Split a String without removing the delimiter in Swift

Question

This might be a duplicate. I couldn\'t find the answer in Swift, so I am not sure.

componentsSeparatedByCharactersInSet removes the delimiter. If you se

Answer 1

This method works on CollectionTypes, rather than Strings, but it should be easy enough to adapt:

extension CollectionType {
  func splitAt(@noescape isSplit: Generator.Element throws -> Bool) rethrows ->  [SubSequence] {
    var p = startIndex
    return try indices
      .filter { i in try isSplit(self[i]) }
      .map { i in
        defer { p = i }
        return self[p..<i]
      } + [suffixFrom(p)]
  }
}

extension CollectionType where Generator.Element : Equatable {
  func splitAt(splitter: Generator.Element) -> [SubSequence] {
    return splitAt { el in el == splitter }
  }
}

You could use it like this:

let sentence = "Hello, my name is oisdk. This should split: but only at punctuation!"

let puncSet = Set("!.,:".characters)

sentence
  .characters
  .splitAt(puncSet.contains)
  .map(String.init)

// ["Hello", ", my name is oisdk", ". This should split", ": but only at punctuation", "!"]

Or, this version, which uses a for-loop, and splits after the delimiter:

extension CollectionType {
  func splitAt(@noescape isSplit: Generator.Element throws -> Bool) rethrows ->  [SubSequence] {
    var p = startIndex
    var result: [SubSequence] = []
    for i in indices where try isSplit(self[i]) {
      result.append(self[p...i])
      p = i.successor()
    }
    if p != endIndex { result.append(suffixFrom(p)) }
    return result
  }
}


extension CollectionType where Generator.Element : Equatable {
  func splitAt(splitter: Generator.Element) -> [SubSequence] {
    return splitAt { el in el == splitter }
  }
}

let sentence = "Hello, my name is oisdk. This should split: but only at punctuation!"

let puncSet = Set("!.,:".characters)

sentence
  .characters
  .splitAt(puncSet.contains)
  .map(String.init)

// ["Hello,", " my name is oisdk.", " This should split:", " but only at punctuation!"]

Or, if you wanted to get the most Swift features into one function (defer, throws, a Protocol extension, an evil flatMap, guard, and Optionals):

extension CollectionType {
  func splitAt(@noescape isSplit: Generator.Element throws -> Bool) rethrows -> [SubSequence] {
    var p = startIndex
    var result: [SubSequence] = try indices.flatMap { i in
      guard try isSplit(self[i]) else { return nil }
      defer { p = i.successor() }
      return self[p...i]
    }
    if p != endIndex { result.append(suffixFrom(p)) }
    return result
  }
}

Answer 2

Swift 3 and 4 Versions

extension Collection {
    func splitAt(isSplit: (Iterator.Element) throws -> Bool) rethrows -> [SubSequence] {
        var p = self.startIndex
        var result:[SubSequence] = try self.indices.flatMap {
            i in
            guard try isSplit(self[i]) else {
                return nil
            }
            defer {
                p = self.index(after: i)
            }
            return self[p...i]
        }
        if p != self.endIndex {
            result.append(suffix(from: p))
        }
        return result
    }
}

Thanks to Oisdk for getting me thinking.

Answer 3

I came here looking for an answer to this question. Didn't find what I was looking for and ended up building this by repeated calls to .split(...) It isn't elegant but you can choose which delimiters are preserved and which aren't. There's probably a way to avoid the String <--> Substring conversions, anyone know?

var input = """
    {All those moments will be (lost in time)},
    like tears [in rain](. ([(Time to)] die))
"""
var separator: Character = "!"
var output: [String] = []
repeat {
    let tokens = input.split(
        maxSplits: 1,
        omittingEmptySubsequences: false,
        whereSeparator: {
            switch $0 {
                case "{", "}", "(", ")", "[", "]": // preserve
                    separator = $0; return true
                case " ", "\n", ",", ".":          // omit
                    separator = " "; return true
                default:
                    return false
            }
        }
    )
    if tokens[0] != ""  { 
        output.append(String(tokens[0])) 
    }
    guard tokens.count == 2 else { break }
    if separator != " " { 
        output.append(String(separator)) 
    }
    input = String(tokens[1])
} while true

for token in output { print("\(token)") }

In the case above, the selectors are not in actual sets. I didn't need that, but if you do, simply make these declarations,

let preservedDelimiters: Set<Character> = [ "{", "}", "(", ")", "[", "]" ]
let omittedDelimiters: Set<Character> = [ " ", "\n", ",", "." ]

and replace the whereSeparator function with:

whereSeparator: {
    if preservedDelimiters.contains($0) {
        separator = $0
        return true
    } else if omittedDelimiters.contains($0) {
        separator = " "
        return true
    } else {
        return false
    }
}