Remove emoji from a String in Swift

Question

How can emojis also be filtered out of string using the following extension?

extension String {
    func removeCharacters() -> String {
        let removeCharacters: CharacterSet = [" ", ";", ",", ":", ".", "!", "%", "$", "&", "#", "*", "^", "@", "(", ")", "/"]
        let filteredString = self.unicodeScalars.filter { !removeCharacters.contains($0) }
        return String(String.UnicodeScalarView(filteredString))
    }
}

I understand isEmojiPresentation == false property can be used, just not sure how to add it to the function. To clarify all emoji should be removed from strings passed through the extension.

Answer 1

I wrote this Swift 5 extension that removes emoji characters:

extension String {
    func withoutEmoji() -> String {
        filter { $0.isASCII }
    }
}

You can run this unit test to see if it works:

func test_removingEmoji_shouldReturnStringWithoutEmoji() {
        // Given
        let inputText = "Hell☺️‍‍‍o World❤️"
        
        // When
        let emojilessText = inputText.withoutEmoji()
        
        // Then
        let expectedString = "Hello World"
        XCTAssertEqual(emojilessText, expectedString, "\(String(describing: asData(emojilessText))) should be equal to \(String(describing: asData(expectedString)))")
    }
    
// Helper function that returns the string as Data. 
// Reveals any invisible characters that might have remained in the string. 
// This can help you to understand why two strings might not be considered equal, even if they look the same.
private func asData(_ text: String) -> NSData? {
    text.data(using: .utf8) as? NSData
}

Limitation: As "hotdougsoup.nl" pointed out in the comment, this also removes international characters.

---

Alternative solution

By building upon Leo Dabus great answer, I wrote this small extension that excludes emoji characters to the visible eye. However, the unit test revealed that some invisible characters remained. That's why I decided to go with the shorter snippet above (.isASCII).

// This does not remove all characters reliably. Use the solution above.
extension String {
    func withoutEmoji() -> String {
        unicodeScalars
            .filter { (!$0.properties.isEmojiPresentation && !$0.properties.isEmoji) || $0.properties.numericType == .decimal }
            .reduce(into: "") { $0 += String($1) }
    }
}

Answer 2

You need to access the Unicode.Scalar's property properties. Btw filter on a UnicodeScalarView already returns a UnicodeScalarView:

extension String {
    func removeCharacters() -> String {
        let removeCharacters: CharacterSet = .init(charactersIn: " ;,:.!%$&#*^@()/")
        return .init(
            unicodeScalars.filter {
                !removeCharacters.contains($0) &&
                $0.properties.isEmojiPresentation == false
            }
        )
    }
}

---

edit/update:

The reason a digit would return true for isEmoji is explained in the docs isEmoji

The final result is true because the ASCII digits have non-default emoji presentations; some platforms render these with an alternate appearance.

You can avoid removing them from your string adding an extra condition to your filter method:

extension RangeReplaceableCollection where Self: StringProtocol {
    var removingEmoji: Self  {
        filter { !($0.unicodeScalars.first?.properties.isEmoji == true && !("0"..."9" ~= $0)) }
    }
}

---

let noEmoji = "abc123".removingEmoji

noEmoji  // "abc123"

---

Expanding on that:

extension Character {
    var isEmoji: Bool { unicodeScalars.first?.properties.isEmoji == true && !isDigit }
    var isDigit: Bool { "0"..."9" ~= self }
    var isNotEmoji: Bool { !isEmoji }
}

---

extension RangeReplaceableCollection where Self: StringProtocol {
    var removingEmoji: Self  { filter(\.isNotEmoji) }
    var emojis: Self  { filter(\.isEmoji) }
}

---

Playground testing:

let notEmoji = "abc✈️123".removingEmoji
notEmoji  // "abc123"
let emojis = "abc✈️123".emojis
emojis  // "✈️"

Answer 3

In addition to the previous responses, it's important not to overlook the ZWJ (Zero Width Joiner) sequence in emojis. Failing to properly remove this can lead to the accumulation of ZWJ codes, so "\u{200D}" should also be removed correctly.

Exactly, it's like this.

func withoutEmoji() -> String {
    unicodeScalars
        .filter { (!$0.properties.isEmojiPresentation && !$0.properties.isEmoji && $0 != "\u{200D}") || $0.properties.numericType == .decimal }
        .reduce(into: "") { $0 += String($1) }
}