DaisyUI for styling
• MLX-Nougat for efficient OCR processing
• SQLite with Drizzle ORM for data persistence
• TypeScript for type safety
• Python sidecar with FastAPI for OCR service

Note: The Python sidecar is automatically rebuilt when changes are detected in the src-python directory. The build process is managed by the pretauri script.

Contributing

Contributions are always welcome!

See contributing.md for ways to get started.

Roadmap

• Enhanced OCR Capabilities

  • Support for handwritten text recognition
  • Table structure recognition and export to Excel/CSV
  • Chemical formula recognition
  • Multi-language support with language auto-detection

• Unified processing backend

• User Experience

  • Interactive tutorial for new users
  • Customizable OCR region presets

• Platform Support

  • Windows support
  • Linux support

Authors

• @julienblanchon

Related

This project is heavily inspired by the following commercial projects:

• Mathpix
• TextSniper

Acknowledgements

• Tauri v2: A framework for building web based desktop apps using Rust.
• tauri-toolkit: A tauri plugin toolkit for menubar apps.
• JacobBolda: A youtube channel with a lot of tauri content
• MrJakob: A youtube channel with a lot of tauri content
• OCRS: A lightweight OCR library project.
• mlx-nougat: Nougat implementation for MLX.
• nougat: Facebook Nougat OCR.
• Fluent Icons: A library of icons for Windows (use for the tray icon)