A modern Swift wrapper for Tesseract OCR, providing a clean API for text recognition in iOS and macOS applications.

Features

• 🚀 Pure Swift API with async/await support
• 📱 iOS 16+ and macOS 13+ support
• 🌍 100+ language support via Tesseract
• 📦 Distributed as Swift Package Manager
• 🎯 Pre-built XCFrameworks for easy integration
• 💾 Built-in language data downloader
• 🖼 Direct CGImage support

Requirements

• macOS: Apple Silicon (ARM64) Macs only
• iOS: Currently not supported (XCFrameworks need iOS slices)
• Architecture: ARM64 only (Intel x86_64 not supported)

Installation

Swift Package Manager

Add TesseractSwift to your project:

dependencies: [
    .package(url: "https://github.com/amebalabs/TesseractSwift.git", from: "1.0.0")
]

Usage

Basic Text Recognition

import TesseractSwift

// Initialize engine
let engine = TesseractEngine(dataPath: "/path/to/tessdata")
try engine.initialize(language: "eng")

// Recognize text from CGImage
let text = try engine.recognize(cgImage: image)
print("Recognized text: \(text)")
print("Confidence: \(engine.confidence())%")

Download Language Data

let downloader = LanguageDownloader.shared
let language = TesseractLanguage(code: "fra", name: "French", script: "Latin", fileSize: nil)

// Download language data
try await downloader.downloadLanguage(language, to: tessdataURL) { progress in
    print("Download progress: \(progress * 100)%")
}

Available Languages

// Get list of common languages
let languages = LanguageDownloader.commonLanguages

// Check downloaded languages
let downloaded = downloader.downloadedLanguages(in: tessdataURL)

Page Segmentation Modes

// Set recognition mode based on your content
engine.setPageSegmentationMode(.auto)        // Automatic detection
engine.setPageSegmentationMode(.singleLine)  // Single line of text
engine.setPageSegmentationMode(.singleBlock) // Uniform block of text

Language Support

TesseractSwift supports 100+ languages. Common languages include:

• English (eng)
• Spanish (spa)
• French (fra)
• German (deu)
• Italian (ita)
• Portuguese (por)
• Russian (rus)
• Chinese Simplified (chi_sim)
• Chinese Traditional (chi_tra)
• Japanese (jpn)
• Korean (kor)
• Arabic (ara)
• Hindi (hin)

Integration with TRex

This package was created to provide Tesseract OCR support for TRex, extending its language support beyond the built-in Vision framework.

Example Integration

import TesseractSwift

class TesseractOCREngine: OCREngine {
    private let engine: TesseractEngine

    init(dataPath: String) {
        self.engine = TesseractEngine(dataPath: dataPath)
    }

    func recognizeText(in image: CGImage) async throws -> String {
        if !engine.isInitialized {
            try engine.initialize(language: "eng")
        }
        return try engine.recognize(cgImage: image)
    }
}

Building from Source

1. Clone the repository
2. Ensure you have the pre-built XCFrameworks in Binaries/
3. Open in Xcode or build via command line:

swift build
swift test

Requirements

• iOS 16.0+ / macOS 13.0+
• Swift 5.9+
• Xcode 15.0+

License

This package includes Tesseract OCR (Apache 2.0) and Leptonica (BSD-style license).