OCR PDF using Swift for iOS

1
//
2
//  Copyright © 2020-2025 PSPDFKit GmbH. All rights reserved.
3
//
4
//  The Nutrient sample applications are licensed with a modified BSD license.
5
//  Please see License for details. This notice may not be removed from this file.
6
//
7

8
import PSPDFKit
9
import PSPDFKitOCR
10
import PSPDFKitUI
11

12
class OCRExample: Example {
13

14
    private var document: Document!
15
    private var tabbedController: OCRTabbedViewController!
16
    private var progress: Foundation.Progress!
17

18
    override init() {
19
        super.init()
20

21
        title = "OCR"
22
        contentDescription = "Performs optical character recognition (OCR) on the document."
23
        category = .componentsExamples
24
        priority = 1
25
    }
26

27
    override func invoke(with delegate: ExampleRunnerDelegate) -> UIViewController? {
28
        document = AssetLoader.document(for: "Remote Work.pdf")
29

30
        let ocrButton = UIBarButtonItem(title: "Perform OCR", style: .plain, target: self, action: #selector(performOCRButtonTapped(_:)))
31

32
        tabbedController = OCRTabbedViewController()
33
        tabbedController.delegate = self
34
        tabbedController.documents = [document]
35
        let pdfController = tabbedController.pdfController
36
        pdfController.navigationItem.setRightBarButtonItems([ocrButton, pdfController.activityButtonItem, pdfController.annotationButtonItem, pdfController.searchButtonItem], animated: false)
37
        return tabbedController
38
    }
39

40
    @objc func performOCRButtonTapped(_ sender: UIBarButtonItem) {
41
        let alert = UIAlertController(title: "OCR", message: "This will convert inaccessible text to real text objects. The operation can take a few seconds.", preferredStyle: .alert)
42
        let continueAction = UIAlertAction(title: "Continue", style: .default) { _ in
43
            self.performOCR(sender)
44
        }
45
        alert.addAction(continueAction)
46
        alert.addAction(UIAlertAction(title: "Cancel", style: .cancel, handler: nil))
47
        alert.preferredAction = continueAction
48
        tabbedController.present(alert, animated: true, completion: nil)
49
    }
50

51
    func performOCR(_ sender: UIBarButtonItem) {
52
        sender.isEnabled = false
53

54
        let controller = tabbedController!
55

56
        let processorConfiguration = Processor.Configuration(document: document)!
57
        processorConfiguration.performOCROnPages(at: IndexSet(integer: 0), options: ProcessorOCROptions(language: .english))
58

59
        let processor = Processor(configuration: processorConfiguration, securityOptions: nil)
60
        controller.processor = processor
61
        processor.delegate = self
62
        let ocrURL = FileHelper.temporaryPDFFileURL(prefix: "ocr")
63

64
        let progress = Progress(totalUnitCount: Int64(document.pageCount + 1))
65
        self.progress = progress
66
        let provider = CoordinatedFileDataProvider(fileURL: ocrURL, progress: progress)
67
        let ocrDocument = Document(dataProviders: [provider])
68
        ocrDocument.title = document.title! + " (OCR)"
69
        controller.addDocument(ocrDocument, makeVisible: false, animated: true)
70

71
        DispatchQueue.global(qos: .userInitiated).async {
72
            do {
73
                let startTime = CFAbsoluteTimeGetCurrent()
74
                try processor.write(toFileURL: ocrURL)
75
                progress.completedUnitCount = progress.totalUnitCount
76
                DispatchQueue.main.async {
77
                    controller.visibleDocument = ocrDocument
78
                    sender.isEnabled = true
79

80
                    let diffTime = CFAbsoluteTimeGetCurrent() - startTime
81
                    print("OCR took \(diffTime) seconds")
82
                }
83
            } catch {
84
                print(error)
85
            }
86
        }
87
    }
88
}
89

90
extension OCRExample: ProcessorDelegate {
91
    nonisolated func processor(_ processor: Processor, didProcessPage currentPage: UInt, totalPages: UInt) {
92
        DispatchQueue.main.async {
93
            self.progress.completedUnitCount = Int64(currentPage + 1)
94
        }
95
    }
96
}
97

98
extension OCRExample: PDFTabbedViewControllerDelegate {
99
    func tabbedPDFController(_ tabbedPDFController: PDFTabbedViewController, shouldClose document: Document) -> Bool {
100
        // Don't allow closing original document.
101
        if document == self.document {
102
            return false
103
        }
104
        // Don't allow closing documents where OCR is currently performed.
105
        return document.progress.isFinished
106
    }
107
}
108

109
private class OCRTabbedViewController: PDFTabbedViewController {
110
    var processor: Processor?
111

112
    override func viewDidDisappear(_ animated: Bool) {
113
        super.viewDidDisappear(animated)
114
        processor?.cancel()
115
    }
116
}