@cbcruk/vision-ocr

CLI and Node.js library for image OCR using the macOS Vision Framework.

Uses node-swift to call Swift code directly from Node.js.

Features

• Free - no API costs
• Offline - no network required
• Fast - native performance
• Korean and English support

Requirements

• macOS 13.0+
• Swift 5.9+
• Node.js 18+

Installation

As a library

npm install @cbcruk/vision-ocr

As a CLI

# Global install
npm install -g @cbcruk/vision-ocr

# Or use without installing
npx @cbcruk/vision-ocr

CLI Usage

# Extract text from clipboard image (also copies result to clipboard)
vision-ocr

# Extract text from a file
vision-ocr screenshot.png

# Output to stdout only, without copying to clipboard
vision-ocr --no-copy

# Pipe to a file
vision-ocr screenshot.png --no-copy > output.txt

API Usage

import {
  recognizeText,
  recognizeTextFromFile,
  recognizeTextFromClipboard,
} from '@cbcruk/vision-ocr'

// From raw image buffer
const text = recognizeText(imageBuffer)

// From file path
const text = recognizeTextFromFile('/path/to/image.png')

// From clipboard
const text = recognizeTextFromClipboard()

License

MIT