4

There is a clever 1st answer given here for splitting swift string with a regex expression

split string answer

However it keeps the searched text within the array of answers. I'm trying to do a similar thing, but ignoring the characters acting as separators (e.g. just like the swift split function, but just with a regex expression as the separator).

As an example: regex would be something like 

"\\\||Z|ZY"

and when applied to string of "hi|thisZshouldZYbe|separated" then you would get back an array 

["hi", "this", "should", "be", "separated"]

NB. The regex is adapted to the swift NSRegularExpression format with the double escape. In regular regex it would just be "\||Z|ZY" Also NB the regex contains the vertical line symbol not the letter "l"

You prob don't need to do too many tweaks to the original to make work.

Poul Bak
  • 10,450
  • 5
  • 32
  • 57
domc
  • 147
  • 1
  • 12

2 Answers2

12

You can define an extension like this:

extension String {
    func split(usingRegex pattern: String) -> [String] {
        //### Crashes when you pass invalid `pattern`
        let regex = try! NSRegularExpression(pattern: pattern)
        let matches = regex.matches(in: self, range: NSRange(0..<utf16.count))
        let ranges = [startIndex..<startIndex] + matches.map{Range($0.range, in: self)!} + [endIndex..<endIndex]
        return (0...matches.count).map {String(self[ranges[$0].upperBound..<ranges[$0+1].lowerBound])}
    }
}

let str = "hi|thisZshouldZYbe|separated"
let separator = "\\||ZY?"
let result = str.split(usingRegex: separator)
print(result) //->["hi", "this", "should", "be", "separated"]

The above code does not work as you expect when you use "\\||Z|ZY", but I think you can modify your pattern to fit into this extension.

OOPer
  • 47,149
  • 6
  • 107
  • 142
  • Thank you. I think I can follow what you've done in the code - and makes sense. Very concise! and I didn't realise you could use map like that on the return statement - so many thanks. It's this conversion between NSRanges and normal ranges which makes it so tricky. I need to test once some other errors corrected. – domc Nov 10 '18 at 18:01
  • I've tested this and it is working well for my purposes. Thank you – domc Nov 13 '18 at 09:24
3

My suggestion is to create an UUID string as separator, then replace the occurrences of the regex pattern with this UUID string and split the string.

let string = "hi|thisZshouldZYbe|separated"
let uuid = UUID().uuidString
let result = string.replacingOccurrences(of: "\\||ZY?", with: uuid, options: .regularExpression).components(separatedBy: uuid)

Your pattern works only in another order of the OR parts ("\\||ZY|Z")

vadian
  • 274,689
  • 30
  • 353
  • 361