PDF Text Redaction using Regex in Swift for iOS

1
//
2
//  Copyright © 2021-2025 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 PSPDFKitUI
10

11
class RedactTextUsingRegexExample: Example {
12

13
    override init() {
14
        super.init()
15

16
        title = "Redact Text Using Regular Expressions"
17
        contentDescription = "Shows how to redact URLs using a regex pattern."
18
        category = .documentProcessing
19
        priority = 16
20
    }
21

22
    override func invoke(with delegate: ExampleRunnerDelegate) -> UIViewController? {
23
        let document = AssetLoader.document(for: .web)
24
        let status = StatusHUDItem.indeterminateProgress(withText: "Processing...")
25
        status.push(animated: true, on: delegate.currentViewController?.view.window)
26

27
        // The regex pattern for URLs.
28
        let urlPattern = #"[-a-zA-Z0-9@:%._\+~#=]{1,256}\.[a-zA-Z0-9()]{1,6}\b([-a-zA-Z0-9()@:%_\+.~#?&//=]*)"#
29
        let urlRegularExpression = try! NSRegularExpression(pattern: urlPattern)
30

31
        // Loop through all the pages in the document to find URLs.
32
        // In production we recommend doing this on a utility queue.
33
        for pageIndex in 0..<document.pageCount {
34
            if let textParser = document.textParserForPage(at: pageIndex) {
35
                textParser.words.forEach { word in
36
                    let wordString = word.stringValue
37
                    let range = NSRange(wordString.startIndex..<wordString.endIndex, in: wordString)
38
                    // Redact all the words that match the regex.
39
                    let isValidURL = urlRegularExpression.numberOfMatches(in: wordString, options: [], range: range) > 0
40
                    if isValidURL {
41
                        let redaction = self.createRedactionAnnotationFor(word: word, pageIndex: pageIndex)
42
                        document.add(annotations: [redaction])
43
                    }
44
                }
45
            }
46
        }
47

48
        DispatchQueue.global(qos: .default).async {
49
            // Use Processor to create the newly redacted document.
50
            let processorConfiguration = Processor.Configuration(document: document)!
51
            processorConfiguration.applyRedactions()
52

53
            let redactedDocumentURL = FileHelper.temporaryPDFFileURL(prefix: "redacted")
54
            let processor = Processor(configuration: processorConfiguration, securityOptions: nil)
55
            try? processor.write(toFileURL: redactedDocumentURL)
56

57
            DispatchQueue.main.async {
58
                status.pop(animated: true)
59
                // Instantiate the redacted document and present it.
60
                let redactedDocument = Document(url: redactedDocumentURL)
61
                let pdfController = PDFViewController(document: redactedDocument)
62
                delegate.currentViewController!.navigationController?.pushViewController(pdfController, animated: true)
63
            }
64
        }
65
        return nil
66
    }
67

68
    // MARK: Private
69
    private func createRedactionAnnotationFor(word: Word, pageIndex: PageIndex) -> RedactionAnnotation {
70
        let redactionRect = word.frame
71
        let redaction = RedactionAnnotation()
72
        redaction.boundingBox = redactionRect
73
        redaction.rects = [redactionRect]
74
        redaction.color = .orange
75
        redaction.fillColor = .black
76
        redaction.overlayText = "REDACTED"
77
        redaction.pageIndex = pageIndex
78
        return redaction
79
    }
80
}